Update guessit with unicode fix

This commit is contained in:
Ruud committed 2013-06-22 00:34:58 +02:00
1 parent b2d9a7675d
commit bc8d8dcd04
35 files changed
+517 -218

No files matched your search

+15 -9
View File
@@ -19,24 +19,30 @@
#
from __future__ import unicode_literals
#from guessit.transfo import SingleNodeGuesser
#from guessit.date import search_year
from guessit.country import Country
from guessit import Guess
import logging
log = logging.getLogger(__name__)
# list of common words which could be interpreted as countries, but which
# are far too common to be able to say they represent a country
country_common_words = frozenset([ 'bt', 'bb' ])
def process(mtree):
for node in mtree.unidentified_leaves():
# only keep explicit groups (enclosed in parentheses/brackets)
if len(node.node_idx) == 2:
try:
country = Country(node.value[1:-1], strict=True)
if node.value[0] + node.value[-1] not in ['()', '[]', '{}']:
continue
node.guess = Guess(country=country, confidence=1.0)
c = node.value[1:-1].lower()
if c in country_common_words:
continue
# only keep explicit groups (enclosed in parentheses/brackets)
if node.value[0] + node.value[-1] not in ['()', '[]', '{}']:
continue
try:
country = Country(c, strict=True)
except ValueError:
pass
continue
node.guess = Guess(country=country, confidence=1.0)