Merge branch 'master' of https://github.com/yt-dlp/yt-dlp into ytdlp

* 'master' of https://github.com/yt-dlp/yt-dlp: [extractor/kompas] Add extractor (#4562) [extractor/vidio] Support embed link (#4564)
ytdl-patched · Aug 5, 2022 · 94852b0 · 94852b0
2 parents baefdf3 + d380fc1
commit 94852b0
Show file tree

Hide file tree

Showing 3 changed files with 97 additions and 3 deletions.
diff --git a/yt_dlp/extractor/_extractors.py b/yt_dlp/extractor/_extractors.py
@@ -775,6 +775,7 @@
 from .kickstarter import KickStarterIE
 from .kinja import KinjaEmbedIE
 from .kinopoisk import KinoPoiskIE
+from .kompas import KompasVideoIE
 from .konserthusetplay import KonserthusetPlayIE
 from .koo import KooIE
 from .kth import KTHIE

diff --git a/yt_dlp/extractor/kompas.py b/yt_dlp/extractor/kompas.py
@@ -0,0 +1,68 @@
+from .common import InfoExtractor
+from ..utils import (
+    clean_html,
+    float_or_none,
+    traverse_obj,
+    try_call,
+)
+
+# Video from www.kompas.tv and video.kompas.com seems use jixie player
+# see [1] https://jixie.atlassian.net/servicedesk/customer/portal/2/article/1339654214?src=-1456335525,
+# [2] https://scripts.jixie.media/jxvideo.3.1.min.js for more info
+
+
+class KompasVideoIE(InfoExtractor):
+    _VALID_URL = r'https?://video\.kompas\.com/\w+/(?P<id>\d+)/(?P<slug>[\w-]+)'
+    _TESTS = [{
+        'url': 'https://video.kompas.com/watch/164474/kim-jong-un-siap-kirim-nuklir-lawan-as-dan-korsel',
+        'info_dict': {
+            'id': '164474',
+            'ext': 'mp4',
+            'title': 'Kim Jong Un Siap Kirim Nuklir Lawan AS dan Korsel',
+            'description': 'md5:262530c4fb7462398235f9a5dba92456',
+            'uploader_id': '9262bf2590d558736cac4fff7978fcb1',
+            'display_id': 'kim-jong-un-siap-kirim-nuklir-lawan-as-dan-korsel',
+            'duration': 85.066667,
+            'categories': ['news'],
+            'thumbnail': 'https://video.jixie.media/1001/164474/164474_1280x720.jpg',
+            'tags': 'count:9',
+        }
+    }]
+
+    def _real_extract(self, url):
+        video_id, display_id = self._match_valid_url(url).group('id', 'slug')
+        webpage = self._download_webpage(url, display_id)
+
+        json_data = self._download_json(
+            'https://apidam.jixie.io/api/public/stream', display_id,
+            query={'metadata': 'full', 'video_id': video_id})['data']
+
+        formats, subtitles = [], {}
+        for stream in json_data['streams']:
+            if stream.get('type') == 'HLS':
+                fmt, sub = self._extract_m3u8_formats_and_subtitles(stream.get('url'), display_id, ext='mp4')
+                formats.extend(fmt)
+                self._merge_subtitles(sub, target=subtitles)
+            else:
+                formats.append({
+                    'url': stream.get('url'),
+                    'width': stream.get('width'),
+                    'height': stream.get('height'),
+                    'ext': 'mp4',
+                })
+
+        self._sort_formats(formats)
+        return {
+            'id': video_id,
+            'display_id': display_id,
+            'formats': formats,
+            'subtitles': subtitles,
+            'title': json_data.get('title') or self._html_search_meta(['og:title', 'twitter:title'], webpage),
+            'description': (clean_html(traverse_obj(json_data, ('metadata', 'description')))
+                            or self._html_search_meta(['description', 'og:description', 'twitter:description'], webpage)),
+            'thumbnails': traverse_obj(json_data, ('metadata', 'thumbnails')),
+            'duration': float_or_none(traverse_obj(json_data, ('metadata', 'duration'))),
+            'tags': try_call(lambda: json_data['metadata']['keywords'].split(',')),
+            'categories': try_call(lambda: json_data['metadata']['categories'].split(',')),
+            'uploader_id': json_data.get('owner_id'),
+        }
diff --git a/yt_dlp/extractor/vidio.py b/yt_dlp/extractor/vidio.py
@@ -67,10 +67,10 @@ def _call_api(self, url, video_id, note=None):
 
 
 class VidioIE(VidioBaseIE):
-    _VALID_URL = r'https?://(?:www\.)?vidio\.com/watch/(?P<id>\d+)-(?P<display_id>[^/?#&]+)'
+    _VALID_URL = r'https?://(?:www\.)?vidio\.com/(watch|embed)/(?P<id>\d+)-(?P<display_id>[^/?#&]+)'
     _TESTS = [{
         'url': 'http://www.vidio.com/watch/165683-dj_ambred-booyah-live-2015',
-        'md5': 'cd2801394afc164e9775db6a140b91fe',
+        'md5': 'abac81b1a205a8d94c609a473b5ea62a',
         'info_dict': {
             'id': '165683',
             'display_id': 'dj_ambred-booyah-live-2015',
@@ -89,7 +89,8 @@ class VidioIE(VidioBaseIE):
             'view_count': int,
             'dislike_count': int,
             'comment_count': int,
-            'tags': 'count:4',
+            'tags': 'count:3',
+            'uploader_url': 'https://www.vidio.com/@twelvepictures',
         },
     }, {
         'url': 'https://www.vidio.com/watch/77949-south-korea-test-fires-missile-that-can-strike-all-of-the-north',
@@ -98,6 +99,30 @@ class VidioIE(VidioBaseIE):
         # Premier-exclusive video
         'url': 'https://www.vidio.com/watch/1550718-stand-by-me-doraemon',
         'only_matching': True
+    }, {
+        # embed url from https://enamplus.liputan6.com/read/5033648/video-fakta-temuan-suspek-cacar-monyet-di-jawa-tengah
+        'url': 'https://www.vidio.com/embed/7115874-fakta-temuan-suspek-cacar-monyet-di-jawa-tengah',
+        'info_dict': {
+            'id': '7115874',
+            'ext': 'mp4',
+            'channel_id': '40172876',
+            'comment_count': int,
+            'uploader_id': 'liputan6',
+            'view_count': int,
+            'dislike_count': int,
+            'upload_date': '20220804',
+            'uploader': 'Liputan6.com',
+            'display_id': 'fakta-temuan-suspek-cacar-monyet-di-jawa-tengah',
+            'channel': 'ENAM PLUS 165',
+            'timestamp': 1659605520,
+            'title': 'Fakta Temuan Suspek Cacar Monyet di Jawa Tengah',
+            'duration': 59,
+            'like_count': int,
+            'tags': ['monkeypox indonesia', 'cacar monyet menyebar', 'suspek cacar monyet di indonesia', 'fakta', 'hoax atau bukan?', 'jawa tengah'],
+            'thumbnail': 'https://thumbor.prod.vidiocdn.com/83PN-_BKm5sS7emLtRxl506MLqQ=/640x360/filters:quality(70)/vidio-web-prod-video/uploads/video/image/7115874/fakta-suspek-cacar-monyet-di-jawa-tengah-24555a.jpg',
+            'uploader_url': 'https://www.vidio.com/@liputan6',
+            'description': 'md5:6d595a18d3b19ee378e335a6f288d5ac',
+        },
     }]
 
     def _real_extract(self, url):