test/extractor/test_arte.py

@@ -0,0 +1,20 @@
+import pytest
+
+from yt_dlp.extractor.arte import ArteTVIE
+
+
+ACCESSIBLE_TESTS = [
+    ({'fr': [{'url': 'https://arte-cmafhls.akamaized.net/am/cmaf/103000/103500/103522-000-A/230804153206/medias/103522-000-A_st_VF-FRA.m3u8', 'ext': 'vtt', 'protocol': 'm3u8_native'}]}, 'fr'),
+    ({'fr': [{'url': 'https://arte-cmafhls.akamaized.net/am/cmaf/103000/103500/103522-000-A/230804153206/medias/103522-000-A_st_VF-MAL.m3u8', 'ext': 'vtt', 'protocol': 'm3u8_native'}]}, 'fr-acc'),
+]
+
+
+@pytest.mark.parametrize('original_subs,expected_locale', ACCESSIBLE_TESTS)
+def test_extract_accessible_subtitles(original_subs, expected_locale):
+    extractor = ArteTVIE()
+
+    subs = extractor._contvert_accessible_subs_locale(original_subs)
+
+    assert len(subs) == 1
+    assert expected_locale in subs
+    assert subs[expected_locale] == original_subs['fr']

yt_dlp/extractor/arte.py

@@ -16,6 +16,7 @@ from ..utils import (
 class ArteTVBaseIE(InfoExtractor):
     _ARTE_LANGUAGES = 'fr|de|en|es|it|pl'
     _API_BASE = 'https://api.arte.tv/api/player/v2'
+    _ARTE_ACCESSIBLE_SUBS_SUFFIX = '-MAL.m3u8'
 
 
 class ArteTVIE(ArteTVBaseIE):
@@ -70,24 +71,7 @@ class ArteTVIE(ArteTVBaseIE):
             'thumbnail': 'https://api-cdn.arte.tv/img/v2/image/q82dTTfyuCXupPsGxXsd7B/940x530',
             'upload_date': '20230930',
             'ext': 'mp4',
-        },
+        }
-    }, {
-        'url': 'https://www.arte.tv/de/videos/085374-003-A/im-hohen-norden-geboren/',
-        'info_dict': {
-            'id': '085374-003-A',
-            'ext': 'mp4',
-            'description': 'md5:ab79ec7cc472a93164415b4e4916abf9',
-            'timestamp': 1702872000,
-            'thumbnail': 'https://api-cdn.arte.tv/img/v2/image/TnyHBfPxv3v2GEY3suXGZP/940x530',
-            'duration': 2594,
-            'title': 'Die kurze Zeit der Jugend',
-            'alt_title': 'Im hohen Norden geboren',
-            'upload_date': '20231218',
-            'subtitles': {
-                'fr': 'mincount:1',
-                'fr-acc': 'mincount:1',
-            },
-        },
     }]
 
     _GEO_BYPASS = True
@@ -138,15 +122,23 @@ class ArteTVIE(ArteTVBaseIE):
         ),
     }
 
-    @staticmethod
+    def _contvert_accessible_subs_locale(self, subs):
-    def _fix_accessible_subs_locale(subs):
+        """Gives a different locale code to accessible subttiles.
-        updated_subs = {}
+
-        for lang, sub_formats in subs.items():
+        Some videos have multiple subtitles for the same language. A normal
-            for format in sub_formats:
+        subtitle and some accessible subtitles. This method will transform
-                if format.get('url', '').endswith('-MAL.m3u8'):
+        the language code for accessible subtitles by appending a '-acc' suffix
-                    lang += '-acc'
+        to its language code.
-                updated_subs.setdefault(lang, []).append(format)
+        """
-        return updated_subs
+        extracted_subs = {}
+
+        for locale, sub_content in subs.items():
+            if any(item.get('url', '').endswith(self._ARTE_ACCESSIBLE_SUBS_SUFFIX) for item in sub_content):
+                locale += '-acc'
+
+            extracted_subs[locale] = sub_content
+
+        return extracted_subs
 
     def _real_extract(self, url):
         mobj = self._match_valid_url(url)
@@ -192,6 +184,7 @@ class ArteTVIE(ArteTVBaseIE):
             if 'HLS' in stream['protocol']:
                 fmts, subs = self._extract_m3u8_formats_and_subtitles(
                     stream['url'], video_id=video_id, ext='mp4', m3u8_id=stream_version_code, fatal=False)
+                subs = self._contvert_accessible_subs_locale(subs)
                 for fmt in fmts:
                     fmt.update({
                         'format_note': f'{stream_version.get("label", "unknown")} [{short_label}]',
@@ -201,7 +194,6 @@ class ArteTVIE(ArteTVBaseIE):
                     secondary_formats.extend(fmts)
                 else:
                     formats.extend(fmts)
-                subs = self._fix_accessible_subs_locale(subs)
                 self._merge_subtitles(subs, target=subtitles)
 
             elif stream['protocol'] in ('HTTPS', 'RTMP'):