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

[addresses] refactoring place dropout into its own method

Browse Source
This commit is contained in:
Al
2017-01-16 19:35:16 -05:00
parent 35dbce59d2
commit 024a6a40b1
1 changed files with 26 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

48
scripts/geodata/addresses/components.py
View File

@@ -1669,6 +1669,31 @@ class AddressComponents(object):
else: else:
return None return None
def dropout_places(self, address_components, osm_components, country, language, population=None, population_from_city=False):
# Population of the city helps us determine if the city can be used
# on its own like "Seattle" or "New York" vs. smaller cities like
# have to be qualified with a state, country, etc.
unambiguous_city = False
if population is None and population_from_city:
population = 0
tagged = self.categorized_osm_components(country, osm_components)
for props, component in (tagged or []):
if component == AddressFormatter.CITY:
if self.unambiguous_wikipedia(props, language):
unambiguous_city = True
if 'population' in props:
try:
population = int(props['population'])
except (ValueError, TypeError):
continue
# Perform dropout on places
address_components = place_config.dropout_components(address_components, osm_components, country=country, population=population, unambiguous_city=unambiguous_city)
return address_components
def dropout_address_level_component(self, address_components, component): def dropout_address_level_component(self, address_components, component):
probability = self.address_level_dropout_probabilities.get(component, None) probability = self.address_level_dropout_probabilities.get(component, None)
if probability is not None and random.random() < probability: if probability is not None and random.random() < probability:
@@ -1793,28 +1818,7 @@ class AddressComponents(object):
self.add_house_number_phrase(address_components, language, country=country) self.add_house_number_phrase(address_components, language, country=country)
if dropout_places: if dropout_places:
# Population of the city helps us determine if the city can be used address_components = self.dropout_places(address_components, all_osm_components, country, language, population=population, population_from_city=population_from_city)
# on its own like "Seattle" or "New York" vs. smaller cities like
# have to be qualified with a state, country, etc.
unambiguous_city = False
if population is None and population_from_city:
population = 0
tagged = self.categorized_osm_components(country, osm_components)
for props, component in (tagged or []):
if component == AddressFormatter.CITY:
if self.unambiguous_wikipedia(props, language):
unambiguous_city = True
if 'population' in props:
try:
population = int(props['population'])
except (ValueError, TypeError):
continue
# Perform dropout on places
address_components = place_config.dropout_components(address_components, all_osm_components, country=country, population=population, unambiguous_city=unambiguous_city)
self.drop_invalid_components(address_components, country) self.drop_invalid_components(address_components, country)