[vimeo] improve format extraction and sorting(closes #25285)

youtube_dl/extractor/vimeo.py

@@ -140,28 +140,28 @@ class VimeoBaseInfoExtractor(InfoExtractor):
             })
 
         # TODO: fix handling of 308 status code returned for live archive manifest requests
+        sep_pattern = r'/sep/video/'
         for files_type in ('hls', 'dash'):
             for cdn_name, cdn_data in config_files.get(files_type, {}).get('cdns', {}).items():
                 manifest_url = cdn_data.get('url')
                 if not manifest_url:
                     continue
                 format_id = '%s-%s' % (files_type, cdn_name)
-                if files_type == 'hls':
+                sep_manifest_urls = []
-                    formats.extend(self._extract_m3u8_formats(
+                if re.search(sep_pattern, manifest_url):
-                        manifest_url, video_id, 'mp4',
+                    for suffix, repl in (('', 'video'), ('_sep', 'sep/video')):
-                        'm3u8' if is_live else 'm3u8_native', m3u8_id=format_id,
+                        sep_manifest_urls.append((format_id + suffix, re.sub(
-                        note='Downloading %s m3u8 information' % cdn_name,
+                            sep_pattern, '/%s/' % repl, manifest_url)))
-                        fatal=False))
+                else:
-                elif files_type == 'dash':
+                    sep_manifest_urls = [(format_id, manifest_url)]
-                    mpd_pattern = r'/%s/(?:sep/)?video/' % video_id
+                for f_id, m_url in sep_manifest_urls:
-                    mpd_manifest_urls = []
+                    if files_type == 'hls':
-                    if re.search(mpd_pattern, manifest_url):
+                        formats.extend(self._extract_m3u8_formats(
-                        for suffix, repl in (('', 'video'), ('_sep', 'sep/video')):
+                            m_url, video_id, 'mp4',
-                            mpd_manifest_urls.append((format_id + suffix, re.sub(
+                            'm3u8' if is_live else 'm3u8_native', m3u8_id=f_id,
-                                mpd_pattern, '/%s/%s/' % (video_id, repl), manifest_url)))
+                            note='Downloading %s m3u8 information' % cdn_name,
-                    else:
+                            fatal=False))
-                        mpd_manifest_urls = [(format_id, manifest_url)]
+                    elif files_type == 'dash':
-                    for f_id, m_url in mpd_manifest_urls:
                         if 'json=1' in m_url:
                             real_m_url = (self._download_json(m_url, video_id, fatal=False) or {}).get('url')
                             if real_m_url:
@@ -170,11 +170,6 @@ class VimeoBaseInfoExtractor(InfoExtractor):
                             m_url.replace('/master.json', '/master.mpd'), video_id, f_id,
                             'Downloading %s MPD information' % cdn_name,
                             fatal=False)
-                        for f in mpd_formats:
-                            if f.get('vcodec') == 'none':
-                                f['preference'] = -50
-                            elif f.get('acodec') == 'none':
-                                f['preference'] = -40
                         formats.extend(mpd_formats)
 
         live_archive = live_event.get('archive') or {}
@@ -186,6 +181,12 @@ class VimeoBaseInfoExtractor(InfoExtractor):
                 'preference': 1,
             })
 
+        for f in formats:
+            if f.get('vcodec') == 'none':
+                f['preference'] = -50
+            elif f.get('acodec') == 'none':
+                f['preference'] = -40
+
         subtitles = {}
         text_tracks = config['request'].get('text_tracks')
         if text_tracks: