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

[states] adding multiple state abbreviations for states that can have periods in the naem like D.C., D.F. in Mexico and Brasil, etc.

Browse Source
This commit is contained in:
Al
2016-12-09 19:48:59 -05:00
parent 675552d254
commit 973466bb13
5 changed files with 55 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

25
scripts/geodata/states/state_abbreviations.py
View File

@@ -1,4 +1,5 @@
import os
import random
import six
import sys
import yaml
@@ -24,30 +25,34 @@ class StateAbbreviations(object):
country = filename.split('.yaml')[0]
country_config = yaml.load(open(os.path.join(base_dir, filename)))
country_abbreviations = defaultdict(dict)
country_full_names = defaultdict(dict)
country_abbreviations = defaultdict(list)
country_full_names = defaultdict(list)
for abbreviation, vals in six.iteritems(country_config):
for language, full_name in six.iteritems(vals):
full_name = safe_decode(full_name)
abbreviation = safe_decode(abbreviation)
country_abbreviations[full_name.lower()][language] = abbreviation
country_full_names[abbreviation.lower()][language] = full_name
country_abbreviations[(full_name.lower(), language)].append(abbreviation)
country_full_names[(abbreviation.lower(), language)].append(full_name)
self.abbreviations[country] = dict(country_abbreviations)
self.full_names[country] = dict(country_full_names)
def get_abbreviation(self, country, language, state, default=None):
value = nested_get(self.abbreviations, (country.lower(), state.lower(), language.lower()))
if value is DoesNotExist:
values = nested_get(self.abbreviations, (country.lower(), (state.lower(), language.lower())))
if values is DoesNotExist:
return default
return value
if len(values) == 1:
return values[0]
return random.choice(values)
def get_full_name(self, country, language, state, default=None):
value = nested_get(self.full_names, (country.lower(), state.lower(), language.lower()))
if value is DoesNotExist:
values = nested_get(self.full_names, (country.lower(), (state.lower(), language.lower())))
if values is DoesNotExist:
return default
return value
if len(values) == 1:
return values[0]
return random.choice(values)
state_abbreviations = StateAbbreviations()