Python parser.NameParser类代码示例

def filterBadReleases(name):
    """
    Filters out non-english and just all-around stupid releases by comparing them
    to the resultFilters contents.
    
    name: the release name to check
    
    Returns: True if the release name is OK, False if it's bad.
    """

    try:
        fp = NameParser()
        parse_result = fp.parse(name)
    except InvalidNameException:
        logger.log(u"Unable to parse the filename "+name+" into a valid episode", logger.WARNING)
        return False

    # if there's no info after the season info then assume it's fine
    if not parse_result.extra_info:
        return True

    # if any of the bad strings are in the name then say no
    for x in resultFilters:
        if re.search('(^|[\W_])'+x+'($|[\W_])', parse_result.extra_info, re.I):
            logger.log(u"Invalid scene release: "+name+" contains "+x+", ignoring it", logger.DEBUG)
            return False

    return True

    def _addCacheEntry(self, name, url, parse_result=None, indexer_id=0):

        # check if we passed in a parsed result or should we try and create one
        if not parse_result:

            # create showObj from indexer_id if available
            showObj=None
            if indexer_id:
                showObj = helpers.findCertainShow(sickbeard.showList, indexer_id)

            try:
                myParser = NameParser(showObj=showObj, convert=True)
                parse_result = myParser.parse(name)
            except InvalidNameException:
                logger.log(u"Unable to parse the filename " + name + " into a valid episode", logger.DEBUG)
                return None
            except InvalidShowException:
                logger.log(u"Unable to parse the filename " + name + " into a valid show", logger.DEBUG)
                return None

            if not parse_result or not parse_result.series_name:
                return None

        # if we made it this far then lets add the parsed result to cache for usager later on
        season = episodes = None
        if parse_result.is_air_by_date or parse_result.is_sports:
            airdate = parse_result.air_date.toordinal() if parse_result.air_date else parse_result.sports_air_date.toordinal()

            myDB = db.DBConnection()
            sql_results = myDB.select(
                "SELECT season, episode FROM tv_episodes WHERE showid = ? AND indexer = ? AND airdate = ?",
                [parse_result.show.indexerid, parse_result.show.indexer, airdate])
            if sql_results > 0:
                season = int(sql_results[0]["season"])
                episodes = [int(sql_results[0]["episode"])]
        else:
            season = parse_result.season_number if parse_result.season_number else 1
            episodes = parse_result.episode_numbers

        if season and episodes:
            # store episodes as a seperated string
            episodeText = "|" + "|".join(map(str, episodes)) + "|"

            # get the current timestamp
            curTimestamp = int(time.mktime(datetime.datetime.today().timetuple()))

            # get quality of release
            quality = parse_result.quality

            if not isinstance(name, unicode):
                name = unicode(name, 'utf-8', 'replace')

            # get release group
            release_group = parse_result.release_group

            logger.log(u"Added RSS item: [" + name + "] to cache: [" + self.providerID + "]", logger.DEBUG)

            return [
                "INSERT OR IGNORE INTO [" + self.providerID + "] (name, season, episodes, indexerid, url, time, quality, release_group) VALUES (?,?,?,?,?,?,?,?)",
                [name, season, episodeText, parse_result.show.indexerid, url, curTimestamp, quality, release_group]]

def filterBadReleases(name, lang="en"):

    try:
        fp = NameParser()
        parse_result = fp.parse(name)
    except InvalidNameException:
        logger.log(u"Unable to parse the filename "+name+" into a valid episode", logger.WARNING)
        return False

    # exclude language exceptions from filters
    lResultFilters = [f for f in resultFilters]
    if lang in filterExceptions and filterExceptions[lang]:
	for exception in filterExceptions[lang]:
	    if exception in lResultFilters:
		lResultFilters.remove(exception)

    # if there's no info after the season info then assume it's fine, unless there is a whitelist for this language
    if not parse_result.extra_info and not lang in requiredFilters:
        return True

    # if any of the bad strings are in the name then say no
    for x in lResultFilters:
        if re.search('(^|[\W_])'+x+'($|[\W_])', parse_result.extra_info, re.I):
            logger.log(u"Invalid scene release: "+name+" contains "+x+", ignoring it", logger.DEBUG)
            return False

    # if whitelist exists for language, allow only releases containing a whitelisted word
    if lang in requiredFilters and requiredFilters[lang]:
	for x in requiredFilters[lang]:
	    if re.search('(^|[\W_])'+x+'($|[\W_])', parse_result.extra_info, re.I):
	    	return True
	logger.log(u"Invalid scene release: "+name+" doesn't contain any of the words "+str(requiredFilters[lang])+", ignoring it", logger.DEBUG)
	return False

    return True

def filterBadReleases(name):
    """
    Filters out non-english and just all-around stupid releases by comparing them
    to the resultFilters contents.
    
    name: the release name to check
    
    Returns: True if the release name is OK, False if it's bad.
    """

    try:
        fp = NameParser()
        parse_result = fp.parse(name)
    except InvalidNameException:
        logger.log(u"Unable to parse the filename " + name + " into a valid episode", logger.WARNING)
        return False

    # if any of the bad strings are in the name then say no
    if sickbeard.IGNORE_WORDS:
        resultFilters.extend(sickbeard.IGNORE_WORDS.split(','))
    filters = [re.compile('(^|[\W_])%s($|[\W_])' % filter.strip(), re.I) for filter in resultFilters]
    for regfilter in filters:
        if regfilter.search(name):
            logger.log(u"Invalid scene release: " + name + " contains pattern: " + regfilter.pattern + ", ignoring it", logger.DEBUG)
            return False

    return True

def filterBadReleases(name, show):
    """
    Filters out non-english and just all-around stupid releases by comparing them
    to the resultFilters contents.
    
    name: the release name to check
    
    Returns: True if the release name is OK, False if it's bad.
    """

    try:
        fp = NameParser()
        parse_result = fp.parse(name)
    except InvalidNameException:
        logger.log(u"Unable to parse the filename "+name+" into a valid episode", logger.WARNING)
        return False

    #if language not english, search for mandatory, else add other languages to ignore list
    if show.lang != "en":
        mandatory = [(langCodes[show.lang])]
        if langCodes[show.lang] in resultFilters:
            resultFilters.remove(langCodes[show.lang])
        logger.log(u"Language for \""+show.name+"\" is "+show.lang+" so im looking for \""+langCodes[show.lang]+"\" in release names", logger.DEBUG)
    elif show.lang == "en":
        for key in langCodes:
            if not langCodes[key] in resultFilters:
                resultFilters.append(langCodes[key])
        mandatory = []
        logger.log(u"Language for \""+show.name+"\" is "+show.lang, logger.DEBUG)


    # use the extra info and the scene group to filter against
    check_string = ''
    if parse_result.extra_info:
        check_string = parse_result.extra_info
    if parse_result.release_group:
        if check_string:
            check_string = check_string + '-' + parse_result.release_group
        else:
            check_string = parse_result.release_group 

    # if there's no info after the season info then assume it's fine
    if not check_string:
        return True

    # if any of the bad strings are in the name then say no
    for x in resultFilters + sickbeard.IGNORE_WORDS.split(','):
        if re.search('(^|[\W_])'+x+'($|[\W_])', check_string, re.I):
            logger.log(u"Invalid scene release: "+name+" contains "+x+", ignoring it", logger.DEBUG)
            return False
    # if every of the mandatory words are in there, say yes
    if mandatory:
        for x in mandatory:
            if not re.search('(^|[\W_])'+x+'($|[\W_])', check_string, re.I):
                logger.log(u"Mandatory string not found: "+name+" doesnt contains "+x+", ignoring it", logger.DEBUG)
                return False

    return True

    def _addCacheEntry(self, name, url, quality=None):

        cacheDB = self._getDB()
        season = None
        episodes = None

        # if we don't have complete info then parse the filename to get it
        try:
            myParser = NameParser(0)
            parse_result = myParser.parse(name).convert()
        except InvalidNameException:
            logger.log(u"Unable to parse the filename " + name + " into a valid episode", logger.DEBUG)
            return None

        if not parse_result:
            logger.log(u"Giving up because I'm unable to parse this name: " + name, logger.DEBUG)
            return None

        if not parse_result.series_name:
            logger.log(u"No series name retrieved from " + name + ", unable to cache it", logger.DEBUG)
            return None

        if not parse_result.show:
            logger.log(u"Couldn't find a show in our databases matching " + name + ", unable to cache it", logger.DEBUG)
            return None

        try:
            myDB = db.DBConnection()
            if parse_result.show.air_by_date:
                airdate = parse_result.sports_event_date.toordinal() if parse_result.show.sports else parse_result.air_date.toordinal()
                sql_results = myDB.select("SELECT season, episode FROM tv_episodes WHERE showid = ? AND airdate = ?",
                                          [parse_result.show.indexerid, airdate])
                if sql_results > 0:
                    season = int(sql_results[0]["season"])
                    episodes = [int(sql_results[0]["episode"])]
            else:
                season = parse_result.season_number
                episodes = parse_result.episode_numbers

            if season and episodes:
                # store episodes as a seperated string
                episodeText = "|" + "|".join(map(str, episodes)) + "|"

                # get the current timestamp
                curTimestamp = int(time.mktime(datetime.datetime.today().timetuple()))

                # get quality of release
                if quality is None:
                    quality = Quality.sceneQuality(name)

                if not isinstance(name, unicode):
                    name = unicode(name, 'utf-8')

                cacheDB.action(
                    "INSERT INTO [" + self.providerID + "] (name, season, episodes, indexerid, url, time, quality) VALUES (?,?,?,?,?,?,?)",
                    [name, season, episodeText, parse_result.show.indexerid, url, curTimestamp, quality])
        except:
            return

    def _addCacheEntry(self, name, url, quality=None):

        try:
            myParser = NameParser(convert=True)
            parse_result = myParser.parse(name)
        except InvalidNameException:
            logger.log(u"Unable to parse the filename " + name + " into a valid episode", logger.DEBUG)
            return None
        except InvalidShowException:
            logger.log(u"Unable to parse the filename " + name + " into a valid show", logger.DEBUG)
            return None

        if not parse_result or not parse_result.series_name:
            return None

        season = episodes = None
        if parse_result.air_by_date or parse_result.sports:
            airdate = parse_result.air_date.toordinal() if parse_result.air_date else parse_result.sports_event_date.toordinal()

            myDB = db.DBConnection()
            sql_results = myDB.select(
                "SELECT season, episode FROM tv_episodes WHERE showid = ? AND indexer = ? AND airdate = ?",
                [parse_result.show.indexerid, parse_result.show.indexer, airdate])
            if sql_results > 0:
                season = int(sql_results[0]["season"])
                episodes = [int(sql_results[0]["episode"])]
        else:
            season = parse_result.season_number if parse_result.season_number != None else 1
            episodes = parse_result.episode_numbers

        if season and episodes:
            # store episodes as a seperated string
            episodeText = "|" + "|".join(map(str, episodes)) + "|"

            # get the current timestamp
            curTimestamp = int(time.mktime(datetime.datetime.today().timetuple()))

            # get quality of release
            if quality is None:
                quality = Quality.sceneQuality(name, parse_result.is_anime)

            if not isinstance(name, unicode):
                name = unicode(name, 'utf-8')

            # get release group
            release_group = parse_result.release_group

            logger.log(u"Added RSS item: [" + name + "] to cache: [" + self.providerID + "]", logger.DEBUG)

            return [
                "INSERT OR IGNORE INTO [" + self.providerID + "] (name, season, episodes, indexerid, url, time, quality, release_group) VALUES (?,?,?,?,?,?,?,?)",
                [name, season, episodeText, parse_result.show.indexerid, url, curTimestamp, quality, release_group]]

    def _addCacheEntry(self, name, url, parse_result=None, indexer_id=0):

        # check if we passed in a parsed result or should we try and create one
        if not parse_result:

            # create showObj from indexer_id if available
            showObj=None
            if indexer_id:
                showObj = helpers.findCertainShow(sickbeard.showList, indexer_id)

            try:
                myParser = NameParser(showObj=showObj, convert=True)
                parse_result = myParser.parse(name)
            except InvalidNameException:
                logger.log(u'Unable to parse the filename ' + name + ' into a valid episode', logger.DEBUG)
                return None
            except InvalidShowException:
                logger.log(u'Unable to parse the filename ' + name + ' into a valid show', logger.DEBUG)
                return None

            if not parse_result or not parse_result.series_name:
                return None

        # if we made it this far then lets add the parsed result to cache for usager later on
        season = parse_result.season_number if parse_result.season_number else 1
        episodes = parse_result.episode_numbers

        if season and episodes:
            # store episodes as a seperated string
            episodeText = '|' + '|'.join(map(str, episodes)) + '|'

            # get the current timestamp
            curTimestamp = int(time.mktime(datetime.datetime.today().timetuple()))

            # get quality of release
            quality = parse_result.quality

            if not isinstance(name, unicode):
                name = unicode(name, 'utf-8', 'replace')

            # get release group
            release_group = parse_result.release_group

            # get version
            version = parse_result.version

            logger.log(u'Added RSS item: [' + name + '] to cache: [' + self.providerID + ']', logger.DEBUG)

            return [
                'INSERT OR IGNORE INTO provider_cache (provider, name, season, episodes, indexerid, url, time, quality, release_group, version) VALUES (?,?,?,?,?,?,?,?,?,?)',
                [self.providerID, name, season, episodeText, parse_result.show.indexerid, url, curTimestamp, quality, release_group, version]]

    def _addCacheEntry(self, name, url, parse_result=None, indexer_id=0):

        # check if we passed in a parsed result or should we try and create one
        if not parse_result:

            # create showObj from indexer_id if available
            showObj = None
            if indexer_id:
                showObj = helpers.findCertainShow(sickbeard.showList, indexer_id)

            try:
                myParser = NameParser(showObj=showObj)
                parse_result = myParser.parse(name)
            except InvalidNameException:
                logger.log(u"Unable to parse the filename " + name + " into a valid episode", logger.DEBUG)
                return None
            except InvalidShowException:
                logger.log(u"Unable to parse the filename " + name + " into a valid show", logger.DEBUG)
                return None

            if not parse_result or not parse_result.series_name:
                return None

        # if we made it this far then lets add the parsed result to cache for usager later on
        season = parse_result.season_number if parse_result.season_number else 1
        episodes = parse_result.episode_numbers

        if season and episodes:
            # store episodes as a seperated string
            episodeText = "|" + "|".join(map(str, episodes)) + "|"

            # get the current timestamp
            curTimestamp = int(time.mktime(datetime.datetime.today().timetuple()))

            # get quality of release
            quality = parse_result.quality

            name = ss(name)

            # get release group
            release_group = parse_result.release_group

            # get version
            version = parse_result.version

            logger.log(u"Added RSS item: [" + name + "] to cache: [" + self.providerID + "]", logger.DEBUG)

            return [
                "INSERT OR IGNORE INTO [" + self.providerID + "] (name, season, episodes, indexerid, url, time, quality, release_group, version) VALUES (?,?,?,?,?,?,?,?,?)",
                [name, season, episodeText, parse_result.show.indexerid, url, curTimestamp, quality, release_group, version]]

def filterBadReleases(name,showLang=u"en"):
    """
    Filters out non-english and just all-around stupid releases by comparing them
    to the resultFilters contents.
    
    name: the release name to check
    
    Returns: True if the release name is OK, False if it's bad.
    """
    
    additionalFilters = []
    if showLang == u"en":
        additionalFilters.append("dub(bed)?")

    try:
        fp = NameParser()
        parse_result = fp.parse(name)
    except InvalidNameException:
        logger.log(u"Unable to parse the filename "+name+" into a valid episode", logger.WARNING)
        return False

    # use the extra info and the scene group to filter against
    check_string = ''
    if parse_result.extra_info:
        check_string = parse_result.extra_info
    if parse_result.release_group:
        if check_string:
            check_string = check_string + '-' + parse_result.release_group
        else:
            check_string = parse_result.release_group 

    # if there's no info after the season info then assume it's fine
    if not check_string:
        check_string = name

    # if any of the bad strings are in the name then say no
    if sickbeard.IGNORE_WORDS == "":
        ignore_words="ztreyfgut"
    else:
        ignore_words=sickbeard.IGNORE_WORDS
    for x in resultFilters + ignore_words.split(',') + additionalFilters:
        if x == showLanguages.get(showLang):
            continue
        if re.search('(^|[\W_])'+x+'($|[\W_])', check_string, re.I):
            logger.log(u"Invalid scene release: "+name+" contains "+x+", ignoring it", logger.DEBUG)
            return False

    return True

def filterBadReleases(name):
    """
    Filters out non-english and just all-around stupid releases by comparing them
    to the resultFilters contents.
    
    name: the release name to check
    
    Returns: True if the release name is OK, False if it's bad.
    """

    try:
        fp = NameParser()
        parse_result = fp.parse(name)
    except InvalidNameException:
        logger.log(u"Unable to parse the filename "+name+" into a valid episode", logger.WARNING)
        return False

    # use the extra info and the scene group to filter against
    check_string = ''
    if parse_result.extra_info:
        check_string = parse_result.extra_info
    if parse_result.release_group:
        if check_string:
            check_string = check_string + '-' + parse_result.release_group
        else:
            check_string = parse_result.release_group 

    # if there's no info after the season info then assume it's fine
    if not check_string:
        return True

    # if any of the bad strings are in the name then say no
    for x in resultFilters + sickbeard.IGNORE_WORDS.split(','):
        if re.search('(^|[\W_])'+x+'($|[\W_])', check_string, re.I):
            logger.log(u"Invalid scene release: "+name+" contains "+x+", ignoring it", logger.DEBUG)
            return False

    # if every of the mandatory words are in there, say yes
    for x in mandatory:
        if not re.search('(^|[\W_])'+x+'($|[\W_])', check_string, re.I):
            logger.log(u"Mandatory string not found: "+name+" doesnt contains "+x+", ignoring it", logger.DEBUG)
            return False

    return True

def filterByRequiredWordsReleases(name, requiredWords):
    """
    Filters out non-english and just all-around stupid releases by comparing them
    to the resultFilters contents.
    
    name: the release name to check
    
    Returns: True if the release name is OK, False if it's bad.
    """

    try:
        fp = NameParser()
        parse_result = fp.parse(name)
    except InvalidNameException:
        logger.log(u"Unable to parse the filename "+name+" into a valid episode", logger.WARNING)
        return False

                 
		
    # use the extra info and the scene group to filter against
    check_string = ''
    if parse_result.extra_info:
        check_string = parse_result.extra_info
    if parse_result.release_group:
        if check_string:
            check_string = check_string + '-' + parse_result.release_group
        else:
            check_string = parse_result.release_group 

    # if there's no info after the season info then assume it's fine
    if not check_string:
        return True

    # if any of the bad strings are in the name then say no
    logger.log(u"filtered requiredWords", logger.DEBUG)
    if ( requiredWords != ""):
        
        for x in requiredWords.split(','):
            if re.search('(^|[\W_])'+x+'($|[\W_])', check_string, re.I):
                logger.log(u"This scene release: "+name+" contains "+x+", add it", logger.DEBUG)
                return True
        logger.log(u"Invalid scene release: "+name+" not contains, ignore it", logger.DEBUG)
        return False
    return True

def filterBadReleases(name):

    try:
        fp = NameParser()
        parse_result = fp.parse(name)
    except InvalidNameException:
        logger.log(u"Unable to parse the filename "+name+" into a valid episode", logger.WARNING)
        return False

    # if there's no info after the season info then assume it's fine
    if not parse_result.extra_info:
        return True

    # if any of the bad strings are in the name then say no
    for x in resultFilters:
        if re.search('(^|[\W_])'+x+'($|[\W_])', parse_result.extra_info, re.I):
            logger.log(u"Invalid scene release: "+name+" contains "+x+", ignoring it", logger.DEBUG)
            return False

    return True

def filterBadReleases(name):
    """
    Filters out non-english and just all-around stupid releases by comparing them
    to the resultFilters contents.
    
    name: the release name to check
    
    Returns: True if the release name is OK, False if it's bad.
    """

    try:
        fp = NameParser()
        parse_result = fp.parse(name)
    except InvalidNameException:
        logger.log(u"Unable to parse the filename " + name + " into a valid episode", logger.WARNING)
        return False

#    # use the extra info and the scene group to filter against
#    check_string = ''
#    if parse_result.extra_info:
#        check_string = parse_result.extra_info
#    if parse_result.release_group:
#        if check_string:
#            check_string = check_string + '-' + parse_result.release_group
#        else:
#            check_string = parse_result.release_group 
#
#    # if there's no info after the season info then assume it's fine
#    if not check_string:
#        return True

    # if any of the bad strings are in the name then say no
    for ignore_word in resultFilters + sickbeard.IGNORE_WORDS.split(','):
        ignore_word = ignore_word.strip()
        if ignore_word:
            if re.search('(^|[\W_])' + ignore_word + '($|[\W_])', name, re.I):
                logger.log(u"Invalid scene release: " + name + " contains " + ignore_word + ", ignoring it", logger.DEBUG)
                return False

    return True

    def _addCacheEntry(self, name, url, quality=None):
        indexerid = None
        in_cache = False

        # if we don't have complete info then parse the filename to get it
        try:
            myParser = NameParser()
            parse_result = myParser.parse(name).convert()
        except InvalidNameException:
            logger.log(u"Unable to parse the filename " + name + " into a valid episode", logger.DEBUG)
            return None

        if not parse_result:
            logger.log(u"Giving up because I'm unable to parse this name: " + name, logger.DEBUG)
            return None

        if not parse_result.series_name:
            logger.log(u"No series name retrieved from " + name + ", unable to cache it", logger.DEBUG)
            return None

        cacheResult = sickbeard.name_cache.retrieveNameFromCache(parse_result.series_name)
        if cacheResult:
            in_cache = True
            indexerid = int(cacheResult)

        if not indexerid:
            showResult = helpers.searchDBForShow(parse_result.series_name)
            if showResult:
                indexerid = int(showResult[0])

        showObj = None
        if indexerid:
            showObj = helpers.findCertainShow(sickbeard.showList, indexerid)

        if not showObj:
            logger.log(u"No match for show: [" + parse_result.series_name + "], not caching ...", logger.DEBUG)
            return None

        season = episodes = None
        if parse_result.air_by_date or parse_result.sports:
            myDB = db.DBConnection()

            airdate = parse_result.air_date.toordinal() or parse_result.sports_event_date.toordinal()
            sql_results = myDB.select(
                "SELECT season, episode FROM tv_episodes WHERE showid = ? AND indexer = ? AND airdate = ?",
                [indexerid, showObj.indexer, airdate])
            if sql_results > 0:
                season = int(sql_results[0]["season"])
                episodes = [int(sql_results[0]["episode"])]
        else:
            season = parse_result.season_number if parse_result.season_number != None else 1
            episodes = parse_result.episode_numbers

        if season and episodes:
            # store episodes as a seperated string
            episodeText = "|" + "|".join(map(str, episodes)) + "|"

            # get the current timestamp
            curTimestamp = int(time.mktime(datetime.datetime.today().timetuple()))

            # get quality of release
            if quality is None:
                quality = Quality.sceneQuality(name)

            if not isinstance(name, unicode):
                name = unicode(name, 'utf-8')

            logger.log(u"Added RSS item: [" + name + "] to cache: [" + self.providerID + "]", logger.DEBUG)

            if not in_cache:
                sickbeard.name_cache.addNameToCache(parse_result.series_name, indexerid)

            return [
                "INSERT INTO [" + self.providerID + "] (name, season, episodes, indexerid, url, time, quality) VALUES (?,?,?,?,?,?,?)",
                [name, season, episodeText, indexerid, url, curTimestamp, quality]]