X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Ftwitter.py;h=89eabe77e0c4cd107aa88ea3ed71b4cd8c2b43be;hb=9be31e771cd9481ea690c01eed398645deadc1de;hp=602538e5cc09ab89214caff794358a5702e45be0;hpb=e68d3a010fcf34455c7922b28a05ccc012381729;p=youtube-dl.git diff --git a/youtube_dl/extractor/twitter.py b/youtube_dl/extractor/twitter.py index 602538e5c..89eabe77e 100644 --- a/youtube_dl/extractor/twitter.py +++ b/youtube_dl/extractor/twitter.py @@ -4,24 +4,46 @@ from __future__ import unicode_literals import re from .common import InfoExtractor +from ..compat import compat_urlparse from ..utils import ( + determine_ext, + dict_get, + ExtractorError, float_or_none, - xpath_text, - remove_end, int_or_none, - ExtractorError, + remove_end, + try_get, + xpath_text, ) +from .periscope import PeriscopeIE + class TwitterBaseIE(InfoExtractor): - def _get_vmap_video_url(self, vmap_url, video_id): + def _extract_formats_from_vmap_url(self, vmap_url, video_id): vmap_data = self._download_xml(vmap_url, video_id) - return xpath_text(vmap_data, './/MediaFile').strip() + video_url = xpath_text(vmap_data, './/MediaFile').strip() + if determine_ext(video_url) == 'm3u8': + return self._extract_m3u8_formats( + video_url, video_id, ext='mp4', m3u8_id='hls', + entry_protocol='m3u8_native') + return [{ + 'url': video_url, + }] + + @staticmethod + def _search_dimensions_in_video_url(a_format, video_url): + m = re.search(r'/(?P\d+)x(?P\d+)/', video_url) + if m: + a_format.update({ + 'width': int(m.group('width')), + 'height': int(m.group('height')), + }) class TwitterCardIE(TwitterBaseIE): IE_NAME = 'twitter:card' - _VALID_URL = r'https?://(?:www\.)?twitter\.com/i/(?:cards/tfw/v1|videos/tweet)/(?P\d+)' + _VALID_URL = r'https?://(?:www\.)?twitter\.com/i/(?:cards/tfw/v1|videos(?:/tweet)?)/(?P\d+)' _TESTS = [ { 'url': 'https://twitter.com/i/cards/tfw/v1/560070183650213889', @@ -30,7 +52,7 @@ class TwitterCardIE(TwitterBaseIE): 'id': '560070183650213889', 'ext': 'mp4', 'title': 'Twitter Card', - 'thumbnail': 're:^https?://.*\.jpg$', + 'thumbnail': r're:^https?://.*\.jpg$', 'duration': 30.033, } }, @@ -41,18 +63,18 @@ class TwitterCardIE(TwitterBaseIE): 'id': '623160978427936768', 'ext': 'mp4', 'title': 'Twitter Card', - 'thumbnail': 're:^https?://.*\.jpg', + 'thumbnail': r're:^https?://.*\.jpg', 'duration': 80.155, }, }, { 'url': 'https://twitter.com/i/cards/tfw/v1/654001591733886977', - 'md5': 'd4724ffe6d2437886d004fa5de1043b3', + 'md5': 'b6d9683dd3f48e340ded81c0e917ad46', 'info_dict': { 'id': 'dq4Oj5quskI', 'ext': 'mp4', 'title': 'Ubuntu 11.10 Overview', - 'description': 'Take a quick peek at what\'s new and improved in Ubuntu 11.10.\n\nOnce installed take a look at 10 Things to Do After Installing: http://www.omgubuntu.co.uk/2011/10/10-things-to-do-after-installing-ubuntu-11-10/', + 'description': 'md5:a831e97fa384863d6e26ce48d1c43376', 'upload_date': '20111013', 'uploader': 'OMG! Ubuntu!', 'uploader_id': 'omgubuntu', @@ -78,11 +100,67 @@ class TwitterCardIE(TwitterBaseIE): 'id': '705235433198714880', 'ext': 'mp4', 'title': 'Twitter web player', - 'thumbnail': 're:^https?://.*\.jpg', + 'thumbnail': r're:^https?://.*\.jpg', }, + }, { + 'url': 'https://twitter.com/i/videos/752274308186120192', + 'only_matching': True, }, ] + def _parse_media_info(self, media_info, video_id): + formats = [] + for media_variant in media_info.get('variants', []): + media_url = media_variant['url'] + if media_url.endswith('.m3u8'): + formats.extend(self._extract_m3u8_formats(media_url, video_id, ext='mp4', m3u8_id='hls')) + elif media_url.endswith('.mpd'): + formats.extend(self._extract_mpd_formats(media_url, video_id, mpd_id='dash')) + else: + vbr = int_or_none(dict_get(media_variant, ('bitRate', 'bitrate')), scale=1000) + a_format = { + 'url': media_url, + 'format_id': 'http-%d' % vbr if vbr else 'http', + 'vbr': vbr, + } + # Reported bitRate may be zero + if not a_format['vbr']: + del a_format['vbr'] + + self._search_dimensions_in_video_url(a_format, media_url) + + formats.append(a_format) + return formats + + def _extract_mobile_formats(self, username, video_id): + webpage = self._download_webpage( + 'https://mobile.twitter.com/%s/status/%s' % (username, video_id), + video_id, 'Downloading mobile webpage', + headers={ + # A recent mobile UA is necessary for `gt` cookie + 'User-Agent': 'Mozilla/5.0 (Android 6.0.1; Mobile; rv:54.0) Gecko/54.0 Firefox/54.0', + }) + main_script_url = self._html_search_regex( + r']+src="([^"]+main\.[^"]+)"', webpage, 'main script URL') + main_script = self._download_webpage( + main_script_url, video_id, 'Downloading main script') + bearer_token = self._search_regex( + r'BEARER_TOKEN\s*:\s*"([^"]+)"', + main_script, 'bearer token') + guest_token = self._search_regex( + r'document\.cookie\s*=\s*decodeURIComponent\("gt=(\d+)', + webpage, 'guest token') + api_data = self._download_json( + 'https://api.twitter.com/2/timeline/conversation/%s.json' % video_id, + video_id, 'Downloading mobile API data', + headers={ + 'Authorization': 'Bearer ' + bearer_token, + 'x-guest-token': guest_token, + }) + media_info = try_get(api_data, lambda o: o['globalObjects']['tweets'][video_id] + ['extended_entities']['media'][0]['video_info']) or {} + return self._parse_media_info(media_info, video_id) + def _real_extract(self, url): video_id = self._match_id(url) @@ -99,33 +177,35 @@ class TwitterCardIE(TwitterBaseIE): return self.url_result(iframe_url) config = self._parse_json(self._html_search_regex( - r'data-(?:player-)?config="([^"]+)"', webpage, 'data player config'), + r'data-(?:player-)?config="([^"]+)"', webpage, + 'data player config', default='{}'), video_id) - def _search_dimensions_in_video_url(a_format, video_url): - m = re.search(r'/(?P\d+)x(?P\d+)/', video_url) - if m: - a_format.update({ - 'width': int(m.group('width')), - 'height': int(m.group('height')), - }) + if config.get('source_type') == 'vine': + return self.url_result(config['player_url'], 'Vine') + + periscope_url = PeriscopeIE._extract_url(webpage) + if periscope_url: + return self.url_result(periscope_url, PeriscopeIE.ie_key()) video_url = config.get('video_url') or config.get('playlist', [{}])[0].get('source') if video_url: - f = { - 'url': video_url, - } + if determine_ext(video_url) == 'm3u8': + formats.extend(self._extract_m3u8_formats(video_url, video_id, ext='mp4', m3u8_id='hls')) + else: + f = { + 'url': video_url, + } - _search_dimensions_in_video_url(f, video_url) + self._search_dimensions_in_video_url(f, video_url) - formats.append(f) + formats.append(f) vmap_url = config.get('vmapUrl') or config.get('vmap_url') if vmap_url: - formats.append({ - 'url': self._get_vmap_video_url(vmap_url, video_id), - }) + formats.extend( + self._extract_formats_from_vmap_url(vmap_url, video_id)) media_info = None @@ -134,29 +214,14 @@ class TwitterCardIE(TwitterBaseIE): media_info = entity['mediaInfo'] if media_info: - for media_variant in media_info['variants']: - media_url = media_variant['url'] - if media_url.endswith('.m3u8'): - formats.extend(self._extract_m3u8_formats(media_url, video_id, ext='mp4', m3u8_id='hls')) - elif media_url.endswith('.mpd'): - formats.extend(self._extract_mpd_formats(media_url, video_id, mpd_id='dash')) - else: - vbr = int_or_none(media_variant.get('bitRate'), scale=1000) - a_format = { - 'url': media_url, - 'format_id': 'http-%d' % vbr if vbr else 'http', - 'vbr': vbr, - } - # Reported bitRate may be zero - if not a_format['vbr']: - del a_format['vbr'] - - _search_dimensions_in_video_url(a_format, media_url) - - formats.append(a_format) - + formats.extend(self._parse_media_info(media_info, video_id)) duration = float_or_none(media_info.get('duration', {}).get('nanos'), scale=1e9) + username = config.get('user', {}).get('screen_name') + if username: + formats.extend(self._extract_mobile_formats(username, video_id)) + + self._remove_duplicate_formats(formats) self._sort_formats(formats) title = self._search_regex(r'([^<]+)', webpage, 'title') @@ -183,7 +248,7 @@ class TwitterIE(InfoExtractor): 'id': '643211948184596480', 'ext': 'mp4', 'title': 'FREE THE NIPPLE - FTN supporters on Hollywood Blvd today!', - 'thumbnail': 're:^https?://.*\.jpg', + 'thumbnail': r're:^https?://.*\.jpg', 'description': 'FREE THE NIPPLE on Twitter: "FTN supporters on Hollywood Blvd today! http://t.co/c7jHH749xJ"', 'uploader': 'FREE THE NIPPLE', 'uploader_id': 'freethenipple', @@ -199,11 +264,12 @@ class TwitterIE(InfoExtractor): 'ext': 'mp4', 'title': 'Gifs - tu vai cai tu vai cai tu nao eh capaz disso tu vai cai', 'description': 'Gifs on Twitter: "tu vai cai tu vai cai tu nao eh capaz disso tu vai cai https://t.co/tM46VHFlO5"', - 'thumbnail': 're:^https?://.*\.png', + 'thumbnail': r're:^https?://.*\.png', 'uploader': 'Gifs', 'uploader_id': 'giphz', }, 'expected_warnings': ['height', 'width'], + 'skip': 'Account suspended', }, { 'url': 'https://twitter.com/starwars/status/665052190608723968', 'md5': '39b7199856dee6cd4432e72c74bc69d4', @@ -236,15 +302,67 @@ class TwitterIE(InfoExtractor): 'info_dict': { 'id': '700207533655363584', 'ext': 'mp4', - 'title': 'jay - BEAT PROD: @suhmeduh #Damndaniel', - 'description': 'jay on Twitter: "BEAT PROD: @suhmeduh https://t.co/HBrQ4AfpvZ #Damndaniel https://t.co/byBooq2ejZ"', - 'thumbnail': 're:^https?://.*\.jpg', - 'uploader': 'jay', + 'title': 'JG - BEAT PROD: @suhmeduh #Damndaniel', + 'description': 'JG on Twitter: "BEAT PROD: @suhmeduh https://t.co/HBrQ4AfpvZ #Damndaniel https://t.co/byBooq2ejZ"', + 'thumbnail': r're:^https?://.*\.jpg', + 'uploader': 'JG', 'uploader_id': 'jaydingeer', }, 'params': { 'skip_download': True, # requires ffmpeg }, + }, { + 'url': 'https://twitter.com/Filmdrunk/status/713801302971588609', + 'md5': '89a15ed345d13b86e9a5a5e051fa308a', + 'info_dict': { + 'id': 'MIOxnrUteUd', + 'ext': 'mp4', + 'title': 'Dr.Pepperの飲み方 #japanese #バカ #ドクペ #電動ガン', + 'uploader': 'TAKUMA', + 'uploader_id': '1004126642786242560', + 'upload_date': '20140615', + }, + 'add_ie': ['Vine'], + }, { + 'url': 'https://twitter.com/captainamerica/status/719944021058060289', + 'info_dict': { + 'id': '719944021058060289', + 'ext': 'mp4', + 'title': 'Captain America - @King0fNerd Are you sure you made the right choice? Find out in theaters.', + 'description': 'Captain America on Twitter: "@King0fNerd Are you sure you made the right choice? Find out in theaters. https://t.co/GpgYi9xMJI"', + 'uploader_id': 'captainamerica', + 'uploader': 'Captain America', + }, + 'params': { + 'skip_download': True, # requires ffmpeg + }, + }, { + 'url': 'https://twitter.com/OPP_HSD/status/779210622571536384', + 'info_dict': { + 'id': '1zqKVVlkqLaKB', + 'ext': 'mp4', + 'title': 'Sgt Kerry Schmidt - LIVE on #Periscope: Road rage, mischief, assault, rollover and fire in one occurrence', + 'description': 'Sgt Kerry Schmidt on Twitter: "LIVE on #Periscope: Road rage, mischief, assault, rollover and fire in one occurrence https://t.co/EKrVgIXF3s"', + 'upload_date': '20160923', + 'uploader_id': 'OPP_HSD', + 'uploader': 'Sgt Kerry Schmidt', + 'timestamp': 1474613214, + }, + 'add_ie': ['Periscope'], + }, { + # has mp4 formats via mobile API + 'url': 'https://twitter.com/news_al3alm/status/852138619213144067', + 'info_dict': { + 'id': '852138619213144067', + 'ext': 'mp4', + 'title': 'عالم الأخبار - كلمة تاريخية بجلسة الجناسي التاريخية.. النائب خالد مؤنس العتيبي للمعارضين : اتقوا الله .. الظلم ظلمات يوم القيامة', + 'description': 'عالم الأخبار on Twitter: "كلمة تاريخية بجلسة الجناسي التاريخية.. النائب خالد مؤنس العتيبي للمعارضين : اتقوا الله .. الظلم ظلمات يوم القيامة https://t.co/xg6OhpyKfN"', + 'uploader': 'عالم الأخبار', + 'uploader_id': 'news_al3alm', + }, + 'params': { + 'format': 'best[format_id^=http-]', + }, }] def _real_extract(self, url): @@ -252,7 +370,11 @@ class TwitterIE(InfoExtractor): user_id = mobj.group('user_id') twid = mobj.group('id') - webpage = self._download_webpage(self._TEMPLATE_URL % (user_id, twid), twid) + webpage, urlh = self._download_webpage_handle( + self._TEMPLATE_URL % (user_id, twid), twid) + + if 'twitter.com/account/suspended' in urlh.geturl(): + raise ExtractorError('Account suspended by Twitter.', expected=True) username = remove_end(self._og_search_title(webpage), ' on Twitter') @@ -269,17 +391,6 @@ class TwitterIE(InfoExtractor): 'title': username + ' - ' + title, } - card_id = self._search_regex( - r'["\']/i/cards/tfw/v1/(\d+)', webpage, 'twitter card url', default=None) - if card_id: - card_url = 'https://twitter.com/i/cards/tfw/v1/' + card_id - info.update({ - '_type': 'url_transparent', - 'ie_key': 'TwitterCard', - 'url': card_url, - }) - return info - mobj = re.search(r'''(?x) ]+class="animated-gif"(?P[^>]+)>\s* ]+video-src="(?P[^"]+)" @@ -302,13 +413,22 @@ class TwitterIE(InfoExtractor): }) return info + twitter_card_url = None if 'class="PlayableMedia' in webpage: + twitter_card_url = '%s//twitter.com/i/videos/tweet/%s' % (self.http_scheme(), twid) + else: + twitter_card_iframe_url = self._search_regex( + r'data-full-card-iframe-url=([\'"])(?P(?:(?!\1).)+)\1', + webpage, 'Twitter card iframe URL', default=None, group='url') + if twitter_card_iframe_url: + twitter_card_url = compat_urlparse.urljoin(url, twitter_card_iframe_url) + + if twitter_card_url: info.update({ '_type': 'url_transparent', 'ie_key': 'TwitterCard', - 'url': '%s//twitter.com/i/videos/tweet/%s' % (self.http_scheme(), twid), + 'url': twitter_card_url, }) - return info raise ExtractorError('There\'s no video in this tweet.') @@ -316,7 +436,7 @@ class TwitterIE(InfoExtractor): class TwitterAmplifyIE(TwitterBaseIE): IE_NAME = 'twitter:amplify' - _VALID_URL = 'https?://amp\.twimg\.com/v/(?P[0-9a-f\-]{36})' + _VALID_URL = r'https?://amp\.twimg\.com/v/(?P[0-9a-f\-]{36})' _TEST = { 'url': 'https://amp.twimg.com/v/0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951', @@ -335,7 +455,7 @@ class TwitterAmplifyIE(TwitterBaseIE): vmap_url = self._html_search_meta( 'twitter:amplify:vmap', webpage, 'vmap url') - video_url = self._get_vmap_video_url(vmap_url, video_id) + formats = self._extract_formats_from_vmap_url(vmap_url, video_id) thumbnails = [] thumbnail = self._html_search_meta( @@ -357,11 +477,10 @@ class TwitterAmplifyIE(TwitterBaseIE): }) video_w, video_h = _find_dimension('player') - formats = [{ - 'url': video_url, + formats[0].update({ 'width': video_w, 'height': video_h, - }] + }) return { 'id': video_id,