def process_str_glob(s, enc):
"""Receive str and encoding, then return the list
of str considering glob processing.
Decode str into unicode and pass it to process_unicode_glob.
When decode failed, return the result splitted per word.
Splitting depends on locale specified by rxGlob_L.
"""
try:
if not isinstance(s, unicode):
uni = s.decode(enc, "strict")
else:
uni = s
except UnicodeDecodeError:
return rxGlob_L.findall(s)
bigrams = process_unicode_glob(uni)
return [x.encode(enc, "strict") for x in bigrams]
def process_str_glob(s, enc='utf-8'):
"""Receive str and encoding, then return the list
of str considering glob processing.
Decode str into unicode and pass it to process_unicode_glob.
When decode failed, return the result splitted per word.
Splitting depends on locale specified by rxGlob_L.
"""
try:
if not isinstance(s, unicode):
uni = s.decode(enc, "strict")
else:
uni = s
except UnicodeDecodeError:
return rxGlob_L.findall(s)
bigrams = process_unicode_glob(uni)
return [x.encode(enc, "strict") for x in bigrams]