X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Ftwitch.py;h=f9164af098d8a7529f290e2297a4974fe2004af5;hb=f0c6c2bce29281d37d2bbd589143b35323e38e3d;hp=6d67bda86d7bc61e92fe11ce5b67a5e6d9327422;hpb=ec85ded83cbfa652ba94cb080aab52d8b270212a;p=youtube-dl diff --git a/youtube_dl/extractor/twitch.py b/youtube_dl/extractor/twitch.py index 6d67bda86..f9164af09 100644 --- a/youtube_dl/extractor/twitch.py +++ b/youtube_dl/extractor/twitch.py @@ -12,7 +12,6 @@ from ..compat import ( compat_str, compat_urllib_parse_urlencode, compat_urllib_parse_urlparse, - compat_urlparse, ) from ..utils import ( clean_html, @@ -24,15 +23,16 @@ from ..utils import ( parse_iso8601, update_url_query, urlencode_postdata, + urljoin, ) class TwitchBaseIE(InfoExtractor): - _VALID_URL_BASE = r'https?://(?:www\.)?twitch\.tv' + _VALID_URL_BASE = r'https?://(?:(?:www|go)\.)?twitch\.tv' _API_BASE = 'https://api.twitch.tv' _USHER_BASE = 'https://usher.ttvnw.net' - _LOGIN_URL = 'http://www.twitch.tv/login' + _LOGIN_URL = 'https://www.twitch.tv/login' _CLIENT_ID = 'jzkbprff40iqj646a697cyrvl0zt2m6' _NETRC_MACHINE = 'twitch' @@ -64,6 +64,35 @@ class TwitchBaseIE(InfoExtractor): 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']+action=(["\'])(?P.+?)\1', page, + 'post url', default=page_url, group='url') + post_url = urljoin(page_url, post_url) + + headers = {'Referer': page_url} + + try: + response = self._download_json( + post_url, None, note, + data=urlencode_postdata(form), + headers=headers) + except ExtractorError as e: + if isinstance(e.cause, compat_HTTPError) and e.cause.code == 400: + response = self._parse_json( + e.cause.read().decode('utf-8'), None) + fail(response['message']) + raise + + redirect_url = urljoin(post_url, response['redirect']) + return self._download_webpage_handle( + redirect_url, None, 'Downloading login redirect page', + headers=headers) + login_page, handle = self._download_webpage_handle( self._LOGIN_URL, None, 'Downloading login page') @@ -71,40 +100,19 @@ class TwitchBaseIE(InfoExtractor): if 'blacklist_message' in login_page: fail(clean_html(login_page)) - login_form = self._hidden_inputs(login_page) - - login_form.update({ - 'username': username, - 'password': password, - }) - - redirect_url = handle.geturl() - - post_url = self._search_regex( - r']+action=(["\'])(?P.+?)\1', login_page, - 'post url', default=redirect_url, group='url') - - if not post_url.startswith('http'): - post_url = compat_urlparse.urljoin(redirect_url, post_url) - - headers = {'Referer': redirect_url} + redirect_page, handle = login_step( + login_page, handle, 'Logging in', { + 'username': username, + 'password': password, + }) - try: - response = self._download_json( - post_url, None, 'Logging in as %s' % username, - data=urlencode_postdata(login_form), - headers=headers) - except ExtractorError as e: - if isinstance(e.cause, compat_HTTPError) and e.cause.code == 400: - response = self._parse_json( - e.cause.read().decode('utf-8'), None) - fail(response['message']) - raise - - if response.get('redirect'): - self._download_webpage( - response['redirect'], None, 'Downloading login redirect page', - headers=headers) + if re.search(r'(?i)]+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: @@ -209,7 +217,7 @@ class TwitchVodIE(TwitchItemBaseIE): _VALID_URL = r'''(?x) https?:// (?: - (?:www\.)?twitch\.tv/[^/]+/v/| + (?:(?:www|go)\.)?twitch\.tv/(?:[^/]+/v|videos)/| player\.twitch\.tv/\?.*?\bvideo=v ) (?P\d+) @@ -259,6 +267,9 @@ class TwitchVodIE(TwitchItemBaseIE): }, { 'url': 'http://player.twitch.tv/?t=5m10s&video=v6528877', 'only_matching': True, + }, { + 'url': 'https://www.twitch.tv/videos/6528877', + 'only_matching': True, }] def _real_extract(self, url): @@ -347,9 +358,16 @@ class TwitchPlaylistBaseIE(TwitchBaseIE): break offset += limit return self.playlist_result( - [self.url_result(entry) for entry in orderedSet(entries)], + [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 [] @@ -444,7 +462,14 @@ class TwitchHighlightsIE(TwitchVideosBaseIE): class TwitchStreamIE(TwitchBaseIE): IE_NAME = 'twitch:stream' - _VALID_URL = r'%s/(?P[^/#?]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE + _VALID_URL = r'''(?x) + https?:// + (?: + (?:(?:www|go)\.)?twitch\.tv/| + player\.twitch\.tv/\?.*?\bchannel= + ) + (?P[^/#?]+) + ''' _TESTS = [{ 'url': 'http://www.twitch.tv/shroomztv', @@ -468,8 +493,28 @@ class TwitchStreamIE(TwitchBaseIE): }, { '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, }] + @classmethod + def suitable(cls, url): + return (False + if any(ie.suitable(url) for ie in ( + TwitchVideoIE, + TwitchChapterIE, + TwitchVodIE, + TwitchProfileIE, + TwitchAllVideosIE, + TwitchUploadsIE, + TwitchPastBroadcastsIE, + TwitchHighlightsIE)) + else super(TwitchStreamIE, cls).suitable(url)) + def _real_extract(self, url): channel_id = self._match_id(url) @@ -571,7 +616,7 @@ class TwitchClipsIE(InfoExtractor): r'(?s)clipInfo\s*=\s*({.+?});', webpage, 'clip info'), video_id, transform_source=js_to_json) - title = clip.get('channel_title') or self._og_search_title(webpage) + title = clip.get('title') or clip.get('channel_title') or self._og_search_title(webpage) formats = [{ 'url': option['source'],