[screenwavemedia] fix info extraction (fixes #6270)

Closes #6330.

youtube_dl/extractor/screenwavemedia.py

@@ -1,12 +1,11 @@
 # encoding: utf-8
 from __future__ import unicode_literals
 
-import re
-
 from .common import InfoExtractor
 from ..utils import (
     int_or_none,
     unified_strdate,
+    js_to_json,
 )
 
 
@@ -22,59 +21,48 @@ class ScreenwaveMediaIE(InfoExtractor):
         video_id = self._match_id(url)
 
         playerdata = self._download_webpage(
-            'http://player.screenwavemedia.com/play/player.php?id=%s' % video_id,
+            'http://player.screenwavemedia.com/player.php?id=%s' % video_id,
             video_id, 'Downloading player webpage')
 
         vidtitle = self._search_regex(
             r'\'vidtitle\'\s*:\s*"([^"]+)"', playerdata, 'vidtitle').replace('\\/', '/')
-        vidurl = self._search_regex(
-            r'\'vidurl\'\s*:\s*"([^"]+)"', playerdata, 'vidurl').replace('\\/', '/')
-
-        videolist_url = None
 
-        mobj = re.search(r"'videoserver'\s*:\s*'(?P<videoserver>[^']+)'", playerdata)
+        playerconfig = self._download_webpage(
-        if mobj:
+            'http://player.screenwavemedia.com/player.js',
-            videoserver = mobj.group('videoserver')
+            video_id, 'Downloading playerconfig webpage')
-            mobj = re.search(r'\'vidid\'\s*:\s*"(?P<vidid>[^\']+)"', playerdata)
+
-            vidid = mobj.group('vidid') if mobj else video_id
+        videoserver = self._search_regex(r"'videoserver'\s*:\s*'([^']+)", playerconfig, 'videoserver')
-            videolist_url = 'http://%s/vod/smil:%s.smil/jwplayer.smil' % (videoserver, vidid)
+
-        else:
+        sources = self._parse_json(
-            mobj = re.search(r"file\s*:\s*'(?P<smil>http.+?/jwplayer\.smil)'", playerdata)
+            js_to_json(
-            if mobj:
+                self._search_regex(
-                videolist_url = mobj.group('smil')
+                    r"sources\s*:\s*(\[[^\]]+?\])", playerconfig,
+                    'sources',
+                ).replace(
+                    "' + thisObj.options.videoserver + '",
+                    videoserver
+                ).replace(
+                    "' + playerVidId + '",
+                    video_id
+                )
+            ),
+            video_id
+        )
 
-        if videolist_url:
-            videolist = self._download_xml(videolist_url, video_id, 'Downloading videolist XML')
         formats = []
-            baseurl = vidurl[:vidurl.rfind('/') + 1]
+        for source in sources:
-            for video in videolist.findall('.//video'):
+            if source['type'] == 'hls':
-                src = video.get('src')
+                formats.extend(self._extract_m3u8_formats(source['file'], video_id))
-                if not src:
-                    continue
-                file_ = src.partition(':')[-1]
-                width = int_or_none(video.get('width'))
-                height = int_or_none(video.get('height'))
-                bitrate = int_or_none(video.get('system-bitrate'), scale=1000)
-                format = {
-                    'url': baseurl + file_,
-                    'format_id': src.rpartition('.')[0].rpartition('_')[-1],
-                }
-                if width or height:
-                    format.update({
-                        'tbr': bitrate,
-                        'width': width,
-                        'height': height,
-                    })
             else:
-                    format.update({
+                format_label = source.get('label')
-                        'abr': bitrate,
+                height = int_or_none(self._search_regex(
-                        'vcodec': 'none',
+                    r'^(\d+)[pP]', format_label, 'height', default=None))
+                formats.append({
+                    'url': source['file'],
+                    'format': format_label,
+                    'ext': source.get('type'),
+                    'height': height,
                 })
-                formats.append(format)
-        else:
-            formats = [{
-                'url': vidurl,
-            }]
         self._sort_formats(formats)
 
         return {