X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fbyutv.py;h=0b11bf11fce42168b0aa6dc7592ae0fa259e7c0e;hb=HEAD;hp=4be175d7039dd845f7c961af552bc1153b73598e;hpb=9218a6b4f5292cc3d7761484e6b5dbe1e7a1a998;p=youtube-dl diff --git a/youtube_dl/extractor/byutv.py b/youtube_dl/extractor/byutv.py index 4be175d70..0b11bf11f 100644 --- a/youtube_dl/extractor/byutv.py +++ b/youtube_dl/extractor/byutv.py @@ -3,29 +3,52 @@ from __future__ import unicode_literals import re from .common import InfoExtractor -from ..utils import ExtractorError +from ..utils import ( + determine_ext, + merge_dicts, + parse_duration, + url_or_none, +) class BYUtvIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?byutv\.org/watch/(?!event/)(?P[0-9a-f-]+)(?:/(?P[^/?#&]+))?' + _VALID_URL = r'https?://(?:www\.)?byutv\.org/(?:watch|player)/(?!event/)(?P[0-9a-f-]+)(?:/(?P[^/?#&]+))?' _TESTS = [{ + # ooyalaVOD 'url': 'http://www.byutv.org/watch/6587b9a3-89d2-42a6-a7f7-fd2f81840a7d/studio-c-season-5-episode-5', 'info_dict': { - 'id': '6587b9a3-89d2-42a6-a7f7-fd2f81840a7d', + 'id': 'ZvanRocTpW-G5_yZFeltTAMv6jxOU9KH', 'display_id': 'studio-c-season-5-episode-5', 'ext': 'mp4', 'title': 'Season 5 Episode 5', - 'description': 'md5:e07269172baff037f8e8bf9956bc9747', - 'thumbnail': 're:^https?://.*\.jpg$', + 'description': 'md5:1d31dc18ef4f075b28f6a65937d22c65', + 'thumbnail': r're:^https?://.*', 'duration': 1486.486, }, 'params': { 'skip_download': True, }, 'add_ie': ['Ooyala'], + }, { + # dvr + 'url': 'https://www.byutv.org/player/8f1dab9b-b243-47c8-b525-3e2d021a3451/byu-softball-pacific-vs-byu-41219---game-2', + 'info_dict': { + 'id': '8f1dab9b-b243-47c8-b525-3e2d021a3451', + 'display_id': 'byu-softball-pacific-vs-byu-41219---game-2', + 'ext': 'mp4', + 'title': 'Pacific vs. BYU (4/12/19)', + 'description': 'md5:1ac7b57cb9a78015910a4834790ce1f3', + 'duration': 11645, + }, + 'params': { + 'skip_download': True + }, }, { 'url': 'http://www.byutv.org/watch/6587b9a3-89d2-42a6-a7f7-fd2f81840a7d', 'only_matching': True, + }, { + 'url': 'https://www.byutv.org/player/27741493-dc83-40b0-8420-e7ae38a2ae98/byu-football-toledo-vs-byu-93016?listid=4fe0fee5-0d3c-4a29-b725-e4948627f472&listindex=0&q=toledo', + 'only_matching': True, }] def _real_extract(self, url): @@ -33,61 +56,62 @@ class BYUtvIE(InfoExtractor): video_id = mobj.group('id') display_id = mobj.group('display_id') or video_id - webpage = self._download_webpage(url, display_id) - episode_code = self._search_regex( - r'(?s)episode:(.*?\}),\s*\n', webpage, 'episode information') - - ep = self._parse_json( - episode_code, display_id, transform_source=lambda s: - re.sub(r'(\n\s+)([a-zA-Z]+):\s+\'(.*?)\'', r'\1"\2": "\3"', s)) - - if ep['providerType'] != 'Ooyala': - raise ExtractorError('Unsupported provider %s' % ep['provider']) - - return { - '_type': 'url_transparent', - 'ie_key': 'Ooyala', - 'url': 'ooyala:%s' % ep['providerId'], + video = self._download_json( + 'https://api.byutv.org/api3/catalog/getvideosforcontent', + display_id, query={ + 'contentid': video_id, + 'channel': 'byutv', + 'x-byutv-context': 'web$US', + }, headers={ + 'x-byutv-context': 'web$US', + 'x-byutv-platformkey': 'xsaaw9c7y5', + }) + + ep = video.get('ooyalaVOD') + if ep: + return { + '_type': 'url_transparent', + 'ie_key': 'Ooyala', + 'url': 'ooyala:%s' % ep['providerId'], + 'id': video_id, + 'display_id': display_id, + 'title': ep.get('title'), + 'description': ep.get('description'), + 'thumbnail': ep.get('imageThumbnail'), + } + + info = {} + formats = [] + for format_id, ep in video.items(): + if not isinstance(ep, dict): + continue + video_url = url_or_none(ep.get('videoUrl')) + if not video_url: + continue + ext = determine_ext(video_url) + if ext == 'm3u8': + formats.extend(self._extract_m3u8_formats( + video_url, video_id, 'mp4', entry_protocol='m3u8_native', + m3u8_id='hls', fatal=False)) + elif ext == 'mpd': + formats.extend(self._extract_mpd_formats( + video_url, video_id, mpd_id='dash', fatal=False)) + else: + formats.append({ + 'url': video_url, + 'format_id': format_id, + }) + merge_dicts(info, { + 'title': ep.get('title'), + 'description': ep.get('description'), + 'thumbnail': ep.get('imageThumbnail'), + 'duration': parse_duration(ep.get('length')), + }) + self._sort_formats(formats) + + return merge_dicts(info, { 'id': video_id, 'display_id': display_id, - 'title': ep['title'], - 'description': ep.get('description'), - 'thumbnail': ep.get('imageThumbnail'), - } - - -class BYUtvEventIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?byutv\.org/watch/event/(?P[0-9a-f-]+)' - _TEST = { - 'url': 'http://www.byutv.org/watch/event/29941b9b-8bf6-48d2-aebf-7a87add9e34b', - 'info_dict': { - 'id': '29941b9b-8bf6-48d2-aebf-7a87add9e34b', - 'ext': 'mp4', - 'title': 'Toledo vs. BYU (9/30/16)', - }, - 'params': { - 'skip_download': True, - }, - 'add_ie': ['Ooyala'], - } - - def _real_extract(self, url): - video_id = self._match_id(url) - - webpage = self._download_webpage(url, video_id) - - ooyala_id = self._search_regex( - r'providerId\s*:\s*(["\'])(?P(?:(?!\1).)+)\1', - webpage, 'ooyala id', group='id') - - title = self._search_regex( - r'class=["\']description["\'][^>]*>\s*

([^<]+)

', webpage, - 'title').strip() - - return { - '_type': 'url_transparent', - 'ie_key': 'Ooyala', - 'url': 'ooyala:%s' % ooyala_id, - 'id': video_id, - 'title': title, - } + 'title': display_id, + 'formats': formats, + })