Imported Upstream version 2016.08.17

[youtubedl] / youtube_dl / extractor / brightcove.py

diff --git a/youtube_dl/extractor/brightcove.py b/youtube_dl/extractor/brightcove.py
index ef560b592792910c2d39174eb2db129cea5590fe..aeb22be168402fd4e7e134c164339db74f95a0f8 100644 (file)
--- a/youtube_dl/extractor/brightcove.py
+++ b/youtube_dl/extractor/brightcove.py
@@ -26,6 +26,8 @@ from ..utils import (
     unescapeHTML,
     unsmuggle_url,
     update_url_query,
     unescapeHTML,
     unsmuggle_url,
     update_url_query,

+    clean_html,
+    mimetype2ext,

 )
 
 
 )
 
 


@@ -90,6 +92,7 @@ class BrightcoveLegacyIE(InfoExtractor):
                 'description': 'md5:363109c02998fee92ec02211bd8000df',
                 'uploader': 'National Ballet of Canada',
             },
                 'description': 'md5:363109c02998fee92ec02211bd8000df',
                 'uploader': 'National Ballet of Canada',
             },

+            'skip': 'Video gone',

         },
         {
             # test flv videos served by akamaihd.net
         },
         {
             # test flv videos served by akamaihd.net


@@ -108,7 +111,7 @@ class BrightcoveLegacyIE(InfoExtractor):
             },
         },
         {
             },
         },
         {

-            # playlist test
+            # playlist with 'videoList'

             # from http://support.brightcove.com/en/video-cloud/docs/playlist-support-single-video-players
             'url': 'http://c.brightcove.com/services/viewer/htmlFederated?playerID=3550052898001&playerKey=AQ%7E%7E%2CAAABmA9XpXk%7E%2C-Kp7jNgisre1fG5OdqpAFUTcs0lP_ZoL',
             'info_dict': {
             # from http://support.brightcove.com/en/video-cloud/docs/playlist-support-single-video-players
             'url': 'http://c.brightcove.com/services/viewer/htmlFederated?playerID=3550052898001&playerKey=AQ%7E%7E%2CAAABmA9XpXk%7E%2C-Kp7jNgisre1fG5OdqpAFUTcs0lP_ZoL',
             'info_dict': {


@@ -117,6 +120,15 @@ class BrightcoveLegacyIE(InfoExtractor):
             },
             'playlist_mincount': 7,
         },
             },
             'playlist_mincount': 7,
         },

+        {
+            # playlist with 'playlistTab' (https://github.com/rg3/youtube-dl/issues/9965)
+            'url': 'http://c.brightcove.com/services/json/experience/runtime/?command=get_programming_for_experience&playerKey=AQ%7E%7E,AAABXlLMdok%7E,NJ4EoMlZ4rZdx9eU1rkMVd8EaYPBBUlg',
+            'info_dict': {
+                'id': '1522758701001',
+                'title': 'Lesson 08',
+            },
+            'playlist_mincount': 10,
+        },

     ]
     FLV_VCODECS = {
         1: 'SORENSON',
     ]
     FLV_VCODECS = {
         1: 'SORENSON',


@@ -298,13 +310,19 @@ class BrightcoveLegacyIE(InfoExtractor):
             info_url, player_key, 'Downloading playlist information')
 
         json_data = json.loads(playlist_info)
             info_url, player_key, 'Downloading playlist information')
 
         json_data = json.loads(playlist_info)

-        if 'videoList' not in json_data:
+        if 'videoList' in json_data:
+            playlist_info = json_data['videoList']
+            playlist_dto = playlist_info['mediaCollectionDTO']
+        elif 'playlistTabs' in json_data:
+            playlist_info = json_data['playlistTabs']
+            playlist_dto = playlist_info['lineupListDTO']['playlistDTOs'][0]
+        else:

             raise ExtractorError('Empty playlist')
             raise ExtractorError('Empty playlist')

-        playlist_info = json_data['videoList']
-        videos = [self._extract_video_info(video_info) for video_info in playlist_info['mediaCollectionDTO']['videoDTOs']]
+
+        videos = [self._extract_video_info(video_info) for video_info in playlist_dto['videoDTOs']]

 
         return self.playlist_result(videos, playlist_id='%s' % playlist_info['id'],
 
         return self.playlist_result(videos, playlist_id='%s' % playlist_info['id'],

-                                    playlist_title=playlist_info['mediaCollectionDTO']['displayName'])
+                                    playlist_title=playlist_dto['displayName'])

 
     def _extract_video_info(self, video_info):
         video_id = compat_str(video_info['id'])
 
     def _extract_video_info(self, video_info):
         video_id = compat_str(video_info['id'])


@@ -528,14 +546,16 @@ class BrightcoveNewIE(InfoExtractor):
         formats = []
         for source in json_data.get('sources', []):
             container = source.get('container')
         formats = []
         for source in json_data.get('sources', []):
             container = source.get('container')

-            source_type = source.get('type')
+            ext = mimetype2ext(source.get('type'))

             src = source.get('src')
             src = source.get('src')

-            if source_type == 'application/x-mpegURL' or container == 'M2TS':
+            if ext == 'ism':
+                continue
+            elif ext == 'm3u8' or container == 'M2TS':

                 if not src:
                     continue
                 formats.extend(self._extract_m3u8_formats(
                     src, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
                 if not src:
                     continue
                 formats.extend(self._extract_m3u8_formats(
                     src, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))

-            elif source_type == 'application/dash+xml':
+            elif ext == 'mpd':

                 if not src:
                     continue
                 formats.extend(self._extract_mpd_formats(src, video_id, 'dash', fatal=False))
                 if not src:
                     continue
                 formats.extend(self._extract_mpd_formats(src, video_id, 'dash', fatal=False))


@@ -551,7 +571,7 @@ class BrightcoveNewIE(InfoExtractor):
                     'tbr': tbr,
                     'filesize': int_or_none(source.get('size')),
                     'container': container,
                     'tbr': tbr,
                     'filesize': int_or_none(source.get('size')),
                     'container': container,

-                    'ext': container.lower(),
+                    'ext': ext or container.lower(),

                 }
                 if width == 0 and height == 0:
                     f.update({
                 }
                 if width == 0 and height == 0:
                     f.update({


@@ -585,6 +605,13 @@ class BrightcoveNewIE(InfoExtractor):
                         'format_id': build_format_id('rtmp'),
                     })
                 formats.append(f)
                         'format_id': build_format_id('rtmp'),
                     })
                 formats.append(f)

+
+        errors = json_data.get('errors')
+        if not formats and errors:
+            error = errors[0]
+            raise ExtractorError(
+                error.get('message') or error.get('error_subcode') or error['error_code'], expected=True)
+

         self._sort_formats(formats)
 
         subtitles = {}
         self._sort_formats(formats)
 
         subtitles = {}


@@ -597,7 +624,7 @@ class BrightcoveNewIE(InfoExtractor):
         return {
             'id': video_id,
             'title': title,
         return {
             'id': video_id,
             'title': title,

-            'description': json_data.get('description'),
+            'description': clean_html(json_data.get('description')),

             'thumbnail': json_data.get('thumbnail') or json_data.get('poster'),
             'duration': float_or_none(json_data.get('duration'), 1000),
             'timestamp': parse_iso8601(json_data.get('published_at')),
             'thumbnail': json_data.get('thumbnail') or json_data.get('poster'),
             'duration': float_or_none(json_data.get('duration'), 1000),
             'timestamp': parse_iso8601(json_data.get('published_at')),