From 83cebb8b7adeff0474b32f12d651837c8353ab59 Mon Sep 17 00:00:00 2001
From: David Wagner <mnem@noiseandheat.com>
Date: Thu, 27 Feb 2014 00:20:34 +0000
Subject: [PATCH] Add support for FLV videos with speaker decks

---
 youtube_dl/extractor/gdcvault.py | 87 +++++++++++++++++++++++---------
 1 file changed, 63 insertions(+), 24 deletions(-)

diff --git a/youtube_dl/extractor/gdcvault.py b/youtube_dl/extractor/gdcvault.py
index a1e4c3221..63527bc85 100644
--- a/youtube_dl/extractor/gdcvault.py
+++ b/youtube_dl/extractor/gdcvault.py
@@ -10,16 +10,26 @@ from ..utils import unified_strdate
 
 class GDCVaultIE(InfoExtractor):
     _VALID_URL = r'https?://(?:www\.)?gdcvault\.com/play/(?P<id>\d+)/(?P<name>(\w|-)+)'
-
-    _TEST = {
-        u'url': u'http://www.gdcvault.com/play/1019721/Doki-Doki-Universe-Sweet-Simple',
-        u'md5': u'7ce8388f544c88b7ac11c7ab1b593704',
-        u'info_dict': {
-            u"id": u"1019721",
-            u"ext": u"mp4",
-            u"title": u"Doki-Doki Universe: Sweet, Simple and Genuine (GDC Next 10)"
-        }
-    }
+    _TESTS = [
+        {
+            u'url': u'http://www.gdcvault.com/play/1015683/Embracing-the-Dark-Art-of',
+            u'md5': u'05763e5edd1a74776999a12b02ee1c4e',
+            u'info_dict': {
+                u"id": u"1015683",
+                u"ext": u"flv",
+                u"title": u"Embracing the Dark Art of Mathematical Modeling in AI"
+            }
+        },
+        {
+            u'url': u'http://www.gdcvault.com/play/1019721/Doki-Doki-Universe-Sweet-Simple',
+            u'md5': u'7ce8388f544c88b7ac11c7ab1b593704',
+            u'info_dict': {
+                u"id": u"1019721",
+                u"ext": u"mp4",
+                u"title": u"Doki-Doki Universe: Sweet, Simple and Genuine (GDC Next 10)"
+            }
+        },
+    ]
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
@@ -32,30 +42,59 @@ class GDCVaultIE(InfoExtractor):
         self.report_extraction(video_id)
 
         xml_root = self._html_search_regex(r'<iframe src="(?P<xml_root>.*?)player.html.*?".*?</iframe>', start_page, 'xml root')
-        xml_name = self._html_search_regex(r'<iframe src=".*?\?xml=(?P<xml_file>.+?\.xml).*?".*?</iframe>', start_page, 'xml filename')
+        xml_name = self._html_search_regex(r'<iframe src=".*?\?xml=(?P<xml_file>.+?\.xml).*?".*?</iframe>', start_page, 'xml filename', None, False)
+        if xml_name is None:
+            # Fallback to the older format
+            xml_name = self._html_search_regex(r'<iframe src=".*?\?xmlURL=xml/(?P<xml_file>.+?\.xml).*?".*?</iframe>', start_page, 'xml filename')
+
         xml_decription_url = xml_root + 'xml/' + xml_name
 
         xml_description = self._download_xml(xml_decription_url, video_id)
 
         video_title = xml_description.find('./metadata/title').text
 
-        mp4_video = xml_description.find('./metadata/mp4video').text
-        mobj = re.match(r'(?P<root>https?://.*?/).*', mp4_video)
-        video_root = mobj.group('root')
-
-        formats = xml_description.findall('./metadata/MBRVideos/MBRVideo')
+        video_details = {
+            'id': video_id,
+            'title': video_title,
+        }
         video_formats = []
-        for format in formats:
-            mobj = re.match(r'mp4\:(?P<path>.*)', format.find('streamName').text)
-            url = video_root + mobj.group('path')
-            vbr = format.find('bitrate').text
+
+        mp4_video = xml_description.find('./metadata/mp4video')
+        if mp4_video is not None:
+            mobj = re.match(r'(?P<root>https?://.*?/).*', mp4_video.text)
+            video_root = mobj.group('root')
+            formats = xml_description.findall('./metadata/MBRVideos/MBRVideo')
+            for format in formats:
+                mobj = re.match(r'mp4\:(?P<path>.*)', format.find('streamName').text)
+                url = video_root + mobj.group('path')
+                vbr = format.find('bitrate').text
+                video_formats.append({
+                    'url': url,
+                    'vbr': int(vbr),
+                })
+            video_details['formats'] = video_formats
+        else:
+            # Fallback to flv
+            akami_url = xml_description.find('./metadata/akamaiHost').text
+            slide_video_path = xml_description.find('./metadata/slideVideo').text
             video_formats.append({
-                'url': url,
-                'vbr': int(vbr),
-            })
+                    'url': 'rtmp://' + akami_url + '/' + slide_video_path,
+                    'format_note': 'slide deck video',
+                    'quality': -2,
+                    'preference': -2,
+                    'format_id': 'slides',
+                })
+            speaker_video_path = xml_description.find('./metadata/speakerVideo').text
+            video_formats.append({
+                    'url': 'rtmp://' + akami_url + '/' + speaker_video_path,
+                    'format_note': 'speaker video',
+                    'quality': -1,
+                    'preference': -1,
+                    'format_id': 'speaker',
+                })
 
         return [{
             'id': video_id,
-            'formats': video_formats,
             'title': video_title,
+            'formats': video_formats,
         }]