ariblibで実装したepgdump互換のXMLを出力するスクリプト

epgdump.py

#!/usr/bin/env python3

from ariblib import tsopen
from ariblib.descriptors import ServiceDescriptor
from ariblib.sections import ActualStreamEventInformationSection, EventInformationSection, ActualStreamServiceDescriptionSection, ServiceDescriptionSection
from ariblib.service import tsid2channel, Service
from ariblib.event import Event

import sys
from datetime import timezone, timedelta
import lxml.etree as etree

CONTENT_TYPE_EN = {
    0x0: 'news',
    0x1: 'sports',
    0x2: 'information',
    0x3: 'drama',
    0x4: 'music',
    0x5: 'variety',
    0x6: 'cinema',
    0x7: 'anime',
    0x8: 'documentary',
    0x9: 'stage',
    0xA: 'hobby',
    0xB: 'welfare',
    0xE: 'etc',
    0xF: 'etc',
}

services = {}
events = {}
events_long = {}
active_service_ids = []

if sys.argv[1] in ('/BS', '/CS'):
    channel_id = None
else:
    channel_id = sys.argv[1]

with tsopen(sys.argv[2]) as ts:
    for section in ts.sections(ActualStreamEventInformationSection, ActualStreamServiceDescriptionSection):
        if isinstance(section, EventInformationSection):
            for e in section.events:
                event = Event(section, e)
                event_id = event.event_id
                if hasattr(event, 'title'):
                    events[event_id] = event
                if hasattr(event, 'longdesc'):
                    events_long[event_id] = event
                if event.service_id not in active_service_ids:
                    active_service_ids.append(event.service_id)
        elif isinstance(section, ServiceDescriptionSection):
            for s in section.services:
                if channel_id is None:
                    channel_id = tsid2channel(section.transport_stream_id)
                service = Service(s, channel_id)
                service_id = service.service_id
                if service_id not in services:
                    service.transport_stream_id = section.transport_stream_id
                    service.original_network_id = section.original_network_id
                    services[service_id] = service

elm_tv = etree.Element('tv', attrib = {'generator-info-name': 'tsEPG2xml', 'generator-info-url': 'http://localhost/'})

for service_id in sorted(active_service_ids):
    service = services[service_id]

    elm_channel = etree.SubElement(elm_tv, 'channel')
    elm_channel.attrib['id'] = '%s_%s' % (service.broadcasting_type, service.service_id)
    elm_channel.attrib['tp'] = service.channel_id

    elm_display_name = etree.SubElement(elm_channel, 'display-name', lang = 'ja_JP')
    elm_display_name.text = str(service.name)

    elm_transport = etree.SubElement(elm_channel, 'transport_stream_id')
    elm_transport.text = str(service.transport_stream_id)

    elm_network = etree.SubElement(elm_channel, 'original_network_id')
    elm_network.text = str(service.original_network_id)

    elm_service_id = etree.SubElement(elm_channel, 'service_id')
    elm_service_id.text = str(service_id)

for event_id, event in events.items():
    start_time = event.start_time
    end_time = event.start_time + event.duration
    service = services[event.service_id]

    elm_programme = etree.SubElement(elm_tv, 'programme')
    elm_programme.attrib['start'] = start_time.strftime('%Y%m%d%H%M%S +0900')
    elm_programme.attrib['stop'] = end_time.strftime('%Y%m%d%H%M%S +0900')
    elm_programme.attrib['channel'] = '%s_%s' % (service.broadcasting_type, service.service_id)
    elm_programme.attrib['event_id'] = str(event.event_id)

    elm_title = etree.SubElement(elm_programme, 'title', lang = 'ja_JP')
    elm_title.text = str(event.title).replace('　', ' ')

    elm_desc = etree.SubElement(elm_programme, 'desc', lang = 'ja_JP')
    if event_id in events_long:
        elm_desc.text = '%s\n\n%s' % (event.desc, events_long[event_id].longdesc)
    else:
        elm_desc.text = str(event.desc)

    elm_category_ja = etree.SubElement(elm_programme, 'category', lang = 'ja_JP')
    elm_category_en = etree.SubElement(elm_programme, 'category', lang = 'en')

    if hasattr(event, 'genre') and event.nibble1[0] < 0x08:
        elm_category_ja.text = str(event.genre[0])
        elm_category_en.text = CONTENT_TYPE_EN[event.nibble1[0]]
    else:
        elm_category_ja.text = 'その他'
        elm_category_en.text = 'etc'

print('<?xml version="1.0" encoding="UTF-8"?>\n')
print('<!DOCTYPE tv SYSTEM "xmltv.dtd">\n\n')
print(etree.tounicode(elm_tv, pretty_print=True))

• /BS /CS オプションは動作未確認
• <desc> に拡張形式イベント記述子の内容を追記しているので、Chinachu で出演者等の情報も閲覧・検索できる
• 元の epgdump と違って英数字が全て半角で出力されるため、既存の録画ルールを修正する必要がある
• 107行目付近で event.nibble1[0] < 0x08 以外を etc にしているのは epgdump のバグに合わせたため