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

[fix] component bitsets

Browse Source
This commit is contained in:
Al
2016-05-24 13:07:32 -04:00
parent 0dbfd79b72
commit 046f445a56
1 changed files with 5 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
scripts/geodata/addresses/components.py
View File

@@ -187,6 +187,9 @@ class AddressComponents(object):
Some components are also more likely to be dropped than others, so in the same config there are Some components are also more likely to be dropped than others, so in the same config there are
dropout probabilities for each. dropout probabilities for each.
''' '''
if not components:
return []
component_bitset = self.component_bitset(components) component_bitset = self.component_bitset(components)
candidates = [c for c in components if c in self.address_level_dropout_probabilities] candidates = [c for c in components if c in self.address_level_dropout_probabilities]
@@ -985,6 +988,8 @@ class AddressComponents(object):
return {c: v for c, v in six.iteritems(address_components) if c != AddressFormatter.POSTCODE} return {c: v for c, v in six.iteritems(address_components) if c != AddressFormatter.POSTCODE}
def drop_invalid_components(self, address_components): def drop_invalid_components(self, address_components):
if not address_components:
return
component_bitset = self.component_bitset(address_components) component_bitset = self.component_bitset(address_components)
for c in list(address_components): for c in list(address_components):
if c in self.component_dependencies and not component_bitset & self.component_dependencies[c]: if c in self.component_dependencies and not component_bitset & self.component_dependencies[c]: