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

[fix] only replace whitespace between words, not for instance whitespace around an existing hyphen, and reducing to one space for spaced hyphens

Browse Source
This commit is contained in:
Al
2016-10-19 01:23:58 -04:00
parent 78f341f4f1
commit 51afc2619b
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
scripts/geodata/addresses/components.py
View File

@@ -1224,8 +1224,8 @@ class AddressComponents(object):
address_components.pop(AddressFormatter.HOUSE_NUMBER, None) address_components.pop(AddressFormatter.HOUSE_NUMBER, None)
name_regex = re.compile('^[\s\-]*(.*?)[\s\-]*$') name_regex = re.compile('^[\s\-]*(.*?)[\s\-]*$')
whitespace_regex = re.compile('[\s]+') whitespace_regex = re.compile('(?<=[\w])[\s]+(?=[\w])')
hyphen_regex = re.compile('[\-]+') hyphen_regex = re.compile('[\s]*[\-]+[\s]*')
def dehyphenate_multiword_name(self, name): def dehyphenate_multiword_name(self, name):
return self.hyphen_regex.sub(six.u(' '), name) return self.hyphen_regex.sub(six.u(' '), name)