[discovery] Rewrite DiscoveryIE (fixes #5898)

Discovery.com now uses a completely different approach for serving
videos. At least in both test cases brightcove are involved. However,
AMF support is necessary for these brightcove videos. As a result, I
try to extract videos from the info page ('?flat=1'). The downloaded
file can be different from the one in browsers.

diff --git a/youtube_dl/extractor/discovery.py b/youtube_dl/extractor/discovery.py
index d3e6675283cddcb8f6a6dfffbfbd1e1ea3da11bc..d6723ecf26ea67356b288df6e5f3bf612141b91a 100644 (file)
--- a/youtube_dl/extractor/discovery.py
+++ b/youtube_dl/extractor/discovery.py
@@ -2,19 +2,19 @@ from __future__ import unicode_literals
 
 from .common import InfoExtractor
 from ..utils import (
+    parse_duration,
     parse_iso8601,
-    int_or_none,
 )
+from ..compat import compat_str
 
 
 class DiscoveryIE(InfoExtractor):
     _VALID_URL = r'http://www\.discovery\.com\/[a-zA-Z0-9\-]*/[a-zA-Z0-9\-]*/videos/(?P<id>[a-zA-Z0-9_\-]*)(?:\.htm)?'
-    _TEST = {
+    _TESTS = [{
         'url': 'http://www.discovery.com/tv-shows/mythbusters/videos/mission-impossible-outtakes.htm',
-        'md5': '3c69d77d9b0d82bfd5e5932a60f26504',
         'info_dict': {
-            'id': 'mission-impossible-outtakes',
-            'ext': 'flv',
+            'id': '20769',
+            'ext': 'mp4',
             'title': 'Mission Impossible Outtakes',
             'description': ('Watch Jamie Hyneman and Adam Savage practice being'
                             ' each other -- to the point of confusing Jamie\'s dog -- and '
@@ -24,22 +24,36 @@ class DiscoveryIE(InfoExtractor):
             'timestamp': 1303099200,
             'upload_date': '20110418',
         },
-    }
+        'params': {
+            'skip_download': True,  # requires ffmpeg
+        }
+    }, {
+        'url': 'http://www.discovery.com/tv-shows/mythbusters/videos/mythbusters-the-simpsons',
+        'info_dict': {
+            'id': 'mythbusters-the-simpsons',
+            'title': 'MythBusters: The Simpsons',
+        },
+        'playlist_count': 9,
+    }]
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
-        webpage = self._download_webpage(url, video_id)
+        info = self._download_json(url + '?flat=1', video_id)
 
-        info = self._parse_json(self._search_regex(
-            r'(?s)<script type="application/ld\+json">(.*?)</script>',
-            webpage, 'video info'), video_id)
+        video_title = info.get('playlist_title') or info.get('video_title')
 
-        return {
-            'id': video_id,
-            'title': info['name'],
-            'url': info['contentURL'],
-            'description': info.get('description'),
-            'thumbnail': info.get('thumbnailUrl'),
-            'timestamp': parse_iso8601(info.get('uploadDate')),
-            'duration': int_or_none(info.get('duration')),
-        }
+        entries = [{
+            'id': compat_str(video_info['id']),
+            'formats': self._extract_m3u8_formats(
+                video_info['src'], video_id, ext='mp4',
+                note='Download m3u8 information for video %d' % (idx + 1)),
+            'title': video_info['title'],
+            'description': video_info.get('description'),
+            'duration': parse_duration(video_info.get('video_length')),
+            'webpage_url': video_info.get('href'),
+            'thumbnail': video_info.get('thumbnailURL'),
+            'alt_title': video_info.get('secondary_title'),
+            'timestamp': parse_iso8601(video_info.get('publishedDate')),
+        } for idx, video_info in enumerate(info['playlist'])]
+
+        return self.playlist_result(entries, video_id, video_title)