[extractor/radiko] Fix format sorting for Time Free (#6159)

Authored by: road-master

diff --git a/yt_dlp/extractor/radiko.py b/yt_dlp/extractor/radiko.py
index f10292203479aed6a43def16c4530500a9ee6230..43eecba5f6cd6e34aa903fa5dfb2ce944bc34813 100644 (file)
--- a/yt_dlp/extractor/radiko.py
+++ b/yt_dlp/extractor/radiko.py
@@ -1,5 +1,4 @@
 import base64
-import re
 import urllib.parse
 
 from .common import InfoExtractor
@@ -15,6 +14,23 @@
 
 class RadikoBaseIE(InfoExtractor):
     _FULL_KEY = None
+    _HOSTS_FOR_TIME_FREE_FFMPEG_UNSUPPORTED = (
+        'https://c-rpaa.smartstream.ne.jp',
+        'https://si-c-radiko.smartstream.ne.jp',
+        'https://tf-f-rpaa-radiko.smartstream.ne.jp',
+        'https://tf-c-rpaa-radiko.smartstream.ne.jp',
+        'https://si-f-radiko.smartstream.ne.jp',
+        'https://rpaa.smartstream.ne.jp',
+    )
+    _HOSTS_FOR_TIME_FREE_FFMPEG_SUPPORTED = (
+        'https://rd-wowza-radiko.radiko-cf.com',
+        'https://radiko.jp',
+        'https://f-radiko.smartstream.ne.jp',
+    )
+    # Following URL forcibly connects not Time Free but Live
+    _HOSTS_FOR_LIVE = (
+        'https://c-radiko.smartstream.ne.jp',
+    )
 
     def _auth_client(self):
         _, auth1_handle = self._download_webpage_handle(
@@ -92,9 +108,9 @@ def _extract_formats(self, video_id, station, is_onair, ft, cursor, auth_token,
         formats = []
         found = set()
         for url_tag in m3u8_urls:
-            pcu = url_tag.find('playlist_create_url')
+            pcu = url_tag.find('playlist_create_url').text
             url_attrib = url_tag.attrib
-            playlist_url = update_url_query(pcu.text, {
+            playlist_url = update_url_query(pcu, {
                 'station_id': station,
                 **query,
                 'l': '15',
@@ -117,10 +133,11 @@ def _extract_formats(self, video_id, station, is_onair, ft, cursor, auth_token,
                     'X-Radiko-AreaId': area_id,
                     'X-Radiko-AuthToken': auth_token,
                 })
+            not_preferred = is_onair and not pcu.startswith(self._HOSTS_FOR_LIVE) or (not is_onair and (pcu.startswith(self._HOSTS_FOR_TIME_FREE_FFMPEG_UNSUPPORTED) or pcu.startswith(self._HOSTS_FOR_LIVE)))
             for sf in subformats:
-                if re.fullmatch(r'[cf]-radiko\.smartstream\.ne\.jp', domain):
-                    # Prioritize live radio vs playback based on extractor
-                    sf['preference'] = 100 if is_onair else -100
+                if not_preferred:
+                    sf['preference'] = -100
+                    sf['format_note'] = 'not preferred'
                 if not is_onair and url_attrib['timefree'] == '1' and time_to_skip:
                     sf['downloader_options'] = {'ffmpeg_args': ['-ss', time_to_skip]}
             formats.extend(subformats)