Merge pull request #1 from dabiboo/universalmusicfrance

Universalmusicfrance

youtube_dl/extractor/__init__.py

@@ -687,6 +687,7 @@ from .udemy import (
 )
 from .udn import UDNEmbedIE
 from .ultimedia import UltimediaIE
+from .universalmusicfrance import UniversalMusicFranceIE
 from .unistra import UnistraIE
 from .urort import UrortIE
 from .ustream import UstreamIE, UstreamChannelIE

youtube_dl/extractor/universalmusicfrance.py

@@ -0,0 +1,72 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+from .common import InfoExtractor
+
+from ..compat import (
+    compat_urllib_request,
+)
+
+from ..utils import (
+    urlencode_postdata,
+)
+
+
+class UniversalMusicFranceIE(InfoExtractor):
+    _VALID_URL = r'https?://www\.universalmusic\.fr/artiste/.*/videos/(?P<id>.*)#?'
+    _TESTS = [
+        {
+            'url': 'http://www.universalmusic.fr/artiste/7415-anna-bergendahl/videos/4555-for-you-remix-lyric-video.iframe',
+            'md5': '159cda7568b9fc1e5e3de6aeca5d4bfc)',
+            'info_dict': {
+                'id': '1881-waiting-for-love-lyric-video',
+                'ext': 'mp4',
+                'title': '1881-waiting-for-love-lyric-video'
+            }
+        }
+        ,
+        {
+            'url': 'https://www.universalmusic.fr/artiste/4428-avicii/videos/1881-waiting-for-love-lyric-video#contentPart',
+            'md5': '159cda7568b9fc1e5e3de6aeca5d4bfc)',
+            'info_dict': {
+                'id': '1881-waiting-for-love-lyric-video',
+                'ext': 'mp4',
+                'title': '1881-waiting-for-love-lyric-video'
+            }
+        }
+        ,
+        {
+            # from http://www.wat.tv/video/anna-bergendahl-for-you-2015-7dvjn_76lkz_.html
+            'url': 'http://www.universalmusic.fr/artiste/7415-anna-bergendahl/videos/4555-for-you-remix-lyric-video',
+            'md5': '159cda7568b9fc1e5e3de6aeca5d4bfc)',
+            'info_dict': {
+                'id': '4555-for-you-remix-lyric-video',
+                'ext': 'mp4',
+                'title': 'anna-bergendahl - for-you'
+            }
+        }
+    ]
+    GET_TOKEN_URL = 'http://www.universalmusic.fr/_/artiste/video/token'
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+        webpage = self._download_webpage(url, video_id)
+
+        urlVideo = self._html_search_regex(r'var urlVideo = \'(.*)\';', webpage, 'urlVideo')
+        title = self._html_search_regex(r'<meta property="?og:title"? content="(.*)"/>', webpage, 'title')
+
+        request = compat_urllib_request.Request(self.GET_TOKEN_URL, urlencode_postdata({'videoUrl': urlVideo}))
+        request.add_header('Content-Type', 'application/x-www-form-urlencoded; charset=UTF-8')
+        request.add_header('X-Requested-With', 'XMLHttpRequest')
+        manifest_json = self._download_webpage(request, None, note='Getting token', errnote='unable to get token')
+
+        manifestUrl = self._parse_json(manifest_json, video_id).get("video")
+        print(manifestUrl);
+        return {
+            'id': video_id,
+            'title': title,
+            'description': title,
+            'formats':
+                self._extract_m3u8_formats(
+                    manifestUrl, video_id, 'mp4')
+        }

youtube_dl/extractor/wat.py

@@ -3,6 +3,8 @@ from __future__ import unicode_literals
 
 import re
 import hashlib
+import universalmusicfrance
+import ultimedia
 
 from .common import InfoExtractor
 from ..utils import (
@@ -15,6 +17,29 @@ class WatIE(InfoExtractor):
     _VALID_URL = r'http://www\.wat\.tv/video/(?P<display_id>.*)-(?P<short_id>.*?)_.*?\.html'
     IE_NAME = 'wat.tv'
     _TESTS = [
+        {
+            'url': 'http://www.wat.tv/video/anna-bergendahl-for-you-2015-7dvjn_76lkz_.html',
+            'md5': '159cda7568b9fc1e5e3de6aeca5d4bfc',
+            'info_dict': {
+                'id': '4555-for-you-remix-lyric-video',
+                'display_id': '4555-for-you-remix-lyric-video',
+                'ext': 'mp4',
+                'title': 'For You - Anna Bergendahl - Universal Music France',
+                'description': 'md5:1bbdde8d44751f43367ba68e8b9966a6'
+            },
+        },
+        {
+            'url': 'http://www.wat.tv/video/david-guetta-titanium-feat-sia-4v6p5_4v69t_.html',
+            'md5': '5c31a70358cd5019595297a26390cd46',
+            'info_dict': {
+                'id': 'qzkfx3',
+                'display_id': 'qzkfx3',
+                'ext': 'mp4',
+                'title': 'David Guetta - Titanium feat. Sia (Clip)',
+                'description': 'md5:bb28f8c4a84586e2eb1c3d092ab94f4b',
+                'upload_date': '20111220'
+            },
+        },
         {
             'url': 'http://www.wat.tv/video/soupe-figues-l-orange-aux-epices-6z1uz_2hvf7_.html',
             'md5': 'ce70e9223945ed26a8056d413ca55dc9',
@@ -57,6 +82,18 @@ class WatIE(InfoExtractor):
         short_id = mobj.group('short_id')
         display_id = mobj.group('display_id')
         webpage = self._download_webpage(url, display_id or short_id)
+        srcIFrame = self._html_search_regex(r'<iframe .* src="(.*?)(.iframe)?"', webpage, 'srcIFrame')
+        if (srcIFrame.__contains__("universalmusic")):
+            #return universalmusicfrance.UniversalMusicFranceIE()._real_extract(srcIFrame);
+            print srcIFrame
+            return self.url_result(srcIFrame)
+        elif (srcIFrame.__contains__("ultimedia")):
+            mobj = re.match(r'http://www\.ultimedia\.com/deliver/musique/iframe/mdtk/[0-9]*/zone/[0-9]/article/(?P<article_id>.*?)/.*', srcIFrame)
+            article_id = mobj.group('article_id')
+            ultimedia_url = "http://www.ultimedia.com/default/index/videomusic/id/" + article_id
+            #return ultimedia.UltimediaIE()._real_extract(ultimedia_url)
+            return self.url_result(ultimedia_url)
+        else:
             real_id = self._search_regex(r'xtpage = ".*-(.*?)";', webpage, 'real id')
 
             video_info = self.download_video_info(real_id)