[osm] adding admin1 ids to the OSM country rtree

scripts/geodata/i18n/languages.py

@@ -22,6 +22,8 @@ road_language_overrides = defaultdict(OrderedDict)
 languages = set()
 all_languages = languages
 
+osm_admin1_ids = set()
+
 languages_initialized = False
 
 
@@ -50,6 +52,8 @@ def init_languages(languages_dir=LANGUAGES_DIR):
     path = os.path.join(languages_dir, 'regional', 'adm1.tsv')
 
     for country, key, value, langs, default in unicode_csv_reader(open(path), delimiter='\t'):
+        if key == 'osm':
+            osm_admin1_ids.add(tuple(value.split(':')))
         for lang in langs.split(','):
             regional_languages[(country, key, value)][lang] = int(default)
             if lang not in country_languages[country]:

scripts/geodata/osm/admin_boundaries.py

@@ -14,8 +14,10 @@ from collections import defaultdict, OrderedDict
 from itertools import izip, combinations
 
 from geodata.coordinates.conversion import latlon_to_decimal
+from geodata.encoding import safe_encode, safe_decode
 from geodata.file_utils import ensure_dir
 from geodata.graph.scc import strongly_connected_components
+from geodata.i18n.languages import osm_admin1_ids
 from geodata.math.floats import isclose
 from geodata.osm.extract import *
 
@@ -312,7 +314,7 @@ class OSMBuildingPolygonReader(OSMPolygonReader):
 
 class OSMCountryPolygonReader(OSMPolygonReader):
     def include_polygon(self, props):
-        return 'ISO3166-1:alpha2' in props
+        return 'ISO3166-1:alpha2' in props or (props.get('type', 'relation'), safe_encode(props.get('id', ''))) in osm_admin1_ids
 
 
 class OSMPostalCodesPolygonReader(OSMPolygonReader):