]> Raphaël G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/twitch.py
Update upstream source from tag 'upstream/2020.07.28'
[youtubedl] / youtube_dl / extractor / twitch.py
index b11a1d5610d0dffe0d98df7d7b05d4228552dfb7..e211cd4c84cb4c713e937cf0e1eef83de7c40a93 100644 (file)
@@ -3,56 +3,43 @@ from __future__ import unicode_literals
 
 import itertools
 import re
 
 import itertools
 import re
+import random
+import json
 
 from .common import InfoExtractor
 from ..compat import (
 
 from .common import InfoExtractor
 from ..compat import (
-    compat_urllib_parse,
-    compat_urllib_request,
+    compat_kwargs,
+    compat_parse_qs,
+    compat_str,
+    compat_urllib_parse_urlencode,
+    compat_urllib_parse_urlparse,
 )
 from ..utils import (
 )
 from ..utils import (
+    clean_html,
     ExtractorError,
     ExtractorError,
+    int_or_none,
+    orderedSet,
+    parse_duration,
     parse_iso8601,
     parse_iso8601,
+    qualities,
+    str_or_none,
+    try_get,
+    unified_timestamp,
+    update_url_query,
+    url_or_none,
+    urljoin,
 )
 
 
 )
 
 
-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>[^/]+)|
-            (?:(?:[^/]+)/v/(?P<vodid>[^/]+))|
-            (?:(?:[^/]+)/b/(?P<videoid>[^/]+))|
-            (?:(?:[^/]+)/c/(?P<chapterid>[^/]+))
-        )
-        /?(?:\#.*)?$
-        """
-    _PAGE_LIMIT = 100
+class TwitchBaseIE(InfoExtractor):
+    _VALID_URL_BASE = r'https?://(?:(?:www|go|m)\.)?twitch\.tv'
+
     _API_BASE = 'https://api.twitch.tv'
     _API_BASE = 'https://api.twitch.tv'
-    _LOGIN_URL = 'https://secure.twitch.tv/user/login'
-    _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,
-    }]
+    _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):
 
     def _handle_error(self, response):
         if not isinstance(response, dict):
@@ -63,35 +50,118 @@ class TwitchIE(InfoExtractor):
                 '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
                 expected=True)
 
                 '%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)
+    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
 
         self._handle_error(response)
         return response
 
-    def _extract_media(self, item, item_id):
-        ITEMS = {
-            'a': 'video',
-            'v': 'vod',
-            '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]))
-
-        if item == 'v':
-            access_token = self._download_json(
-                '%s/api/vods/%s/access_token' % (self._API_BASE, item_id), item_id,
-                'Downloading %s access token' % ITEMS[item])
-            formats = self._extract_m3u8_formats(
-                'http://usher.twitch.tv/vod/%s?nauth=%s&nauthsig=%s'
-                % (item_id, access_token['token'], access_token['sig']),
-                item_id, 'mp4')
-            info['formats'] = formats
-            return info
+    def _real_initialize(self):
+        self._login()
 
 
-        response = self._download_json(
-            '%s/api/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
-            'Downloading %s playlist JSON' % ITEMS[item])
+    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)
+
+
+class TwitchItemBaseIE(TwitchBaseIE):
+    def _download_info(self, item, item_id):
+        return self._extract_info(self._call_api(
+            'kraken/videos/%s%s' % (item, item_id), item_id,
+            'Downloading %s info JSON' % self._ITEM_TYPE))
+
+    def _extract_media(self, item_id):
+        info = self._download_info(self._ITEM_SHORTCUT, item_id)
+        response = self._call_api(
+            'api/videos/%s%s' % (self._ITEM_SHORTCUT, item_id), item_id,
+            'Downloading %s playlist JSON' % self._ITEM_TYPE)
         entries = []
         chunks = response['chunks']
         qualities = list(chunks.keys())
         entries = []
         chunks = response['chunks']
         qualities = list(chunks.keys())
@@ -117,131 +187,610 @@ class TwitchIE(InfoExtractor):
         return self.playlist_result(entries, info['id'], info['title'])
 
     def _extract_info(self, info):
         return self.playlist_result(entries, info['id'], info['title'])
 
     def _extract_info(self, 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'],
         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'],
+            '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,
         }
 
         }
 
-    def _real_initialize(self):
-        self._login()
+    def _real_extract(self, url):
+        return self._extract_media(self._match_id(url))
 
 
-    def _login(self):
-        (username, password) = self._get_login_info()
-        if username is None:
-            return
 
 
-        login_page = self._download_webpage(
-            self._LOGIN_URL, None, 'Downloading login page')
-
-        authenticity_token = self._search_regex(
-            r'<input name="authenticity_token" type="hidden" value="([^"]+)"',
-            login_page, 'authenticity token')
-
-        login_form = {
-            'utf8': '✓'.encode('utf-8'),
-            'authenticity_token': authenticity_token,
-            'redirect_on_login': '',
-            'embed_form': 'false',
-            'mp_source_action': '',
-            'follow': '',
-            'user[login]': username,
-            'user[password]': password,
-        }
+class TwitchVideoIE(TwitchItemBaseIE):
+    IE_NAME = 'twitch:video'
+    _VALID_URL = r'%s/[^/]+/b/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
+    _ITEM_TYPE = 'video'
+    _ITEM_SHORTCUT = 'a'
 
 
-        request = compat_urllib_request.Request(
-            self._LOGIN_URL, compat_urllib_parse.urlencode(login_form).encode('utf-8'))
-        request.add_header('Referer', self._LOGIN_URL)
-        response = self._download_webpage(
-            request, None, 'Logging in as %s' % username)
+    _TEST = {
+        'url': 'http://www.twitch.tv/riotgames/b/577357806',
+        'info_dict': {
+            'id': 'a577357806',
+            'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
+        },
+        'playlist_mincount': 12,
+        'skip': 'HTTP Error 404: Not Found',
+    }
 
 
-        m = re.search(
-            r"id=([\"'])login_error_message\1[^>]*>(?P<msg>[^<]+)", response)
-        if m:
-            raise ExtractorError(
-                'Unable to login: %s' % m.group('msg').strip(), expected=True)
+
+class TwitchChapterIE(TwitchItemBaseIE):
+    IE_NAME = 'twitch:chapter'
+    _VALID_URL = r'%s/[^/]+/c/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
+    _ITEM_TYPE = 'chapter'
+    _ITEM_SHORTCUT = 'c'
+
+    _TESTS = [{
+        'url': 'http://www.twitch.tv/acracingleague/c/5285812',
+        'info_dict': {
+            'id': 'c5285812',
+            'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
+        },
+        'playlist_mincount': 3,
+        'skip': 'HTTP Error 404: Not Found',
+    }, {
+        'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361',
+        'only_matching': True,
+    }]
+
+
+class TwitchVodIE(TwitchItemBaseIE):
+    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+)
+                    '''
+    _ITEM_TYPE = 'vod'
+    _ITEM_SHORTCUT = 'v'
+
+    _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$',
+            'duration': 17208,
+            'timestamp': 1435131709,
+            'upload_date': '20150624',
+            'uploader': 'Riot Games',
+            'uploader_id': 'riotgames',
+            'view_count': int,
+            'start_time': 310,
+        },
+        'params': {
+            # m3u8 download
+            'skip_download': True,
+        },
+    }, {
+        # Untitled broadcast (title is None)
+        'url': 'http://www.twitch.tv/belkao_o/v/11230755',
+        'info_dict': {
+            'id': 'v11230755',
+            'ext': 'mp4',
+            'title': 'Untitled Broadcast',
+            'thumbnail': r're:^https?://.*\.jpg$',
+            '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 _real_extract(self, url):
+        item_id = self._match_id(url)
+
+        info = self._download_info(self._ITEM_SHORTCUT, item_id)
+        access_token = self._call_api(
+            'api/vods/%s/access_token' % item_id, item_id,
+            'Downloading %s access token' % self._ITEM_TYPE)
+
+        formats = self._extract_m3u8_formats(
+            '%s/vod/%s.m3u8?%s' % (
+                self._USHER_BASE, item_id,
+                compat_urllib_parse_urlencode({
+                    'allow_source': 'true',
+                    'allow_audio_only': 'true',
+                    'allow_spectre': 'true',
+                    'player': 'twitchweb',
+                    'playlist_include_framerate': 'true',
+                    'nauth': access_token['token'],
+                    'nauthsig': access_token['sig'],
+                })),
+            item_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' % item_id, {
+                            'client_id': self._CLIENT_ID,
+                        }),
+                    'ext': 'json',
+                }],
+            }
+
+        return info
+
+
+class TwitchPlaylistBaseIE(TwitchBaseIE):
+    _PLAYLIST_PATH = 'kraken/channels/%s/videos/?offset=%d&limit=%d'
+    _PAGE_LIMIT = 100
+
+    def _extract_playlist(self, channel_id):
+        info = self._call_api(
+            'kraken/channels/%s' % 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
+        broken_paging_detected = False
+        counter_override = None
+        for counter in itertools.count(1):
+            response = self._call_api(
+                self._PLAYLIST_PATH % (channel_id, offset, limit),
+                channel_id,
+                'Downloading %s JSON page %s'
+                % (self._PLAYLIST_TYPE, counter_override or counter))
+            page_entries = self._extract_playlist_page(response)
+            if not page_entries:
+                break
+            total = int_or_none(response.get('_total'))
+            # Since the beginning of March 2016 twitch's paging mechanism
+            # is completely broken on the twitch side. It simply ignores
+            # a limit and returns the whole offset number of videos.
+            # Working around by just requesting all videos at once.
+            # Upd: pagination bug was fixed by twitch on 15.03.2016.
+            if not broken_paging_detected and total and len(page_entries) > limit:
+                self.report_warning(
+                    'Twitch pagination is broken on twitch side, requesting all videos at once',
+                    channel_id)
+                broken_paging_detected = True
+                offset = total
+                counter_override = '(all at once)'
+                continue
+            entries.extend(page_entries)
+            if broken_paging_detected or total and len(page_entries) >= total:
+                break
+            offset += limit
+        return self.playlist_result(
+            [self._make_url_result(entry) for entry in orderedSet(entries)],
+            channel_id, channel_name)
+
+    def _make_url_result(self, url):
+        try:
+            video_id = 'v%s' % TwitchVodIE._match_id(url)
+            return self.url_result(url, TwitchVodIE.ie_key(), video_id=video_id)
+        except AssertionError:
+            return self.url_result(url)
+
+    def _extract_playlist_page(self, response):
+        videos = response.get('videos')
+        return [video['url'] for video in videos] if videos else []
 
     def _real_extract(self, url):
 
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        if mobj.group('chapterid'):
-            return self._extract_media('c', mobj.group('chapterid'))
+        return self._extract_playlist(self._match_id(url))
+
+
+class TwitchProfileIE(TwitchPlaylistBaseIE):
+    IE_NAME = 'twitch:profile'
+    _VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
+    _PLAYLIST_TYPE = 'profile'
+
+    _TESTS = [{
+        'url': 'http://www.twitch.tv/vanillatv/profile',
+        'info_dict': {
+            'id': 'vanillatv',
+            'title': 'VanillaTV',
+        },
+        'playlist_mincount': 412,
+    }, {
+        'url': 'http://m.twitch.tv/vanillatv/profile',
+        'only_matching': True,
+    }]
+
+
+class TwitchVideosBaseIE(TwitchPlaylistBaseIE):
+    _VALID_URL_VIDEOS_BASE = r'%s/(?P<id>[^/]+)/videos' % TwitchBaseIE._VALID_URL_BASE
+    _PLAYLIST_PATH = TwitchPlaylistBaseIE._PLAYLIST_PATH + '&broadcast_type='
 
 
-            """
-            webpage = self._download_webpage(url, chapter_id)
-            m = re.search(r'PP\.archive_id = "([0-9]+)";', webpage)
+
+class TwitchAllVideosIE(TwitchVideosBaseIE):
+    IE_NAME = 'twitch:videos:all'
+    _VALID_URL = r'%s/all' % TwitchVideosBaseIE._VALID_URL_VIDEOS_BASE
+    _PLAYLIST_PATH = TwitchVideosBaseIE._PLAYLIST_PATH + 'archive,upload,highlight'
+    _PLAYLIST_TYPE = 'all videos'
+
+    _TESTS = [{
+        'url': 'https://www.twitch.tv/spamfish/videos/all',
+        'info_dict': {
+            'id': 'spamfish',
+            'title': 'Spamfish',
+        },
+        'playlist_mincount': 869,
+    }, {
+        'url': 'https://m.twitch.tv/spamfish/videos/all',
+        'only_matching': True,
+    }]
+
+
+class TwitchUploadsIE(TwitchVideosBaseIE):
+    IE_NAME = 'twitch:videos:uploads'
+    _VALID_URL = r'%s/uploads' % TwitchVideosBaseIE._VALID_URL_VIDEOS_BASE
+    _PLAYLIST_PATH = TwitchVideosBaseIE._PLAYLIST_PATH + 'upload'
+    _PLAYLIST_TYPE = 'uploads'
+
+    _TESTS = [{
+        'url': 'https://www.twitch.tv/spamfish/videos/uploads',
+        'info_dict': {
+            'id': 'spamfish',
+            'title': 'Spamfish',
+        },
+        'playlist_mincount': 0,
+    }, {
+        'url': 'https://m.twitch.tv/spamfish/videos/uploads',
+        'only_matching': True,
+    }]
+
+
+class TwitchPastBroadcastsIE(TwitchVideosBaseIE):
+    IE_NAME = 'twitch:videos:past-broadcasts'
+    _VALID_URL = r'%s/past-broadcasts' % TwitchVideosBaseIE._VALID_URL_VIDEOS_BASE
+    _PLAYLIST_PATH = TwitchVideosBaseIE._PLAYLIST_PATH + 'archive'
+    _PLAYLIST_TYPE = 'past broadcasts'
+
+    _TESTS = [{
+        'url': 'https://www.twitch.tv/spamfish/videos/past-broadcasts',
+        'info_dict': {
+            'id': 'spamfish',
+            'title': 'Spamfish',
+        },
+        'playlist_mincount': 0,
+    }, {
+        'url': 'https://m.twitch.tv/spamfish/videos/past-broadcasts',
+        'only_matching': True,
+    }]
+
+
+class TwitchHighlightsIE(TwitchVideosBaseIE):
+    IE_NAME = 'twitch:videos:highlights'
+    _VALID_URL = r'%s/highlights' % TwitchVideosBaseIE._VALID_URL_VIDEOS_BASE
+    _PLAYLIST_PATH = TwitchVideosBaseIE._PLAYLIST_PATH + 'highlight'
+    _PLAYLIST_TYPE = 'highlights'
+
+    _TESTS = [{
+        'url': 'https://www.twitch.tv/spamfish/videos/highlights',
+        'info_dict': {
+            'id': 'spamfish',
+            'title': 'Spamfish',
+        },
+        'playlist_mincount': 805,
+    }, {
+        'url': 'https://m.twitch.tv/spamfish/videos/highlights',
+        'only_matching': True,
+    }]
+
+
+class TwitchStreamIE(TwitchBaseIE):
+    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 (
+                    TwitchVideoIE,
+                    TwitchChapterIE,
+                    TwitchVodIE,
+                    TwitchProfileIE,
+                    TwitchAllVideosIE,
+                    TwitchUploadsIE,
+                    TwitchPastBroadcastsIE,
+                    TwitchHighlightsIE,
+                    TwitchClipsIE))
+                else super(TwitchStreamIE, cls).suitable(url))
+
+    def _real_extract(self, url):
+        channel_name = self._match_id(url)
+
+        access_token = self._call_api(
+            'api/channels/%s/access_token' % channel_name, channel_name,
+            'Downloading access token JSON')
+
+        token = access_token['token']
+        channel_id = compat_str(self._parse_json(
+            token, channel_name)['channel_id'])
+
+        stream = self._call_api(
+            'kraken/streams/%s?stream_type=all' % channel_id,
+            channel_id, 'Downloading stream JSON').get('stream')
+
+        if not stream:
+            raise ExtractorError('%s is offline' % channel_id, expected=True)
+
+        # Channel name may be typed if different case than the original channel name
+        # (e.g. http://www.twitch.tv/TWITCHPLAYSPOKEMON) that will lead to constructing
+        # an invalid m3u8 URL. Working around by use of original channel name from stream
+        # JSON and fallback to lowercase if it's not available.
+        channel_name = try_get(
+            stream, lambda x: x['channel']['name'],
+            compat_str) or channel_name.lower()
+
+        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['sig'].encode('utf-8'),
+            'token': token.encode('utf-8'),
+        }
+        formats = self._extract_m3u8_formats(
+            '%s/api/channel/hls/%s.m3u8?%s'
+            % (self._USHER_BASE, channel_name, compat_urllib_parse_urlencode(query)),
+            channel_id, 'mp4')
+        self._prefer_source(formats)
+
+        view_count = stream.get('viewers')
+        timestamp = parse_iso8601(stream.get('created_at'))
+
+        channel = stream['channel']
+        title = self._live_title(channel.get('display_name') or channel.get('name'))
+        description = channel.get('status')
+
+        thumbnails = []
+        for thumbnail_key, thumbnail_url in stream['preview'].items():
+            m = re.search(r'(?P<width>\d+)x(?P<height>\d+)\.jpg$', thumbnail_key)
             if not m:
             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'],
+                continue
+            thumbnails.append({
+                'url': thumbnail_url,
+                'width': int(m.group('width')),
+                'height': int(m.group('height')),
+            })
+
+        return {
+            'id': str_or_none(stream.get('_id')) or channel_id,
+            'display_id': channel_name,
+            'title': title,
+            'description': description,
+            'thumbnails': thumbnails,
+            'uploader': channel.get('display_name'),
+            'uploader_id': channel.get('name'),
+            'timestamp': timestamp,
+            'view_count': view_count,
+            'formats': formats,
+            'is_live': True,
+        }
+
+
+class TwitchClipsIE(TwitchBaseIE):
+    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_json(
+            'https://gql.twitch.tv/gql', video_id, data=json.dumps({
+                '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,
+            }).encode(), headers={
+                'Client-ID': self._CLIENT_ID,
+            })['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,
             }
             }
-            return info
-            """
-        elif mobj.group('videoid'):
-            return self._extract_media('a', mobj.group('videoid'))
-        elif mobj.group('vodid'):
-            return self._extract_media('v', mobj.group('vodid'))
-        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)
+            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),
+        }