404 lines
11 KiB
YAML
404 lines
11 KiB
YAML
# 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
|