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

[gazetteers] Adding new gazetteer types/address components

Browse Source
This commit is contained in:
Al
2016-05-28 19:19:18 -04:00
parent 95b239a5f9
commit e800f21f06
3 changed files with 52 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
src/gazetteer_data.c
View File

@@ -3,29 +3,45 @@ gazetteer_t gazetteer_config[] = {
{DICTIONARY_ACADEMIC_DEGREE, ADDRESS_NAME},
{DICTIONARY_AMBIGUOUS_EXPANSION, ADDRESS_NONE},
{DICTIONARY_BUILDING_TYPE, ADDRESS_NAME | ADDRESS_HOUSE_NUMBER | ADDRESS_STREET | ADDRESS_UNIT},
{DICTIONARY_CATEGORY, ADDRESS_CATEGORY},
{DICTIONARY_CHAIN, ADDRESS_NAME},
{DICTIONARY_COMPANY_TYPE, ADDRESS_NAME},
{DICTIONARY_CONCATENATED_PREFIX_SEPARABLE, ADDRESS_STREET},
{DICTIONARY_CONCATENATED_SUFFIX_INSEPARABLE, ADDRESS_STREET},
{DICTIONARY_CONCATENATED_SUFFIX_SEPARABLE, ADDRESS_STREET},
{DICTIONARY_CROSS_STREET, ADDRESS_STREET},
{DICTIONARY_DIRECTIONAL, ADDRESS_ANY},
{DICTIONARY_ELISION, ADDRESS_ANY},
{DICTIONARY_ENTRANCE, ADDRESS_ENTRANCE},
{DICTIONARY_GIVEN_NAME, ADDRESS_STREET | ADDRESS_NAME},
{DICTIONARY_LEVEL, ADDRESS_HOUSE_NUMBER | ADDRESS_STREET | ADDRESS_UNIT},
{DICTIONARY_NO_ADDRESS, ADDRESS_HOUSE_NUMBER | ADDRESS_STREET},
{DICTIONARY_HOUSE_NUMBER, ADDRESS_HOUSE_NUMBER},
{DICTIONARY_LEVEL_NUMBERED, ADDRESS_LEVEL},
{DICTIONARY_LEVEL_STANDALONE, ADDRESS_LEVEL},
{DICTIONARY_LEVEL_MEZZANINE, ADDRESS_LEVEL},
{DICTIONARY_LEVEL_BASEMENT, ADDRESS_LEVEL},
{DICTIONARY_LEVEL_SUB_BASEMENT, ADDRESS_LEVEL},
{DICTIONARY_NULL, ADDRESS_ANY},
{DICTIONARY_NAMED_ORGANIZATION, ADDRESS_NAME},
{DICTIONARY_NAMED_PERSON, ADDRESS_NAME | ADDRESS_STREET},
{DICTIONARY_NO_NUMBER, ADDRESS_HOUSE_NUMBER},
{DICTIONARY_NUMBER, ADDRESS_HOUSE_NUMBER | ADDRESS_UNIT | ADDRESS_LEVEL | ADDRESS_STAIRCASE | ADDRESS_ENTRANCE},
{DICTIONARY_PERSONAL_SUFFIX, ADDRESS_NAME | ADDRESS_STREET},
{DICTIONARY_PERSONAL_TITLE, ADDRESS_NAME | ADDRESS_STREET},
{DICTIONARY_PLACE_NAME, ADDRESS_NAME | ADDRESS_STREET},
{DICTIONARY_POST_OFFICE, ADDRESS_HOUSE_NUMBER | ADDRESS_STREET},
{DICTIONARY_POSTAL_CODE, ADDRESS_POSTAL_CODE},
{DICTIONARY_QUALIFIER, ADDRESS_STREET},
{DICTIONARY_STAIRCASE, ADDRESS_STAIRCASE},
{DICTIONARY_STOPWORD, ADDRESS_ANY},
{DICTIONARY_STREET_TYPE, ADDRESS_STREET},
{DICTIONARY_SURNAME, ADDRESS_STREET | ADDRESS_NAME},
{DICTIONARY_SYNONYM, ADDRESS_ANY},
{DICTIONARY_TOPONYM, ADDRESS_NAME | ADDRESS_STREET | ADDRESS_LOCALITY | ADDRESS_ADMIN1 | ADDRESS_ADMIN2 | ADDRESS_ADMIN3 | ADDRESS_ADMIN4 | ADDRESS_ADMIN_OTHER | ADDRESS_NEIGHBORHOOD},
{DICTIONARY_UNIT, ADDRESS_NAME | ADDRESS_HOUSE_NUMBER | ADDRESS_STREET | ADDRESS_UNIT}
{DICTIONARY_TOPONYM, ADDRESS_NAME | ADDRESS_STREET | ADDRESS_TOPONYM},
{DICTIONARY_UNIT_NUMBERED, ADDRESS_UNIT},
{DICTIONARY_UNIT_STANDALONE, ADDRESS_UNIT},
{DICTIONARY_UNIT_DIRECTION, ADDRESS_UNIT}
};
#define NUM_DICTIONARY_TYPES sizeof(gazetteer_config) / sizeof(gazetteer_t)