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

[fix] refactoring geonames script a bit

Browse Source
This commit is contained in:
Al
2016-08-11 21:31:39 -04:00
parent 48755ec218
commit 22123b80ba
2 changed files with 25 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
resources/boundaries/osm/gb.yaml
View File

@@ -1,8 +1,8 @@
--- ---
admin_level: admin_level:
"2": "country" "2": "country"
"4": "country_region" "4": "state"
"5": "state" "5": "state_district"
"6": "state_district" "6": "state_district"
"8": "city" "8": "city"
"10": "suburb" "10": "suburb"

49
scripts/geodata/geonames/create_geonames_tsv.py
View File

@@ -141,38 +141,32 @@ geonames_fields = [
GeonamesField('a4.geonames_id as a4_gn_id', 'GEONAMES_ADMIN4_ID'), GeonamesField('a4.geonames_id as a4_gn_id', 'GEONAMES_ADMIN4_ID'),
] ]
DUMMY_BOUNDARY_TYPE_INDEX = [i for i, f in enumerate(geonames_fields) def geonames_field_index(s):
if f.c_constant == 'GEONAMES_BOUNDARY_TYPE'][0] for i, f in enumerate(geonames_fields):
if f.c_constant == s:
return i
return None
DUMMY_HAS_WIKIPEDIA_ENTRY_INDEX = [i for i, f in enumerate(geonames_fields)
if f.c_constant == 'GEONAMES_HAS_WIKIPEDIA_ENTRY'][0]
GEONAMES_ID_INDEX = [i for i, f in enumerate(geonames_fields) DUMMY_BOUNDARY_TYPE_INDEX = geonames_field_index('GEONAMES_BOUNDARY_TYPE')
if f.c_constant == 'GEONAMES_ID'][0] DUMMY_HAS_WIKIPEDIA_ENTRY_INDEX = geonames_field_index('GEONAMES_HAS_WIKIPEDIA_ENTRY')
LANGUAGE_INDEX = [i for i, f in enumerate(geonames_fields) GEONAMES_ID_INDEX = geonames_field_index('GEONAMES_ID')
if f.c_constant == 'GEONAMES_ISO_LANGUAGE'][0] LANGUAGE_INDEX = geonames_field_index('GEONAMES_ISO_LANGUAGE')
DUMMY_LANGUAGE_PRIORITY_INDEX = [i for i, f in enumerate(geonames_fields) DUMMY_LANGUAGE_PRIORITY_INDEX = geonames_field_index('GEONAMES_LANGUAGE_PRIORITY')
if f.c_constant == 'GEONAMES_LANGUAGE_PRIORITY'][0]
CANONICAL_NAME_INDEX = [i for i, f in enumerate(geonames_fields) CANONICAL_NAME_INDEX = geonames_field_index('GEONAMES_CANONICAL')
if f.c_constant == 'GEONAMES_CANONICAL'][0]
NAME_INDEX = [i for i, f in enumerate(geonames_fields) NAME_INDEX = geonames_field_index('GEONAMES_NAME')
if f.c_constant == 'GEONAMES_NAME'][0]
COUNTRY_CODE_INDEX = [i for i, f in enumerate(geonames_fields) COUNTRY_CODE_INDEX = geonames_field_index('GEONAMES_COUNTRY_CODE')
if f.c_constant == 'GEONAMES_COUNTRY_CODE'][0]
POPULATION_INDEX = [i for i, f in enumerate(geonames_fields) POPULATION_INDEX = geonames_field_index('GEONAMES_POPULATION')
if f.c_constant == 'GEONAMES_POPULATION'][0]
PREFERRED_INDEX = [i for i, f in enumerate(geonames_fields) PREFERRED_INDEX = geonames_field_index('GEONAMES_IS_PREFERRED_NAME')
if f.c_constant == 'GEONAMES_IS_PREFERRED_NAME'][0]
HISTORICAL_INDEX = [i for i, f in enumerate(geonames_fields) HISTORICAL_INDEX = geonames_field_index('GEONAMES_IS_HISTORICAL')
if f.c_constant == 'GEONAMES_IS_HISTORICAL'][0]
geonames_admin_joins = ''' geonames_admin_joins = '''
@@ -241,11 +235,14 @@ postal_code_fields = [
GeonamesField('group_concat(distinct a3.geonames_id) admin3_ids', 'GN_POSTAL_ADMIN3_IDS'), GeonamesField('group_concat(distinct a3.geonames_id) admin3_ids', 'GN_POSTAL_ADMIN3_IDS'),
] ]
POSTAL_CODE_INDEX = [i for i, f in enumerate(postal_code_fields) def postal_code_field_index(s):
if f.c_constant == 'GN_POSTAL_CODE'][0] for i, f in enumerate(postal_code_fields):
if f.c_constant == s:
return i
return None
POSTAL_CODE_POP_INDEX = [i for i, f in enumerate(postal_code_fields) POSTAL_CODE_INDEX = postal_code_field_index('GN_POSTAL_CODE')
if f.c_constant == 'GN_POSTAL_COUNTRY_POPULATION'][0] POSTAL_CODE_POP_INDEX = postal_code_field_index('GN_POSTAL_COUNTRY_POPULATION')
postal_codes_query = ''' postal_codes_query = '''
select select