debian/changelog: Finalize release.

[youtubedl] / youtube_dl / extractor / theplatform.py

diff --git a/youtube_dl/extractor/theplatform.py b/youtube_dl/extractor/theplatform.py
index 23172143ec41ecc48e88c333dfdd476db330e1b8..b6b2dba9ca9e6ee02c7dc6b2cf01d3601874a6b2 100644 (file)
--- a/youtube_dl/extractor/theplatform.py
+++ b/youtube_dl/extractor/theplatform.py
@@ -1,3 +1,5 @@
+from __future__ import unicode_literals
+

 import re
 import json
 
 import re
 import json
 


@@ -11,34 +13,35 @@ _x = lambda p: xpath_with_ns(p, {'smil': 'http://www.w3.org/2005/SMIL21/Language
 
 
 class ThePlatformIE(InfoExtractor):
 
 
 class ThePlatformIE(InfoExtractor):

-    _VALID_URL = r'(?:https?://link\.theplatform\.com/s/[^/]+/|theplatform:)(?P<id>[^/\?]+)'
+    _VALID_URL = r'''(?x)
+        (?:https?://(?:link|player)\.theplatform\.com/[sp]/[^/]+/
+           (?P<config>(?:[^/\?]+/(?:swf|config)|onsite)/select/)?
+         |theplatform:)(?P<id>[^/\?&]+)'''

 
     _TEST = {
         # from http://www.metacafe.com/watch/cb-e9I_cZgTgIPd/blackberrys_big_bold_z30/
 
     _TEST = {
         # from http://www.metacafe.com/watch/cb-e9I_cZgTgIPd/blackberrys_big_bold_z30/

-        u'url': u'http://link.theplatform.com/s/dJ5BDC/e9I_cZgTgIPd/meta.smil?format=smil&Tracking=true&mbr=true',
-        u'info_dict': {
-            u'id': u'e9I_cZgTgIPd',
-            u'ext': u'flv',
-            u'title': u'Blackberry\'s big, bold Z30',
-            u'description': u'The Z30 is Blackberry\'s biggest, baddest mobile messaging device yet.',
-            u'duration': 247,
+        'url': 'http://link.theplatform.com/s/dJ5BDC/e9I_cZgTgIPd/meta.smil?format=smil&Tracking=true&mbr=true',
+        'info_dict': {
+            'id': 'e9I_cZgTgIPd',
+            'ext': 'flv',
+            'title': 'Blackberry\'s big, bold Z30',
+            'description': 'The Z30 is Blackberry\'s biggest, baddest mobile messaging device yet.',
+            'duration': 247,

         },
         },

-        u'params': {
+        'params': {

             # rtmp download
             # rtmp download

-            u'skip_download': True,
+            'skip_download': True,

         },
     }
 
         },
     }
 

-    def _get_info(self, video_id):
-        smil_url = ('http://link.theplatform.com/s/dJ5BDC/{0}/meta.smil?'
-            'format=smil&mbr=true'.format(video_id))
+    def _get_info(self, video_id, smil_url):

         meta = self._download_xml(smil_url, video_id)
 
         try:
             error_msg = next(
                 n.attrib['abstract']
                 for n in meta.findall(_x('.//smil:ref'))
         meta = self._download_xml(smil_url, video_id)
 
         try:
             error_msg = next(
                 n.attrib['abstract']
                 for n in meta.findall(_x('.//smil:ref'))

-                if n.attrib.get('title') == u'Geographic Restriction')
+                if n.attrib.get('title') == 'Geographic Restriction')

         except StopIteration:
             pass
         else:
         except StopIteration:
             pass
         else:


@@ -50,26 +53,39 @@ class ThePlatformIE(InfoExtractor):
 
         head = meta.find(_x('smil:head'))
         body = meta.find(_x('smil:body'))
 
         head = meta.find(_x('smil:head'))
         body = meta.find(_x('smil:body'))

-        base_url = head.find(_x('smil:meta')).attrib['base']
-        switch = body.find(_x('smil:switch'))
-        formats = []
-        for f in switch.findall(_x('smil:video')):
-            attr = f.attrib
-            width = int(attr['width'])
-            height = int(attr['height'])
-            vbr = int(attr['system-bitrate']) // 1000
-            format_id = '%dx%d_%dk' % (width, height, vbr)
-            formats.append({
-                'format_id': format_id,
-                'url': base_url,
-                'play_path': 'mp4:' + attr['src'],
-                'ext': 'flv',
-                'width': width,
-                'height': height,
-                'vbr': vbr,
-            })

 
 

-        self._sort_formats(formats)
+        f4m_node = body.find(_x('smil:seq//smil:video'))
+        if f4m_node is not None:
+            f4m_url = f4m_node.attrib['src']
+            if 'manifest.f4m?' not in f4m_url:
+                f4m_url += '?'
+            # the parameters are from syfy.com, other sites may use others,
+            # they also work for nbc.com
+            f4m_url += '&g=UXWGVKRWHFSP&hdcore=3.0.3'
+            formats = [{
+                'ext': 'flv',
+                'url': f4m_url,
+            }]
+        else:
+            base_url = head.find(_x('smil:meta')).attrib['base']
+            switch = body.find(_x('smil:switch'))
+            formats = []
+            for f in switch.findall(_x('smil:video')):
+                attr = f.attrib
+                width = int(attr['width'])
+                height = int(attr['height'])
+                vbr = int(attr['system-bitrate']) // 1000
+                format_id = '%dx%d_%dk' % (width, height, vbr)
+                formats.append({
+                    'format_id': format_id,
+                    'url': base_url,
+                    'play_path': 'mp4:' + attr['src'],
+                    'ext': 'flv',
+                    'width': width,
+                    'height': height,
+                    'vbr': vbr,
+                })
+            self._sort_formats(formats)

 
         return {
             'id': video_id,
 
         return {
             'id': video_id,


@@ -83,4 +99,13 @@ class ThePlatformIE(InfoExtractor):
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('id')
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('id')

-        return self._get_info(video_id)
+        if mobj.group('config'):
+            config_url = url+ '&form=json'
+            config_url = config_url.replace('swf/', 'config/')
+            config_url = config_url.replace('onsite/', 'onsite/config/')
+            config = self._download_json(config_url, video_id, 'Downloading config')
+            smil_url = config['releaseUrl'] + '&format=SMIL&formats=MPEG4&manifest=f4m'
+        else:
+            smil_url = ('http://link.theplatform.com/s/dJ5BDC/{0}/meta.smil?'
+                'format=smil&mbr=true'.format(video_id))
+        return self._get_info(video_id, smil_url)