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

[languages] Use English formats for Romanized CJK

Browse Source
This commit is contained in:
Al
2016-05-24 12:13:58 -04:00
parent ad4b197ead
commit 890268aa87
2 changed files with 11 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
resources/parser/default.yaml
View File

@@ -7,6 +7,11 @@ languages:
# Replace user-tagged admin components with the non-local language version # Replace user-tagged admin components with the non-local language version
replace_non_local_probability: 0.4 replace_non_local_probability: 0.4
language_code_replacements:
ja_rm: en
ko_rm: en
zh_pinyin: en
# Dependencies for including each component in an "address" # Dependencies for including each component in an "address"
# Two-way dependencies are not an issue # Two-way dependencies are not an issue
component_dependencies: component_dependencies:

6
scripts/geodata/addresses/components.py
View File

@@ -136,6 +136,8 @@ class AddressComponents(object):
# Non-admin component dropout # Non-admin component dropout
self.address_level_dropout_probabilities = {k: v['probability'] for k, v in six.iteritems(self.config['dropout'])} self.address_level_dropout_probabilities = {k: v['probability'] for k, v in six.iteritems(self.config['dropout'])}
self.language_code_replacements = nested_get(self.config, ('languages', 'language_code_replacements'))
self.osm_admin_rtree = osm_admin_rtree self.osm_admin_rtree = osm_admin_rtree
self.language_rtree = language_rtree self.language_rtree = language_rtree
self.neighborhoods_rtree = neighborhoods_rtree self.neighborhoods_rtree = neighborhoods_rtree
@@ -1105,6 +1107,10 @@ class AddressComponents(object):
self.drop_invalid_components(address_components) self.drop_invalid_components(address_components)
if language_suffix:
suffix_lang = language_suffix.lstrip(':')
language = self.language_code_replacements.get(suffix_lang, language)
return address_components, country, language return address_components, country, language
def limited(self, address_components, latitude, longitude): def limited(self, address_components, latitude, longitude):