X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Ffrancetv.py;h=d7e9aef90db5b7c8f8027b0c072420d5ec346f66;hb=9e1a5b845586a0a5431fb72467142046d8571e6f;hp=086cafca027e3b99967f50f3220d1632f1cc8033;hpb=d2f9cdb2058fe3530ba42ba26fe843fb94021f70;p=youtube-dl diff --git a/youtube_dl/extractor/francetv.py b/youtube_dl/extractor/francetv.py index 086cafca0..d7e9aef90 100644 --- a/youtube_dl/extractor/francetv.py +++ b/youtube_dl/extractor/francetv.py @@ -1,38 +1,80 @@ # encoding: utf-8 + +from __future__ import unicode_literals + import re -import xml.etree.ElementTree import json from .common import InfoExtractor from ..utils import ( compat_urlparse, + ExtractorError, + clean_html, + parse_duration, + compat_urllib_parse_urlparse, + int_or_none, ) class FranceTVBaseInfoExtractor(InfoExtractor): - def _extract_video(self, video_id): - xml_desc = self._download_webpage( - 'http://www.francetvinfo.fr/appftv/webservices/video/' - 'getInfosOeuvre.php?id-diffusion=' - + video_id, video_id, 'Downloading XML config') - info = xml.etree.ElementTree.fromstring(xml_desc.encode('utf-8')) - - manifest_url = info.find('videos/video/url').text - video_url = manifest_url.replace('manifest.f4m', 'index_2_av.m3u8') - video_url = video_url.replace('/z/', '/i/') - thumbnail_path = info.find('image').text - - return {'id': video_id, - 'ext': 'mp4', - 'url': video_url, - 'title': info.find('titre').text, - 'thumbnail': compat_urlparse.urljoin('http://pluzz.francetv.fr', thumbnail_path), - 'description': info.find('synopsis').text, - } + def _extract_video(self, video_id, catalogue): + info = self._download_json( + 'http://webservices.francetelevisions.fr/tools/getInfosOeuvre/v2/?idDiffusion=%s&catalogue=%s' + % (video_id, catalogue), + video_id, 'Downloading video JSON') + + if info.get('status') == 'NOK': + raise ExtractorError( + '%s returned error: %s' % (self.IE_NAME, info['message']), expected=True) + + formats = [] + for video in info['videos']: + if video['statut'] != 'ONLINE': + continue + video_url = video['url'] + if not video_url: + continue + format_id = video['format'] + if video_url.endswith('.f4m'): + video_url_parsed = compat_urllib_parse_urlparse(video_url) + f4m_url = self._download_webpage( + 'http://hdfauth.francetv.fr/esi/urltokengen2.html?url=%s' % video_url_parsed.path, + video_id, 'Downloading f4m manifest token', fatal=False) + if f4m_url: + f4m_formats = self._extract_f4m_formats(f4m_url, video_id) + for f4m_format in f4m_formats: + f4m_format['preference'] = 1 + formats.extend(f4m_formats) + elif video_url.endswith('.m3u8'): + formats.extend(self._extract_m3u8_formats(video_url, video_id, 'mp4')) + 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, + }) + self._sort_formats(formats) + + return { + 'id': video_id, + 'title': info['titre'], + 'description': clean_html(info['synopsis']), + 'thumbnail': compat_urlparse.urljoin('http://pluzz.francetv.fr', info['image']), + 'duration': parse_duration(info['duree']), + 'timestamp': int_or_none(info['diffusion']['timestamp']), + 'formats': formats, + } class PluzzIE(FranceTVBaseInfoExtractor): - IE_NAME = u'pluzz.francetv.fr' + IE_NAME = 'pluzz.francetv.fr' _VALID_URL = r'https?://pluzz\.francetv\.fr/videos/(.*?)\.html' # Can't use tests, videos expire in 7 days @@ -42,81 +84,147 @@ class PluzzIE(FranceTVBaseInfoExtractor): webpage = self._download_webpage(url, title) video_id = self._search_regex( r'data-diffusion="(\d+)"', webpage, 'ID') - return self._extract_video(video_id) + return self._extract_video(video_id, 'Pluzz') class FranceTvInfoIE(FranceTVBaseInfoExtractor): - IE_NAME = u'francetvinfo.fr' - _VALID_URL = r'https?://www\.francetvinfo\.fr/replay.*/(?P.+).html' + IE_NAME = 'francetvinfo.fr' + _VALID_URL = r'https?://(?:www|mobile)\.francetvinfo\.fr/.*/(?P<title>.+)\.html' - _TEST = { - u'url': u'http://www.francetvinfo.fr/replay-jt/france-3/soir-3/jt-grand-soir-3-lundi-26-aout-2013_393427.html', - u'file': u'84981923.mp4', - u'info_dict': { - u'title': u'Soir 3', + _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', + 'title': 'Soir 3', + 'upload_date': '20130826', + 'timestamp': 1377548400, }, - u'params': { - u'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': { + 'id': 'EV_20019', + 'ext': 'mp4', + 'title': 'Débat des candidats à la Commission européenne', + 'description': 'Débat des candidats à la Commission européenne', }, - } + 'params': { + 'skip_download': 'HLS (reqires ffmpeg)' + }, + 'skip': 'Ce direct est terminé et sera disponible en rattrapage dans quelques minutes.', + }] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) page_title = mobj.group('title') webpage = self._download_webpage(url, page_title) - video_id = self._search_regex(r'id-video=(\d+?)"', webpage, u'video id') - return self._extract_video(video_id) + video_id, catalogue = self._search_regex( + r'id-video=([^@]+@[^"]+)', webpage, 'video id').split('@') + return self._extract_video(video_id, catalogue) -class France2IE(FranceTVBaseInfoExtractor): - IE_NAME = u'france2.fr' - _VALID_URL = r'''(?x)https?://www\.france2\.fr/ +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/(?P<id>\d+) - | emission/(?P<key>[^/?]+) + emissions/.*?/(videos|emissions)/(?P<id>[^/?]+) + | (emissions?|jt)/(?P<key>[^/?]+) )''' - _TEST = { - u'url': u'http://www.france2.fr/emissions/13h15-le-samedi-le-dimanche/videos/75540104', - u'file': u'75540104.mp4', - u'info_dict': { - u'title': u'13h15, le samedi...', - u'description': u'md5:2e5b58ba7a2d3692b35c792be081a03d', + _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, + }, }, - u'params': { - u'skip_download': True, + # 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/92837968', + 'md5': '78f0f4064f9074438e660785bbf2c5d9', + 'info_dict': { + 'id': '108961659', + 'ext': 'flv', + 'title': 'C à dire ?!', + 'description': 'md5:1a4aeab476eb657bf57c4ff122129f81', + 'upload_date': '20140915', + 'timestamp': 1410795000, + }, + }, + # franceo + { + 'url': 'http://www.franceo.fr/jt/info-afrique/04-12-2013', + 'md5': '52f0bfe202848b15915a2f39aaa8981b', + 'info_dict': { + 'id': '108634970', + 'ext': 'flv', + 'title': 'Infô Afrique', + 'description': 'md5:ebf346da789428841bee0fd2a935ea55', + 'upload_date': '20140915', + 'timestamp': 1410822000, + }, + }, + ] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) - if mobj.group('key'): - webpage = self._download_webpage(url, mobj.group('key')) - video_id = self._html_search_regex( - r'''(?x)<div\s+class="video-player">\s* - <a\s+href="http://videos.francetv.fr/video/([0-9]+)"\s+ - class="francetv-video-player">''', - webpage, u'video ID') - else: - video_id = mobj.group('id') - return self._extract_video(video_id) + webpage = self._download_webpage(url, mobj.group('key') or mobj.group('id')) + video_id, catalogue = self._html_search_regex( + r'href="http://videos\.francetv\.fr/video/([^@]+@[^"]+)"', + webpage, 'video ID').split('@') + return self._extract_video(video_id, catalogue) class GenerationQuoiIE(InfoExtractor): - IE_NAME = u'france2.fr:generation-quoi' + IE_NAME = 'france2.fr:generation-quoi' _VALID_URL = r'https?://generation-quoi\.france2\.fr/portrait/(?P<name>.*)(\?|$)' _TEST = { - u'url': u'http://generation-quoi.france2.fr/portrait/garde-a-vous', - u'file': u'k7FJX8VBcvvLmX4wA5Q.mp4', - u'info_dict': { - u'title': u'Génération Quoi - Garde à Vous', - u'uploader': u'Génération Quoi', + 'url': 'http://generation-quoi.france2.fr/portrait/garde-a-vous', + 'file': 'k7FJX8VBcvvLmX4wA5Q.mp4', + 'info_dict': { + 'title': 'Génération Quoi - Garde à Vous', + 'uploader': 'Génération Quoi', }, - u'params': { + 'params': { # It uses Dailymotion - u'skip_download': True, + 'skip_download': True, }, + 'skip': 'Only available from France', } def _real_extract(self, url): @@ -126,4 +234,31 @@ class GenerationQuoiIE(InfoExtractor): info_json = self._download_webpage(info_url, name) info = json.loads(info_json) return self.url_result('http://www.dailymotion.com/video/%s' % info['id'], - ie='Dailymotion') + ie='Dailymotion') + + +class CultureboxIE(FranceTVBaseInfoExtractor): + IE_NAME = 'culturebox.francetvinfo.fr' + _VALID_URL = r'https?://(?:m\.)?culturebox\.francetvinfo\.fr/(?P<name>.*?)(\?|$)' + + _TEST = { + 'url': 'http://culturebox.francetvinfo.fr/festivals/dans-les-jardins-de-william-christie/dans-les-jardins-de-william-christie-le-camus-162553', + 'md5': '5ad6dec1ffb2a3fbcb20cc4b744be8d6', + 'info_dict': { + 'id': 'EV_22853', + 'ext': 'flv', + 'title': 'Dans les jardins de William Christie - Le Camus', + 'description': 'md5:4710c82315c40f0c865ca8b9a68b5299', + 'upload_date': '20140829', + 'timestamp': 1409317200, + }, + } + + def _real_extract(self, url): + mobj = re.match(self._VALID_URL, url) + name = mobj.group('name') + webpage = self._download_webpage(url, name) + video_id, catalogue = self._search_regex( + r'"http://videos\.francetv\.fr/video/([^@]+@[^"]+)"', webpage, 'video id').split('@') + + return self._extract_video(video_id, catalogue)