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

[fix] unit phrases

Browse Source
This commit is contained in:
Al
2016-08-27 00:23:47 -04:00
parent a4bd7ec3ad
commit 6723fff9b4
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
scripts/geodata/openaddresses/formatter.py
View File

@@ -42,8 +42,8 @@ class OpenAddressesFormatter(object):
for (lang, dictionary_type), values in six.iteritems(address_phrase_dictionaries.phrases): for (lang, dictionary_type), values in six.iteritems(address_phrase_dictionaries.phrases):
if dictionary_type == 'unit_types_numbered': if dictionary_type == 'unit_types_numbered':
unit_phrases = itertools.chain(*[safe_encode(p) for p in values if len(p) > 2]) unit_phrases = [safe_encode(p) for p in itertools.chain(*values) if len(p) > 2]
pattern = re.compile(r'\b(?:{})\s+(?:#?\s*)(?:[\d]+|[a-z]|[a-z]\-?[\d]+|[\d]+\-?[a-z])\s*$'.format(six.u('|').join(unit_phrases)), pattern = re.compile(r'\b(?:{})\s+(?:#?\s*)(?:[\d]+|[a-z]|[a-z]\-?[\d]+|[\d]+\-?[a-z])\s*$'.format(safe_encode('|').join(unit_phrases)),
re.I | re.UNICODE) re.I | re.UNICODE)
unit_type_regexes[lang] = pattern unit_type_regexes[lang] = pattern