[yt-dlp.git] / youtube_dl / extractor / twitch.py

from __future__ import unicode_literals

import itertools
import re

from .common import InfoExtractor
from ..utils import (
    ExtractorError,
    parse_iso8601,
)


class TwitchIE(InfoExtractor):
    # TODO: One broadcast may be split into multiple videos. The key
    # 'broadcast_id' is the same for all parts, and 'broadcast_part'
    # starts at 1 and increases. Can we treat all parts as one video?
    _VALID_URL = r"""(?x)^(?:http://)?(?:www\.)?twitch\.tv/
        (?:
            (?P<channelid>[^/]+)|
            (?:(?:[^/]+)/b/(?P<videoid>[^/]+))|
            (?:(?:[^/]+)/c/(?P<chapterid>[^/]+))
        )
        /?(?:\#.*)?$
        """
    _PAGE_LIMIT = 100
    _API_BASE = 'https://api.twitch.tv'
    _TESTS = [{
        'url': 'http://www.twitch.tv/riotgames/b/577357806',
        'info_dict': {
            'id': 'a577357806',
            'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
        },
        'playlist_mincount': 12,
    }, {
        'url': 'http://www.twitch.tv/acracingleague/c/5285812',
        'info_dict': {
            'id': 'c5285812',
            'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
        },
        'playlist_mincount': 3,
    }, {
        'url': 'http://www.twitch.tv/vanillatv',
        'info_dict': {
            'id': 'vanillatv',
            'title': 'VanillaTV',
        },
        'playlist_mincount': 412,
    }]

    def _handle_error(self, response):
        if not isinstance(response, dict):
            return
        error = response.get('error')
        if error:
            raise ExtractorError(
                '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
                expected=True)

    def _download_json(self, url, video_id, note='Downloading JSON metadata'):
        response = super(TwitchIE, self)._download_json(url, video_id, note)
        self._handle_error(response)
        return response

    def _extract_media(self, item, item_id):
        ITEMS = {
            'a': 'video',
            'c': 'chapter',
        }
        info = self._extract_info(self._download_json(
            '%s/kraken/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
            'Downloading %s info JSON' % ITEMS[item]))
        response = self._download_json(
            '%s/api/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
            'Downloading %s playlist JSON' % ITEMS[item])
        entries = []
        chunks = response['chunks']
        qualities = list(chunks.keys())
        for num, fragment in enumerate(zip(*chunks.values()), start=1):
            formats = []
            for fmt_num, fragment_fmt in enumerate(fragment):
                format_id = qualities[fmt_num]
                fmt = {
                    'url': fragment_fmt['url'],
                    'format_id': format_id,
                    'quality': 1 if format_id == 'live' else 0,
                }
                m = re.search(r'^(?P<height>\d+)[Pp]', format_id)
                if m:
                    fmt['height'] = int(m.group('height'))
                formats.append(fmt)
            self._sort_formats(formats)
            entry = dict(info)
            entry['id'] = '%s_%d' % (entry['id'], num)
            entry['title'] = '%s part %d' % (entry['title'], num)
            entry['formats'] = formats
            entries.append(entry)
        return self.playlist_result(entries, info['id'], info['title'])

    def _extract_info(self, info):
        return {
            'id': info['_id'],
            'title': info['title'],
            'description': info['description'],
            'duration': info['length'],
            'thumbnail': info['preview'],
            'uploader': info['channel']['display_name'],
            'uploader_id': info['channel']['name'],
            'timestamp': parse_iso8601(info['recorded_at']),
            'view_count': info['views'],
        }

    def _real_extract(self, url):
        mobj = re.match(self._VALID_URL, url)
        if mobj.group('chapterid'):
            return self._extract_media('c', mobj.group('chapterid'))

            """
            webpage = self._download_webpage(url, chapter_id)
            m = re.search(r'PP\.archive_id = "([0-9]+)";', webpage)
            if not m:
                raise ExtractorError('Cannot find archive of a chapter')
            archive_id = m.group(1)

            api = api_base + '/broadcast/by_chapter/%s.xml' % chapter_id
            doc = self._download_xml(
                api, chapter_id,
                note='Downloading chapter information',
                errnote='Chapter information download failed')
            for a in doc.findall('.//archive'):
                if archive_id == a.find('./id').text:
                    break
            else:
                raise ExtractorError('Could not find chapter in chapter information')

            video_url = a.find('./video_file_url').text
            video_ext = video_url.rpartition('.')[2] or 'flv'

            chapter_api_url = 'https://api.twitch.tv/kraken/videos/c' + chapter_id
            chapter_info = self._download_json(
                chapter_api_url, 'c' + chapter_id,
                note='Downloading chapter metadata',
                errnote='Download of chapter metadata failed')

            bracket_start = int(doc.find('.//bracket_start').text)
            bracket_end = int(doc.find('.//bracket_end').text)

            # TODO determine start (and probably fix up file)
            #  youtube-dl -v http://www.twitch.tv/firmbelief/c/1757457
            #video_url += '?start=' + TODO:start_timestamp
            # bracket_start is 13290, but we want 51670615
            self._downloader.report_warning('Chapter detected, but we can just download the whole file. '
                                            'Chapter starts at %s and ends at %s' % (formatSeconds(bracket_start), formatSeconds(bracket_end)))

            info = {
                'id': 'c' + chapter_id,
                'url': video_url,
                'ext': video_ext,
                'title': chapter_info['title'],
                'thumbnail': chapter_info['preview'],
                'description': chapter_info['description'],
                'uploader': chapter_info['channel']['display_name'],
                'uploader_id': chapter_info['channel']['name'],
            }
            return info
            """
        elif mobj.group('videoid'):
            return self._extract_media('a', mobj.group('videoid'))
        elif mobj.group('channelid'):
            channel_id = mobj.group('channelid')
            info = self._download_json(
                '%s/kraken/channels/%s' % (self._API_BASE, channel_id),
                channel_id, 'Downloading channel info JSON')
            channel_name = info.get('display_name') or info.get('name')
            entries = []
            offset = 0
            limit = self._PAGE_LIMIT
            for counter in itertools.count(1):
                response = self._download_json(
                    '%s/kraken/channels/%s/videos/?offset=%d&limit=%d'
                    % (self._API_BASE, channel_id, offset, limit),
                    channel_id, 'Downloading channel videos JSON page %d' % counter)
                videos = response['videos']
                if not videos:
                    break
                entries.extend([self.url_result(video['url'], 'Twitch') for video in videos])
                offset += limit
            return self.playlist_result(entries, channel_id, channel_name)
Commit	Line	Data
ee1e1996 PH	1	from __future__ import unicode_literals
ee1e1996 PH	2
3182f3e2	3	import itertools
79e93125	4	import re
79e93125 PH	5
	6	from .common import InfoExtractor
	7	from ..utils import (
	8	ExtractorError,
355d074f	9	parse_iso8601,
79e93125 PH	10	)
	11
	12
46fd0dd5	13	class TwitchIE(InfoExtractor):
79e93125 PH	14	# TODO: One broadcast may be split into multiple videos. The key
	15	# 'broadcast_id' is the same for all parts, and 'broadcast_part'
	16	# starts at 1 and increases. Can we treat all parts as one video?
46fd0dd5	17	_VALID_URL = r"""(?x)^(?:http://)?(?:www\.)?twitch\.tv/
79e93125 PH	18	(?:
	19	(?P<channelid>[^/]+)\|
	20	(?:(?:[^/]+)/b/(?P<videoid>[^/]+))\|
	21	(?:(?:[^/]+)/c/(?P<chapterid>[^/]+))
	22	)
	23	/?(?:\#.*)?$
	24	"""
46fd0dd5 S	25	_PAGE_LIMIT = 100
46fd0dd5 S	26	_API_BASE = 'https://api.twitch.tv'
013bfdd8 S	27	_TESTS = [{
013bfdd8 S	28	'url': 'http://www.twitch.tv/riotgames/b/577357806',
ee1e1996	29	'info_dict': {
013bfdd8 S	30	'id': 'a577357806',
	31	'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
	32	},
	33	'playlist_mincount': 12,
	34	}, {
	35	'url': 'http://www.twitch.tv/acracingleague/c/5285812',
	36	'info_dict': {
	37	'id': 'c5285812',
	38	'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
	39	},
	40	'playlist_mincount': 3,
	41	}, {
	42	'url': 'http://www.twitch.tv/vanillatv',
	43	'info_dict': {
	44	'id': 'vanillatv',
	45	'title': 'VanillaTV',
	46	},
	47	'playlist_mincount': 412,
	48	}]
79e93125	49
355d074f S	50	def _handle_error(self, response):
	51	if not isinstance(response, dict):
	52	return
	53	error = response.get('error')
	54	if error:
	55	raise ExtractorError(
	56	'%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
	57	expected=True)
	58
	59	def _download_json(self, url, video_id, note='Downloading JSON metadata'):
46fd0dd5	60	response = super(TwitchIE, self)._download_json(url, video_id, note)
355d074f S	61	self._handle_error(response)
	62	return response
	63
	64	def _extract_media(self, item, item_id):
355d074f S	65	ITEMS = {
	66	'a': 'video',
	67	'c': 'chapter',
	68	}
355d074f S	69	info = self._extract_info(self._download_json(
	70	'%s/kraken/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
	71	'Downloading %s info JSON' % ITEMS[item]))
355d074f S	72	response = self._download_json(
	73	'%s/api/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
	74	'Downloading %s playlist JSON' % ITEMS[item])
355d074f S	75	entries = []
	76	chunks = response['chunks']
	77	qualities = list(chunks.keys())
	78	for num, fragment in enumerate(zip(*chunks.values()), start=1):
	79	formats = []
	80	for fmt_num, fragment_fmt in enumerate(fragment):
	81	format_id = qualities[fmt_num]
	82	fmt = {
	83	'url': fragment_fmt['url'],
	84	'format_id': format_id,
	85	'quality': 1 if format_id == 'live' else 0,
	86	}
	87	m = re.search(r'^(?P<height>\d+)[Pp]', format_id)
	88	if m:
	89	fmt['height'] = int(m.group('height'))
	90	formats.append(fmt)
	91	self._sort_formats(formats)
	92	entry = dict(info)
159444a6	93	entry['id'] = '%s_%d' % (entry['id'], num)
355d074f S	94	entry['title'] = '%s part %d' % (entry['title'], num)
	95	entry['formats'] = formats
	96	entries.append(entry)
013bfdd8	97	return self.playlist_result(entries, info['id'], info['title'])
355d074f S	98
	99	def _extract_info(self, info):
	100	return {
	101	'id': info['_id'],
	102	'title': info['title'],
	103	'description': info['description'],
	104	'duration': info['length'],
	105	'thumbnail': info['preview'],
	106	'uploader': info['channel']['display_name'],
	107	'uploader_id': info['channel']['name'],
	108	'timestamp': parse_iso8601(info['recorded_at']),
	109	'view_count': info['views'],
	110	}
	111
79e93125 PH	112	def _real_extract(self, url):
79e93125 PH	113	mobj = re.match(self._VALID_URL, url)
46fd0dd5	114	if mobj.group('chapterid'):
355d074f	115	return self._extract_media('c', mobj.group('chapterid'))
79e93125	116
355d074f	117	"""
79e93125 PH	118	webpage = self._download_webpage(url, chapter_id)
	119	m = re.search(r'PP\.archive_id = "([0-9]+)";', webpage)
	120	if not m:
ee1e1996	121	raise ExtractorError('Cannot find archive of a chapter')
79e93125 PH	122	archive_id = m.group(1)
	123
	124	api = api_base + '/broadcast/by_chapter/%s.xml' % chapter_id
ee1e1996 PH	125	doc = self._download_xml(
	126	api, chapter_id,
	127	note='Downloading chapter information',
	128	errnote='Chapter information download failed')
79e93125 PH	129	for a in doc.findall('.//archive'):
	130	if archive_id == a.find('./id').text:
	131	break
	132	else:
ee1e1996	133	raise ExtractorError('Could not find chapter in chapter information')
79e93125 PH	134
79e93125 PH	135	video_url = a.find('./video_file_url').text
ee1e1996	136	video_ext = video_url.rpartition('.')[2] or 'flv'
79e93125	137
ee1e1996 PH	138	chapter_api_url = 'https://api.twitch.tv/kraken/videos/c' + chapter_id
	139	chapter_info = self._download_json(
	140	chapter_api_url, 'c' + chapter_id,
	141	note='Downloading chapter metadata',
	142	errnote='Download of chapter metadata failed')
79e93125 PH	143
	144	bracket_start = int(doc.find('.//bracket_start').text)
	145	bracket_end = int(doc.find('.//bracket_end').text)
	146
	147	# TODO determine start (and probably fix up file)
	148	# youtube-dl -v http://www.twitch.tv/firmbelief/c/1757457
ee1e1996	149	#video_url += '?start=' + TODO:start_timestamp
79e93125	150	# bracket_start is 13290, but we want 51670615
ee1e1996 PH	151	self._downloader.report_warning('Chapter detected, but we can just download the whole file. '
ee1e1996 PH	152	'Chapter starts at %s and ends at %s' % (formatSeconds(bracket_start), formatSeconds(bracket_end)))
79e93125 PH	153
79e93125 PH	154	info = {
ee1e1996	155	'id': 'c' + chapter_id,
79e93125 PH	156	'url': video_url,
	157	'ext': video_ext,
	158	'title': chapter_info['title'],
	159	'thumbnail': chapter_info['preview'],
	160	'description': chapter_info['description'],
	161	'uploader': chapter_info['channel']['display_name'],
	162	'uploader_id': chapter_info['channel']['name'],
	163	}
ee1e1996	164	return info
355d074f	165	"""
46fd0dd5	166	elif mobj.group('videoid'):
355d074f	167	return self._extract_media('a', mobj.group('videoid'))
46fd0dd5 S	168	elif mobj.group('channelid'):
	169	channel_id = mobj.group('channelid')
	170	info = self._download_json(
	171	'%s/kraken/channels/%s' % (self._API_BASE, channel_id),
	172	channel_id, 'Downloading channel info JSON')
	173	channel_name = info.get('display_name') or info.get('name')
	174	entries = []
	175	offset = 0
	176	limit = self._PAGE_LIMIT
	177	for counter in itertools.count(1):
	178	response = self._download_json(
	179	'%s/kraken/channels/%s/videos/?offset=%d&limit=%d'
	180	% (self._API_BASE, channel_id, offset, limit),
	181	channel_id, 'Downloading channel videos JSON page %d' % counter)
	182	videos = response['videos']
	183	if not videos:
	184	break
	185	entries.extend([self.url_result(video['url'], 'Twitch') for video in videos])
	186	offset += limit
	187	return self.playlist_result(entries, channel_id, channel_name)