[jwplatform] Parse video_id from JWPlayer data

[youtube-dl] / youtube_dl / extractor / cbc.py

diff --git a/youtube_dl/extractor/cbc.py b/youtube_dl/extractor/cbc.py
index daf237ca8e887a1afe0e1383e45a9e238fba14a2..a87e971406b9a45d665d4bfe14e655c851f5a899 100644 (file)
--- a/youtube_dl/extractor/cbc.py
+++ b/youtube_dl/extractor/cbc.py
@@ -4,9 +4,11 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
+from ..compat import compat_str
 from ..utils import (
     js_to_json,
     smuggle_url,
+    try_get,
 )
 
 
@@ -25,9 +27,24 @@ class CBCIE(InfoExtractor):
             'upload_date': '20160203',
             'uploader': 'CBCC-NEW',
         },
+        'skip': 'Geo-restricted to Canada',
     }, {
-        # with clipId
+        # with clipId, feed available via tpfeed.cbc.ca and feed.theplatform.com
+        'url': 'http://www.cbc.ca/22minutes/videos/22-minutes-update/22-minutes-update-episode-4',
+        'md5': '162adfa070274b144f4fdc3c3b8207db',
+        'info_dict': {
+            'id': '2414435309',
+            'ext': 'mp4',
+            'title': '22 Minutes Update: What Not To Wear Quebec',
+            'description': "This week's latest Canadian top political story is What Not To Wear Quebec.",
+            'upload_date': '20131025',
+            'uploader': 'CBCC-NEW',
+            'timestamp': 1382717907,
+        },
+    }, {
+        # with clipId, feed only available via tpfeed.cbc.ca
         'url': 'http://www.cbc.ca/archives/entry/1978-robin-williams-freestyles-on-90-minutes-live',
+        'md5': '0274a90b51a9b4971fe005c63f592f12',
         'info_dict': {
             'id': '2487345465',
             'ext': 'mp4',
@@ -63,6 +80,7 @@ class CBCIE(InfoExtractor):
                 'uploader': 'CBCC-NEW',
             },
         }],
+        'skip': 'Geo-restricted to Canada',
     }]
 
     @classmethod
@@ -80,9 +98,15 @@ class CBCIE(InfoExtractor):
             media_id = player_info.get('mediaId')
             if not media_id:
                 clip_id = player_info['clipId']
-                media_id = self._download_json(
-                    'http://feed.theplatform.com/f/h9dtGB/punlNGjMlc1F?fields=id&byContent=byReleases%3DbyId%253D' + clip_id,
-                    clip_id)['entries'][0]['id'].split('/')[-1]
+                feed = self._download_json(
+                    'http://tpfeed.cbc.ca/f/ExhSPC/vms_5akSXx4Ng_Zn?byCustomValue={:mpsReleases}{%s}' % clip_id,
+                    clip_id, fatal=False)
+                if feed:
+                    media_id = try_get(feed, lambda x: x['entries'][0]['guid'], compat_str)
+                if not media_id:
+                    media_id = self._download_json(
+                        'http://feed.theplatform.com/f/h9dtGB/punlNGjMlc1F?fields=id&byContent=byReleases%3DbyId%253D' + clip_id,
+                        clip_id)['entries'][0]['id'].split('/')[-1]
             return self.url_result('cbcplayer:%s' % media_id, 'CBCPlayer', media_id)
         else:
             entries = [self.url_result('cbcplayer:%s' % media_id, 'CBCPlayer', media_id) for media_id in re.findall(r'<iframe[^>]+src="[^"]+?mediaId=(\d+)"', webpage)]
@@ -91,8 +115,9 @@ class CBCIE(InfoExtractor):
 
 class CBCPlayerIE(InfoExtractor):
     _VALID_URL = r'(?:cbcplayer:|https?://(?:www\.)?cbc\.ca/(?:player/play/|i/caffeine/syndicate/\?mediaId=))(?P<id>\d+)'
-    _TEST = {
+    _TESTS = [{
         'url': 'http://www.cbc.ca/player/play/2683190193',
+        'md5': '64d25f841ddf4ddb28a235338af32e2c',
         'info_dict': {
             'id': '2683190193',
             'ext': 'mp4',
@@ -102,7 +127,34 @@ class CBCPlayerIE(InfoExtractor):
             'upload_date': '20160210',
             'uploader': 'CBCC-NEW',
         },
-    }
+        'skip': 'Geo-restricted to Canada',
+    }, {
+        # Redirected from http://www.cbc.ca/player/AudioMobile/All%20in%20a%20Weekend%20Montreal/ID/2657632011/
+        'url': 'http://www.cbc.ca/player/play/2657631896',
+        'md5': 'e5e708c34ae6fca156aafe17c43e8b75',
+        'info_dict': {
+            'id': '2657631896',
+            'ext': 'mp3',
+            'title': 'CBC Montreal is organizing its first ever community hackathon!',
+            'description': 'The modern technology we tend to depend on so heavily, is never without it\'s share of hiccups and headaches. Next weekend - CBC Montreal will be getting members of the public for its first Hackathon.',
+            'timestamp': 1425704400,
+            'upload_date': '20150307',
+            'uploader': 'CBCC-NEW',
+        },
+    }, {
+        # available only when we add `formats=MPEG4,FLV,MP3` to theplatform url
+        'url': 'http://www.cbc.ca/player/play/2164402062',
+        'md5': '17a61eb813539abea40618d6323a7f82',
+        'info_dict': {
+            'id': '2164402062',
+            'ext': 'flv',
+            'title': 'Cancer survivor four times over',
+            'description': 'Tim Mayer has beaten three different forms of cancer four times in five years.',
+            'timestamp': 1320410746,
+            'upload_date': '20111104',
+            'uploader': 'CBCC-NEW',
+        },
+    }]
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
@@ -110,7 +162,7 @@ class CBCPlayerIE(InfoExtractor):
             '_type': 'url_transparent',
             'ie_key': 'ThePlatform',
             'url': smuggle_url(
-                'http://link.theplatform.com/s/ExhSPC/media/guid/2655402169/%s?mbr=true' % video_id, {
+                'http://link.theplatform.com/s/ExhSPC/media/guid/2655402169/%s?mbr=true&formats=MPEG4,FLV,MP3' % video_id, {
                     'force_smil_url': True
                 }),
             'id': video_id,