tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[addresses] Fixes for English/French Canadian apartment numbers

Browse Source
This commit is contained in:
Al
2016-06-01 11:43:42 -04:00
parent 7d5d54bd29
commit fdba7b138d
2 changed files with 76 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
resources/addresses/en.yaml
View File

@@ -1357,16 +1357,19 @@ countries:
components: components:
combinations: combinations:
house_number_unit: house_number_unit:
probability: 0.0
unit_house_number:
components: components:
- unit - unit
- house_number - house_number
label: house_number
separators: separators:
- separator: / - separator: /
probability: 0.1 probability: 0.04
- separator: "-" - separator: "-"
probability: 0.8 probability: 0.95
- separator: " - " - separator: " - "
probability: 0.1 probability: 0.01
probability: 0.1 probability: 0.1
levels: levels:
# Note: Canadian English uses "storey" keeping with the British convention, so no need to change that # Note: Canadian English uses "storey" keeping with the British convention, so no need to change that

116
resources/addresses/fr.yaml
View File

@@ -4,9 +4,8 @@
components: components:
level: level:
# If no floor number is specified # If no floor number is specified
null_probability: 0.6 null_probability: 0.8
alphanumeric_probability: 0.35 alphanumeric_probability: 0.2
standalone_probability: 0.05
staircase: staircase:
null_probability: 0.99 null_probability: 0.99
@@ -18,12 +17,10 @@ components:
unit: unit:
# If no unit number is specified # If no unit number is specified
null_probability: 0.3 null_probability: 0.8
alphanumeric_probability: 0.65 alphanumeric_probability: 0.2
standalone_probability: 0.05
combinations: combinations:
# For unit types like 2/34 (more common in Canada and Australia)
house_number_unit: house_number_unit:
components: components:
- house_number - house_number
@@ -92,9 +89,8 @@ levels:
sample_probability: 0.2 sample_probability: 0.2
numeric: numeric:
direction: left direction: left
direction_probability: 0.8
add_number_phrase: true add_number_phrase: true
add_number_phrase_probability: 0.2 add_number_phrase_probability: 0.05
ordinal: ordinal:
direction: right direction: right
numeric_probability: 0.75 numeric_probability: 0.75
@@ -106,9 +102,8 @@ levels:
sample_probability: 0.1 sample_probability: 0.1
numeric: numeric:
direction: left direction: left
direction_probability: 0.8
add_number_phrase: true add_number_phrase: true
add_number_phrase_probability: 0.2 add_number_phrase_probability: 0.05
ordinal: ordinal:
direction: right direction: right
numeric_probability: 0.75 numeric_probability: 0.75
@@ -419,7 +414,7 @@ entrances:
directional: directional:
modifier: modifier:
direction: right # e.g. Entrance Nord direction: right # e.g. Entrance Nord
direction_probability: 0.9 direction_probability: 0.95
alternatives: alternatives:
- alternative: *nord - alternative: *nord
- alternative: *sud - alternative: *sud
@@ -456,8 +451,8 @@ staircases:
whitespace_probability: 0.1 whitespace_probability: 0.1
directional: directional:
direction: right # e.g. Escalera Izq direction: right # e.g. Escalier Izq
direction_probability: 0.8 direction_probability: 0.9
modifier: modifier:
alternatives: alternatives:
- alternative: *nord - alternative: *nord
@@ -637,6 +632,27 @@ units:
probability: 0.095 probability: 0.095
- alternative: *chambre - alternative: *chambre
probability: 0.005 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: zones:
residential: *unit_alphanumeric residential: *unit_alphanumeric
@@ -763,37 +779,6 @@ categories:
- alternative: - alternative:
canonical: a proximite de la canonical: a proximite de la
probability: 0.05 probability: 0.05
alternatives:
- alternative:
canonical: próximo
probability: 0.05
- alternative:
canonical: proximo
probability: 0.05
- alternative:
canonical: cerca de aquí
probability: 0.05
- alternative:
canonical: cerca de aqui
probability: 0.05
- alternative:
canonical: acá
probability: 0.05
- alternative:
canonical: aca
probability: 0.05
- alternative:
canonical: cerca de acá
probability: 0.05
- alternative:
canonical: cerca de aca
probability: 0.05
- alternative:
canonical: por aquí
probability: 0.05
- alternative:
canonical: por aqui
probability: 0.05
near_me: near_me:
default: default:
canonical: proche de chez moi canonical: proche de chez moi
@@ -834,11 +819,50 @@ categories:
countries: countries:
ca: ca:
components:
unit:
null_probability: 0.6
alphanumeric_probability: 0.4
combinations:
house_number_unit:
probability: 0.0
unit_house_number:
components:
- unit
- house_number
label: house_number
separators:
- separator: /
probability: 0.04
- separator: "-"
probability: 0.95
- separator: " - "
probability: 0.01
probability: 0.1
levels: levels:
numbering_starts_at: 1 numbering_starts_at: 1
aliases: aliases:
"1": "1":
default: *rez_de_chaussee # 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
be: be:
units: units:
alphanumeric: alphanumeric: