[ie/rai] Fix m3u8 formats extraction (#9291)

Closes #887
Authored by: nixxo

diff --git a/yt_dlp/extractor/rai.py b/yt_dlp/extractor/rai.py
index f6219c2dbd9a01ab421ee64de10371717bc1af32..c1fc65c81fd95a35106f533471939b45106ef60d 100644 (file)
--- a/yt_dlp/extractor/rai.py
+++ b/yt_dlp/extractor/rai.py
@@ -28,6 +28,29 @@ class RaiBaseIE(InfoExtractor):
     _GEO_COUNTRIES = ['IT']
     _GEO_BYPASS = False
 
+    def _fix_m3u8_formats(self, media_url, video_id):
+        fmts = self._extract_m3u8_formats(
+            media_url, video_id, 'mp4', m3u8_id='hls', fatal=False)
+
+        # Fix malformed m3u8 manifests by setting audio-only/video-only formats
+        for f in fmts:
+            if not f.get('acodec'):
+                f['acodec'] = 'mp4a'
+            if not f.get('vcodec'):
+                f['vcodec'] = 'avc1'
+            man_url = f['url']
+            if re.search(r'chunklist(?:_b\d+)*_ao[_.]', man_url):  # audio only
+                f['vcodec'] = 'none'
+            elif re.search(r'chunklist(?:_b\d+)*_vo[_.]', man_url):  # video only
+                f['acodec'] = 'none'
+            else:  # video+audio
+                if f['acodec'] == 'none':
+                    f['acodec'] = 'mp4a'
+                if f['vcodec'] == 'none':
+                    f['vcodec'] = 'avc1'
+
+        return fmts
+
     def _extract_relinker_info(self, relinker_url, video_id, audio_only=False):
         def fix_cdata(s):
             # remove \r\n\t before and after <![CDATA[ ]]> to avoid
@@ -69,8 +92,7 @@ def fix_cdata(s):
                 'format_id': 'https-mp3',
             })
         elif ext == 'm3u8' or 'format=m3u8' in media_url:
-            formats.extend(self._extract_m3u8_formats(
-                media_url, video_id, 'mp4', m3u8_id='hls', fatal=False))
+            formats.extend(self._fix_m3u8_formats(media_url, video_id))
         elif ext == 'f4m':
             # very likely no longer needed. Cannot find any url that uses it.
             manifest_url = update_url_query(
@@ -153,10 +175,10 @@ def get_format_info(tbr):
                 'format_id': f'https-{tbr}',
                 'width': format_copy.get('width'),
                 'height': format_copy.get('height'),
-                'tbr': format_copy.get('tbr'),
-                'vcodec': format_copy.get('vcodec'),
-                'acodec': format_copy.get('acodec'),
-                'fps': format_copy.get('fps'),
+                'tbr': format_copy.get('tbr') or tbr,
+                'vcodec': format_copy.get('vcodec') or 'avc1',
+                'acodec': format_copy.get('acodec') or 'mp4a',
+                'fps': format_copy.get('fps') or 25,
             } if format_copy else {
                 'format_id': f'https-{tbr}',
                 'width': _QUALITY[tbr][0],
@@ -245,7 +267,7 @@ class RaiPlayIE(RaiBaseIE):
             'series': 'Report',
             'season': '2013/14',
             'subtitles': {'it': 'count:4'},
-            'release_year': 2022,
+            'release_year': 2024,
             'episode': 'Espresso nel caffè - 07/04/2014',
             'timestamp': 1396919880,
             'upload_date': '20140408',
@@ -253,7 +275,7 @@ class RaiPlayIE(RaiBaseIE):
         },
         'params': {'skip_download': True},
     }, {
-        # 1080p direct mp4 url
+        # 1080p
         'url': 'https://www.raiplay.it/video/2021/11/Blanca-S1E1-Senza-occhi-b1255a4a-8e72-4a2f-b9f3-fc1308e00736.html',
         'md5': 'aeda7243115380b2dd5e881fd42d949a',
         'info_dict': {
@@ -274,7 +296,7 @@ class RaiPlayIE(RaiBaseIE):
             'episode': 'Senza occhi',
             'timestamp': 1637318940,
             'upload_date': '20211119',
-            'formats': 'count:12',
+            'formats': 'count:7',
         },
         'params': {'skip_download': True},
         'expected_warnings': ['Video not available. Likely due to geo-restriction.']
@@ -527,7 +549,7 @@ class RaiPlaySoundPlaylistIE(InfoExtractor):
         'info_dict': {
             'id': 'ilruggitodelconiglio',
             'title': 'Il Ruggito del Coniglio',
-            'description': 'md5:48cff6972435964284614d70474132e6',
+            'description': 'md5:62a627b3a2d0635d08fa8b6e0a04f27e',
         },
         'playlist_mincount': 65,
     }, {
@@ -634,19 +656,20 @@ def _real_extract(self, url):
         }
 
 
-class RaiNewsIE(RaiIE):  # XXX: Do not subclass from concrete IE
+class RaiNewsIE(RaiBaseIE):
     _VALID_URL = rf'https?://(www\.)?rainews\.it/(?!articoli)[^?#]+-(?P<id>{RaiBaseIE._UUID_RE})(?:-[^/?#]+)?\.html'
     _EMBED_REGEX = [rf'<iframe[^>]+data-src="(?P<url>/iframe/[^?#]+?{RaiBaseIE._UUID_RE}\.html)']
     _TESTS = [{
         # new rainews player (#3911)
-        'url': 'https://www.rainews.it/rubriche/24mm/video/2022/05/24mm-del-29052022-12cf645d-1ffd-4220-b27c-07c226dbdecf.html',
+        'url': 'https://www.rainews.it/video/2024/02/membri-della-croce-rossa-evacuano-gli-abitanti-di-un-villaggio-nella-regione-ucraina-di-kharkiv-il-filmato-dallucraina--31e8017c-845c-43f5-9c48-245b43c3a079.html',
         'info_dict': {
-            'id': '12cf645d-1ffd-4220-b27c-07c226dbdecf',
+            'id': '31e8017c-845c-43f5-9c48-245b43c3a079',
             'ext': 'mp4',
-            'title': 'Puntata del 29/05/2022',
-            'duration': 1589,
-            'upload_date': '20220529',
+            'title': 'md5:1e81364b09de4a149042bac3c7d36f0b',
+            'duration': 196,
+            'upload_date': '20240225',
             'uploader': 'rainews',
+            'formats': 'count:2',
         },
         'params': {'skip_download': True},
     }, {
@@ -659,7 +682,8 @@ class RaiNewsIE(RaiIE):  # XXX: Do not subclass from concrete IE
             'description': 'I film in uscita questa settimana.',
             'thumbnail': r're:^https?://.*\.png$',
             'duration': 833,
-            'upload_date': '20161103'
+            'upload_date': '20161103',
+            'formats': 'count:8',
         },
         'params': {'skip_download': True},
         'expected_warnings': ['unable to extract player_data'],
@@ -684,7 +708,7 @@ def _real_extract(self, url):
         if not relinker_url:
             # fallback on old implementation for some old content
             try:
-                return self._extract_from_content_id(video_id, url)
+                return RaiIE._real_extract(self, url)
             except GeoRestrictedError:
                 raise
             except ExtractorError as e: