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

[fix] convert newlines

Browse Source
This commit is contained in:
Al
2016-10-27 12:01:48 -04:00
parent e48f207d10
commit e9106698d2
1 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
scripts/geodata/addresses/components.py
View File

@@ -1223,6 +1223,7 @@ class AddressComponents(object):
else: else:
address_components.pop(AddressFormatter.HOUSE_NUMBER, None) address_components.pop(AddressFormatter.HOUSE_NUMBER, None)
newline_regex = re.compile('[\n]+')
name_regex = re.compile('^[\s\-]*(.*?)[\s\-]*$') name_regex = re.compile('^[\s\-]*(.*?)[\s\-]*$')
whitespace_regex = re.compile('(?<=[\w])[\s]+(?=[\w])') whitespace_regex = re.compile('(?<=[\w])[\s]+(?=[\w])')
hyphen_regex = re.compile('[\s]*[\-]+[\s]*') hyphen_regex = re.compile('[\s]*[\-]+[\s]*')
@@ -1237,6 +1238,7 @@ class AddressComponents(object):
@classmethod @classmethod
def strip_whitespace_and_hyphens(cls, name): def strip_whitespace_and_hyphens(cls, name):
name = cls.newline_regex.sub(six.u(' '), name)
return cls.name_regex.match(name).group(1) return cls.name_regex.match(name).group(1)
def name_hyphens(self, name, hyphenate_multiword_probability=None, remove_hyphen_probability=None): def name_hyphens(self, name, hyphenate_multiword_probability=None, remove_hyphen_probability=None):