[extractor/tempo] Add IVXPlayer extractor (#5837)

Authored by: HobbyistDev

yt_dlp/extractor/_extractors.py

@@ -1858,7 +1858,7 @@ from .telequebec import (
 )
 from .teletask import TeleTaskIE
 from .telewebion import TelewebionIE
-from .tempo import TempoIE
+from .tempo import TempoIE, IVXPlayerIE
 from .tencent import (
     IflixEpisodeIE,
     IflixSeriesIE,

yt_dlp/extractor/tempo.py

@@ -1,5 +1,81 @@
+import re
+
 from .common import InfoExtractor
-from ..utils import int_or_none, parse_iso8601, str_or_none, traverse_obj
+from ..utils import (
+    int_or_none,
+    parse_iso8601,
+    traverse_obj,
+    try_call
+)
+
+
+class IVXPlayerIE(InfoExtractor):
+    _VALID_URL = r'ivxplayer:(?P<video_id>\d+):(?P<player_key>\w+)'
+    _TESTS = [{
+        'url': 'ivxplayer:2366065:4a89dfe6bc8f002596b1dfbd600730b1',
+        'info_dict': {
+            'id': '2366065',
+            'ext': 'mp4',
+            'duration': 112,
+            'upload_date': '20221204',
+            'title': 'Film Indonesia di Disney Content Showcase Asia Pacific 2022',
+            'timestamp': 1670151746,
+            'thumbnail': 'https://ivx-image.ivideosmart.com/serve/image/video/2366065?width=300'
+        }
+    }]
+    _WEBPAGE_TESTS = [{
+        'url': 'https://www.cantika.com/video/31737/film-indonesia-di-disney-content-showcase-asia-pacific-2022',
+        'info_dict': {
+            'id': '2374200',
+            'ext': 'mp4',
+            'duration': 110,
+            'title': 'Serial Indonesia di Disney Content Showcase Asia Pacific 2022',
+            'timestamp': 1670639416,
+            'upload_date': '20221210',
+            'thumbnail': 'https://ivx-image.ivideosmart.com/serve/image/video/2374200?width=300'
+        }
+    }, {
+        'url': 'https://www.gooto.com/video/11437/wuling-suv-ramai-dikunjungi-di-giias-2018',
+        'info_dict': {
+            'id': '892109',
+            'ext': 'mp4',
+            'title': 'Wuling SUV Ramai Dikunjungi di GIIAS 2018',
+            'upload_date': '20180811',
+            'description': 'md5:6d901483d0aacc664aecb4489719aafa',
+            'duration': 75,
+            'timestamp': 1534011263,
+            'thumbnail': 'https://ivx-image.ivideosmart.com/serve/image/video/892109?width=300'
+        }
+    }]
+
+    @classmethod
+    def _extract_embed_urls(cls, url, webpage):
+        # more info at https://player.ivideosmart.com/ivsplayer/v4/dist/js/loader.js
+        mobj = re.search(
+            r'<ivs-player\s*[^>]+data-ivs-key\s*=\s*"(?P<player_key>[\w]+)\s*[^>]+\bdata-ivs-vid="(?P<video_id>[\w-]+)',
+            webpage)
+        if mobj:
+            yield f'ivxplayer:{mobj.group("video_id")}:{mobj.group("player_key")}'
+            raise cls.StopExtraction()
+
+    def _real_extract(self, url):
+        video_id, player_key = self._match_valid_url(url).group('video_id', 'player_key')
+        json_data = self._download_json(
+            f'https://ivxplayer.ivideosmart.com/prod/video/{video_id}?key={player_key}', video_id)
+
+        formats, subtitles = self._extract_m3u8_formats_and_subtitles(
+            json_data['player']['video_url'], video_id)
+
+        return {
+            'id': str(json_data['ivx']['id']),
+            'title': traverse_obj(json_data, ('ivx', 'name')),
+            'description': traverse_obj(json_data, ('ivx', 'description')),
+            'duration': int_or_none(traverse_obj(json_data, ('ivx', 'duration'))),
+            'timestamp': parse_iso8601(traverse_obj(json_data, ('ivx', 'published_at'))),
+            'formats': formats,
+            'subtitles': subtitles,
+            'thumbnail': traverse_obj(json_data, ('ivx', 'thumbnail_url'))
+        }
 
 
 class TempoIE(InfoExtractor):
@@ -7,14 +83,14 @@ class TempoIE(InfoExtractor):
     _TESTS = [{
         'url': 'https://video.tempo.co/read/30058/anies-baswedan-ajukan-banding-putusan-ptun-batalkan-ump-dki',
         'info_dict': {
-            'id': '2144438',
+            'id': '2144275',
+            'display_id': 'anies-baswedan-ajukan-banding-putusan-ptun-batalkan-ump-dki',
             'ext': 'mp4',
             'title': 'Anies Baswedan Ajukan Banding Putusan PTUN Batalkan UMP DKI',
-            'display_id': 'anies-baswedan-ajukan-banding-putusan-ptun-batalkan-ump-dki',
+            'duration': 85,
-            'duration': 84,
             'description': 'md5:a6822b7c4c874fa7e5bd63e96a387b66',
             'thumbnail': 'https://statik.tempo.co/data/2022/07/27/id_1128287/1128287_720.jpg',
-            'timestamp': 1658911277,
+            'timestamp': 1658907970,
             'upload_date': '20220727',
             'tags': ['Anies Baswedan', ' PTUN', ' PTUN | Pengadilan Tata Usaha Negara', ' PTUN Batalkan UMP DKI', ' UMP DKI'],
         }
@@ -24,30 +100,15 @@ class TempoIE(InfoExtractor):
         display_id = self._match_id(url)
         webpage = self._download_webpage(url, display_id)
 
-        player_key, widget_id = self._search_regex(
+        _, video_id, player_key = next(IVXPlayerIE._extract_embed_urls(url, webpage)).split(':')
-            r'<ivs-player\s*[^>]+data-ivs-key\s*=\s*"(?P<player_key>[\w]+)[^>]+\bdata-ivs-wid="(?P<widget_id>[\w-]+)',
-            webpage, 'player_key, widget_id', group=('player_key', 'widget_id'))
 
         json_ld_data = self._search_json_ld(webpage, display_id)
 
-        json_data = self._download_json(
+        return self.url_result(
-            f'https://ivxplayer.ivideosmart.com/prod/widget/{widget_id}',
+            f'ivxplayer:{video_id}:{player_key}', display_id=display_id,
-            display_id, query={'key': player_key})
+            thumbnail=self._html_search_meta('twitter:image:src', webpage) or self._og_search_thumbnail(webpage),
-        formats, subtitles = self._extract_m3u8_formats_and_subtitles(
+            tags=try_call(lambda: self._html_search_meta('keywords', webpage).split(',')),
-            json_data['player']['video_url'], display_id, ext='mp4')
+            description=(json_ld_data.get('description')
-
+                         or self._html_search_meta(('description', 'twitter:description'), webpage)
-        return {
+                         or self._og_search_description(webpage)),
-            'id': str(json_data['ivx']['id']),
+            url_transparent=True)
-            'display_id': display_id,
-            'formats': formats,
-            'subtitles': subtitles,
-            'title': (self._html_search_meta('twitter:title', webpage) or self._og_search_title(webpage)
-                      or traverse_obj(json_data, ('ivx', 'name'))),
-            'duration': int_or_none(traverse_obj(json_data, ('ivx', 'duration'))),
-            'thumbnail': (self._html_search_meta('twitter:image:src', webpage) or self._og_search_thumbnail(webpage)
-                          or traverse_obj(json_data, ('ivx', 'thumbnail_url'))),
-            'description': (json_ld_data.get('description') or self._html_search_meta(['description', 'twitter:description'], webpage)
-                            or self._og_search_description(webpage)),
-            'timestamp': parse_iso8601(traverse_obj(json_data, ('ivx', 'created_at'))),
-            'tags': str_or_none(self._html_search_meta('keywords', webpage), '').split(','),
-        }