392 lines
10 KiB
YAML
392 lines
10 KiB
YAML
# 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
|