def __init__(self):
Transformer.__init__(self, 15)
of_separators = ['of', 'sur', '/', '\\']
of_separators_re = re.compile(build_or_pattern(of_separators, escape=True), re.IGNORECASE)
self.container = PropertiesContainer(enhance=False, canonical_from_pattern=False)
episode_words = ['episodes?']
def _formater(episode_number):
epnum = parse_numeral(episode_number)
if not valid_year(epnum):
if epnum > 100:
season, epnum = epnum // 100, epnum % 100
# episodes which have a season > 50 are most likely errors
# (Simpson is at 25!)
if season > 50:
return None
return {'season': season, 'episodeNumber': epnum}
else:
return epnum
self.container.register_property(['episodeNumber', 'season'], '[0-9]{2,4}', confidence=0.6, formatter=_formater, disabler=lambda options: options.get('episode_prefer_number') if options else False)
self.container.register_property(['episodeNumber', 'season'], '[0-9]{4}', confidence=0.6, formatter=_formater)
self.container.register_property('episodeNumber', '[^0-9](\d{1,3})', confidence=0.6, formatter=parse_numeral, disabler=lambda options: not options.get('episode_prefer_number') if options else True)
self.container.register_property(None, '(' + build_or_pattern(episode_words) + sep + '?(?P<episodeNumber>' + numeral + '))[^0-9]', confidence=0.4, formatter=parse_numeral)
self.container.register_property(None, r'(?P<episodeNumber>' + numeral + ')' + sep + '?' + of_separators_re.pattern + sep + '?(?P<episodeCount>' + numeral +')', confidence=0.6, formatter=parse_numeral)
self.container.register_property('episodeNumber', r'^' + sep + '?(\d{1,3})' + sep, confidence=0.4, formatter=parse_numeral, disabler=lambda options: not options.get('episode_prefer_number') if options else True)
self.container.register_property('episodeNumber', sep + r'(\d{1,3})' + sep + '?$', confidence=0.4, formatter=parse_numeral, disabler=lambda options: not options.get('episode_prefer_number') if options else True)
def __init__(self):
Transformer.__init__(self, 45)
self.container = PropertiesContainer(enhance=False,
canonical_from_pattern=False)
tlds_pattern = build_or_pattern(
TLDS) # All registered domain extension
safe_tlds_pattern = build_or_pattern(
['com', 'org', 'net']) # For sure a website extension
safe_subdomains_pattern = build_or_pattern(
['www']) # For sure a website subdomain
safe_prefix_tlds_pattern = build_or_pattern(
['co', 'com', 'org', 'net']) # Those words before a tlds are sure
self.container.register_property(
'website', '(?:' + safe_subdomains_pattern + '\.)+' +
r'(?:[a-z-]+\.)+' + r'(?:' + tlds_pattern + r')+')
self.container.register_property(
'website', '(?:' + safe_subdomains_pattern + '\.)*' +
r'[a-z-]+\.' + r'(?:' + safe_tlds_pattern + r')+')
self.container.register_property(
'website',
'(?:' + safe_subdomains_pattern + '\.)*' + r'[a-z-]+\.' + r'(?:' +
safe_prefix_tlds_pattern + r'\.)+' + r'(?:' + tlds_pattern + r')+')
def __init__(self):
Transformer.__init__(self, 45)
self.container = PropertiesContainer(enhance=False, canonical_from_pattern=False)
tlds = []
f = resource_stream('guessit', 'tlds-alpha-by-domain.txt')
f.readline()
next(f)
for tld in f:
tld = tld.strip()
if b'--' in tld:
continue
tlds.append(tld.decode("utf-8"))
f.close()
tlds_pattern = build_or_pattern(tlds) # All registered domain extension
safe_tlds_pattern = build_or_pattern(['com', 'org', 'net']) # For sure a website extension
safe_subdomains_pattern = build_or_pattern(['www']) # For sure a website subdomain
safe_prefix_tlds_pattern = build_or_pattern(['co', 'com', 'org', 'net']) # Those words before a tlds are sure
self.container.register_property('website', '(?:' + safe_subdomains_pattern + '\.)+' + r'(?:[a-z-]+\.)+' + r'(?:' + tlds_pattern + r')+')
self.container.register_property('website', '(?:' + safe_subdomains_pattern + '\.)*' + r'[a-z-]+\.' + r'(?:' + safe_tlds_pattern + r')+')
self.container.register_property('website', '(?:' + safe_subdomains_pattern + '\.)*' + r'[a-z-]+\.' + r'(?:' + safe_prefix_tlds_pattern + r'\.)+' + r'(?:' + tlds_pattern + r')+')
def __init__(self):
Transformer.__init__(self, 45)
self.container = PropertiesContainer(enhance=False, canonical_from_pattern=False)
tlds_pattern = build_or_pattern(TLDS) # All registered domain extension
safe_tlds_pattern = build_or_pattern(['com', 'org', 'net']) # For sure a website extension
safe_subdomains_pattern = build_or_pattern(['www']) # For sure a website subdomain
safe_prefix_tlds_pattern = build_or_pattern(['co', 'com', 'org', 'net']) # Those words before a tlds are sure
self.container.register_property('website', '(?:' + safe_subdomains_pattern + '\.)+' + r'(?:[a-z-]+\.)+' + r'(?:' + tlds_pattern + r')+')
self.container.register_property('website', '(?:' + safe_subdomains_pattern + '\.)*' + r'[a-z-]+\.' + r'(?:' + safe_tlds_pattern + r')+')
self.container.register_property('website', '(?:' + safe_subdomains_pattern + '\.)*' + r'[a-z-]+\.' + r'(?:' + safe_prefix_tlds_pattern + r'\.)+' + r'(?:' + tlds_pattern + r')+')
def __init__(self):
Transformer.__init__(self, 45)
self.container = PropertiesContainer(enhance=False,
canonical_from_pattern=False)
tlds = []
f = resource_stream('guessit', 'tlds-alpha-by-domain.txt')
f.readline()
next(f)
for tld in f:
tld = tld.strip()
if b'--' in tld:
continue
tlds.append(tld.decode("utf-8"))
f.close()
tlds_pattern = build_or_pattern(
tlds) # All registered domain extension
safe_tlds_pattern = build_or_pattern(
['com', 'org', 'net']) # For sure a website extension
safe_subdomains_pattern = build_or_pattern(
['www']) # For sure a website subdomain
safe_prefix_tlds_pattern = build_or_pattern(
['co', 'com', 'org', 'net']) # Those words before a tlds are sure
self.container.register_property(
'website', '(?:' + safe_subdomains_pattern + '\.)+' +
r'(?:[a-z-]+\.)+' + r'(?:' + tlds_pattern + r')+')
self.container.register_property(
'website', '(?:' + safe_subdomains_pattern + '\.)*' +
r'[a-z-]+\.' + r'(?:' + safe_tlds_pattern + r')+')
self.container.register_property(
'website',
'(?:' + safe_subdomains_pattern + '\.)*' + r'[a-z-]+\.' + r'(?:' +
safe_prefix_tlds_pattern + r'\.)+' + r'(?:' + tlds_pattern + r')+')
def __init__(self):
Transformer.__init__(self, 35)
self.container = PropertiesContainer()
self.qualities = QualitiesContainer()
def register_property(propname, props, **kwargs):
"""props a dict of {value: [patterns]}"""
for canonical_form, patterns in props.items():
if isinstance(patterns, tuple):
patterns2, pattern_kwarg = patterns
if kwargs:
current_kwarg = dict(kwargs)
current_kwarg.update(pattern_kwarg)
else:
current_kwarg = dict(pattern_kwarg)
current_kwarg['canonical_form'] = canonical_form
self.container.register_property(propname, *patterns2, **current_kwarg)
elif kwargs:
current_kwarg = dict(kwargs)
current_kwarg['canonical_form'] = canonical_form
self.container.register_property(propname, *patterns, **current_kwarg)
else:
self.container.register_property(propname, *patterns, canonical_form=canonical_form)
def register_quality(propname, quality_dict):
"""props a dict of {canonical_form: quality}"""
for canonical_form, quality in quality_dict.items():
self.qualities.register_quality(propname, canonical_form, quality)
register_property('container', {'mp4': ['MP4']})
# http://en.wikipedia.org/wiki/Pirated_movie_release_types
register_property('format', {'VHS': ['VHS', 'VHS-Rip'],
'Cam': ['CAM', 'CAMRip', 'HD-CAM'],
#'Telesync': ['TELESYNC', 'PDVD'],
'Telesync': (['TS', 'HD-TS'], {'confidence': 0.4}),
'Workprint': ['WORKPRINT', 'WP'],
'Telecine': ['TELECINE', 'TC'],
'PPV': ['PPV', 'PPV-Rip'], # Pay Per View
'TV': ['SD-TV', 'SD-TV-Rip', 'Rip-SD-TV', 'TV-Rip', 'Rip-TV'],
'DVB': ['DVB-Rip', 'DVB', 'PD-TV'],
'DVD': ['DVD', 'DVD-Rip', 'VIDEO-TS', 'DVD-R', 'DVD-9', 'DVD-5'],
'HDTV': ['HD-TV', 'TV-RIP-HD', 'HD-TV-RIP', 'HD-RIP'],
'VOD': ['VOD', 'VOD-Rip'],
'WEBRip': ['WEB-Rip'],
'WEB-DL': ['WEB-DL', 'WEB-HD', 'WEB'],
'HD-DVD': ['HD-DVD-Rip', 'HD-DVD'],
'BluRay': ['Blu-ray(?:-Rip)?', 'B[DR]', 'B[DR]-Rip', 'BD[59]', 'BD25', 'BD50']
})
register_quality('format', {'VHS': -100,
'Cam': -90,
'Telesync': -80,
'Workprint': -70,
'Telecine': -60,
'PPV': -50,
'TV': -30,
'DVB': -20,
'DVD': 0,
'HDTV': 20,
'VOD': 40,
'WEBRip': 50,
'WEB-DL': 60,
'HD-DVD': 80,
'BluRay': 100
})
register_property('screenSize', {'360p': ['(?:\d{3,}(?:\\|\/|x|\*))?360(?:i|p?x?)'],
'368p': ['(?:\d{3,}(?:\\|\/|x|\*))?368(?:i|p?x?)'],
'480p': ['(?:\d{3,}(?:\\|\/|x|\*))?480(?:i|p?x?)'],
#'480p': (['hr'], {'confidence': 0.2}), # duplicate dict key
'576p': ['(?:\d{3,}(?:\\|\/|x|\*))?576(?:i|p?x?)'],
'720p': ['(?:\d{3,}(?:\\|\/|x|\*))?720(?:i|p?x?)'],
'900p': ['(?:\d{3,}(?:\\|\/|x|\*))?900(?:i|p?x?)'],
'1080i': ['(?:\d{3,}(?:\\|\/|x|\*))?1080i'],
'1080p': ['(?:\d{3,}(?:\\|\/|x|\*))?1080p?x?'],
'4K': ['(?:\d{3,}(?:\\|\/|x|\*))?2160(?:i|p?x?)']
},
validator=ChainedValidator(DefaultValidator(), OnlyOneValidator()))
class ResolutionValidator(object):
"""Make sure our match is surrounded by separators, or by another entry"""
@staticmethod
def validate(prop, string, node, match, entry_start, entry_end):
"""
span = _get_span(prop, match)
span = _trim_span(span, string[span[0]:span[1]])
start, end = span
sep_start = start <= 0 or string[start - 1] in sep
sep_end = end >= len(string) or string[end] in sep
start_by_other = start in entry_end
end_by_other = end in entry_start
if (sep_start or start_by_other) and (sep_end or end_by_other):
return True
return False
"""
return True
_digits_re = re.compile('\d+')
def resolution_formatter(value):
digits = _digits_re.findall(value)
return 'x'.join(digits)
self.container.register_property('screenSize', '\d{3,4}-?[x\*]-?\d{3,4}', canonical_from_pattern=False, formatter=resolution_formatter, validator=ChainedValidator(DefaultValidator(), ResolutionValidator()))
register_quality('screenSize', {'360p': -300,
'368p': -200,
'480p': -100,
'576p': 0,
'720p': 100,
'900p': 130,
'1080i': 180,
'1080p': 200,
'4K': 400
})
_videoCodecProperty = {'Real': ['Rv\d{2}'], # http://en.wikipedia.org/wiki/RealVideo
'Mpeg2': ['Mpeg2'],
'DivX': ['DVDivX', 'DivX'],
'XviD': ['XviD'],
'h264': ['[hx]-264(?:-AVC)?', 'MPEG-4(?:-AVC)'],
'h265': ['[hx]-265(?:-HEVC)?', 'HEVC']
}
register_property('videoCodec', _videoCodecProperty)
register_quality('videoCodec', {'Real': -50,
'Mpeg2': -30,
'DivX': -10,
'XviD': 0,
'h264': 100,
'h265': 150
})
# http://blog.mediacoderhq.com/h264-profiles-and-levels/
# http://fr.wikipedia.org/wiki/H.264
self.container.register_property('videoProfile', 'BP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', 'XP', 'EP', canonical_form='XP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', 'MP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', 'HP', 'HiP', canonical_form='HP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', '10.?bit', 'Hi10P', canonical_form='10bit')
self.container.register_property('videoProfile', '8.?bit', canonical_form='8bit')
self.container.register_property('videoProfile', 'Hi422P', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', 'Hi444PP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
register_quality('videoProfile', {'BP': -20,
'XP': -10,
'MP': 0,
'HP': 10,
'10bit': 15,
'Hi422P': 25,
'Hi444PP': 35
})
# has nothing to do here (or on filenames for that matter), but some
# releases use it and it helps to identify release groups, so we adapt
register_property('videoApi', {'DXVA': ['DXVA']})
register_property('audioCodec', {'MP3': ['MP3', 'LAME', 'LAME(?:\d)+-(?:\d)+'],
'DolbyDigital': ['DD'],
'AAC': ['AAC'],
'AC3': ['AC3'],
'Flac': ['FLAC'],
'DTS': (['DTS'], {'validator': LeftValidator()}),
'TrueHD': ['True-HD']
})
register_quality('audioCodec', {'MP3': 10,
'DolbyDigital': 30,
'AAC': 35,
'AC3': 40,
'Flac': 45,
'DTS': 60,
'TrueHD': 70
})
self.container.register_property('audioProfile', 'HD', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'DTS']))
self.container.register_property('audioProfile', 'HD-MA', canonical_form='HDMA', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'DTS']))
self.container.register_property('audioProfile', 'HE', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'AAC']))
self.container.register_property('audioProfile', 'LC', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'AAC']))
self.container.register_property('audioProfile', 'HQ', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'AC3']))
register_quality('audioProfile', {'HD': 20,
'HDMA': 50,
'LC': 0,
'HQ': 0,
'HE': 20
})
register_property('audioChannels', {'7.1': ['7[\W_]1', '7ch', '8ch'],
'5.1': ['5[\W_]1', '5ch', '6ch'],
'2.0': ['2[\W_]0', '2ch', 'stereo'],
'1.0': ['1[\W_]0', '1ch', 'mono']
})
register_quality('audioChannels', {'7.1': 200,
'5.1': 100,
'2.0': 0,
'1.0': -100
})
self.container.register_property('episodeFormat', r'Minisodes?', canonical_form='Minisode')
self.container.register_property('crc32', '(?:[a-fA-F]|[0-9]){8}', enhance=False, canonical_from_pattern=False)
part_words = ['pt', 'part']
self.container.register_property(None, '(' + build_or_pattern(part_words) + sep + '?(?P<part>' + numeral + '))[^0-9]', enhance=False, canonical_from_pattern=False, confidence=0.4, formatter=parse_numeral)
register_property('other', {'AudioFix': ['Audio-Fix', 'Audio-Fixed'],
'SyncFix': ['Sync-Fix', 'Sync-Fixed'],
'DualAudio': ['Dual-Audio'],
'WideScreen': ['ws', 'wide-screen'],
'Netflix': ['Netflix', 'NF']
})
self.container.register_property('other', 'Real', 'Fix', canonical_form='Proper', validator=ChainedValidator(FullMatchValidator(), NeighborValidator()))
self.container.register_property('other', 'Proper', 'Repack', 'Rerip', canonical_form='Proper')
self.container.register_property('other', 'Fansub', canonical_form='Fansub', validator=ChainedValidator(FullMatchValidator(), NeighborValidator()))
self.container.register_property('other', 'Fastsub', canonical_form='Fastsub', validator=ChainedValidator(FullMatchValidator(), NeighborValidator()))
self.container.register_property('other', '(?:Seasons?' + sep + '?)?Complete', canonical_form='Complete')
self.container.register_property('other', 'R5', 'RC', canonical_form='R5')
self.container.register_property('other', 'Pre-Air', 'Preair', canonical_form='Preair')
self.container.register_property('other', 'CC') # Close Caption
self.container.register_property('other', 'LD', 'MD') # Line/Mic Dubbed
self.container.register_canonical_properties('other', 'Screener', 'Remux', '3D', 'HD', 'mHD', 'HDLight', 'HQ',
'DDC',
'HR', 'PAL', 'SECAM', 'NTSC')
self.container.register_canonical_properties('other', 'Limited', 'Complete', 'Classic', 'Unrated', 'LiNE', 'Bonus', 'Trailer', validator=WeakValidator())
for prop in self.container.get_properties('format'):
self.container.register_property('other', prop.pattern + '(-?Scr(?:eener)?)', canonical_form='Screener')
for exts in (subtitle_exts, info_exts, video_exts):
for container in exts:
self.container.register_property('container', container, confidence=0.3)
def __init__(self):
Transformer.__init__(self, 20)
range_separators = ['-', 'to', 'a']
discrete_separators = ['&', 'and', 'et']
of_separators = ['of', 'sur', '/', '\\']
season_words = ['seasons?', 'saisons?', 'series?']
episode_words = ['episodes?']
season_markers = ['s']
episode_markers = ['e', 'ep']
discrete_sep = sep
for range_separator in range_separators:
discrete_sep = discrete_sep.replace(range_separator, '')
discrete_separators.append(discrete_sep)
all_separators = list(range_separators)
all_separators.extend(discrete_separators)
self.container = PropertiesContainer(enhance=False,
canonical_from_pattern=False)
range_separators_re = re.compile(build_or_pattern(range_separators),
re.IGNORECASE)
discrete_separators_re = re.compile(
build_or_pattern(discrete_separators), re.IGNORECASE)
all_separators_re = re.compile(build_or_pattern(all_separators),
re.IGNORECASE)
of_separators_re = re.compile(
build_or_pattern(of_separators, escape=True), re.IGNORECASE)
season_words_re = re.compile(build_or_pattern(season_words),
re.IGNORECASE)
episode_words_re = re.compile(build_or_pattern(episode_words),
re.IGNORECASE)
season_markers_re = re.compile(build_or_pattern(season_markers),
re.IGNORECASE)
episode_markers_re = re.compile(build_or_pattern(episode_markers),
re.IGNORECASE)
def list_parser(value,
propertyListName,
discrete_separators_re=discrete_separators_re,
range_separators_re=range_separators_re,
allow_discrete=False,
fill_gaps=False):
discrete_elements = filter(lambda x: x != '',
discrete_separators_re.split(value))
discrete_elements = [x.strip() for x in discrete_elements]
proper_discrete_elements = []
i = 0
while i < len(discrete_elements):
if i < len(discrete_elements) - 2 and range_separators_re.match(
discrete_elements[i + 1]):
proper_discrete_elements.append(discrete_elements[i] +
discrete_elements[i + 1] +
discrete_elements[i + 2])
i += 3
else:
match = range_separators_re.search(discrete_elements[i])
if match and match.start() == 0:
proper_discrete_elements[i -
1] = proper_discrete_elements[
i -
1] + discrete_elements[i]
elif match and match.end() == len(discrete_elements[i]):
proper_discrete_elements.append(discrete_elements[i] +
discrete_elements[i +
1])
else:
proper_discrete_elements.append(discrete_elements[i])
i += 1
discrete_elements = proper_discrete_elements
ret = []
for discrete_element in discrete_elements:
range_values = filter(
lambda x: x != '',
range_separators_re.split(discrete_element))
range_values = [x.strip() for x in range_values]
if len(range_values) > 1:
for x in range(0, len(range_values) - 1):
start_range_ep = parse_numeral(range_values[x])
end_range_ep = parse_numeral(range_values[x + 1])
for range_ep in range(start_range_ep,
end_range_ep + 1):
if range_ep not in ret:
ret.append(range_ep)
else:
discrete_value = parse_numeral(discrete_element)
if discrete_value not in ret:
ret.append(discrete_value)
if len(ret) > 1:
if not allow_discrete:
valid_ret = []
# replace discrete elements by ranges
valid_ret.append(ret[0])
for i in range(0, len(ret) - 1):
previous = valid_ret[len(valid_ret) - 1]
if ret[i + 1] < previous:
pass
else:
valid_ret.append(ret[i + 1])
ret = valid_ret
if fill_gaps:
ret = list(range(min(ret), max(ret) + 1))
if len(ret) > 1:
return {None: ret[0], propertyListName: ret}
if len(ret) > 0:
return ret[0]
return None
def episode_parser_x(value):
return list_parser(value,
'episodeList',
discrete_separators_re=re.compile(
'x', re.IGNORECASE))
def episode_parser_e(value):
return list_parser(value,
'episodeList',
discrete_separators_re=re.compile(
'e', re.IGNORECASE),
fill_gaps=True)
def episode_parser(value):
return list_parser(value, 'episodeList')
def season_parser(value):
return list_parser(value, 'seasonList')
class ResolutionCollisionValidator(object):
def validate(self, prop, string, node, match, entry_start,
entry_end):
return len(match.group(2)) < 3 #limit
self.container.register_property(None,
r'(' + season_words_re.pattern + sep +
'?(?P<season>' + numeral + ')' + sep +
'?' + season_words_re.pattern + '?)',
confidence=1.0,
formatter=parse_numeral)
self.container.register_property(
None,
r'(' + season_words_re.pattern + sep + '?(?P<season>' +
digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern +
sep + '?' + digital_numeral + ')*)' + sep + '?' +
season_words_re.pattern + '?)' + sep,
confidence=1.0,
formatter={
None: parse_numeral,
'season': season_parser
},
validator=ChainedValidator(
DefaultValidator(),
FormatterValidator(
'season', lambda x: len(x) > 1
if hasattr(x, '__len__') else False)))
self.container.register_property(
None,
r'(' + season_markers_re.pattern + '(?P<season>' +
digital_numeral + ')[^0-9]?' + sep + '?(?P<episodeNumber>(?:e' +
digital_numeral + '(?:' + sep + '?[e-]' + digital_numeral +
')*)))',
confidence=1.0,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser_e,
'season': season_parser
},
validator=NoValidator())
#self.container.register_property(None, r'[^0-9]((?P<season>' + digital_numeral + ')[^0-9 .-]?-?(?P<episodeNumber>(?:x' + digital_numeral + '(?:' + sep + '?[x-]' + digital_numeral + ')*)))', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser_x, 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), ResolutionCollisionValidator()))
self.container.register_property(
None,
sep + r'((?P<season>' + digital_numeral + ')' + sep + '' +
'(?P<episodeNumber>(?:x' + sep + digital_numeral + '(?:' + sep +
'[x-]' + digital_numeral + ')*)))',
confidence=1.0,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser_x,
'season': season_parser
},
validator=ChainedValidator(DefaultValidator(),
ResolutionCollisionValidator()))
self.container.register_property(
None,
r'((?P<season>' + digital_numeral + ')' +
'(?P<episodeNumber>(?:x' + digital_numeral + '(?:[x-]' +
digital_numeral + ')*)))',
confidence=1.0,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser_x,
'season': season_parser
},
validator=ChainedValidator(DefaultValidator(),
ResolutionCollisionValidator()))
self.container.register_property(
None,
r'(' + season_markers_re.pattern + '(?P<season>' +
digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern +
sep + '?' + digital_numeral + ')*))',
confidence=0.6,
formatter={
None: parse_numeral,
'season': season_parser
},
validator=NoValidator())
self.container.register_property(None,
r'((?P<episodeNumber>' +
digital_numeral + ')' + sep +
'?v(?P<version>\d+))',
confidence=0.6,
formatter=parse_numeral)
self.container.register_property(
None,
r'(ep' + sep + r'?(?P<episodeNumber>' + numeral + ')' + sep + '?)',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(
None,
r'(ep' + sep + r'?(?P<episodeNumber>' + numeral + ')' + sep +
'?v(?P<version>\d+))',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(
None,
r'(' + episode_markers_re.pattern + '(?P<episodeNumber>' +
digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern +
sep + '?' + digital_numeral + ')*))',
confidence=0.6,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser
})
self.container.register_property(
None,
r'(' + episode_words_re.pattern + sep + '?(?P<episodeNumber>' +
digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern +
sep + '?' + digital_numeral + ')*)' + sep + '?' +
episode_words_re.pattern + '?)',
confidence=0.8,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser
})
self.container.register_property(
None,
r'(' + episode_markers_re.pattern + '(?P<episodeNumber>' +
digital_numeral + ')' + sep + '?v(?P<version>\d+))',
confidence=0.6,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser
})
self.container.register_property(
None,
r'(' + episode_words_re.pattern + sep + '?(?P<episodeNumber>' +
digital_numeral + ')' + sep + '?v(?P<version>\d+))',
confidence=0.8,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser
})
self.container.register_property('episodeNumber',
r'^ ?(\d{2})' + sep,
confidence=0.4,
formatter=parse_numeral)
self.container.register_property('episodeNumber',
r'^ ?(\d{2})' + sep,
confidence=0.4,
formatter=parse_numeral)
self.container.register_property('episodeNumber',
r'^ ?0(\d{1,2})' + sep,
confidence=0.4,
formatter=parse_numeral)
self.container.register_property('episodeNumber',
sep + r'(\d{2}) ?$',
confidence=0.4,
formatter=parse_numeral)
self.container.register_property('episodeNumber',
sep + r'0(\d{1,2}) ?$',
confidence=0.4,
formatter=parse_numeral)
self.container.register_property(
None,
r'((?P<episodeNumber>' + numeral + ')' + sep + '?' +
of_separators_re.pattern + sep + '?(?P<episodeCount>' + numeral +
')(?:' + sep + '?(?:episodes?|eps?))?)',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(
None,
r'((?:episodes?|eps?)' + sep + '?(?P<episodeNumber>' + numeral +
')' + sep + '?' + of_separators_re.pattern + sep +
'?(?P<episodeCount>' + numeral + '))',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(None,
r'((?:seasons?|saisons?|s)' + sep +
'?(?P<season>' + numeral + ')' + sep +
'?' + of_separators_re.pattern + sep +
'?(?P<seasonCount>' + numeral + '))',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(
None,
r'((?P<season>' + numeral + ')' + sep + '?' +
of_separators_re.pattern + sep + '?(?P<seasonCount>' + numeral +
')' + sep + '?(?:seasons?|saisons?|s))',
confidence=0.7,
formatter=parse_numeral)
self.container.register_canonical_properties('other',
'FiNAL',
'Complete',
validator=WeakValidator())
self.container.register_property(
None,
r'[^0-9]((?P<season>' + digital_numeral +
')[^0-9 .-]?-?(?P<other>xAll))',
confidence=1.0,
formatter={
None: parse_numeral,
'other': lambda x: 'Complete',
'season': season_parser
},
validator=ChainedValidator(DefaultValidator(),
ResolutionCollisionValidator()))
def __init__(self):
Transformer.__init__(self, 35)
self.container = PropertiesContainer()
self.qualities = QualitiesContainer()
def register_property(propname, props, **kwargs):
"""props a dict of {value: [patterns]}"""
for canonical_form, patterns in props.items():
if isinstance(patterns, tuple):
patterns2, pattern_kwarg = patterns
if kwargs:
current_kwarg = dict(kwargs)
current_kwarg.update(pattern_kwarg)
else:
current_kwarg = dict(pattern_kwarg)
current_kwarg['canonical_form'] = canonical_form
self.container.register_property(propname, *patterns2, **current_kwarg)
elif kwargs:
current_kwarg = dict(kwargs)
current_kwarg['canonical_form'] = canonical_form
self.container.register_property(propname, *patterns, **current_kwarg)
else:
self.container.register_property(propname, *patterns, canonical_form=canonical_form)
def register_quality(propname, quality_dict):
"""props a dict of {canonical_form: quality}"""
for canonical_form, quality in quality_dict.items():
self.qualities.register_quality(propname, canonical_form, quality)
register_property('container', {'mp4': ['MP4']})
# http://en.wikipedia.org/wiki/Pirated_movie_release_types
register_property('format', {'VHS': ['VHS', 'VHS-Rip'],
'Cam': ['CAM', 'CAMRip', 'HD-CAM'],
#'Telesync': ['TELESYNC', 'PDVD'],
'Telesync': (['TS', 'HD-TS'], {'confidence': 0.4}),
'Workprint': ['WORKPRINT', 'WP'],
'Telecine': ['TELECINE', 'TC'],
'PPV': ['PPV', 'PPV-Rip'], # Pay Per View
'TV': ['SD-TV', 'SD-TV-Rip', 'Rip-SD-TV', 'TV-Rip', 'Rip-TV'],
'DVB': ['DVB-Rip', 'DVB', 'PD-TV'],
'DVD': ['DVD', 'DVD-Rip', 'VIDEO-TS', 'DVD-R', 'DVD-9', 'DVD-5'],
'HDTV': ['HD-TV', 'TV-RIP-HD', 'HD-TV-RIP', 'HD-RIP'],
'VOD': ['VOD', 'VOD-Rip'],
'WEBRip': ['WEB-Rip'],
'WEB-DL': ['WEB-DL', 'WEB-HD', 'WEB'],
'HD-DVD': ['HD-DVD-Rip', 'HD-DVD'],
'BluRay': ['Blu-ray(?:-Rip)?', 'B[DR]', 'B[DR]-Rip', 'BD[59]', 'BD25', 'BD50']
})
register_quality('format', {'VHS': -100,
'Cam': -90,
'Telesync': -80,
'Workprint': -70,
'Telecine': -60,
'PPV': -50,
'TV': -30,
'DVB': -20,
'DVD': 0,
'HDTV': 20,
'VOD': 40,
'WEBRip': 50,
'WEB-DL': 60,
'HD-DVD': 80,
'BluRay': 100
})
register_property('screenSize', {'360p': ['(?:\d{3,}(?:\\|\/|x|\*))?360(?:i|p?x?)'],
'368p': ['(?:\d{3,}(?:\\|\/|x|\*))?368(?:i|p?x?)'],
'480p': ['(?:\d{3,}(?:\\|\/|x|\*))?480(?:i|p?x?)'],
#'480p': (['hr'], {'confidence': 0.2}), # duplicate dict key
'576p': ['(?:\d{3,}(?:\\|\/|x|\*))?576(?:i|p?x?)'],
'720p': ['(?:\d{3,}(?:\\|\/|x|\*))?720(?:i|p?x?)'],
'900p': ['(?:\d{3,}(?:\\|\/|x|\*))?900(?:i|p?x?)'],
'1080i': ['(?:\d{3,}(?:\\|\/|x|\*))?1080i'],
'1080p': ['(?:\d{3,}(?:\\|\/|x|\*))?1080p?x?'],
'4K': ['(?:\d{3,}(?:\\|\/|x|\*))?2160(?:i|p?x?)']
},
validator=ChainedValidator(DefaultValidator(), OnlyOneValidator()))
class ResolutionValidator(object):
"""Make sure our match is surrounded by separators, or by another entry"""
@staticmethod
def validate(prop, string, node, match, entry_start, entry_end):
"""
span = _get_span(prop, match)
span = _trim_span(span, string[span[0]:span[1]])
start, end = span
sep_start = start <= 0 or string[start - 1] in sep
sep_end = end >= len(string) or string[end] in sep
start_by_other = start in entry_end
end_by_other = end in entry_start
if (sep_start or start_by_other) and (sep_end or end_by_other):
return True
return False
"""
return True
_digits_re = re.compile('\d+')
def resolution_formatter(value):
digits = _digits_re.findall(value)
return 'x'.join(digits)
self.container.register_property('screenSize', '\d{3,4}-?[x\*]-?\d{3,4}', canonical_from_pattern=False, formatter=resolution_formatter, validator=ChainedValidator(DefaultValidator(), ResolutionValidator()))
register_quality('screenSize', {'360p': -300,
'368p': -200,
'480p': -100,
'576p': 0,
'720p': 100,
'900p': 130,
'1080i': 180,
'1080p': 200,
'4K': 400
})
_videoCodecProperty = {'Real': ['Rv\d{2}'], # http://en.wikipedia.org/wiki/RealVideo
'Mpeg2': ['Mpeg2'],
'DivX': ['DVDivX', 'DivX'],
'XviD': ['XviD'],
'h264': ['[hx]-264(?:-AVC)?', 'MPEG-4(?:-AVC)'],
'h265': ['[hx]-265(?:-HEVC)?', 'HEVC']
}
register_property('videoCodec', _videoCodecProperty)
register_quality('videoCodec', {'Real': -50,
'Mpeg2': -30,
'DivX': -10,
'XviD': 0,
'h264': 100,
'h265': 150
})
# http://blog.mediacoderhq.com/h264-profiles-and-levels/
# http://fr.wikipedia.org/wiki/H.264
self.container.register_property('videoProfile', 'BP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', 'XP', 'EP', canonical_form='XP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', 'MP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', 'HP', 'HiP', canonical_form='HP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', '10.?bit', 'Hi10P', canonical_form='10bit')
self.container.register_property('videoProfile', '8.?bit', canonical_form='8bit')
self.container.register_property('videoProfile', 'Hi422P', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
self.container.register_property('videoProfile', 'Hi444PP', validator=LeavesValidator(lambdas=[lambda node: 'videoCodec' in node.guess]))
register_quality('videoProfile', {'BP': -20,
'XP': -10,
'MP': 0,
'HP': 10,
'10bit': 15,
'Hi422P': 25,
'Hi444PP': 35
})
# has nothing to do here (or on filenames for that matter), but some
# releases use it and it helps to identify release groups, so we adapt
register_property('videoApi', {'DXVA': ['DXVA']})
register_property('audioCodec', {'MP3': ['MP3', 'LAME', 'LAME(?:\d)+-(?:\d)+'],
'DolbyDigital': ['DD'],
'AAC': ['AAC'],
'AC3': ['AC3'],
'Flac': ['FLAC'],
'DTS': (['DTS'], {'validator': LeftValidator()}),
'TrueHD': ['True-HD']
})
register_quality('audioCodec', {'MP3': 10,
'DolbyDigital': 30,
'AAC': 35,
'AC3': 40,
'Flac': 45,
'DTS': 60,
'TrueHD': 70
})
self.container.register_property('audioProfile', 'HD', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'DTS']))
self.container.register_property('audioProfile', 'HD-MA', canonical_form='HDMA', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'DTS']))
self.container.register_property('audioProfile', 'HE', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'AAC']))
self.container.register_property('audioProfile', 'LC', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'AAC']))
self.container.register_property('audioProfile', 'HQ', validator=LeavesValidator(lambdas=[lambda node: node.guess.get('audioCodec') == 'AC3']))
register_quality('audioProfile', {'HD': 20,
'HDMA': 50,
'LC': 0,
'HQ': 0,
'HE': 20
})
register_property('audioChannels', {'7.1': ['7[\W_]1', '7ch', '8ch'],
'5.1': ['5[\W_]1', '5ch', '6ch'],
'2.0': ['2[\W_]0', '2ch', 'stereo'],
'1.0': ['1[\W_]0', '1ch', 'mono']
})
register_quality('audioChannels', {'7.1': 200,
'5.1': 100,
'2.0': 0,
'1.0': -100
})
self.container.register_property('episodeFormat', r'Minisodes?', canonical_form='Minisode')
self.container.register_property('crc32', '(?:[a-fA-F]|[0-9]){8}', enhance=False, canonical_from_pattern=False)
weak_episode_words = ['pt', 'part']
self.container.register_property(None, '(' + build_or_pattern(weak_episode_words) + sep + '?(?P<part>' + numeral + '))[^0-9]', enhance=False, canonical_from_pattern=False, confidence=0.4, formatter=parse_numeral)
register_property('other', {'AudioFix': ['Audio-Fix', 'Audio-Fixed'],
'SyncFix': ['Sync-Fix', 'Sync-Fixed'],
'DualAudio': ['Dual-Audio'],
'WideScreen': ['ws', 'wide-screen'],
'Netflix': ['Netflix', 'NF']
})
self.container.register_property('other', 'Real', 'Fix', canonical_form='Proper', validator=NeighborValidator())
self.container.register_property('other', 'Proper', 'Repack', 'Rerip', canonical_form='Proper')
self.container.register_property('other', 'Fansub', canonical_form='Fansub')
self.container.register_property('other', 'Fastsub', canonical_form='Fastsub')
self.container.register_property('other', '(?:Seasons?' + sep + '?)?Complete', canonical_form='Complete')
self.container.register_property('other', 'R5', 'RC', canonical_form='R5')
self.container.register_property('other', 'Pre-Air', 'Preair', canonical_form='Preair')
self.container.register_canonical_properties('other', 'Screener', 'Remux', '3D', 'HD', 'mHD', 'HDLight', 'HQ',
'DDC',
'HR', 'PAL', 'SECAM', 'NTSC')
self.container.register_canonical_properties('other', 'Limited', 'Complete', 'Classic', 'Unrated', 'LiNE', 'Bonus', 'Trailer', validator=WeakValidator())
for prop in self.container.get_properties('format'):
self.container.register_property('other', prop.pattern + '(-?Scr(?:eener)?)', canonical_form='Screener')
for exts in (subtitle_exts, info_exts, video_exts):
for container in exts:
self.container.register_property('container', container, confidence=0.3)
def __init__(self):
Transformer.__init__(self, 20)
range_separators = ['-', 'to', 'a']
discrete_separators = ['&', 'and', 'et']
of_separators = ['of', 'sur', '/', '\\']
season_words = ['seasons?', 'saisons?', 'series?']
episode_words = ['episodes?']
season_markers = ['s']
episode_markers = ['e', 'ep']
discrete_sep = sep
for range_separator in range_separators:
discrete_sep = discrete_sep.replace(range_separator, '')
discrete_separators.append(discrete_sep)
all_separators = list(range_separators)
all_separators.extend(discrete_separators)
self.container = PropertiesContainer(enhance=False, canonical_from_pattern=False)
range_separators_re = re.compile(build_or_pattern(range_separators), re.IGNORECASE)
discrete_separators_re = re.compile(build_or_pattern(discrete_separators), re.IGNORECASE)
all_separators_re = re.compile(build_or_pattern(all_separators), re.IGNORECASE)
of_separators_re = re.compile(build_or_pattern(of_separators, escape=True), re.IGNORECASE)
season_words_re = re.compile(build_or_pattern(season_words), re.IGNORECASE)
episode_words_re = re.compile(build_or_pattern(episode_words), re.IGNORECASE)
season_markers_re = re.compile(build_or_pattern(season_markers), re.IGNORECASE)
episode_markers_re = re.compile(build_or_pattern(episode_markers), re.IGNORECASE)
def list_parser(value, propertyListName, discrete_separators_re=discrete_separators_re, range_separators_re=range_separators_re, allow_discrete=False, fill_gaps=False):
discrete_elements = filter(lambda x: x != '', discrete_separators_re.split(value))
discrete_elements = [x.strip() for x in discrete_elements]
proper_discrete_elements = []
i = 0
while i < len(discrete_elements):
if i < len(discrete_elements) - 2 and range_separators_re.match(discrete_elements[i+1]):
proper_discrete_elements.append(discrete_elements[i] + discrete_elements[i+1] + discrete_elements[i+2])
i += 3
else:
match = range_separators_re.search(discrete_elements[i])
if match and match.start() == 0:
proper_discrete_elements[i-1] = proper_discrete_elements[i-1] + discrete_elements[i]
elif match and match.end() == len(discrete_elements[i]):
proper_discrete_elements.append(discrete_elements[i] + discrete_elements[i + 1])
else:
proper_discrete_elements.append(discrete_elements[i])
i += 1
discrete_elements = proper_discrete_elements
ret = []
for discrete_element in discrete_elements:
range_values = filter(lambda x: x != '', range_separators_re.split(discrete_element))
range_values = [x.strip() for x in range_values]
if len(range_values) > 1:
for x in range(0, len(range_values) - 1):
start_range_ep = parse_numeral(range_values[x])
end_range_ep = parse_numeral(range_values[x+1])
for range_ep in range(start_range_ep, end_range_ep + 1):
if range_ep not in ret:
ret.append(range_ep)
else:
discrete_value = parse_numeral(discrete_element)
if discrete_value not in ret:
ret.append(discrete_value)
if len(ret) > 1:
if not allow_discrete:
valid_ret = []
# replace discrete elements by ranges
valid_ret.append(ret[0])
for i in range(0, len(ret) - 1):
previous = valid_ret[len(valid_ret) - 1]
if ret[i+1] < previous:
pass
else:
valid_ret.append(ret[i+1])
ret = valid_ret
if fill_gaps:
ret = list(range(min(ret), max(ret) + 1))
if len(ret) > 1:
return {None: ret[0], propertyListName: ret}
if len(ret) > 0:
return ret[0]
return None
def episode_parser_x(value):
return list_parser(value, 'episodeList', discrete_separators_re=re.compile('x', re.IGNORECASE))
def episode_parser_e(value):
return list_parser(value, 'episodeList', discrete_separators_re=re.compile('e',re.IGNORECASE), fill_gaps=True)
def episode_parser(value):
return list_parser(value, 'episodeList')
def season_parser(value):
return list_parser(value, 'seasonList')
class ResolutionCollisionValidator(object):
def validate(self, prop, string, node, match, entry_start, entry_end):
return len(match.group(2)) < 3 #limit
self.container.register_property(None, r'(' + season_words_re.pattern + sep + '?(?P<season>' + numeral + ')' + sep + '?' + season_words_re.pattern + '?)', confidence=1.0, formatter=parse_numeral)
self.container.register_property(None, r'(' + season_words_re.pattern + sep + '?(?P<season>' + digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + digital_numeral + ')*)' + sep + '?' + season_words_re.pattern + '?)' + sep, confidence=1.0, formatter={None: parse_numeral, 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), FormatterValidator('season', lambda x: len(x) > 1 if hasattr(x, '__len__') else False)))
self.container.register_property(None, r'(' + season_markers_re.pattern + '(?P<season>' + digital_numeral + ')[^0-9]?' + sep + '?(?P<episodeNumber>(?:e' + digital_numeral + '(?:' + sep + '?[e-]' + digital_numeral + ')*)))', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser_e, 'season': season_parser}, validator=NoValidator())
#self.container.register_property(None, r'[^0-9]((?P<season>' + digital_numeral + ')[^0-9 .-]?-?(?P<episodeNumber>(?:x' + digital_numeral + '(?:' + sep + '?[x-]' + digital_numeral + ')*)))', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser_x, 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), ResolutionCollisionValidator()))
self.container.register_property(None, sep + r'((?P<season>' + digital_numeral + ')' + sep + '' + '(?P<episodeNumber>(?:x' + sep + digital_numeral + '(?:' + sep + '[x-]' + digital_numeral + ')*)))', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser_x, 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), ResolutionCollisionValidator()))
self.container.register_property(None, r'((?P<season>' + digital_numeral + ')' + '(?P<episodeNumber>(?:x' + digital_numeral + '(?:[x-]' + digital_numeral + ')*)))', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser_x, 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), ResolutionCollisionValidator()))
self.container.register_property(None, r'(' + season_markers_re.pattern + '(?P<season>' + digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + digital_numeral + ')*))', confidence=0.6, formatter={None: parse_numeral, 'season': season_parser}, validator=NoValidator())
self.container.register_property(None, r'((?P<episodeNumber>' + digital_numeral + ')' + sep + '?v(?P<version>\d+))', confidence=0.6, formatter=parse_numeral)
self.container.register_property(None, r'(ep' + sep + r'?(?P<episodeNumber>' + numeral + ')' + sep + '?)', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'(ep' + sep + r'?(?P<episodeNumber>' + numeral + ')' + sep + '?v(?P<version>\d+))', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'(' + episode_markers_re.pattern + '(?P<episodeNumber>' + digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + digital_numeral + ')*))', confidence=0.6, formatter={None: parse_numeral, 'episodeNumber': episode_parser})
self.container.register_property(None, r'(' + episode_words_re.pattern + sep + '?(?P<episodeNumber>' + digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + digital_numeral + ')*)' + sep + '?' + episode_words_re.pattern + '?)', confidence=0.8, formatter={None: parse_numeral, 'episodeNumber': episode_parser})
self.container.register_property(None, r'(' + episode_markers_re.pattern + '(?P<episodeNumber>' + digital_numeral + ')' + sep + '?v(?P<version>\d+))', confidence=0.6, formatter={None: parse_numeral, 'episodeNumber': episode_parser})
self.container.register_property(None, r'(' + episode_words_re.pattern + sep + '?(?P<episodeNumber>' + digital_numeral + ')' + sep + '?v(?P<version>\d+))', confidence=0.8, formatter={None: parse_numeral, 'episodeNumber': episode_parser})
self.container.register_property('episodeNumber', r'^ ?(\d{2})' + sep, confidence=0.4, formatter=parse_numeral)
self.container.register_property('episodeNumber', r'^ ?(\d{2})' + sep, confidence=0.4, formatter=parse_numeral)
self.container.register_property('episodeNumber', r'^ ?0(\d{1,2})' + sep, confidence=0.4, formatter=parse_numeral)
self.container.register_property('episodeNumber', sep + r'(\d{2}) ?$', confidence=0.4, formatter=parse_numeral)
self.container.register_property('episodeNumber', sep + r'0(\d{1,2}) ?$', confidence=0.4, formatter=parse_numeral)
self.container.register_property(None, r'((?P<episodeNumber>' + numeral + ')' + sep + '?' + of_separators_re.pattern + sep + '?(?P<episodeCount>' + numeral + ')(?:' + sep + '?(?:episodes?|eps?))?)', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'((?:episodes?|eps?)' + sep + '?(?P<episodeNumber>' + numeral + ')' + sep + '?' + of_separators_re.pattern + sep + '?(?P<episodeCount>' + numeral + '))', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'((?:seasons?|saisons?|s)' + sep + '?(?P<season>' + numeral + ')' + sep + '?' + of_separators_re.pattern + sep + '?(?P<seasonCount>' + numeral + '))', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'((?P<season>' + numeral + ')' + sep + '?' + of_separators_re.pattern + sep + '?(?P<seasonCount>' + numeral + ')' + sep + '?(?:seasons?|saisons?|s))', confidence=0.7, formatter=parse_numeral)
self.container.register_canonical_properties('other', 'FiNAL', 'Complete', validator=WeakValidator())
self.container.register_property(None, r'[^0-9]((?P<season>' + digital_numeral + ')[^0-9 .-]?-?(?P<other>xAll))', confidence=1.0, formatter={None: parse_numeral, 'other': lambda x: 'Complete', 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), ResolutionCollisionValidator()))
container.register_property('format', 'HDTV', 'HD-TV')
container.register_property('format', 'DVB', 'DVB-Rip', 'DVB', 'PD-TV')
container.register_property('format', 'WEBRip', 'WEB-Rip')
container.register_property('format', 'VHS', 'VHS')
container.register_property('format', 'WEB-DL', 'WEB-DL')
container.register_property('screenSize', '360p', '(?:\d{3,}(?:\\|\/|x|\*))?360(?:i|p?x?)')
container.register_property('screenSize', '368p', '(?:\d{3,}(?:\\|\/|x|\*))?368(?:i|p?x?)')
container.register_property('screenSize', '480p', '(?:\d{3,}(?:\\|\/|x|\*))?480(?:i|p?x?)')
container.register_property('screenSize', '576p', '(?:\d{3,}(?:\\|\/|x|\*))?576(?:i|p?x?)')
container.register_property('screenSize', '720p', '(?:\d{3,}(?:\\|\/|x|\*))?720(?:i|p?x?)')
container.register_property('screenSize', '1080i', '(?:\d{3,}(?:\\|\/|x|\*))?1080i(?:i|p?x?)')
container.register_property('screenSize', '1080p', '(?:\d{3,}(?:\\|\/|x|\*))?1080(?:i|p?x?)')
container.register_property('screenSize', '4K', '(?:\d{3,}(?:\\|\/|x|\*))?2160(?:i|p?x?)')
profile_pattern = build_or_pattern(["BS", "EP", "MP", "HP", "AVC"])
container.register_property('videoCodec', 'XviD', 'XviD', 'XviD-' + profile_pattern)
container.register_property('videoCodec', 'DivX', 'DVDivX', 'DivX', 'DivX-' + profile_pattern)
container.register_property('videoCodec', 'h264', '[hx]-264', '[hx]-264-' + profile_pattern)
container.register_property('videoCodec', 'Rv10', 'Rv10')
container.register_property('videoCodec', 'Mpeg2', 'Mpeg2')
# has nothing to do here (or on filenames for that matter), but some
# releases use it and it helps to identify release groups, so we adapt
container.register_property('videoApi', 'DXVA', 'DXVA')
container.register_property('audioCodec', 'AC3', 'AC3')
container.register_property('audioCodec', 'DTS', 'DTS', 'DTS-HD')
container.register_property('audioCodec', 'AAC', 'HE-AAC', 'AAC-HE', 'LC-AAC', 'AAC-LC', 'AAC')
def __init__(self):
Transformer.__init__(self, 20)
of_separators = ['of', 'sur', '/', '\\']
of_separators_re = re.compile(build_or_pattern(of_separators, escape=True), re.IGNORECASE)
season_words = ['seasons?', 'saisons?', 'series?']
episode_words = ['episodes?']
season_markers = ['s']
episode_markers = ['e', 'ep']
self.container = PropertiesContainer(enhance=False, canonical_from_pattern=False)
season_words_re = re.compile(build_or_pattern(season_words), re.IGNORECASE)
episode_words_re = re.compile(build_or_pattern(episode_words), re.IGNORECASE)
season_markers_re = re.compile(build_or_pattern(season_markers), re.IGNORECASE)
episode_markers_re = re.compile(build_or_pattern(episode_markers), re.IGNORECASE)
def episode_parser_x(value):
return list_parser(value, 'episodeList', discrete_separators_re=re.compile('x', re.IGNORECASE))
def episode_parser_e(value):
return list_parser(value, 'episodeList', discrete_separators_re=re.compile('e', re.IGNORECASE), fill_gaps=True)
def episode_parser(value):
return list_parser(value, 'episodeList')
def season_parser(value):
return list_parser(value, 'seasonList')
class ResolutionCollisionValidator(object):
@staticmethod
def validate(prop, string, node, match, entry_start, entry_end):
# Invalidate when season or episode is more than 100.
try:
season_value = season_parser(match.group(2))
episode_value = episode_parser_x(match.group(3))
return season_value < 100 or episode_value < 100
except:
# This may occur for 1xAll or patterns like this.
return True
self.container.register_property(None, r'(' + season_words_re.pattern + sep + '?(?P<season>' + numeral + ')' + sep + '?' + season_words_re.pattern + '?)', confidence=1.0, formatter=parse_numeral)
self.container.register_property(None, r'(' + season_words_re.pattern + sep + '?(?P<season>' + digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + digital_numeral + ')*)' + sep + '?' + season_words_re.pattern + '?)' + sep, confidence=1.0, formatter={None: parse_numeral, 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), FormatterValidator('season', lambda x: len(x) > 1 if hasattr(x, '__len__') else False)))
self.container.register_property(None, r'(' + season_markers_re.pattern + '(?P<season>' + digital_numeral + ')[^0-9]?' + sep + '?(?P<episodeNumber>(?:e' + digital_numeral + '(?:' + sep + '?[e-]' + digital_numeral + ')*)))', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser_e, 'season': season_parser}, validator=NoValidator())
self.container.register_property(None, r'(' + season_markers_re.pattern + '(?P<season>' + digital_numeral + ')[^0-9]?' + sep + '?(?P<episodeNumber>(?:e' + digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + digital_numeral + ')*)))', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser, 'season': season_parser}, validator=NoValidator())
self.container.register_property(None, sep + r'((?P<season>' + digital_numeral + ')' + sep + '' + '(?P<episodeNumber>(?:x' + sep + digital_numeral + '(?:' + sep + '[x-]' + digital_numeral + ')*)))', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser_x, 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), ResolutionCollisionValidator()))
self.container.register_property(None, r'((?P<season>' + digital_numeral + ')' + '(?P<episodeNumber>(?:x' + digital_numeral + '(?:[x-]' + digital_numeral + ')*)))', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser_x, 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), ResolutionCollisionValidator()))
self.container.register_property(None, r'(' + season_markers_re.pattern + '(?P<season>' + digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + digital_numeral + ')*))', confidence=0.6, formatter={None: parse_numeral, 'season': season_parser}, validator=NoValidator())
self.container.register_property(None, r'((?P<episodeNumber>' + digital_numeral + ')' + sep + '?v(?P<version>\d+))', confidence=0.6, formatter=parse_numeral)
self.container.register_property('version', sep + r'(V\d+)' + sep, confidence=0.6, formatter=parse_numeral, validator=NoValidator())
self.container.register_property(None, r'(ep' + sep + r'?(?P<episodeNumber>' + digital_numeral + ')' + sep + '?)', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'(ep' + sep + r'?(?P<episodeNumber>' + digital_numeral + ')' + sep + '?v(?P<version>\d+))', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'(' + episode_markers_re.pattern + '(?P<episodeNumber>' + digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + digital_numeral + ')*))', confidence=0.6, formatter={None: parse_numeral, 'episodeNumber': episode_parser})
self.container.register_property(None, r'(' + episode_words_re.pattern + sep + '?(?P<episodeNumber>' + digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + digital_numeral + ')*)' + sep + '?' + episode_words_re.pattern + '?)', confidence=0.8, formatter={None: parse_numeral, 'episodeNumber': episode_parser})
self.container.register_property(None, r'(' + episode_markers_re.pattern + '(?P<episodeNumber>' + digital_numeral + ')' + sep + '?v(?P<version>\d+))', confidence=0.6, formatter={None: parse_numeral, 'episodeNumber': episode_parser})
self.container.register_property(None, r'(' + episode_words_re.pattern + sep + '?(?P<episodeNumber>' + digital_numeral + ')' + sep + '?v(?P<version>\d+))', confidence=0.8, formatter={None: parse_numeral, 'episodeNumber': episode_parser})
self.container.register_property('episodeNumber', r'^' + sep + '+(\d{2}' + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + '\d{2}' + ')*)' + sep, confidence=0.4, formatter=episode_parser)
self.container.register_property('episodeNumber', r'^' + sep + '+0(\d{1,2}' + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + '0\d{1,2}' + ')*)' + sep, confidence=0.4, formatter=episode_parser)
self.container.register_property('episodeNumber', sep + r'(\d{2}' + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + r'\d{2}' + ')*)' + sep + '+$', confidence=0.4, formatter=episode_parser)
self.container.register_property('episodeNumber', sep + r'0(\d{1,2}' + '(?:' + sep + '?' + all_separators_re.pattern + sep + '?' + r'0\d{1,2}' + ')*)' + sep + '+$', confidence=0.4, formatter=episode_parser)
self.container.register_property(None, r'((?P<episodeNumber>' + numeral + ')' + sep + '?' + of_separators_re.pattern + sep + '?(?P<episodeCount>' + numeral + ')(?:' + sep + '?(?:episodes?|eps?))?)', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'((?:episodes?|eps?)' + sep + '?(?P<episodeNumber>' + numeral + ')' + sep + '?' + of_separators_re.pattern + sep + '?(?P<episodeCount>' + numeral + '))', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'((?:seasons?|saisons?|s)' + sep + '?(?P<season>' + numeral + ')' + sep + '?' + of_separators_re.pattern + sep + '?(?P<seasonCount>' + numeral + '))', confidence=0.7, formatter=parse_numeral)
self.container.register_property(None, r'((?P<season>' + numeral + ')' + sep + '?' + of_separators_re.pattern + sep + '?(?P<seasonCount>' + numeral + ')' + sep + '?(?:seasons?|saisons?|s))', confidence=0.7, formatter=parse_numeral)
self.container.register_canonical_properties('other', 'FiNAL', 'Complete', validator=WeakValidator())
self.container.register_property(None, r'[^0-9]((?P<season>' + digital_numeral + ')[^0-9 .-]?-?(?P<other>xAll))', confidence=1.0, formatter={None: parse_numeral, 'other': lambda x: 'Complete', 'season': season_parser}, validator=ChainedValidator(DefaultValidator(), ResolutionCollisionValidator()))
from guessit.patterns.numeral import parse_numeral
range_separators = ['-', 'to', 'a']
discrete_separators = ['&', 'and', 'et']
excluded_separators = ['.'] # Dot cannot serve as a discrete_separator
discrete_sep = sep
for range_separator in range_separators:
discrete_sep = discrete_sep.replace(range_separator, '')
for excluded_separator in excluded_separators:
discrete_sep = discrete_sep.replace(excluded_separator, '')
discrete_separators.append(discrete_sep)
all_separators = list(range_separators)
all_separators.extend(discrete_separators)
range_separators_re = re.compile(build_or_pattern(range_separators), re.IGNORECASE)
discrete_separators_re = re.compile(build_or_pattern(discrete_separators), re.IGNORECASE)
all_separators_re = re.compile(build_or_pattern(all_separators), re.IGNORECASE)
def list_parser(value, property_list_name, discrete_separators_re=discrete_separators_re, range_separators_re=range_separators_re, allow_discrete=False, fill_gaps=False):
discrete_elements = filter(lambda x: x != '', discrete_separators_re.split(value))
discrete_elements = [x.strip() for x in discrete_elements]
proper_discrete_elements = []
i = 0
while i < len(discrete_elements):
if i < len(discrete_elements) - 2 and range_separators_re.match(discrete_elements[i+1]):
proper_discrete_elements.append(discrete_elements[i] + discrete_elements[i+1] + discrete_elements[i+2])
i += 3
else:
def __init__(self):
Transformer.__init__(self, 15)
of_separators = ["of", "sur", "/", "\\"]
of_separators_re = re.compile(build_or_pattern(of_separators, escape=True), re.IGNORECASE)
self.container = PropertiesContainer(enhance=False, canonical_from_pattern=False, remove_duplicates=True)
episode_words = ["episodes?"]
def episode_list_parser(value):
return list_parser(value, "episodeList")
def season_episode_parser(episode_number):
epnum = parse_numeral(episode_number)
if not valid_year(epnum):
if epnum > 100:
season, epnum = epnum // 100, epnum % 100
# episodes which have a season > 50 are most likely errors
# (Simpson is at 25!)
if season > 50:
return None
return {"season": season, "episodeNumber": epnum}
else:
return epnum
self.container.register_property(
["episodeNumber", "season"],
"[0-9]{2,4}",
confidence=0.6,
formatter=season_episode_parser,
disabler=lambda options: options.get("episode_prefer_number") if options else False,
)
self.container.register_property(
["episodeNumber", "season"], "[0-9]{4}", confidence=0.6, formatter=season_episode_parser
)
self.container.register_property(
None,
"(" + build_or_pattern(episode_words) + sep + "?(?P<episodeNumber>" + numeral + "))[^0-9]",
confidence=0.4,
formatter=parse_numeral,
)
self.container.register_property(
None,
r"(?P<episodeNumber>"
+ numeral
+ ")"
+ sep
+ "?"
+ of_separators_re.pattern
+ sep
+ "?(?P<episodeCount>"
+ numeral
+ ")",
confidence=0.6,
formatter=parse_numeral,
)
self.container.register_property(
"episodeNumber",
"[^0-9](\d{2,3}" + "(?:" + sep + "?" + all_separators_re.pattern + sep + "?" + "\d{2,3}" + ")*)",
confidence=0.4,
formatter=episode_list_parser,
disabler=lambda options: not options.get("episode_prefer_number") if options else True,
)
self.container.register_property(
"episodeNumber",
r"^"
+ sep
+ "?(\d{2,3}"
+ "(?:"
+ sep
+ "?"
+ all_separators_re.pattern
+ sep
+ "?"
+ "\d{2,3}"
+ ")*)"
+ sep,
confidence=0.4,
formatter=episode_list_parser,
disabler=lambda options: not options.get("episode_prefer_number") if options else True,
)
self.container.register_property(
"episodeNumber",
sep
+ r"(\d{2,3}"
+ "(?:"
+ sep
+ "?"
+ all_separators_re.pattern
+ sep
+ "?"
+ "\d{2,3}"
+ ")*)"
+ sep
+ "?$",
confidence=0.4,
formatter=episode_list_parser,
disabler=lambda options: not options.get("episode_prefer_number") if options else True,
)
def __init__(self):
Transformer.__init__(self, 20)
of_separators = ['of', 'sur', '/', '\\']
of_separators_re = re.compile(
build_or_pattern(of_separators, escape=True), re.IGNORECASE)
season_words = ['seasons?', 'saisons?', 'series?']
episode_words = ['episodes?']
season_markers = ['s']
episode_markers = ['e', 'ep']
self.container = PropertiesContainer(enhance=False,
canonical_from_pattern=False)
season_words_re = re.compile(build_or_pattern(season_words),
re.IGNORECASE)
episode_words_re = re.compile(build_or_pattern(episode_words),
re.IGNORECASE)
season_markers_re = re.compile(build_or_pattern(season_markers),
re.IGNORECASE)
episode_markers_re = re.compile(build_or_pattern(episode_markers),
re.IGNORECASE)
def episode_parser_x(value):
return list_parser(value,
'episodeList',
discrete_separators_re=re.compile(
'x', re.IGNORECASE))
def episode_parser_e(value):
return list_parser(value,
'episodeList',
discrete_separators_re=re.compile(
'e', re.IGNORECASE),
fill_gaps=True)
def episode_parser(value):
return list_parser(value, 'episodeList')
def season_parser(value):
return list_parser(value, 'seasonList')
class ResolutionCollisionValidator(object):
@staticmethod
def validate(prop, string, node, match, entry_start, entry_end):
return len(match.group(2)) < 3 # limit
self.container.register_property(None,
r'(' + season_words_re.pattern + sep +
'?(?P<season>' + numeral + ')' + sep +
'?' + season_words_re.pattern + '?)',
confidence=1.0,
formatter=parse_numeral)
self.container.register_property(
None,
r'(' + season_words_re.pattern + sep + '?(?P<season>' +
digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern +
sep + '?' + digital_numeral + ')*)' + sep + '?' +
season_words_re.pattern + '?)' + sep,
confidence=1.0,
formatter={
None: parse_numeral,
'season': season_parser
},
validator=ChainedValidator(
DefaultValidator(),
FormatterValidator(
'season', lambda x: len(x) > 1
if hasattr(x, '__len__') else False)))
self.container.register_property(
None,
r'(' + season_markers_re.pattern + '(?P<season>' +
digital_numeral + ')[^0-9]?' + sep + '?(?P<episodeNumber>(?:e' +
digital_numeral + '(?:' + sep + '?[e-]' + digital_numeral +
')*)))',
confidence=1.0,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser_e,
'season': season_parser
},
validator=NoValidator())
self.container.register_property(
None,
r'(' + season_markers_re.pattern + '(?P<season>' +
digital_numeral + ')[^0-9]?' + sep + '?(?P<episodeNumber>(?:e' +
digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern +
sep + '?' + digital_numeral + ')*)))',
confidence=1.0,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser,
'season': season_parser
},
validator=NoValidator())
self.container.register_property(
None,
sep + r'((?P<season>' + digital_numeral + ')' + sep + '' +
'(?P<episodeNumber>(?:x' + sep + digital_numeral + '(?:' + sep +
'[x-]' + digital_numeral + ')*)))',
confidence=1.0,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser_x,
'season': season_parser
},
validator=ChainedValidator(DefaultValidator(),
ResolutionCollisionValidator()))
self.container.register_property(
None,
r'((?P<season>' + digital_numeral + ')' +
'(?P<episodeNumber>(?:x' + digital_numeral + '(?:[x-]' +
digital_numeral + ')*)))',
confidence=1.0,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser_x,
'season': season_parser
},
validator=ChainedValidator(DefaultValidator(),
ResolutionCollisionValidator()))
self.container.register_property(
None,
r'(' + season_markers_re.pattern + '(?P<season>' +
digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern +
sep + '?' + digital_numeral + ')*))',
confidence=0.6,
formatter={
None: parse_numeral,
'season': season_parser
},
validator=NoValidator())
self.container.register_property(None,
r'((?P<episodeNumber>' +
digital_numeral + ')' + sep +
'?v(?P<version>\d+))',
confidence=0.6,
formatter=parse_numeral)
self.container.register_property('version',
sep + r'(V\d+)' + sep,
confidence=0.6,
formatter=parse_numeral,
validator=NoValidator())
self.container.register_property(
None,
r'(ep' + sep + r'?(?P<episodeNumber>' + digital_numeral + ')' +
sep + '?)',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(
None,
r'(ep' + sep + r'?(?P<episodeNumber>' + digital_numeral + ')' +
sep + '?v(?P<version>\d+))',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(
None,
r'(' + episode_markers_re.pattern + '(?P<episodeNumber>' +
digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern +
sep + '?' + digital_numeral + ')*))',
confidence=0.6,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser
})
self.container.register_property(
None,
r'(' + episode_words_re.pattern + sep + '?(?P<episodeNumber>' +
digital_numeral + '(?:' + sep + '?' + all_separators_re.pattern +
sep + '?' + digital_numeral + ')*)' + sep + '?' +
episode_words_re.pattern + '?)',
confidence=0.8,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser
})
self.container.register_property(
None,
r'(' + episode_markers_re.pattern + '(?P<episodeNumber>' +
digital_numeral + ')' + sep + '?v(?P<version>\d+))',
confidence=0.6,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser
})
self.container.register_property(
None,
r'(' + episode_words_re.pattern + sep + '?(?P<episodeNumber>' +
digital_numeral + ')' + sep + '?v(?P<version>\d+))',
confidence=0.8,
formatter={
None: parse_numeral,
'episodeNumber': episode_parser
})
self.container.register_property('episodeNumber',
r'^' + sep + '+(\d{2}' + '(?:' + sep +
'?' + all_separators_re.pattern +
sep + '?' + '\d{2}' + ')*)' + sep,
confidence=0.4,
formatter=episode_parser)
self.container.register_property(
'episodeNumber',
r'^' + sep + '+0(\d{1,2}' + '(?:' + sep + '?' +
all_separators_re.pattern + sep + '?' + '0\d{1,2}' + ')*)' + sep,
confidence=0.4,
formatter=episode_parser)
self.container.register_property('episodeNumber',
sep + r'(\d{2}' + '(?:' + sep + '?' +
all_separators_re.pattern + sep +
'?' + r'\d{2}' + ')*)' + sep + '+$',
confidence=0.4,
formatter=episode_parser)
self.container.register_property(
'episodeNumber',
sep + r'0(\d{1,2}' + '(?:' + sep + '?' +
all_separators_re.pattern + sep + '?' + r'0\d{1,2}' + ')*)' + sep +
'+$',
confidence=0.4,
formatter=episode_parser)
self.container.register_property(
None,
r'((?P<episodeNumber>' + numeral + ')' + sep + '?' +
of_separators_re.pattern + sep + '?(?P<episodeCount>' + numeral +
')(?:' + sep + '?(?:episodes?|eps?))?)',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(
None,
r'((?:episodes?|eps?)' + sep + '?(?P<episodeNumber>' + numeral +
')' + sep + '?' + of_separators_re.pattern + sep +
'?(?P<episodeCount>' + numeral + '))',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(None,
r'((?:seasons?|saisons?|s)' + sep +
'?(?P<season>' + numeral + ')' + sep +
'?' + of_separators_re.pattern + sep +
'?(?P<seasonCount>' + numeral + '))',
confidence=0.7,
formatter=parse_numeral)
self.container.register_property(
None,
r'((?P<season>' + numeral + ')' + sep + '?' +
of_separators_re.pattern + sep + '?(?P<seasonCount>' + numeral +
')' + sep + '?(?:seasons?|saisons?|s))',
confidence=0.7,
formatter=parse_numeral)
self.container.register_canonical_properties('other',
'FiNAL',
'Complete',
validator=WeakValidator())
self.container.register_property(
None,
r'[^0-9]((?P<season>' + digital_numeral +
')[^0-9 .-]?-?(?P<other>xAll))',
confidence=1.0,
formatter={
None: parse_numeral,
'other': lambda x: 'Complete',
'season': season_parser
},
validator=ChainedValidator(DefaultValidator(),
ResolutionCollisionValidator()))
def __init__(self):
Transformer.__init__(self, 35)
self.container = PropertiesContainer()
self.qualities = QualitiesContainer()
def register_property(propname, props, **kwargs):
"""props a dict of {value: [patterns]}"""
for canonical_form, patterns in props.items():
if isinstance(patterns, tuple):
patterns2, pattern_kwarg = patterns
if kwargs:
current_kwarg = dict(kwargs)
current_kwarg.update(pattern_kwarg)
else:
current_kwarg = dict(pattern_kwarg)
current_kwarg["canonical_form"] = canonical_form
self.container.register_property(propname, *patterns2, **current_kwarg)
elif kwargs:
current_kwarg = dict(kwargs)
current_kwarg["canonical_form"] = canonical_form
self.container.register_property(propname, *patterns, **current_kwarg)
else:
self.container.register_property(propname, *patterns, canonical_form=canonical_form)
def register_quality(propname, quality_dict):
"""props a dict of {canonical_form: quality}"""
for canonical_form, quality in quality_dict.items():
self.qualities.register_quality(propname, canonical_form, quality)
# http://en.wikipedia.org/wiki/Pirated_movie_release_types
register_property(
"format",
{
"VHS": ["VHS", "VHS-Rip"],
"Cam": ["CAM", "CAMRip", "HD-CAM"],
#'Telesync': ['TELESYNC', 'PDVD'],
"Telesync": (["TS", "HD-TS"], {"confidence": 0.4}),
"Workprint": ["WORKPRINT", "WP"],
"Telecine": ["TELECINE", "TC"],
"PPV": ["PPV", "PPV-Rip"], # Pay Per View
"TV": ["SD-TV", "SD-TV-Rip", "Rip-SD-TV", "TV-Rip", "Rip-TV"],
"DVB": ["DVB-Rip", "DVB", "PD-TV"],
"DVD": ["DVD", "DVD-Rip", "VIDEO-TS", "DVD-R", "DVD-9", "DVD-5"],
"HDTV": ["HD-TV", "TV-RIP-HD", "HD-TV-RIP", "HD-RIP"],
"VOD": ["VOD", "VOD-Rip"],
"WEBRip": ["WEB-Rip"],
"WEB-DL": ["WEB-DL", "WEB-HD", "WEB"],
"HD-DVD": ["HD-DVD-Rip", "HD-DVD"],
"BluRay": ["Blu-ray(?:-Rip)?", "B[DR]", "B[DR]-Rip", "BD[59]", "BD25", "BD50"],
},
)
register_quality(
"format",
{
"VHS": -100,
"Cam": -90,
"Telesync": -80,
"Workprint": -70,
"Telecine": -60,
"PPV": -50,
"TV": -30,
"DVB": -20,
"DVD": 0,
"HDTV": 20,
"VOD": 40,
"WEBRip": 50,
"WEB-DL": 60,
"HD-DVD": 80,
"BluRay": 100,
},
)
register_property(
"screenSize",
{
"360p": ["(?:\d{3,}(?:\\|\/|x|\*))?360(?:i|p?x?)"],
"368p": ["(?:\d{3,}(?:\\|\/|x|\*))?368(?:i|p?x?)"],
"480p": ["(?:\d{3,}(?:\\|\/|x|\*))?480(?:i|p?x?)"],
#'480p': (['hr'], {'confidence': 0.2}), # duplicate dict key
"576p": ["(?:\d{3,}(?:\\|\/|x|\*))?576(?:i|p?x?)"],
"720p": ["(?:\d{3,}(?:\\|\/|x|\*))?720(?:i|p?x?)"],
"900p": ["(?:\d{3,}(?:\\|\/|x|\*))?900(?:i|p?x?)"],
"1080i": ["(?:\d{3,}(?:\\|\/|x|\*))?1080i"],
"1080p": ["(?:\d{3,}(?:\\|\/|x|\*))?1080p?x?"],
"4K": ["(?:\d{3,}(?:\\|\/|x|\*))?2160(?:i|p?x?)"],
},
validator=ChainedValidator(DefaultValidator(), OnlyOneValidator()),
)
_digits_re = re.compile("\d+")
def resolution_formatter(value):
digits = _digits_re.findall(value)
return "x".join(digits)
self.container.register_property(
"screenSize", "\d{3,4}-?[x\*]-?\d{3,4}", canonical_from_pattern=False, formatter=resolution_formatter
)
register_quality(
"screenSize",
{
"360p": -300,
"368p": -200,
"480p": -100,
"576p": 0,
"720p": 100,
"900p": 130,
"1080i": 180,
"1080p": 200,
"4K": 400,
},
)
_videoCodecProperty = {
"Real": ["Rv\d{2}"], # http://en.wikipedia.org/wiki/RealVideo
"Mpeg2": ["Mpeg2"],
"DivX": ["DVDivX", "DivX"],
"XviD": ["XviD"],
"h264": ["[hx]-264(?:-AVC)?", "MPEG-4(?:-AVC)"],
"h265": ["[hx]-265(?:-HEVC)?", "HEVC"],
}
register_property("videoCodec", _videoCodecProperty)
register_quality("videoCodec", {"Real": -50, "Mpeg2": -30, "DivX": -10, "XviD": 0, "h264": 100, "h265": 150})
# http://blog.mediacoderhq.com/h264-profiles-and-levels/
# http://fr.wikipedia.org/wiki/H.264
self.container.register_property(
"videoProfile", "BP", validator=LeavesValidator(lambdas=[lambda node: "videoCodec" in node.guess])
)
self.container.register_property(
"videoProfile",
"XP",
"EP",
canonical_form="XP",
validator=LeavesValidator(lambdas=[lambda node: "videoCodec" in node.guess]),
)
self.container.register_property(
"videoProfile", "MP", validator=LeavesValidator(lambdas=[lambda node: "videoCodec" in node.guess])
)
self.container.register_property(
"videoProfile",
"HP",
"HiP",
canonical_form="HP",
validator=LeavesValidator(lambdas=[lambda node: "videoCodec" in node.guess]),
)
self.container.register_property("videoProfile", "10.?bit", "Hi10P", canonical_form="10bit")
self.container.register_property("videoProfile", "8.?bit", canonical_form="8bit")
self.container.register_property(
"videoProfile", "Hi422P", validator=LeavesValidator(lambdas=[lambda node: "videoCodec" in node.guess])
)
self.container.register_property(
"videoProfile", "Hi444PP", validator=LeavesValidator(lambdas=[lambda node: "videoCodec" in node.guess])
)
register_quality(
"videoProfile", {"BP": -20, "XP": -10, "MP": 0, "HP": 10, "10bit": 15, "Hi422P": 25, "Hi444PP": 35}
)
# has nothing to do here (or on filenames for that matter), but some
# releases use it and it helps to identify release groups, so we adapt
register_property("videoApi", {"DXVA": ["DXVA"]})
register_property(
"audioCodec",
{
"MP3": ["MP3", "LAME", "LAME(?:\d)+-(?:\d)+"],
"DolbyDigital": ["DD"],
"AAC": ["AAC"],
"AC3": ["AC3"],
"Flac": ["FLAC"],
"DTS": (["DTS"], {"validator": LeftValidator()}),
"TrueHD": ["True-HD"],
},
)
register_quality(
"audioCodec", {"MP3": 10, "DolbyDigital": 30, "AAC": 35, "AC3": 40, "Flac": 45, "DTS": 60, "TrueHD": 70}
)
self.container.register_property(
"audioProfile",
"HD",
validator=LeavesValidator(lambdas=[lambda node: node.guess.get("audioCodec") == "DTS"]),
)
self.container.register_property(
"audioProfile",
"HD-MA",
canonical_form="HDMA",
validator=LeavesValidator(lambdas=[lambda node: node.guess.get("audioCodec") == "DTS"]),
)
self.container.register_property(
"audioProfile",
"HE",
validator=LeavesValidator(lambdas=[lambda node: node.guess.get("audioCodec") == "AAC"]),
)
self.container.register_property(
"audioProfile",
"LC",
validator=LeavesValidator(lambdas=[lambda node: node.guess.get("audioCodec") == "AAC"]),
)
self.container.register_property(
"audioProfile",
"HQ",
validator=LeavesValidator(lambdas=[lambda node: node.guess.get("audioCodec") == "AC3"]),
)
register_quality("audioProfile", {"HD": 20, "HDMA": 50, "LC": 0, "HQ": 0, "HE": 20})
register_property(
"audioChannels",
{
"7.1": ["7[\W_]1", "7ch", "8ch"],
"5.1": ["5[\W_]1", "5ch", "6ch"],
"2.0": ["2[\W_]0", "2ch", "stereo"],
"1.0": ["1[\W_]0", "1ch", "mono"],
},
)
register_quality("audioChannels", {"7.1": 200, "5.1": 100, "2.0": 0, "1.0": -100})
self.container.register_property("episodeFormat", r"Minisodes?", canonical_form="Minisode")
self.container.register_property("crc32", "(?:[a-fA-F]|[0-9]){8}", enhance=False, canonical_from_pattern=False)
part_words = ["pt", "part"]
self.container.register_property(
None,
"(" + build_or_pattern(part_words) + sep + "?(?P<part>" + numeral + "))[^0-9]",
enhance=False,
canonical_from_pattern=False,
confidence=0.4,
formatter=parse_numeral,
)
register_property(
"other",
{
"AudioFix": ["Audio-Fix", "Audio-Fixed"],
"SyncFix": ["Sync-Fix", "Sync-Fixed"],
"DualAudio": ["Dual-Audio"],
"WideScreen": ["ws", "wide-screen"],
"Netflix": ["Netflix", "NF"],
},
)
self.container.register_property(
"other",
"Real",
"Fix",
canonical_form="Proper",
validator=ChainedValidator(FullMatchValidator(), NeighborValidator()),
)
self.container.register_property("other", "Proper", "Repack", "Rerip", canonical_form="Proper")
self.container.register_property(
"other",
"Fansub",
canonical_form="Fansub",
validator=ChainedValidator(FullMatchValidator(), NeighborValidator()),
)
self.container.register_property(
"other",
"Fastsub",
canonical_form="Fastsub",
validator=ChainedValidator(FullMatchValidator(), NeighborValidator()),
)
self.container.register_property("other", "(?:Seasons?" + sep + "?)?Complete", canonical_form="Complete")
self.container.register_property("other", "R5", "RC", canonical_form="R5")
self.container.register_property("other", "Pre-Air", "Preair", canonical_form="Preair")
self.container.register_property("other", "CC") # Close Caption
self.container.register_property("other", "LD", "MD") # Line/Mic Dubbed
self.container.register_canonical_properties(
"other", "Screener", "Remux", "3D", "HD", "mHD", "HDLight", "HQ", "DDC", "HR", "PAL", "SECAM", "NTSC"
)
self.container.register_canonical_properties(
"other", "Limited", "Complete", "Classic", "Unrated", "LiNE", "Bonus", "Trailer", validator=WeakValidator()
)
for prop in self.container.get_properties("format"):
self.container.register_property("other", prop.pattern + "(-?Scr(?:eener)?)", canonical_form="Screener")
for exts in (subtitle_exts, info_exts, video_exts):
for container in exts:
self.container.register_property("container", container, confidence=0.3)
