Python containers.PropertiesContainer类代码示例

class GuessEpisodeSpecial(Transformer):
    def __init__(self):
        Transformer.__init__(self, -205)
        self.container = PropertiesContainer()
        self.container.register_property('special', 'Special', 'Bonus', 'Omake', 'Ova', 'Oav', 'Pilot', 'Unaired')
        self.container.register_property('special', 'Extras?', canonical_form='Extras')

    def guess_special(self, string, node=None, options=None):
        properties = self.container.find_properties(string, node, 'special', multiple=True)
        guesses = self.container.as_guess(properties, multiple=True)
        return guesses

    def second_pass_options(self, mtree, options=None):
        if not mtree.guess.get('type', '').startswith('episode'):
            for unidentified_leaf in mtree.unidentified_leaves():
                properties = self.container.find_properties(unidentified_leaf.value, unidentified_leaf, 'special')
                guess = self.container.as_guess(properties)
                if guess:
                    return {'type': 'episode'}
        return None

    def supported_properties(self):
        return self.container.get_supported_properties()

    def process(self, mtree, options=None):
        if mtree.guess.get('type', '').startswith('episode') and (not mtree.info.get('episodeNumber') or mtree.info.get('season') == 0):
            for title_leaf in mtree.leaves_containing('title'):
                guesses = self.guess_special(title_leaf.value, title_leaf, options)
                for guess in guesses:
                    found_guess(title_leaf, guess, update_guess=False)
            for unidentified_leaf in mtree.unidentified_leaves():
                guesses = self.guess_special(unidentified_leaf.value, unidentified_leaf, options)
                for guess in guesses:
                    found_guess(unidentified_leaf, guess, update_guess=False)
        return None

 def guess_regexps_id(self, string, node=None, options=None):
     container = PropertiesContainer(
         enhance=False,
         canonical_from_pattern=False)
     for regexp in options.get("id_regexps"):
         container.register_property(
             'regexpId',
             regexp,
             confidence=1.0,
             validator=NoValidator())
     found = container.find_properties(string, node, options)
     return container.as_guess(found, string)

 def __init__(self):
     Transformer.__init__(self, -205)
     self.container = PropertiesContainer()
     self.container.register_property(
         "episodeDetails", "Special", "Bonus", "Omake", "Ova", "Oav", "Pilot", "Unaired"
     )
     self.container.register_property("episodeDetails", "Extras?", canonical_form="Extras")

    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, 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, -190)
        self.container = PropertiesContainer(canonical_from_pattern=False)
        self._allowed_groupname_pattern = '[\[email protected]#€£$&]'
        self._forbidden_groupname_lambda = [lambda elt: elt in ['rip', 'by', 'for', 'par', 'pour', 'bonus'],
                               lambda elt: self._is_number(elt),
                               ]
        # If the previous property in this list, the match will be considered as safe
        # and group name can contain a separator.
        self.previous_safe_properties = ['videoCodec', 'format', 'videoApi', 'audioCodec', 'audioProfile', 'videoProfile', 'audioChannels']

        self.container.sep_replace_char = '-'
        self.container.canonical_from_pattern = False
        self.container.enhance = True
        self.container.register_property('releaseGroup', self._allowed_groupname_pattern + '+')
        self.container.register_property('releaseGroup', self._allowed_groupname_pattern + '+-' + self._allowed_groupname_pattern + '+')

    def expected_series(self, string, node=None, options=None):
        container = PropertiesContainer(enhance=True, canonical_from_pattern=False)

        for expected_serie in options.get('expected_series'):
            if expected_serie.startswith('re:'):
                expected_serie = expected_serie[3:]
                expected_serie = expected_serie.replace(' ', '-')
                container.register_property('series', expected_serie, enhance=True)
            else:
                expected_serie = re.escape(expected_serie)
                container.register_property('series', expected_serie, enhance=False)

        found = container.find_properties(string, node, options)
        return container.as_guess(found, string)

class GuessEpisodeDetails(Transformer):
    def __init__(self):
        Transformer.__init__(self, -205)
        self.container = PropertiesContainer()
        self.container.register_property(
            "episodeDetails", "Special", "Bonus", "Omake", "Ova", "Oav", "Pilot", "Unaired"
        )
        self.container.register_property("episodeDetails", "Extras?", canonical_form="Extras")

    def guess_details(self, string, node=None, options=None):
        properties = self.container.find_properties(string, node, options, "episodeDetails", multiple=True)
        guesses = self.container.as_guess(properties, multiple=True)
        return guesses

    def second_pass_options(self, mtree, options=None):
        if not mtree.guess.get("type", "").startswith("episode"):
            for unidentified_leaf in mtree.unidentified_leaves():
                properties = self.container.find_properties(
                    unidentified_leaf.value, unidentified_leaf, options, "episodeDetails"
                )
                guess = self.container.as_guess(properties)
                if guess:
                    return {"type": "episode"}
        return None

    def supported_properties(self):
        return self.container.get_supported_properties()

    def process(self, mtree, options=None):
        if mtree.guess.get("type", "").startswith("episode") and (
            not mtree.info.get("episodeNumber") or mtree.info.get("season") == 0
        ):

            for leaf in itertools.chain(mtree.leaves_containing("title"), mtree.unidentified_leaves()):
                guesses = self.guess_details(leaf.value, leaf, options)
                for guess in guesses:
                    found_guess(leaf, guess, update_guess=False)

        return None

    def __init__(self):
        Transformer.__init__(self, 25)

        self.container = PropertiesContainer(canonical_from_pattern=False)

        self.container.register_property(None, 'cd' + _psep + '(?P<cdNumber>[0-9])(?:' + _psep + 'of' + _psep + '(?P<cdNumberTotal>[0-9]))?', confidence=1.0, enhance=False, global_span=True, formatter=parse_numeral)
        self.container.register_property('cdNumberTotal', '([1-9])' + _psep + 'cds?', confidence=0.9, enhance=False, formatter=parse_numeral)

        self.container.register_property('bonusNumber', 'x([0-9]{1,2})', enhance=False, global_span=True, formatter=parse_numeral)

        self.container.register_property('filmNumber', 'f([0-9]{1,2})', enhance=False, global_span=True, formatter=parse_numeral)

        self.container.register_property('edition', 'collector', 'collector-edition', 'edition-collector', canonical_form='Collector Edition')
        self.container.register_property('edition', 'special-edition', 'edition-special', canonical_form='Special Edition')
        self.container.register_property('edition', 'criterion', 'criterion-edition', 'edition-criterion', canonical_form='Criterion Edition')
        self.container.register_property('edition', 'deluxe', 'cdeluxe-edition', 'edition-deluxe', canonical_form='Deluxe Edition')
        self.container.register_property('edition', 'director\'?s?-cut', 'director\'?s?-cut-edition', 'edition-director\'?s?-cut', canonical_form='Director\'s cut')

    def __init__(self):
        Transformer.__init__(self, 20)

        self.container = PropertiesContainer(enhance=False, canonical_from_pattern=False)

        def episode_parser(value):
            values = re_split('[a-zA-Z]', value)
            values = [x for x in values if x]
            ret = []
            for letters_elt in values:
                dashed_values = letters_elt.split('-')
                dashed_values = [x for x in dashed_values if x]
                if len(dashed_values) > 1:
                    for _ in range(0, len(dashed_values) - 1):
                        start_dash_ep = parse_numeral(dashed_values[0])
                        end_dash_ep = parse_numeral(dashed_values[1])
                        for dash_ep in range(start_dash_ep, end_dash_ep + 1):
                            ret.append(dash_ep)
                else:
                    ret.append(parse_numeral(letters_elt))
            if len(ret) > 1:
                return {None: ret[0], 'episodeList': ret}  # TODO: Should support seasonList also
            elif len(ret) > 0:
                return ret[0]
            else:
                return None

        class ResolutionCollisionValidator(object):
            def validate(self, prop, string, node, match, entry_start, entry_end):
                return len(match.group(2)) < 3

        self.container.register_property(None, r'((?:season|saison)' + sep + '?(?P<season>' + numeral + '))', confidence=1.0, formatter=parse_numeral)
        self.container.register_property(None, r'(s(?P<season>' + digital_numeral + ')[^0-9]?' + sep + '?(?P<episodeNumber>(?:e' + digital_numeral + '(?:' + sep + '?[e-]' + digital_numeral + ')*)))[^0-9]', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_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 + ')*)))[^0-9]', confidence=1.0, formatter={None: parse_numeral, 'episodeNumber': episode_parser}, validator=ChainedValidator(DefaultValidator(), ResolutionCollisionValidator()))
        self.container.register_property(None, r'(s(?P<season>' + digital_numeral + '))[^0-9]', confidence=0.6, formatter=parse_numeral, validator=NoValidator())
        self.container.register_property(None, r'((?P<episodeNumber>' + digital_numeral + ')v[23])', confidence=0.6, formatter=parse_numeral)
        self.container.register_property(None, r'((?:ep)' + sep + r'(?P<episodeNumber>' + numeral + '))[^0-9]', confidence=0.7, formatter=parse_numeral)
        self.container.register_property(None, r'(e(?P<episodeNumber>' + digital_numeral + '))', confidence=0.6, formatter=parse_numeral)
        self.container.register_property(None, r'\A ?((?P<episodeNumber>' + '\d{2}' + '))' + sep, confidence=0.4, formatter=parse_numeral)
        self.container.register_property(None, r'\A ?(0(?P<episodeNumber>' + '\d+' + '))' + sep, confidence=0.4, formatter=parse_numeral)

        self.container.register_canonical_properties('other', 'FiNAL', 'Complete', validator=WeakValidator())

    def __init__(self):
        Transformer.__init__(self, 15)

        self.properties = PropertiesContainer(enhance=False, canonical_from_pattern=False)

        def _formater(episodeNumber):
            epnum = parse_numeral(episodeNumber)
            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.properties.register_property(['episodeNumber', 'season'], '[0-9]{2,4}', confidence=0.6, formatter=_formater)
        self.properties.register_property('episodeNumber', '(?:episode)' + sep + '(' + numeral + ')[^0-9]', confidence=0.3)

    def __init__(self):
        Transformer.__init__(self, -190)

        self.container = PropertiesContainer(canonical_from_pattern=False)
        self._allowed_groupname_pattern = '[\[email protected]#€£$&!\?]'
        self._forbidden_groupname_lambda = [lambda elt: elt in ['rip', 'by', 'for', 'par', 'pour', 'bonus'],
                                            lambda elt: self._is_number(elt)]
        # If the previous property in this list, the match will be considered as safe
        # and group name can contain a separator.
        self.previous_safe_properties = ['videoCodec', 'format', 'videoApi', 'audioCodec', 'audioProfile', 'videoProfile', 'audioChannels', 'screenSize', 'other']
        self.previous_safe_values = {'other': ['Complete']}
        self.next_safe_properties = ['extension', 'website']
        self.next_safe_values = {'format': ['Telesync']}
        self.next_unsafe_properties = list(self.previous_safe_properties)
        self.next_unsafe_properties.extend(['episodeNumber', 'season'])
        self.container.sep_replace_char = '-'
        self.container.canonical_from_pattern = False
        self.container.enhance = True
        self.container.register_property('releaseGroup', self._allowed_groupname_pattern + '+')
        self.container.register_property('releaseGroup', self._allowed_groupname_pattern + '+-' + self._allowed_groupname_pattern + '+')
        self.re_sep = re.compile('(' + sep + ')')

class GuessWeakEpisodesRexps(Transformer):
    def __init__(self):
        Transformer.__init__(self, 15)

        self.properties = PropertiesContainer(enhance=False, canonical_from_pattern=False)

        def _formater(episodeNumber):
            epnum = parse_numeral(episodeNumber)
            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.properties.register_property(['episodeNumber', 'season'], '[0-9]{2,4}', confidence=0.6, formatter=_formater)
        self.properties.register_property('episodeNumber', '(?:episode)' + sep + '(' + numeral + ')[^0-9]', confidence=0.3)

    def supported_properties(self):
        return self.properties.get_supported_properties()

    def guess_weak_episodes_rexps(self, string, node=None, options=None):
        if node and 'episodeNumber' in node.root.info:
            return None

        properties = self.properties.find_properties(string, node)
        guess = self.properties.as_guess(properties, string)

        return guess

    def should_process(self, mtree, options=None):
        return mtree.guess.get('type', '').startswith('episode')

    def process(self, mtree, options=None):
        GuessFinder(self.guess_weak_episodes_rexps, 0.6, self.log, options).process_nodes(mtree.unidentified_leaves())

class GuessWebsite(Transformer):
    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 supported_properties(self):
        return self.container.get_supported_properties()

    def guess_website(self, string, node=None, options=None):
        found = self.container.find_properties(string, node, 'website')
        return self.container.as_guess(found, string)

    def process(self, mtree, options=None):
        GuessFinder(self.guess_website, 1.0, self.log, options).process_nodes(mtree.unidentified_leaves())