Solve the bug of parsing titles with unicode (cyrillic)

This commit is contained in:
Art Zhitnik 2012-11-01 23:06:46 +10:00
parent b7a34316d2
commit 39973a0236

View File

@ -333,11 +333,15 @@ class FileDownloader(object):
title = info_dict['title']
matchtitle = self.params.get('matchtitle', False)
if matchtitle and not re.search(matchtitle, title, re.IGNORECASE):
return u'[download] "' + title + '" title did not match pattern "' + matchtitle + '"'
if matchtitle:
matchtitle = matchtitle.decode('utf8')
if not re.search(matchtitle, title, re.IGNORECASE):
return u'[download] "' + title + '" title did not match pattern "' + matchtitle + '"'
rejecttitle = self.params.get('rejecttitle', False)
if rejecttitle and re.search(rejecttitle, title, re.IGNORECASE):
return u'"' + title + '" title matched reject pattern "' + rejecttitle + '"'
if rejecttitle:
rejecttitle = rejecttitle.decode('utf8')
if re.search(rejecttitle, title, re.IGNORECASE):
return u'"' + title + '" title matched reject pattern "' + rejecttitle + '"'
return None
def process_info(self, info_dict):