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

[places] better handling of population exceptions in places config

Browse Source
This commit is contained in:
Al
2016-11-25 00:37:57 -08:00
parent 4e10dc47e1
commit 5c3ccc3bc6
2 changed files with 41 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files

72
scripts/geodata/places/config.py
View File

@@ -34,11 +34,11 @@ class PlaceConfig(object):
AddressFormatter.WORLD_REGION,
}
numeric_ops = (('lte', operator.le),
('gt', operator.gt),
('lt', operator.lt),
('gte', operator.ge),
)
numeric_ops = {'lte': operator.le,
'gt': operator.gt,
'lt': operator.lt,
'gte': operator.ge,
}
def __init__(self, config_file=PLACE_CONFIG_FILE):
self.cache = {}
@@ -82,7 +82,34 @@ class PlaceConfig(object):
return nested_get(config, key, default=default)
def include_component(self, component, containing_ids, country=None, population=None):
def include_by_population_exceptions(self, population_exceptions, population):
if population_exceptions:
try:
population = int(population)
except (TypeError, ValueError):
population = 0
for exc in population_exceptions:
support = 0
for k in exc:
op = self.numeric_ops.get(k)
if not op:
continue
res = op(population, exc[k])
if not res:
support = 0
break
support += 1
if support > 0:
probability = exc.get('probability', 0.0)
if random.random() < probability:
return True
return False
def include_component_simple(self, component, containing_ids, country=None):
containing = self.get_property(('components', component, 'containing'), country=country, default=None)
if containing is not None:
@@ -90,36 +117,17 @@ class PlaceConfig(object):
if (c['type'], safe_encode(c['id'])) in containing_ids:
return random.random() < c['probability']
if population is not None:
population_exceptions = self.get_property(('components', component, 'population'), country=country, default=None)
if population_exceptions:
try:
population = int(population)
except (TypeError, ValueError):
population = 0
for exc in population_exceptions:
support = 0
for k, op in self.numeric_ops:
if k not in exc:
continue
res = op(population, exc[k])
if not res:
support = 0
break
support += 1
if support > 0:
probability = exc.get('probability', 0.0)
return random.random() < probability
probability = self.get_property(('components', component, 'probability'), country=country, default=0.0)
return random.random() < probability
def include_component(self, component, containing_ids, country=None, population=None, check_population=True):
if check_population:
population_exceptions = self.get_property(('components', component, 'population'), country=country, default=None)
if population_exceptions and self.include_by_population_exceptions(population_exceptions, population=population or 0):
return True
return self.include_component_simple(component, containing_ids, country=country)
def drop_invalid_components(self, address_components, country, original_bitset=None):
if not address_components:
return