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

[polygons] Adding Belgium regional languages

Browse Source
This commit is contained in:
Al
2015-07-17 00:53:25 -04:00
parent d5ac816066
commit 1d7247d7e1
2 changed files with 68 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
data/language/regional/adm1.tsv
View File

@@ -1,3 +1,42 @@
be name Liège fr 1
be gn_id 2792411 fr 1
be osm relation:1407192 de 1
be name Liège de 1
be gn_id 2792411 de 1
be osm relation:1407192 de 1
be name Brabant wallon fr 1
be gn_id 3333251 fr 1
be osm relation:78748 fr 1
be name Namur fr 1
be gn_id 2790469 fr 1
be osm relation:1311816 fr 1
be name Limburg nl 1
be gn_id 2792347 nl 1
be osm relation:53142 nl 1
be name West-Vlanderen nl 1
be gn_id 2783770 nl 1
be osm relation:416271 nl 1
be qs_a1r Brussels Hoofstedelijk Gewest#R\xe9gion de Bruxelles-Capitale fr 1
be gn_id 2800867 fr 1
be osm relation:54094 fr 1
be qs_a1r Brussels Hoofstedelijk Gewest#R\xe9gion de Bruxelles-Capitale nl 1
be gn_id 2800867 nl 1
be osm relation:54094 nl 1
be name Antwerpen nl 1
be gn_id 2803136 nl 1
be osm relation:53114 nl 1
be name Hainaut fr 1
be gn_id 2796741 fr 1
be osm relation:157559 fr 1
be name Luxembourg fr 1
be gn_id 2791993 fr 1
be osm relation:1412581 fr 1
be name Vlaams Brabant nl 1
be gn_id 3333250 nl 1
be osm relation:58004 nl 1
be name Oost-Vlanderen nl 1
be gn_id 2789733 nl 1
be osm relation:53135 nl 1
ca name Quebec fr 1 ca name Quebec fr 1
ca gn_id 6115047 fr 1 ca gn_id 6115047 fr 1
ca osm relation:61549 fr 1 ca osm relation:61549 fr 1
1 ca be name Quebec Liège fr 1
1 be name Liège fr 1
2 be gn_id 2792411 fr 1
3 be osm relation:1407192 de 1
4 be name Liège de 1
5 be gn_id 2792411 de 1
6 be osm relation:1407192 de 1
7 be name Brabant wallon fr 1
8 be gn_id 3333251 fr 1
9 be osm relation:78748 fr 1
10 be name Namur fr 1
11 be gn_id 2790469 fr 1
12 be osm relation:1311816 fr 1
13 be name Limburg nl 1
14 be gn_id 2792347 nl 1
15 be osm relation:53142 nl 1
16 be name West-Vlanderen nl 1
17 be gn_id 2783770 nl 1
18 be osm relation:416271 nl 1
19 be qs_a1r Brussels Hoofstedelijk Gewest#R\xe9gion de Bruxelles-Capitale fr 1
20 be gn_id 2800867 fr 1
21 be osm relation:54094 fr 1
22 be qs_a1r Brussels Hoofstedelijk Gewest#R\xe9gion de Bruxelles-Capitale nl 1
23 be gn_id 2800867 nl 1
24 be osm relation:54094 nl 1
25 be name Antwerpen nl 1
26 be gn_id 2803136 nl 1
27 be osm relation:53114 nl 1
28 be name Hainaut fr 1
29 be gn_id 2796741 fr 1
30 be osm relation:157559 fr 1
31 be name Luxembourg fr 1
32 be gn_id 2791993 fr 1
33 be osm relation:1412581 fr 1
34 be name Vlaams Brabant nl 1
35 be gn_id 3333250 nl 1
36 be osm relation:58004 nl 1
37 be name Oost-Vlanderen nl 1
38 be gn_id 2789733 nl 1
39 be osm relation:53135 nl 1
40 ca ca name Quebec Quebec fr 1
41 ca ca gn_id 6115047 6115047 fr 1
42 ca ca osm relation:61549 relation:61549 fr 1

59
src/gazetteers.h
View File

@@ -34,6 +34,7 @@ typedef enum dictionary_type {
DICTIONARY_SYNONYM = 2, DICTIONARY_SYNONYM = 2,
DICTIONARY_STOPWORD = 3, DICTIONARY_STOPWORD = 3,
DICTIONARY_ELISION = 4, DICTIONARY_ELISION = 4,
DICTIONARY_AMBIGUOUS_EXPANSION = 5,
DICTIONARY_STREET_NAME = 10, DICTIONARY_STREET_NAME = 10,
DICTIONARY_STREET_TYPE = 11, DICTIONARY_STREET_TYPE = 11,
@@ -84,38 +85,36 @@ typedef struct named_gazetteer {
} named_gazetteer_t; } named_gazetteer_t;
// Only need these for the in-memory dictionaries // Only need these for the in-memory dictionaries
named_gazetteer_t gazetteer_config[] = { gazetteer_t gazetteer_config[] = {
{"academic_degrees", {DICTIONARY_ACADEMIC_DEGREE, ADDRESS_NAME}}, {DICTIONARY_ACADEMIC_DEGREE, ADDRESS_NAME},
{"building_types", {DICTIONARY_BUILDING_TYPE, ADDRESS_NAME | ADDRESS_HOUSE_NUMBER | ADDRESS_STREET | ADDRESS_UNIT}}, {DICTIONARY_AMBIGUOUS_EXPANSION, ADDRESS_ANY},
{"company_types", {DICTIONARY_COMPANY_TYPE, ADDRESS_NAME}}, {DICTIONARY_BUILDING_TYPE, ADDRESS_NAME | ADDRESS_HOUSE_NUMBER | ADDRESS_STREET | ADDRESS_UNIT},
{"concatenated_prefixes_inseparable", {DICTIONARY_CONCATENATED_PREFIX_SEPARABLE, ADDRESS_STREET}}, {DICTIONARY_COMPANY_TYPE, ADDRESS_NAME},
{"concatenated_suffixes_inseparable", {DICTIONARY_CONCATENATED_SUFFIX_INSEPARABLE, ADDRESS_STREET}}, {DICTIONARY_CONCATENATED_PREFIX_SEPARABLE, ADDRESS_STREET},
{"concatenated_suffixes_separable", {DICTIONARY_CONCATENATED_SUFFIX_SEPARABLE, ADDRESS_STREET}}, {DICTIONARY_CONCATENATED_SUFFIX_INSEPARABLE, ADDRESS_STREET},
{"directionals", {DICTIONARY_DIRECTIONAL, ADDRESS_ANY}}, {DICTIONARY_CONCATENATED_SUFFIX_SEPARABLE, ADDRESS_STREET},
{"elisions", {DICTIONARY_ELISION, ADDRESS_ANY}}, {DICTIONARY_DIRECTIONAL, ADDRESS_ANY},
{"given_names", {DICTIONARY_GIVEN_NAME, ADDRESS_STREET | ADDRESS_NAME}}, {DICTIONARY_ELISION, ADDRESS_ANY},
{"level_types", {DICTIONARY_LEVEL, ADDRESS_HOUSE_NUMBER | ADDRESS_STREET | ADDRESS_UNIT}}, {DICTIONARY_GIVEN_NAME, ADDRESS_STREET | ADDRESS_NAME},
{"no_number", {DICTIONARY_NO_ADDRESS, ADDRESS_HOUSE_NUMBER | ADDRESS_STREET}}, {DICTIONARY_LEVEL, ADDRESS_HOUSE_NUMBER | ADDRESS_STREET | ADDRESS_UNIT},
{"nulls", {DICTIONARY_NULL, ADDRESS_ANY}}, {DICTIONARY_NO_ADDRESS, ADDRESS_HOUSE_NUMBER | ADDRESS_STREET},
{"organizations", {DICTIONARY_NAMED_ORGANIZATION, ADDRESS_NAME}}, {DICTIONARY_NULL, ADDRESS_ANY},
{"people", {DICTIONARY_NAMED_PERSON, ADDRESS_NAME | ADDRESS_STREET}}, {DICTIONARY_NAMED_ORGANIZATION, ADDRESS_NAME},
{"personal_suffixes", {DICTIONARY_PERSONAL_SUFFIX, ADDRESS_NAME | ADDRESS_STREET}}, {DICTIONARY_NAMED_PERSON, ADDRESS_NAME | ADDRESS_STREET},
{"personal_titles", {DICTIONARY_PERSONAL_TITLE, ADDRESS_NAME | ADDRESS_STREET}}, {DICTIONARY_PERSONAL_SUFFIX, ADDRESS_NAME | ADDRESS_STREET},
{"place_names", {DICTIONARY_PLACE_NAME, ADDRESS_NAME | ADDRESS_STREET}}, {DICTIONARY_PERSONAL_TITLE, ADDRESS_NAME | ADDRESS_STREET},
{"post_office", {DICTIONARY_POST_OFFICE, ADDRESS_HOUSE_NUMBER | ADDRESS_STREET}}, {DICTIONARY_PLACE_NAME, ADDRESS_NAME | ADDRESS_STREET},
{"qualifiers", {DICTIONARY_QUALIFIER, ADDRESS_STREET}}, {DICTIONARY_POST_OFFICE, ADDRESS_HOUSE_NUMBER | ADDRESS_STREET},
{"stopwords", {DICTIONARY_STOPWORD, ADDRESS_ANY}}, {DICTIONARY_QUALIFIER, ADDRESS_STREET},
{"street_types", {DICTIONARY_STREET_TYPE, ADDRESS_STREET}}, {DICTIONARY_STOPWORD, ADDRESS_ANY},
{"surnames", {DICTIONARY_SURNAME, ADDRESS_STREET | ADDRESS_NAME}}, {DICTIONARY_STREET_TYPE, ADDRESS_STREET},
{"synonyms", {DICTIONARY_SYNONYM, ADDRESS_ANY}}, {DICTIONARY_SURNAME, ADDRESS_STREET | ADDRESS_NAME},
{"toponyms", {DICTIONARY_TOPONYM, ADDRESS_LOCALITY | ADDRESS_ADMIN1 | ADDRESS_ADMIN2 | ADDRESS_ADMIN3 | ADDRESS_ADMIN4 | ADDRESS_ADMIN_OTHER | ADDRESS_NEIGHBORHOOD}}, {DICTIONARY_SYNONYM, ADDRESS_ANY},
{"unit_types", {DICTIONARY_UNIT, ADDRESS_NAME | ADDRESS_HOUSE_NUMBER | ADDRESS_STREET}} {DICTIONARY_TOPONYM, ADDRESS_LOCALITY | ADDRESS_ADMIN1 | ADDRESS_ADMIN2 | ADDRESS_ADMIN3 | ADDRESS_ADMIN4 | ADDRESS_ADMIN_OTHER | ADDRESS_NEIGHBORHOOD},
{DICTIONARY_UNIT, ADDRESS_NAME | ADDRESS_HOUSE_NUMBER | ADDRESS_STREET}
}; };
#define NUM_DICTIONARY_TYPES sizeof(gazetteer_config) / sizeof(gazetteer_t)
VECTOR_INIT(gazetteer_array, gazetteer_t)
#define NUM_DICTIONARY_TYPES sizeof(gazetteer_config) / sizeof(named_gazetteer_t)
#ifdef __cplusplus #ifdef __cplusplus
} }