Пример #1
0
    def getPlays(leporello_info):
        plays = []
        
        # Initializing the Leporello Assistent with meta information for web scraping.
#        Lepistant.setInfo(leporello_info)
        
        leporelloFileNameOnDisk = Lepistant.createFilePath(
                                                Lepistant.REL_PATH_DOWNLOADS_FOLDER, 
                                                Lepistant.FILE_NAME_LEPORELLO, 
                                                'leporello')
        
        soup = Lepistant.getSoup(leporello_info[URL_LEPORELLO], 
                                 leporelloFileNameOnDisk)
        
        playItems = Lepistant.getTagsByClass(soup, 'div', 
                                             leporello_info[CSS_CLASS_PLAY_ITEM])
        leporello_info[PLAY_ITEMS] = playItems
        
        # Get only one play item for testing purposes
#        playItem = playItems[20]    # WirAlleAnders
        playItem = playItems[12]    # AltArmArbeitslos
        play = Play(playItem)
        play.link = Lepistant.getURLFromTagContent(playItem)
        formatted_title = Lepistant.removeNonAlphanumericCharacters(play.title)
        play.file_path_on_disk = Lepistant.REL_PATH_PLAYS_FOLDER + formatted_title + '/'
        play.file_name_on_disk = Lepistant.createFilePath(
                                                play.file_path_on_disk, 
                                                play.title, 
                                                'html')
        soup = Lepistant.getSoup(play.link, play.file_name_on_disk)
        play.setPlayDetails(soup)
        plays.append(play)
        
        
#        for playItem in playItems:
#            logger.info('>>>>>>>>>>>>>>>>>>>>>>>> Fetching new play <<<<<<<<<<<<<<<<<<<<<<<<<')
#            play = Play(playItem)
#            play.link = Lepistant.getURLFromTagContent(playItem)
#            formatted_title = Lepistant.removeNonAlphanumericCharacters(play.title)
#            play.file_path_on_disk = Lepistant.REL_PATH_PLAYS_FOLDER + formatted_title + '/'
#            play.file_name_on_disk = Lepistant.createFilePath(
#                                                play.file_path_on_disk, 
#                                                play.title, 
#                                                'html')
#            soup = Lepistant.getSoup(play.link, play.file_name_on_disk)
#            play.setPlayDetails(soup)
#            
#            logger.info('')
#            
#            plays.append(play)
        
        return plays
Пример #2
0
    def getPlays(leporello_info):
        plays = []

        # Initializing the Leporello Assistent with meta information for web scraping.
        #        Lepistant.setInfo(leporello_info)

        leporelloFileNameOnDisk = Lepistant.createFilePath(
            Lepistant.REL_PATH_DOWNLOADS_FOLDER, Lepistant.FILE_NAME_LEPORELLO,
            'leporello')

        soup = Lepistant.getSoup(leporello_info[URL_LEPORELLO],
                                 leporelloFileNameOnDisk)

        playItems = Lepistant.getTagsByClass(
            soup, 'div', leporello_info[CSS_CLASS_PLAY_ITEM])
        leporello_info[PLAY_ITEMS] = playItems

        # Get only one play item for testing purposes
        #        playItem = playItems[20]    # WirAlleAnders
        playItem = playItems[12]  # AltArmArbeitslos
        play = Play(playItem)
        play.link = Lepistant.getURLFromTagContent(playItem)
        formatted_title = Lepistant.removeNonAlphanumericCharacters(play.title)
        play.file_path_on_disk = Lepistant.REL_PATH_PLAYS_FOLDER + formatted_title + '/'
        play.file_name_on_disk = Lepistant.createFilePath(
            play.file_path_on_disk, play.title, 'html')
        soup = Lepistant.getSoup(play.link, play.file_name_on_disk)
        play.setPlayDetails(soup)
        plays.append(play)

        #        for playItem in playItems:
        #            logger.info('>>>>>>>>>>>>>>>>>>>>>>>> Fetching new play <<<<<<<<<<<<<<<<<<<<<<<<<')
        #            play = Play(playItem)
        #            play.link = Lepistant.getURLFromTagContent(playItem)
        #            formatted_title = Lepistant.removeNonAlphanumericCharacters(play.title)
        #            play.file_path_on_disk = Lepistant.REL_PATH_PLAYS_FOLDER + formatted_title + '/'
        #            play.file_name_on_disk = Lepistant.createFilePath(
        #                                                play.file_path_on_disk,
        #                                                play.title,
        #                                                'html')
        #            soup = Lepistant.getSoup(play.link, play.file_name_on_disk)
        #            play.setPlayDetails(soup)
        #
        #            logger.info('')
        #
        #            plays.append(play)

        return plays
Пример #3
0
    def getSpecificTheaterPlays(leporello_info, file_name, play_type, url):
        theater_plays = []
        
        leporello_file_name_on_disk = Lepistant.createFilePath(
                                               Lepistant.REL_PATH_DOWNLOADS_FOLDER, 
                                               file_name, 
                                               'html')
        
        soup = Lepistant.getSoup(url, leporello_file_name_on_disk)
        
        theater_play_items = Lepistant.getTagsByClass(soup, 'div', leporello_info[CSS_CLASS_PLAY_ITEM])
        
#        for play_item in theater_play_items:
#            title = getTitleFromPlayItem(play_item)
#            theater_plays.append(title)
            
        for play_item in theater_play_items:
            logger.info('>>>>>>>>>>>>>>>>>>>>>>>> Fetching new play <<<<<<<<<<<<<<<<<<<<<<<<<')
            play = Play(play_item)
            play.setType(play_type)
            play.link = Lepistant.getURLFromTagContent(play_item)
            formatted_title = Lepistant.removeNonAlphanumericCharacters(play.title)
            play.file_path_on_disk = Lepistant.REL_PATH_PLAYS_FOLDER + formatted_title + '/'
            play.file_name_on_disk = Lepistant.createFilePath(
                                                play.file_path_on_disk, 
                                                play.title, 
                                                'html')
            soup = Lepistant.getSoup(play.link, play.file_name_on_disk)
            play.setPlayDetails(soup)
            
            setDefaultTimeForPerformancesInPlay(play)
            
            logger.info('')
            
            theater_plays.append(play)
            
        return theater_plays
Пример #4
0
    def getSpecificTheaterPlays(leporello_info, file_name, play_type, url):
        theater_plays = []

        leporello_file_name_on_disk = Lepistant.createFilePath(
            Lepistant.REL_PATH_DOWNLOADS_FOLDER, file_name, 'html')

        soup = Lepistant.getSoup(url, leporello_file_name_on_disk)

        theater_play_items = Lepistant.getTagsByClass(
            soup, 'div', leporello_info[CSS_CLASS_PLAY_ITEM])

        #        for play_item in theater_play_items:
        #            title = getTitleFromPlayItem(play_item)
        #            theater_plays.append(title)

        for play_item in theater_play_items:
            logger.info(
                '>>>>>>>>>>>>>>>>>>>>>>>> Fetching new play <<<<<<<<<<<<<<<<<<<<<<<<<'
            )
            play = Play(play_item)
            play.setType(play_type)
            play.link = Lepistant.getURLFromTagContent(play_item)
            formatted_title = Lepistant.removeNonAlphanumericCharacters(
                play.title)
            play.file_path_on_disk = Lepistant.REL_PATH_PLAYS_FOLDER + formatted_title + '/'
            play.file_name_on_disk = Lepistant.createFilePath(
                play.file_path_on_disk, play.title, 'html')
            soup = Lepistant.getSoup(play.link, play.file_name_on_disk)
            play.setPlayDetails(soup)

            setDefaultTimeForPerformancesInPlay(play)

            logger.info('')

            theater_plays.append(play)

        return theater_plays