v1 milestone

2026-03-25 11:20:05 +07:00
commit 4ec369c18e
7307 changed files with 873141 additions and 0 deletions
--- a/venv/lib/python3.12/site-packages/yt_dlp/extractor/s4c.py
+++ b/venv/lib/python3.12/site-packages/yt_dlp/extractor/s4c.py
@@ -0,0 +1,104 @@
+from .common import InfoExtractor
+from ..utils import traverse_obj, url_or_none
+
+
+class S4CIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?s4c\.cymru/clic/programme/(?P<id>\d+)'
+    _TESTS = [{
+        'url': 'https://www.s4c.cymru/clic/programme/861362209',
+        'info_dict': {
+            'id': '861362209',
+            'ext': 'mp4',
+            'title': 'Y Swn',
+            'description': 'md5:f7681a30e4955b250b3224aa9fe70cf0',
+            'duration': 5340,
+            'thumbnail': 'https://www.s4c.cymru/amg/1920x1080/Y_Swn_2023S4C_099_ii.jpg',
+        },
+    }, {
+        # Geo restricted to the UK
+        'url': 'https://www.s4c.cymru/clic/programme/886303048',
+        'info_dict': {
+            'id': '886303048',
+            'ext': 'mp4',
+            'title': 'Pennod 1',
+            'description': 'md5:7e3f364b70f61fcdaa8b4cb4a3eb3e7a',
+            'duration': 2880,
+            'thumbnail': 'https://www.s4c.cymru/amg/1920x1080/Stad_2025S4C_P1_210053.jpg',
+        },
+    }]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+        details = self._download_json(
+            f'https://www.s4c.cymru/df/full_prog_details?lang=e&programme_id={video_id}',
+            video_id, fatal=False)
+
+        player_config = self._download_json(
+            'https://player-api.s4c-cdn.co.uk/player-configuration/prod', video_id, query={
+                'programme_id': video_id,
+                'signed': '0',
+                'lang': 'en',
+                'mode': 'od',
+                'appId': 'clic',
+                'streamName': '',
+            }, note='Downloading player config JSON')
+        subtitles = {}
+        for sub in traverse_obj(player_config, ('subtitles', lambda _, v: url_or_none(v['0']))):
+            subtitles.setdefault(sub.get('3', 'en'), []).append({
+                'url': sub['0'],
+                'name': sub.get('1'),
+            })
+        m3u8_url = self._download_json(
+            'https://player-api.s4c-cdn.co.uk/streaming-urls/prod', video_id, query={
+                'mode': 'od',
+                'application': 'clic',
+                'region': 'UK' if player_config.get('application') == 's4chttpl' else 'WW',
+                'extra': 'false',
+                'thirdParty': 'false',
+                'filename': player_config['filename'],
+            }, note='Downloading streaming urls JSON')['hls']
+
+        return {
+            'id': video_id,
+            'formats': self._extract_m3u8_formats(m3u8_url, video_id, 'mp4', m3u8_id='hls'),
+            'subtitles': subtitles,
+            'thumbnail': url_or_none(player_config.get('poster')),
+            **traverse_obj(details, ('full_prog_details', 0, {
+                'title': (('programme_title', 'series_title'), {str}),
+                'description': ('full_billing', {str.strip}),
+                'duration': ('duration', {lambda x: int(x) * 60}),
+            }), get_all=False),
+        }
+
+
+class S4CSeriesIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?s4c\.cymru/clic/series/(?P<id>\d+)'
+    _TESTS = [{
+        'url': 'https://www.s4c.cymru/clic/series/864982911',
+        'playlist_mincount': 6,
+        'info_dict': {
+            'id': '864982911',
+            'title': 'Iaith ar Daith',
+        },
+    }, {
+        'url': 'https://www.s4c.cymru/clic/series/866852587',
+        'playlist_mincount': 8,
+        'info_dict': {
+            'id': '866852587',
+            'title': 'FFIT Cymru',
+        },
+    }]
+
+    def _real_extract(self, url):
+        series_id = self._match_id(url)
+        series_details = self._download_json(
+            'https://www.s4c.cymru/df/series_details', series_id, query={
+                'lang': 'e',
+                'series_id': series_id,
+                'show_prog_in_series': 'Y',
+            }, note='Downloading series details JSON')
+
+        return self.playlist_result(
+            [self.url_result(f'https://www.s4c.cymru/clic/programme/{episode_id}', S4CIE, episode_id)
+             for episode_id in traverse_obj(series_details, ('other_progs_in_series', ..., 'id'))],
+            series_id, traverse_obj(series_details, ('full_prog_details', 0, 'series_title', {str})))