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

[fix] prefix/suffix regexes

Browse Source
This commit is contained in:
Al
2015-12-05 18:41:22 -05:00
parent 5fcb6d2c30
commit dd8f8b4d7b
1 changed files with 3 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
scripts/geodata/names/normalization.py
View File

@@ -6,6 +6,7 @@ from geodata.encoding import safe_decode
name_prefixes = ['{} '.format(s) for s in ( name_prefixes = ['{} '.format(s) for s in (
'city of', 'city of',
'township of', 'township of',
'regional municipality of',
'municipality of', 'municipality of',
'borough of', 'borough of',
'london borough of', 'london borough of',
@@ -17,8 +18,8 @@ name_suffixes = [' {}'.format(s) for s in (
'municipality', 'municipality',
)] )]
name_prefix_regex = re.compile('^{}'.format('|'.join(name_prefixes)), re.I | re.UNICODE) name_prefix_regex = re.compile('^(?:{})'.format('|'.join(name_prefixes)), re.I | re.UNICODE)
name_suffix_regex = re.compile('{}$'.format('|'.join(name_suffixes)), re.I | re.UNICODE) name_suffix_regex = re.compile('(?:{})$'.format('|'.join(name_suffixes)), re.I | re.UNICODE)
def replace_name_prefixes(name): def replace_name_prefixes(name):