1
0
mirror of https://github.com/moparisthebest/SickRage synced 2024-11-14 13:25:11 -05:00
SickRage/lib/guessit/transfo/guess_movie_title_from_position.py
echel0n 0d9fbc1ad7 Welcome to our SickBeard-TVRage Edition ...
This version of SickBeard uses both TVDB and TVRage to search and gather it's series data from allowing you to now have access to and download shows that you couldn't before because of being locked into only what TheTVDB had to offer.

Also this edition is based off the code we used in our XEM editon so it does come with scene numbering support as well as all the other features our XEM edition has to offer.

Please before using this with your existing database (sickbeard.db) please make a backup copy of it and delete any other database files such as cache.db and failed.db if present, we HIGHLY recommend starting out with no database files at all to make this a fresh start but the choice is at your own risk!

Enjoy!
2014-03-09 22:39:12 -07:00

174 lines
6.4 KiB
Python

#!/usr/bin/env python2
# -*- coding: utf-8 -*-
#
# GuessIt - A library for guessing information from filenames
# Copyright (c) 2012 Nicolas Wack <wackou@gmail.com>
#
# GuessIt is free software; you can redistribute it and/or modify it under
# the terms of the Lesser GNU General Public License as published by
# the Free Software Foundation; either version 3 of the License, or
# (at your option) any later version.
#
# GuessIt is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# Lesser GNU General Public License for more details.
#
# You should have received a copy of the Lesser GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
#
from __future__ import unicode_literals
from guessit import Guess
import unicodedata
import logging
log = logging.getLogger(__name__)
def process(mtree):
def found_property(node, name, value, confidence):
node.guess = Guess({ name: value },
confidence=confidence)
log.debug('Found with confidence %.2f: %s' % (confidence, node.guess))
def found_title(node, confidence):
found_property(node, 'title', node.clean_value, confidence)
basename = mtree.node_at((-2,))
all_valid = lambda leaf: len(leaf.clean_value) > 0
basename_leftover = basename.unidentified_leaves(valid=all_valid)
try:
folder = mtree.node_at((-3,))
folder_leftover = folder.unidentified_leaves()
except ValueError:
folder = None
folder_leftover = []
log.debug('folder: %s' % folder_leftover)
log.debug('basename: %s' % basename_leftover)
# specific cases:
# if we find the same group both in the folder name and the filename,
# it's a good candidate for title
if (folder_leftover and basename_leftover and
folder_leftover[0].clean_value == basename_leftover[0].clean_value):
found_title(folder_leftover[0], confidence=0.8)
return
# specific cases:
# if the basename contains a number first followed by an unidentified
# group, and the folder only contains 1 unidentified one, then we have
# a series
# ex: Millenium Trilogy (2009)/(1)The Girl With The Dragon Tattoo(2009).mkv
try:
series = folder_leftover[0]
filmNumber = basename_leftover[0]
title = basename_leftover[1]
basename_leaves = basename.leaves()
num = int(filmNumber.clean_value)
log.debug('series: %s' % series.clean_value)
log.debug('title: %s' % title.clean_value)
if (series.clean_value != title.clean_value and
series.clean_value != filmNumber.clean_value and
basename_leaves.index(filmNumber) == 0 and
basename_leaves.index(title) == 1):
found_title(title, confidence=0.6)
found_property(series, 'filmSeries',
series.clean_value, confidence=0.6)
found_property(filmNumber, 'filmNumber',
num, confidence=0.6)
return
except Exception:
pass
# specific cases:
# - movies/tttttt (yyyy)/tttttt.ccc
try:
if mtree.node_at((-4, 0)).value.lower() == 'movies':
folder = mtree.node_at((-3,))
# Note:too generic, might solve all the unittests as they all
# contain 'movies' in their path
#
#if containing_folder.is_leaf() and not containing_folder.guess:
# containing_folder.guess =
# Guess({ 'title': clean_string(containing_folder.value) },
# confidence=0.7)
year_group = folder.first_leaf_containing('year')
groups_before = folder.previous_unidentified_leaves(year_group)
found_title(groups_before[0], confidence=0.8)
return
except Exception:
pass
# if we have either format or videoCodec in the folder containing the file
# or one of its parents, then we should probably look for the title in
# there rather than in the basename
try:
props = mtree.previous_leaves_containing(mtree.children[-2],
[ 'videoCodec', 'format',
'language' ])
except IndexError:
props = []
if props:
group_idx = props[0].node_idx[0]
if all(g.node_idx[0] == group_idx for g in props):
# if they're all in the same group, take leftover info from there
leftover = mtree.node_at((group_idx,)).unidentified_leaves()
if leftover:
found_title(leftover[0], confidence=0.7)
return
# look for title in basename if there are some remaining undidentified
# groups there
if basename_leftover:
title_candidate = basename_leftover[0]
# if basename is only one word and the containing folder has at least
# 3 words in it, we should take the title from the folder name
# ex: Movies/Alice in Wonderland DVDRip.XviD-DiAMOND/dmd-aw.avi
# ex: Movies/Somewhere.2010.DVDRip.XviD-iLG/i-smwhr.avi <-- TODO: gets caught here?
if (title_candidate.clean_value.count(' ') == 0 and
folder_leftover and
folder_leftover[0].clean_value.count(' ') >= 2):
found_title(folder_leftover[0], confidence=0.7)
return
# if there are only 2 unidentified groups, the first of which is inside
# brackets or parentheses, we take the second one for the title:
# ex: Movies/[阿维达].Avida.2006.FRENCH.DVDRiP.XViD-PROD.avi
if len(basename_leftover) == 2 and basename_leftover[0].is_explicit():
found_title(basename_leftover[1], confidence=0.8)
return
# if all else fails, take the first remaining unidentified group in the
# basename as title
found_title(title_candidate, confidence=0.6)
return
# if there are no leftover groups in the basename, look in the folder name
if folder_leftover:
found_title(folder_leftover[0], confidence=0.5)
return
# if nothing worked, look if we have a very small group at the beginning
# of the basename
basename = mtree.node_at((-2,))
basename_leftover = basename.unidentified_leaves(valid=lambda leaf: True)
if basename_leftover:
found_title(basename_leftover[0], confidence=0.4)
return