[jwplatform] Improved JWPlayer support

[youtube-dl] / youtube_dl / extractor / jwplatform.py

diff --git a/youtube_dl/extractor/jwplatform.py b/youtube_dl/extractor/jwplatform.py
index 60a09044f0593339845f7b4e60f1f721fae22b46..e44e31104f55fc72daa6884f09bcf8faab390568 100644 (file)
--- a/youtube_dl/extractor/jwplatform.py
+++ b/youtube_dl/extractor/jwplatform.py
@@ -4,43 +4,95 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..utils import int_or_none
+from ..utils import (
+    determine_ext,
+    float_or_none,
+    int_or_none,
+)
 
 
 class JWPlatformBaseIE(InfoExtractor):
-    def _parse_jwplayer_data(self, jwplayer_data, video_id):
+    @staticmethod
+    def _find_jwplayer_data(webpage):
+        # TODO: Merge this with JWPlayer-related codes in generic.py
+
+        mobj = re.search(
+            'jwplayer\((?P<quote>[\'"])[^\'" ]+(?P=quote)\)\.setup\((?P<options>[^)]+)\)',
+            webpage)
+        if mobj:
+            return mobj.group('options')
+
+    def _extract_jwplayer_data(self, webpage, video_id, *args, **kwargs):
+        jwplayer_data = self._parse_json(
+            self._find_jwplayer_data(webpage), video_id)
+        return self._parse_jwplayer_data(
+            jwplayer_data, video_id, *args, **kwargs)
+
+    def _parse_jwplayer_data(self, jwplayer_data, video_id, require_title=True, m3u8_id=None, rtmp_params=None):
+        # JWPlayer backward compatibility: flattened playlists
+        # https://github.com/jwplayer/jwplayer/blob/v7.4.3/src/js/api/config.js#L81-L96
+        if 'playlist' not in jwplayer_data:
+            jwplayer_data = {'playlist': [jwplayer_data]}
+
         video_data = jwplayer_data['playlist'][0]
-        subtitles = {}
-        for track in video_data['tracks']:
-            if track['kind'] == 'captions':
-                subtitles[track['label']] = [{'url': self._proto_relative_url(track['file'])}]
+
+        # JWPlayer backward compatibility: flattened sources
+        # https://github.com/jwplayer/jwplayer/blob/v7.4.3/src/js/playlist/item.js#L29-L35
+        if 'sources' not in video_data:
+            video_data['sources'] = [video_data]
 
         formats = []
         for source in video_data['sources']:
             source_url = self._proto_relative_url(source['file'])
             source_type = source.get('type') or ''
-            if source_type == 'application/vnd.apple.mpegurl':
+            if source_type in ('application/vnd.apple.mpegurl', 'hls') or determine_ext(source_url) == 'm3u8':
                 formats.extend(self._extract_m3u8_formats(
-                    source_url, video_id, 'mp4', 'm3u8_native', fatal=False))
+                    source_url, video_id, 'mp4', 'm3u8_native', m3u8_id=m3u8_id, fatal=False))
             elif source_type.startswith('audio'):
                 formats.append({
                     'url': source_url,
                     'vcodec': 'none',
                 })
             else:
-                formats.append({
+                a_format = {
                     'url': source_url,
                     'width': int_or_none(source.get('width')),
                     'height': int_or_none(source.get('height')),
-                })
+                }
+                if source_url.startswith('rtmp'):
+                    a_format['ext'] = 'flv',
+
+                    # See com/longtailvideo/jwplayer/media/RTMPMediaProvider.as
+                    # of jwplayer.flash.swf
+                    rtmp_url_parts = re.split(
+                        r'((?:mp4|mp3|flv):)', source_url, 1)
+                    if len(rtmp_url_parts) == 3:
+                        rtmp_url, prefix, play_path = rtmp_url_parts
+                        a_format.update({
+                            'url': rtmp_url,
+                            'play_path': prefix + play_path,
+                        })
+                    if rtmp_params:
+                        a_format.update(rtmp_params)
+                formats.append(a_format)
         self._sort_formats(formats)
 
+        subtitles = {}
+        tracks = video_data.get('tracks')
+        if tracks and isinstance(tracks, list):
+            for track in tracks:
+                if track.get('file') and track.get('kind') == 'captions':
+                    subtitles.setdefault(track.get('label') or 'en', []).append({
+                        'url': self._proto_relative_url(track['file'])
+                    })
+
         return {
             'id': video_id,
-            'title': video_data['title'],
+            'title': video_data['title'] if require_title else video_data.get('title'),
             'description': video_data.get('description'),
             'thumbnail': self._proto_relative_url(video_data.get('image')),
             'timestamp': int_or_none(video_data.get('pubdate')),
+            'duration': float_or_none(jwplayer_data.get('duration')),
             'subtitles': subtitles,
             'formats': formats,
         }