youtube-dl

commit 756f574e4e7160ca5b39c6e18ec5168beb4a8eb1
parent 78294e6a9ce2c9a294d663ac79936df7353b9980
Author: Yen Chi Hsuan <yan12125@gmail.com>
Date:   Sun, 21 Jun 2015 21:30:34 +0800

[dailymotion/generic] Add DailymotionCloudIE

Diffstat:
M
youtube_dl/extractor/__init__.py
 | 
1
+
M
youtube_dl/extractor/dailymotion.py
 | 
42
++++++++++++++++++++++++++++++++++++++++++
M
youtube_dl/extractor/generic.py
 | 
17
+++++++++++++++++

3 files changed, 60 insertions(+), 0 deletions(-)
diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py
@@ -106,6 +106,7 @@ from .dailymotion import (
     DailymotionIE,
     DailymotionPlaylistIE,
     DailymotionUserIE,
+    DailymotionCloudIE,
 )
 from .daum import DaumIE
 from .dbtv import DBTVIE
diff --git a/youtube_dl/extractor/dailymotion.py b/youtube_dl/extractor/dailymotion.py
@@ -251,3 +251,45 @@ class DailymotionUserIE(DailymotionPlaylistIE):
             'title': full_user,
             'entries': self._extract_entries(user),
         }
+
+
+class DailymotionCloudIE(DailymotionBaseInfoExtractor):
+    _VALID_URL = r'http://api\.dmcloud\.net/embed/[^/]+/(?P<id>[^/?]+)'
+
+    _TEST = {
+        # From http://www.francetvinfo.fr/economie/entreprises/les-entreprises-familiales-le-secret-de-la-reussite_933271.html
+        # Tested at FranceTvInfo_2
+        'url': 'http://api.dmcloud.net/embed/4e7343f894a6f677b10006b4/556e03339473995ee145930c?auth=1464865870-0-jyhsm84b-ead4c701fb750cf9367bf4447167a3db&autoplay=1',
+        'only_matching': True,
+    }
+
+    @classmethod
+    def _extract_dmcloud_url(self, webpage):
+        mobj = re.search(r'<iframe[^>]+src=[\'"](http://api\.dmcloud\.net/embed/[^/]+/[^\'"]+)[\'"]', webpage)
+        if mobj:
+            return mobj.group(1)
+
+        mobj = re.search(r'<input[^>]+id=[\'"]dmcloudUrlEmissionSelect[\'"][^>]+value=[\'"](http://api\.dmcloud\.net/embed/[^/]+/[^\'"]+)[\'"]', webpage)
+        if mobj:
+            return mobj.group(1)
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        request = self._build_request(url)
+        webpage = self._download_webpage(request, video_id)
+
+        title = self._html_search_regex(r'<title>([^>]+)</title>', webpage, 'title')
+
+        video_info = self._parse_json(self._search_regex(
+            r'var\s+info\s*=\s*([^;]+);', webpage, 'video info'), video_id)
+
+        # TODO: parse ios_url, which is in fact a manifest
+        video_url = video_info['mp4_url']
+
+        return {
+            'id': video_id,
+            'url': video_url,
+            'title': title,
+            'thumbnail': video_info.get('thumbnail_url'),
+        }
diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
@@ -44,6 +44,7 @@ from .bliptv import BlipTVIE
 from .svt import SVTIE
 from .pornhub import PornHubIE
 from .vimeo import VimeoIE
+from .dailymotion import DailymotionCloudIE
 
 
 class GenericIE(InfoExtractor):
@@ -813,6 +814,17 @@ class GenericIE(InfoExtractor):
                 'description': 'To understand why he was the Toronto Blue Jays’ top off-season priority is to appreciate his background and upbringing in Montreal, where he first developed his baseball skills. Written and narrated by Stephen Brunt.',
                 'uploader': 'Rogers Sportsnet',
             },
+        },
+        # Dailymotion Cloud video
+        {
+            'url': 'http://replay.publicsenat.fr/vod/le-debat/florent-kolandjian,dominique-cena,axel-decourtye,laurence-abeille,bruno-parmentier/175910',
+            'md5': '49444254273501a64675a7e68c502681',
+            'info_dict': {
+                'id': '5585de919473990de4bee11b',
+                'ext': 'mp4',
+                'title': 'Le débat',
+                'thumbnail': 're:^https?://.*\.jpe?g$',
+            }
         }
     ]
 
@@ -1486,6 +1498,11 @@ class GenericIE(InfoExtractor):
         if senate_isvp_url:
             return self.url_result(senate_isvp_url, 'SenateISVP')
 
+        # Look for Dailymotion Cloud videos
+        dmcloud_url = DailymotionCloudIE._extract_dmcloud_url(webpage)
+        if dmcloud_url:
+            return self.url_result(dmcloud_url, 'DailymotionCloud')
+
         def check_video(vurl):
             if YoutubeIE.suitable(vurl):
                 return True