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

[osm] use containing ids in component mapping

Browse Source
This commit is contained in:
Al
2016-07-24 21:57:04 -04:00
parent 2a9185874a
commit bfb89adaab
1 changed files with 17 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
scripts/geodata/addresses/components.py
View File

@@ -223,16 +223,18 @@ class AddressComponents(object):
def categorized_osm_components(self, country, osm_components): def categorized_osm_components(self, country, osm_components):
components = defaultdict(list) components = defaultdict(list)
for props in osm_components: for i, props in enumerate(osm_components):
name = props.get('name') name = props.get('name')
if not name: if not name:
continue continue
for k, v in props.iteritems(): containing_ids = [(c['type'], c['id']) for c in osm_components[i + 1:]]
normalized_key = osm_address_components.get_component(country, k, v)
if normalized_key: component = osm_address_components.component_from_properties(country, props, containing=containing_ids)
components[normalized_key].append(props)
break if component is not None:
components[component].append(props)
return components return components
@classmethod @classmethod
@@ -308,20 +310,22 @@ class AddressComponents(object):
names = set() names = set()
components = defaultdict(set) components = defaultdict(set)
for props in osm_components:
for i, props in enumerate(osm_components):
component_names = set(self.all_names(props, languages or [])) component_names = set(self.all_names(props, languages or []))
names |= component_names names |= component_names
is_state = False is_state = False
for k, v in six.iteritems(props):
normalized_key = osm_address_components.get_component(country, k, v) containing_ids = [(c['type'], c['id']) for c in osm_components[i + 1:]]
if not normalized_key:
continue component = osm_address_components.component_from_properties(country, props, containing=containing_ids)
if component is not None:
for cn in component_names: for cn in component_names:
components[cn.lower()].add(normalized_key) components[cn.lower()].add(normalized_key)
if normalized_key == AddressFormatter.STATE and not is_state: if not is_state:
is_state = True is_state = component == AddressFormatter.STATE
if is_state: if is_state:
for state in component_names: for state in component_names: