X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/3477c644417600d9ec8f8d2a44f82da0a4b15eb5..5a9d00a5190582368d533f12db7638e735d7f7c8:/youtube_dl/extractor/francetv.py diff --git a/youtube_dl/extractor/francetv.py b/youtube_dl/extractor/francetv.py index 8e60cf6..095bb39 100644 --- a/youtube_dl/extractor/francetv.py +++ b/youtube_dl/extractor/francetv.py @@ -1,9 +1,8 @@ -# encoding: utf-8 +# coding: utf-8 from __future__ import unicode_literals import re -import json from .common import InfoExtractor from ..compat import compat_urlparse @@ -14,15 +13,17 @@ from ..utils import ( parse_duration, determine_ext, ) -from .dailymotion import DailymotionCloudIE +from .dailymotion import DailymotionIE class FranceTVBaseInfoExtractor(InfoExtractor): - def _extract_video(self, video_id, catalogue): + def _extract_video(self, video_id, catalogue=None): info = self._download_json( - 'http://webservices.francetelevisions.fr/tools/getInfosOeuvre/v2/?idDiffusion=%s&catalogue=%s' - % (video_id, catalogue), - video_id, 'Downloading video JSON') + 'https://sivideo.webservices.francetelevisions.fr/tools/getInfosOeuvre/v2/', + video_id, 'Downloading video JSON', query={ + 'idDiffusion': video_id, + 'catalogue': catalogue or '', + }) if info.get('status') == 'NOK': raise ExtractorError( @@ -60,28 +61,31 @@ class FranceTVBaseInfoExtractor(InfoExtractor): video_id, 'Downloading f4m manifest token', fatal=False) if f4m_url: formats.extend(self._extract_f4m_formats( - f4m_url + '&hdcore=3.7.0&plugin=aasp-3.7.0.39.44', video_id, 1, format_id)) + f4m_url + '&hdcore=3.7.0&plugin=aasp-3.7.0.39.44', + video_id, f4m_id=format_id, fatal=False)) elif ext == 'm3u8': - formats.extend(self._extract_m3u8_formats(video_url, video_id, 'mp4', m3u8_id=format_id)) + formats.extend(self._extract_m3u8_formats( + video_url, video_id, 'mp4', entry_protocol='m3u8_native', + m3u8_id=format_id, fatal=False)) elif video_url.startswith('rtmp'): formats.append({ 'url': video_url, 'format_id': 'rtmp-%s' % format_id, 'ext': 'flv', - 'preference': 1, }) else: - formats.append({ - 'url': video_url, - 'format_id': format_id, - 'preference': -1, - }) + if self._is_valid_url(video_url, video_id, format_id): + formats.append({ + 'url': video_url, + 'format_id': format_id, + }) self._sort_formats(formats) title = info['titre'] subtitle = info.get('sous_titre') if subtitle: title += ' - %s' % subtitle + title = title.strip() subtitles = {} subtitles_list = [{ @@ -103,35 +107,108 @@ class FranceTVBaseInfoExtractor(InfoExtractor): } -class PluzzIE(FranceTVBaseInfoExtractor): - IE_NAME = 'pluzz.francetv.fr' - _VALID_URL = r'https?://(?:m\.)?pluzz\.francetv\.fr/videos/(?P.+?)\.html' +class FranceTVIE(FranceTVBaseInfoExtractor): + _VALID_URL = r'https?://(?:(?:www\.)?france\.tv|mobile\.france\.tv)/(?:[^/]+/)*(?P[^/]+)\.html' - # Can't use tests, videos expire in 7 days + _TESTS = [{ + 'url': 'https://www.france.tv/france-2/13h15-le-dimanche/140921-les-mysteres-de-jesus.html', + 'info_dict': { + 'id': '157550144', + 'ext': 'mp4', + 'title': '13h15, le dimanche... - Les mystères de Jésus', + 'description': 'md5:75efe8d4c0a8205e5904498ffe1e1a42', + 'timestamp': 1494156300, + 'upload_date': '20170507', + }, + 'params': { + # m3u8 downloads + 'skip_download': True, + }, + }, { + # france3 + 'url': 'https://www.france.tv/france-3/des-chiffres-et-des-lettres/139063-emission-du-mardi-9-mai-2017.html', + 'only_matching': True, + }, { + # france4 + 'url': 'https://www.france.tv/france-4/hero-corp/saison-1/134151-apres-le-calme.html', + 'only_matching': True, + }, { + # france5 + 'url': 'https://www.france.tv/france-5/c-a-dire/saison-10/137013-c-a-dire.html', + 'only_matching': True, + }, { + # franceo + 'url': 'https://www.france.tv/france-o/archipels/132249-mon-ancetre-l-esclave.html', + 'only_matching': True, + }, { + # france2 live + 'url': 'https://www.france.tv/france-2/direct.html', + 'only_matching': True, + }, { + 'url': 'https://www.france.tv/documentaires/histoire/136517-argentine-les-500-bebes-voles-de-la-dictature.html', + 'only_matching': True, + }, { + 'url': 'https://www.france.tv/jeux-et-divertissements/divertissements/133965-le-web-contre-attaque.html', + 'only_matching': True, + }, { + 'url': 'https://mobile.france.tv/france-5/c-dans-l-air/137347-emission-du-vendredi-12-mai-2017.html', + 'only_matching': True, + }, { + 'url': 'https://www.france.tv/142749-rouge-sang.html', + 'only_matching': True, + }] def _real_extract(self, url): display_id = self._match_id(url) webpage = self._download_webpage(url, display_id) - video_id = self._html_search_meta( - 'id_video', webpage, 'video id', default=None) + catalogue = None + video_id = self._search_regex( + r'data-main-video=(["\'])(?P(?:(?!\1).)+)\1', + webpage, 'video id', default=None, group='id') + if not video_id: - video_id = self._search_regex( - r'data-diffusion=["\'](\d+)', webpage, 'video id') + video_id, catalogue = self._html_search_regex( + r'(?:href=|player\.setVideo\(\s*)"http://videos?\.francetv\.fr/video/([^@]+@[^"]+)"', + webpage, 'video ID').split('@') + return self._extract_video(video_id, catalogue) + + +class FranceTVEmbedIE(FranceTVBaseInfoExtractor): + _VALID_URL = r'https?://embed\.francetv\.fr/*\?.*?\bue=(?P[^&]+)' + + _TEST = { + 'url': 'http://embed.francetv.fr/?ue=7fd581a2ccf59d2fc5719c5c13cf6961', + 'info_dict': { + 'id': 'NI_983319', + 'ext': 'mp4', + 'title': 'Le Pen Reims', + 'upload_date': '20170505', + 'timestamp': 1493981780, + 'duration': 16, + }, + } + + def _real_extract(self, url): + video_id = self._match_id(url) + + video = self._download_json( + 'http://api-embed.webservices.francetelevisions.fr/key/%s' % video_id, + video_id) - return self._extract_video(video_id, 'Pluzz') + return self._extract_video(video['video_id'], video.get('catalog')) -class FranceTvInfoIE(FranceTVBaseInfoExtractor): +class FranceTVInfoIE(FranceTVBaseInfoExtractor): IE_NAME = 'francetvinfo.fr' - _VALID_URL = r'https?://(?:www|mobile)\.francetvinfo\.fr/.*/(?P.+)\.html' + _VALID_URL = r'https?://(?:www|mobile|france3-regions)\.francetvinfo\.fr/(?:[^/]+/)*(?P<title>[^/?#&.]+)' _TESTS = [{ 'url': 'http://www.francetvinfo.fr/replay-jt/france-3/soir-3/jt-grand-soir-3-lundi-26-aout-2013_393427.html', 'info_dict': { 'id': '84981923', - 'ext': 'flv', + 'ext': 'mp4', 'title': 'Soir 3', 'upload_date': '20130826', 'timestamp': 1377548400, @@ -139,6 +216,10 @@ class FranceTvInfoIE(FranceTVBaseInfoExtractor): 'fr': 'mincount:2', }, }, + 'params': { + # m3u8 downloads + 'skip_download': True, + }, }, { 'url': 'http://www.francetvinfo.fr/elections/europeennes/direct-europeennes-regardez-le-debat-entre-les-candidats-a-la-presidence-de-la-commission_600639.html', 'info_dict': { @@ -155,11 +236,50 @@ class FranceTvInfoIE(FranceTVBaseInfoExtractor): 'url': 'http://www.francetvinfo.fr/economie/entreprises/les-entreprises-familiales-le-secret-de-la-reussite_933271.html', 'md5': 'f485bda6e185e7d15dbc69b72bae993e', 'info_dict': { - 'id': '556e03339473995ee145930c', + 'id': 'NI_173343', 'ext': 'mp4', 'title': 'Les entreprises familiales : le secret de la réussite', - 'thumbnail': 're:^https?://.*\.jpe?g$', - } + 'thumbnail': r're:^https?://.*\.jpe?g$', + 'timestamp': 1433273139, + 'upload_date': '20150602', + }, + 'params': { + # m3u8 downloads + 'skip_download': True, + }, + }, { + 'url': 'http://france3-regions.francetvinfo.fr/bretagne/cotes-d-armor/thalassa-echappee-breizh-ce-venredi-dans-les-cotes-d-armor-954961.html', + 'md5': 'f485bda6e185e7d15dbc69b72bae993e', + 'info_dict': { + 'id': 'NI_657393', + 'ext': 'mp4', + 'title': 'Olivier Monthus, réalisateur de "Bretagne, le choix de l’Armor"', + 'description': 'md5:a3264114c9d29aeca11ced113c37b16c', + 'thumbnail': r're:^https?://.*\.jpe?g$', + 'timestamp': 1458300695, + 'upload_date': '20160318', + }, + 'params': { + 'skip_download': True, + }, + }, { + # Dailymotion embed + 'url': 'http://www.francetvinfo.fr/politique/notre-dame-des-landes/video-sur-france-inter-cecile-duflot-denonce-le-regard-meprisant-de-patrick-cohen_1520091.html', + 'md5': 'ee7f1828f25a648addc90cb2687b1f12', + 'info_dict': { + 'id': 'x4iiko0', + 'ext': 'mp4', + 'title': 'NDDL, référendum, Brexit : Cécile Duflot répond à Patrick Cohen', + 'description': 'Au lendemain de la victoire du "oui" au référendum sur l\'aéroport de Notre-Dame-des-Landes, l\'ancienne ministre écologiste est l\'invitée de Patrick Cohen. Plus d\'info : https://www.franceinter.fr/emissions/le-7-9/le-7-9-27-juin-2016', + 'timestamp': 1467011958, + 'upload_date': '20160627', + 'uploader': 'France Inter', + 'uploader_id': 'x2q2ez', + }, + 'add_ie': ['Dailymotion'], + }, { + 'url': 'http://france3-regions.francetvinfo.fr/limousin/emissions/jt-1213-limousin', + 'only_matching': True, }] def _real_extract(self, url): @@ -167,158 +287,45 @@ class FranceTvInfoIE(FranceTVBaseInfoExtractor): page_title = mobj.group('title') webpage = self._download_webpage(url, page_title) - dmcloud_url = DailymotionCloudIE._extract_dmcloud_url(webpage) - if dmcloud_url: - return self.url_result(dmcloud_url, 'DailymotionCloud') + dailymotion_urls = DailymotionIE._extract_urls(webpage) + if dailymotion_urls: + return self.playlist_result([ + self.url_result(dailymotion_url, DailymotionIE.ie_key()) + for dailymotion_url in dailymotion_urls]) video_id, catalogue = self._search_regex( - r'id-video=([^@]+@[^"]+)', webpage, 'video id').split('@') - return self._extract_video(video_id, catalogue) - - -class FranceTVIE(FranceTVBaseInfoExtractor): - IE_NAME = 'francetv' - IE_DESC = 'France 2, 3, 4, 5 and Ô' - _VALID_URL = r'''(?x) - https?:// - (?: - (?:www\.)?france[2345o]\.fr/ - (?: - emissions/[^/]+/(?:videos|diffusions)| - emission/[^/]+| - videos| - jt - ) - /| - embed\.francetv\.fr/\?ue= - ) - (?P<id>[^/?]+) - ''' - - _TESTS = [ - # france2 - { - 'url': 'http://www.france2.fr/emissions/13h15-le-samedi-le-dimanche/videos/75540104', - 'md5': 'c03fc87cb85429ffd55df32b9fc05523', - 'info_dict': { - 'id': '109169362', - 'ext': 'flv', - 'title': '13h15, le dimanche...', - 'description': 'md5:9a0932bb465f22d377a449be9d1a0ff7', - 'upload_date': '20140914', - 'timestamp': 1410693600, - }, - }, - # france3 - { - 'url': 'http://www.france3.fr/emissions/pieces-a-conviction/diffusions/13-11-2013_145575', - 'md5': '679bb8f8921f8623bd658fa2f8364da0', - 'info_dict': { - 'id': '000702326_CAPP_PicesconvictionExtrait313022013_120220131722_Au', - 'ext': 'mp4', - 'title': 'Le scandale du prix des médicaments', - 'description': 'md5:1384089fbee2f04fc6c9de025ee2e9ce', - 'upload_date': '20131113', - 'timestamp': 1384380000, - }, - }, - # france4 - { - 'url': 'http://www.france4.fr/emissions/hero-corp/videos/rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4', - 'md5': 'a182bf8d2c43d88d46ec48fbdd260c1c', - 'info_dict': { - 'id': 'rhozet_herocorp_bonus_1_20131106_1923_06112013172108_F4', - 'ext': 'mp4', - 'title': 'Hero Corp Making of - Extrait 1', - 'description': 'md5:c87d54871b1790679aec1197e73d650a', - 'upload_date': '20131106', - 'timestamp': 1383766500, - }, - }, - # france5 - { - 'url': 'http://www.france5.fr/emissions/c-a-dire/videos/quels_sont_les_enjeux_de_cette_rentree_politique__31-08-2015_908948?onglet=tous&page=1', - 'md5': 'f6c577df3806e26471b3d21631241fd0', - 'info_dict': { - 'id': '123327454', - 'ext': 'flv', - 'title': 'C à dire ?! - Quels sont les enjeux de cette rentrée politique ?', - 'description': 'md5:4a0d5cb5dce89d353522a84462bae5a4', - 'upload_date': '20150831', - 'timestamp': 1441035120, - }, - }, - # franceo - { - 'url': 'http://www.franceo.fr/jt/info-soir/18-07-2015', - 'md5': '47d5816d3b24351cdce512ad7ab31da8', - 'info_dict': { - 'id': '125377621', - 'ext': 'flv', - 'title': 'Infô soir', - 'description': 'md5:01b8c6915a3d93d8bbbd692651714309', - 'upload_date': '20150718', - 'timestamp': 1437241200, - 'duration': 414, - }, - }, - { - # francetv embed - 'url': 'http://embed.francetv.fr/?ue=8d7d3da1e3047c42ade5a5d7dfd3fc87', - 'info_dict': { - 'id': 'EV_30231', - 'ext': 'flv', - 'title': 'Alcaline, le concert avec Calogero', - 'description': 'md5:61f08036dcc8f47e9cfc33aed08ffaff', - 'upload_date': '20150226', - 'timestamp': 1424989860, - 'duration': 5400, - }, - }, - { - 'url': 'http://www.france4.fr/emission/highlander/diffusion-du-17-07-2015-04h05', - 'only_matching': True, - }, - { - 'url': 'http://www.franceo.fr/videos/125377617', - 'only_matching': True, - } - ] - - def _real_extract(self, url): - video_id = self._match_id(url) - webpage = self._download_webpage(url, video_id) - video_id, catalogue = self._html_search_regex( - r'href="http://videos?\.francetv\.fr/video/([^@]+@[^"]+)"', - webpage, 'video ID').split('@') + (r'id-video=([^@]+@[^"]+)', + r'<a[^>]+href="(?:https?:)?//videos\.francetv\.fr/video/([^@]+@[^"]+)"'), + webpage, 'video id').split('@') return self._extract_video(video_id, catalogue) -class GenerationQuoiIE(InfoExtractor): - IE_NAME = 'france2.fr:generation-quoi' - _VALID_URL = r'https?://generation-quoi\.france2\.fr/portrait/(?P<id>[^/?#]+)' +class GenerationWhatIE(InfoExtractor): + IE_NAME = 'france2.fr:generation-what' + _VALID_URL = r'https?://generation-what\.francetv\.fr/[^/]+/video/(?P<id>[^/?#]+)' - _TEST = { - 'url': 'http://generation-quoi.france2.fr/portrait/garde-a-vous', + _TESTS = [{ + 'url': 'http://generation-what.francetv.fr/portrait/video/present-arms', 'info_dict': { - 'id': 'k7FJX8VBcvvLmX4wA5Q', + 'id': 'wtvKYUG45iw', 'ext': 'mp4', - 'title': 'Génération Quoi - Garde à Vous', - 'uploader': 'Génération Quoi', - }, - 'params': { - # It uses Dailymotion - 'skip_download': True, + 'title': 'Generation What - Garde à vous - FRA', + 'uploader': 'Generation What', + 'uploader_id': 'UCHH9p1eetWCgt4kXBYCb3_w', + 'upload_date': '20160411', }, - } + }, { + 'url': 'http://generation-what.francetv.fr/europe/video/present-arms', + 'only_matching': True, + }] def _real_extract(self, url): display_id = self._match_id(url) - info_url = compat_urlparse.urljoin(url, '/medias/video/%s.json' % display_id) - info_json = self._download_webpage(info_url, display_id) - info = json.loads(info_json) - return self.url_result('http://www.dailymotion.com/video/%s' % info['id'], - ie='Dailymotion') + webpage = self._download_webpage(url, display_id) + youtube_id = self._search_regex( + r"window\.videoURL\s*=\s*'([0-9A-Za-z_-]{11})';", + webpage, 'youtube id') + return self.url_result(youtube_id, 'Youtube', youtube_id) class CultureboxIE(FranceTVBaseInfoExtractor): @@ -349,6 +356,7 @@ class CultureboxIE(FranceTVBaseInfoExtractor): raise ExtractorError('Video %s is not available' % name, expected=True) video_id, catalogue = self._search_regex( - r'"http://videos\.francetv\.fr/video/([^@]+@[^"]+)"', webpage, 'video id').split('@') + r'["\'>]https?://videos\.francetv\.fr/video/([^@]+@.+?)["\'<]', + webpage, 'video id').split('@') return self._extract_video(video_id, catalogue)