def rules():
"""Return all custom rules to be applied to guessit default api.
IMPORTANT! The order is important.
Certain rules needs to be executed first, and others should be executed at the end.
DO NOT define priority or dependency in each rule, it can become a mess. Better to just define the correct order
in this method.
Builder for rebulk object.
:return: Created Rebulk object
:rtype: Rebulk
"""
return Rebulk().rules(
BlacklistedReleaseGroup,
FixTvChaosUkWorkaround,
FixAnimeReleaseGroup,
FixInvalidAbsoluteReleaseGroups,
AnimeWithSeasonAbsoluteEpisodeNumbers,
AnimeAbsoluteEpisodeNumbers,
AbsoluteEpisodeNumbers,
PartsAsEpisodeNumbers,
RemoveInvalidEpisodeSeparator,
CreateAliasWithAlternativeTitles,
CreateAliasWithCountryOrYear,
ReleaseGroupPostProcessor,
FixParentFolderReplacingTitle,
FixMultipleSources,
AudioCodecStandardizer,
SourceStandardizer,
VideoEncoderRule,
CreateProperTags,
AvoidMultipleValuesRule,
)
def other():
"""Other property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
rebulk.defaults(name='other', validator=seps_surround)
# https://github.com/guessit-io/guessit/issues/300
rebulk.regex(r'Re-?Enc(?:oded)?', value='Re-Encoded')
rebulk.regex('DIRFIX', value='DirFix')
rebulk.regex('INTERNAL', value='Internal')
rebulk.regex(r'(?:HD)?iTunes(?:HD)?', value='iTunes')
rebulk.regex(r'UNCENSORED', value='Uncensored')
rebulk.regex(r'MULTi', value='Multi Language')
rebulk.regex('HC', value='Hardcoded subtitles')
# Discarded:
rebulk.regex('DownRev', 'small-size', private=True)
rebulk.rules(ValidateHardcodedSubs)
return rebulk
def container():
"""Builder for rebulk object.
:return: Created Rebulk object
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE).string_defaults(ignore_case=True)
rebulk.defaults(name='container',
tags=['extension'],
conflict_solver=lambda match, other: other
if other.name in ['format', 'video_codec'] or
other.name == 'container' and 'extension' not in other.tags
else '__default__')
nzb = ['nzb']
rebulk.regex(r'\.' + build_or_pattern(nzb) + '$', exts=nzb, tags=['extension', 'torrent'])
rebulk.defaults(name='container',
validator=seps_surround,
formatter=lambda s: s.upper(),
conflict_solver=lambda match, other: match
if other.name in ['format',
'video_codec'] or other.name == 'container' and 'extension' in other.tags
else '__default__')
rebulk.string(*nzb, tags=['nzb'])
return rebulk
def subtitle_language():
"""Subtitle language property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE | re.UNICODE,
abbreviations=[alt_dash])
rebulk.defaults(name='subtitle_language', validator=seps_surround)
# special handling
rebulk.regex(r'Legenda(?:s|do)?@PT-?BR',
value=babelfish.Language('por', 'BR'))
rebulk.regex(r'Legenda(?:s|do)?@PT(?!-?BR)',
value=babelfish.Language('por'))
rebulk.regex('Subtitulado@ESP(?:a[nñ]ol)?@Spanish',
'Subtitulado@ESP(?:a[nñ]ol)?',
value=babelfish.Language('spa'),
conflict_solver=lambda match, other: other
if other.name == 'language' else '__default__')
# undefined language
rebulk.regex('Subtitles',
'Legenda(?:s|do)',
'Subbed',
'Sub(?:title)?s?@Latino',
value='und',
formatter=babelfish.Language,
tags='subtitle.undefined')
rebulk.rules(RemoveSubtitleUndefined)
return rebulk
def format_():
"""Format property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
rebulk.defaults(name='format', tags='video-codec-prefix')
# More accurate formats
rebulk.regex('BD-?Rip',
'BD(?=-?Mux)',
value='BDRip',
conflict_solver=lambda match, other: other
if other.name == 'format' else '__default__')
rebulk.regex('BD(?!\d)',
value='BDRip',
validator=seps_surround,
conflict_solver=lambda match, other: other
if other.name == 'format' else '__default__')
rebulk.regex('BR-?Rip',
'BR(?=-?Mux)',
value='BRRip',
conflict_solver=lambda match, other: other
if other.name == 'format' else '__default__')
rebulk.regex('DVD-?Rip',
value='DVDRip',
conflict_solver=lambda match, other: other
if other.name == 'format' else '__default__')
rebulk.regex('DVD\d', value='DVD')
return rebulk
def format_():
"""Format property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
rebulk.defaults(name='format')
# https://github.com/guessit-io/guessit/issues/307
rebulk.regex('HDTV-?Mux', value='HDTV')
rebulk.regex('B[RD]-?Mux', 'Blu-?ray-?Mux', value='BluRay')
rebulk.regex('DVD-?Mux', value='DVD')
rebulk.regex('WEB-?Mux',
'DL-?WEB-?Mux',
'WEB-?DL-?Mux',
'DL-?Mux',
value='WEB-DL')
# https://github.com/guessit-io/guessit/issues/315
rebulk.regex('WEB-?DL-?Rip', value='WEBRip')
rebulk.regex('WEB-?Cap', value='WEBCap')
rebulk.regex('DSR', 'DS-?Rip', 'SAT-?Rip', 'DTH-?Rip', value='DSRip')
rebulk.regex('LDTV', value='TV')
rebulk.regex('DVD\d', value='DVD')
return rebulk
def audio_codec():
"""Audio codec property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(name='audio_codec', flags=re.IGNORECASE)
rebulk.regex(r'DDP', value='AC3')
rebulk.rules(Ac3Rule, AudioValidatorRule, HqConflictRule)
return rebulk
def screen_size():
"""Screen size property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE)
rebulk.defaults(name='screen_size', validator=seps_surround)
rebulk.regex(r'(?:\d{3,}(?:x|\*))?4320(?:p?x?)', value='4320p')
return rebulk
def screen_size():
"""Screen size property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE)
rebulk.defaults(name='screen_size', validator=seps_surround)
rebulk.regex('NetflixUHD', value='4k')
return rebulk
def other():
"""Other property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
rebulk.defaults(name='other', validator=seps_surround)
rebulk.regex('F1', value='Formula One',
conflict_solver=lambda match, other: other if other.name == 'film' else '__default__')
# Discarded:
rebulk.regex('DownRev', 'small-size', private=True)
return rebulk
def language():
"""Language property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
rebulk.defaults(name='language', validator=seps_surround)
rebulk.regex('SPANISH-?AUDIO',
r'(?:Espa[.]ol-)?castellano',
value=babelfish.Language('spa'))
rebulk.regex('german-dubbed',
'dubbed-german',
value=babelfish.Language('deu'))
rebulk.regex('english-dubbed', value=babelfish.Language('eng'))
rebulk.regex('dublado', value='und', formatter=babelfish.Language)
return rebulk
def size():
"""Size property.
Remove when https://github.com/guessit-io/guessit/issues/299 is fixed.
:return:
:rtype: Rebulk
"""
def format_size(value):
return re.sub(r'(?<=\d)[.](?=[^\d])', '', value.upper())
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
rebulk.defaults(name='size', validator=seps_surround)
rebulk.regex(r'\d+\.?[mgt]b',
r'\d+\.\d+[mgt]b',
formatter=format_size,
tags=['release-group-prefix'])
return rebulk
def blacklist():
"""Blacklisted patterns.
All blacklisted patterns.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
rebulk.defaults(name='blacklist', validator=seps_surround,
conflict_solver=lambda match, other: other if other.name != 'blacklist' else '__default__')
rebulk.regex(r'(?:(?:\[\d+/\d+\])-+)?\.".+".*', tags=['blacklist-01'])
rebulk.regex(r'vol\d{2,3}\+\d{2,3}.*', tags=['blacklist-02'])
rebulk.regex(r'(?:nzb|par2)-+\d+\.of\.\d+.*', tags=['blacklist-03'])
rebulk.regex(r'(?:(?:nzb|par2)-+)?\d+\.of\.\d+.*', tags=['blacklist-03', 'should-have-container-before'])
rebulk.rules(ValidateBlacklist, RemoveBlacklisted)
return rebulk
def format_():
"""Format property.
:return:
:rtype: Rebulk
"""
rebulk = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
rebulk.defaults(name='format')
# More accurate formats
rebulk.regex('BD(?!\d)',
'BD-?Rip',
'BD-?Mux',
'BD-?Rip-?Mux',
value='BDRip',
validator=seps_surround,
conflict_solver=lambda match, other: other
if other.name == 'format' else '__default__')
rebulk.regex('BR-?Rip',
'BR-?Mux',
'BR-?Rip-?Mux',
value='BRRip',
conflict_solver=lambda match, other: other
if other.name == 'format' else '__default__')
rebulk.regex('DVD-?Rip',
value='DVDRip',
conflict_solver=lambda match, other: other
if other.name == 'format' else '__default__')
# https://github.com/guessit-io/guessit/issues/307
rebulk.regex('HDTV-?Mux', value='HDTV')
rebulk.regex('Blu-?ray-?Mux', value='BluRay')
rebulk.regex('DVD-?Mux', value='DVD')
rebulk.regex('WEB-?Mux',
'DL-?WEB-?Mux',
'WEB-?DL-?Mux',
'DL-?Mux',
value='WEB-DL')
rebulk.regex('DVD\d', value='DVD')
return rebulk
