Fix #1176622 (txt files - 'dict' object has no attribute 'lower' (0.9.29))

This commit is contained in:
Kovid Goyal 2013-05-06 15:29:09 +05:30
parent 25501e8574
commit 49578b23e6

View File

@ -91,14 +91,15 @@ class TXTInput(InputFormatPlugin):
log.debug('Using user specified input encoding of %s' % ienc) log.debug('Using user specified input encoding of %s' % ienc)
else: else:
det_encoding = detect(txt) det_encoding = detect(txt)
det_encoding, confidence = det_encoding['encoding'], det_encoding['confidence']
if det_encoding and det_encoding.lower().replace('_', '-').strip() in ( if det_encoding and det_encoding.lower().replace('_', '-').strip() in (
'gb2312', 'chinese', 'csiso58gb231280', 'euc-cn', 'euccn', 'gb2312', 'chinese', 'csiso58gb231280', 'euc-cn', 'euccn',
'eucgb2312-cn', 'gb2312-1980', 'gb2312-80', 'iso-ir-58'): 'eucgb2312-cn', 'gb2312-1980', 'gb2312-80', 'iso-ir-58'):
# Microsoft Word exports to HTML with encoding incorrectly set to # Microsoft Word exports to HTML with encoding incorrectly set to
# gb2312 instead of gbk. gbk is a superset of gb2312, anyway. # gb2312 instead of gbk. gbk is a superset of gb2312, anyway.
det_encoding = 'gbk' det_encoding = 'gbk'
ienc = det_encoding['encoding'] ienc = det_encoding
log.debug('Detected input encoding as %s with a confidence of %s%%' % (ienc, det_encoding['confidence'] * 100)) log.debug('Detected input encoding as %s with a confidence of %s%%' % (ienc, confidence * 100))
if not ienc: if not ienc:
ienc = 'utf-8' ienc = 'utf-8'
log.debug('No input encoding specified and could not auto detect using %s' % ienc) log.debug('No input encoding specified and could not auto detect using %s' % ienc)