youtube-dl/youtube_dl/extractor/twitch.py

# coding: utf-8
from __future__ import unicode_literals

import collections
import itertools
import json
import random
import re

from .common import InfoExtractor
from ..compat import (
    compat_kwargs,
    compat_parse_qs,
    compat_str,
    compat_urlparse,
    compat_urllib_parse_urlencode,
    compat_urllib_parse_urlparse,
)
from ..utils import (
    clean_html,
    ExtractorError,
    float_or_none,
    int_or_none,
    parse_duration,
    parse_iso8601,
    qualities,
    try_get,
    unified_timestamp,
    update_url_query,
    url_or_none,
    urljoin,
)


class TwitchBaseIE(InfoExtractor):
    _VALID_URL_BASE = r'https?://(?:(?:www|go|m)\.)?twitch\.tv'

    _API_BASE = 'https://api.twitch.tv'
    _USHER_BASE = 'https://usher.ttvnw.net'
    _LOGIN_FORM_URL = 'https://www.twitch.tv/login'
    _LOGIN_POST_URL = 'https://passport.twitch.tv/login'
    _CLIENT_ID = 'kimne78kx3ncx6brgo4mv6wki5h1ko'
    _NETRC_MACHINE = 'twitch'

    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 _call_api(self, path, item_id, *args, **kwargs):
        headers = kwargs.get('headers', {}).copy()
        headers.update({
            'Accept': 'application/vnd.twitchtv.v5+json; charset=UTF-8',
            'Client-ID': self._CLIENT_ID,
        })
        kwargs.update({
            'headers': headers,
            'expected_status': (400, 410),
        })
        response = self._download_json(
            '%s/%s' % (self._API_BASE, path), item_id,
            *args, **compat_kwargs(kwargs))
        self._handle_error(response)
        return response

    def _real_initialize(self):
        self._login()

    def _login(self):
        username, password = self._get_login_info()
        if username is None:
            return

        def fail(message):
            raise ExtractorError(
                'Unable to login. Twitch said: %s' % message, expected=True)

        def login_step(page, urlh, note, data):
            form = self._hidden_inputs(page)
            form.update(data)

            page_url = urlh.geturl()
            post_url = self._search_regex(
                r'<form[^>]+action=(["\'])(?P<url>.+?)\1', page,
                'post url', default=self._LOGIN_POST_URL, group='url')
            post_url = urljoin(page_url, post_url)

            headers = {
                'Referer': page_url,
                'Origin': page_url,
                'Content-Type': 'text/plain;charset=UTF-8',
            }

            response = self._download_json(
                post_url, None, note, data=json.dumps(form).encode(),
                headers=headers, expected_status=400)
            error = response.get('error_description') or response.get('error_code')
            if error:
                fail(error)

            if 'Authenticated successfully' in response.get('message', ''):
                return None, None

            redirect_url = urljoin(
                post_url,
                response.get('redirect') or response['redirect_path'])
            return self._download_webpage_handle(
                redirect_url, None, 'Downloading login redirect page',
                headers=headers)

        login_page, handle = self._download_webpage_handle(
            self._LOGIN_FORM_URL, None, 'Downloading login page')

        # Some TOR nodes and public proxies are blocked completely
        if 'blacklist_message' in login_page:
            fail(clean_html(login_page))

        redirect_page, handle = login_step(
            login_page, handle, 'Logging in', {
                'username': username,
                'password': password,
                'client_id': self._CLIENT_ID,
            })

        # Successful login
        if not redirect_page:
            return

        if re.search(r'(?i)<form[^>]+id="two-factor-submit"', redirect_page) is not None:
            # TODO: Add mechanism to request an SMS or phone call
            tfa_token = self._get_tfa_info('two-factor authentication token')
            login_step(redirect_page, handle, 'Submitting TFA token', {
                'authy_token': tfa_token,
                'remember_2fa': 'true',
            })

    def _prefer_source(self, formats):
        try:
            source = next(f for f in formats if f['format_id'] == 'Source')
            source['quality'] = 10
        except StopIteration:
            for f in formats:
                if '/chunked/' in f['url']:
                    f.update({
                        'quality': 10,
                        'format_note': 'Source',
                    })
        self._sort_formats(formats)

    def _download_access_token(self, channel_name):
        return self._call_api(
            'api/channels/%s/access_token' % channel_name, channel_name,
            'Downloading access token JSON')

    def _extract_channel_id(self, token, channel_name):
        return compat_str(self._parse_json(token, channel_name)['channel_id'])


class TwitchGraphQLBaseIE(TwitchBaseIE):
    _PAGE_LIMIT = 100

    _OPERATION_HASHES = {
        'CollectionSideBar': '27111f1b382effad0b6def325caef1909c733fe6a4fbabf54f8d491ef2cf2f14',
        'FilterableVideoTower_Videos': 'a937f1d22e269e39a03b509f65a7490f9fc247d7f83d6ac1421523e3b68042cb',
        'ClipsCards__User': 'b73ad2bfaecfd30a9e6c28fada15bd97032c83ec77a0440766a56fe0bd632777',
        'ChannelCollectionsContent': '07e3691a1bad77a36aba590c351180439a40baefc1c275356f40fc7082419a84',
        'StreamMetadata': '1c719a40e481453e5c48d9bb585d971b8b372f8ebb105b17076722264dfa5b3e',
        'ComscoreStreamingQuery': 'e1edae8122517d013405f237ffcc124515dc6ded82480a88daef69c83b53ac01',
        'VideoPreviewOverlay': '3006e77e51b128d838fa4e835723ca4dc9a05c5efd4466c1085215c6e437e65c',
        'VideoMetadata': '226edb3e692509f727fd56821f5653c05740242c82b0388883e0c0e75dcbf687',
    }

    def _download_base_gql(self, video_id, ops, note, fatal=True):
        return self._download_json(
            'https://gql.twitch.tv/gql', video_id, note,
            data=json.dumps(ops).encode(),
            headers={
                'Content-Type': 'text/plain;charset=UTF-8',
                'Client-ID': self._CLIENT_ID,
            }, fatal=fatal)

    def _download_gql(self, video_id, ops, note, fatal=True):
        for op in ops:
            op['extensions'] = {
                'persistedQuery': {
                    'version': 1,
                    'sha256Hash': self._OPERATION_HASHES[op['operationName']],
                }
            }
        return self._download_base_gql(video_id, ops, note)

    def _download_access_token_gql(self, video_id, token_kind, param_name):
        method = '%sPlaybackAccessToken' % token_kind
        ops = {
            'query': '''{
              %s(
                %s: "%s",
                params: {
                  platform: "web",
                  playerBackend: "mediaplayer",
                  playerType: "site"
                }
              )
              {
                value
                signature
              }
            }''' % (method, param_name, video_id),
        }
        return self._download_base_gql(
            video_id, ops,
            'Downloading %s access token GraphQL' % token_kind)['data'][method]


class TwitchVodIE(TwitchGraphQLBaseIE):
    IE_NAME = 'twitch:vod'
    _VALID_URL = r'''(?x)
                    https?://
                        (?:
                            (?:(?:www|go|m)\.)?twitch\.tv/(?:[^/]+/v(?:ideo)?|videos)/|
                            player\.twitch\.tv/\?.*?\bvideo=v?
                        )
                        (?P<id>\d+)
                    '''

    _TESTS = [{
        'url': 'http://www.twitch.tv/riotgames/v/6528877?t=5m10s',
        'info_dict': {
            'id': 'v6528877',
            'ext': 'mp4',
            'title': 'LCK Summer Split - Week 6 Day 1',
            'thumbnail': r're:^https?://.*\.jpg$',
            'duration': 17208,
            'timestamp': 1435131734,
            'upload_date': '20150624',
            'uploader': 'Riot Games',
            'uploader_id': 'riotgames',
            'view_count': int,
            'start_time': 310,
        },
        'params': {
            # m3u8 download
            'skip_download': True,
        },
    }, {
        # Untitled broadcast (title is None)
        'url': 'http://www.twitch.tv/belkao_o/v/11230755',
        'info_dict': {
            'id': 'v11230755',
            'ext': 'mp4',
            'title': 'Untitled Broadcast',
            'thumbnail': r're:^https?://.*\.jpg$',
            'duration': 1638,
            'timestamp': 1439746708,
            'upload_date': '20150816',
            'uploader': 'BelkAO_o',
            'uploader_id': 'belkao_o',
            'view_count': int,
        },
        'params': {
            # m3u8 download
            'skip_download': True,
        },
        'skip': 'HTTP Error 404: Not Found',
    }, {
        'url': 'http://player.twitch.tv/?t=5m10s&video=v6528877',
        'only_matching': True,
    }, {
        'url': 'https://www.twitch.tv/videos/6528877',
        'only_matching': True,
    }, {
        'url': 'https://m.twitch.tv/beagsandjam/v/247478721',
        'only_matching': True,
    }, {
        'url': 'https://www.twitch.tv/northernlion/video/291940395',
        'only_matching': True,
    }, {
        'url': 'https://player.twitch.tv/?video=480452374',
        'only_matching': True,
    }]

    def _download_info(self, item_id):
        return self._extract_info(
            self._call_api(
                'kraken/videos/%s' % item_id, item_id,
                'Downloading video info JSON'))

    def _download_info_gql(self, item_id):
        data = self._download_gql(
            item_id, [{
                'operationName': 'VideoMetadata',
                'variables': {
                    'channelLogin': '',
                    'videoID': item_id,
                },
            }],
            'Downloading stream metadata GraphQL')[0]['data']
        video = data.get('video')
        if video is None:
            raise ExtractorError(
                'Video %s does not exist' % item_id, expected=True)
        return self._extract_info_gql(video, item_id)

    @staticmethod
    def _extract_info(info):
        status = info.get('status')
        if status == 'recording':
            is_live = True
        elif status == 'recorded':
            is_live = False
        else:
            is_live = None
        _QUALITIES = ('small', 'medium', 'large')
        quality_key = qualities(_QUALITIES)
        thumbnails = []
        preview = info.get('preview')
        if isinstance(preview, dict):
            for thumbnail_id, thumbnail_url in preview.items():
                thumbnail_url = url_or_none(thumbnail_url)
                if not thumbnail_url:
                    continue
                if thumbnail_id not in _QUALITIES:
                    continue
                thumbnails.append({
                    'url': thumbnail_url,
                    'preference': quality_key(thumbnail_id),
                })
        return {
            'id': info['_id'],
            'title': info.get('title') or 'Untitled Broadcast',
            'description': info.get('description'),
            'duration': int_or_none(info.get('length')),
            'thumbnails': thumbnails,
            'uploader': info.get('channel', {}).get('display_name'),
            'uploader_id': info.get('channel', {}).get('name'),
            'timestamp': parse_iso8601(info.get('recorded_at')),
            'view_count': int_or_none(info.get('views')),
            'is_live': is_live,
        }

    @staticmethod
    def _extract_info_gql(info, item_id):
        vod_id = info.get('id') or item_id
        # id backward compatibility for download archives
        if vod_id[0] != 'v':
            vod_id = 'v%s' % vod_id
        thumbnail = url_or_none(info.get('previewThumbnailURL'))
        if thumbnail:
            for p in ('width', 'height'):
                thumbnail = thumbnail.replace('{%s}' % p, '0')
        return {
            'id': vod_id,
            'title': info.get('title') or 'Untitled Broadcast',
            'description': info.get('description'),
            'duration': int_or_none(info.get('lengthSeconds')),
            'thumbnail': thumbnail,
            'uploader': try_get(info, lambda x: x['owner']['displayName'], compat_str),
            'uploader_id': try_get(info, lambda x: x['owner']['login'], compat_str),
            'timestamp': unified_timestamp(info.get('publishedAt')),
            'view_count': int_or_none(info.get('viewCount')),
        }

    def _real_extract(self, url):
        vod_id = self._match_id(url)

        info = self._download_info_gql(vod_id)
        access_token = self._download_access_token_gql(vod_id, 'video', 'id')

        formats = self._extract_m3u8_formats(
            '%s/vod/%s.m3u8?%s' % (
                self._USHER_BASE, vod_id,
                compat_urllib_parse_urlencode({
                    'allow_source': 'true',
                    'allow_audio_only': 'true',
                    'allow_spectre': 'true',
                    'player': 'twitchweb',
                    'playlist_include_framerate': 'true',
                    'nauth': access_token['value'],
                    'nauthsig': access_token['signature'],
                })),
            vod_id, 'mp4', entry_protocol='m3u8_native')

        self._prefer_source(formats)
        info['formats'] = formats

        parsed_url = compat_urllib_parse_urlparse(url)
        query = compat_parse_qs(parsed_url.query)
        if 't' in query:
            info['start_time'] = parse_duration(query['t'][0])

        if info.get('timestamp') is not None:
            info['subtitles'] = {
                'rechat': [{
                    'url': update_url_query(
                        'https://api.twitch.tv/v5/videos/%s/comments' % vod_id, {
                            'client_id': self._CLIENT_ID,
                        }),
                    'ext': 'json',
                }],
            }

        return info


def _make_video_result(node):
    assert isinstance(node, dict)
    video_id = node.get('id')
    if not video_id:
        return
    return {
        '_type': 'url_transparent',
        'ie_key': TwitchVodIE.ie_key(),
        'id': video_id,
        'url': 'https://www.twitch.tv/videos/%s' % video_id,
        'title': node.get('title'),
        'thumbnail': node.get('previewThumbnailURL'),
        'duration': float_or_none(node.get('lengthSeconds')),
        'view_count': int_or_none(node.get('viewCount')),
    }


class TwitchCollectionIE(TwitchGraphQLBaseIE):
    _VALID_URL = r'https?://(?:(?:www|go|m)\.)?twitch\.tv/collections/(?P<id>[^/]+)'

    _TESTS = [{
        'url': 'https://www.twitch.tv/collections/wlDCoH0zEBZZbQ',
        'info_dict': {
            'id': 'wlDCoH0zEBZZbQ',
            'title': 'Overthrow Nook, capitalism for children',
        },
        'playlist_mincount': 13,
    }]

    _OPERATION_NAME = 'CollectionSideBar'

    def _real_extract(self, url):
        collection_id = self._match_id(url)
        collection = self._download_gql(
            collection_id, [{
                'operationName': self._OPERATION_NAME,
                'variables': {'collectionID': collection_id},
            }],
            'Downloading collection GraphQL')[0]['data']['collection']
        title = collection.get('title')
        entries = []
        for edge in collection['items']['edges']:
            if not isinstance(edge, dict):
                continue
            node = edge.get('node')
            if not isinstance(node, dict):
                continue
            video = _make_video_result(node)
            if video:
                entries.append(video)
        return self.playlist_result(
            entries, playlist_id=collection_id, playlist_title=title)


class TwitchPlaylistBaseIE(TwitchGraphQLBaseIE):
    def _entries(self, channel_name, *args):
        cursor = None
        variables_common = self._make_variables(channel_name, *args)
        entries_key = '%ss' % self._ENTRY_KIND
        for page_num in itertools.count(1):
            variables = variables_common.copy()
            variables['limit'] = self._PAGE_LIMIT
            if cursor:
                variables['cursor'] = cursor
            page = self._download_gql(
                channel_name, [{
                    'operationName': self._OPERATION_NAME,
                    'variables': variables,
                }],
                'Downloading %ss GraphQL page %s' % (self._NODE_KIND, page_num),
                fatal=False)
            if not page:
                break
            edges = try_get(
                page, lambda x: x[0]['data']['user'][entries_key]['edges'], list)
            if not edges:
                break
            for edge in edges:
                if not isinstance(edge, dict):
                    continue
                if edge.get('__typename') != self._EDGE_KIND:
                    continue
                node = edge.get('node')
                if not isinstance(node, dict):
                    continue
                if node.get('__typename') != self._NODE_KIND:
                    continue
                entry = self._extract_entry(node)
                if entry:
                    cursor = edge.get('cursor')
                    yield entry
            if not cursor or not isinstance(cursor, compat_str):
                break

    # Deprecated kraken v5 API
    def _entries_kraken(self, channel_name, broadcast_type, sort):
        access_token = self._download_access_token(channel_name)
        channel_id = self._extract_channel_id(access_token['token'], channel_name)
        offset = 0
        counter_override = None
        for counter in itertools.count(1):
            response = self._call_api(
                'kraken/channels/%s/videos/' % channel_id,
                channel_id,
                'Downloading video JSON page %s' % (counter_override or counter),
                query={
                    'offset': offset,
                    'limit': self._PAGE_LIMIT,
                    'broadcast_type': broadcast_type,
                    'sort': sort,
                })
            videos = response.get('videos')
            if not isinstance(videos, list):
                break
            for video in videos:
                if not isinstance(video, dict):
                    continue
                video_url = url_or_none(video.get('url'))
                if not video_url:
                    continue
                yield {
                    '_type': 'url_transparent',
                    'ie_key': TwitchVodIE.ie_key(),
                    'id': video.get('_id'),
                    'url': video_url,
                    'title': video.get('title'),
                    'description': video.get('description'),
                    'timestamp': unified_timestamp(video.get('published_at')),
                    'duration': float_or_none(video.get('length')),
                    'view_count': int_or_none(video.get('views')),
                    'language': video.get('language'),
                }
            offset += self._PAGE_LIMIT
            total = int_or_none(response.get('_total'))
            if total and offset >= total:
                break


class TwitchVideosIE(TwitchPlaylistBaseIE):
    _VALID_URL = r'https?://(?:(?:www|go|m)\.)?twitch\.tv/(?P<id>[^/]+)/(?:videos|profile)'

    _TESTS = [{
        # All Videos sorted by Date
        'url': 'https://www.twitch.tv/spamfish/videos?filter=all',
        'info_dict': {
            'id': 'spamfish',
            'title': 'spamfish - All Videos sorted by Date',
        },
        'playlist_mincount': 924,
    }, {
        # All Videos sorted by Popular
        'url': 'https://www.twitch.tv/spamfish/videos?filter=all&sort=views',
        'info_dict': {
            'id': 'spamfish',
            'title': 'spamfish - All Videos sorted by Popular',
        },
        'playlist_mincount': 931,
    }, {
        # Past Broadcasts sorted by Date
        'url': 'https://www.twitch.tv/spamfish/videos?filter=archives',
        'info_dict': {
            'id': 'spamfish',
            'title': 'spamfish - Past Broadcasts sorted by Date',
        },
        'playlist_mincount': 27,
    }, {
        # Highlights sorted by Date
        'url': 'https://www.twitch.tv/spamfish/videos?filter=highlights',
        'info_dict': {
            'id': 'spamfish',
            'title': 'spamfish - Highlights sorted by Date',
        },
        'playlist_mincount': 901,
    }, {
        # Uploads sorted by Date
        'url': 'https://www.twitch.tv/esl_csgo/videos?filter=uploads&sort=time',
        'info_dict': {
            'id': 'esl_csgo',
            'title': 'esl_csgo - Uploads sorted by Date',
        },
        'playlist_mincount': 5,
    }, {
        # Past Premieres sorted by Date
        'url': 'https://www.twitch.tv/spamfish/videos?filter=past_premieres',
        'info_dict': {
            'id': 'spamfish',
            'title': 'spamfish - Past Premieres sorted by Date',
        },
        'playlist_mincount': 1,
    }, {
        'url': 'https://www.twitch.tv/spamfish/videos/all',
        'only_matching': True,
    }, {
        'url': 'https://m.twitch.tv/spamfish/videos/all',
        'only_matching': True,
    }, {
        'url': 'https://www.twitch.tv/spamfish/videos',
        'only_matching': True,
    }]

    Broadcast = collections.namedtuple('Broadcast', ['type', 'label'])

    _DEFAULT_BROADCAST = Broadcast(None, 'All Videos')
    _BROADCASTS = {
        'archives': Broadcast('ARCHIVE', 'Past Broadcasts'),
        'highlights': Broadcast('HIGHLIGHT', 'Highlights'),
        'uploads': Broadcast('UPLOAD', 'Uploads'),
        'past_premieres': Broadcast('PAST_PREMIERE', 'Past Premieres'),
        'all': _DEFAULT_BROADCAST,
    }

    _DEFAULT_SORTED_BY = 'Date'
    _SORTED_BY = {
        'time': _DEFAULT_SORTED_BY,
        'views': 'Popular',
    }

    _OPERATION_NAME = 'FilterableVideoTower_Videos'
    _ENTRY_KIND = 'video'
    _EDGE_KIND = 'VideoEdge'
    _NODE_KIND = 'Video'

    @classmethod
    def suitable(cls, url):
        return (False
                if any(ie.suitable(url) for ie in (
                    TwitchVideosClipsIE,
                    TwitchVideosCollectionsIE))
                else super(TwitchVideosIE, cls).suitable(url))

    @staticmethod
    def _make_variables(channel_name, broadcast_type, sort):
        return {
            'channelOwnerLogin': channel_name,
            'broadcastType': broadcast_type,
            'videoSort': sort.upper(),
        }

    @staticmethod
    def _extract_entry(node):
        return _make_video_result(node)

    def _real_extract(self, url):
        channel_name = self._match_id(url)
        qs = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query)
        filter = qs.get('filter', ['all'])[0]
        sort = qs.get('sort', ['time'])[0]
        broadcast = self._BROADCASTS.get(filter, self._DEFAULT_BROADCAST)
        return self.playlist_result(
            self._entries(channel_name, broadcast.type, sort),
            playlist_id=channel_name,
            playlist_title='%s - %s sorted by %s'
            % (channel_name, broadcast.label,
               self._SORTED_BY.get(sort, self._DEFAULT_SORTED_BY)))


class TwitchVideosClipsIE(TwitchPlaylistBaseIE):
    _VALID_URL = r'https?://(?:(?:www|go|m)\.)?twitch\.tv/(?P<id>[^/]+)/(?:clips|videos/*?\?.*?\bfilter=clips)'

    _TESTS = [{
        # Clips
        'url': 'https://www.twitch.tv/vanillatv/clips?filter=clips&range=all',
        'info_dict': {
            'id': 'vanillatv',
            'title': 'vanillatv - Clips Top All',
        },
        'playlist_mincount': 1,
    }, {
        'url': 'https://www.twitch.tv/dota2ruhub/videos?filter=clips&range=7d',
        'only_matching': True,
    }]

    Clip = collections.namedtuple('Clip', ['filter', 'label'])

    _DEFAULT_CLIP = Clip('LAST_WEEK', 'Top 7D')
    _RANGE = {
        '24hr': Clip('LAST_DAY', 'Top 24H'),
        '7d': _DEFAULT_CLIP,
        '30d': Clip('LAST_MONTH', 'Top 30D'),
        'all': Clip('ALL_TIME', 'Top All'),
    }

    # NB: values other than 20 result in skipped videos
    _PAGE_LIMIT = 20

    _OPERATION_NAME = 'ClipsCards__User'
    _ENTRY_KIND = 'clip'
    _EDGE_KIND = 'ClipEdge'
    _NODE_KIND = 'Clip'

    @staticmethod
    def _make_variables(channel_name, filter):
        return {
            'login': channel_name,
            'criteria': {
                'filter': filter,
            },
        }

    @staticmethod
    def _extract_entry(node):
        assert isinstance(node, dict)
        clip_url = url_or_none(node.get('url'))
        if not clip_url:
            return
        return {
            '_type': 'url_transparent',
            'ie_key': TwitchClipsIE.ie_key(),
            'id': node.get('id'),
            'url': clip_url,
            'title': node.get('title'),
            'thumbnail': node.get('thumbnailURL'),
            'duration': float_or_none(node.get('durationSeconds')),
            'timestamp': unified_timestamp(node.get('createdAt')),
            'view_count': int_or_none(node.get('viewCount')),
            'language': node.get('language'),
        }

    def _real_extract(self, url):
        channel_name = self._match_id(url)
        qs = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query)
        range = qs.get('range', ['7d'])[0]
        clip = self._RANGE.get(range, self._DEFAULT_CLIP)
        return self.playlist_result(
            self._entries(channel_name, clip.filter),
            playlist_id=channel_name,
            playlist_title='%s - Clips %s' % (channel_name, clip.label))


class TwitchVideosCollectionsIE(TwitchPlaylistBaseIE):
    _VALID_URL = r'https?://(?:(?:www|go|m)\.)?twitch\.tv/(?P<id>[^/]+)/videos/*?\?.*?\bfilter=collections'

    _TESTS = [{
        # Collections
        'url': 'https://www.twitch.tv/spamfish/videos?filter=collections',
        'info_dict': {
            'id': 'spamfish',
            'title': 'spamfish - Collections',
        },
        'playlist_mincount': 3,
    }]

    _OPERATION_NAME = 'ChannelCollectionsContent'
    _ENTRY_KIND = 'collection'
    _EDGE_KIND = 'CollectionsItemEdge'
    _NODE_KIND = 'Collection'

    @staticmethod
    def _make_variables(channel_name):
        return {
            'ownerLogin': channel_name,
        }

    @staticmethod
    def _extract_entry(node):
        assert isinstance(node, dict)
        collection_id = node.get('id')
        if not collection_id:
            return
        return {
            '_type': 'url_transparent',
            'ie_key': TwitchCollectionIE.ie_key(),
            'id': collection_id,
            'url': 'https://www.twitch.tv/collections/%s' % collection_id,
            'title': node.get('title'),
            'thumbnail': node.get('thumbnailURL'),
            'duration': float_or_none(node.get('lengthSeconds')),
            'timestamp': unified_timestamp(node.get('updatedAt')),
            'view_count': int_or_none(node.get('viewCount')),
        }

    def _real_extract(self, url):
        channel_name = self._match_id(url)
        return self.playlist_result(
            self._entries(channel_name), playlist_id=channel_name,
            playlist_title='%s - Collections' % channel_name)


class TwitchStreamIE(TwitchGraphQLBaseIE):
    IE_NAME = 'twitch:stream'
    _VALID_URL = r'''(?x)
                    https?://
                        (?:
                            (?:(?:www|go|m)\.)?twitch\.tv/|
                            player\.twitch\.tv/\?.*?\bchannel=
                        )
                        (?P<id>[^/#?]+)
                    '''

    _TESTS = [{
        'url': 'http://www.twitch.tv/shroomztv',
        'info_dict': {
            'id': '12772022048',
            'display_id': 'shroomztv',
            'ext': 'mp4',
            'title': 're:^ShroomzTV [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
            'description': 'H1Z1 - lonewolfing with ShroomzTV | A3 Battle Royale later - @ShroomzTV',
            'is_live': True,
            'timestamp': 1421928037,
            'upload_date': '20150122',
            'uploader': 'ShroomzTV',
            'uploader_id': 'shroomztv',
            'view_count': int,
        },
        'params': {
            # m3u8 download
            'skip_download': True,
        },
    }, {
        'url': 'http://www.twitch.tv/miracle_doto#profile-0',
        'only_matching': True,
    }, {
        'url': 'https://player.twitch.tv/?channel=lotsofs',
        'only_matching': True,
    }, {
        'url': 'https://go.twitch.tv/food',
        'only_matching': True,
    }, {
        'url': 'https://m.twitch.tv/food',
        'only_matching': True,
    }]

    @classmethod
    def suitable(cls, url):
        return (False
                if any(ie.suitable(url) for ie in (
                    TwitchVodIE,
                    TwitchCollectionIE,
                    TwitchVideosIE,
                    TwitchVideosClipsIE,
                    TwitchVideosCollectionsIE,
                    TwitchClipsIE))
                else super(TwitchStreamIE, cls).suitable(url))

    def _real_extract(self, url):
        channel_name = self._match_id(url).lower()

        gql = self._download_gql(
            channel_name, [{
                'operationName': 'StreamMetadata',
                'variables': {'channelLogin': channel_name},
            }, {
                'operationName': 'ComscoreStreamingQuery',
                'variables': {
                    'channel': channel_name,
                    'clipSlug': '',
                    'isClip': False,
                    'isLive': True,
                    'isVodOrCollection': False,
                    'vodID': '',
                },
            }, {
                'operationName': 'VideoPreviewOverlay',
                'variables': {'login': channel_name},
            }],
            'Downloading stream GraphQL')

        user = gql[0]['data']['user']

        if not user:
            raise ExtractorError(
                '%s does not exist' % channel_name, expected=True)

        stream = user['stream']

        if not stream:
            raise ExtractorError('%s is offline' % channel_name, expected=True)

        access_token = self._download_access_token_gql(
            channel_name, 'stream', 'channelName')
        token = access_token['value']

        stream_id = stream.get('id') or channel_name
        query = {
            'allow_source': 'true',
            'allow_audio_only': 'true',
            'allow_spectre': 'true',
            'p': random.randint(1000000, 10000000),
            'player': 'twitchweb',
            'playlist_include_framerate': 'true',
            'segment_preference': '4',
            'sig': access_token['signature'].encode('utf-8'),
            'token': token.encode('utf-8'),
        }
        formats = self._extract_m3u8_formats(
            '%s/api/channel/hls/%s.m3u8' % (self._USHER_BASE, channel_name),
            stream_id, 'mp4', query=query)
        self._prefer_source(formats)

        view_count = stream.get('viewers')
        timestamp = unified_timestamp(stream.get('createdAt'))

        sq_user = try_get(gql, lambda x: x[1]['data']['user'], dict) or {}
        uploader = sq_user.get('displayName')
        description = try_get(
            sq_user, lambda x: x['broadcastSettings']['title'], compat_str)

        thumbnail = url_or_none(try_get(
            gql, lambda x: x[2]['data']['user']['stream']['previewImageURL'],
            compat_str))

        title = uploader or channel_name
        stream_type = stream.get('type')
        if stream_type in ['rerun', 'live']:
            title += ' (%s)' % stream_type

        return {
            'id': stream_id,
            'display_id': channel_name,
            'title': self._live_title(title),
            'description': description,
            'thumbnail': thumbnail,
            'uploader': uploader,
            'uploader_id': channel_name,
            'timestamp': timestamp,
            'view_count': view_count,
            'formats': formats,
            'is_live': stream_type == 'live',
        }


class TwitchClipsIE(TwitchGraphQLBaseIE):
    IE_NAME = 'twitch:clips'
    _VALID_URL = r'''(?x)
                    https?://
                        (?:
                            clips\.twitch\.tv/(?:embed\?.*?\bclip=|(?:[^/]+/)*)|
                            (?:(?:www|go|m)\.)?twitch\.tv/[^/]+/clip/
                        )
                        (?P<id>[^/?#&]+)
                    '''

    _TESTS = [{
        'url': 'https://clips.twitch.tv/FaintLightGullWholeWheat',
        'md5': '761769e1eafce0ffebfb4089cb3847cd',
        'info_dict': {
            'id': '42850523',
            'ext': 'mp4',
            'title': 'EA Play 2016 Live from the Novo Theatre',
            'thumbnail': r're:^https?://.*\.jpg',
            'timestamp': 1465767393,
            'upload_date': '20160612',
            'creator': 'EA',
            'uploader': 'stereotype_',
            'uploader_id': '43566419',
        },
    }, {
        # multiple formats
        'url': 'https://clips.twitch.tv/rflegendary/UninterestedBeeDAESuppy',
        'only_matching': True,
    }, {
        'url': 'https://www.twitch.tv/sergeynixon/clip/StormyThankfulSproutFutureMan',
        'only_matching': True,
    }, {
        'url': 'https://clips.twitch.tv/embed?clip=InquisitiveBreakableYogurtJebaited',
        'only_matching': True,
    }, {
        'url': 'https://m.twitch.tv/rossbroadcast/clip/ConfidentBraveHumanChefFrank',
        'only_matching': True,
    }, {
        'url': 'https://go.twitch.tv/rossbroadcast/clip/ConfidentBraveHumanChefFrank',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        video_id = self._match_id(url)

        clip = self._download_base_gql(
            video_id, {
                'query': '''{
  clip(slug: "%s") {
    broadcaster {
      displayName
    }
    createdAt
    curator {
      displayName
      id
    }
    durationSeconds
    id
    tiny: thumbnailURL(width: 86, height: 45)
    small: thumbnailURL(width: 260, height: 147)
    medium: thumbnailURL(width: 480, height: 272)
    title
    videoQualities {
      frameRate
      quality
      sourceURL
    }
    viewCount
  }
}''' % video_id}, 'Downloading clip GraphQL')['data']['clip']

        if not clip:
            raise ExtractorError(
                'This clip is no longer available', expected=True)

        formats = []
        for option in clip.get('videoQualities', []):
            if not isinstance(option, dict):
                continue
            source = url_or_none(option.get('sourceURL'))
            if not source:
                continue
            formats.append({
                'url': source,
                'format_id': option.get('quality'),
                'height': int_or_none(option.get('quality')),
                'fps': int_or_none(option.get('frameRate')),
            })
        self._sort_formats(formats)

        thumbnails = []
        for thumbnail_id in ('tiny', 'small', 'medium'):
            thumbnail_url = clip.get(thumbnail_id)
            if not thumbnail_url:
                continue
            thumb = {
                'id': thumbnail_id,
                'url': thumbnail_url,
            }
            mobj = re.search(r'-(\d+)x(\d+)\.', thumbnail_url)
            if mobj:
                thumb.update({
                    'height': int(mobj.group(2)),
                    'width': int(mobj.group(1)),
                })
            thumbnails.append(thumb)

        return {
            'id': clip.get('id') or video_id,
            'title': clip.get('title') or video_id,
            'formats': formats,
            'duration': int_or_none(clip.get('durationSeconds')),
            'views': int_or_none(clip.get('viewCount')),
            'timestamp': unified_timestamp(clip.get('createdAt')),
            'thumbnails': thumbnails,
            'creator': try_get(clip, lambda x: x['broadcaster']['displayName'], compat_str),
            'uploader': try_get(clip, lambda x: x['curator']['displayName'], compat_str),
            'uploader_id': try_get(clip, lambda x: x['curator']['id'], compat_str),
        }