[ie/theatercomplextown] Add extractors (#8560)

Closes #8491
Authored by: bashonly

yt_dlp/extractor/_extractors.py

@@ -1911,6 +1911,8 @@
 from .stacommu import (
     StacommuLiveIE,
     StacommuVODIE,
+    TheaterComplexTownVODIE,
+    TheaterComplexTownPPVIE,
 )
 from .stanfordoc import StanfordOpenClassroomIE
 from .startv import StarTVIE

yt_dlp/extractor/stacommu.py

@@ -38,9 +38,48 @@ def _extract_hls_key(self, data, path, decrypt):
             return None
         return traverse_obj(encryption_data, {'key': ('key', {decrypt}), 'iv': ('iv', {decrypt})})
 
+    def _extract_vod(self, url):
+        video_id = self._match_id(url)
+        video_info = self._download_metadata(
+            url, video_id, 'ja', ('dehydratedState', 'queries', 0, 'state', 'data'))
+        hls_info, decrypt = self._call_encrypted_api(
+            video_id, ':watch', 'stream information', data={'method': 1})
+
+        return {
+            'id': video_id,
+            'formats': self._get_formats(hls_info, ('protocolHls', 'url', {url_or_none}), video_id),
+            'hls_aes': self._extract_hls_key(hls_info, 'protocolHls', decrypt),
+            **traverse_obj(video_info, {
+                'title': ('displayName', {str}),
+                'description': ('description', {str}),
+                'timestamp': ('watchStartTime', {int_or_none}),
+                'thumbnail': ('keyVisualUrl', {url_or_none}),
+                'cast': ('casts', ..., 'displayName', {str}),
+                'duration': ('duration', {int}),
+            }),
+        }
+
+    def _extract_ppv(self, url):
+        video_id = self._match_id(url)
+        video_info = self._call_api(video_id, msg='video information', query={'al': 'ja'}, auth=False)
+        hls_info, decrypt = self._call_encrypted_api(
+            video_id, ':watchArchive', 'stream information', data={'method': 1})
+
+        return {
+            'id': video_id,
+            'formats': self._get_formats(hls_info, ('hls', 'urls', ..., {url_or_none}), video_id),
+            'hls_aes': self._extract_hls_key(hls_info, 'hls', decrypt),
+            **traverse_obj(video_info, {
+                'title': ('displayName', {str}),
+                'timestamp': ('startTime', {int_or_none}),
+                'thumbnail': ('keyVisualUrl', {url_or_none}),
+                'duration': ('duration', {int_or_none}),
+            }),
+        }
+
 
 class StacommuVODIE(StacommuBaseIE):
-    _VALID_URL = r'https?://www\.stacommu\.jp/videos/episodes/(?P<id>[\da-zA-Z]+)'
+    _VALID_URL = r'https?://www\.stacommu\.jp/(?:en/)?videos/episodes/(?P<id>[\da-zA-Z]+)'
     _TESTS = [{
         # not encrypted
         'url': 'https://www.stacommu.jp/videos/episodes/aXcVKjHyAENEjard61soZZ',
@@ -79,34 +118,19 @@ class StacommuVODIE(StacommuBaseIE):
         'params': {
             'skip_download': 'm3u8',
         },
+    }, {
+        'url': 'https://www.stacommu.jp/en/videos/episodes/aXcVKjHyAENEjard61soZZ',
+        'only_matching': True,
     }]
 
     _API_PATH = 'videoEpisodes'
 
     def _real_extract(self, url):
-        video_id = self._match_id(url)
+        return self._extract_vod(url)
-        video_info = self._download_metadata(
-            url, video_id, 'ja', ('dehydratedState', 'queries', 0, 'state', 'data'))
-        hls_info, decrypt = self._call_encrypted_api(
-            video_id, ':watch', 'stream information', data={'method': 1})
-
-        return {
-            'id': video_id,
-            'formats': self._get_formats(hls_info, ('protocolHls', 'url', {url_or_none}), video_id),
-            'hls_aes': self._extract_hls_key(hls_info, 'protocolHls', decrypt),
-            **traverse_obj(video_info, {
-                'title': ('displayName', {str}),
-                'description': ('description', {str}),
-                'timestamp': ('watchStartTime', {int_or_none}),
-                'thumbnail': ('keyVisualUrl', {url_or_none}),
-                'cast': ('casts', ..., 'displayName', {str}),
-                'duration': ('duration', {int}),
-            }),
-        }
 
 
 class StacommuLiveIE(StacommuBaseIE):
-    _VALID_URL = r'https?://www\.stacommu\.jp/live/(?P<id>[\da-zA-Z]+)'
+    _VALID_URL = r'https?://www\.stacommu\.jp/(?:en/)?live/(?P<id>[\da-zA-Z]+)'
     _TESTS = [{
         'url': 'https://www.stacommu.jp/live/d2FJ3zLnndegZJCAEzGM3m',
         'info_dict': {
@@ -125,24 +149,83 @@ class StacommuLiveIE(StacommuBaseIE):
         'params': {
             'skip_download': 'm3u8',
         },
+    }, {
+        'url': 'https://www.stacommu.jp/en/live/d2FJ3zLnndegZJCAEzGM3m',
+        'only_matching': True,
     }]
 
     _API_PATH = 'events'
 
     def _real_extract(self, url):
-        video_id = self._match_id(url)
+        return self._extract_ppv(url)
-        video_info = self._call_api(video_id, msg='video information', query={'al': 'ja'}, auth=False)
-        hls_info, decrypt = self._call_encrypted_api(
-            video_id, ':watchArchive', 'stream information', data={'method': 1})
 
-        return {
+
-            'id': video_id,
+class TheaterComplexTownBaseIE(StacommuBaseIE):
-            'formats': self._get_formats(hls_info, ('hls', 'urls', ..., {url_or_none}), video_id),
+    _NETRC_MACHINE = 'theatercomplextown'
-            'hls_aes': self._extract_hls_key(hls_info, 'hls', decrypt),
+    _API_HOST = 'api.theater-complex.town'
-            **traverse_obj(video_info, {
+    _LOGIN_QUERY = {'key': 'AIzaSyAgNCqToaIz4a062EeIrkhI_xetVfAOrfc'}
-                'title': ('displayName', {str}),
+    _LOGIN_HEADERS = {
-                'timestamp': ('startTime', {int_or_none}),
+        'Accept': '*/*',
-                'thumbnail': ('keyVisualUrl', {url_or_none}),
+        'Content-Type': 'application/json',
-                'duration': ('duration', {int_or_none}),
+        'X-Client-Version': 'Chrome/JsCore/9.23.0/FirebaseCore-web',
-            }),
+        'Referer': 'https://www.theater-complex.town/',
+        'Origin': 'https://www.theater-complex.town',
     }
+
+
+class TheaterComplexTownVODIE(TheaterComplexTownBaseIE):
+    _VALID_URL = r'https?://(?:www\.)?theater-complex\.town/(?:en/)?videos/episodes/(?P<id>\w+)'
+    IE_NAME = 'theatercomplextown:vod'
+    _TESTS = [{
+        'url': 'https://www.theater-complex.town/videos/episodes/hoxqidYNoAn7bP92DN6p78',
+        'info_dict': {
+            'id': 'hoxqidYNoAn7bP92DN6p78',
+            'ext': 'mp4',
+            'title': '演劇ドラフトグランプリ2023　劇団『恋のぼり』〜劇団名決定秘話ラジオ',
+            'description': 'md5:a7e2e9cf570379ea67fb630f345ff65d',
+            'cast': ['玉城 裕規', '石川 凌雅'],
+            'thumbnail': 'https://image.theater-complex.town/5URnXX6KCeDysuFrPkP38o/5URnXX6KCeDysuFrPkP38o',
+            'upload_date': '20231103',
+            'timestamp': 1699016400,
+            'duration': 868,
+        },
+        'params': {
+            'skip_download': 'm3u8',
+        },
+    }, {
+        'url': 'https://www.theater-complex.town/en/videos/episodes/6QT7XYwM9dJz5Gf9VB6K5y',
+        'only_matching': True,
+    }]
+
+    _API_PATH = 'videoEpisodes'
+
+    def _real_extract(self, url):
+        return self._extract_vod(url)
+
+
+class TheaterComplexTownPPVIE(TheaterComplexTownBaseIE):
+    _VALID_URL = r'https?://(?:www\.)?theater-complex\.town/(?:en/)?ppv/(?P<id>\w+)'
+    IE_NAME = 'theatercomplextown:ppv'
+    _TESTS = [{
+        'url': 'https://www.theater-complex.town/ppv/wytW3X7khrjJBUpKuV3jen',
+        'info_dict': {
+            'id': 'wytW3X7khrjJBUpKuV3jen',
+            'ext': 'mp4',
+            'title': 'BREAK FREE STARS　11月5日（日）12:30千秋楽公演',
+            'thumbnail': 'https://image.theater-complex.town/5GWEB31JcTUfjtgdeV5t6o/5GWEB31JcTUfjtgdeV5t6o',
+            'upload_date': '20231105',
+            'timestamp': 1699155000,
+            'duration': 8378,
+        },
+        'params': {
+            'skip_download': 'm3u8',
+        },
+    }, {
+        'url': 'https://www.theater-complex.town/en/ppv/wytW3X7khrjJBUpKuV3jen',
+        'only_matching': True,
+    }]
+
+    _API_PATH = 'events'
+
+    def _real_extract(self, url):
+        return self._extract_ppv(url)