[ie/crunchyroll] Fix stream extraction (#10005)

[yt-dlp.git] / yt_dlp / extractor / substack.py

diff --git a/yt_dlp/extractor/substack.py b/yt_dlp/extractor/substack.py
index 70cf10515c8fd0d57252bf34298daef162032f3e..6ee3f75e1a6cdd0f29cd79b46268382e8194d3f8 100644 (file)
--- a/yt_dlp/extractor/substack.py
+++ b/yt_dlp/extractor/substack.py
@@ -2,7 +2,7 @@
 import urllib.parse
 
 from .common import InfoExtractor
-from ..utils import str_or_none, traverse_obj
+from ..utils import js_to_json, str_or_none, traverse_obj
 
 
 class SubstackIE(InfoExtractor):
@@ -14,7 +14,7 @@ class SubstackIE(InfoExtractor):
             'id': '47660949',
             'ext': 'mp4',
             'title': 'I MADE A VLOG',
-            'description': 'md5:10c01ff93439a62e70ce963b2aa0b7f6',
+            'description': 'md5:9248af9a759321e1027226f988f54d96',
             'thumbnail': 'md5:bec758a34d8ee9142d43bcebdf33af18',
             'uploader': 'Maybe Baby',
             'uploader_id': '33628',
@@ -46,19 +46,20 @@ class SubstackIE(InfoExtractor):
     }]
 
     @classmethod
-    def _extract_url(cls, webpage, url):
+    def _extract_embed_urls(cls, url, webpage):
         if not re.search(r'<script[^>]+src=["\']https://substackcdn.com/[^"\']+\.js', webpage):
             return
 
-        mobj = re.search(r'{[^}]*["\']subdomain["\']\s*:\s*["\'](?P<subdomain>[^"]+)', webpage)
+        mobj = re.search(r'{[^}]*\\?["\']subdomain\\?["\']\s*:\s*\\?["\'](?P<subdomain>[^\\"\']+)', webpage)
         if mobj:
             parsed = urllib.parse.urlparse(url)
-            return parsed._replace(netloc=f'{mobj.group("subdomain")}.substack.com').geturl()
+            yield parsed._replace(netloc=f'{mobj.group("subdomain")}.substack.com').geturl()
+            raise cls.StopExtraction()
 
-    def _extract_video_formats(self, video_id, username):
+    def _extract_video_formats(self, video_id, url):
         formats, subtitles = [], {}
         for video_format in ('hls', 'mp4'):
-            video_url = f'https://{username}.substack.com/api/v1/video/upload/{video_id}/src?type={video_format}'
+            video_url = urllib.parse.urljoin(url, f'/api/v1/video/upload/{video_id}/src?type={video_format}')
 
             if video_format == 'hls':
                 fmts, subs = self._extract_m3u8_formats_and_subtitles(video_url, video_id, 'mp4', fatal=False)
@@ -76,18 +77,24 @@ def _real_extract(self, url):
         display_id, username = self._match_valid_url(url).group('id', 'username')
         webpage = self._download_webpage(url, display_id)
 
-        webpage_info = self._search_json(r'<script[^>]*>\s*window\._preloads\s*=', webpage, 'preloads', display_id)
+        webpage_info = self._parse_json(self._search_json(
+            r'window\._preloads\s*=\s*JSON\.parse\(', webpage, 'json string',
+            display_id, transform_source=js_to_json, contains_pattern=r'"{(?s:.+)}"'), display_id)
+
+        canonical_url = url
+        domain = traverse_obj(webpage_info, ('domainInfo', 'customDomain', {str}))
+        if domain:
+            canonical_url = urllib.parse.urlparse(url)._replace(netloc=domain).geturl()
 
         post_type = webpage_info['post']['type']
         formats, subtitles = [], {}
         if post_type == 'podcast':
             formats, subtitles = [{'url': webpage_info['post']['podcast_url']}], {}
         elif post_type == 'video':
-            formats, subtitles = self._extract_video_formats(webpage_info['post']['videoUpload']['id'], username)
+            formats, subtitles = self._extract_video_formats(webpage_info['post']['videoUpload']['id'], canonical_url)
         else:
             self.raise_no_formats(f'Page type "{post_type}" is not supported')
 
-        self._sort_formats(formats)
         return {
             'id': str(webpage_info['post']['id']),
             'formats': formats,
@@ -97,4 +104,5 @@ def _real_extract(self, url):
             'thumbnail': traverse_obj(webpage_info, ('post', 'cover_image')),
             'uploader': traverse_obj(webpage_info, ('pub', 'name')),
             'uploader_id': str_or_none(traverse_obj(webpage_info, ('post', 'publication_id'))),
+            'webpage_url': canonical_url,
         }