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

[dictionaries] Adding new dictionary types to generator script

Browse Source
This commit is contained in:
Al
2016-05-28 17:16:43 -04:00
parent 7aa06c4535
commit 9561f771ce
1 changed files with 5 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
scripts/geodata/address_expansions/address_dictionaries.py
View File

@@ -68,6 +68,7 @@ gazetteer_types = {
'building_types': 'DICTIONARY_BUILDING_TYPE', 'building_types': 'DICTIONARY_BUILDING_TYPE',
'categories': 'DICTIONARY_CATEGORY', 'categories': 'DICTIONARY_CATEGORY',
'categories_plural': 'DICTIONARY_CATEGORY_PLURAL', 'categories_plural': 'DICTIONARY_CATEGORY_PLURAL',
'chains': 'DICTIONARY_CHAIN',
'company_types': 'DICTIONARY_COMPANY_TYPE', 'company_types': 'DICTIONARY_COMPANY_TYPE',
'concatenated_prefixes_separable': 'DICTIONARY_CONCATENATED_PREFIX_SEPARABLE', 'concatenated_prefixes_separable': 'DICTIONARY_CONCATENATED_PREFIX_SEPARABLE',
'concatenated_suffixes_inseparable': 'DICTIONARY_CONCATENATED_SUFFIX_INSEPARABLE', 'concatenated_suffixes_inseparable': 'DICTIONARY_CONCATENATED_SUFFIX_INSEPARABLE',
@@ -92,9 +93,9 @@ gazetteer_types = {
'personal_titles': 'DICTIONARY_PERSONAL_TITLE', 'personal_titles': 'DICTIONARY_PERSONAL_TITLE',
'place_names': 'DICTIONARY_PLACE_NAME', 'place_names': 'DICTIONARY_PLACE_NAME',
'post_office': 'DICTIONARY_POST_OFFICE', 'post_office': 'DICTIONARY_POST_OFFICE',
'postcodes': 'DICTIONARY_POSTCODE', 'postcodes': 'DICTIONARY_POSTAL_CODE',
'qualifiers': 'DICTIONARY_QUALIFIER', 'qualifiers': 'DICTIONARY_QUALIFIER',
'staircase': 'DICTIONARY_STAIRCASE', 'staircases': 'DICTIONARY_STAIRCASE',
'stopwords': 'DICTIONARY_STOPWORD', 'stopwords': 'DICTIONARY_STOPWORD',
'street_types': 'DICTIONARY_STREET_TYPE', 'street_types': 'DICTIONARY_STREET_TYPE',
'surnames': 'DICTIONARY_SURNAME', 'surnames': 'DICTIONARY_SURNAME',
@@ -151,8 +152,8 @@ class AddressPhraseDictionaries(object):
raise InvalidAddressFileException(u'Invalid extension for file {}/{}, must be .txt'.format(language_dir, filename)) raise InvalidAddressFileException(u'Invalid extension for file {}/{}, must be .txt'.format(language_dir, filename))
dictionary_name = filename.split('.')[0].lower() dictionary_name = filename.split('.')[0].lower()
'''if dictionary_name not in gazetteer_types: if dictionary_name not in gazetteer_types:
raise InvalidAddressFileException(u'Invalid filename for file {}/{}. Must be one of {{{}}}'.format(language_dir, filename, ', '.join(sorted(gazetteer_types))))''' raise InvalidAddressFileException(u'Invalid filename for file {}/{}. Must be one of {{{}}}'.format(language_dir, filename, ', '.join(sorted(gazetteer_types))))
self.language_dictionaries[language].append(dictionary_name) self.language_dictionaries[language].append(dictionary_name)
path = os.path.join(language_dir, filename) path = os.path.join(language_dir, filename)