X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/9dc487f48b50767cf540fa36c3de2c386fd74c04..0bec2c9f91a3b4f834e00c8ea4790481e257d0a2:/youtube_dl/extractor/arte.py diff --git a/youtube_dl/extractor/arte.py b/youtube_dl/extractor/arte.py index 049f1fa..ffc3218 100644 --- a/youtube_dl/extractor/arte.py +++ b/youtube_dl/extractor/arte.py @@ -1,4 +1,4 @@ -# encoding: utf-8 +# coding: utf-8 from __future__ import unicode_literals import re @@ -6,15 +6,18 @@ import re from .common import InfoExtractor from ..compat import ( compat_parse_qs, + compat_str, compat_urllib_parse_urlparse, ) from ..utils import ( + ExtractorError, find_xpath_attr, - unified_strdate, get_element_by_attribute, int_or_none, NO_DEFAULT, qualities, + try_get, + unified_strdate, ) # There are different sources of video in arte.tv, the extraction process @@ -79,6 +82,16 @@ class ArteTVBaseIE(InfoExtractor): info = self._download_json(json_url, video_id) player_info = info['videoJsonPlayer'] + vsr = try_get(player_info, lambda x: x['VSR'], dict) + if not vsr: + error = None + if try_get(player_info, lambda x: x['custom_msg']['type']) == 'error': + error = try_get( + player_info, lambda x: x['custom_msg']['msg'], compat_str) + if not error: + error = 'Video %s is not available' % player_info.get('VID') or video_id + raise ExtractorError(error, expected=True) + upload_date_str = player_info.get('shootingDate') if not upload_date_str: upload_date_str = (player_info.get('VRA') or player_info.get('VDA') or '').split(' ')[0] @@ -107,7 +120,7 @@ class ArteTVBaseIE(InfoExtractor): langcode = LANGS.get(lang, lang) formats = [] - for format_id, format_dict in player_info['VSR'].items(): + for format_id, format_dict in vsr.items(): f = dict(format_dict) versionCode = f.get('versionCode') l = re.escape(langcode) @@ -180,7 +193,7 @@ class ArteTVBaseIE(InfoExtractor): class ArteTVPlus7IE(ArteTVBaseIE): IE_NAME = 'arte.tv:+7' - _VALID_URL = r'https?://(?:(?:www|sites)\.)?arte\.tv/[^/]+/(?Pfr|de|en|es)/(?:[^/]+/)*(?P[^/?#&]+)' + _VALID_URL = r'https?://(?:(?:www|sites)\.)?arte\.tv/(?:[^/]+/)?(?Pfr|de|en|es)/(?:videos/)?(?:[^/]+/)*(?P[^/?#&]+)' _TESTS = [{ 'url': 'http://www.arte.tv/guide/de/sendungen/XEN/xenius/?vid=055918-015_PLUS7-D', @@ -188,6 +201,9 @@ class ArteTVPlus7IE(ArteTVBaseIE): }, { 'url': 'http://sites.arte.tv/karambolage/de/video/karambolage-22', 'only_matching': True, + }, { + 'url': 'http://www.arte.tv/de/videos/048696-000-A/der-kluge-bauch-unser-zweites-gehirn', + 'only_matching': True, }] @classmethod @@ -410,6 +426,22 @@ class ArteTVEmbedIE(ArteTVPlus7IE): return self._extract_from_json_url(json_url, video_id, lang) +class TheOperaPlatformIE(ArteTVPlus7IE): + IE_NAME = 'theoperaplatform' + _VALID_URL = r'https?://(?:www\.)?theoperaplatform\.eu/(?Pfr|de|en|es)/(?P[^/?#&]+)' + + _TESTS = [{ + 'url': 'http://www.theoperaplatform.eu/de/opera/verdi-otello', + 'md5': '970655901fa2e82e04c00b955e9afe7b', + 'info_dict': { + 'id': '060338-009-A', + 'ext': 'mp4', + 'title': 'Verdi - OTELLO', + 'upload_date': '20160927', + }, + }] + + class ArteTVPlaylistIE(ArteTVBaseIE): IE_NAME = 'arte.tv:playlist' _VALID_URL = r'https?://(?:www\.)?arte\.tv/guide/(?Pfr|de|en|es)/[^#]*#collection/(?PPL-\d+)' @@ -419,6 +451,7 @@ class ArteTVPlaylistIE(ArteTVBaseIE): 'info_dict': { 'id': 'PL-013263', 'title': 'Areva & Uramin', + 'description': 'md5:a1dc0312ce357c262259139cfd48c9bf', }, 'playlist_mincount': 6, }, {