X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/90cd207d52d77e075e62cc256cc39dceee05bbf3..6a680aa147c68717d8224546bf1bcbb737b78ac7:/youtube_dl/extractor/eitb.py diff --git a/youtube_dl/extractor/eitb.py b/youtube_dl/extractor/eitb.py index 4ba3231..ee5ead1 100644 --- a/youtube_dl/extractor/eitb.py +++ b/youtube_dl/extractor/eitb.py @@ -1,37 +1,88 @@ -# encoding: utf-8 -import re +# coding: utf-8 +from __future__ import unicode_literals from .common import InfoExtractor -from .brightcove import BrightcoveIE -from ..utils import ExtractorError +from ..utils import ( + float_or_none, + int_or_none, + parse_iso8601, + sanitized_Request, +) class EitbIE(InfoExtractor): - IE_NAME = u'eitb.tv' - _VALID_URL = r'https?://www\.eitb\.tv/(eu/bideoa|es/video)/[^/]+/(?P\d+)/(?P\d+)' + IE_NAME = 'eitb.tv' + _VALID_URL = r'https?://(?:www\.)?eitb\.tv/(?:eu/bideoa|es/video)/[^/]+/\d+/(?P\d+)' _TEST = { - u'add_ie': ['Brightcove'], - u'url': u'http://www.eitb.tv/es/video/60-minutos-60-minutos-2013-2014/2677100210001/2743577154001/lasa-y-zabala-30-anos/', - u'md5': u'edf4436247185adee3ea18ce64c47998', - u'info_dict': { - u'id': u'2743577154001', - u'ext': u'mp4', - u'title': u'60 minutos (Lasa y Zabala, 30 años)', - # All videos from eitb has this description in the brightcove info - u'description': u'.', - u'uploader': u'Euskal Telebista', + 'url': 'http://www.eitb.tv/es/video/60-minutos-60-minutos-2013-2014/4104995148001/4090227752001/lasa-y-zabala-30-anos/', + 'md5': 'edf4436247185adee3ea18ce64c47998', + 'info_dict': { + 'id': '4090227752001', + 'ext': 'mp4', + 'title': '60 minutos (Lasa y Zabala, 30 años)', + 'description': 'Programa de reportajes de actualidad.', + 'duration': 3996.76, + 'timestamp': 1381789200, + 'upload_date': '20131014', + 'tags': list, }, } def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - chapter_id = mobj.group('chapter_id') - webpage = self._download_webpage(url, chapter_id) - bc_url = BrightcoveIE._extract_brightcove_url(webpage) - if bc_url is None: - raise ExtractorError(u'Could not extract the Brightcove url') - # The BrightcoveExperience object doesn't contain the video id, we set - # it manually - bc_url += '&%40videoPlayer={0}'.format(chapter_id) - return self.url_result(bc_url, BrightcoveIE.ie_key()) + video_id = self._match_id(url) + + video = self._download_json( + 'http://mam.eitb.eus/mam/REST/ServiceMultiweb/Video/MULTIWEBTV/%s/' % video_id, + video_id, 'Downloading video JSON') + + media = video['web_media'][0] + + formats = [] + for rendition in media['RENDITIONS']: + video_url = rendition.get('PMD_URL') + if not video_url: + continue + tbr = float_or_none(rendition.get('ENCODING_RATE'), 1000) + format_id = 'http' + if tbr: + format_id += '-%d' % int(tbr) + formats.append({ + 'url': rendition['PMD_URL'], + 'format_id': format_id, + 'width': int_or_none(rendition.get('FRAME_WIDTH')), + 'height': int_or_none(rendition.get('FRAME_HEIGHT')), + 'tbr': tbr, + }) + + hls_url = media.get('HLS_SURL') + if hls_url: + request = sanitized_Request( + 'http://mam.eitb.eus/mam/REST/ServiceMultiweb/DomainRestrictedSecurity/TokenAuth/', + headers={'Referer': url}) + token_data = self._download_json( + request, video_id, 'Downloading auth token', fatal=False) + if token_data: + token = token_data.get('token') + if token: + formats.extend(self._extract_m3u8_formats( + '%s?hdnts=%s' % (hls_url, token), video_id, m3u8_id='hls', fatal=False)) + + hds_url = media.get('HDS_SURL') + if hds_url: + formats.extend(self._extract_f4m_formats( + '%s?hdcore=3.7.0' % hds_url.replace('euskalsvod', 'euskalvod'), + video_id, f4m_id='hds', fatal=False)) + + self._sort_formats(formats) + + return { + 'id': video_id, + 'title': media.get('NAME_ES') or media.get('name') or media['NAME_EU'], + 'description': media.get('SHORT_DESC_ES') or video.get('desc_group') or media.get('SHORT_DESC_EU'), + 'thumbnail': media.get('STILL_URL') or media.get('THUMBNAIL_URL'), + 'duration': float_or_none(media.get('LENGTH'), 1000), + 'timestamp': parse_iso8601(media.get('BROADCST_DATE'), ' '), + 'tags': media.get('TAGS'), + 'formats': formats, + }