SickGear/lib/subliminal/subtitle.py

278 lines
10 KiB
Python

# -*- coding: utf-8 -*-
from __future__ import unicode_literals
import logging
import os.path
import babelfish
import chardet
import guessit.matchtree
import guessit.transfo
import pysrt
from .video import Episode, Movie
logger = logging.getLogger(__name__)
class Subtitle(object):
"""Base class for subtitle
:param language: language of the subtitle
:type language: :class:`babelfish.Language`
:param bool hearing_impaired: `True` if the subtitle is hearing impaired, `False` otherwise
:param page_link: link to the web page from which the subtitle can be downloaded, if any
:type page_link: string or None
"""
def __init__(self, language, hearing_impaired=False, page_link=None):
self.language = language
self.hearing_impaired = hearing_impaired
self.page_link = page_link
#: Content as bytes
self.content = None
#: Encoding to decode with when accessing :attr:`text`
self.encoding = None
@property
def guessed_encoding(self):
"""Guessed encoding using the language, falling back on chardet"""
# always try utf-8 first
encodings = ['utf-8']
# add language-specific encodings
if self.language.alpha3 == 'zho':
encodings.extend(['gb18030', 'big5'])
elif self.language.alpha3 == 'jpn':
encodings.append('shift-jis')
elif self.language.alpha3 == 'ara':
encodings.append('windows-1256')
elif self.language.alpha3 == 'heb':
encodings.append('windows-1255')
elif self.language.alpha3 == 'tur':
encodings.extend(['iso-8859-9', 'windows-1254'])
else:
encodings.append('latin-1')
# try to decode
for encoding in encodings:
try:
self.content.decode(encoding)
return encoding
except UnicodeDecodeError:
pass
# fallback on chardet
logger.warning('Could not decode content with encodings %r', encodings)
return chardet.detect(self.content)['encoding']
@property
def text(self):
"""Content as string
If :attr:`encoding` is None, the encoding is guessed with :attr:`guessed_encoding`
"""
if not self.content:
return ''
return self.content.decode(self.encoding or self.guessed_encoding, errors='replace')
@property
def is_valid(self):
"""Check if a subtitle text is a valid SubRip format"""
try:
pysrt.from_string(self.text, error_handling=pysrt.ERROR_RAISE)
return True
except pysrt.Error as e:
if e.args[0] > 80:
return True
except:
logger.exception('Unexpected error when validating subtitle')
return False
def compute_matches(self, video):
"""Compute the matches of the subtitle against the `video`
:param video: the video to compute the matches against
:type video: :class:`~subliminal.video.Video`
:return: matches of the subtitle
:rtype: set
"""
raise NotImplementedError
def compute_score(self, video):
"""Compute the score of the subtitle against the `video`
There are equivalent matches so that a provider can match one element or its equivalent. This is
to give all provider a chance to have a score in the same range without hurting quality.
* Matching :class:`~subliminal.video.Video`'s `hashes` is equivalent to matching everything else
* Matching :class:`~subliminal.video.Episode`'s `season` and `episode`
is equivalent to matching :class:`~subliminal.video.Episode`'s `title`
* Matching :class:`~subliminal.video.Episode`'s `tvdb_id` is equivalent to matching
:class:`~subliminal.video.Episode`'s `series`
:param video: the video to compute the score against
:type video: :class:`~subliminal.video.Video`
:return: score of the subtitle
:rtype: int
"""
score = 0
# compute matches
initial_matches = self.compute_matches(video)
matches = initial_matches.copy()
# hash is the perfect match
if 'hash' in matches:
score = video.scores['hash']
else:
# remove equivalences
if isinstance(video, Episode):
if 'imdb_id' in matches:
matches -= {'series', 'tvdb_id', 'season', 'episode', 'title', 'year'}
if 'tvdb_id' in matches:
matches -= {'series', 'year'}
if 'title' in matches:
matches -= {'season', 'episode'}
# add other scores
score += sum((video.scores[match] for match in matches))
logger.info('Computed score %d with matches %r', score, initial_matches)
return score
def __repr__(self):
return '<%s [%s]>' % (self.__class__.__name__, self.language)
def get_subtitle_path(video_path, language=None):
"""Create the subtitle path from the given `video_path` and `language`
:param string video_path: path to the video
:param language: language of the subtitle to put in the path
:type language: :class:`babelfish.Language` or None
:return: path of the subtitle
:rtype: string
"""
subtitle_path = os.path.splitext(video_path)[0]
if language is not None:
try:
return subtitle_path + '.%s.%s' % (language.alpha2, 'srt')
except babelfish.LanguageConvertError:
return subtitle_path + '.%s.%s' % (language.alpha3, 'srt')
return subtitle_path + '.srt'
def compute_guess_matches(video, guess):
"""Compute matches between a `video` and a `guess`
:param video: the video to compute the matches on
:type video: :class:`~subliminal.video.Video`
:param guess: the guess to compute the matches on
:type guess: :class:`guessit.Guess`
:return: matches of the `guess`
:rtype: set
"""
matches = set()
if isinstance(video, Episode):
# series
if video.series and 'series' in guess and guess['series'].lower() == video.series.lower():
matches.add('series')
# season
if video.season and 'seasonNumber' in guess and guess['seasonNumber'] == video.season:
matches.add('season')
# episode
if video.episode and 'episodeNumber' in guess and guess['episodeNumber'] == video.episode:
matches.add('episode')
# year
if video.year == guess.get('year'): # count "no year" as an information
matches.add('year')
elif isinstance(video, Movie):
# year
if video.year and 'year' in guess and guess['year'] == video.year:
matches.add('year')
# title
if video.title and 'title' in guess and guess['title'].lower() == video.title.lower():
matches.add('title')
# release group
if video.release_group and 'releaseGroup' in guess and guess['releaseGroup'].lower() == video.release_group.lower():
matches.add('release_group')
# screen size
if video.resolution and 'screenSize' in guess and guess['screenSize'] == video.resolution:
matches.add('resolution')
# format
if video.format and 'format' in guess and guess['format'].lower() == video.format.lower():
matches.add('format')
# video codec
if video.video_codec and 'videoCodec' in guess and guess['videoCodec'] == video.video_codec:
matches.add('video_codec')
# audio codec
if video.audio_codec and 'audioCodec' in guess and guess['audioCodec'] == video.audio_codec:
matches.add('audio_codec')
return matches
def compute_guess_properties_matches(video, string, propertytype):
"""Compute matches between a `video` and properties of a certain property type
:param video: the video to compute the matches on
:type video: :class:`~subliminal.video.Video`
:param string string: the string to check for a certain property type
:param string propertytype: the type of property to check (as defined in guessit)
:return: matches of a certain property type (but will only be 1 match because we are checking for 1 property type)
:rtype: set
Supported property types: result of guessit.transfo.guess_properties.GuessProperties().supported_properties()
[u'audioProfile',
u'videoCodec',
u'container',
u'format',
u'episodeFormat',
u'videoApi',
u'screenSize',
u'videoProfile',
u'audioChannels',
u'other',
u'audioCodec']
"""
matches = set()
# We only check for the property types relevant for us
if propertytype == 'screenSize' and video.resolution:
for prop in guess_properties(string, propertytype):
if prop.lower() == video.resolution.lower():
matches.add('resolution')
elif propertytype == 'format' and video.format:
for prop in guess_properties(string, propertytype):
if prop.lower() == video.format.lower():
matches.add('format')
elif propertytype == 'videoCodec' and video.video_codec:
for prop in guess_properties(string, propertytype):
if prop.lower() == video.video_codec.lower():
matches.add('video_codec')
elif propertytype == 'audioCodec' and video.audio_codec:
for prop in guess_properties(string, propertytype):
if prop.lower() == video.audio_codec.lower():
matches.add('audio_codec')
return matches
def guess_properties(string, propertytype):
properties = set()
if string:
tree = guessit.matchtree.MatchTree(string)
guessit.transfo.guess_properties.GuessProperties().process(tree)
properties = set(n.guess[propertytype] for n in tree.nodes() if propertytype in n.guess)
return properties
def fix_line_endings(content):
"""Fix line ending of `content` by changing it to \n
:param bytes content: content of the subtitle
:return: the content with fixed line endings
:rtype: bytes
"""
return content.replace(b'\r\n', b'\n').replace(b'\r', b'\n')