From 1c97b0a777f52c520587e93e7e61721fa6195977 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Sat, 16 May 2015 20:00:40 +0600 Subject: [PATCH] [ooyala:external] Add extractor --- youtube_dl/extractor/__init__.py | 5 +- youtube_dl/extractor/ooyala.py | 153 ++++++++++++++++++++----------- 2 files changed, 103 insertions(+), 55 deletions(-) diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py index f293bc2a4..1731f4fb2 100644 --- a/youtube_dl/extractor/__init__.py +++ b/youtube_dl/extractor/__init__.py @@ -376,7 +376,10 @@ from .nytimes import ( from .nuvid import NuvidIE from .odnoklassniki import OdnoklassnikiIE from .oktoberfesttv import OktoberfestTVIE -from .ooyala import OoyalaIE +from .ooyala import ( + OoyalaIE, + OoyalaExternalIE, +) from .openfilm import OpenFilmIE from .orf import ( ORFTVthekIE, diff --git a/youtube_dl/extractor/ooyala.py b/youtube_dl/extractor/ooyala.py index c0e6d643d..9f4fe5b29 100644 --- a/youtube_dl/extractor/ooyala.py +++ b/youtube_dl/extractor/ooyala.py @@ -12,50 +12,7 @@ from ..utils import ( ) -class OoyalaIE(InfoExtractor): - _VALID_URL = r'(?:ooyala:|https?://.+?\.ooyala\.com/.*?(?:embedCode|ec)=)(?P.+?)(&|$)' - - _TESTS = [ - { - # From http://it.slashdot.org/story/13/04/25/178216/recovering-data-from-broken-hard-drives-and-ssds-video - 'url': 'http://player.ooyala.com/player.js?embedCode=pxczE2YjpfHfn1f3M-ykG_AmJRRn0PD8', - 'info_dict': { - 'id': 'pxczE2YjpfHfn1f3M-ykG_AmJRRn0PD8', - 'ext': 'mp4', - 'title': 'Explaining Data Recovery from Hard Drives and SSDs', - 'description': 'How badly damaged does a drive have to be to defeat Russell and his crew? Apparently, smashed to bits.', - }, - }, { - # Only available for ipad - 'url': 'http://player.ooyala.com/player.js?embedCode=x1b3lqZDq9y_7kMyC2Op5qo-p077tXD0', - 'info_dict': { - 'id': 'x1b3lqZDq9y_7kMyC2Op5qo-p077tXD0', - 'ext': 'mp4', - 'title': 'Simulation Overview - Levels of Simulation', - 'description': '', - }, - }, - { - # Information available only through SAS api - # From http://community.plm.automation.siemens.com/t5/News-NX-Manufacturing/Tool-Path-Divide/ba-p/4187 - 'url': 'http://player.ooyala.com/player.js?embedCode=FiOG81ZTrvckcchQxmalf4aQj590qTEx', - 'md5': 'a84001441b35ea492bc03736e59e7935', - 'info_dict': { - 'id': 'FiOG81ZTrvckcchQxmalf4aQj590qTEx', - 'ext': 'mp4', - 'title': 'Ooyala video', - } - } - ] - - @staticmethod - def _url_for_embed_code(embed_code): - return 'http://player.ooyala.com/player.js?embedCode=%s' % embed_code - - @classmethod - def _build_url_result(cls, embed_code): - return cls.url_result(cls._url_for_embed_code(embed_code), - ie=cls.ie_key()) +class OoyalaBaseIE(InfoExtractor): def _extract_result(self, info, more_info): embedCode = info['embedCode'] @@ -77,11 +34,8 @@ class OoyalaIE(InfoExtractor): 'thumbnail': more_info['promo'], } - def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - embedCode = mobj.group('id') - player_url = 'http://player.ooyala.com/player.js?embedCode=%s' % embedCode - player = self._download_webpage(player_url, embedCode) + def _extract(self, player_url, video_id): + player = self._download_webpage(player_url, video_id) mobile_url = self._search_regex(r'mobile_player_url="(.+?)&device="', player, 'mobile player url') # Looks like some videos are only available for particular devices @@ -94,7 +48,7 @@ class OoyalaIE(InfoExtractor): devices.insert(0, 'unknown') for device in devices: mobile_player = self._download_webpage( - '%s&device=%s' % (mobile_url, device), embedCode, + '%s&device=%s' % (mobile_url, device), video_id, 'Downloading mobile player JS for %s device' % device) videos_info = self._search_regex( r'var streams=window.oo_testEnv\?\[\]:eval\("\((\[{.*?}\])\)"\);', @@ -106,9 +60,9 @@ class OoyalaIE(InfoExtractor): formats = [] auth_data = self._download_json( 'http://player.ooyala.com/sas/player_api/v1/authorization/embed_code/%s/%s?domain=www.example.org&supportedFormats=mp4,webm' % (embedCode, embedCode), - embedCode) + video_id) - cur_auth_data = auth_data['authorization_data'][embedCode] + cur_auth_data = auth_data['authorization_data'][video_id] for stream in cur_auth_data['streams']: formats.append({ @@ -123,7 +77,7 @@ class OoyalaIE(InfoExtractor): }) if formats: return { - 'id': embedCode, + 'id': video_id, 'formats': formats, 'title': 'Ooyala video', } @@ -143,9 +97,100 @@ class OoyalaIE(InfoExtractor): videos = [self._extract_result(info, more_info) for (info, more_info) in zip(videos_info, videos_more_info['lineup'])] return { '_type': 'playlist', - 'id': embedCode, + 'id': video_id, 'title': unescapeHTML(videos_more_info['title']), 'entries': videos, } else: return self._extract_result(videos_info[0], videos_more_info) + + +class OoyalaIE(OoyalaBaseIE): + _VALID_URL = r'(?:ooyala:|https?://.+?\.ooyala\.com/.*?(?:embedCode|ec)=)(?P.+?)(&|$)' + + _TESTS = [ + { + # From http://it.slashdot.org/story/13/04/25/178216/recovering-data-from-broken-hard-drives-and-ssds-video + 'url': 'http://player.ooyala.com/player.js?embedCode=pxczE2YjpfHfn1f3M-ykG_AmJRRn0PD8', + 'info_dict': { + 'id': 'pxczE2YjpfHfn1f3M-ykG_AmJRRn0PD8', + 'ext': 'mp4', + 'title': 'Explaining Data Recovery from Hard Drives and SSDs', + 'description': 'How badly damaged does a drive have to be to defeat Russell and his crew? Apparently, smashed to bits.', + }, + }, { + # Only available for ipad + 'url': 'http://player.ooyala.com/player.js?embedCode=x1b3lqZDq9y_7kMyC2Op5qo-p077tXD0', + 'info_dict': { + 'id': 'x1b3lqZDq9y_7kMyC2Op5qo-p077tXD0', + 'ext': 'mp4', + 'title': 'Simulation Overview - Levels of Simulation', + 'description': '', + }, + }, + { + # Information available only through SAS api + # From http://community.plm.automation.siemens.com/t5/News-NX-Manufacturing/Tool-Path-Divide/ba-p/4187 + 'url': 'http://player.ooyala.com/player.js?embedCode=FiOG81ZTrvckcchQxmalf4aQj590qTEx', + 'md5': 'a84001441b35ea492bc03736e59e7935', + 'info_dict': { + 'id': 'FiOG81ZTrvckcchQxmalf4aQj590qTEx', + 'ext': 'mp4', + 'title': 'Ooyala video', + } + } + ] + + @staticmethod + def _url_for_embed_code(embed_code): + return 'http://player.ooyala.com/player.js?embedCode=%s' % embed_code + + @classmethod + def _build_url_result(cls, embed_code): + return cls.url_result(cls._url_for_embed_code(embed_code), + ie=cls.ie_key()) + + def _real_extract(self, url): + embed_code = self._match_id(url) + player_url = 'http://player.ooyala.com/player.js?embedCode=%s' % embed_code + return self._extract(player_url, embed_code) + + +class OoyalaExternalIE(OoyalaBaseIE): + _VALID_URL = r'''(?x) + (?: + ooyalaexternal:| + https?://.+?\.ooyala\.com/.*?\bexternalId= + ) + (?P[^:]+) + : + (?P.+) + (?: + :| + .*?&pcode= + ) + (?P.+?) + (&|$) + ''' + + _TEST = { + 'url': 'https://player.ooyala.com/player.js?externalId=espn:10365079&pcode=1kNG061cgaoolOncv54OAO1ceO-I&adSetCode=91cDU6NuXTGKz3OdjOxFdAgJVtQcKJnI&callback=handleEvents&hasModuleParams=1&height=968&playerBrandingId=7af3bd04449c444c964f347f11873075&targetReplaceId=videoPlayer&width=1656&wmode=opaque&allowScriptAccess=always', + 'info_dict': { + 'id': 'FkYWtmazr6Ed8xmvILvKLWjd4QvYZpzG', + 'ext': 'mp4', + 'title': 'dm_140128_30for30Shorts___JudgingJewellv2', + 'description': '', + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, + } + + def _real_extract(self, url): + mobj = re.match(self._VALID_URL, url) + partner_id = mobj.group('partner_id') + video_id = mobj.group('id') + pcode = mobj.group('pcode') + player_url = 'http://player.ooyala.com/player.js?externalId=%s:%s&pcode=%s' % (partner_id, video_id, pcode) + return self._extract(player_url, video_id)