Python parser.NameParser类代码示例

def validate_name(pattern, multi=None, anime_type=None, file_only=False, abd=False, sports=False):
    """
    See if we understand a name

    :param pattern: Name to analyse
    :param multi: Is this a multi-episode name
    :param anime_type: Is this anime
    :param file_only: Is this just a file or a dir
    :param abd: Is air-by-date enabled
    :param sports: Is this sports
    :return: True if valid name, False if not
    """
    ep = generate_sample_ep(multi, abd, sports, anime_type)

    new_name = ep.formatted_filename(pattern, multi, anime_type) + '.ext'
    new_path = ep.formatted_dir(pattern, multi)
    if not file_only:
        new_name = ek(os.path.join, new_path, new_name)

    if not new_name:
        logger.log(u"Unable to create a name out of " + pattern, logger.DEBUG)
        return False

    logger.log(u"Trying to parse " + new_name, logger.DEBUG)

    parser = NameParser(True, showObj=ep.show, naming_pattern=True)

    try:
        result = parser.parse(new_name)
    except Exception, e:
        logger.log(u"Unable to parse " + new_name + ", not valid", logger.DEBUG)
        return False

    def _analyze_name(self, name, resource=True):
        """
        Takes a name and tries to figure out a show, season, and episode from it.

        name: A string which we want to analyze to determine show info from (unicode)

        Returns a (indexer_id, season, [episodes]) tuple. The first two may be None and episodes may be []
        if none were found.
        """

        logger.log(u'Analyzing name ' + repr(name))

        to_return = (None, None, [], None)

        if not name:
            return to_return

        # parse the name to break it into show name, season, and episode
        np = NameParser(resource, try_scene_exceptions=True, convert=True)
        parse_result = np.parse(name)
        self._log(u'Parsed %s<br />.. from %s' % (str(parse_result).decode('utf-8', 'xmlcharrefreplace'), name), logger.DEBUG)

        if parse_result.is_air_by_date:
            season = -1
            episodes = [parse_result.air_date]
        else:
            season = parse_result.season_number
            episodes = parse_result.episode_numbers

        # show object
        show = parse_result.show
        to_return = (show, season, episodes, parse_result.quality)

        self._finalize(parse_result)
        return to_return

    def process(self):
        self._log(u"Failed download detected: (" + str(self.nzb_name) + ", " + str(self.dir_name) + ")")

        releaseName = show_name_helpers.determineReleaseName(self.dir_name, self.nzb_name)
        if releaseName is None:
            self._log(u"Warning: unable to find a valid release name.", logger.WARNING)
            raise exceptions.FailedProcessingFailed()

        try:
            parser = NameParser(False, showObj=self.show, convert=True)
            parsed = parser.parse(releaseName)
        except InvalidNameException:
            self._log(u"Error: release name is invalid: " + releaseName, logger.DEBUG)
            raise exceptions.FailedProcessingFailed()
        except InvalidShowException:
            self._log(u"Error: unable to parse release name " + releaseName + " into a valid show", logger.DEBUG)
            raise exceptions.FailedProcessingFailed()

        logger.log(u"name_parser info: ", logger.DEBUG)
        logger.log(u" - " + str(parsed.series_name), logger.DEBUG)
        logger.log(u" - " + str(parsed.season_number), logger.DEBUG)
        logger.log(u" - " + str(parsed.episode_numbers), logger.DEBUG)
        logger.log(u" - " + str(parsed.extra_info), logger.DEBUG)
        logger.log(u" - " + str(parsed.release_group), logger.DEBUG)
        logger.log(u" - " + str(parsed.air_date), logger.DEBUG)

        for episode in parsed.episode_numbers:
            segment = parsed.show.getEpisode(parsed.season_number, episode)

            cur_failed_queue_item = search_queue.FailedQueueItem(parsed.show, [segment])
            sickbeard.searchQueueScheduler.action.add_item(cur_failed_queue_item)

        return True

def validate_name(pattern, multi=None, file_only=False, abd=False):
    ep = _generate_sample_ep(multi, abd)

    parser = NameParser(True)

    new_name = ep.formatted_filename(pattern, multi) + '.ext'
    new_path = ep.formatted_dir(pattern, multi)
    if not file_only:
        new_name = ek.ek(os.path.join, new_path, new_name)

    logger.log(u"Trying to parse "+new_name, logger.DEBUG)

    try:
        result = parser.parse(new_name)
    except InvalidNameException:
        logger.log(u"Unable to parse "+new_name+", not valid", logger.DEBUG)
        return False
    
    logger.log(new_name + " vs " + str(result), logger.DEBUG)

    if abd:
        if result.air_date != ep.airdate:
            return False
    else:
        if result.season_number != ep.season:
            return False
        if result.episode_numbers != [x.episode for x in [ep] + ep.relatedEps]:
            return False

    return True

    def test_formating(self):
        release = "UFC.172.26th.April.2014.HDTV.x264.720p-Sir.Paul[rartv]"
        # parse the name to break it into show name, season, and episode
        np = NameParser(file)
        parse_result = np.parse(release).convert()

        print(parse_result)

    def _find_season_quality(self, title, torrent_id, ep_number):
        """ Return the modified title of a Season Torrent with the quality found inspecting torrent file list """

        mediaExtensions = ['avi', 'mkv', 'wmv', 'divx',
                           'vob', 'dvr-ms', 'wtv', 'ts'
                                                   'ogv', 'rar', 'zip', 'mp4']

        quality = Quality.UNKNOWN

        fileName = None

        fileURL = self.proxy._buildURL(self.url + 'ajax_details_filelist.php?id=' + str(torrent_id))

        data = self.getURL(fileURL)
        if not data:
            return None

        filesList = re.findall('<td.+>(.*?)</td>', data)

        if not filesList:
            logger.log(u"Unable to get the torrent file list for " + title, logger.ERROR)

        videoFiles = filter(lambda x: x.rpartition(".")[2].lower() in mediaExtensions, filesList)

        #Filtering SingleEpisode/MultiSeason Torrent
        if len(videoFiles) < ep_number or len(videoFiles) > float(ep_number * 1.1):
            logger.log(
                u"Result " + title + " have " + str(ep_number) + " episode and episodes retrived in torrent are " + str(
                    len(videoFiles)), logger.DEBUG)
            logger.log(u"Result " + title + " Seem to be a Single Episode or MultiSeason torrent, skipping result...",
                       logger.DEBUG)
            return None

        if Quality.sceneQuality(title) != Quality.UNKNOWN:
            return title

        for fileName in videoFiles:
            quality = Quality.sceneQuality(os.path.basename(fileName))
            if quality != Quality.UNKNOWN: break

        if fileName is not None and quality == Quality.UNKNOWN:
            quality = Quality.assumeQuality(os.path.basename(fileName))

        if quality == Quality.UNKNOWN:
            logger.log(u"Unable to obtain a Season Quality for " + title, logger.DEBUG)
            return None

        try:
            myParser = NameParser()
            parse_result = myParser.parse(fileName)
        except InvalidNameException:
            return None

        logger.log(u"Season quality for " + title + " is " + Quality.qualityStrings[quality], logger.DEBUG)

        if parse_result.series_name and parse_result.season_number:
            title = parse_result.series_name + ' S%02d' % int(parse_result.season_number) + ' ' + self._reverseQuality(
                quality)

        return title

    def _parseItem(self, item):
        ltvdb_api_parms = sickbeard.TVDB_API_PARMS.copy()
        ltvdb_api_parms['search_all_languages'] = True

        (title, url) = self.provider._get_title_and_url(item)

        if not title or not url:
            logger.log(u"The XML returned from the Ethor RSS feed is incomplete, this result is unusable", logger.ERROR)
            return
            
        try:
            myParser = NameParser()
            parse_result = myParser.parse(title)
        except InvalidNameException:
            logger.log(u"Unable to parse the filename "+title+" into a valid episode", logger.DEBUG)
            return

        try:
            t = tvdb_api.Tvdb(**ltvdb_api_parms)
            showObj = t[parse_result.series_name]
        except tvdb_exceptions.tvdb_error:
            logger.log(u"TVDB timed out, unable to update episodes from TVDB", logger.ERROR)
            return

        logger.log(u"Adding item from RSS to cache: " + title, logger.DEBUG)

        self._addCacheEntry(name=title, url=url, tvdb_id=showObj['id'])

def history_snatched_proper_fix():
    my_db = db.DBConnection()
    if not my_db.has_flag('history_snatch_proper'):
        logger.log('Updating history items with status Snatched Proper in a background process...')
        sql_result = my_db.select('SELECT rowid, resource, quality, showid'
                                  ' FROM history'
                                  ' WHERE action LIKE "%%%02d"' % SNATCHED +
                                  ' AND (UPPER(resource) LIKE "%PROPER%"'
                                  ' OR UPPER(resource) LIKE "%REPACK%"'
                                  ' OR UPPER(resource) LIKE "%REAL%")')
        if sql_result:
            cl = []
            for r in sql_result:
                show_obj = None
                try:
                    show_obj = helpers.findCertainShow(sickbeard.showList, int(r['showid']))
                except (StandardError, Exception):
                    pass
                np = NameParser(False, showObj=show_obj, testing=True)
                try:
                    pr = np.parse(r['resource'])
                except (StandardError, Exception):
                    continue
                if 0 < Quality.get_proper_level(pr.extra_info_no_name(), pr.version, pr.is_anime):
                    cl.append(['UPDATE history SET action = ? WHERE rowid = ?',
                               [Quality.compositeStatus(SNATCHED_PROPER, int(r['quality'])),
                                r['rowid']]])
            if cl:
                my_db.mass_action(cl)
            logger.log('Completed the history table update with status Snatched Proper.')
        my_db.add_flag('history_snatch_proper')

    def findEpisode(self, episode, manualSearch=False):

        self._checkAuth()

        logger.log(u"Searching "+self.name+" for " + episode.prettyName())

        self.cache.updateCache()
        results = self.cache.searchCache(episode, manualSearch)
        logger.log(u"Cache results: "+str(results), logger.DEBUG)

        # if we got some results then use them no matter what.
        # OR
        # return anyway unless we're doing a manual search
        if results or not manualSearch:
            return results

        itemList = []

        for cur_search_string in self._get_episode_search_strings(episode):
            itemList += self._doSearch(cur_search_string)

        for item in itemList:

            (title, url) = self._get_title_and_url(item)

            # parse the file name
            try:
                myParser = NameParser()
                parse_result = myParser.parse(title)
            except InvalidNameException:
                logger.log(u"Unable to parse the filename "+title+" into a valid episode", logger.WARNING)
                continue

            if episode.show.air_by_date:
                if parse_result.air_date != episode.airdate:
                    logger.log("Episode "+title+" didn't air on "+str(episode.airdate)+", skipping it", logger.DEBUG)
                    continue
            elif parse_result.season_number != episode.season or episode.episode not in parse_result.episode_numbers:
                logger.log("Episode "+title+" isn't "+str(episode.season)+"x"+str(episode.episode)+", skipping it", logger.DEBUG)
                continue

            quality = self.getQuality(item)

            if not episode.show.wantEpisode(episode.season, episode.episode, quality, manualSearch):
                logger.log(u"Ignoring result "+title+" because we don't want an episode that is "+Quality.qualityStrings[quality], logger.DEBUG)
                continue

            logger.log(u"Found result " + title + " at " + url, logger.DEBUG)

            result = self.getResult([episode])
            result.url = url
            result.name = title
            result.quality = quality
            result.provider = self
            result.content = None 
            
            results.append(result)

        return results

    def _find_season_quality(self, title, torrent_link, ep_number):
        """ Return the modified title of a Season Torrent with the quality found inspecting torrent file list """

        quality = Quality.UNKNOWN

        file_name = None

        data = self.get_url(torrent_link)
        if not data:
            return None

        try:
            with BS4Parser(data, features=['html5lib', 'permissive']) as soup:
                file_table = soup.find('table', attrs={'class': 'torrentFileList'})

                if not file_table:
                    return None

                files = [x.text for x in file_table.find_all('td', attrs={'class': 'torFileName'})]
                video_files = filter(lambda i: i.rpartition('.')[2].lower() in mediaExtensions, files)

                # Filtering SingleEpisode/MultiSeason Torrent
                if len(video_files) < ep_number or len(video_files) > float(ep_number * 1.1):
                    logger.log(u'Result %s lists %s episodes with %s episodes retrieved in torrent'
                               % (title, ep_number, len(video_files)), logger.DEBUG)
                    logger.log(u'Result %s seem to be a single episode or multi-season torrent, skipping result...'
                               % title, logger.DEBUG)
                    return None

                if Quality.UNKNOWN != Quality.sceneQuality(title):
                    return title

                for file_name in video_files:
                    quality = Quality.sceneQuality(os.path.basename(file_name))
                    if Quality.UNKNOWN != quality:
                        break

                if None is not file_name and Quality.UNKNOWN == quality:
                    quality = Quality.assumeQuality(os.path.basename(file_name))

                if Quality.UNKNOWN == quality:
                    logger.log(u'Unable to obtain a Season Quality for ' + title, logger.DEBUG)
                    return None

                try:
                    my_parser = NameParser(showObj=self.show)
                    parse_result = my_parser.parse(file_name)
                except (InvalidNameException, InvalidShowException):
                    return None

                logger.log(u'Season quality for %s is %s' % (title, Quality.qualityStrings[quality]), logger.DEBUG)

                if parse_result.series_name and parse_result.season_number:
                    title = parse_result.series_name + ' S%02d %s' % (int(parse_result.season_number),
                                                                      self._reverse_quality(quality))
                return title

        except Exception:
            logger.log(u'Failed to quality parse ' + self.name + ' Traceback: ' + traceback.format_exc(), logger.ERROR)

def already_postprocessed(dirName, videofile, force, result):
    """
    Check if we already post processed a file

    :param dirName: Directory a file resides in
    :param videofile: File name
    :param force: Force checking when already checking (currently unused)
    :param result: True if file is already postprocessed, False if not
    :return:
    """
    if force:
        return False

    # Avoid processing the same dir again if we use a process method <> move
    myDB = db.DBConnection()
    sqlResult = myDB.select("SELECT * FROM tv_episodes WHERE release_name = ?", [dirName])
    if sqlResult:
        # result.output += logHelper(u"You're trying to post process a dir that's already been processed, skipping", logger.DEBUG)
        return True

    else:
        sqlResult = myDB.select("SELECT * FROM tv_episodes WHERE release_name = ?", [videofile.rpartition(".")[0]])
        if sqlResult:
            # result.output += logHelper(u"You're trying to post process a video that's already been processed, skipping", logger.DEBUG)
            return True

        # Needed if we have downloaded the same episode @ different quality
        # But we need to make sure we check the history of the episode we're going to PP, and not others
        np = NameParser(dirName, tryIndexers=True)
        try:  # if it fails to find any info (because we're doing an unparsable folder (like the TV root dir) it will throw an exception, which we want to ignore
            parse_result = np.parse(dirName)
        except Exception:  # ignore the exception, because we kind of expected it, but create parse_result anyway so we can perform a check on it.
            parse_result = False

        search_sql = "SELECT tv_episodes.indexerid, history.resource FROM tv_episodes INNER JOIN history ON history.showid=tv_episodes.showid"  # This part is always the same
        search_sql += " WHERE history.season=tv_episodes.season and history.episode=tv_episodes.episode"
        # If we find a showid, a season number, and one or more episode numbers then we need to use those in the query
        if parse_result and (
            parse_result.show.indexerid and parse_result.episode_numbers and parse_result.season_number
        ):
            search_sql += (
                " and tv_episodes.showid = '"
                + str(parse_result.show.indexerid)
                + "' and tv_episodes.season = '"
                + str(parse_result.season_number)
                + "' and tv_episodes.episode = '"
                + str(parse_result.episode_numbers[0])
                + "'"
            )

        search_sql += " and tv_episodes.status IN (" + ",".join([str(x) for x in common.Quality.DOWNLOADED]) + ")"
        search_sql += " and history.resource LIKE ?"
        sqlResult = myDB.select(search_sql, ["%" + videofile])
        if sqlResult:
            # result.output += logHelper(u"You're trying to post process a video that's already been processed, skipping", logger.DEBUG)
            return True

    return False

    def _find_season_quality(self,title, torrent_link, ep_number):
        """ Return the modified title of a Season Torrent with the quality found inspecting torrent file list """

        mediaExtensions = ['avi', 'mkv', 'wmv', 'divx',
                           'vob', 'dvr-ms', 'wtv', 'ts'
                           'ogv', 'rar', 'zip', 'mp4'] 
        
        quality = Quality.UNKNOWN        
        
        fileName = None

        data = self.getURL(torrent_link)
        
        if not data:
            return None
        
        try: 
            soup = BeautifulSoup(data, features=["html5lib", "permissive"])
            file_table = soup.find('table', attrs = {'class': 'torrentFileList'})

            if not file_table:
                return None
            
            files = [x.text for x in file_table.find_all('td', attrs = {'class' : 'torFileName'} )]
            videoFiles = filter(lambda x: x.rpartition(".")[2].lower() in mediaExtensions, files)

            #Filtering SingleEpisode/MultiSeason Torrent
            if len(videoFiles) < ep_number or len(videoFiles) > float(ep_number * 1.1 ):
                logger.log(u"Result " + title + " Seem to be a Single Episode or MultiSeason torrent, skipping result...", logger.DEBUG)
                return None

            for fileName in videoFiles:
                quality = Quality.sceneQuality(os.path.basename(fileName))
                if quality != Quality.UNKNOWN: break
    
            if fileName!=None and quality == Quality.UNKNOWN:
                quality = Quality.assumeQuality(os.path.basename(fileName))            
    
            if quality == Quality.UNKNOWN:
                logger.log(u"Unable to obtain a Season Quality for " + title, logger.DEBUG)
                return None
    
            try:
                myParser = NameParser()
                parse_result = myParser.parse(fileName)
            except InvalidNameException:
                return None
            
            logger.log(u"Season quality for "+title+" is "+Quality.qualityStrings[quality], logger.DEBUG)
            
            if parse_result.series_name and parse_result.season_number: 
                title = parse_result.series_name+' S%02d' % int(parse_result.season_number)+' '+self._reverseQuality(quality)
            
            return title
            
        except Exception, e:
            logger.log(u"Failed parsing " + self.name + " Traceback: "  + traceback.format_exc(), logger.ERROR)

    def test_formating(self):
        self.loadFromDB()

        release = "d:\\Downloads\\newdownload\\2.Broke.Girls.S03E10.And.the.First.Day.of.School.720p.WEB-DL.DD5.1.H.264-BS.mkv"
        # parse the name to break it into show name, season, and episode
        np = NameParser(file)
        parse_result = np.parse(release).convert()

        print(parse_result)

    def _find_season_quality(self, title, torrent_id):
        """ Rewrite the title of a Season Torrent with the quality found inspecting torrent file list """

        mediaExtensions = ["avi", "mkv", "wmv", "divx", "vob", "dvr-ms", "wtv", "ts" "ogv", "rar", "zip"]

        quality = Quality.UNKNOWN

        fileName = None

        fileURL = self.proxy._buildURL(self.url + "ajax_details_filelist.php?id=" + str(torrent_id))

        data = self.getURL(fileURL)

        if not data:
            return None

        filesList = re.findall("<td.+>(.*?)</td>", data)

        if not filesList:
            logger.log(u"Unable to get the torrent file list for " + title, logger.ERROR)

        #        for fileName in filter(lambda x: x.rpartition(".")[2].lower() in mediaExtensions, filesList):
        #            quality = Quality.nameQuality(os.path.basename(fileName))
        #            if quality != Quality.UNKNOWN: break

        for fileName in filesList:
            sepFile = fileName.rpartition(".")
            if fileName.rpartition(".")[2].lower() in mediaExtensions:
                quality = Quality.nameQuality(fileName)
                if quality != Quality.UNKNOWN:
                    break

        if fileName != None and quality == Quality.UNKNOWN:
            quality = Quality.assumeQuality(os.path.basename(fileName))

        if quality == Quality.UNKNOWN:
            logger.log(u"No Season quality for " + title, logger.DEBUG)
            return None

        try:
            myParser = NameParser()
            parse_result = myParser.parse(fileName)
        except InvalidNameException:
            return None

        logger.log(u"Season quality for " + title + " is " + Quality.qualityStrings[quality], logger.DEBUG)

        if parse_result.series_name and parse_result.season_number:
            title = (
                parse_result.series_name
                + " S%02d" % int(parse_result.season_number)
                + " "
                + self._reverseQuality(quality)
            )

        return title

    def _find_season_quality(self, title, torrent_id, ep_number):
        """ Return the modified title of a Season Torrent with the quality found inspecting torrent file list """

        if not self.url:
            return False

        quality = Quality.UNKNOWN
        file_name = None
        data = self.get_url('%sajax_details_filelist.php?id=%s' % (self.url, torrent_id))
        if self.should_skip() or not data:
            return None

        files_list = re.findall('<td.+>(.*?)</td>', data)

        if not files_list:
            logger.log(u'Unable to get the torrent file list for ' + title, logger.ERROR)

        video_files = filter(lambda x: x.rpartition('.')[2].lower() in mediaExtensions, files_list)

        # Filtering SingleEpisode/MultiSeason Torrent
        if ep_number > len(video_files) or float(ep_number * 1.1) < len(video_files):
            logger.log(u'Result %s has episode %s and total episodes retrieved in torrent are %s'
                       % (title, str(ep_number), str(len(video_files))), logger.DEBUG)
            logger.log(u'Result %s seems to be a single episode or multiseason torrent, skipping result...'
                       % title, logger.DEBUG)
            return None

        if Quality.UNKNOWN != Quality.sceneQuality(title):
            return title

        for file_name in video_files:
            quality = Quality.sceneQuality(os.path.basename(file_name))
            if Quality.UNKNOWN != quality:
                break

        if None is not file_name and Quality.UNKNOWN == quality:
            quality = Quality.assumeQuality(os.path.basename(file_name))

        if Quality.UNKNOWN == quality:
            logger.log(u'Unable to obtain a Season Quality for ' + title, logger.DEBUG)
            return None

        try:
            my_parser = NameParser(showObj=self.show, indexer_lookup=False)
            parse_result = my_parser.parse(file_name)
        except (InvalidNameException, InvalidShowException):
            return None

        logger.log(u'Season quality for %s is %s' % (title, Quality.qualityStrings[quality]), logger.DEBUG)

        if parse_result.series_name and parse_result.season_number:
            title = '%s S%02d %s' % (parse_result.series_name,
                                     int(parse_result.season_number),
                                     self._reverse_quality(quality))

        return title