part_of_epg2xml.py

# Write Program
buf = {}
def writeProgram(programdata):
    global buf
    ChannelId = programdata['channelId']
    startTime = programdata['startTime']
    endTime = programdata['endTime']
    programName = escape(programdata['programName']).strip()
    subprogramName = escape(programdata['subprogramName']).strip()
    matches = re.match('(.*) \(?(\d+부)\)?', unescape(programName.encode('utf-8', 'ignore')))
    if not(matches is None):
        programName = escape(matches.group(1)).strip();
        subprogramName = escape(matches.group(2)) + ' ' + subprogramName
        subprogramName = subprogramName.strip()
    if programName is None:
        programName = subprogramName
    actors = escape(programdata['actors'])
    producers = escape(programdata['producers'])
    category = escape(programdata['category'])
    episode = programdata['episode']
    if episode:
        try:
            episode_ns = int(episode) - 1
            episode_ns = '0'+ '.' +  str(episode_ns) + '.' + '0' + '/' + '0'
        except ValueError as ex:
            episode_ns = int(episode.split(',', 1)[0]) - 1
            episode_ns = '0'+ '.' +  str(episode_ns) + '.' + '0' + '/' + '0'
        episode_on = episode
    else:
        episode_ns = ''
        episode_on = ''
    rebroadcast = programdata['rebroadcast']
    if episode and addepisode  == 'y': programName = programName + ' ('+ str(episode) + '회)'
    if rebroadcast  == True and addrebroadcast == 'y' : programName = programName + ' (재)'
    if programdata['rating'] == 0 :
        rating = '전체 관람가'
    else :
        rating = '%s세 이상 관람가' % (programdata['rating'])
    if addverbose == 'y':
        desc = programName
        if subprogramName : desc = desc + '\n부제 : ' + subprogramName
        if rebroadcast == True and addrebroadcast == 'y' : desc = desc + '\n방송 : 재방송'
        if episode : desc = desc + '\n회차 : ' + str(episode) + '회'
        if category : desc = desc + '\n장르 : ' + category
        if actors : desc = desc + '\n출연 : ' + actors.strip()
        if producers : desc = desc + '\n제작 : ' + producers.strip()
        desc = desc + '\n등급 : ' + rating
    else:
        desc =''
    if programdata['desc'] : desc = desc + '\n' + escape(programdata['desc'])
    desc = re.sub(' +',' ', desc)
    contentTypeDict={'교양':'Arts / Culture (without music)', '만화':'Cartoons / Puppets', '교육':'Education / Science / Factual topics', '취미':'Leisure hobbies', '드라마':'Movie / Drama', '영화':'Movie / Drama', '음악':'Music / Ballet / Dance', '뉴스':'News / Current affairs', '다큐':'Documentary', '라이프':'Documentary', '시사/다큐':'Documentary', '연예':'Show / Game show', '스포츠':'Sports', '홈쇼핑':'Advertisement / Shopping'}
    contentType = ''
    for key, value in contentTypeDict.iteritems():
        if key in category:
            contentType = value
    try:
        splitProgram = buf['ChannelId'] == ChannelId and programName != '' and buf['programName'] == programName and buf['episode'] == episode
    except KeyError:
        splitProgram = False

    if splitProgram:
        buf['endTime'] = endTime
        buf['subprogramName'] = ''
        return
    elif buf:
        printBuffer()

    buf = {
        'startTime': startTime,
        'endTime': endTime,
        'ChannelId': ChannelId,
        'programName': programName,
        'subprogramName': subprogramName,
        'desc': desc,
        'actors': actors,
        'producers': producers,
        'category': category,
        'contentType': contentType,
        'episode': episode,
        'episode_ns': episode_ns,
        'episode_on': episode_on,
        'rebroadcast': rebroadcast,
        'rating': rating,
    }

def printBuffer():
    if not buf:
        return

    print('  <programme start="%s +0900" stop="%s +0900" channel="%s">' % (buf['startTime'], buf['endTime'], buf['ChannelId']))
    print('    <title lang="kr">%s</title>' % (buf['programName']))
    if buf['subprogramName']:
        print('    <sub-title lang="kr">%s</sub-title>' % (buf['subprogramName']))
    if addverbose == 'y':
        print('    <desc lang="kr">%s</desc>' % (buf['desc']))
        if buf['actors'] or buf['producers']:
            print('    <credits>')
            if buf['actors']:
                for actor in buf['actors'].split(','):
                    if actor.strip(): print('      <actor>%s</actor>' % (actor.strip()))
            if buf['producers']:
                for producer in buf['producers'].split(','):
                    if producer.strip(): print('      <producer>%s</producer>' % (producer).strip())
            print('    </credits>')
    if buf['category']: print('    <category lang="kr">%s</category>' % (buf['category']))
    if buf['contentType']: print('    <category lang="en">%s</category>' % (buf['contentType']))
    if buf['episode'] and addxmltvns == 'y': print('    <episode-num system="xmltv_ns">%s</episode-num>' % (buf['episode_ns']))
    if buf['episode'] and addxmltvns != 'y': print('    <episode-num system="onscreen">%s</episode-num>' % (buf['episode_on']))
    if buf['rebroadcast']: print('    <previously-shown />')
    if buf['rating']:
        print('    <rating system="KMRB">')
        print('      <value>%s</value>' % (buf['rating']))
        print('    </rating>')
    print('  </programme>')