Initial fork commit

resources/addresses/bs.yaml

@@ -0,0 +1,585 @@
+# bs.yaml
+# -------
+# Bosnian language specification
+
+components:
+    level:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.7
+        alphanumeric_probability: 0.3
+
+    combinations:
+        -
+            components:
+                - house_number
+                - staircase
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.1
+        # For unit types like 2/34
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+
+
+numbers:
+    no_number: 
+        default:
+            canonical: bez broja
+            abbreviated: bb
+            sample: true
+            canonical_probability: 0.3
+            abbreviated_probability: 0.4
+            sample_probability: 0.3
+
+    default: &broj
+        canonical: broj
+        abbreviated: br
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.6
+        sample_probability: 0.1
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "br."
+            whitespace_probability: 0.6
+            direction: left
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+    alphanumeric_phrase_probability: 0.05
+    no_number_probability: 0.05
+
+
+and:
+    default: &i
+        canonical: i
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+cross_streets:
+    i: *i
+    at: &na
+        canonical: na
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    corner: &ugao
+        canonical: ugao
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    corner_of: &uglu
+        canonical: uglu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    na_uglu: &na_uglu
+        canonical: na uglu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *i
+        probability: 0.65
+        alternatives:
+            - alternative: *na
+              probability: 0.1
+            - alternative: *uglu
+              probability: 0.1
+            - alternative: *na_uglu
+              probability: 0.1
+            - alternative: *ugao
+              probability: 0.05
+
+    izmedu: &izmedu
+        canonical: između
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+    between:
+        default: *izmedu
+
+levels:
+    sprat: &sprat
+        canonical: sprat
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    kat: &kat
+        canonical: kat
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    prizemlje: &prizemlje
+        canonical: prizemlje
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    parter: &parter
+        canonical: parter
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    mezanino: &mezanin
+        canonical: mezanin
+        half_floors: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        sample: true
+        # e.g. mezanin 2
+        numeric:
+            direction: left
+        # e.g. 2. mezanin
+        ordinal:
+            direction: right
+        numeric_probability: 0.1
+        ordinal_probability: 0.2
+        standalone_probability: 0.6
+    podrum: &podrum
+        canonical: podrum
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        # e.g. podrum 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. 1. podrum
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+
+    aliases:
+        "<-1":
+            default: *podrum
+        "-1":
+            default: *podrum
+        # Special token for half-floors
+        half_floors:
+            default: *mezanin
+        "0":
+            default: *prizemlje
+            probability: 0.5
+            alternatives:
+                - alternative: *parter
+                  probability: 0.4
+                - alternative: *kat
+                  probability: 0.05
+                - alternative: *sprat
+                  probability: 0.05
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *kat
+        probability: 0.5
+        alternatives:
+            - alternative: *sprat
+              probability: 0.5
+        numeric_probability: 0.69 # With this probability, pick an integer
+        roman_numeral_probability: 0.3 # Pick a Roman numeral for the actual value
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+
+categories:
+    near:
+        default:
+            canonical: u blizini
+    nearby:
+        default:
+            canonical: u blizini
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: u blizini ovdje
+              probability: 0.3
+            - alternative:
+                canonical: ovde
+              probability: 0.1
+
+    near_me:
+        default:
+            canonical: u blizini mene
+
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: u
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+directions:
+    right: &desno
+        canonical: desno
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &lijevo
+        canonical: lijevo
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *desno
+          probability: 0.5
+        - alternative: *lijevo
+          probability: 0.5
+
+cardinal_directions:
+    east: &istok
+        canonical: istok
+        abbreviated: i
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: i
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &zapad
+        canonical: zapad
+        abbreviated: z
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: z
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &sjever
+        canonical: sjever
+        abbreviated: s
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &jug
+        canonical: jug
+        abbreviated: j
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: j
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *sjever
+          probability: 0.25
+        - alternative: *istok
+          probability: 0.23
+        - alternative: *jug
+          probability: 0.23
+        - alternative: *zapad
+          probability: 0.23
+
+entrances:
+    ulaz: &ulaz
+        canonical: ulaz
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Ulaz 1, Ulaz A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *ulaz
+        numeric_probability: 0.1 # e.g. Ulaz 1
+        alpha_probability: 0.85 # e.g. Ulaz A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+
+staircases:
+    stubiste: &stubiste
+        canonical: stubište
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+
+    alphanumeric: &staircase_alphanumeric
+        default: *stubiste
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: right
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *desno
+                      probability: 0.2
+                    - alternative: *lijevo
+                      probability: 0.2
+                    - alternative: *sjever
+                      probability: 0.15
+                    - alternative: *jug
+                      probability: 0.15
+                    - alternative: *istok
+                      probability: 0.15
+                    - alternative: *zapad
+                      probability: 0.15
+
+po_boxes:
+    postanski_pretinac: &postanski_pretinac
+        canonical: poštanski pretinac
+        abbreviated: p.p
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.4
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+
+    alphanumeric:
+        default: *postanski_pretinac
+        numeric_probability: 0.9 # pp 123
+        alpha_probability: 0.05 # p.p A
+        numeric_plus_alpha_probability: 0.04 # pp 123G
+        alpha_plus_numeric_probability: 0.01 # pp A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    stan: &stan
+        canonical: stan
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    apartman: &apartman
+        canonical: apartman
+        abbreviated: ap
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.2
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    soba: &soba
+        canonical: soba
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    ured: &ured
+        canonical: ured
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    alphanumeric: &unit_alphanumeric
+        default: *stan
+        probability: 0.6
+        alternatives:
+            - alternative: *apartman
+              probability: 0.3
+            - alternative: *soba
+              probability: 0.1
+        numeric_probability: 0.9 # e.g. stan. 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. stan A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.05
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *soba
+            probability: 0.6
+            alternatives:
+                - alternative: *ured
+                  probability: 0.4
+            numeric_probability: 0.95 # e.g. soba 1
+            numeric_plus_alpha_probability: 0.01 # e.g. soba 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. soba A1
+            alpha_probability: 0.03 # e.g. soba A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university:
+            default: *soba
+            numeric_probability: 0.95 # e.g. soba 1
+            numeric_plus_alpha_probability: 0.01 # e.g. soba 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. soba A1
+            alpha_probability: 0.03 # e.g. soba A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/ca.yaml

@@ -0,0 +1,856 @@
+# ca.yaml
+# -------
+# Catalan language specification
+
+components:
+    level:
+        # If no floor number is specified
+        null_probability: 0.6
+        alphanumeric_probability: 0.35
+        standalone_probability: 0.05
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        # If no unit number is specified
+        null_probability: 0.3
+        alphanumeric_probability: 0.65
+        standalone_probability: 0.05
+
+numbers:
+    default: &numero
+        canonical: número
+        abbreviated: "nº"
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.7
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#" # e.g. #3, #2F, etc.
+            probability: 0.5
+            alternatives:
+                - alternative: 
+            direction: left # affix goes on the number's left
+
+        # Probabilities for numbers
+        numeric_probability: 0.7
+        numeric_affix_probability: 0.3
+
+and:
+    default: &i
+        canonical: i
+        abbreviated: "&"
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability:  0.4
+        sample_probability: 0.1
+
+house_numbers:
+    # sense número (s/n) addresses
+    no_number: 
+        default:
+            canonical: sense número
+            abbreviated: s/n
+            sample: true
+            canonical_probability: 0.1
+            abbreviated_probability: 0.7
+            sample_probability: 0.2
+    alphanumeric:
+        default: *numero
+
+    alphanumeric_phrase_probability: 0.01
+    no_number_probability: 0.1 # With this probability, use sense número if no house_number is specified
+
+
+
+levels:
+    # Everywhere except Spain
+    floor: &pis
+        canonical: pis
+        abbreviated: p
+        sample: true
+        canonical_probability: 0.8
+        abbreviated_probability: 0.1
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true # Occasionally add variation of "number", e.g. Pis No 2
+            add_number_phrase_probability: 0.05
+        numeric_affix:
+            affix: p
+            direction: left # P2
+        # e.g. 2o piso
+        ordinal:
+            direction: right
+            direction_probability: 0.95 # Let it vary occasionally e.g. Pis 2o
+            standalone_probability: 0.2 # Let e.g. 5º be the entire floor string
+            # If ordinal is selected, chance of e.g. just using 2o without Piso
+            null_phrase_probability: 0.6
+        numeric_probability: 0.2
+        numeric_affix_probability: 0.05
+        ordinal_probability: 0.75
+    # Ground floor
+    baixos: &baixos
+        canonical: baixos
+        abbreviated: bxs
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.3
+        sample_probability: 0.1
+    pis_baix: &pis_baix
+        canonical: pis baix
+        abbreviated: pb
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.5
+        sample_probability: 0.1
+    sota: &sota
+        canonical: sota
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    # Used when floor number is < 0 (starts at -1 in all countries)
+    soterrani: &soterrani
+        canonical: soterrani
+        abbreviated: so
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.3
+        sample_probability: 0.2
+        # e.g. soterrani 1
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: so
+            direction: left
+        # e.g. segon soterrani
+        ordinal:
+            direction: right
+        standalone_probability: 0.985
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.005
+        ordinal_probability: 0.005
+    sub_soterrani: &sub_soterrani
+        canonical: sub soterrani
+        abbreviated: ss
+        sample: true
+        # e.g. sub soterrani 1
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: ss
+            direction: left
+        # e.g. segon sub soterrani
+        ordinal:
+            direction: right
+        number_abs_value: true
+        number_min_abs_value: 2
+        # Soterrani 2 == Sub-soterrani 1
+        number_subtract_abs_value: 1
+        standalone_probability: 0.985
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.005
+        ordinal_probability: 0.005
+    entresol: &entresol
+        canonical: entresòl
+        abbreviated: entl
+        half_floors: true
+        sample: true
+        canonical_probability: 0.7
+        abbreviated_probability: 0.2
+        sample_probability: 0.1
+        # e.g. entresòl 2
+        numeric:
+            direction: left
+        # e.g. ent2
+        numeric_affix:
+            affix: ent
+            direction: left
+        # e.g. segon entresòl
+        ordinal:
+            direction: right
+        numeric_probability: 0.1
+        numeric_affix_probability: 0.1
+        ordinal_probability: 0.2
+        standalone_probability: 0.6
+    pis_principal: &pis_principal
+        canonical: pis principal
+        abbreviated: pis pral
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.3
+        sample_probability: 0.5
+    principal: &principal
+        canonical: principal
+        abbreviated: pral
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.6
+        sample_probability: 0.2
+    atic: &atic
+        canonical: àtic
+        abbreviated: át
+        sample: true
+        canonical_probability: 0.7
+        abbreviated_probability: 0.1
+        sample_probability: 0.2
+    sobreatic: &sobreatic
+        canonical: sobreàtic
+    aliases:
+        "<-1":
+            default: *soterrani
+            probability: 0.6
+            alternatives:
+                - alternative: *sub_soterrani
+                  probability: 0.3995
+                - alternative: *pis
+                  probability: 0.0005
+        "-1":
+            default: *soterrani
+            probability: 0.9995
+            alternatives:
+                - alternative: *pis
+                  probability: 0.0005
+        # Special token for half-floors
+        half_floors:
+            default: *entresol
+        "0":
+            default: *baixos
+            probability: 0.495
+            alternatives:
+                - alternative: *pis_baix
+                  probability: 0.395
+                - alternative: *sota
+                  probability: 0.1
+                - alternative: *pis
+                  # Piso 0 is uncommon
+                  probability: 0.01
+        top:
+            default: *pis
+            probability: 0.85
+            alternatives:
+                - alternative: *atic
+                  probability: 0.1
+                - alternative: *sobreatic
+                  probability: 0.05
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *pis
+        add_number_phrase: true
+        add_number_phrase_probability: 0.05
+        numeric_probability: 0.99
+        alpha_probability: 0.01
+
+blocks:
+    alphanumeric:
+        default:
+            canonical: bloc
+            abbreviated: bl
+            sample: true
+            canonical_probability: 0.6
+            abbreviated_probability: 0.2
+            sample_probability: 0.2
+            numeric:
+                direction: left
+
+categories:
+    near:
+        default:
+            canonical: a prop de
+        probability: 0.5
+        alternatives:
+            - alternative:
+                  canonical: prop de
+              probability: 0.2
+            - alternative:
+                  canonical: prop
+              probability: 0.1
+            - alternative:
+                  canonical: a prop
+              probability: 0.1
+            - alternative:
+                  canonical: proper
+              probability: 0.05
+            - alternative:
+                  canonical: proper a
+              probability: 0.05
+
+    nearby:
+        default:
+            canonical: proper
+        probability: 0.5
+        alternatives:
+            - alternative:
+                  canonical: a prop
+              probability: 0.1
+            - alternative:
+                  canonical: a prop d'aquí
+              probability: 0.1
+            - alternative:
+                  canonical: a prop d'aqui
+              probability: 0.1
+            - alternative:
+                  canonical: aquí
+              probability: 0.1
+            - alternative:
+                  canonical: aqui
+              probability: 0.1
+    near_me:
+        default:
+            canonical: a prop meu
+    in:
+        default:
+            canonical: a
+        probability: 0.6
+        alternatives:
+            - alternative:
+                  canonical: dins
+              probability: 0.2
+            - alternative:
+                  canonical: en
+              probability: 0.2
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+cross_streets:
+    and: *i
+    amb: &amb
+        canonical: amb
+    a: &a
+        canonical: a
+    corner_of: &cantonada_de
+        canonical: cantonada de
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+    at_the_corner_of: &a_la_cantonada_de
+        canonical: a la cantonada de
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+    corner: &cantonada
+        canonical: cantonada
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+
+    intersection:
+        default: *i
+        probability: 0.55
+        alternatives:
+            - alternative: *amb
+              probability: 0.2
+            - alternative: *a
+              probability: 0.1
+            - alternative: *cantonada_de
+              probability: 0.09
+            - alternative: *a_la_cantonada_de
+              probability: 0.05
+            - alternative: *cantonada
+              probability: 0.01
+
+    between:
+        canonical: entre
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probabililty: 0.5
+
+
+po_boxes:
+    apartat: &apartat
+        canonical: apartat
+        abbreviated: apt
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.3
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.4 # Apt No 1234
+        numeric_probability: 1.0
+    alphanumeric:
+        sample: false
+        default: *apartat
+        numeric_probability: 0.9 # Apt 123
+        alpha_probability: 0.05 # Apt A
+        numeric_plus_alpha_probability: 0.04 # Apt 123G
+        alpha_plus_numeric_probability: 0.01 # Apt A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+postcodes:
+    alphanumeric:
+        default:
+            canonical: codi postal
+            abbreviated: cp
+            sample: true
+            canonical_probability: 0.01
+            abbreviated_probability: 0.95
+            sample_probability: 0.04
+
+            numeric:
+                # Postcodes in Spain and Latin America are sometimes prefixed by CP
+                direction: left
+
+            numeric_affix:
+                affix: cp
+                direction: left
+            # null_probability means the chance of doing nothing e.g. just the postal code
+            null_probability: 0.7
+            numeric_probability: 0.18
+            numeric_affix_probability: 0.12
+            strict_numeric: true
+
+directions:
+    right: &dreta
+        canonical: dreta
+        abbreviated: dta
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: d
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+    left: &esquerra
+        canonical: esquerra
+        abbreviated: esq
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: e
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+    rear: &posterior
+        canonical: posterior
+        abbreviated: pos
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.2
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    front: &front
+        canonical: front
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *dreta
+          probability: 0.45
+        - alternative: *esquerra
+          probability: 0.45
+        - alternative: *posterior
+          probability: 0.05
+        - alternative: *front
+          probability: 0.05
+
+    anteroposterior:
+        alternatives:
+            - alternative: *front
+              probability: 0.5
+            - alternative: *posterior
+              probability: 0.5
+
+    lateral:
+        alternatives:
+            - alternative: *dreta
+              probability: 0.5
+            - alternative: *esquerra
+              probability: 0.5
+
+
+
+
+cardinal_directions:
+    east: &est
+        canonical: est
+        abbreviated: e
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6 
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: e
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &oest
+        canonical: oest
+        abbreviated: w
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: w
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &nord
+        canonical: nord
+        abbreviated: n
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &sud
+        canonical: sud
+        abbreviated: s
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *nord
+          probability: 0.25
+        - alternative: *est
+          probability: 0.25
+        - alternative: *sud
+          probability: 0.25
+        - alternative: *oest
+          probability: 0.25
+
+entrances:
+    entrada: &entrada
+        canonical: entrada
+        abbreviated: entr
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.2
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    # Entrance 1, Entrance A, etc.
+    alphanumeric:
+        default: *entrada
+        numeric_probability: 0.1 # e.g. Entrance 1
+        alpha_probability: 0.85 # e.g. Entrnace A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            modifier:
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *sud
+                    - alternative: *est
+                    - alternative: *oest
+                    - alternative: *dreta
+                    - alternative: *esquerra
+                    - alternative: *posterior
+                    - alternative: *front
+
+staircases:
+    escala: &escala
+        canonical: escala
+        abbreviated: esc
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    alphanumeric:
+        # For alphanumerics, Stair A, Stair 1, etc.
+        default: *escala
+        numeric_probability: 0.6 # e.g. Escalera 1
+        alpha_probability: 0.35 # e.g. Escalera A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: right # e.g. Escalera Izq
+            direction_probability: 0.8
+            modifier:
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *sud
+                    - alternative: *est
+                    - alternative: *oest
+                    - alternative: *dreta
+                    - alternative: *esquerra
+                    - alternative: *posterior
+                    - alternative: *front
+
+units:
+    flat: &apartament
+        canonical: apartament
+        abbreviated: apmt
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    door: &porta
+        canonical: porta
+        abbreviated: pta
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            # If it's just puerta B, many times it's just e.g. 3o B for "tercero piso puerta B"
+            null_phrase_probability: 0.15
+        ordinal:
+            direction: right
+            gender: f
+            direction_probability: 0.95 # Let it vary occasionally e.g. Porta 2a
+            null_phrase_probability: 0.8 # Let e.g. 5a be the entire unit string
+        numeric_probability: 0.25
+        ordinal_probability: 0.75
+    lletra: &lletra
+        canonical: lletra
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    office: &oficina
+        canonical: oficina
+        abbreviated: of
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.3
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    # Another word for unit, used more in Colombia
+    unitat: &unitat
+        canonical: unitat
+        abbreviated: un
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+    lot: &lot
+        canonical: lot
+        abbreviated: lt
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.2
+        sample_probability: 0.2
+    parcel: &parcella
+        canonical: parcel·la
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    habitacio: &habitacio
+        canonical: habitació
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    casa: &casa
+        canonical: casa
+        numeric:
+            direction: left
+    room: &sala
+        canonical: sala
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *porta
+        probability: 0.8
+        sample: true
+        alternatives:
+            - alternative: *apartament
+              probability: 0.1
+            - alternative: *casa
+              probability: 0.1
+
+        # Separate random probability for adding directions like 2o Izq, 2 Dcha, etc.
+        add_direction: true
+        add_direction_probability: 0.1
+        add_direction_numeric: true # Only for numbers
+        add_direction_standalone: true # A unit can be as simple as "D"
+
+        numeric_probability: 0.7 # e.g. Porta 1a
+        numeric_plus_alpha_probability: 0.01 # e.g. Porta 1A
+        alpha_plus_numeric_probability: 0.01 # e.g. Porta A1
+        alpha_probability: 0.28 # e.g. Porta A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        alpha:
+            default: *porta
+            probability: 0.8
+            alternatives:
+                - alternative: *lletra
+                  probability: 0.12
+                - alternative: *apartament
+                  probability: 0.05
+                - alternative: *casa
+                  probability: 0.01
+                - alternative: *unitat
+                  probability: 0.01
+                - alternative: *habitacio
+                  probability: 0.01
+
+    zones:
+        residential: *unit_alphanumeric
+        commercial:
+            default: *oficina
+            probability: 0.8
+            alternatives:
+                - alternative: *sala
+                  probability: 0.2
+
+            numeric_probability: 0.9 # e.g. Oficina 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Oficina 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Oficina A1
+            alpha_probability: 0.08 # e.g. Oficina A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+            alpha:
+                default: *oficina
+                probability: 0.8
+                alternatives:
+                    - alternative: *sala
+                      probability: 0.15
+                    - alternative: *lletra
+                      probability: 0.05
+
+        industrial:
+            default: *lot
+            probability: 0.5
+            alternatives:
+                - alternative: *oficina
+                  probability: 0.3
+                - alternative: *unitat
+                  probability: 0.19
+                - alternative: *parcella
+                  probability: 0.01
+
+            numeric_probability: 0.9 # e.g. Lote 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Lote 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Lote A1
+            alpha_probability: 0.08 # e.g. Lote A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+        university:
+            default: *sala
+            probability: 0.9
+            alternatives:
+                - alternative: *porta
+                  probability: 0.1
+
+            numeric_probability: 0.9 # e.g. Sala 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Sala 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Sala A1
+            alpha_probability: 0.08 # e.g. Sala A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+            alpha:
+                default: *sala
+                probability: 0.9
+                alternatives:
+                    - alternative: *porta
+                      probability: 0.08
+                    - alternative: *lletra
+                      probability: 0.02
+
+    allotments:
+        lot:
+            default: *lot
+            numeric_probability: 0.8
+            alphanumeric_probability: 0.1
+            alpha_probability: 0.1
+        parcel:
+            default: *parcella
+            numeric_probability: 0.3
+            alphanumeric_probability: 0.3
+            alpha_probability: 0.4
+        lot_probability: 0.9
+        parcel_probability: 0.06
+        lot_plus_parcel_probability: 0.02
+        parcel_plus_lot_probability: 0.02

resources/addresses/cs.yaml

@@ -0,0 +1,570 @@
+# cs.yaml
+# -------
+# Czech language specification
+
+
+components:
+    level:
+        null_probability: 0.95
+        alphanumeric_probability: 0.04
+        standalone_probability: 0.01
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    # Note: no combinations because of the house numbering scheme
+
+numbers:
+    default: &cislo
+        canonical: číslo
+        abbreviated: č
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.6
+        sample_probability: 0.1
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "č."
+            direction: left
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+and:
+    default: &a
+        canonical: a
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+conscription_numbers:
+    alphanumeric:
+        default: 
+            canonical: číslo popisné
+            abbreviated: "č.p."
+            canonical_probability: 0.05
+            abbreviated_probability: 0.85
+            sample: true
+            sample_probability: 0.1
+            numeric:
+                direction: left
+
+cross_streets:
+    and: *a
+    at: &na
+        canonical: na
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2        
+    corner_of: &rohu
+        canonical: rohu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    corner: &roh
+        canonical: roh
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    at_the_corner_of: &na_rohu
+        canonical: na rohu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *a
+        probability: 0.6
+        alternatives:
+            - alternative: *na
+              probability: 0.1
+            - alternative: *rohu
+              probability: 0.1
+            - alternative: *roh
+              probability: 0.1
+            - alternative: *na_rohu
+              probability: 0.1
+
+    between:
+        canonical: mezi
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &patro
+        canonical: patro
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    nadzemni_podlazi: &nadzemni_podlazi
+        canonical: nadzemní podlaží
+        abbreviated: np
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.8
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    etaz: &etaz
+        canonical: etáž
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    prizemi: &prizemi
+        canonical: přízemí
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    podzemni_podlazi: &podzemni_podlazi
+        canonical: podzemní podlaží
+        abbreviated: pp
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.2
+        sample_probability: 0.3
+        # e.g. podzemní podlaží 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. pp1
+        numeric_affix:
+            affix: pp
+            direction: left
+        # e.g. 1. podzemní podlaží
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        standalone_probability: 0.985
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.005
+        ordinal_probability: 0.005
+    aliases:
+        "<-1":
+            default: *podzemni_podlazi
+        "-1":
+            default: *podzemni_podlazi
+        "0":
+            default: *prizemi
+            probability: 0.9
+            alternatives:
+                - alternative: *patro
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *patro
+        probability: 0.8
+        alternatives:
+            - alternative: *nadzemni_podlazi
+              probability: 0.19
+            - alternative: *etaz
+              probability: 0.01
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+categories:
+    near:
+        default:
+            canonical: poblíž
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.75
+        alternatives:
+            - alternative:
+                canonical: v blízkém okolí
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: u
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: kolem
+                sample: true
+                canonical_probability: 0.7
+                sample_probability: 0.3
+              probability: 0.05
+    nearby:
+        default:
+            canonical: poblíž
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.45
+        alternatives:
+            - alternative:
+                canonical: blízko
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: v blízkosti
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: tady poblíž
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: tady
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05
+            - alternative:
+                canonical: okolo
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05
+            - alternative:
+                canonical: v okolí
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05
+    near_me:
+        default:
+            canonical: v blízkosti mně
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: v
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: ve
+              probability: 0.3
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+directions:
+    right: &prava
+        canonical: pravá
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &leva
+        canonical: levá
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *prava
+          probability: 0.5
+        - alternative: *leva
+          probability: 0.5
+
+cardinal_directions:
+    east: &vychod
+        canonical: východ
+        abbreviated: v
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &zapad
+        canonical: západ
+        abbreviated: z
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: z
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &sever
+        canonical: sever
+        abbreviated: s
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &jih
+        canonical: jih
+        abbreviated: j
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: j
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *sever
+          probability: 0.25
+        - alternative: *vychod
+          probability: 0.25
+        - alternative: *jih
+          probability: 0.25
+        - alternative: *zapad
+          probability: 0.25
+entrances:
+    vchod: &vchod
+        canonical: vchod
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Wejście 1, Wejście A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *vchod
+        numeric_probability: 0.1 # e.g. Wejście 1
+        alpha_probability: 0.85 # e.g. Wejście A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    schodiste: &schodiste
+        canonical: schodiště
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *schodiste
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *sever
+                    - alternative: *jih
+                    - alternative: *vychod
+                    - alternative: *zapad
+
+po_boxes:
+    postovni_prihradka: &postovni_prihradka
+        canonical: poštovní přihrádka
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # poštovní přihrádka 1234
+    alphanumeric:
+        default: *postovni_prihradka
+        numeric_probability: 0.9 # poštovní přihrádka 123
+        alpha_probability: 0.05 # poštovní přihrádka A
+        numeric_plus_alpha_probability: 0.04 # poštovní přihrádka 123G
+        alpha_plus_numeric_probability: 0.01 # poštovní přihrádka A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    apartaman: &apartaman
+        canonical: apartmán
+        abbreviated: apt
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.5
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    pokoj: &pokoj
+        canonical: pokoj
+        abbreviated: pok
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.5
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    kancelar: &kancelar
+        canonical: kancelář
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    alphanumeric: &unit_alphanumeric
+        default: *apartaman
+        probability: 0.9
+        alternatives:
+            - alternative: *pokoj
+              probability: 0.1
+        numeric_probability: 0.9 # e.g. apt. 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. apt. A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.01
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *pokoj
+            probability: 0.6
+            alternatives:
+                - alternative: *kancelar
+                  probability: 0.4
+            numeric_probability: 0.95 # e.g. pokoj 1
+            numeric_plus_alpha_probability: 0.01 # e.g. pokoj 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. pokoj A1
+            alpha_probability: 0.03 # e.g. pokoj A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university:
+            default: *pokoj
+            numeric_probability: 0.95 # e.g. pokoj 1
+            numeric_plus_alpha_probability: 0.01 # e.g. pok 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. pokoj A1
+            alpha_probability: 0.03 # e.g. pokoj A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/da.yaml

@@ -0,0 +1,599 @@
+# da.yaml
+# -------
+# Danish language specification.
+
+components:
+    level:
+        null_probability: 0.85
+        alphanumeric_probability: 0.1
+        standalone_probability: 0.05
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+    combinations:
+        -
+            components:
+                - level
+                - unit
+            label: unit
+            separators:
+                - separator: "-"
+                  probability: 0.9
+                - separator: " - "
+                  probability: 0.1
+            probability: 0.005
+        -
+            components:
+                - entrance
+                - unit
+            label: unit
+            separators:
+                - separator: "-"
+                  probability: 0.9
+                - separator: " - "
+                  probability: 0.1
+            probability: 0.001
+
+
+numbers:
+    default: &nummer
+        canonical: nummer
+        abbreviated: nr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+house_numbers:
+    alphanumeric:
+        default: *nummer
+
+    alphanumeric_phrase_probability: 0.0001
+
+
+and:
+    default: &og
+        canonical: og
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+cross_streets:
+    and: *og
+    corner_of: &hjorne_af
+        canonical: hjørne af
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    at_the_corner_of: &pa_hjornet_af
+        canonical: på hjørnet af
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *og
+        probability: 0.7
+        alternatives:
+            - alternative: *hjorne_af
+              probability: 0.15
+            - alternative: *pa_hjornet_af
+              probability: 0.15
+
+    between:
+        canonical: mellem
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &sal
+        canonical: sal
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+            direction_probability: 0.9
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    etage: &etage
+        canonical: etage
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    stuen: &stuen
+        canonical: stuen
+        abbreviated: st
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.6
+        sample_probability: 0.1
+    stueetage: &stueetage
+        canonical: stueetage
+        sample: true
+        canonical_probability: 0.3
+        sample_probability: 0.7
+    kaelderen: &kaelderen
+        canonical: kælderen
+        abbreviated: kl
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.6
+        sample_probability: 0.2
+        # e.g. 1 kælderen
+        numeric:
+            direction: right
+            direction_probability: 0.8
+        # e.g. k1
+        numeric_affix:
+            affix: k
+            direction: left
+        # e.g. 1. kl
+        ordinal:
+            direction: right
+        standalone_probability: 0.985
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.005
+        ordinal_probability: 0.005
+    aliases:
+        "<-1":
+            default: *kaelderen
+        "-1":
+            default: *kaelderen
+        "0":
+            default: *stuen
+            probability: 0.9
+            alternatives:
+                - alternative: *stueetage
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *sal
+        probability: 0.7
+        alternatives:
+            - alternative: *etage
+              probability: 0.3
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+categories:
+    near:
+        default:
+            canonical: i nærheden af
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: tæt på
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: tæt ved
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+    nearby:
+        default:
+            canonical: i nærheden
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.4
+        alternatives:
+            - alternative:
+                canonical: rundt her
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: nær her
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: nær
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative: 
+                canonical: omkring her
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: tæt på her
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+    near_me:
+        default:
+            canonical: nær ​​mig
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.8
+        alternatives:
+            - alternative:
+                canonical: i nærheden af ​​mig
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: tæt på mig
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: i
+        probability: 0.8
+        alternatives:
+            - alternative:
+                canonical: om
+              probability: 0.1
+            - alternative:
+                canonical: på
+              probability: 0.1
+
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+
+directions:
+    right: &til_hojre
+        canonical: til højre
+        abbreviated: t.h
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.5
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: t.h
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    left: &til_venstre
+        canonical: til venstre
+        abbreviated: t.v
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.6
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: t.v
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    middle: &midt_for
+        canonical: midt for
+        abbreviated: m.f
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.6
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: m.f
+            direction: right
+            whitespace_probability: 0.1
+    alternatives:
+        - alternative: *til_hojre
+          probability: 0.45
+        - alternative: *til_venstre
+          probability: 0.45
+        - alternative: *midt_for
+          probability: 0.1
+
+
+cardinal_directions:
+    east: &ost
+        canonical: øst
+        abbreviated: ø
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: ø
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &vest
+        canonical: vest
+        abbreviated: v
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &nord
+        canonical: nord
+        abbreviated: n
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &syd
+        canonical: syd
+        abbreviated: s
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *nord
+          probability: 0.25
+        - alternative: *ost
+          probability: 0.25
+        - alternative: *syd
+          probability: 0.25
+        - alternative: *vest
+          probability: 0.25
+
+
+entrances:
+    indgang: &indgang
+        canonical: indgang
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Eingang 1, Eingang A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *indgang
+        numeric_probability: 0.1 # e.g. Eingang 1
+        alpha_probability: 0.85 # e.g. Eingang A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    stiege: &stiege
+        canonical: stiege
+        abbreviated: stg
+        sample: true
+        canonical_probability: 0.7
+        abbreviated_probability: 0.2
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    trappe: &trappe
+        canonical: trappe
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *trappe
+        probability: 0.8
+        alternatives:
+            - alternative: *stiege
+              probability: 0.2
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *syd
+                    - alternative: *ost
+                    - alternative: *vest
+
+po_boxes:
+    postboks: &postboks
+        canonical: postboks
+        abbreviated: pb
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.2
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # Pb No 1234
+    boks: &boks
+        canonical: boks
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # Boks No 1234
+    alphanumeric:
+        sample: false
+        default: *postboks
+        probability: 0.9
+        alternatives:
+            - alternative: *boks
+              probability: 0.1
+        numeric_probability: 0.9 # Pb 123
+        alpha_probability: 0.05 # Pb A
+        numeric_plus_alpha_probability: 0.04 # Pb 123G
+        alpha_plus_numeric_probability: 0.01 # Pb A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    lejlighed: &lejlighed
+        canonical: lejlighed
+        abbreviated: ljd
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.1
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            null_phrase_probability: 0.5
+        # Lejlighed nummer 4
+        add_number_phrase: true
+        add_number_phrase_probability: 0.05
+    hus: &hus
+        canonical: hus
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    vaerelse: &vaerelse
+        canonical: værelse
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *lejlighed
+        probability: 0.8
+        alternatives:
+            - alternative: *hus
+              probability: 0.1
+            - alternative: *vaerelse
+              probability: 0.1
+        numeric_probability: 0.9 # e.g. Lejlighed 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. Lejl A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # Separate random probability for adding directions like 2R, 2L, etc.
+        add_direction: true
+        add_direction_probability: 0.5
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+        # Add direction only e.g. Lejlighed Rechts
+        add_direction_standalone: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.1

resources/addresses/de.yaml

@@ -0,0 +1,728 @@
+# de.yaml
+# -------
+# Note: this will only apply to the German language code, which encompasses Germany,
+# Austria, Switzerland (but not Swiss-German, which has its own language code),
+# Lichtenstein, Luxembourg (Luxembourgish has its own language code), and part of Belgium.
+
+
+components:
+    level:
+        null_probability: 0.85
+        alphanumeric_probability: 0.1
+        standalone_probability: 0.05
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    combinations:
+        # e.g. 2/34, more common way to specify a unit number in German
+        # if unit exists in the first place
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: /
+                  probability: 0.8
+                - separator: "-"
+                  probability: 0.1
+                - separator: " - "
+                  probability: 0.1
+            probability: 0.05
+
+
+numbers:
+    default: &nummer
+        canonical: nummer
+        abbreviated: nr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+house_numbers:
+    gebaude: &gebaude
+        canonical: gebäude
+        abbreviated: geb
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.5
+        sample_probability: 0.05
+        numeric:
+            direction: left
+    alphanumeric:
+        default: *nummer
+        probability: 0.95
+        alternatives:
+            - alternative: *gebaude
+              probability: 0.05
+
+    alphanumeric_phrase_probability: 0.05
+
+conscription_numbers:
+    alphanumeric:
+        default: 
+            canonical: konskriptionsnummer
+            abbreviated: konskr. nr
+            canonical_probability: 0.15
+            abbreviated_probability: 0.65
+            sample: true
+            sample_probability: 0.2
+            numeric:
+                direction: left
+
+and:
+    default: &und
+        canonical: und
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+cross_streets:
+    and: *und
+    corner_of: &ecke_von
+        canonical: ecke von
+    at_the_corner_of: &an_der_ecke_von
+        canonical: an der ecke von
+    intersection:
+        default: *und
+        probability: 0.7
+        alternatives:
+            - alternative: *ecke_von
+              probability: 0.15
+            - alternative: *an_der_ecke_von
+              probability: 0.15
+
+    between:
+        canonical: zwischen
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &obergeschoss
+        canonical: obergeschoss
+        abbreviated: og
+        sample: true
+        add_number_phrase: true
+        add_number_phrase_probability: 0.1
+        canonical_probability: 0.5
+        abbreviated_probability: 0.4
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: og
+            direction: right
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.3
+        numeric_affix_probability: 0.5
+        ordinal_probability: 0.2
+    etage: &etage
+        canonical: etage
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    stock: &stock
+        canonical: stock
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.1
+        ordinal_probability: 0.9
+    erdgeschoss: &erdgeschoss
+        canonical: erdgeschoss
+        abbreviated: eg
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.6
+        sample_probability: 0.1
+    untergeschoss: &untergeschoss
+        canonical: untergeschoss
+        abbreviated: ug
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.5
+        sample_probability: 0.1
+        # e.g. Basement 1
+        numeric:
+            direction: left
+        # e.g. 1ug
+        numeric_affix:
+            affix: ug
+            direction: left
+        # e.g. 1. UG
+        ordinal:
+            direction: right
+        standalone_probability: 0.985
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.005
+        ordinal_probability: 0.005
+    unterste_etage: &unterste_etage
+        canonical: unterste etage
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    oberste_etage: &oberste_etage
+        canonical: oberste etage
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+
+    aliases:
+        "<-1":
+            default: *untergeschoss
+        "-1":
+            default: *untergeschoss
+        "0":
+            default: *erdgeschoss
+            probability: 0.9
+            alternatives:
+                - alternative: *unterste_etage
+                  probability: 0.1
+        "top":
+            default: *obergeschoss
+            probability: 0.75
+            alternatives:
+                - alternative: *stock
+                  probability: 0.1
+                - alternative: *etage
+                  probability: 0.05
+                - alternative: *oberste_etage
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *obergeschoss
+        probability: 0.85
+        alternatives:
+            - alternative: *stock
+              probability: 0.1
+            - alternative: *etage
+              probability: 0.05
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+categories:
+    near:
+        default:
+            canonical: nähe
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.5
+        alternatives:
+            - alternative:
+                canonical: bei
+              probability: 0.3
+            - alternative:
+                canonical: nah
+              probability: 0.15
+            - alternative:
+                canonical: nahe an
+              probability: 0.05
+    nearby:
+        default:
+            canonical: hier in der nähe
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.4
+        alternatives:
+            - alternative:
+                canonical: in der nähe
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.3
+            - alternative:
+                canonical: in der nähe hier
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative: 
+                canonical: in der nähe von
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: nahe gelegen
+              probability: 0.05
+            - alternative:
+                canonical: hier in der gegend
+              probability: 0.05
+
+    near_me:
+        default:
+            canonical: in meiner nähe
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.9
+        alternatives:
+            - alternative:
+                canonical: in der nähe zu mir
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: in
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: im
+              probability: 0.2
+            - alternative:
+                canonical: um
+              probability: 0.2
+
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+
+directions:
+    right: &rechts
+        canonical: rechts
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: r
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    left: &links
+        canonical: links
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: l
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+    alternatives:
+        - alternative: *rechts
+          probability: 0.5
+        - alternative: *links
+          probability: 0.5
+
+
+cardinal_directions:
+    east: &ost
+        canonical: ost
+        abbreviated: o
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: o
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &west
+        canonical: west
+        abbreviated: w
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: w
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &nord
+        canonical: nord
+        abbreviated: n
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &sud
+        canonical: süd
+        abbreviated: s
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *nord
+          probability: 0.25
+        - alternative: *ost
+          probability: 0.25
+        - alternative: *sud
+          probability: 0.25
+        - alternative: *west
+          probability: 0.25
+
+
+entrances:
+    eingang: &eingang
+        canonical: eingang
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Eingang 1, Eingang A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *eingang
+        numeric_probability: 0.1 # e.g. Eingang 1
+        alpha_probability: 0.85 # e.g. Eingang A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    stiege: &stiege
+        canonical: stiege
+        abbreviated: stg
+        sample: true
+        canonical_probability: 0.7
+        abbreviated_probability: 0.2
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    treppe: &treppe
+        canonical: treppe
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *stiege
+        probability: 0.6
+        alternatives:
+            - alternative: *treppe
+              probability: 0.4
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *sud
+                    - alternative: *ost
+                    - alternative: *west
+
+po_boxes:
+    postfach: &postfach
+        canonical: postfach
+        abbreviated: pf
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.2
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # PF No 1234
+        numeric_probability: 1.0
+    alphanumeric:
+        sample: false
+        default: *postfach
+        numeric_probability: 0.9 # Apdo 123
+        alpha_probability: 0.05 # Apdo A
+        numeric_plus_alpha_probability: 0.04 # Apdo 123G
+        alpha_plus_numeric_probability: 0.01 # Apdo A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    halle: &halle
+        canonical: halle
+        numeric:
+            direction: left
+    wohnung: &wohnung
+        canonical: wohnung
+        abbreviated: whg
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.1
+        sample_probability: 0.3
+        plural:
+            canonical: wohnungen
+        numeric:
+            direction: left
+        # Wohnung nummer 4
+        add_number_phrase: true
+        add_number_phrase_probability: 0.2
+    haus: &haus
+        canonical: haus
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    wohnungsnummer: &wohnungsnummer
+        canonical: wohnungsnummer
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+    appartement: &appartement
+        canonical: appartement
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    buro: &buro
+        canonical: büro
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    zimmer: &zimmer
+        canonical: zimmer
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *wohnung
+        probability: 0.8
+        alternatives:
+            - alternative: *wohnungsnummer
+              probability: 0.1
+            - alternative: *appartement
+              probability: 0.05
+            - alternative: *haus
+              probability: 0.05
+
+        numeric_probability: 0.9 # e.g. Wohnung 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. Wohnung A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # Separate random probability for adding directions like 2R, 2L, etc.
+        add_direction: true
+        add_direction_probability: 0.1
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+        # Add direction only e.g. Wohnung Rechts
+        add_direction_standalone: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.1
+
+    zone:
+        residential: *unit_alphanumeric
+        commercial:
+            default: *buro
+            probability: 0.9
+            alternatives:
+                - alternative: *zimmer
+                  probability: 0.1
+        university:
+            default: *halle
+            probability: 0.9
+            alternatives:
+                - alternative: *zimmer
+                  probability: 0.1
+
+
+countries:
+    # Austria
+    at:
+        # Staircase and entrance numbers more common
+        components:
+            level:
+                null_probability: 0.6
+                alphanumeric_probability: 0.3
+                standalone_probability: 0.1
+            staircase:
+                null_probability: 0.9
+                alphanumeric_probability: 0.1
+
+            entrance:
+                null_probability: 0.99
+                alphanumeric_probability: 0.01
+
+            unit:
+                null_probability: 0.4
+                alphanumeric_probability: 0.6
+
+            # Combined apartment numbers are very common      
+            combinations:
+                # e.g. Neubaugasse 55/A/1/5
+                -
+                    components:
+                        - house_number
+                        - entrance
+                        - staircase
+                        - unit
+                    label: house_number
+                    separators:
+                        - separator: /
+                          probability: 0.98
+                        - separator: "-"
+                          probability: 0.02
+                    probability: 0.9
+                # e.g. Neubaugasse 55/1/5
+                -
+                    components:
+                        - house_number
+                        - staircase
+                        - unit
+                    label: house_number
+                    separators:
+                        - separator: /
+                          probability: 0.98
+                        - separator: "-"
+                          probability: 0.02
+                    probability: 0.8
+                # e.g. Neubaugasse 55/5
+                -
+                    components:
+                        - house_number
+                        - unit
+                    label: house_number
+                    probability: 0.7
+                    separators:
+                        - separator: /
+                          probability: 0.98
+                        - separator: "-"
+                          probability: 0.02
+
+        units:
+            top: &top
+                canonical: top
+                numeric:
+                    direction: left
+            alphanumeric: &austria_units_alphanumeric
+                default: *top
+                probability: 0.75
+                alternatives:
+                    - alternative: *haus
+                      probability: 0.15
+                    - alternative: *wohnung
+                      probability: 0.05
+                    - alternative: *wohnungsnummer
+                      probability: 0.025
+                    - alternative: *appartement
+                      probability: 0.025

resources/addresses/el.yaml

@@ -0,0 +1,368 @@
+# el.yaml
+# -------
+# Greek language specification
+
+
+alphabet: αβγδεζηθικλμνξοπρστυφχψω
+alphabet_probability: 0.8
+
+components:
+    level:
+        null_probability: 0.95
+        alphanumeric_probability: 0.05
+
+    entrance:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    unit:
+        null_probability: 0.6
+        alphanumeric_probability: 0.4
+
+
+    combinations:
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.1
+
+levels:
+    orofos: &orofos
+        canonical: όροφος
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: left
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    orofos_latin: &orofos_latin
+        canonical: órofos
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+        ordinal:
+            direction: left
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+
+    isogelo: &isogelo
+        canonical: ισόγειο
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    isogelo_latin: &isogelo_latin
+        canonical: isógeio
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+    imiorofos: &imiorofos
+        canonical: ημιώροφος
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    imiorofos_latin: &imiorofos_latin
+        canonical: imiórofos
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+
+    ypogeio: &ypogeio
+        canonical: υπόγειο
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: left
+        number_abs_value: true
+        number_min_abs_value: 1
+        standalone_probability: 0.985
+        numeric_probability: 0.01
+        ordinal_probability: 0.005
+    ypogeio_latin: &ypogeio_latin
+        canonical: ypógeio
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: left
+        number_abs_value: true
+        number_min_abs_value: 1
+        standalone_probability: 0.985
+        numeric_probability: 0.01
+        ordinal_probability: 0.005
+
+    aliases:
+        "<-1":
+            default: *ypogeio
+            probability: 0.9
+            alternatives:
+                - alternative: *ypogeio_latin
+                  probability: 0.1
+        "-1":
+            default: *ypogeio
+            probability: 0.9
+            alternatives:
+                - alternative: *ypogeio_latin
+                  probability: 0.1
+
+        half_floors:
+            default: *imiorofos
+            probability: 0.9
+            alternatives:
+                - alternative: *imiorofos_latin
+                  probability: 0.1
+
+        "0":
+            default: *isogelo
+            probability: 0.9
+            alternatives:
+                - alternative: *isogelo_latin
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *orofos
+        probability: 0.9
+        alternatives:
+            - alternative: *orofos_latin
+              probability: 0.1
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+entrances:
+    eisodos: &eisodos
+        canonical: είσοδος
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    eisodos_latin: &eisodos_latin
+        canonical: eísodos
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # είσοδος 1, etc.
+    alphanumeric:
+        default: *eisodos
+        probability: 0.99
+        alternatives:
+            - alternative: *eisodos_latin
+              probability: 0.01
+        numeric_probability: 0.1
+        alpha_probability: 0.9
+
+staircases:
+    skala: &skala
+        canonical: σκάλα
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    skala_latin: &skala_latin
+        canonical: skála
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    alphanumeric:
+        # For alphanumerics, skála A, skála 1, etc.
+        default: *skala
+        probability: 0.9
+        alternatives:
+            - alternative: *skala_latin
+              probability: 0.1
+        numeric_probability: 0.6 # e.g. skála 1
+        alpha_probability: 0.35 # e.g. skála A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+po_boxes:
+    tachydromiki_thyrida: &tachydromiki_thyrida
+        canonical: ταχυδρομική θυρίδα
+        abbreviated: τ.θ
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    tachydromiki_thyrida_latin: &tachydromiki_thyrida_latin
+        canonical: tachydromikí thyrída
+        abbreviated: t.th
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric:
+        default: *tachydromiki_thyrida
+        probability: 0.8
+        alternatives:
+            - alternative: *tachydromiki_thyrida_latin
+              probability: 0.2
+        numeric_probability: 0.9 # t.th 123
+        alpha_probability: 0.05 # t.th А
+        numeric_plus_alpha_probability: 0.04 # t.th 123А
+        alpha_plus_numeric_probability: 0.01 # t.th А123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    diamerisma: &diamerisma
+        canonical: διαμέρισμα
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        numeric_probability: 0.6
+        ordinal_probability: 0.4
+    diamerisma_latin: &diamerisma_latin
+        canonical: diamérisma
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        numeric_probability: 0.6
+        ordinal_probability: 0.4
+
+    domatio: &domatio
+        canonical: δωμάτιο
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        numeric_probability: 0.6
+        ordinal_probability: 0.4
+    domatio_latin: &domatio_latin
+        canonical: domátio
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        numeric_probability: 0.6
+        ordinal_probability: 0.4
+
+    grafeiou: &grafeiou
+        canonical: γραφείου
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        numeric_probability: 0.6
+        ordinal_probability: 0.4
+    grafeiou_latin: &grafeiou_latin
+        canonical: grafeíou
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        numeric_probability: 0.6
+        ordinal_probability: 0.4
+
+    alphanumeric: &unit_alphanumeric
+        default: *diamerisma
+        probability: 0.8
+        alternatives:
+            - alternative: *diamerisma_latin
+              probability: 0.1
+            - alternative: *domatio
+              probability: 0.09
+            - alternative: *domatio_latin
+              probability: 0.01
+
+        numeric_probability: 0.9 # e.g. diamérisma 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1А
+        alpha_plus_numeric_probability: 0.03 # e.g. AА1
+        alpha_probability: 0.04 # e.g. διαμέρισμα А
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.1
+
+    zone:
+        residential: *unit_alphanumeric
+        commercial:
+            default: *grafeiou
+            probability: 0.9
+            alternatives:
+                - alternative: *grafeiou_latin
+                  probability: 0.1
+        university:
+            default: *domatio
+            probability: 0.9
+            alternatives:
+                - alternative: *domatio_latin
+                  probability: 0.1

resources/addresses/et.yaml

@@ -0,0 +1,470 @@
+# et.yaml
+# -------
+# Estonian language specification.
+
+components:
+    level:
+        null_probability: 0.97
+        alphanumeric_probability: 0.02
+        standalone_probability: 0.01
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+    combinations:
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "-"
+                  probability: 0.95
+                - separator: " - "
+                  probability: 0.05
+            probability: 0.7
+
+
+numbers:
+    default: &number
+        canonical: number
+        abbreviated: nbr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+house_numbers:
+    alphanumeric:
+        default: *number
+
+    alphanumeric_phrase_probability: 0.0001
+
+
+and:
+    default: &ja
+        canonical: ja
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+cross_streets:
+    and: *ja
+    corner_of: &nurgas
+        canonical: nurgas
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    at_the_corner_of: &nurgal
+        canonical: nurgal
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *ja
+        probability: 0.7
+        alternatives:
+            - alternative: *nurgas
+              probability: 0.15
+            - alternative: *nurgal
+              probability: 0.15
+
+    between:
+        canonical: vahel
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &korrusel
+        canonical: korrusel
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+            direction_probability: 0.9
+        ordinal:
+            direction: right
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    parter: &parter
+        canonical: parter
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    kelder: &kelder
+        canonical: kelder
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        standalone_probability: 1.0
+    keldris: &keldris
+        canonical: keldris
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        # e.g. 1 keldris
+        numeric:
+            direction: right
+            direction_probability: 0.8
+        # e.g. k1
+        numeric_affix:
+            affix: k
+            direction: left
+        # e.g. 1. keldris
+        ordinal:
+            direction: right
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.05
+        numeric_affix_probability: 0.9
+        ordinal_probability: 0.05
+    aliases:
+        "<-1":
+            default: *kelder
+            probability: 0.85
+            alternatives:
+                - alternative: *keldris
+                  probability: 0.15            
+        "-1":
+            default: *kelder
+            probability: 0.85
+            alternatives:
+                - alternative: *keldris
+                  probability: 0.1
+                - alternative: *korrusel
+                  probability: 0.05
+        "1":
+            default: *parter
+            probability: 0.5
+            alternatives:
+                - alternative: *korrusel
+                  probability: 0.5
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *korrusel
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+
+
+
+categories:
+    near:
+        default:
+            canonical: lähedal
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+    nearby:
+        default:
+            canonical: lähedal
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: siin lähedal
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: siinkandis
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+
+    near_me:
+        default:
+            canonical: lähedal mulle
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+
+    # Probabilities of each phrase
+    near_probability: 0.7
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+
+directions:
+    right: &paremal
+        canonical: paremal
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: p
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    paramale: &paremale
+        canonical: paremale
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: p
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    left: &vasakul
+        canonical: vasakul
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    vasakule: &vasakule
+        canonical: vasakule
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    alternatives:
+        - alternative: *paremal
+          probability: 0.25
+        - alternative: *paremale
+          probability: 0.25
+        - alternative: *vasakul
+          probability: 0.25
+        - alternative: *vasakule
+          probability: 0.25
+
+cardinal_directions:
+    east: &ida
+        canonical: ida
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    west: &laas
+        canonical: lääs
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    north: &pohi
+        canonical: põhi
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+
+    south: &louna
+        canonical: lõuna
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+
+    alternatives:
+        - alternative: *pohi
+          probability: 0.25
+        - alternative: *ida
+          probability: 0.25
+        - alternative: *louna
+          probability: 0.25
+        - alternative: *laas
+          probability: 0.25
+
+
+entrances:
+    sissepaas: &sissepaas
+        canonical: sissepääs
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Eingang 1, Eingang A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *sissepaas
+        numeric_probability: 0.1 # e.g. Eingang 1
+        alpha_probability: 0.85 # e.g. Eingang A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    trepikoda: &trepikoda
+        canonical: trepikoda
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *trepikoda
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *pohi
+                    - alternative: *louna
+                    - alternative: *ida
+                    - alternative: *laas
+
+po_boxes:
+    postboks: &abonementpostkast
+        canonical: abonementpostkast
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # abonementpostkast #1234
+    kast: &kast
+        canonical: kast
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # Kast #1234
+    alphanumeric:
+        sample: false
+        default: *abonementpostkast
+        probability: 0.9
+        alternatives:
+            - alternative: *kast
+              probability: 0.1
+        numeric_probability: 0.9 # 123
+        alpha_probability: 0.05 # A
+        numeric_plus_alpha_probability: 0.04 # 123G
+        alpha_plus_numeric_probability: 0.01 # A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    korter: &korter
+        canonical: korter
+        abbreviated: k
+        sample: true
+        canonical_probability: 0.8
+        abbreviated_probability: 0.1
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            null_phrase_probability: 0.3
+            # Lejlighed nummer 4
+            add_number_phrase: true
+            add_number_phrase_probability: 0.05
+    ruumi: &ruumi
+        canonical: ruumi
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *korter
+        probability: 0.9
+        alternatives:
+            - alternative: *ruumi
+              probability: 0.1
+        numeric_probability: 1.0 # e.g. korter 1
+
+        # Separate random probability for adding directions like 2P, 2V, etc.
+        add_direction: true
+        add_direction_probability: 0.005
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+        # Add direction only e.g. Korter vasakule
+        add_direction_standalone: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.05

resources/addresses/eu.yaml

@@ -0,0 +1,375 @@
+# eu.yaml
+# -------
+# Basque language specification
+
+components:
+    level:
+        # If no floor number is specified
+        null_probability: 0.8
+        alphanumeric_probability: 0.2
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        # If no unit number is specified
+        null_probability: 0.4
+        alphanumeric_probability: 0.6
+
+    combinations:
+        -
+            components:
+                - level
+                - unit
+            label: unit
+            separators:
+                - separator: "-"
+                  probability: 0.85
+                - separator: "/"
+                  probability: 0.15
+            probability: 0.7
+
+
+and:
+    default: &eta
+        canonical: eta
+        abbreviated: "&"
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability:  0.4
+        sample_probability: 0.1
+
+house_numbers:
+    # zenbakirik gabe (zk.g) addresses
+    no_number: 
+        default:
+            canonical: zenbakirik gabe
+            abbreviated: zk.g
+            sample: true
+            canonical_probability: 0.1
+            abbreviated_probability: 0.6
+            sample_probability: 0.3
+
+    no_number_probability: 0.1 # With this probability, use sense número if no house_number is specified
+
+levels:
+    floor: &solairua
+        canonical: solairua
+        abbreviated: sol
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.3
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        # e.g. 2. solairua
+        ordinal:
+            direction: right
+        numeric_probability: 0.25
+        ordinal_probability: 0.75
+    # Ground floor
+    beheko_solairua: &beheko_solairua
+        canonical: beheko solairua
+        abbreviated: beheko sol
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.3
+        sample_probability: 0.1
+    behe_solairua: &behe_solairua
+        canonical: behe-solairua
+        abbreviated: behe-sol
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.3
+        sample_probability: 0.1
+    aliases:
+        "0":
+            default: *beheko_solairua
+            probability: 0.5
+            alternatives:
+                - alternative: *behe_solairua
+                  probability: 0.4
+                - alternative: *solairua
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *solairua
+        numeric_probability: 0.99
+        alpha_probability: 0.01
+
+blocks:
+    alphanumeric:
+        default:
+            canonical: blokea
+            abbreviated: bl
+            sample: true
+            canonical_probability: 0.6
+            abbreviated_probability: 0.2
+            sample_probability: 0.2
+            numeric:
+                direction: left
+            ordinal:
+                direction: right
+            numeric_probability: 0.2
+            ordinal_probability: 0.8
+
+categories:
+    near:
+        default:
+            canonical: gertu
+
+    nearby:
+        default:
+            canonical: gertuko
+        probability: 0.7
+        alternatives:
+            - alternative:
+                  canonical: hemen gertu
+              probability: 0.2
+            - alternative:
+                  canonical: hemen
+              probability: 0.1
+    near_me:
+        default:
+            canonical: me gertu
+
+    # Probabilities of each phrase
+    near_probability: 0.7
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+
+cross_streets:
+    and: *eta
+    txoko: &txoko
+        canonical: txoko
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+
+    intersection:
+        default: *eta
+        probability: 0.8
+        alternatives:
+            - alternative: *txoko
+              probability: 0.2
+
+    between:
+        canonical: arteko
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probabililty: 0.5
+
+
+po_boxes:
+    posta_kutxa: &posta_kutxa
+        canonical: posta-kutxa
+        abbreviated: p.-ku
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.4
+        sample_probability: 0.4
+        numeric:
+            direction: left
+        numeric_probability: 1.0
+    alphanumeric:
+        sample: false
+        default: *posta_kutxa
+        numeric_probability: 0.9 # P.-Ku  123
+        alpha_probability: 0.05 # P.-Ku  A
+        numeric_plus_alpha_probability: 0.04 # P.-Ku  123G
+        alpha_plus_numeric_probability: 0.01 # P.-Ku  A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+
+postcodes:
+    alphanumeric:
+        default:
+            canonical: posta-kodea
+            abbreviated: p.-k
+            sample: true
+            canonical_probability: 0.01
+            abbreviated_probability: 0.9
+            sample_probability: 0.09
+
+            numeric:
+                direction: left
+
+            numeric_affix:
+                affix: p.-k.
+                direction: left
+            # null_probability means the chance of doing nothing e.g. just the postal code
+            null_probability: 0.7
+            numeric_probability: 0.18
+            numeric_affix_probability: 0.12
+            strict_numeric: true
+
+directions:
+    right: &eskuina
+        canonical: eskuina
+        abbreviated: esk
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: esk.
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.9
+        numeric_affix_probability: 0.1
+    left: &ezkerkada
+        canonical: ezkerkada
+        abbreviated: ezk
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: ezk.
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.9
+        numeric_affix_probability: 0.1
+    ezkerreko: &ezkerreko
+        canonical: ezkerreko
+        abbreviated: ezk.-ko
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.5
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    alternatives:
+        - alternative: *eskuina
+          probability: 0.5
+        - alternative: *ezkerkada
+          probability: 0.5
+
+
+entrances:
+    sarrera: &sarrera
+        canonical: sarrera
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Sarrera 1, Sarrera A, etc.
+    alphanumeric:
+        default: *sarrera
+        numeric_probability: 0.1 # e.g. Sarrera 1
+        alpha_probability: 0.85 # e.g. Sarrera A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            modifier:
+                alternatives:
+                    - alternative: *eskuina
+                    - alternative: *ezkerreko
+
+staircases:
+    eskailera: &eskailera
+        canonical: eskailera
+        abbreviated: eskra
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    alphanumeric:
+        # For alphanumerics, Eskra A, Eskra 1, etc.
+        default: *eskailera
+        numeric_probability: 0.6 # e.g. Eskra 1
+        alpha_probability: 0.35 # e.g. Eskra A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left # e.g. Ezk.-ko Eskra
+            modifier:
+                alternatives:
+                    - alternative: *eskuina
+                    - alternative: *ezkerreko
+
+units:
+    flat: &apartamentu
+        canonical: apartamentu
+        abbreviated: aptu
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            # If it's just puerta B, many times it's just e.g. 3o B for "tercero piso puerta B"
+            null_phrase_probability: 0.15
+        ordinal:
+            direction: right
+        numeric_probability: 0.6
+        ordinal_probability: 0.4
+
+    alphanumeric: &unit_alphanumeric
+        default: *apartamentu
+
+        # Separate random probability for adding directions like 2. Ezk, 2 Esk, etc.
+        add_direction: true
+        add_direction_probability: 0.1
+        add_direction_numeric: true # Only for numbers
+        add_direction_standalone: true # A unit can be as simple as "D"
+
+        numeric_probability: 0.7 # e.g. 1
+        numeric_plus_alpha_probability: 0.01 # e.g. 1A
+        alpha_plus_numeric_probability: 0.01 # e.g. A1
+        alpha_probability: 0.28 # e.g. A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1

resources/addresses/fi.yaml

@@ -0,0 +1,403 @@
+# fi.yaml
+# -------
+# Finnish language specification.
+
+components:
+    level:
+        null_probability: 0.97
+        alphanumeric_probability: 0.02
+        standalone_probability: 0.01
+
+    staircase:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+    combinations:
+        -
+            components:
+                - staircase
+                - unit
+            label: unit
+            separators:
+                - separator: " "
+                  probability: 0.8
+                - separator: "-"
+                  probability: 0.1
+                - separator: "/"
+                  probability: 0.05
+                - separator: " - "
+                  probability: 0.05
+            probability: 0.85
+
+numbers:
+    default: &numero
+        canonical: numero
+        abbreviated: nro
+        sample: true
+        # Probabilities
+        canonical_probability: 0.1
+        abbreviated_probability: 0.5
+        sample_probability: 0.4
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.7
+        numeric_affix_probability: 0.3
+
+house_numbers:
+    alphanumeric:
+        default: *numero
+
+    alphanumeric_phrase_probability: 0.0001
+
+
+and:
+    default: &ja
+        canonical: ja
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+cross_streets:
+    and: *ja
+    corner_of: &kulmassa
+        canonical: kulmassa
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *ja
+        probability: 0.7
+        alternatives:
+            - alternative: *kulmassa
+              probability: 0.3
+
+    between:
+        canonical: välillä
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &kerros
+        canonical: kerros
+        abbreviated: krs
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.2
+        sample_probability: 0.2
+        numeric:
+            direction: right
+            direction_probability: 0.9
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *kerros
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+
+
+
+categories:
+    near:
+        default:
+            canonical: lähellä
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+    nearby:
+        default:
+            canonical: lähistöllä
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: lähellä
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: tässä lähellä
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: täällä
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+
+    near_me:
+        default:
+            canonical: lähellä minua
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+
+    # Probabilities of each phrase
+    near_probability: 0.7
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+
+directions:
+    right: &oikea
+        canonical: oikea
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: o
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    oikealla: &oikealla
+        canonical: oikealla
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: o
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    left: &vasen
+        canonical: vasen
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    vasemmalla: &vasemmalla
+        canonical: vasemmalla
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    alternatives:
+        - alternative: *oikea
+          probability: 0.25
+        - alternative: *oikealla
+          probability: 0.25
+        - alternative: *vasen
+          probability: 0.25
+        - alternative: *vasemmalla
+          probability: 0.25
+
+cardinal_directions:
+    east: &itaan
+        canonical: itään
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    west: &lansi
+        canonical: länsi
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    north: &pohja
+        canonical: pohja
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+
+    south: &etela
+        canonical: etelä
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+
+    alternatives:
+        - alternative: *pohja
+          probability: 0.25
+        - alternative: *itaan
+          probability: 0.25
+        - alternative: *etela
+          probability: 0.25
+        - alternative: *lansi
+          probability: 0.25
+
+
+entrances:
+    sissepaas: &sisaankaynti
+        canonical: sisäänkäynti
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Portaikko 1, Portaikko A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *sisaankaynti
+        numeric_probability: 0.1 # e.g. Portaikko 1
+        alpha_probability: 0.85 # e.g. Portaikko A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    portaikko: &portaikko
+        canonical: portaikko
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *portaikko
+        alpha_probability: 1.0
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *pohja
+                    - alternative: *etela
+                    - alternative: *itaan
+                    - alternative: *lansi
+
+po_boxes:
+    postilokero: &postilokero
+        canonical: postilokero
+        abbreviated: pl
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.6
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # PL #1234
+    alphanumeric:
+        sample: false
+        default: *postilokero
+        numeric_probability: 0.9 # 123
+        alpha_probability: 0.05 # A
+        numeric_plus_alpha_probability: 0.04 # 123G
+        alpha_plus_numeric_probability: 0.01 # A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    asunto: &asunto
+        canonical: asunto
+        abbreviated: as
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.7
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            null_phrase_probability: 0.3
+            # as nro 4
+            add_number_phrase: true
+            add_number_phrase_probability: 0.05
+    ruumi: &huone
+        canonical: huone
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *asunto
+        probability: 0.9
+        alternatives:
+            - alternative: *huone
+              probability: 0.1
+        numeric_probability: 1.0 # e.g. as 1
+
+        # Separate random probability for adding directions like 2O, 2V, etc.
+        add_direction: true
+        add_direction_probability: 0.005
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+        # Add direction only e.g. asunto 
+        add_direction_standalone: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.05

resources/addresses/fr.yaml

@@ -0,0 +1,951 @@
+# Note: default config is for France. Canadian, Swiss, Belgian, and other
+# conventions go in country overrides
+
+components:
+    level:
+        # If no floor number is specified
+        null_probability: 0.8
+        alphanumeric_probability: 0.2
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        # If no unit number is specified
+        null_probability: 0.8
+        alphanumeric_probability: 0.2
+
+    combinations:
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: /
+                  probability: 0.8
+                - separator: "-"
+                  probability: 0.1
+                - separator: " - "
+                  probability: 0.1
+            probability: 0.005
+
+numbers:
+    default: &numero
+        canonical: numéro
+        abbreviated: "nº"
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.5
+        sample_probability: 0.3
+        sample_exclude:
+            - "#" # Used in numeric affix. Needs to be quoted, otherwise it's a comment
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+        # Probabilities for numbers
+        numeric_probability: 0.7
+        numeric_affix_probability: 0.3
+
+and:
+    default: &and
+        canonical: et
+        abbreviated: "&"
+        canonical_probability: 0.7
+        abbreviated_probability: 0.25
+        sample: true
+        sample_probability: 0.05
+
+house_numbers:
+    # sans numéro (s/n) addresses
+    no_number: 
+        canonical: sans numéro
+        abbreviated: s/n
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.7
+        sample_probability: 0.2
+
+    alphanumeric:
+        default: *numero
+
+    alphanumeric_phrase_probability: 0.01
+    no_number_probability: 0.05 # With this probability, use sin número if no house_number is specified
+
+levels:
+    floor: &etage
+        canonical: étage
+        abbreviated: ét
+        sample: true
+        canonical_probability: 0.7
+        abbreviated_probability: 0.1
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.05
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.75
+        ordinal_probability: 0.25
+    niveau: &niveau
+        canonical: niveau
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.05
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.75
+        ordinal_probability: 0.25
+    bel_etage: &bel_etage
+        canonical: bel étage
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    etage_noble: &etage_noble
+        canonical: étage noble
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    dernier_etage: &dernier_etage
+        canonical: dernier étage
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    basement: &sous_sol
+        canonical: sous-sol
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        number_abs_value: true
+        number_min_abs_value: 1
+        standalone_probability: 0.99
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+    sub_basement: &soubassement
+        canonical: soubassement
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        number_abs_value: true
+        number_min_abs_value: 2
+        number_subtract_abs_value: 1
+        standalone_probability: 0.99
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+    mezzanine: &entresol
+        canonical: entresol
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    # Ground floor
+    rez_de_chaussee: &rez_de_chaussee
+        canonical: rez-de-chaussée
+        abbreviated: rdc
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.3
+        sample_probability: 0.3
+    rez_de_chaussee_bas: &rez_de_chaussee_bas
+        canonical: rez-de-chaussée bas
+        abbreviated: rcb
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+    rez_de_chaussee_haut: &rez_de_chaussee_haut
+        canonical: rez-de-chaussée haut
+        abbreviated: rch
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+    parterre: &parterre
+        canonical: parterre
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    rez_de_jardin: &rez_de_jardin
+        canonical: rez-de-jardin
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    aliases:
+        "<-1":
+            default: *sous_sol
+            probability: 0.6
+            alternatives:
+                - alternative: *soubassement
+                  probability: 0.3995
+                - alternative: *etage
+                  probability: 0.0005
+        "-1":
+            default: *sous_sol
+            probability: 0.9995
+            alternatives:
+                - alternative: *etage
+                  probability: 0.0005
+        half_floors:
+            default: *entresol
+        "0":
+            default: *rez_de_chaussee
+            probability: 0.74
+            alternatives:
+                - alternative: *rez_de_jardin
+                  probability: 0.01
+                - alternative: *rez_de_chaussee_bas
+                  probability: 0.1
+                - alternative: *rez_de_chaussee_haut
+                  probability: 0.1
+                - alternative: *etage
+                  probability: 0.05
+        "1":
+            default: *etage
+            probability: 0.8
+            alternatives:
+                - alternative: *bel_etage
+                  probability: 0.1
+                - alternative: *etage_noble
+                  probability: 0.1
+        top:
+            default: *etage
+            probability: 0.9
+            alternatives:
+                - alternative: *dernier_etage
+                  probability: 0.1
+
+    alphanumeric:
+        default: *etage
+        probability: 0.95
+        alternatives:
+            - alternative: *niveau
+              probability: 0.05
+        numeric_probability: 0.99
+        alpha_probability: 0.01
+
+    numbering_starts_at: 0
+
+
+cross_streets:
+    # 26th & 6th Avenue
+    and: *and
+    # 26th @ Broadway
+    a: &a
+        canonical: à
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    au: &au
+        canonical: au
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    corner_of: &langle_de
+        canonical: l'angle de
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    at_the_corner_of: &a_langle_de
+        canonical: à l'angle de
+
+    intersection:
+        default: *and
+        probability: 0.7
+        alternatives:
+            - alternative: *a
+              probability: 0.025
+            - alternative: *au
+              probability: 0.025
+            - alternative: *langle_de
+              probability: 0.15
+            - alternative: *a_langle_de
+              probability: 0.1
+
+    # 26th betw 5th Ave and 6th Ave
+    between:
+        canonical: entre
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5 # Probability of using parentheses e.g. (between 5th and 6th)
+
+directions:
+    right: &droit
+        canonical: droit
+        abbreviated: dr
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.2
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: d
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.7
+        numeric_affix_probability: 0.3
+    left: &gauche
+        canonical: gauche
+        abbreviated: g
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: g
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+    rear: &arriere
+        canonical: arrière
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    front: &avant
+        canonical: avant
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *droit
+          probability: 0.49
+        - alternative: *gauche
+          probability: 0.49
+        - alternative: *arriere
+          probability: 0.01
+        - alternative: *avant
+          probability: 0.01
+
+    anteroposterior:
+        alternatives:
+            - alternative: *avant
+              probability: 0.5
+            - alternative: *arriere
+              probability: 0.5
+
+    lateral:
+        alternatives:
+            - alternative: *droit
+              probability: 0.5
+            - alternative: *gauche
+              probability: 0.5
+
+cardinal_directions:
+    east: &est
+        canonical: est
+        abbreviated: e
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6 
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: e
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &ouest
+        canonical: ouest
+        abbreviated: o
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: o
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &nord
+        canonical: nord
+        abbreviated: n
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &sud
+        canonical: sud
+        abbreviated: s
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *nord
+          probability: 0.25
+        - alternative: *est
+          probability: 0.25
+        - alternative: *sud
+          probability: 0.25
+        - alternative: *ouest
+          probability: 0.25
+
+entrances:
+    entrance: &entrance
+        canonical: entrance
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Entrance 1, Entrance A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *entrance
+        numeric_probability: 0.1 # e.g. Entrance 1
+        alpha_probability: 0.85 # e.g. Entrnace A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            modifier:
+                direction: right # e.g. Entrance Nord
+                direction_probability: 0.95
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *sud
+                    - alternative: *est
+                    - alternative: *ouest
+                    - alternative: *droit
+                    - alternative: *gauche
+                    - alternative: *arriere
+                    - alternative: *avant
+
+staircases:
+    escalier: &escalier
+        canonical: escalier
+        abbreviated: esc
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    alphanumeric:
+        # For alphanumerics, Escalier A, Esc 1, etc.
+        default: *escalier
+        numeric_probability: 0.6 # e.g. Escalier 1
+        alpha_probability: 0.35 # e.g. Escalier A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: right # e.g. Escalier Izq
+            direction_probability: 0.9
+            modifier:
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *sud
+                    - alternative: *est
+                    - alternative: *ouest
+                    - alternative: *droit
+                    - alternative: *gauche
+                    - alternative: *arriere
+                    - alternative: *avant
+
+
+po_boxes:
+    boite_postal: &boite_postal
+        canonical: boîte postale
+        abbreviated: bp
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # BP No 1234
+        numeric_probability: 1.0
+    case_postal: &case_postal
+        canonical: case postale
+        abbreviated: cp
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # CP No 1234
+        numeric_probability: 1.0
+    alphanumeric:
+        sample: false
+        default: *boite_postal
+        numeric_probability: 0.9 # BP 123
+        alpha_probability: 0.05 # BP A
+        numeric_plus_alpha_probability: 0.04 # BP 123G
+        alpha_plus_numeric_probability: 0.01 # BP A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+
+units:
+    flat: &appartement
+        canonical: appartement
+        abbreviated: app
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.5
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    unit: &unite
+        canonical: unité
+        abbreviated: u
+        sample: true
+        canonical_probability: 0.8
+        abbreviated_probability: 0.1
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    suite: &suite
+        canonical: suite
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.4
+    office: &bureau
+        canonical: bureau
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.3
+    door: &porte
+        canonical: porte
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    room: &salle
+        canonical: salle
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    chambre: &chambre
+        canonical: chambre
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    boite: &boite
+        canonical: boîte
+        abbreviated: bte
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    lot: &lotissement
+        canonical: lotissement
+        abbreviated: lot
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.5
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    parcelle: &parcelle
+        canonical: parcelle
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+
+    allotments:
+        lot:
+            default: *lotissement
+            numeric_probability: 0.8
+            alphanumeric_probability: 0.1
+            alpha_probability: 0.1
+        parcel:
+            default: *parcelle
+            numeric_probability: 0.3
+            alphanumeric_probability: 0.3
+            alpha_probability: 0.4
+        lot_probability: 0.9
+        parcel_probability: 0.06
+        lot_plus_parcel_probability: 0.02
+        parcel_plus_lot_probability: 0.02
+
+    alphanumeric: &unit_alphanumeric
+        default: *appartement
+        probability: 0.85
+        alternatives:
+              # e.g. just plain #3 or No. 4
+            - alternative: *numero
+              probability: 0.05
+            - alternative: *porte
+              probability: 0.095
+            - alternative: *chambre
+              probability: 0.005
+        numeric_probability: 0.9 # e.g. Appartement 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. Appartement A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # Separate random probability for adding directions like 2D, 2G, etc.
+        add_direction: true
+        add_direction_probability: 0.1
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+        # Add direction only e.g. Unité Gauche
+        add_direction_standalone: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.1
+
+    zones:
+        residential: *unit_alphanumeric
+        commercial:
+            default: *bureau
+            probability: 0.8
+            alternatives:
+                - alternative: *suite
+                  probability: 0.2
+
+            numeric_probability: 0.9 # e.g. Bureau 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Bureau 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Bureau A1
+            alpha_probability: 0.08 # e.g. Bureau A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+        industrial:
+            default: *lotissement
+            probability: 0.5
+            alternatives:
+                - alternative: *bureau
+                  probability: 0.3
+                - alternative: *unite
+                  probability: 0.19
+                - alternative: *parcelle
+                  probability: 0.01
+
+            numeric_probability: 0.9 # e.g. Lote 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Lote 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Lote A1
+            alpha_probability: 0.08 # e.g. Lote A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+        university:
+            default: *salle
+            probability: 0.9
+            alternatives:
+                - alternative: *porte
+                  probability: 0.1
+
+            numeric_probability: 0.9 # e.g. Salle 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Salle 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Salle A1
+            alpha_probability: 0.08 # e.g. Salle A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+categories:
+    near:
+        default:
+            canonical: près de
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: à coté de
+              probability: 0.05
+            - alternative:
+                canonical: proche de
+              probability: 0.05
+            - alternative:
+                canonical: proches de
+              probability: 0.05
+            - alternative:
+                canonical: a cote de
+              probability: 0.05
+            - alternative:
+                canonical: pres de
+              probability: 0.05
+            - alternative:
+                canonical: aux environs de
+              probability: 0.05
+            - alternative:
+                canonical: à proximité de
+              probability: 0.05
+            - alternative:
+                canonical: a proximite de
+              probability: 0.05
+    nearby:
+        default:
+            canonical: proche
+        probability: 0.4
+        alternatives:
+            - alternative:
+                canonical: à coté
+              probability: 0.05
+            - alternative:
+                canonical: a cote
+              probability: 0.05
+            - alternative:
+                canonical: près d'ici
+              probability: 0.05
+            - alternative:
+                canonical: près dici
+              probability: 0.05
+            - alternative:
+                canonical: pres d'ici
+              probability: 0.05
+            - alternative:
+                canonical: pres dici
+              probability: 0.05
+            - alternative:
+                canonical: près de là
+              probability: 0.05
+            - alternative:
+                canonical: pres de la
+              probability: 0.05
+            - alternative:
+                canonical: par ici
+              probability: 0.05
+            - alternative:
+                canonical: dans les alentours
+              probability: 0.05
+            - alternative:
+                canonical: à proximité de là
+              probability: 0.05
+            - alternative:
+                canonical: a proximite de la
+              probability: 0.05
+    near_me:
+        default:
+            canonical: proche de chez moi
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: près de moi
+              probability: 0.1
+            - alternative:
+                canonical: pres de moi
+              probability: 0.1
+            - alternative:
+                canonical: à proximité de moi
+              probability: 0.1
+            - alternative:
+                canonical: a proximite de moi
+              probability: 0.1
+    in:
+        default:
+            canonical: à
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: en
+              probability: 0.1
+            - alternative:
+                canonical: a
+              probability: 0.1
+            - alternative:
+                canonical: dans
+              probability: 0.1
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+countries:
+    # Belgium
+    be:
+        units:
+            alphanumeric:
+                default: *boite
+                probability: 0.75
+                alternatives:
+                    - alternative: *appartement
+                      probability: 0.1
+                      # e.g. just plain #3 or No. 4
+                    - alternative: *numero
+                      probability: 0.05
+                    - alternative: *porte
+                      probability: 0.095
+                    - alternative: *chambre
+                      probability: 0.005
+    # Canada
+    ca:
+        components:
+
+            unit:
+                null_probability: 0.6
+                alphanumeric_probability: 0.4
+            combinations:
+                -
+                    components:
+                        - unit
+                        - house_number
+                    label: house_number
+                    separators:
+                        - separator: /
+                          probability: 0.04
+                        - separator: "-"
+                          probability: 0.95
+                        - separator: " - "
+                          probability: 0.01
+                    probability: 0.1
+        levels:
+            numbering_starts_at: 1
+            aliases:
+                "1":
+                    # Have to do this because etage is numeric
+                    # and has keys like "numeric_probability" which 
+                    # we don't want to infect rez_de_chausee when doing
+                    # a recursive merge
+                    default: *etage
+                    probability: 0.1
+                    alternatives:
+                        - alternative: *rez_de_chaussee
+                          probability: 0.8
+                        - alternative: *bel_etage
+                          probability: 0.05
+                        - alternative: *etage_noble
+                          probability: 0.05
+
+        units:
+            alphanumeric:
+                # More common to use in in Canada, as in the US
+                use_floor_probability: 0.35
+
+        po_boxes:
+            alphanumeric:
+                default: *case_postal
+    # Switzerland
+    ch:
+        levels:
+            aliases:
+                "0":
+                    default: *parterre
+                    probability: 0.9
+                    alternatives:
+                        - alternative: *rez_de_chaussee
+                          probability: 0.05
+                        - alternative: *etage
+                          probability: 0.05
+        po_boxes:
+            alphanumeric:
+                default: *case_postal

resources/addresses/he.yaml

@@ -0,0 +1,269 @@
+# he.yaml
+# -------
+# Hebrew language specification
+
+
+alphabet: אבגדהוזחטיכךלמםנןסעפףצץקרשת
+alphabet_probability: 0.8
+
+components:
+    level:
+        null_probability: 0.95
+        alphanumeric_probability: 0.05
+
+    entrance:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    unit:
+        null_probability: 0.6
+        alphanumeric_probability: 0.4
+
+
+    combinations:
+        -
+            components:
+                - house_number
+                - entrance
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.7
+        -
+            components:
+                - house_number
+                - entrance
+            label: house_number
+            separators:
+                - separator: " "
+                  probability: 0.5
+                - separator: ""
+                  probability: 0.2
+                - separator: "/"
+                  probability: 0.1
+                - separator: "-"
+                  probability: 0.1
+                - separator: " - "
+                  probability: 0.1
+            probability: 0.7
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.1
+
+levels:
+    koma: &koma
+        canonical: קומה
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: left
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    koma_latin: &koma_latin
+        canonical: koma
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+        ordinal:
+            direction: left
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+
+    komat_karka: &komat_karka
+        canonical: קומת קרקע
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    komat_karka_latin: &komat_karka_latin
+        canonical: komát karká
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+
+    martef: &martef
+        canonical: מרתף
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: left
+        number_abs_value: true
+        number_min_abs_value: 1
+        standalone_probability: 0.985
+        numeric_probability: 0.01
+        ordinal_probability: 0.005
+    martef_latin: &martef_latin
+        canonical: martef
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: left
+        number_abs_value: true
+        number_min_abs_value: 1
+        standalone_probability: 0.985
+        numeric_probability: 0.01
+        ordinal_probability: 0.005
+
+    aliases:
+        "<-1":
+            default: *martef
+            probability: 0.9
+            alternatives:
+                - alternative: *martef_latin
+                  probability: 0.1
+        "-1":
+            default: *martef
+            probability: 0.9
+            alternatives:
+                - alternative: *martef_latin
+                  probability: 0.1
+        "0":
+            default: *komat_karka
+            probability: 0.9
+            alternatives:
+                - alternative: *komat_karka_latin
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *koma
+        probability: 0.9
+        alternatives:
+            - alternative: *koma_latin
+              probability: 0.1
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+entrances:
+    knisa: &knisa
+        canonical: כניסה
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    knisa_latin: &knisa_latin
+        canonical: knisa
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # вход 1, вход A, etc.
+    alphanumeric:
+        default: *knisa
+        probability: 0.99
+        alternatives:
+            - alternative: *knisa_latin
+              probability: 0.01
+        numeric_probability: 0.1
+        alpha_probability: 0.9
+
+po_boxes:
+    ta_doar: &ta_doar
+        canonical: תיבת דואר
+        abbreviated: ת.ד.
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    ta_doar_latin: &ta_doar_latin
+        canonical: abonementnyy pochtovyy yashchik
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric:
+        default: *ta_doar
+        probability: 0.8
+        alternatives:
+            - alternative: *ta_doar_latin
+              probability: 0.2
+        numeric_probability: 0.9 # ta doar 123
+        alpha_probability: 0.05 # ta doar А
+        numeric_plus_alpha_probability: 0.04 # ta doar 123А
+        alpha_plus_numeric_probability: 0.01 # ta doar А123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    dira: &dira
+        canonical: דירה
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    dira_latin: &dira_latin
+        canonical: dira
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &unit_alphanumeric
+        default: *dira
+        probability: 0.9
+        alternatives:
+            - alternative: *dira_latin
+              probability: 0.1
+
+        numeric_probability: 0.9 # e.g. dira 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1А
+        alpha_plus_numeric_probability: 0.03 # e.g. AА1
+        alpha_probability: 0.04 # e.g. dira А
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.1

resources/addresses/hr.yaml

@@ -0,0 +1,586 @@
+# hr.yaml
+# -------
+# Croatian language specification
+
+components:
+    level:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.7
+        alphanumeric_probability: 0.3
+
+    combinations:
+        -
+            components:
+                - house_number
+                - staircase
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.1
+        # For unit types like 2/34
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+
+
+numbers:
+    no_number: 
+        default:
+            canonical: bez broja
+            abbreviated: bb
+            sample: true
+            canonical_probability: 0.3
+            abbreviated_probability: 0.4
+            sample_probability: 0.3
+
+    default: &broj
+        canonical: broj
+        abbreviated: br
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.6
+        sample_probability: 0.1
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "br."
+            whitespace_probability: 0.6
+            direction: left
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+    alphanumeric_phrase_probability: 0.05
+    no_number_probability: 0.05
+
+
+and:
+    default: &i
+        canonical: i
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+
+cross_streets:
+    i: *i
+    at: &na
+        canonical: na
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    corner: &ugao
+        canonical: ugao
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    corner_of: &uglu
+        canonical: uglu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    na_uglu: &na_uglu
+        canonical: na uglu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *i
+        probability: 0.65
+        alternatives:
+            - alternative: *na
+              probability: 0.1
+            - alternative: *uglu
+              probability: 0.1
+            - alternative: *na_uglu
+              probability: 0.1
+            - alternative: *ugao
+              probability: 0.05
+
+    izmedu: &izmedu
+        canonical: između
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+    between:
+        default: *izmedu
+
+levels:
+    kat: &kat
+        canonical: kat
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    etaza: &etaza
+        canonical: etaža
+        abbreviated: et
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    prizemlje: &prizemlje
+        canonical: prizemlje
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    parter: &parter
+        canonical: parter
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    mezanino: &polukat
+        canonical: polukat
+        half_floors: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        sample: true
+        # e.g. polukat 2
+        numeric:
+            direction: left
+        # e.g. 2. entresuelo
+        ordinal:
+            direction: right
+        numeric_probability: 0.1
+        ordinal_probability: 0.2
+        standalone_probability: 0.6
+    podrum: &podrum
+        canonical: podrum
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        # e.g. подрум 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. 1. подрум
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+
+    aliases:
+        "<-1":
+            default: *podrum
+        "-1":
+            default: *podrum
+        # Special token for half-floors
+        half_floors:
+            default: *polukat
+        "0":
+            default: *prizemlje
+            probability: 0.5
+            alternatives:
+                - alternative: *parter
+                  probability: 0.4
+                - alternative: *kat
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *kat
+        probability: 0.95
+        alternatives:
+            - alternative: *etaza
+              probability: 0.05
+        numeric_probability: 0.69 # With this probability, pick an integer
+        roman_numeral_probability: 0.3 # Pick a Roman numeral for the actual value
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+
+categories:
+    near:
+        default:
+            canonical: u blizini
+    nearby:
+        default:
+            canonical: u blizini
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: u blizini ovdje
+              probability: 0.3
+            - alternative:
+                canonical: oko ovdje
+              probability: 0.1
+
+    near_me:
+        default:
+            canonical: u blizini mene
+
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: u
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+directions:
+    right: &desno
+        canonical: desno
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &lijevo
+        canonical: lijevo
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *desno
+          probability: 0.5
+        - alternative: *lijevo
+          probability: 0.5
+
+cardinal_directions:
+    east: &istok
+        canonical: istok
+        abbreviated: i
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: i
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &zapad
+        canonical: zapad
+        abbreviated: z
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: z
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &sjever
+        canonical: sjever
+        abbreviated: s
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &jug
+        canonical: jug
+        abbreviated: j
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: j
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *sjever
+          probability: 0.25
+        - alternative: *istok
+          probability: 0.23
+        - alternative: *jug
+          probability: 0.23
+        - alternative: *zapad
+          probability: 0.23
+
+entrances:
+    ulaz: &ulaz
+        canonical: ulaz
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Ulaz 1, Ulaz A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *ulaz
+        numeric_probability: 0.1 # e.g. Ulaz 1
+        alpha_probability: 0.85 # e.g. Ulaz A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+
+staircases:
+    stubiste: &stubiste
+        canonical: stubište
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+
+    alphanumeric: &staircase_alphanumeric
+        default: *stubiste
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: right
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *desno
+                      probability: 0.2
+                    - alternative: *lijevo
+                      probability: 0.2
+                    - alternative: *sjever
+                      probability: 0.15
+                    - alternative: *jug
+                      probability: 0.15
+                    - alternative: *istok
+                      probability: 0.15
+                    - alternative: *zapad
+                      probability: 0.15
+
+po_boxes:
+    postanski_pretinac: &postanski_pretinac
+        canonical: poštanski pretinac
+        abbreviated: p.p
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.4
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+
+    alphanumeric:
+        default: *postanski_pretinac
+        numeric_probability: 0.9 # pp 123
+        alpha_probability: 0.05 # p.p A
+        numeric_plus_alpha_probability: 0.04 # pp 123G
+        alpha_plus_numeric_probability: 0.01 # pp A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    stan: &stan
+        canonical: stan
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    apartman: &apartman
+        canonical: apartman
+        abbreviated: ap
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.2
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    soba: &soba
+        canonical: soba
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    ured: &ured
+        canonical: ured
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    alphanumeric: &unit_alphanumeric
+        default: *stan
+        probability: 0.6
+        alternatives:
+            - alternative: *apartman
+              probability: 0.3
+            - alternative: *soba
+              probability: 0.1
+        numeric_probability: 0.9 # e.g. stan. 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. stan A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.05
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *soba
+            probability: 0.6
+            alternatives:
+                - alternative: *ured
+                  probability: 0.4
+            numeric_probability: 0.95 # e.g. soba 1
+            numeric_plus_alpha_probability: 0.01 # e.g. soba 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. soba A1
+            alpha_probability: 0.03 # e.g. soba A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university:
+            default: *soba
+            numeric_probability: 0.95 # e.g. soba 1
+            numeric_plus_alpha_probability: 0.01 # e.g. soba 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. soba A1
+            alpha_probability: 0.03 # e.g. soba A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/hu.yaml

@@ -0,0 +1,439 @@
+# hu.yaml
+# -------
+# Hungarian language specification.
+
+components:
+    level:
+        null_probability: 0.75
+        alphanumeric_probability: 0.2
+        standalone_probability: 0.05
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+    combinations:
+        -
+            components:
+                - level
+                - unit
+            label: unit
+            separators:
+                - separator: "/"
+                  probability: 0.55
+                - separator: " "
+                  probability: 0.4
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.8
+
+
+numbers:
+    default: &szam
+        canonical: szám
+        sample: true
+        # Probabilities
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+and:
+    default: &es
+        canonical: és
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+    probability: 0.6
+    alternatives:
+        - alternative: &es_a
+            canonical: és a
+            canonical_probability: 0.9
+            sample: true
+            sample_probability: 0.1
+          probability: 0.2
+        - alternative: &es_az
+            canonical: és az
+            canonical_probability: 0.9
+            sample: true
+            sample_probability: 0.1
+          probability: 0.2
+
+cross_streets:
+    and: *es
+    corner_of: &sarkan
+        canonical: sarkán
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *es
+        probability: 0.6
+        alternatives:
+            - alternative: *es_a
+              probability: 0.1
+            - alternative: *es_az
+              probability: 0.1
+            - alternative: *sarkan
+              probability: 0.2
+
+    between:
+        canonical: között
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &emelet
+        canonical: emelet
+        abbreviated: em
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.85
+        sample_probability: 0.05
+        numeric:
+            direction: left
+            direction_probability: 0.9
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.2
+                roman_numeral_probability: 0.8
+        numeric_probability: 0.1
+        ordinal_probability: 0.9
+    foldszint: &foldszint
+        canonical: földszint
+        abbreviated: fszt
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.6
+        sample_probability: 0.2
+    felemelet: &felemelet
+        canonical: félemelet
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    magasfoldszint: &magasfoldszint
+        canonical: magasföldszint
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    pince: &pince
+        canonical: pince
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+    alagsor: &alagsor
+        canonical: alagsor
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+    felszuteren: &felszuteren
+        canonical: félszuterén
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+    szuteren: &szuteren
+        canonical: szuterén
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+    aliases:
+        "<-1":
+            default: *alagsor
+            probability: 0.6
+            alternatives:
+                - alternative: *pince
+                  probability: 0.3
+                - alternative: *szuteren
+                  probability: 0.1
+        "-1":
+            default: *alagsor
+            probability: 0.5
+            alternatives:
+                - alternative: *pince
+                  probability: 0.3
+                - alternative: *szuteren
+                  probability: 0.1
+                - alternative: *felszuteren
+                  probability: 0.1
+
+        "0":
+            default: *foldszint
+            probability: 0.9
+            alternatives:
+                - alternative: *emelet
+                  probability: 0.1
+
+        "1":
+            default: *emelet
+            probability: 0.9
+            alternatives:
+                - alternative: *felemelet
+                  probability: 0.1
+
+        "2":
+            default: *emelet
+            probability: 0.9
+            alternatives:
+                - alternative: *magasfoldszint
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *emelet
+        numeric_probability: 0.59 # With this probability, pick an integer
+        roman_numeral_probability: 0.4 # Pick a Roman numeral for the actual value
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+categories:
+    near:
+        default:
+            canonical: közelében
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+    nearby:
+        default:
+            canonical: közelben
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+    near_me:
+        default:
+            canonical: közelemben
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+    # Probabilities of each phrase
+    near_probability: 0.7
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+
+directions:
+    right: &jobb
+        canonical: jobb
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &bal
+        canonical: bal
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *jobb
+          probability: 0.5
+        - alternative: *bal
+          probability: 0.5
+
+cardinal_directions:
+    east: &kelet
+        canonical: kelet
+        abbreviated: k
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: k
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &nyugat
+        canonical: nyugat
+        abbreviated: n
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &eszak
+        canonical: észak
+        abbreviated: e
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: e
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &del
+        canonical: dél
+        abbreviated: d
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: d
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *eszak
+          probability: 0.25
+        - alternative: *kelet
+          probability: 0.25
+        - alternative: *del
+          probability: 0.25
+        - alternative: *nyugat
+          probability: 0.25
+
+
+po_boxes:
+    postafiok: &postafiok
+        canonical: postafiók
+        abbreviated: pf
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.7
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    alphanumeric:
+        default: *postafiok
+        numeric_probability: 0.9 # Pf 123
+        alpha_probability: 0.05 # Pf A
+        numeric_plus_alpha_probability: 0.04 # Pf 123G
+        alpha_plus_numeric_probability: 0.01 # Pf A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    lakas: &lakas
+        canonical: lakás
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        numeric_probability: 0.3
+        ordinal_probability: 0.7
+    iroda: &iroda
+        canonical: iroda
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    szoba: &szoba
+        canonical: szoba
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &unit_alphanumeric
+        default: *lakas
+        probability: 0.9
+        alternatives:
+            - alternative: *szoba
+              probability: 0.1
+        numeric_probability: 0.95 # e.g. m. 1
+        numeric_plus_alpha_probability: 0.005 # e.g. 1A
+        alpha_plus_numeric_probability: 0.005 # e.g. A1
+        alpha_probability: 0.04 # e.g. m. A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.2
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *iroda
+            numeric_probability: 0.95 # e.g. pokój 1
+            numeric_plus_alpha_probability: 0.01 # e.g. pokój 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. pokój A1
+            alpha_probability: 0.03 # e.g. pokój A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university: *commercial_unit_types

resources/addresses/is.yaml

@@ -0,0 +1,459 @@
+# is.yaml
+# -------
+# Icelandic language specification.
+
+components:
+    level:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+    combinations:
+        -
+            components:
+                - level
+                - unit
+            label: unit
+            separators:
+                - separator: "-"
+                  probability: 0.9
+                - separator: " - "
+                  probability: 0.1
+            probability: 0.005
+        -
+            components:
+                - entrance
+                - unit
+            label: unit
+            separators:
+                - separator: "-"
+                  probability: 0.9
+                - separator: " - "
+                  probability: 0.1
+            probability: 0.001
+
+
+numbers:
+    default: &numer
+        canonical: númer
+        abbreviated: nr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+house_numbers:
+    alphanumeric:
+        default: *numer
+
+    alphanumeric_phrase_probability: 0.0001
+
+
+and:
+    default: &og
+        canonical: og
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+cross_streets:
+    and: *og
+    corner_of: &horn_of
+        canonical: horn af
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    at_the_corner_of: &a_horinu_a
+        canonical: á horninu á
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *og
+        probability: 0.7
+        alternatives:
+            - alternative: *horn_of
+              probability: 0.15
+            - alternative: *a_horinu_a
+              probability: 0.15
+
+    between:
+        canonical: milli
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &haeo
+        canonical: hæð
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: right
+            direction_probability: 0.9
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+
+    jarohaeo: &jarohaeo
+        canonical: jarðhæð
+        sample: true
+        canonical_probability: 0.3
+        sample_probability: 0.7
+    kjallara: &kjallara
+        canonical: kjallara
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        # e.g. 1 kjallara
+        numeric:
+            direction: right
+            direction_probability: 0.8
+        # e.g. k1
+        numeric_affix:
+            affix: k
+            direction: left
+        # e.g. 1. kjallara
+        ordinal:
+            direction: right
+        standalone_probability: 0.985
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.005
+        ordinal_probability: 0.005
+    aliases:
+        "<-1":
+            default: *kjallara
+        "-1":
+            default: *kjallara
+        "0":
+            default: *jarohaeo
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *haeo
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+categories:
+    near:
+        default:
+            canonical: nálægt
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+    nearby:
+        default:
+            canonical: nálægt
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: nálægt hér
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: hérna
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: hér
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+    near_me:
+        default:
+            canonical: nálægt mér
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: í
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+
+
+directions:
+    right: &til_haegri
+        canonical: til hægri
+        abbreviated: t.h
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.5
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: t.h
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    left: &til_vinstri
+        canonical: til vinstri
+        abbreviated: t.v
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.6
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: t.v
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    alternatives:
+        - alternative: *til_haegri
+          probability: 0.5
+        - alternative: *til_vinstri
+          probability: 0.5
+
+cardinal_directions:
+    east: &austur
+        canonical: austur
+        abbreviated: a
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: a
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &vestur
+        canonical: vestur
+        abbreviated: v
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &norour
+        canonical: norður
+        abbreviated: n
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &suour
+        canonical: suður
+        abbreviated: s
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *norour
+          probability: 0.25
+        - alternative: *austur
+          probability: 0.25
+        - alternative: *suour
+          probability: 0.25
+        - alternative: *vestur
+          probability: 0.25
+
+
+entrances:
+    inngangur: &inngangur
+        canonical: inngangur
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Inngangur 1, Inngangur A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *inngangur
+        numeric_probability: 0.1 # e.g. Inngangur 1
+        alpha_probability: 0.85 # e.g. Inngangur A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    stiege: &stigi
+        canonical: stigi
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *stigi
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *norour
+                    - alternative: *suour
+                    - alternative: *austur
+                    - alternative: *vestur
+
+po_boxes:
+    postholf: &postholf
+        canonical: pósthólf
+        abbreviated: ph
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.2
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # Ph Nr 1234
+    alphanumeric:
+        sample: false
+        default: *postholf
+        numeric_probability: 0.9 # Ph 123
+        alpha_probability: 0.05 # Ph A
+        numeric_plus_alpha_probability: 0.04 # Ph 123G
+        alpha_plus_numeric_probability: 0.01 # Ph A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    ibuo: &ibuo
+        canonical: íbúð
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            null_phrase_probability: 0.5
+        # íbúð nummer 4
+        add_number_phrase: true
+        add_number_phrase_probability: 0.05
+
+    alphanumeric: &unit_alphanumeric
+        default: *ibuo
+        numeric_probability: 0.9 # e.g. íbúð 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. íbúð A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # Separate random probability for adding directions like 2R, 2L, etc.
+        add_direction: true
+        add_direction_probability: 0.1
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.1

resources/addresses/it.yaml

@@ -0,0 +1,673 @@
+# it.yaml
+# -------
+# Italian language specification
+
+components:
+    level:
+        # If no floor number is specified
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        # If no unit number is specified
+        null_probability: 0.8
+        alphanumeric_probability: 0.2
+
+    combinations:
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: /
+                  probability: 1.0
+            probability: 0.5
+
+numbers:
+    default: &numero
+        canonical: numero
+        abbreviated: "nº"
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.3
+        sample_probability: 0.5
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "n."
+            direction: left
+        # Probabilities for numbers
+        numeric_probability: 0.7
+        numeric_affix_probability: 0.3
+
+and:
+    default: &e
+        canonical: e
+        abbreviated: "&"
+        canonical_probability: 0.7
+        abbreviated_probability: 0.25
+        sample: true
+        sample_probability: 0.05
+
+house_numbers:
+    # sans numéro (s/n) addresses
+    no_number: 
+        canonical: senza numero civico
+        abbreviated: snc
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.7
+        sample_probability: 0.2
+
+    alphanumeric:
+        default: *numero
+
+    alphanumeric_phrase_probability: 0.01
+    no_number_probability: 0.05 # With this probability, use sin número if no house_number is specified
+
+levels:
+    floor: &piano
+        canonical: piano
+        abbreviated: pº
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.15
+        sample_probability: 0.25
+        numeric:
+            direction: left
+            direction_probability: 0.95
+            add_number_phrase: true
+            add_number_phrase_probability: 0.05
+            digits:
+                ascii_probability: 0.9
+                roman_numeral_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.5
+                spellout_probability: 0.2
+                roman_numeral_probability: 0.3
+        numeric_probability: 0.55
+        ordinal_probability: 0.45
+    livello: &livello
+        canonical: livello
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.05
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        numeric_probability: 0.75
+        ordinal_probability: 0.25
+    piano_nobile: &piano_nobile
+        canonical: piano nobile
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    piano_terra: &piano_terra
+        canonical: piano terra
+        abbreviated: p.t
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.25
+        sample_probability: 0.25
+    basement: &seminterrato
+        canonical: seminterrato
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+        ordinal:
+            direction: right
+        number_abs_value: true
+        number_min_abs_value: 1
+        standalone_probability: 0.99
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+    aliases:
+        "<-1":
+            default: *seminterrato
+            probability: 0.995
+            alternatives:
+                - alternative: *piano
+                  probability: 0.005
+        "-1":
+            default: *seminterrato
+            probability: 0.9995
+            alternatives:
+                - alternative: *piano
+                  probability: 0.0005
+        "0":
+            default: *piano_terra
+            probability: 0.95
+            alternatives:
+                - alternative: *piano
+                  probability: 0.05
+        "1":
+            default: *piano
+            probability: 0.9
+            alternatives:
+                - alternative: *piano_nobile
+                  probability: 0.1
+
+    alphanumeric:
+        default: *piano
+        probability: 0.95
+        alternatives:
+            - alternative: *livello
+              probability: 0.05
+        numeric_probability: 0.99
+        alpha_probability: 0.01
+
+    numbering_starts_at: 0
+
+cross_streets:
+    # 26th & 6th Avenue
+    and: *e
+    # 26th @ Broadway
+    a: &a
+        canonical: a
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    corner_of: &angolo_di
+        canonical: angolo di
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    corner: &angolo
+        canonical: angolo
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    at_the_corner_of: &all_angolo_tra
+        canonical: all'angolo tra
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    intersection:
+        default: *e
+        probability: 0.7
+        alternatives:
+            - alternative: *a
+              probability: 0.05
+            - alternative: *angolo_di
+              probability: 0.15
+            - alternative: *all_angolo_tra
+              probability: 0.1
+
+    # 26th betw 5th Ave and 6th Ave
+    between:
+        canonical: tra
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5 # Probability of using parentheses e.g. (between 5th and 6th)
+
+categories:
+    near:
+        default:
+            canonical: vicino a
+        probability: 0.75
+        alternatives:
+            - alternative:
+                canonical: presso a
+              probability: 0.25
+    nearby:
+        default:
+            canonical: vicino
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: qui vicino
+              probability: 0.1
+            - alternative:
+                canonical: nelle vicinanze
+              probability: 0.1
+            - alternative:
+                canonical: intorno a qui
+              probability: 0.1
+
+    near_me:
+        default:
+            canonical: vicino a me
+
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: a
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: ad
+              probability: 0.15
+            - alternative:
+                canonical: in
+              probability: 0.15
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+
+directions:
+    right: &destra
+        canonical: destra
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: right
+    left: &sinistra
+        canonical: sinistra
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: right
+    rear: &posteriore
+        canonical: posteriore
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    front: &anteriore
+        canonical: anteriore
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *destra
+          probability: 0.49
+        - alternative: *sinistra
+          probability: 0.49
+        - alternative: *posteriore
+          probability: 0.01
+        - alternative: *anteriore
+          probability: 0.01
+
+    anteroposterior:
+        alternatives:
+            - alternative: *anteriore
+              probability: 0.5
+            - alternative: *posteriore
+              probability: 0.5
+
+    lateral:
+        alternatives:
+            - alternative: *destra
+              probability: 0.5
+            - alternative: *sinistra
+              probability: 0.5
+
+cardinal_directions:
+    east: &est
+        canonical: est
+        abbreviated: e
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6 
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: e
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &ovest
+        canonical: ovest
+        abbreviated: o
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: o
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &nord
+        canonical: nord
+        abbreviated: n
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &sud
+        canonical: sud
+        abbreviated: s
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *nord
+          probability: 0.25
+        - alternative: *est
+          probability: 0.25
+        - alternative: *sud
+          probability: 0.25
+        - alternative: *ovest
+          probability: 0.25
+
+entrances:
+    entrance: &ingresso
+        canonical: ingresso
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Ingresso 1, Ingresso A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *ingresso
+        numeric_probability: 0.1 # e.g. Ingresso 1
+        alpha_probability: 0.85 # e.g. Ingresso A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            modifier:
+                direction: right # e.g. Ingresso Nord
+                direction_probability: 0.95
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *sud
+                    - alternative: *est
+                    - alternative: *ovest
+                    - alternative: *destra
+                    - alternative: *sinistra
+                    - alternative: *posteriore
+                    - alternative: *anteriore
+
+staircases:
+    scala: &scala
+        canonical: scala
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    alphanumeric:
+        # For alphanumerics, Scala A, Scala 1, etc.
+        default: *scala
+        numeric_probability: 0.6 # e.g. Scala 1
+        alpha_probability: 0.35 # e.g. Scala A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: right # e.g. Scala Destra
+            direction_probability: 0.9
+            modifier:
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *sud
+                    - alternative: *est
+                    - alternative: *ovest
+                    - alternative: *destra
+                    - alternative: *sinistra
+                    - alternative: *posteriore
+                    - alternative: *anteriore
+
+
+po_boxes:
+    casella_postale: &casella_postale
+        canonical: casella postale
+        abbreviated: cp
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # CP No 1234
+        numeric_probability: 1.0
+    alphanumeric:
+        default: *casella_postale
+        numeric_probability: 0.9 # CP 123
+        alpha_probability: 0.05 # CP A
+        numeric_plus_alpha_probability: 0.04 # CP 123G
+        alpha_plus_numeric_probability: 0.01 # CP A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+
+units:
+    flat: &appartamento
+        canonical: appartamento
+        abbreviated: app
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.5
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    casa: &casa
+        canonical: casa
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    unit: &unita
+        canonical: unità
+        abbreviated: u
+        sample: true
+        canonical_probability: 0.8
+        abbreviated_probability: 0.1
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    office: &officina
+        canonical: officina
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.3
+    lotto: &lotto
+        canonical: lotto
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    door: &porta
+        canonical: porta
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    interno: &interno
+        canonical: interno
+        abbreviated: int
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    room: &sala
+        canonical: sala
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+
+    alphanumeric: &unit_alphanumeric
+        default: *appartamento
+        probability: 0.75
+        alternatives:
+            - alternative: *interno
+              probability: 0.1
+              # e.g. just plain #3 or No. 4
+            - alternative: *numero
+              probability: 0.05
+            - alternative: *casa
+              probability: 0.05
+            - alternative: *porta
+              probability: 0.045
+            - alternative: *sala
+              probability: 0.005
+        numeric_probability: 0.9 # e.g. Appartement 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. Appartement A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # Separate random probability for adding directions like 2D, 2G, etc.
+        add_direction: true
+        add_direction_probability: 0.1
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+        # Add direction only e.g. Unité Gauche
+        add_direction_standalone: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.1
+
+    zones:
+        residential: *unit_alphanumeric
+        commercial:
+            default: *officina
+            probability: 0.8
+            alternatives:
+                - alternative: *sala
+                  probability: 0.2
+
+            numeric_probability: 0.9 # e.g. Bureau 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Bureau 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Bureau A1
+            alpha_probability: 0.08 # e.g. Bureau A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+        industrial:
+            default: *lotto
+            probability: 0.5
+            alternatives:
+                - alternative: *officina
+                  probability: 0.3
+                - alternative: *unita
+                  probability: 0.2
+
+            numeric_probability: 0.9 # e.g. Lotto 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Lotto 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Lotto A1
+            alpha_probability: 0.08 # e.g. Lotto A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+        university:
+            default: *sala
+            probability: 0.9
+            alternatives:
+                - alternative: *porta
+                  probability: 0.1
+
+            numeric_probability: 0.9 # e.g. Salle 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Salle 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Salle A1
+            alpha_probability: 0.08 # e.g. Salle A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+

resources/addresses/ja.yaml

@@ -0,0 +1,161 @@
+# ja.yaml
+# -------
+# Japanese language specification
+
+whitespace: false
+
+components:
+    level:
+        null_probability: 0.95 # Probability of doing nothing if no floor number is specified
+        alphanumeric_probability: 0.05
+
+    unit:
+        # If no unit number is specified
+        null_probability: 1.0
+        conditional:
+            - component: level
+              probabilities:
+                  null_probability: 0.95
+                  alphanumeric_probability: 0.05
+            - component: house_number
+              probabilities:
+                  null_probability: 0.6
+                  alphanumeric_probability: 0.4
+
+    combinations:
+        # Unit is just appended onto the house number
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "-"
+                  probability: 1.0
+            probability: 1.0
+
+numbers:
+    default: &go
+        canonical: 号
+        numeric_affix:
+            affix: 号
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+
+blocks:
+    alphanumeric:
+        default: &ban
+            canonical: 番
+            numeric_affix:
+                affix: 番
+                direction: right
+            numeric_probability: 0.0
+            numeric_affix_probability: 1.0
+        probability: 0.85
+        alternatives:
+            - alternative: &banchi
+                canonical: 番地
+                numeric_affix:
+                    affix: 番地
+                    direction: right
+                numeric_probability: 0.0
+                numeric_affix_probability: 1.0
+              probability: 0.1
+            - alternative: &banchi_no
+                canonical: 番地の
+                numeric_affix:
+                    affix: 番地の
+                    direction: right
+                numeric_probability: 0.0
+                numeric_affix_probability: 1.0
+              probability: 0.05
+        numeric_probability: 1.0
+    alphanumeric_phrase_probability: 0.4
+
+house_numbers:
+    alphanumeric:
+        default: *go
+    alphanumeric_phrase_probability: 0.4
+
+levels:
+    kai: &kai
+        canonical: 階
+        numeric_affix:
+            affix: 階
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                unicode_full_width_probability: 0.5
+                spellout_probability: 0.2
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *kai
+        numeric_probability: 1.0
+
+po_boxes:
+    shishobako: &shishobako
+        canonical: 私書箱
+        numeric_affix:
+            affix: 私書箱
+            direction: left
+            digits:
+                unicode_full_width_probability: 0.5
+                spellout_probability: 0.2
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    alphanumeric:
+        default: *shishobako
+        numeric_probability: 1.0
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+metro_stations:
+    alphanumeric:
+        default: &eki
+            canonical: 駅
+            numeric:
+                direction: right
+            numeric_affix:
+                affix: 駅
+                direction: right
+            numeric_affix_probability: 1.0
+
+    alphanumeric_phrase_probability: 1.0
+
+postcodes:
+    alphanumeric:
+        default:
+            canonical: 〒
+            numeric_affix:
+                affix: 〒
+                direction: left
+            # null_probability means the chance of doing nothing e.g. just the postal code
+            null_probability: 0.1
+            numeric_probability: 0.0
+            numeric_affix_probability: 0.9
+
+units:
+    alphanumeric:
+        numeric_probability: 1.0
+        use_positive_numbers_probability: 1.0
+        # If we have a floor number (from building:levels), use it
+        use_floor_probability: 0.8

resources/addresses/ja_rm.yaml

@@ -0,0 +1,180 @@
+# ja_rm.yaml
+# ----------
+# Romaji (Romanized Japanese) language specification
+
+components:
+    level:
+        null_probability: 0.95 # Probability of doing nothing if no floor number is specified
+        alphanumeric_probability: 0.05
+
+    unit:
+        # If no unit number is specified
+        null_probability: 1.0
+        conditional:
+            - component: level
+              probabilities:
+                  null_probability: 0.95
+                  alphanumeric_probability: 0.05
+            - component: house_number
+              probabilities:
+                  null_probability: 0.6
+                  alphanumeric_probability: 0.4
+
+
+    combinations:
+        # Unit is just appended onto the house number
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "-"
+                  probability: 1.0
+            probability: 1.0
+
+numbers:
+    default: &go
+        canonical: go
+        numeric_affix:
+            affix: -go
+            upper_case: false
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+blocks:
+    alphanumeric:
+        default: &ban
+            canonical: ban
+            numeric_affix:
+                affix: -ban
+                upper_case: false
+                direction: right
+            numeric_probability: 0.0
+            numeric_affix_probability: 1.0
+        probability: 0.85
+        alternatives:
+            - alternative: &banchi
+                canonical: banchi
+                numeric_affix:
+                    affix: -ban
+                    upper_case: false
+                    direction: right
+                numeric_probability: 0.0
+                numeric_affix_probability: 1.0
+              probability: 0.1
+            - alternative: &banchi_no
+                canonical: banchi-no
+                numeric_affix:
+                    affix: -banchi-no
+                    upper_case: false
+                    direction: right
+                numeric_probability: 0.0
+                numeric_affix_probability: 1.0
+              probability: 0.05
+        numeric_probability: 1.0
+    alphanumeric_phrase_probability: 0.4
+
+house_numbers:
+    alphanumeric:
+        default: *go
+    alphanumeric_phrase_probability: 0.4
+
+levels:
+    kai: &kai
+        canonical: kai
+        numeric_affix:
+            affix: -kai
+            upper_case: false
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                unicode_full_width_probability: 0.5
+                spellout_probability: 0.2
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    gai: &gai
+        canonical: gai
+        numeric_affix:
+            affix: -gai
+            upper_case: false
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                unicode_full_width_probability: 0.5
+                spellout_probability: 0.2
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *kai
+        probability: 0.6
+        alternatives:
+            - alternative: *gai
+              probability: 0.4
+        numeric_probability: 1.0
+
+po_boxes:
+    shishobako: &shishobako
+        canonical: shishobako
+        numeric:
+            direction: left
+        numeric_probability: 1.0
+
+    alphanumeric:
+        default: *shishobako
+        numeric_probability: 1.0
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+
+metro_stations:
+    alphanumeric:
+        default: &eki
+            canonical: eki
+            numeric:
+                direction: right
+                title_case: false
+            numeric_affix:
+                affix: -eki
+                title_case: false
+                direction: right
+            numeric_affix_probability: 1.0
+    alphanumeric_phrase_probability: 1.0
+
+
+postcodes:
+    alphanumeric:
+        # This should still be the default in Romaji
+        default:
+            canonical: 〒
+            numeric_affix:
+                affix: 〒
+                direction: left
+            # null_probability means the chance of doing nothing e.g. just the postal code
+            null_probability: 0.1
+            numeric_probability: 0.0
+            numeric_affix_probability: 0.9
+
+units:
+    alphanumeric:
+        numeric_probability: 1.0
+        use_positive_numbers_probability: 1.0
+        # If we have a floor number (from building:levels), use it
+        use_floor_probability: 0.8

resources/addresses/ko.yaml

@@ -0,0 +1,122 @@
+# ko.yaml
+# -------
+# Korean language specification
+
+whitespace: false
+
+components:
+    level:
+        null_probability: 0.85 # Probability of doing nothing if no floor number is specified
+        alphanumeric_probability: 0.15
+
+    unit:
+        # If no unit number is specified
+        null_probability: 0.6
+        alphanumeric_probability: 0.4
+
+numbers:
+    combinations:
+        # Unit is just appended onto the house number
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "-"
+                  probability: 1.0
+            probability: 1.0
+
+numbers:
+    default: &ho
+        canonical: 호
+        numeric_affix:
+            affix: 호
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    probability: 0.9
+    alternatives:
+        - alternative: &ho_traditional
+              canonical: 號
+              numeric_affix:
+                  affix: 號
+                  direction: right
+              numeric_probability: 0.0
+              numeric_affix_probability: 1.0
+          probability: 0.1
+
+levels:
+    cheung: &cheung
+        canonical: 층
+        numeric_affix:
+            affix: 층
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                unicode_full_width_probability: 0.5
+                spellout_probability: 0.2
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *cheung
+        numeric_probability: 1.0
+
+po_boxes:
+    saseoham: &saseoham
+        canonical: 사서함
+        numeric_affix:
+            affix: 사서함
+            direction: left
+            digits:
+                ascii_probability: 0.7
+                unicode_full_width_probability: 0.1
+                spellout_probability: 0.2
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    alphanumeric:
+        default: *saseoham
+        numeric_probability: 1.0
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+
+postcodes:
+    alphanumeric:
+        default: &upyeon_beonho
+            canonical: 우편번호
+            numeric_affix:
+                affix: 우편번호
+                direction: left
+            # null_probability means the chance of doing nothing e.g. just the postal code
+            null_probability: 0.9
+            numeric_probability: 0.0
+            numeric_affix_probability: 0.1
+
+units:
+    alphanumeric:
+        default: *ho
+        probability: 0.9
+        alternatives:
+            - alternative: *ho_traditional
+              probability: 0.1
+        numeric_probability: 1.0
+        use_positive_numbers_probability: 1.0
+        # If we have a floor number (from building:levels), use it
+        use_floor_probability: 0.8

resources/addresses/ko_rm.yaml

@@ -0,0 +1,90 @@
+# ko_rm.yaml
+# ----------
+# Romanized Korean language specification
+
+whitespace: false
+
+components:
+    level:
+        null_probability: 0.85 # Probability of doing nothing if no floor number is specified
+        alphanumeric_probability: 0.15
+
+    unit:
+        # If no unit number is specified
+        null_probability: 0.6
+        alphanumeric_probability: 0.4
+
+numbers:
+    combinations:
+        # Unit is just appended onto the house number
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "-"
+                  probability: 1.0
+            probability: 1.0
+
+numbers:
+    default: &ho
+        canonical: ho
+        numeric_affix:
+            affix: -ho
+            upper_case: false
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+levels:
+    cheung: &cheung
+        canonical: cheung
+        numeric_affix:
+            affix: -cheung
+            upper_case: false
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                unicode_full_width_probability: 0.5
+                spellout_probability: 0.2
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *cheung
+        numeric_probability: 1.0
+
+po_boxes:
+    saseoham: &saseoham
+        canonical: saseoham
+        numeric:
+            direction: left
+
+    alphanumeric:
+        default: *saseoham
+        numeric_probability: 1.0
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    alphanumeric:
+        default: *ho
+        numeric_probability: 1.0
+        use_positive_numbers_probability: 1.0
+        # If we have a floor number (from building:levels), use it
+        use_floor_probability: 0.8

resources/addresses/lt.yaml

@@ -0,0 +1,391 @@
+# lt.yaml
+# -------
+# Lithuanian language specification.
+
+components:
+    level:
+        null_probability: 0.97
+        alphanumeric_probability: 0.02
+        standalone_probability: 0.01
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+    combinations:
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "-"
+                  probability: 0.95
+                - separator: " - "
+                  probability: 0.05
+            probability: 0.8
+
+
+numbers:
+    default: &numeris
+        canonical: numeris
+        abbreviated: nr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+and:
+    default: &ir
+        canonical: ir
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+
+cross_streets:
+    and: *ir
+    corner_of: &kampelis
+        canonical: kampelis
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *ir
+        probability: 0.7
+        alternatives:
+            - alternative: *kampelis
+              probability: 0.3
+
+    between:
+        canonical: nuo
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+
+levels:
+    aukstas: &aukstas
+        canonical: aukštas
+        abbreviated: auk
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.3
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+        numeric_probability: 0.2
+        ordinal_probability: 0.8
+    aukste: &aukste
+        <<: *aukstas
+        canonical: aukšte
+    # Ground floor
+    pirmas_aukstas: &pirmas_aukstas
+        canonical: pirmas aukštas
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    rusys: &rusys
+        canonical: rūsys
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        standalone_probability: 1.0
+        number_abs_value: true
+        number_min_abs_value: 1
+    rusyje: &rusyje
+        canonical: rūsyje
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        # e.g. rūsyje 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. r1
+        numeric_affix:
+            affix: r
+            direction: left
+        # e.g. 1. rūsyje
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.1
+        ordinal_probability: 0.4
+    aliases:
+        "<-1":
+            default: *rusyje
+        "-1":
+            default: *rusys
+        "0": &ground_floor
+            default: *pirmas_aukstas
+            probability: 0.6
+            alternatives:
+                - alternative: *aukste
+                  probability: 0.3
+                - alternative: *aukstas
+                  probability: 0.1
+        "1": *ground_floor
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *aukstas
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+directions:
+    right: &desineje
+        canonical: dešinėje
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &kaireje
+        canonical: kairėje
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *desineje
+          probability: 0.5
+        - alternative: *kaireje
+          probability: 0.5
+
+cardinal_directions:
+    east: &rytai
+        canonical: rytai
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+
+    west: &vakarai
+        canonical: vakarai
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+
+    north: &siaure
+        canonical: šiaurė
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+
+    south: &pietus
+        canonical: pietūs
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+
+    alternatives:
+        - alternative: *siaure
+          probability: 0.25
+        - alternative: *rytai
+          probability: 0.25
+        - alternative: *pietus
+          probability: 0.25
+        - alternative: *vakarai
+          probability: 0.25
+
+
+entrances:
+    wejscie: &iejimas
+        canonical: įėjimas
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # įėjimas 1, įėjimas A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *iejimas
+        numeric_probability: 0.1 # e.g. įėjimas 1
+        alpha_probability: 0.85 # e.g. įėjimas A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    laiptai: &laiptai
+        canonical: laiptai
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *laiptai
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *siaure
+                    - alternative: *rytai
+                    - alternative: *pietus
+                    - alternative: *vakarai
+
+
+po_boxes:
+    pasto_dezute: &pasto_dezute
+        canonical: pašto dėžutė
+        abbreviated: p d
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.4
+        sample_probability: 0.5
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # pašto dėžutė 1234
+    alphanumeric:
+        default: *pasto_dezute
+        numeric_probability: 0.95 # P. d. 123
+        alpha_probability: 0.01 # pašto dėžutė A
+        numeric_plus_alpha_probability: 0.03 # P. d. 123G
+        alpha_plus_numeric_probability: 0.01 # P. d. A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    butas: &butas
+        canonical: butas
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    biuro: &biuro
+        canonical: biuro
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    kambarys: &kambarys
+        canonical: kambarys
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *butas
+        numeric_probability: 0.9 # e.g. butas 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. butas A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.01
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *biuro
+            numeric_probability: 0.95 # e.g. biuro 1
+            numeric_plus_alpha_probability: 0.01 # e.g. biuro 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. biuro A1
+            alpha_probability: 0.03 # e.g. biuro A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university:
+            default: *kambarys
+            numeric_probability: 0.95 # e.g. kambarys 1
+            numeric_plus_alpha_probability: 0.01 # e.g. kambarys 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. kambarys A1
+            alpha_probability: 0.03 # e.g. kambarys A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/lv.yaml

@@ -0,0 +1,403 @@
+# lv.yaml
+# -------
+# Latvian language specification.
+
+components:
+    level:
+        null_probability: 0.97
+        alphanumeric_probability: 0.02
+        standalone_probability: 0.01
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+    combinations:
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "-"
+                  probability: 0.95
+                - separator: " - "
+                  probability: 0.05
+            probability: 0.2
+
+
+numbers:
+    default: &numurs
+        canonical: numurs
+        abbreviated: nr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+and:
+    default: &un
+        canonical: un
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+
+cross_streets:
+    and: *un
+    corner_of: &sturis
+        canonical: stūris
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    at_the_corner_of: &sturi
+        canonical: stūrī
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *un
+        probability: 0.7
+        alternatives:
+            - alternative: *sturi
+              probability: 0.2
+            - alternative: *sturis
+              probability: 0.1
+
+    between:
+        canonical: starp
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+
+levels:
+    stavs: &stavs
+        canonical: stāvs
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        ordinal:
+            direction: right
+            whitespace_probability: 0.5 # sometimes should be 2.stāvs
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+                # Needs to be 1.0 so we don't get e.g. IIstāvs
+                ordinal_suffix_probability: 1.0
+        numeric_probability: 0.2
+        ordinal_probability: 0.8
+
+    # Ground floor
+    pirmais_stavs: &pirmais_stavs
+        canonical: pirmais stāvs
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    pagrabs: &pagrabs
+        canonical: pagrabs
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        standalone_probability: 1.0
+        number_abs_value: true
+        number_min_abs_value: 1
+    pagraba: &pagraba
+        canonical: pagraba
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        # e.g. pagraba 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. p1
+        numeric_affix:
+            affix: p
+            direction: left
+        # e.g. 1. pagraba
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.1
+        ordinal_probability: 0.4
+    aliases:
+        "<-1":
+            default: *pagraba
+        "-1":
+            default: *pagrabs
+        "0": &ground_floor
+            default: *pirmais_stavs
+            probability: 0.6
+            alternatives:
+                - alternative: *stavs
+                  probability: 0.4
+        "1": *ground_floor
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *stavs
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+directions:
+    right: &pa_labi
+        canonical: pa labi
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &pa_kreisi
+        canonical: pa kreisi
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *pa_labi
+          probability: 0.5
+        - alternative: *pa_kreisi
+          probability: 0.5
+
+cardinal_directions:
+    east: &austrumu
+        canonical: austrumu
+        abbreviated: a
+        sample: true
+        canonical_probability: 0.8
+        abbreviated_probability: 0.05
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: a
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &rietumu
+        canonical: rietumu
+        abbreviated: r
+        sample: true
+        canonical_probability: 0.8
+        abbreviated_probability: 0.05
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: r
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &ziemelu
+        canonical: ziemeļu
+        abbreviated: z
+        sample: true
+        canonical_probability: 0.8
+        abbreviated_probability: 0.05
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: z
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+
+    south: &dienvidu
+        canonical: dienvidu
+        abbreviated: d
+        sample: true
+        canonical_probability: 0.8
+        abbreviated_probability: 0.05
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: d
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *ziemelu
+          probability: 0.25
+        - alternative: *dienvidu
+          probability: 0.25
+        - alternative: *austrumu
+          probability: 0.25
+        - alternative: *rietumu
+          probability: 0.25
+
+
+entrances:
+    ieeja: &ieeja
+        canonical: ieeja
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # ieeja 1, ieeja A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *ieeja
+        numeric_probability: 0.1 # e.g. ieeja 1
+        alpha_probability: 0.85 # e.g. ieeja A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    kapnu: &kapnu
+        canonical: kāpņu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    kapnu_telpa: &kapnu_telpa
+        canonical: kāpņu telpa
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *kapnu
+        probability: 0.6
+        alternatives:
+            - alternative: *kapnu_telpa
+              probability: 0.4
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *pa_labi
+                    - alternative: *pa_kreisi
+                    - alternative: *ziemelu
+                    - alternative: *dienvidu
+                    - alternative: *austrumu
+                    - alternative: *rietumu
+
+
+units:
+    dzivoklis: &dzivoklis
+        canonical: dzīvoklis
+        abbreviated: dz
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.8
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    birojs: &birojs
+        canonical: birojs
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    istaba: &istaba
+        canonical: istaba
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *dzivoklis
+        numeric_probability: 0.9 # e.g. m. 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. m. A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.01
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *birojs
+            numeric_probability: 0.95 # e.g. birojs 1
+            numeric_plus_alpha_probability: 0.01 # e.g. birojs 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. birojs A1
+            alpha_probability: 0.03 # e.g. birojs A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university:
+            default: *istaba
+            numeric_probability: 0.95 # e.g. istaba 1
+            numeric_plus_alpha_probability: 0.01 # e.g. istaba 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. istaba A1
+            alpha_probability: 0.03 # e.g. istaba A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/nb.yaml

@@ -0,0 +1,563 @@
+# nb.yaml
+# -------
+# Norwegian language specification.
+
+components:
+    level:
+        null_probability: 0.85
+        alphanumeric_probability: 0.1
+        standalone_probability: 0.05
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+    combinations:
+        # Bolignummer
+        -
+            components:
+                - level
+                - unit
+            label: unit
+            zero_pad_digits: 2
+            separators:
+                - separator: ""
+                  probability: 1.0
+            probability: 0.05
+
+
+numbers:
+    default: &nummer
+        canonical: nummer
+        abbreviated: nr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+house_numbers:
+    alphanumeric:
+        default: *nummer
+
+    alphanumeric_phrase_probability: 0.0001
+
+
+and:
+    default: &og
+        canonical: og
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+cross_streets:
+    and: *og
+    corner_of: &hjorne_av
+        canonical: hjørne av
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    at_the_corner_of: &pa_hjornet_av
+        canonical: på hjørnet av
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *og
+        probability: 0.7
+        alternatives:
+            - alternative: *hjorne_av
+              probability: 0.15
+            - alternative: *pa_hjornet_av
+              probability: 0.15
+
+    between:
+        canonical: mellom
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &etasje
+        canonical: etasje
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+            direction_probability: 0.9
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    hovedetasje: &hovedetasje
+        canonical: hovedetasje
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: h
+            direction: left
+            zero_pad: 2
+        numeric_probability: 0.1
+        numeric_affix_probability: 0.9
+    underetasje: &underetasje
+        canonical: underetasje
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: u
+            direction: left
+            zero_pad: 2
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.1
+        numeric_affix_probability: 0.9
+    loftsetasje: &loftsetasje
+        canonical: loftsetasje
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: l
+            direction: left
+            zero_pad: 2
+        numeric_probability: 0.1
+        numeric_affix_probability: 0.9
+    loft: &loft
+        canonical: loft
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    kjeller: &kjeller
+        canonical: kjeller
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        # e.g. 1 kjeller
+        numeric:
+            direction: right
+            direction_probability: 0.8
+        # e.g. k01
+        numeric_affix:
+            affix: k
+            direction: left
+            zero_pad: 2
+        # e.g. 1. k
+        ordinal:
+            direction: right
+        standalone_probability: 0.9
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.09
+        ordinal_probability: 0.005
+    aliases:
+        "<-1":
+            default: *kjeller
+        "-1":
+            default: *kjeller
+            probability: 0.85
+            alternatives:
+                - alternative: *etasje
+                  probability: 0.05
+                - alternative: *underetasje
+                  probability: 0.1
+
+        "top":
+            default: *etasje
+            probability: 0.85
+            alternatives:
+                - alternative: *loftsetasje
+                  probability: 0.1
+                - alternative: *loft
+                  probability: 0.05
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *etasje
+        probability: 0.95
+        alternatives:
+            - alternative: *hovedetasje
+              probability: 0.05
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+
+categories:
+    near:
+        default:
+            canonical: i nærheten av
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.9
+        alternatives:
+            - alternative:
+                canonical: nær
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+    nearby:
+        default:
+            canonical: i nærheten
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: rundt her
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: nær
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+    near_me:
+        default:
+            canonical: nær meg
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: i nærheten av meg
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.4
+
+    in:
+        default:
+            canonical: i
+
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+directions:
+    right: &hoyre
+        canonical: høyre
+        sample: true
+        canonical_probability: 0.1
+        sample_probability: 0.9
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: h
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    left: &venstre
+        canonical: venstre
+        sample: true
+        canonical_probability: 0.1
+        sample_probability: 0.9
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    alternatives:
+        - alternative: *hoyre
+          probability: 0.5
+        - alternative: *venstre
+          probability: 0.5
+
+
+cardinal_directions:
+    east: &ost
+        canonical: øst
+        abbreviated: ø
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: ø
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &vest
+        canonical: vest
+        abbreviated: v
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &nord
+        canonical: nord
+        abbreviated: n
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &syd
+        canonical: syd
+        abbreviated: s
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *nord
+          probability: 0.25
+        - alternative: *ost
+          probability: 0.25
+        - alternative: *syd
+          probability: 0.25
+        - alternative: *vest
+          probability: 0.25
+
+
+entrances:
+    inngang: &inngang
+        canonical: inngang
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Eingang 1, Eingang A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *inngang
+        numeric_probability: 0.1 # e.g. Eingang 1
+        alpha_probability: 0.85 # e.g. Eingang A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    stiege: &stiege
+        canonical: stiege
+        abbreviated: stg
+        sample: true
+        canonical_probability: 0.7
+        abbreviated_probability: 0.2
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    trapp: &trapp
+        canonical: trapp
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+
+
+    alphanumeric: &staircase_alphanumeric
+        default: *trapp
+        probability: 0.8
+        alternatives:
+            - alternative: *stiege
+              probability: 0.2
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *syd
+                    - alternative: *ost
+                    - alternative: *vest
+
+po_boxes:
+    postboks: &postboks
+        canonical: postboks
+        abbreviated: pb
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.2
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # Pb No 1234
+    boks: &boks
+        canonical: boks
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # Boks No 1234
+    alphanumeric:
+        sample: false
+        default: *postboks
+        probability: 0.9
+        alternatives:
+            - alternative: *boks
+              probability: 0.1
+        numeric_probability: 0.9 # Pb 123
+        alpha_probability: 0.05 # Pb A
+        numeric_plus_alpha_probability: 0.04 # Pb 123G
+        alpha_plus_numeric_probability: 0.01 # Pb A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    leilighet: &leilighet
+        canonical: leilighet
+        abbreviated: leil
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.1
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            null_phrase_probability: 0.3
+            # Lejlighed nummer 4
+            add_number_phrase: true
+            add_number_phrase_probability: 0.05
+    hus: &hus
+        canonical: hus
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    vaerelse: &vaerelse
+        canonical: værelse
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *leilighet
+        probability: 0.8
+        alternatives:
+            - alternative: *hus
+              probability: 0.1
+            - alternative: *vaerelse
+              probability: 0.1
+        numeric_probability: 0.95 # e.g. Lejlighed 1
+        alpha_probability: 0.05 # e.g. Lejl A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # Separate random probability for adding directions like 2H, 2V, etc.
+        add_direction: true
+        add_direction_probability: 0.005
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+        # Add direction only e.g. Lejlighed Venstre
+        add_direction_standalone: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.2
+
+        # Use the actual floor phrase as long as the whole phrase is numeric
+        # Has the effect of creating Bolignummer-style units
+        use_floor_affix_unit_num_digits: 2

resources/addresses/nl.yaml

@@ -0,0 +1,572 @@
+# nl.yaml
+# -------
+# Note: base config covers Dutch as spoken in the Netherlands
+# Belgium overrides go in country configs
+
+components:
+    level:
+        null_probability: 0.85
+        alphanumeric_probability: 0.1
+        standalone_probability: 0.05
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.8
+        alphanumeric_probability: 0.2
+
+    combinations:
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: /
+                  probability: 0.9
+                - separator: "-"
+                  probability: 0.1
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+            label: house_number
+            separators:
+                - separator: "-"
+                  probability: 0.9
+                - separator: /
+                  probability: 0.1
+            probability: 0.01
+
+
+and:
+    default: &en
+        canonical: en
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+numbers:
+    default: &nummer
+        canonical: nummer
+        abbreviated: nr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+house_numbers:
+    alphanumeric:
+        default: *nummer
+    alphanumeric_phrase_probability: 0.01
+
+levels:
+    verdieping: &verdieping
+        canonical: verdieping
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            digits:
+                ascii_probability: 0.8
+                roman_numeral_probability: 0.2
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.5
+                roman_numeral_probability: 0.3
+                spellout_probability: 0.2
+        numeric_probability: 0.7
+        ordinal_probability: 0.3
+    etage: &etage
+        canonical: etage
+        abbreviated: et
+        sample: true
+        canonical_probability: 0.7
+        abbreviated_probability: 0.1
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            digits:
+                ascii_probability: 0.8
+                roman_numeral_probability: 0.2
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.5
+                roman_numeral_probability: 0.3
+                spellout_probability: 0.2
+        numeric_probability: 0.7
+        ordinal_probability: 0.3
+    begane_grond: &begane_grond
+        canonical: begane grond
+        abbreviated: bg
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.2
+        sample_probability: 0.3
+    benedenverdieping: &benedenverdieping
+        canonical: benedenverdieping
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    parterre: &parterre
+        canonical: parterre
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    gelijkvloers: &gelijkvloers
+        canonical: gelijkvloers
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    het_gelijkvloers: &het_gelijkvloers
+        canonical: het gelijkvloers
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    aliases:
+        "0":
+            default: *begane_grond
+            probability: 0.6
+            alternatives:
+                - alternative: *benedenverdieping
+                  probability: 0.35
+                - alternative: *parterre
+                  probability: 0.04
+                - alternative: *het_gelijkvloers
+                  probability: 0.005
+                - alternative: *gelijkvloers
+                  probability: 0.005
+    alphanumeric:
+        default: *verdieping
+        probability: 0.99
+        alternatives:
+            - alternative: *etage
+              probability: 0.01
+        numeric_probability: 0.79 # With this probability, pick an integer
+        roman_numeral_probability: 0.2 # Pick a Roman numeral for the actual value
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+categories:
+    near:
+        default:
+            canonical: in de buurt van
+        probability: 0.8
+        alternatives:
+            - alternative:
+                canonical: bij
+              probability: 0.1
+            - alternative:
+                canonical: nabij
+              probability: 0.1
+    nearby:
+        default:
+            canonical: in de buurt
+    near_me:
+        default:
+            canonical: in de buurt van me
+
+    in:
+        default:
+            canonical: in
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: te
+              probability: 0.4
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+
+
+cross_streets:
+    and: *en
+    corner_of: &hoek_van
+        canonical: hoek van
+    at_the_corner_of: &op_de_hoek_van
+        canonical: op de hoek van
+    intersection:
+        default: *en
+        probability: 0.7
+        alternatives:
+            - alternative: *hoek_van
+              probability: 0.15
+            - alternative: *op_de_hoek_van
+              probability: 0.15
+
+    between:
+        canonical: tussen
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+
+entrances:
+    ingang: &ingang
+        canonical: ingang
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Eingang 1, Eingang A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *ingang
+        numeric_probability: 0.1 # e.g. Eingang 1
+        alpha_probability: 0.85 # e.g. Eingang A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+po_boxes:
+    postbus: &postbus
+        canonical: postbus
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    antwoordnummer: &antwoordnummer
+        canonical: antwoordnummer
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    alphanumeric:
+        sample: false
+        default: *postbus
+        probability: 0.8
+        alternatives:
+            - alternative: *antwoordnummer
+              probability: 0.2
+        numeric_probability: 0.9 # 123
+        alpha_probability: 0.05 # A
+        numeric_plus_alpha_probability: 0.04 # 123G
+        alpha_plus_numeric_probability: 0.01 # A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+directions:
+    right: &rechts
+        canonical: rechts
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: r
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    left: &links
+        canonical: links
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: l
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+    alternatives:
+        - alternative: *rechts
+          probability: 0.5
+        - alternative: *links
+          probability: 0.5
+
+
+cardinal_directions:
+    east: &oost
+        canonical: oost
+        abbreviated: o
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: o
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    oosten: &oosten
+        <<: *oost
+        canonical: oosten
+
+    oostelijke: &oostelijke
+        canonical: oostelijke
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    west: &west
+        canonical: west
+        abbreviated: w
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: w
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    westen: &westen
+        <<: *west
+        canonical: westen
+
+    westelijke: &westelijke
+        canonical: westelijke
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    north: &noord
+        canonical: noord
+        abbreviated: n
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    noorden: &noorden
+        <<: *noord
+        canonical: noorden
+
+    noordelijke: &noordelijke
+        canonical: noordelijke
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    south: &zuid
+        canonical: zuid
+        abbreviated: z
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: z
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    zuiden: &zuiden
+        <<: *zuid
+        canonical: zuiden
+
+    zuidelijke: &zuidelijke
+        canonical: zuidelijke
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+    alternatives:
+        - alternative: *noord
+          probability: 0.25
+        - alternative: *oost
+          probability: 0.25
+        - alternative: *zuid
+          probability: 0.25
+        - alternative: *west
+          probability: 0.25
+
+
+staircases:
+    stiege: &stiege
+        canonical: stiege
+        abbreviated: stg
+        sample: true
+        canonical_probability: 0.7
+        abbreviated_probability: 0.2
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    trap: &trap
+        canonical: trap
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    alphanumeric: &staircase_alphanumeric
+        default: *trap
+        probability: 0.6
+        alternatives:
+            - alternative: *stiege
+              probability: 0.4
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+units:
+    appartement: &appartement
+        canonical: appartement
+        abbreviated: apt
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    kamer: &kamer
+        canonical: kamer
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *appartement
+        probability: 0.6
+        alternatives:
+            - alternative: *kamer
+              probability: 0.4
+        numeric_probability: 0.9 # e.g. Apt 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. Apt A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # Separate random probability for adding directions like 2R, 2L, etc.
+        add_direction: true
+        add_direction_probability: 0.1
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+        # Add direction only e.g. Apt Rechts
+        add_direction_standalone: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.1
+
+
+countries:
+    be:
+        components:
+            unit:
+                null_probability: 0.65
+                alphanumeric_probability: 0.35
+
+        levels:
+            verdieping: &verdieping_flemish
+                canonical: verdieping
+                abbreviated: verdiep
+                sample: true
+                canonical_probability: 0.3
+                abbreviated_probability: 0.5
+                sample_probability: 0.2
+                numeric:
+                    direction: left
+                ordinal:
+                    direction: right
+                    digits:
+                        ascii_probability: 0.8
+                        spellout_probability: 0.2
+                numeric_probability: 0.7
+                ordinal_probability: 0.3
+
+            aliases:
+                "0":
+                    default: *het_gelijkvloers
+                    probability: 0.5
+                    alternatives:
+                        - alternative: *gelijkvloers
+                          probability: 0.5
+            alphanumeric:
+                default: *verdieping_flemish
+                probability: 0.9
+                alternatives:
+                    - alternative: *etage
+                      probability: 0.1
+
+        units:
+            bus: &bus
+                canonical: bus
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+                numeric:
+                    direction: left
+            alphanumeric:
+                default: *appartement
+                probability: 0.1
+                alternatives:
+                    - alternative: *bus
+                      probability: 0.7
+                    - alternative: *kamer
+                      probability: 0.2

resources/addresses/pl.yaml

@@ -0,0 +1,509 @@
+# pl.yaml
+# -------
+# Polish language specification.
+
+components:
+    level:
+        null_probability: 0.95
+        alphanumeric_probability: 0.04
+        standalone_probability: 0.01
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+    combinations:
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.9
+                - separator: "-"
+                  probability: 0.05
+                - separator: " - "
+                  probability: 0.05
+            probability: 0.01
+
+numbers:
+    default: &numer
+        canonical: numer
+        abbreviated: nr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+house_numbers:
+    dom: &dom
+        canonical: dom
+        abbreviated: d
+        sample: true
+        canonical_probability: 0.8
+        abbreviated_probability: 0.1
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    alphanumeric:
+        default: *numer
+        probability: 0.6
+        alternatives:
+            - alternative: *dom
+              probability: 0.4
+
+    alphanumeric_phrase_probability: 0.0001
+
+and:
+    default: &i
+        canonical: i
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+cross_streets:
+    and: *i
+    at: &w
+        canonical: w
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2        
+    corner_of: &rogu
+        canonical: rogu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    at_the_corner_of: &na_rogu
+        canonical: na rogu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *i
+        probability: 0.7
+        alternatives:
+            - alternative: *w
+              probability: 0.1
+            - alternative: *rogu
+              probability: 0.1
+            - alternative: *na_rogu
+              probability: 0.1
+
+    between:
+        canonical: pomiędzy
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &pietro
+        canonical: piętro
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+                ordinal_suffix_probability: 0.6
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    parter: &parter
+        canonical: parter
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    suterena: &suterena
+        canonical: suterena
+        # e.g. suterena 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. s1
+        numeric_affix:
+            affix: s
+            direction: left
+        # e.g. 1. suterena
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        standalone_probability: 0.985
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.005
+        ordinal_probability: 0.005
+    aliases:
+        "<-1":
+            default: *suterena
+        "-1":
+            default: *suterena
+        "0":
+            default: *parter
+            probability: 0.9
+            alternatives:
+                - alternative: *pietro
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *pietro
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+categories:
+    near:
+        default:
+            canonical: w pobliżu
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: blisko
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: koło
+                sample: true
+                canonical_probability: 0.7
+                sample_probability: 0.3
+              probability: 0.05
+            - alternative:
+                canonical: niedaleko
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05
+            - alternative:
+                canonical: obok
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05
+            - alternative:
+                canonical: przy
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05
+    nearby:
+        default:
+            canonical: w pobliżu
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: w pobliżu tutaj
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: wokół tutaj
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: blisko
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+    near_me:
+        default:
+            canonical: w pobliżu mnie
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: w
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: we
+              probability: 0.3
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+
+directions:
+    right: &prawo
+        canonical: prawo
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &lewo
+        canonical: lewo
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *prawo
+          probability: 0.5
+        - alternative: *lewo
+          probability: 0.5
+
+cardinal_directions:
+    east: &wschod
+        canonical: wschód
+        abbreviated: w
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: w
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &zachod
+        canonical: zachód
+        abbreviated: z
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: z
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &polnoc
+        canonical: północ
+        abbreviated: pn
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: pn
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &poludnie
+        canonical: południe
+        abbreviated: pd
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: pd
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *polnoc
+          probability: 0.25
+        - alternative: *wschod
+          probability: 0.25
+        - alternative: *poludnie
+          probability: 0.25
+        - alternative: *zachod
+          probability: 0.25
+
+
+entrances:
+    wejscie: &wejscie
+        canonical: wejście
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Wejście 1, Wejście A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *wejscie
+        numeric_probability: 0.1 # e.g. Wejście 1
+        alpha_probability: 0.85 # e.g. Wejście A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    schody: &schody
+        canonical: schody
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *schody
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *polnoc
+                    - alternative: *poludnie
+                    - alternative: *wschod
+                    - alternative: *zachod
+
+
+po_boxes:
+    skrytka_pocztowa: &skrytka_pocztowa
+        canonical: skrytka pocztowa
+        abbreviated: skr poczt
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # Skr Poczt 1234
+    alphanumeric:
+        default: *skrytka_pocztowa
+        numeric_probability: 0.9 # Skr Poczt 123
+        alpha_probability: 0.05 # Skr Poczt A
+        numeric_plus_alpha_probability: 0.04 # Skr Poczt 123G
+        alpha_plus_numeric_probability: 0.01 # Skr Poczt A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    mieszkanie: &mieszkanie
+        canonical: mieszkanie
+        abbreviated: m
+        sample: true
+        canonical_probability: 0.05
+        abbreviated_probability: 0.9
+        sample_probability: 0.05
+        numeric:
+            direction: left
+    pokoj: &pokoj
+        canonical: pokój
+        abbreviated: pok
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.5
+        sample_probability: 0.1
+        numeric:
+            direction: left
+
+    alphanumeric: &unit_alphanumeric
+        default: *mieszkanie
+        probability: 0.9
+        alternatives:
+            - alternative: *pokoj
+              probability: 0.1
+        numeric_probability: 0.9 # e.g. m. 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. m. A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.01
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *pokoj
+            numeric_probability: 0.95 # e.g. pokój 1
+            numeric_plus_alpha_probability: 0.01 # e.g. pokój 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. pokój A1
+            alpha_probability: 0.03 # e.g. pokój A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university: *commercial_unit_types

resources/addresses/ro.yaml

@@ -0,0 +1,504 @@
+# ro.yaml
+# -------
+# Romanian language specification
+
+components:
+    level:
+        # If no floor number is specified
+        null_probability: 0.6
+        alphanumeric_probability: 0.35
+        standalone_probability: 0.05
+
+    staircase:
+        null_probability: 0.95
+        alphanumeric_probability: 0.05
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        # If no unit number is specified
+        null_probability: 0.3
+        alphanumeric_probability: 0.65
+        standalone_probability: 0.05
+
+numbers:
+    default: &numar
+        canonical: număr
+        abbreviated: nr
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.7
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#" # e.g. #3, #2F, etc.
+            probability: 0.5
+            alternatives:
+                - alternative: 
+            direction: left # affix goes on the number's left
+
+        # Probabilities for numbers
+        numeric_probability: 0.9
+        numeric_affix_probability: 0.1
+
+and:
+    default: &si
+        canonical: și
+        abbreviated: "&"
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability:  0.4
+        sample_probability: 0.1
+
+cross_streets:
+    and: *si
+    corner_of: &colt
+        canonical: colț
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+    at_the_corner_of: &la_coltul_de_pe
+        canonical: la colțul de pe
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+    intersection:
+        default: *si
+        probability: 0.7
+        alternatives:
+            - alternative: *colt
+              probability: 0.2
+            - alternative: *la_coltul_de_pe
+              probability: 0.1
+
+    between:
+        canonical: între
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        parentheses_probabililty: 0.5
+
+
+house_numbers:
+    # fara numar (FN) addresses
+    no_number: 
+        default:
+            canonical: fără număr
+            abbreviated: fn
+            sample: true
+            canonical_probability: 0.1
+            abbreviated_probability: 0.7
+            sample_probability: 0.2
+    alphanumeric:
+        default: *numar
+
+    alphanumeric_phrase_probability: 0.7
+    no_number_probability: 0.1 # With this probability, use fara numar if no house_number is specified
+
+
+
+levels:
+    floor: &etaj
+        canonical: etaj
+        abbreviated: et
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.4
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            add_number_phrase: true # Occasionally add variation of "number", e.g. et. nr 2
+            add_number_phrase_probability: 0.05
+            digits:
+                ascii_probability: 0.8
+                roman_numeral_probability: 0.2
+    # Ground floor
+    parter: &parter
+        canonical: parter
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    aliases:
+        "0":
+            default: *parter
+            probability: 0.9
+            alternatives:
+                - alternative: *etaj
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *etaj
+        add_number_phrase: true
+        add_number_phrase_probability: 0.05
+        numeric_probability: 0.99
+        alpha_probability: 0.01
+
+blocks:
+    alphanumeric:
+        default:
+            canonical: bloc
+            abbreviated: bl
+            sample: true
+            canonical_probability: 0.6
+            abbreviated_probability: 0.2
+            sample_probability: 0.2
+            numeric:
+                direction: left
+
+
+
+categories:
+    near:
+        default:
+            canonical: in apropiere de
+
+    nearby:
+        default:
+            canonical: în apropiere
+        probability: 0.5
+        alternatives:
+            - alternative:
+                  canonical: in apropiere
+              probability: 0.2
+            - alternative:
+                  canonical: aproape de aici
+              probability: 0.1
+            - alternative:
+                  canonical: aici
+              probability: 0.1
+            - alternative:
+                canonical: în jurul aici
+              probability: 0.05
+            - alternative:
+                canonical: in jurul aici
+              probability: 0.05
+    near_me:
+        default:
+            canonical: lângă mine
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: langa mine
+              probability: 0.3
+    in:
+        default:
+            canonical: din
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+directions:
+    right: &dreapta
+        canonical: dreapta
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: d
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+    left: &stanga
+        canonical: stânga
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+    alternatives:
+        - alternative: *dreapta
+          probability: 0.5
+        - alternative: *stanga
+          probability: 0.5
+
+
+cardinal_directions:
+    east: &est
+        canonical: est
+        abbreviated: e
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6 
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: e
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &vest
+        canonical: vest
+        abbreviated: v
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &nord
+        canonical: nord
+        abbreviated: n
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &sud
+        canonical: sud
+        abbreviated: s
+        canonical_probability: 0.4
+        abbreviated_probability: 0.6
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *nord
+          probability: 0.25
+        - alternative: *est
+          probability: 0.25
+        - alternative: *sud
+          probability: 0.25
+        - alternative: *vest
+          probability: 0.25
+
+entrances:
+    entrada: &intrare
+        canonical: intrare
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    # Intrare 1, Intare A, etc.
+    alphanumeric:
+        default: *intrare
+        numeric_probability: 0.1 # e.g. Intrare 1
+        alpha_probability: 0.85 # e.g. Intrare A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            modifier:
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *sud
+                    - alternative: *est
+                    - alternative: *vest
+                    - alternative: *dreapta
+                    - alternative: *stanga
+
+staircases:
+    scara: &scara
+        canonical: scara
+        abbreviated: sc
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: left
+
+    alphanumeric:
+        # For alphanumerics, Scara A, Scara 1, etc.
+        default: *scara
+        numeric_probability: 0.35 # e.g. Scara 1
+        alpha_probability: 0.6 # e.g. Scara A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: right # e.g. Scara Nord
+            direction_probability: 0.8
+            modifier:
+                alternatives:
+                    - alternative: *nord
+                    - alternative: *sud
+                    - alternative: *est
+                    - alternative: *vest
+                    - alternative: *dreapta
+                    - alternative: *stanga
+
+po_boxes:
+    casuta_postala: &casuta_postala
+        canonical: căsuță poștală
+        abbreviated: cp
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.3
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.4 # Apdo No 1234
+        numeric_probability: 1.0
+    alphanumeric:
+        sample: false
+        default: *casuta_postala
+        numeric_probability: 0.9 # Apdo 123
+        alpha_probability: 0.05 # Apdo A
+        numeric_plus_alpha_probability: 0.04 # Apdo 123G
+        alpha_plus_numeric_probability: 0.01 # Apdo A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    apartament: &apartament
+        canonical: apartament
+        abbreviated: ap
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.6
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    sala: &sala
+        canonical: sală
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    birou: &birou
+        canonical: birou
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    lotul: &lotul
+        canonical: lotul
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *apartament
+        probability: 0.9
+        sample: true
+        alternatives:
+            - alternative: *sala
+              probability: 0.1
+
+        # Separate random probability for adding directions like 2o Izq, 2 Dcha, etc.
+        add_direction: true
+        add_direction_probability: 0.1
+        add_direction_numeric: true # Only for numbers
+
+        numeric_probability: 0.9 # e.g. ap 1
+        numeric_plus_alpha_probability: 0.01 # e.g. ap 1A
+        alpha_plus_numeric_probability: 0.01 # e.g. ap A1
+        alpha_probability: 0.08 # e.g. ap A
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+
+    zones:
+        residential: *unit_alphanumeric
+        commercial:
+            default: *birou
+            numeric_probability: 0.9 # e.g. Birou 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Birou 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Birou A1
+            alpha_probability: 0.08 # e.g. Birou A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+        industrial:
+            default: *lotul
+            probability: 0.5
+            alternatives:
+                - alternative: *birou
+                  probability: 0.3
+                - alternative: *sala
+                  probability: 0.2
+
+            numeric_probability: 0.9 # e.g. Lotul 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Lotul 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Lotul A1
+            alpha_probability: 0.08 # e.g. Lotul A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+
+        university:
+            default: *sala
+            probability: 0.9
+            alternatives:
+                - alternative: *birou
+                  probability: 0.1
+            numeric_probability: 0.9 # e.g. Sala 1
+            numeric_plus_alpha_probability: 0.01 # e.g. Sala 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. Sala A1
+            alpha_probability: 0.08 # e.g. Sala A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/sk.yaml

@@ -0,0 +1,603 @@
+# sk.yaml
+# -------
+# Slovakian language specification
+
+components:
+    level:
+        null_probability: 0.95
+        alphanumeric_probability: 0.04
+        standalone_probability: 0.01
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    # Note: no combinations because of the house numbering scheme
+
+
+numbers:
+    default: &cislo
+        canonical: číslo
+        abbreviated: č
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.6
+        sample_probability: 0.1
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "č."
+            direction: left
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+and:
+    default: &a
+        canonical: a
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+conscription_numbers:
+    alphanumeric:
+        default: 
+            canonical: súpisné číslo
+            abbreviated: s.č.
+            canonical_probability: 0.05
+            abbreviated_probability: 0.85
+            sample: true
+            sample_probability: 0.1
+            numeric:
+                direction: left
+
+cross_streets:
+    and: *a
+    at: &na
+        canonical: na
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2        
+    corner_of: &rohu
+        canonical: rohu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    corner: &roh
+        canonical: roh
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    at_the_corner_of: &na_rohu
+        canonical: na rohu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *a
+        probability: 0.6
+        alternatives:
+            - alternative: *na
+              probability: 0.1
+            - alternative: *roh
+              probability: 0.1
+            - alternative: *rohu
+              probability: 0.1
+            - alternative: *na_rohu
+              probability: 0.1
+
+    between:
+        canonical: medzi
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+levels:
+    floor: &poschodie
+        canonical: poschodie
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    podlazie: &podlazie
+        canonical: podlažie
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+
+    nadzemne_podlazie: &nadzemne_podlazie
+        canonical: nadzemné podlažie
+        abbreviated: np
+        sample: true
+        canonical_probability: 0.1
+        abbreviated_probability: 0.8
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    etaz: &etaz
+        canonical: etáž
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    prizemie: &prizemie
+        canonical: prízemie
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    podzemne_podlazie: &podzemne_podlazie
+        canonical: podzemné podlažie
+        abbreviated: pp
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.2
+        sample_probability: 0.3
+        # e.g. podzemné podlažie 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. pp1
+        numeric_affix:
+            affix: pp
+            direction: left
+        # e.g. 1. podzemné podlažie
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        standalone_probability: 0.985
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.005
+        ordinal_probability: 0.005
+    aliases:
+        "<-1":
+            default: *podzemne_podlazie
+        "-1":
+            default: *podzemne_podlazie
+        "0":
+            default: *prizemie
+            probability: 0.9
+            alternatives:
+                - alternative: *poschodie
+                  probability: 0.05
+                - alternative: *podlazie
+                  probability: 0.05
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *poschodie
+        probability: 0.45
+        alternatives:
+            - alternative: *podlazie
+              probability: 0.35
+            - alternative: *nadzemne_podlazie
+              probability: 0.19
+            - alternative: *etaz
+              probability: 0.01
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+categories:
+    near:
+        default:
+            canonical: v blízkosti
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: u
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: v okolí
+                sample: true
+                canonical_probability: 0.7
+                sample_probability: 0.3
+              probability: 0.05
+            - alternative:
+                canonical: okolo
+                sample: true
+                canonical_probability: 0.7
+                sample_probability: 0.3
+              probability: 0.05
+    nearby:
+        default:
+            canonical: blízkosti
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.4
+        alternatives:
+            - alternative:
+                canonical: blízko
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: v blízkosti
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: tady blízkosti
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: tady
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05
+            - alternative:
+                canonical: tu
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05            
+            - alternative:
+                canonical: v blízkosti tu
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05
+            - alternative:
+                canonical: v okolí
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.05
+    near_me:
+        default:
+            canonical: v blízkosti mne
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: v
+        probability: 0.7
+        alternatives:
+            - alternative:
+                canonical: vo
+              probability: 0.3
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+directions:
+    right: &prava
+        canonical: pravá
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &lava
+        canonical: ľavá
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *prava
+          probability: 0.5
+        - alternative: *lava
+          probability: 0.5
+
+cardinal_directions:
+    east: &vychod
+        canonical: východ
+        abbreviated: v
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &zapad
+        canonical: západ
+        abbreviated: z
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: z
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &sever
+        canonical: sever
+        abbreviated: s
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &juh
+        canonical: juh
+        abbreviated: j
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: j
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *sever
+          probability: 0.25
+        - alternative: *vychod
+          probability: 0.25
+        - alternative: *juh
+          probability: 0.25
+        - alternative: *zapad
+          probability: 0.25
+
+entrances:
+    vchod: &vchod
+        canonical: vchod
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Wejście 1, Wejście A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *vchod
+        numeric_probability: 0.1 # e.g. Wejście 1
+        alpha_probability: 0.85 # e.g. Wejście A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    schodisko: &schodisko
+        canonical: schodisko
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    alphanumeric: &staircase_alphanumeric
+        default: *schodisko
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *sever
+                    - alternative: *juh
+                    - alternative: *vychod
+                    - alternative: *zapad
+
+po_boxes:
+    postova_priehradka: &postova_priehradka
+        canonical: poštová priehradka
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # poštová priehradka 1234
+    alphanumeric:
+        default: *postova_priehradka
+        numeric_probability: 0.9 # poštová priehradka 123
+        alpha_probability: 0.05 # poštová priehradka A
+        numeric_plus_alpha_probability: 0.04 # poštová priehradka 123G
+        alpha_plus_numeric_probability: 0.01 # poštová priehradka A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    apartaman: &apartaman
+        canonical: apartmán
+        abbreviated: apt
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.5
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    izba: &izba
+        canonical: izba
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    kancelaria: &kancelaria
+        canonical: kancelária
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    alphanumeric: &unit_alphanumeric
+        default: *apartaman
+        probability: 0.9
+        alternatives:
+            - alternative: *izba
+              probability: 0.1
+        numeric_probability: 0.9 # e.g. apt. 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. apt. A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.01
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *izba
+            probability: 0.6
+            alternatives:
+                - alternative: *kancelaria
+                  probability: 0.4
+            numeric_probability: 0.95 # e.g. pokoj 1
+            numeric_plus_alpha_probability: 0.01 # e.g. pokoj 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. pokoj A1
+            alpha_probability: 0.03 # e.g. pokoj A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university:
+            default: *izba
+            numeric_probability: 0.95 # e.g. pokoj 1
+            numeric_plus_alpha_probability: 0.01 # e.g. pok 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. pokoj A1
+            alpha_probability: 0.03 # e.g. pokoj A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/sl.yaml

@@ -0,0 +1,539 @@
+# sl.yaml
+# -------
+# Slovenian language specification
+
+components:
+    level:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.7
+        alphanumeric_probability: 0.3
+
+    combinations:
+        -
+            components:
+                - house_number
+                - staircase
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.1
+        # For unit types like 2/34
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+
+
+numbers:
+    no_number: 
+        default:
+            canonical: brez številke
+            abbreviated: brez št
+            sample: true
+            canonical_probability: 0.5
+            abbreviated_probability: 0.3
+            sample_probability: 0.2
+
+    default: &stevilke
+        canonical: številke
+        abbreviated: št
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.6
+        sample_probability: 0.1
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "št."
+            whitespace_probability: 0.6
+            direction: left
+        numeric_probability: 0.6
+        numeric_affix_probability: 0.4
+
+    alphanumeric_phrase_probability: 0.05
+    no_number_probability: 0.05
+
+
+and:
+    default: &in
+        canonical: in
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+
+cross_streets:
+    i: *in
+    at: &na
+        canonical: na
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    corner: &vogalu
+        canonical: vogalu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    na_vogalu: &na_vogalu
+        canonical: na vogalu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *in
+        probability: 0.7
+        alternatives:
+            - alternative: *na
+              probability: 0.1
+            - alternative: *vogalu
+              probability: 0.15
+            - alternative: *na_vogalu
+              probability: 0.05
+
+    med: &med
+        canonical: med
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+    between:
+        default: *med
+
+levels:
+    nadstropje: &nadstropje
+        canonical: nadstropje
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+
+    pritlicje: &pritlicje
+        canonical: pritličje
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    parter: &parter
+        canonical: parter
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    kleti: &kleti
+        canonical: kleti
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        # e.g. kleti 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. 1. kleti
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+
+    aliases:
+        "<-1":
+            default: *kleti
+        "-1":
+            default: *kleti
+        "0":
+            default: *pritlicje
+            probability: 0.5
+            alternatives:
+                - alternative: *parter
+                  probability: 0.4
+                - alternative: *nadstropje
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *nadstropje
+        numeric_probability: 0.69 # With this probability, pick an integer
+        roman_numeral_probability: 0.3 # Pick a Roman numeral for the actual value
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+
+categories:
+    near:
+        default:
+            canonical: v bližini
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.6
+        alternatives:
+            - alternative:
+                canonical: pri
+              probability: 0.4
+
+    nearby:
+        default:
+            canonical: v bližini
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2            
+        probability: 0.5
+        alternatives:
+            - alternative:
+                canonical: v bližini tukaj
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.3
+            - alternative:
+                canonical: okoli tukaj
+              probability: 0.1
+            - alternative:
+                canonical: tukaj
+              probability: 0.1
+
+    near_me:
+        default:
+            canonical: blizu mene
+
+    # Don't worry about agreement
+    in:
+        default:
+            canonical: v
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+directions:
+    right: &prav
+        canonical: prav
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &levo
+        canonical: levo
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *prav
+          probability: 0.5
+        - alternative: *levo
+          probability: 0.5
+
+cardinal_directions:
+    east: &vzhod
+        canonical: vzhod
+        abbreviated: v
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &zahod
+        canonical: zahod
+        abbreviated: z
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: z
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &sever
+        canonical: sever
+        abbreviated: s
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &jug
+        canonical: jug
+        abbreviated: j
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: j
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *sever
+          probability: 0.25
+        - alternative: *vzhod
+          probability: 0.23
+        - alternative: *jug
+          probability: 0.23
+        - alternative: *zahod
+          probability: 0.23
+
+entrances:
+    vhod: &vhod
+        canonical: vhod
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Vhod 1, Vhod A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *vhod
+        numeric_probability: 0.1 # e.g. Vhod 1
+        alpha_probability: 0.85 # e.g. Vhod A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+
+staircases:
+    stopnisce: &stopnisce
+        canonical: stopnišče
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+
+    alphanumeric: &staircase_alphanumeric
+        default: *stopnisce
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: right
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *prav
+                      probability: 0.2
+                    - alternative: *levo
+                      probability: 0.2
+                    - alternative: *sever
+                      probability: 0.15
+                    - alternative: *jug
+                      probability: 0.15
+                    - alternative: *vzhod
+                      probability: 0.15
+                    - alternative: *zahod
+                      probability: 0.15
+
+po_boxes:
+    postni_predal: &postni_predal
+        canonical: poštni predal
+        abbreviated: p.p
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.4
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+
+    alphanumeric:
+        default: *postni_predal
+        numeric_probability: 0.9 # pp 123
+        alpha_probability: 0.05 # p.p A
+        numeric_plus_alpha_probability: 0.04 # pp 123G
+        alpha_plus_numeric_probability: 0.01 # pp A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    stanovanje: &stanovanje
+        canonical: stanovanje
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    soba: &soba
+        canonical: soba
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    urad: &urad
+        canonical: urad
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    alphanumeric: &unit_alphanumeric
+        default: *stanovanje
+        probability: 0.9
+        alternatives:
+            - alternative: *soba
+              probability: 0.1
+        numeric_probability: 0.9 # e.g. stanovanje 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. stanovanje A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.05
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *soba
+            probability: 0.6
+            alternatives:
+                - alternative: *urad
+                  probability: 0.4
+            numeric_probability: 0.95 # e.g. soba 1
+            numeric_plus_alpha_probability: 0.01 # e.g. soba 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. soba A1
+            alpha_probability: 0.03 # e.g. soba A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university:
+            default: *soba
+            numeric_probability: 0.95 # e.g. soba 1
+            numeric_plus_alpha_probability: 0.01 # e.g. soba 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. soba A1
+            alpha_probability: 0.03 # e.g. soba A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/sr.yaml

@@ -0,0 +1,953 @@
+# sr.yaml
+# -------
+# Serbian language specification
+
+alphabet: абвгдђежзијклљмнњопрстћуфхцчџш
+alphanumeric_probability: 0.7
+
+components:
+    level:
+        null_probability: 0.8
+        alphanumeric_probability: 0.2
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.7
+        alphanumeric_probability: 0.3
+
+
+    combinations:
+        -
+            components:
+                - house_number
+                - staircase
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.1
+        # For unit types like 2/34
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+
+
+
+numbers:
+    default: &broj
+        canonical: број
+        abbreviated: бр
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.6
+        sample_probability: 0.1
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "бр."
+            direction: left
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+    alternatives:
+        - alternative: &broj_latin
+            canonical: broj
+            abbreviated: br
+            sample: true
+            canonical_probability: 0.3
+            abbreviated_probability: 0.6
+            sample_probability: 0.1
+            numeric:
+                direction: left
+            numeric_affix:
+                affix: "br."
+                direction: left
+            numeric_probability: 0.4
+            numeric_affix_probability: 0.6
+
+
+and:
+    default: &i
+        canonical: и
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    probability: 0.9
+    alternatives:
+        - alternative: &i_latin
+            canonical: i
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+          probability: 0.1
+
+
+cross_streets:
+    i: *i
+    i_latin: *i_latin
+    at: &na
+        canonical: на
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2        
+    na_latin: &na_latin
+        canonical: na
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    corner: &ugao
+        canonical: угао
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    ugao_latin: &ugao_latin
+        canonical: ugao
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    na_uglu: &na_uglu
+        canonical: на углу
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    na_uglu_latin: &na_uglu_latin
+        canonical: na uglu
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *i
+        probability: 0.65
+        alternatives:
+            - alternative: *i_latin
+              probability: 0.05
+            - alternative: *na
+              probability: 0.075
+            - alternative: *na_latin
+              probability: 0.025
+            - alternative: *ugao
+              probability: 0.1
+            - alternative: *ugao_latin
+              probability: 0.05
+            - alternative: *na_uglu
+              probability: 0.025
+            - alternative: *na_uglu_latin
+              probability: 0.025
+    izmedu: &izmedu
+        canonical: између
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+    izmedu_latin: &izmedu_latin
+        canonical: između
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+    between:
+        default: *izmedu
+        probability: 0.9
+        alternatives:
+            - alternative: *izmedu_latin
+              probability: 0.1
+
+levels:
+    sprat: &sprat
+        canonical: спрат
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    sprat_latin: &sprat_latin
+        canonical: sprat
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    kat: &kat
+        canonical: кат
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    kat_latin: &kat_latin
+        canonical: kat
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    etaza: &etaza
+        canonical: етажа
+        abbreviated: ет
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    etaza_latin: &etaza_latin
+        canonical: etaža
+        abbreviated: et
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+    prizemlje: &prizemlje
+        canonical: приземље
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    prizemlje_latin: &prizemlje_latin
+        canonical: prizemlje
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    parter: &parter
+        canonical: партер
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+    parter_latin: &parter_latin
+        canonical: parter
+        sample: true
+        canonical_probability: 0.9
+        sample_probability: 0.1
+
+    podrum: &podrum
+        canonical: подрум
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        # e.g. подрум 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. 1. подрум
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+    podrum_latin: &podrum_latin
+        canonical: podrum
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        # e.g. подрум 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. 1. подрум
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+
+    aliases:
+        "<-1":
+            default: *podrum
+            probability: 0.8
+            alternatives:
+                - alternative: *podrum_latin
+                  probability: 0.2
+        "-1":
+            default: *podrum
+            probability: 0.8
+            alternatives:
+                - alternative: *podrum_latin
+                  probability: 0.2
+        "0":
+            default: *prizemlje
+            probability: 0.45
+            alternatives:
+                - alternative: *prizemlje_latin
+                  probability: 0.05
+                - alternative: *parter
+                  probability: 0.35
+                - alternative: *parter_latin
+                  probability: 0.05
+                - alternative: *sprat
+                  probability: 0.04
+                - alternative: *sprat_latin
+                  probability: 0.01
+                - alternative: *kat
+                  probability: 0.04
+                - alternative: *kat_latin
+                  probability: 0.01
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *sprat
+        probability: 0.65
+        alternatives:
+            - alternative: *sprat_latin
+              probability: 0.1
+            - alternative: *kat
+              probability: 0.15
+            - alternative: *kat_latin
+              probability: 0.05
+            - alternative: *etaza
+              probability: 0.04
+            - alternative: *etaza_latin
+              probability: 0.01
+        numeric_probability: 0.69 # With this probability, pick an integer
+        roman_numeral_probability: 0.3 # Pick a Roman numeral for the actual value
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+directions:
+    right: &desno
+        canonical: десно
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    desno_latin: &desno_latin
+        canonical: desno
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &levo
+        canonical: лево
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    levo_latin: &levo_latin
+        canonical: levo
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *desno
+          probability: 0.45
+        - alternative: *desno_latin
+          probability: 0.05
+        - alternative: *levo
+          probability: 0.45
+        - alternative: *levo_latin
+          probability: 0.05
+
+cardinal_directions:
+    east: &istok
+        canonical: исток
+        abbreviated: и
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: и
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    istok_latin: &istok_latin
+        canonical: istok
+        abbreviated: i
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: i
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &zapad
+        canonical: запад
+        abbreviated: з
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: з
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    zapad_latin: &zapad_latin
+        canonical: zapad
+        abbreviated: z
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: z
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &sever
+        canonical: север
+        abbreviated: с
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: с
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    sever_latin: &sever_latin
+        canonical: sever
+        abbreviated: s
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &jug
+        canonical: југ
+        abbreviated: ј
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: ј
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    jug_latin: &jug_latin
+        canonical: jug
+        abbreviated: j
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: j
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *sever
+          probability: 0.23
+        - alternative: *sever_latin
+          probability: 0.02
+        - alternative: *istok
+          probability: 0.23
+        - alternative: *istok_latin
+          probability: 0.02
+        - alternative: *jug
+          probability: 0.23
+        - alternative: *jug_latin
+          probability: 0.02
+        - alternative: *zapad
+          probability: 0.23
+        - alternative: *zapad_latin
+          probability: 0.02
+
+entrances:
+    ulaz: &ulaz
+        canonical: улаз
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    ulaz_latin: &ulaz_latin
+        canonical: ulaz
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Ulaz 1, Ulaz A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *ulaz
+        probability: 0.8
+        alternatives:
+            - alternative: *ulaz_latin
+              probability: 0.2
+        numeric_probability: 0.1 # e.g. Ulaz 1
+        alpha_probability: 0.85 # e.g. Ulaz A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+
+
+staircases:
+    stepeniste: &stepeniste
+        canonical: степениште
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    stepeniste_latin: &stepeniste_latin
+        canonical: stepenište
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+
+    alphanumeric: &staircase_alphanumeric
+        default: *stepeniste
+        probability: 0.8
+        alternatives:
+            - alternative: *stepeniste_latin
+              probability: 0.2
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: right
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *desno
+                      probability: 0.19
+                    - alternative: *desno_latin
+                      probability: 0.01
+                    - alternative: *levo
+                      probability: 0.19
+                    - alternative: *levo_latin
+                      probability: 0.01
+                    - alternative: *sever
+                      probability: 0.14
+                    - alternative: *sever_latin
+                      probability: 0.01
+                    - alternative: *jug
+                      probability: 0.14
+                    - alternative: *jug_latin
+                      probability: 0.01
+                    - alternative: *istok
+                      probability: 0.14
+                    - alternative: *istok_latin
+                      probability: 0.01
+                    - alternative: *zapad
+                      probability: 0.14
+                    - alternative: *zapad_latin
+                      probability: 0.01
+
+po_boxes:
+    postanski_fah: &postanski_fah
+        canonical: поштански фах
+        abbreviated: пф
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # poštanski fah br. 1234
+    postanski_fah_latin: &postanski_fah_latin
+        canonical: poštanski fah
+        abbreviated: pf
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # poštanski fah br. 1234
+    postanski_pretinac: &postanski_pretinac
+        canonical: поштански претинац
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.5
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    postanski_pretinac_latin: &postanski_pretinac_latin
+        canonical: poštanski pretinac
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    postanski_pregradak: &postanski_pregradak
+        canonical: поштански преградак
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.5
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+    postanski_pregradak_latin: &postanski_pregradak_latin
+        canonical: poštanski pregradak
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+
+    alphanumeric:
+        default: *postanski_fah
+        probability: 0.7
+        alternatives:
+            - alternative: *postanski_fah_latin
+              probability: 0.05
+            - alternative: *postanski_pretinac
+              probability: 0.1
+            - alternative: *postanski_pretinac_latin
+              probability: 0.05
+            - alternative: *postanski_pregradak
+              probability: 0.075
+            - alternative: *postanski_pregradak_latin
+              probability: 0.025
+        numeric_probability: 0.9 # pf 123
+        alpha_probability: 0.05 # pf A
+        numeric_plus_alpha_probability: 0.04 # pf 123G
+        alpha_plus_numeric_probability: 0.01 # pf A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+units:
+    stan: &stan
+        canonical: стан
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    stan_latin: &stan_latin
+        canonical: stan
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    apartman: &apartman
+        canonical: апартман
+        abbreviated: ап
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.2
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    apartman_latin: &apartman_latin
+        canonical: apartman
+        abbreviated: ap
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.2
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    soba: &soba
+        canonical: соба
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    soba_latin: &soba_latin
+        canonical: soba
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    kancelarija: &kancelarija
+        canonical: канцеларија
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    kancelarija_latin: &kancelarija_latin
+        canonical: kancelarija
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    alphanumeric: &unit_alphanumeric
+        default: *stan
+        probability: 0.5
+        alternatives:
+            - alternative: *stan_latin
+              probability: 0.1
+            - alternative: *apartman
+              probability: 0.2
+            - alternative: *apartman_latin
+              probability: 0.05
+            - alternative: *soba
+              probability: 0.1
+            - alternative: *soba_latin
+              probability: 0.05
+        numeric_probability: 0.9 # e.g. stan. 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. stan A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.01
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *soba
+            probability: 0.55
+            alternatives:
+                - alternative: *soba_latin
+                  probability: 0.05
+                - alternative: *kancelarija
+                  probability: 0.35
+                - alternative: *kancelarija_latin
+                  probability: 0.05
+            numeric_probability: 0.95 # e.g. soba 1
+            numeric_plus_alpha_probability: 0.01 # e.g. soba 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. soba A1
+            alpha_probability: 0.03 # e.g. soba A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university:
+            default: *soba
+            probability: 0.9
+            alternatives:
+                - alternative: *soba_latin
+                  probability: 0.1
+            numeric_probability: 0.95 # e.g. soba 1
+            numeric_plus_alpha_probability: 0.01 # e.g. soba 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. soba A1
+            alpha_probability: 0.03 # e.g. soba A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/sv.yaml

@@ -0,0 +1,795 @@
+# sv.yaml
+# -------
+# Swedish language specification.
+
+components:
+    level:
+        null_probability: 0.85
+        alphanumeric_probability: 0.1
+        standalone_probability: 0.05
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.75
+        alphanumeric_probability: 0.25
+
+numbers:
+    default: &nummer
+        canonical: nummer
+        abbreviated: nr
+        sample: true
+        # Probabilities
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        sample_exclude:
+            - "#"
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "#"
+            direction: left
+
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+
+house_numbers:
+    alphanumeric:
+        default: *nummer
+
+    alphanumeric_phrase_probability: 0.0001
+
+
+and:
+    default: &och
+        canonical: och
+        abbreviated: "&"
+        canonical_probability: 0.2
+        abbreviated_probability: 0.75
+        sample: true
+        sample_probability: 0.05
+
+cross_streets:
+    and: *och
+    corner_of: &hornet_av
+        canonical: hörnet av
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    at_the_corner_of: &i_hornet_av
+        canonical: i hörnet av
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *och
+        probability: 0.7
+        alternatives:
+            - alternative: *hornet_av
+              probability: 0.15
+            - alternative: *i_hornet_av
+              probability: 0.15
+
+    between:
+        canonical: mellan
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+
+
+levels:
+    vaningen: &vaningen
+        canonical: våningen
+        abbreviated: vån
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.3
+        sample_probability: 0.2
+        ordinal:
+            direction: right
+        numeric_probability: 0.0
+        ordinal_probability: 1.0
+    vaning: &vaning
+        canonical: våning
+        abbreviated: vån
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.3
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        ordinal:
+            direction: left
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        numeric_probability: 0.8
+        ordinal_probability: 0.2
+    plan: &plan
+        canonical: plan
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    entreplan: &entreplan
+        canonical: entréplan
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    trappa_upp: &trappa_upp
+        canonical: trappa upp
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        number_min_abs_value: 2
+        number_max_abs_value: 2
+        number_subtract_abs_value: 1
+        numeric_probability: 0.8
+        ordinal_probability: 0.2
+    trappor_upp: &trappor_upp
+        canonical: trappor upp
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        number_min_abs_value: 3
+        number_subtract_abs_value: 1
+        numeric_probability: 0.8
+        ordinal_probability: 0.2
+    trappa: &trappa
+        canonical: trappa
+        abbreviated: tr
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.6
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        number_min_abs_value: 2
+        number_max_abs_value: 2
+        number_subtract_abs_value: 1
+        numeric_probability: 0.8
+        ordinal_probability: 0.2
+    trappor: &trappor
+        canonical: trappor
+        abbreviated: tr
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.6
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.8
+                spellout_probability: 0.2
+        number_min_abs_value: 3
+        number_subtract_abs_value: 1
+        numeric_probability: 0.8
+        ordinal_probability: 0.2
+    bottenvaning: &bottenvaning
+        canonical: bottenvåning
+        abbreviated: bv
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+    vindsvaningen: &vindsvaningen
+        canonical: vindsvåningen
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        standalone_probability: 1.0
+    vinds: &vinds
+        canonical: vinds
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        standalone_probability: 1.0
+    kallare: &kallare
+        canonical: källare
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        # e.g. 1 källare
+        numeric:
+            direction: right
+            direction_probability: 0.8
+        # e.g. k1
+        numeric_affix:
+            affix: k
+            direction: left
+        # e.g. 1:a k
+        ordinal:
+            direction: right
+        standalone_probability: 0.9
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        numeric_affix_probability: 0.09
+        ordinal_probability: 0.005
+    aliases:
+        "<-1":
+            default: *kallare
+            probability: 0.95
+            alternatives:
+                - alternative: *vaning
+                  probability: 0.025
+                - alternative: *vaningen
+                  probability: 0.025
+        "-1":
+            default: *kallare
+            probability: 0.9
+            alternatives:
+                - alternative: *vaning
+                  probability: 0.05
+                - alternative: *vaningen
+                  probability: 0.05
+        "0":
+            default: *bottenvaning
+            probability: 0.6
+            alternatives:
+                - alternative: *entreplan
+                  probability: 0.2
+                - alternative: *vaningen
+                  probability: 0.1
+                - alternative: *vaning
+                  probability: 0.1
+        "1":
+            default: *bottenvaning
+            probability: 0.6
+            alternatives:
+                - alternative: *entreplan
+                  probability: 0.2
+                - alternative: *vaningen
+                  probability: 0.1
+                - alternative: *vaning
+                  probability: 0.1
+        "top":
+            default: *vaningen
+            probability: 0.35
+            alternatives:
+                - alternative: *vaning
+                  probability: 0.35
+                - alternative: *trappor_upp
+                  probability: 0.1
+                - alternative: *trappor
+                  probability: 0.1
+                - alternative: *vindsvaningen
+                  probability: 0.05
+                - alternative: *vinds
+                  probability: 0.05
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *vaningen
+        probability: 0.25
+        alternatives:
+            - alternative: *vaning
+              probability: 0.2
+            - alternative: *plan
+              probability: 0.05
+            - alternative: *trappa_upp
+              probability: 0.125
+            - alternative: *trappa
+              probability: 0.125
+            - alternative: *trappor_upp
+              probability: 0.125
+            - alternative: *trappor
+              probability: 0.125
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+
+categories:
+    near:
+        default:
+            canonical: i närheten av
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.8
+        alternatives:
+            - alternative:
+                canonical: nära
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+    nearby:
+        default:
+            canonical: i närheten
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.4
+        alternatives:
+            - alternative:
+                canonical: runt här
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+            - alternative:
+                canonical: nära här
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: nära här
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative:
+                canonical: nära
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+            - alternative: 
+                canonical: omkring här
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.1
+    near_me:
+        default:
+            canonical: nära mig
+            sample: true
+            canonical_probability: 0.8
+            sample_probability: 0.2
+        probability: 0.8
+        alternatives:
+            - alternative:
+                canonical: i närheten av mig
+                sample: true
+                canonical_probability: 0.8
+                sample_probability: 0.2
+              probability: 0.2
+
+    in:
+        default:
+            canonical: i
+        probability: 0.8
+        alternatives:
+            - alternative:
+                canonical: på
+              probability: 0.2
+
+
+    # Probabilities of each phrase
+    near_probability: 0.35
+    nearby_probability: 0.2
+    near_me_probability: 0.1
+    in_probability: 0.35
+
+
+directions:
+    right: &hoger
+        canonical: höger
+        sample: true
+        canonical_probability: 0.1
+        sample_probability: 0.9
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: h
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    left: &vanster
+        canonical: vänster
+        sample: true
+        canonical_probability: 0.1
+        sample_probability: 0.9
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+            whitespace_probability: 0.1
+        numeric_probability: 0.8
+        numeric_affix_probability: 0.2
+    alternatives:
+        - alternative: *hoger
+          probability: 0.5
+        - alternative: *vanster
+          probability: 0.5
+
+cardinal_directions:
+    east: &ost
+        canonical: öst
+        abbreviated: ö
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: ö
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    eastern: &ostra
+        canonical: östra
+        abbreviated: ö:a
+        canonical_probability: 0.9
+        abbreviated_probability: 0.1
+        numeric:
+            direction: right
+
+    west: &vast
+        canonical: väst
+        abbreviated: v
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: v
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    western: &vastra
+        canonical: västra
+        abbreviated: v:a
+        canonical_probability: 0.9
+        abbreviated_probability: 0.1
+        numeric:
+            direction: right
+
+    north: &norr
+        canonical: norr
+        abbreviated: n
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: n
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    northern: &norra
+        canonical: norra
+        abbreviated: n:a
+        canonical_probability: 0.9
+        abbreviated_probability: 0.1
+
+    south: &sod
+        canonical: söd
+        abbreviated: s
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: s
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    southern: &sodra
+        canonical: södra
+        abbreviated: s:a
+        canonical_probability: 0.9
+        abbreviated_probability: 0.1
+
+    alternatives:
+        - alternative: *norr
+          probability: 0.25
+        - alternative: *ost
+          probability: 0.25
+        - alternative: *sod
+          probability: 0.25
+        - alternative: *vast
+          probability: 0.25
+
+entrances:
+    ingang: &ingang
+        canonical: ingång
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    entre: &entre
+        canonical: entré
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # Eingang 1, Eingang A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *ingang
+        probability: 0.6
+        alternatives:
+            - alternative: *entre
+              probability: 0.4
+        numeric_probability: 0.1 # e.g. Eingang 1
+        alpha_probability: 0.85 # e.g. Eingang A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+staircases:
+    uppgang: &uppgang
+        canonical: uppgång
+        abbreviated: u
+        sample: true
+        canonical_probability: 0.7
+        abbreviated_probability: 0.2
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    uppgang_hoger: &uppgang_hoger
+        canonical: uppgång höger
+        abbreviated: uh
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.5
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    uppgang_vanster: &uppgang_vanster
+        canonical: uppgång vänster
+        abbreviated: uv
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.5
+        sample_probability: 0.1
+        numeric:
+            direction: left
+    alphanumeric: &staircase_alphanumeric
+        default: *uppgang
+        probability: 0.6
+        alternatives:
+            - alternative: *uppgang_hoger
+              probability: 0.2
+            - alternative: *uppgang_vanster
+              probability: 0.2
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: left
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *norr
+                    - alternative: *sod
+                    - alternative: *ost
+                    - alternative: *vast
+
+po_boxes:
+    box: &box
+        canonical: box
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # Box No 1234
+    postlada: &postlada
+        canonical: postlåda
+        abbreviated: pl
+        sample: true
+        canonical_probability: 0.6
+        abbreviated_probability: 0.2
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2 # Pl No 1234
+
+    alphanumeric:
+        sample: false
+        default: *box
+        probability: 0.9
+        alternatives:
+            - alternative: *postlada
+              probability: 0.1
+        numeric_probability: 0.9 # Box 123
+        alpha_probability: 0.05 # Box A
+        numeric_plus_alpha_probability: 0.04 # Box 123G
+        alpha_plus_numeric_probability: 0.01 # Box A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.1
+        - length: 5
+          probability: 0.5
+        - length: 6
+          probability: 0.05
+
+
+
+units:
+    lagenhet: &lagenhet
+        canonical: lägenhet
+        abbreviated: lgh
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.5
+        sample_probability: 0.3
+        numeric:
+            direction: left
+            null_phrase_probability: 0.1
+            # Lejlighed nummer 4
+            add_number_phrase: true
+            add_number_phrase_probability: 0.05
+    bostad: &bostad
+        canonical: bostad
+        abbreviated: bst
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.5
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.05
+    lagenhetsnummer: &lagenhetsnummer
+        canonical: lägenhetsnummer
+        abbreviated: lgh nr
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    hus: &hus
+        canonical: hus
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+    rum: &rum
+        canonical: rum
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        numeric:
+            direction: left
+    alphanumeric: &unit_alphanumeric
+        default: *lagenhet
+        probability: 0.75
+        alternatives:
+            - alternative: *lagenhetsnummer
+              probability: 0.05
+            - alternative: *hus
+              probability: 0.1
+            - alternative: *rum
+              probability: 0.1
+        numeric_probability: 0.95 # e.g. Lägenhet 1
+        alpha_probability: 0.05 # e.g. Lgh A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # Separate random probability for adding directions like 2H, 2V, etc.
+        add_direction: true
+        add_direction_probability: 0.005
+
+        # Add directions for plain numbers
+        add_direction_numeric: true
+        # Add direction only e.g. Lejlighed Igjen
+        add_direction_standalone: true
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.2
+
+        # Use the actual floor phrase as long as the whole phrase is numeric
+        # Has the effect of creating Bolignummer-style units
+        use_floor_affix_unit_num_digits: 2
+
+        # In Swedish addresses, the ground level is 10, floors are 11, 12, ... basements are 9, 8, ...
+        use_floor_ground_starts_at: 10
+        # For single digit floors, use 09, 08, etc.
+        use_floor_floor_num_digits: 2
+
+
+countries:
+    # Swedish addresses in Finland
+    fi:
+        units:
+            alphanumeric:
+                default: *bostad
+                probability: 1.0
+                alternatives: []
+
+                add_direction: false
+                add_direction_numeric: false
+                add_direction_standalone: false
+
+                use_floor_probability: 0.1
+
+                use_floor_affix_unit_num_digits: 0
+
+                use_floor_ground_starts_at: 1
+                use_floor_floor_num_digits: 2

resources/addresses/tr.yaml

@@ -0,0 +1,503 @@
+# tr.yaml
+# -------
+# Turkish language specification
+
+components:
+    level:
+        null_probability: 0.9
+        alphanumeric_probability: 0.1
+
+    staircase:
+        null_probability: 0.99
+        alphanumeric_probability: 0.01
+
+    entrance:
+        null_probability: 0.999
+        alphanumeric_probability: 0.001
+
+    unit:
+        null_probability: 0.7
+        alphanumeric_probability: 0.3
+
+    combinations:
+        -
+            components:
+                - house_number
+                - staircase
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+        -
+            components:
+                - house_number
+                - level
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.1
+        # For unit types like 2/34
+        -
+            components:
+                - house_number
+                - unit
+            label: house_number
+            separators:
+                - separator: "/"
+                  probability: 0.95
+                - separator: "-"
+                  probability: 0.05
+            probability: 0.005
+
+
+numbers:
+
+    default: &numara
+        canonical: numara
+        abbreviated: "no:"
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.6
+        sample_probability: 0.1
+        numeric:
+            direction: left
+        numeric_affix:
+            affix: "no:"
+            whitespace_probability: 0.6
+            direction: left
+        numeric_probability: 0.4
+        numeric_affix_probability: 0.6
+
+    alphanumeric_phrase_probability: 0.05
+    no_number_probability: 0.05
+
+
+and:
+    default: &ve
+        canonical: ve
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+
+
+cross_streets:
+    ve: *ve
+    corner_of: &kose
+        canonical: köşe
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    kosesinde: &kosesinde
+        canonical: köşesinde
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+    intersection:
+        default: *ve
+        probability: 0.8
+        alternatives:
+            - alternative: *kose
+              probability: 0.1
+            - alternative: *kosesinde
+              probability: 0.1
+
+    arasinda: &arasinda
+        canonical: arasında
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        parentheses_probability: 0.5
+    between:
+        default: *arasinda
+
+levels:
+    kat: &kat
+        canonical: kat
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            direction_probability: 0.9
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.3
+                roman_numeral_probability: 0.7
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+        numeric_probability: 0.4
+        ordinal_probability: 0.6
+
+    zemin_kat: &zemin_kat
+        canonical: zemin kat
+        abbreviated: zk
+        sample: true
+        canonical_probability: 0.3
+        abbreviated_probability: 0.4
+        sample_probability: 0.3
+    asma_kat: &asma_kat
+        canonical: asma kat
+        half_floors: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        sample: true
+        # e.g. asma kat 2
+        numeric:
+            direction: left
+        # e.g. 2. asma kat
+        ordinal:
+            direction: right
+        numeric_probability: 0.1
+        ordinal_probability: 0.2
+        standalone_probability: 0.6
+    bodrum: &bodrum
+        canonical: bodrum
+        sample: true
+        canonical_probability: 0.7
+        sample_probability: 0.3
+        # e.g. bodrum 1
+        numeric:
+            direction: left
+            direction_probability: 0.8
+        # e.g. 1. bodrum
+        ordinal:
+            direction: right
+            digits:
+                ascii_probability: 0.7
+                roman_numeral_probability: 0.3
+        standalone_probability: 0.99
+        number_abs_value: true
+        number_min_abs_value: 1
+        numeric_probability: 0.005
+        ordinal_probability: 0.005
+
+    aliases:
+        "<-1":
+            default: *bodrum
+        "-1":
+            default: *bodrum
+        # Special token for half-floors
+        half_floors:
+            default: *asma_kat
+        "0":
+            default: *zemin_kat
+            probability: 0.9
+            alternatives:
+                - alternative: *kat
+                  probability: 0.1
+
+    numbering_starts_at: 0
+
+    alphanumeric:
+        default: *kat
+        numeric_probability: 0.99 # With this probability, pick an integer
+        alpha_probability: 0.0098 # With this probability, pick a letter e.g. A
+        numeric_plus_alpha_probability: 0.0001 # e.g. 2A
+        alpha_plus_numeric_probability: 0.0001 # e.g. A2
+
+
+directions:
+    right: &sag
+        canonical: sağ
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    left: &sol
+        canonical: sol
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: right
+    alternatives:
+        - alternative: *sag
+          probability: 0.5
+        - alternative: *sol
+          probability: 0.5
+
+cardinal_directions:
+    east: &dogu
+        canonical: doğu
+        abbreviated: d
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: d
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    west: &bati
+        canonical: batı
+        abbreviated: b
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: b
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    north: &kuzey
+        canonical: kuzey
+        abbreviated: k
+        canonical_probability: 0.95
+        abbreviated_probability: 0.05
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: k
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    south: &guney
+        canonical: güney
+        abbreviated: g
+        sample: true
+        canonical_probability: 0.75
+        abbreviated_probability: 0.1
+        sample_probability: 0.15
+        numeric:
+            direction: right
+        numeric_affix:
+            affix: g
+            direction: right
+        numeric_probability: 0.5
+        numeric_affix_probability: 0.5
+
+    alternatives:
+        - alternative: *kuzey
+          probability: 0.25
+        - alternative: *dogu
+          probability: 0.23
+        - alternative: *guney
+          probability: 0.23
+        - alternative: *bati
+          probability: 0.23
+
+entrances:
+    giris: &giris
+        canonical: giriş
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+    # giriş 1, giriş A, etc.
+    alphanumeric: &entrance_alphanumeric
+        default: *giris
+        numeric_probability: 0.1 # e.g. giriş 1
+        alpha_probability: 0.85 # e.g. giriş A
+        numeric_plus_alpha_probability: 0.025 # e.g. 1A
+        alpha_plus_numeric_probability: 0.025 # e.g. A1
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+
+staircases:
+    merdiven: &merdiven
+        canonical: merdiven
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+
+
+    alphanumeric: &staircase_alphanumeric
+        default: *merdiven
+        numeric_probability: 0.75
+        alpha_probability: 0.2
+        numeric_plus_alpha_probability: 0.025
+        alpha_plus_numeric_probability: 0.025
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        directional:
+            direction: right
+            direction_probability: 0.85
+            modifier:
+                alternatives:
+                    - alternative: *sag
+                      probability: 0.2
+                    - alternative: *sol
+                      probability: 0.2
+                    - alternative: *kuzey
+                      probability: 0.15
+                    - alternative: *guney
+                      probability: 0.15
+                    - alternative: *dogu
+                      probability: 0.15
+                    - alternative: *bati
+                      probability: 0.15
+
+po_boxes:
+    posta_kutusu: &posta_kutusu
+        canonical: posta kutusu
+        abbreviated: pk
+        sample: true
+        canonical_probability: 0.2
+        abbreviated_probability: 0.4
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.2
+
+    alphanumeric:
+        default: *posta_kutusu
+        numeric_probability: 0.9 # pp 123
+        alpha_probability: 0.05 # p.p A
+        numeric_plus_alpha_probability: 0.04 # pp 123G
+        alpha_plus_numeric_probability: 0.01 # pp A123
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+
+units:
+    daire: &daire
+        canonical: daire
+        abbreviated: d
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.4
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    apartman: &apartman
+        canonical: apartman
+        abbreviated: apt
+        sample: true
+        canonical_probability: 0.4
+        abbreviated_probability: 0.2
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    oda: &oda
+        canonical: oda
+        sample: true
+        canonical_probability: 0.8
+        sample_probability: 0.2
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+    ofis: &ofis
+        canonical: ofis
+        sample: true
+        canonical_probability: 0.6
+        sample_probability: 0.4
+        numeric:
+            direction: left
+            add_number_phrase: true
+            add_number_phrase_probability: 0.1
+
+    alphanumeric: &unit_alphanumeric
+        default: *daire
+        probability: 0.6
+        alternatives:
+            - alternative: *apartman
+              probability: 0.3
+            - alternative: *oda
+              probability: 0.1
+        numeric_probability: 0.9 # e.g. d. 1
+        numeric_plus_alpha_probability: 0.03 # e.g. 1A
+        alpha_plus_numeric_probability: 0.03 # e.g. A1
+        alpha_probability: 0.04 # e.g. daire A
+
+        alpha_plus_numeric:
+            whitespace_probability: 0.1
+        numeric_plus_alpha:
+            whitespace_probability: 0.1
+
+        # If there are 10 floors, create unit numbers like #301 or #1032
+        use_floor_probability: 0.05
+
+    zones:
+        commercial: &commercial_unit_types
+            default: *oda
+            probability: 0.6
+            alternatives:
+                - alternative: *ofis
+                  probability: 0.4
+            numeric_probability: 0.95 # e.g. oda 1
+            numeric_plus_alpha_probability: 0.01 # e.g. oda 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. oda A1
+            alpha_probability: 0.03 # e.g. oda A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1
+        university:
+            default: *oda
+            numeric_probability: 0.95 # e.g. oda 1
+            numeric_plus_alpha_probability: 0.01 # e.g. oda 1A
+            alpha_plus_numeric_probability: 0.01 # e.g. oda A1
+            alpha_probability: 0.03 # e.g. oda A
+            alpha_plus_numeric:
+                whitespace_probability: 0.1
+            numeric_plus_alpha:
+                whitespace_probability: 0.1

resources/addresses/zh.yaml

@@ -0,0 +1,292 @@
+# zh.yaml
+# -------
+# Chinese language specification (default is mainland China, Hong Kong below)
+
+whitespace: false
+
+components:
+    level:
+        null_probability: 0.85 # Probability of doing nothing if no floor number is specified
+        alphanumeric_probability: 0.15
+
+    unit:
+        # If no unit number is specified
+        null_probability: 0.6
+        alphanumeric_probability: 0.4
+
+numbers:
+    default: &hao
+        canonical: 号
+        numeric_affix:
+            affix: 号
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    probability: 0.8
+    alternatives:
+        - alternative: &hao_traditional
+            canonical: 號
+            numeric_affix:
+                affix: 號
+                direction: right
+            numeric_probability: 0.0
+            numeric_affix_probability: 1.0
+          probability: 0.2
+
+house_numbers:
+    alphanumeric:
+        default: *hao
+        probability: 0.8
+        alternatives:
+            - alternative: *hao_traditional
+              probability: 0.2
+    alphanumeric_phrase_probability: 0.6
+
+levels:
+    lou: &lou
+        canonical: 楼
+        numeric_affix:
+            affix: 楼
+            direction: right
+            add_number_phrase: true
+            add_number_phrase_probability: 0.5
+            digits:
+                ascii_probability: 0.6
+                unicode_full_width_probability: 0.1
+                spellout_probability: 0.3
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    lou_traditional: &lou_traditional
+        canonical: 樓
+        numeric_affix:
+            affix: 樓
+            direction: right
+            add_number_phrase: true
+            add_number_phrase_probability: 0.5
+            digits:
+                ascii_probability: 0.6
+                unicode_full_width_probability: 0.1
+                spellout_probability: 0.3
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    ceng: &ceng
+        canonical: 层
+        numeric_affix:
+            affix: 层
+            direction: right
+            add_number_phrase: true
+            add_number_phrase_probability: 0.5
+            digits:
+                ascii_probability: 0.6
+                unicode_full_width_probability: 0.1
+                spellout_probability: 0.3
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    ceng_traditional: &ceng_traditional
+        canonical: 層
+        numeric_affix:
+            affix: 層
+            direction: right
+            add_number_phrase: true
+            add_number_phrase_probability: 0.5
+            digits:
+                ascii_probability: 0.6
+                unicode_full_width_probability: 0.1
+                spellout_probability: 0.3
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *lou
+        probability: 0.85
+        alternatives:
+            - alternative: *lou_traditional
+              probability: 0.05
+            - alternative: *ceng
+              probability: 0.08
+            - alternative: *ceng_traditional
+              probability: 0.02
+        numeric_probability: 1.0
+
+po_boxes:
+    youzheng_xinxiang: &youzheng_xinxiang
+        canonical: 邮政信箱
+        numeric_affix:
+            affix: 邮政信箱
+            direction: left
+            digits:
+                ascii_probability: 0.3
+                unicode_full_width_probability: 0.5
+                spellout_probability: 0.2
+            use_number_phrase: true
+            use_number_phrase_probability: 0.8
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    youzheng_xinxiang_traditional: &youzheng_xinxiang_traditional
+        canonical: 郵政信箱
+        numeric_affix:
+            affix: 郵政信箱
+            direction: left
+            digits:
+                ascii_probability: 0.3
+                unicode_full_width_probability: 0.5
+                spellout_probability: 0.2
+            use_number_phrase: true
+            use_number_phrase_probability: 0.8
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+
+    alphanumeric:
+        default: *youzheng_xinxiang
+        probability: 0.9
+        alternatives:
+            - alternative: *youzheng_xinxiang_traditional
+              probability: 0.1
+        numeric_probability: 1.0
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+postcodes:
+    alphanumeric:
+        default: &youbian
+            canonical: 邮编
+            numeric_affix:
+                affix: 邮编
+                direction: left
+            # null_probability means the chance of doing nothing e.g. just the postal code
+            null_probability: 0.9
+            numeric_probability: 0.0
+            numeric_affix_probability: 0.1
+        probability: 0.9
+        alternatives:
+            - alternative: &youbian_traditional
+                canonical: 郵編
+                numeric_affix:
+                    affix: 郵編
+                    direction: left
+                # null_probability means the chance of doing nothing e.g. just the postal code
+                null_probability: 0.9
+                numeric_probability: 0.0
+                numeric_affix_probability: 0.1
+              probability: 0.1
+
+units:
+    shi: &shi
+        canonical: 室
+        numeric_affix:
+            affix: 室
+            direction: right
+            add_number_phrase: true
+            add_number_phrase_probability: 0.5
+            digits:
+                ascii_probability: 0.6
+                unicode_full_width_probability: 0.1
+                spellout_probability: 0.3
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    alphanumeric:
+        default: *shi
+        numeric_probability: 1.0
+        use_positive_numbers_probability: 1.0
+        # If we have a floor number (from building:levels), use it
+        use_floor_probability: 0.8
+
+
+countries:
+    # Hong Kong
+    hk:
+        components:
+            # Floor number a little more common in Hong Kong than mainland China
+            level:
+                null_probability: 0.75
+                alphanumeric_probability: 0.25
+
+        numbers: &numbers_prefer_traditional
+            default: *hao_traditional
+            probability: 0.7
+            alternatives:
+                - alternative: *hao
+                  probability: 0.3
+
+        house_numbers: &house_number_prefer_traditional
+            alphanumeric:
+                default: *hao_traditional
+                probability: 0.7
+                alternatives:
+                    - alternative: *hao
+                      probability: 0.3
+            alphanumeric_phrase_probability: 0.6
+
+        levels: &levels_prefer_traditional
+            alphanumeric:
+                default: *lou_traditional
+                probability: 0.75
+                alternatives:
+                    - alternative: *lou
+                      probability: 0.15
+                    - alternative: *ceng_traditional
+                      probability: 0.06
+                    - alternative: *ceng
+                      probability: 0.04
+                numeric_probability: 1.0
+
+        po_boxes: &po_boxes_prefer_traditional
+            alphanumeric:
+                default: *youzheng_xinxiang_traditional
+                probability: 0.75
+                alternatives:
+                    - alternative: *youzheng_xinxiang
+                      probability: 0.25
+                numeric_probability: 1.0
+
+
+        postcodes: &postcodes_prefer_traditional
+            alphanumeric:
+                default: *youbian_traditional
+                probability: 0.75
+                alternatives:
+                    - alternative: *youbian
+                      probability: 0.25
+
+    # Macau
+    mo:
+        numbers: *numbers_prefer_traditional
+        house_numbers: *house_number_prefer_traditional
+        levels: *levels_prefer_traditional
+        po_boxes: *po_boxes_prefer_traditional
+        postcodes: *postcodes_prefer_traditional
+
+        units:
+            alphanumeric_probability:
+                numeric_probability: 0.9
+                alpha_probability: 0.1
+
+
+    # Taiwan
+    tw:
+        numbers: *numbers_prefer_traditional
+        house_numbers: *house_number_prefer_traditional
+        levels: *levels_prefer_traditional
+        po_boxes: *po_boxes_prefer_traditional
+        postcodes: *postcodes_prefer_traditional
+
+        units:
+            alphanumeric_probability:
+                numeric_probability: 0.9
+                alpha_probability: 0.1

resources/addresses/zh_pinyin.yaml

@@ -0,0 +1,153 @@
+# zh_pinyin.yaml
+# --------------
+# Chinese (Pinyin)
+
+whitespace: false
+
+components:
+    level:
+        null_probability: 0.85 # Probability of doing nothing if no floor number is specified
+        alphanumeric_probability: 0.15
+
+    unit:
+        # If no unit number is specified
+        null_probability: 0.6
+        alphanumeric_probability: 0.4
+
+numbers:
+    default: &hao
+        canonical: hao
+        numeric_affix:
+            affix: -hao
+            upper_case: false
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+house_numbers:
+    alphanumeric:
+        default: *hao
+    alphanumeric_phrase_probability: 0.6
+
+levels:
+    lou: &lou
+        canonical: lóu
+        numeric_affix:
+            affix: -lóu
+            upper_case: false
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    lou_no_accent: &lou_no_accent
+        canonical: lou
+        numeric_affix:
+            affix: -lou
+            upper_case: false
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    ceng: &ceng
+        canonical: céng
+        numeric_affix:
+            affix: -céng
+            upper_case: false
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+    ceng_no_accent: &ceng_no_accent
+        canonical: ceng
+        numeric_affix:
+            affix: -ceng
+            upper_case: false
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    numbering_starts_at: 1
+
+    alphanumeric:
+        default: *lou
+        probability: 0.85
+        alternatives:
+            - alternative: *lou_no_accent
+              probability: 0.05
+            - alternative: *ceng
+              probability: 0.08
+            - alternative: *ceng_no_accent
+              probability: 0.02
+        numeric_probability: 1.0
+
+po_boxes:
+    youzheng_xinxiang: &youzheng_xinxiang
+        canonical: youzheng xinxiang
+        numeric:
+            direction: left
+        numeric_probability: 1.0
+
+    alphanumeric:
+        default: *youzheng_xinxiang
+        numeric_probability: 1.0
+
+    digits:
+        - length: 1
+          probability: 0.05
+        - length: 2
+          probability: 0.1
+        - length: 3
+          probability: 0.2
+        - length: 4
+          probability: 0.5
+        - length: 5
+          probability: 0.1
+        - length: 6
+          probability: 0.05
+
+postcodes:
+    alphanumeric:
+        default: &youbian
+            canonical: yóubiān
+            numeric:
+                direction: left
+            # null_probability means the chance of doing nothing e.g. just the postal code
+            null_probability: 0.9
+            numeric_probability: 0.1
+        probability: 0.9
+        alternatives:
+            - alternative: &youbian_no_accent
+                canonical: youbian
+                numeric:
+                    direction: left
+                # null_probability means the chance of doing nothing e.g. just the postal code
+                null_probability: 0.9
+                numeric_probability: 0.1
+              probability: 0.1
+
+units:
+    shi: &shi
+        canonical: shì
+        numeric_affix:
+            affix: -shì
+            upper_case: false
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    shi_no_accent: &shi_no_accent
+        canonical: shi
+        numeric_affix:
+            affix: -shi
+            upper_case: false
+            direction: right
+        numeric_probability: 0.0
+        numeric_affix_probability: 1.0
+
+    alphanumeric:
+        default: *shi
+        probability: 0.8
+        alternatives:
+            - alternative: *shi_no_accent
+              probability: 0.2
+        numeric_probability: 1.0
+        use_positive_numbers_probability: 1.0
+        # If we have a floor number (from building:levels), use it
+        use_floor_probability: 0.8

resources/boundaries/geonames/ad.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: state

resources/boundaries/geonames/ar.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/at.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    # admin2 is a mix of state_district and city, need to list specifically

resources/boundaries/geonames/au.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    # admin2 is a mix of state_district and city, need to list specifically

resources/boundaries/geonames/ax.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: city

resources/boundaries/geonames/bd.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    # unclear what admin2 is, maybe city

resources/boundaries/geonames/be.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/bg.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: city

resources/boundaries/geonames/br.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: city

resources/boundaries/geonames/ca.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/ch.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/cz.yaml

@@ -0,0 +1,4 @@
+admin_codes:
+    # The GeoNames admin1 boundaries are admin_level=5 or 6 in OSM
+    # However, they do appear to be states, might need to update Czech OSM config
+    admin1: state_district

resources/boundaries/geonames/de.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/dk.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    # admin2 is a mix of city and island, need to list specifically

resources/boundaries/geonames/do.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: city

resources/boundaries/geonames/dz.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/es.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/fi.yaml

@@ -0,0 +1,5 @@
+admin_codes:
+    # The GeoNames admin1 boundaries are admin_level=6 in OSM
+    # However, they do appear to be states, might need to update Finnish OSM config
+    admin1: state_district
+    admin2: state_district

resources/boundaries/geonames/fo.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: city

resources/boundaries/geonames/fr.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/gb.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/gt.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/gu.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: city

resources/boundaries/geonames/hr.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    # admin2 is a mix of city and city_district, need to list specifically

resources/boundaries/geonames/hu.yaml

@@ -0,0 +1,4 @@
+admin_codes:
+    # The GeoNames admin1 boundaries are admin_level=6 in OSM
+    # However, they do appear to be states, might need to update Hungary OSM config
+    admin1: state_district

resources/boundaries/geonames/ie.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/im.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: city

resources/boundaries/geonames/in.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/is.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: city

resources/boundaries/geonames/it.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/je.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: state

resources/boundaries/geonames/jp.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    # admin2 is a mix of state_district and city, need to list specifically

resources/boundaries/geonames/li.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: city

resources/boundaries/geonames/lk.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/lt.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    # admin2 is a mix of state_district and city, need to list specifically

resources/boundaries/geonames/lu.yaml

@@ -0,0 +1,4 @@
+admin_codes:
+    # The admin1 names don't appear to exist in OSM, but would be states otherwise
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/md.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: state_district

resources/boundaries/geonames/mp.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: state_district

resources/boundaries/geonames/mt.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: city

resources/boundaries/geonames/mx.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/my.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/nl.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: city

resources/boundaries/geonames/no.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: city

resources/boundaries/geonames/nz.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/ph.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: country_region
+    # admin2 is a mix of state_district and city, need to list specifically

resources/boundaries/geonames/pk.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/pl.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/pr.yaml

@@ -0,0 +1,6 @@
+admin_codes:
+    admin1: state_district
+    # The notion of a "barrio" in the official sense in PR is not quite a
+    # municipality, and has no current official purpose, but might be useful
+    # to have the name + "barrio" version available in libpostal
+    admin2: city

resources/boundaries/geonames/pt.yaml

@@ -0,0 +1,8 @@
+admin_codes:
+    admin1: state_district
+    admin2: city
+
+overrides:
+    id:
+        "2593105": "state" # Madeira
+        "3411865": "state" # Azores

resources/boundaries/geonames/ro.yaml

@@ -0,0 +1,4 @@
+admin_codes:
+    admin1: state
+    # These are mostly admin_level=6, which maps to city in OSM
+    admin2: city

resources/boundaries/geonames/ru.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/se.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: city

resources/boundaries/geonames/si.yaml

@@ -0,0 +1,22 @@
+admin_codes:
+    admin1: city
+
+overrides:
+    id:
+        # Districts of Ljubljana (suburbs in OSM)
+        "3196350": "suburb" # Opština Ljubljana-Vič-Rudnik
+        "3196352": "suburb" # Opština [historical] Ljubljana-Šiška
+        "3196355": "suburb" # Opština Ljubljana-Moste-Polje
+        "3196356": "suburb" # Opština Ljubljana-Center
+        "3196357": "suburb" # Opčina Ljubljana-Bežigrad
+        "9794374": "suburb" # Črnuče District
+        "9794375": "suburb" # Dravlje District
+        "9794376": "suburb" # Golovec District
+        "9794377": "suburb" # Jarše District
+        "9794378": "suburb" # Posavje District
+        "9794379": "suburb" # Rožnik District
+        "9794380": "suburb" # Sostro District
+        "9794381": "suburb" # Šentvid District
+        "9794382": "suburb" # Šmarna Gora District
+        "9794384": "suburb" # Trnovo District
+        "9794386": "suburb" # Vič District

resources/boundaries/geonames/sk.yaml

@@ -0,0 +1,17 @@
+admin_codes:
+    admin1: state
+    # admin2 is a mix of state_district and city, need to list specifically
+    admin2: state_district
+overrides:
+    id:
+        # Districts of Bratislava
+        "8986283": "city_district" # Okres Bratislava I
+        "8986339": "city_district" # Okres Bratislava II
+        "8986340": "city_district" # Okres Bratislava III
+        "8986341": "city_district" # Okres Bratislava IV
+        "8986342": "city_district" # Okres Bratislava V
+        # Districts of Košice
+        "8986335": "city_district" # Košice I
+        "8986336": "city_district" # Košice II
+        "8986337": "city_district" # Košice III
+        "8986338": "city_district" # Košice IV

resources/boundaries/geonames/sm.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: city

resources/boundaries/geonames/th.yaml

@@ -0,0 +1,8 @@
+admin_codes:
+    admin1: state
+
+overrides:
+    id:
+        # Bangkok the state is treated as a city
+        # Note: we do this in OSM to get the boundary, so duplicate in GeoNames
+        "1609348": "city"

resources/boundaries/geonames/tr.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/geonames/us.yaml

@@ -0,0 +1,16 @@
+admin_codes:
+    admin1: state
+    admin2: state_district
+
+overrides:
+    id:
+        # Manhattan (Island)
+        "8479493": "city_district"
+        # Brooklyn
+        "5110300": "city_district"
+        # Bronx
+        "5110266": "city_district"
+        # Queens
+        "5133266": "city_district"
+        # Staten Island
+        "5139568": "city_district"

resources/boundaries/geonames/yt.yaml

@@ -0,0 +1,2 @@
+admin_codes:
+    admin1: city

resources/boundaries/geonames/za.yaml

@@ -0,0 +1,3 @@
+admin_codes:
+    admin1: state
+    admin2: state_district

resources/boundaries/names/global.yaml

@@ -0,0 +1,120 @@
+names:
+    keys:
+        default: name
+        probability: 0.75
+        alternatives:
+            - alternative: short_name # e.g. NYC
+              probability: 0.12
+            - alternative: alt_name # e.g. New York (instead of New York City)
+              probability: 0.12
+            - alternative: official_name # e.g. United Kingdom of Great Britain and Northern Ireland
+              probability: 0.01
+
+    components:
+        country:
+            keys:
+                default: name
+                probability: 0.87
+                alternatives:
+                    - alternative: ISO3166-1:alpha2
+                      probability: 0.02
+                    - alternative: ISO3166-1:alpha3
+                      probability: 0.01
+                    - alternative: short_name
+                      probability: 0.04
+                    - alternative: alt_name
+                      probability: 0.04
+                    - alternative: int_name
+                      probability: 0.01
+                    - alternative: official_name # e.g. United Kingdom of Great Britain and Northern Ireland
+                      probability: 0.01
+
+    regex_replacements:
+        - country: fr
+          pattern: "(?:lyon|paris|marseilles?) ([\\d]+(?:e|er|ème|eme) arrondissement)"
+          replace_with_group: 1
+          replace_probability: 0.5
+
+    prefixes:
+        language:
+            ru:
+                city:
+                    default:
+                        prefix: г.
+                    probability: 0.35
+                    alternatives:
+                        - alternative:
+                              prefix: г
+                          probability: 0.1
+                        - alternative:
+                              prefix: город
+                          probability: 0.05
+
+    # This section overrides place names
+    exceptions:
+        # Boroughs of New York City
+        - id: 2552485 # New York County (don't use Manhattan)
+          type: relation
+          default: name # New York County
+          probability: 1.0
+        - id: 369518 # Kings County (don't use Brooklyn)
+          type: relation
+          default: name # Kings County
+          probability: 1.0
+        - id: 369519 # Queens County (don't use Queens)
+          type: relation
+          default: name # Queens County
+          probability: 1.0
+        - id: 2552450 # Bronx County (don't use The Bronx)
+          type: relation
+          default: name # Bronx County
+          probability: 1.0
+        - id: 962876 # Richmond County (don't use Staten Island)
+          type: relation
+          default: name # Richmond County
+          probability: 1.0
+        - id: 6577227 # Kingston Parish (always use Kingston)
+          type: relation
+          default: name # Kingston
+          probability: 1.0
+        - id: 30674098 # Sao Paulo
+          type: node
+          default: name
+          probability: 0.9
+          alternatives:
+              - alternative: alt_name
+                probability: 0.09
+              - alternative: official_name
+                probability: 0.01
+        - id: 298285 # Sao Paulo (relation)
+          type: relation
+          default: name
+          probability: 0.9
+          alternatives:
+              - alternative: alt_name
+                probability: 0.09
+              - alternative: official_name
+                probability: 0.01
+        - id: 556706 # New Zealand
+          type: relation
+          default: name:en
+          probability: 0.77
+          alternatives:
+              - alternative: name
+                probability: 0.1
+              - alternative: ISO3166-1:alpha2
+                probability: 0.02
+              - alternative: ISO3166-1:alpha3
+                probability: 0.01
+              - alternative: short_name
+                probability: 0.04
+              - alternative: alt_name
+                probability: 0.04
+              - alternative: int_name
+                probability: 0.01
+              - alternative: official_name
+                probability: 0.01
+        - id: 2383266 # Melbourne (city center)
+          type: relation
+          default: alt_name # Melbourne
+          probability: 1.0

resources/boundaries/names/languages/de.yaml

@@ -0,0 +1,11 @@
+# Prefixes which can be stripped to normalize a place name
+prefixes:
+    - stadtteil
+    - stadtbezirk
+    - gemeinde
+    - landkreis
+    - kreis
+    - grenze
+    - freistaat
+    - regierungsbezirk
+    - gemeindefreies gebiet