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

[fix] regexes

Browse Source
This commit is contained in:
Al
2015-07-14 14:04:32 -04:00
parent d494963dcd
commit d57f9df7ed
1 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
scripts/geodata/osm/osm_address_training_data.py
View File

@@ -238,6 +238,8 @@ def normalize_osm_name_tag(tag, script=False):
WAYS_LANGUAGE_DATA_FILENAME = 'streets_by_language.tsv' WAYS_LANGUAGE_DATA_FILENAME = 'streets_by_language.tsv'
beginning_re = re.compile('^[^0-9\-]+', re.UNICODE)
end_re = re.compile('[^0-9]+$', re.UNICODE)
latitude_dms_regex = re.compile(ur'^(-?[0-9]{1,2})[ ]*[ :°ºd][ ]*([0-5]?[0-9])?[ ]*[:\'\u2032m]?[ ]*([0-5]?[0-9](?:\.\d+)?)?[ ]*[:\?\"\u2033s]?[ ]*(N|n|S|s)?$', re.I | re.UNICODE) latitude_dms_regex = re.compile(ur'^(-?[0-9]{1,2})[ ]*[ :°ºd][ ]*([0-5]?[0-9])?[ ]*[:\'\u2032m]?[ ]*([0-5]?[0-9](?:\.\d+)?)?[ ]*[:\?\"\u2033s]?[ ]*(N|n|S|s)?$', re.I | re.UNICODE)
longitude_dms_regex = re.compile(ur'^(-?1[0-8][0-9]|0?[0-9]{1,2})[ ]*[ :°ºd][ ]*([0-5]?[0-9])?[ ]*[:\'\u2032m]?[ ]*([0-5]?[0-9](?:\.\d+)?)?[ ]*[:\?\"\u2033s]?[ ]*(E|e|W|w)?$', re.I | re.UNICODE) longitude_dms_regex = re.compile(ur'^(-?1[0-8][0-9]|0?[0-9]{1,2})[ ]*[ :°ºd][ ]*([0-5]?[0-9])?[ ]*[:\'\u2032m]?[ ]*([0-5]?[0-9](?:\.\d+)?)?[ ]*[:\?\"\u2033s]?[ ]*(E|e|W|w)?$', re.I | re.UNICODE)