From 464c919ea82aefdf35f138a1ab2dd0bb8fb7fd0e Mon Sep 17 00:00:00 2001
From: bashonly <88596187+bashonly@users.noreply.github.com>
Date: Sat, 24 Feb 2024 17:13:26 -0600
Subject: [PATCH] [ie/CloudflareStream] Improve embed detection (#9287)

Partially addresses #7858
Authored by: bashonly
---
 yt_dlp/extractor/cloudflarestream.py | 32 ++++++++++++++++++----------
 1 file changed, 21 insertions(+), 11 deletions(-)

diff --git a/yt_dlp/extractor/cloudflarestream.py b/yt_dlp/extractor/cloudflarestream.py
index 0c5f4fb40..a812c24af 100644
--- a/yt_dlp/extractor/cloudflarestream.py
+++ b/yt_dlp/extractor/cloudflarestream.py
@@ -4,27 +4,25 @@ from .common import InfoExtractor
 
 
 class CloudflareStreamIE(InfoExtractor):
+    _SUBDOMAIN_RE = r'(?:(?:watch|iframe|customer-\w+)\.)?'
     _DOMAIN_RE = r'(?:cloudflarestream\.com|(?:videodelivery|bytehighway)\.net)'
-    _EMBED_RE = r'embed\.%s/embed/[^/]+\.js\?.*?\bvideo=' % _DOMAIN_RE
+    _EMBED_RE = rf'embed\.{_DOMAIN_RE}/embed/[^/]+\.js\?.*?\bvideo='
     _ID_RE = r'[\da-f]{32}|[\w-]+\.[\w-]+\.[\w-]+'
-    _VALID_URL = r'''(?x)
-                    https?://
-                        (?:
-                            (?:[\w-]+\.)?%s/|
-                            %s
-                        )
-                        (?P<id>%s)
-                    ''' % (_DOMAIN_RE, _EMBED_RE, _ID_RE)
-    _EMBED_REGEX = [fr'<script[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//{_EMBED_RE}(?:{_ID_RE}).*?)\1']
+    _VALID_URL = rf'https?://(?:{_SUBDOMAIN_RE}{_DOMAIN_RE}/|{_EMBED_RE})(?P<id>{_ID_RE})'
+    _EMBED_REGEX = [
+        rf'<script[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//{_EMBED_RE}(?:{_ID_RE}).*?)\1',
+        rf'<iframe[^>]+\bsrc=["\'](?P<url>https?://{_SUBDOMAIN_RE}{_DOMAIN_RE}/[\da-f]{{32}})',
+    ]
     _TESTS = [{
         'url': 'https://embed.cloudflarestream.com/embed/we4g.fla9.latest.js?video=31c9291ab41fac05471db4e73aa11717',
         'info_dict': {
             'id': '31c9291ab41fac05471db4e73aa11717',
             'ext': 'mp4',
             'title': '31c9291ab41fac05471db4e73aa11717',
+            'thumbnail': 'https://videodelivery.net/31c9291ab41fac05471db4e73aa11717/thumbnails/thumbnail.jpg',
         },
         'params': {
-            'skip_download': True,
+            'skip_download': 'm3u8',
         },
     }, {
         'url': 'https://watch.cloudflarestream.com/9df17203414fd1db3e3ed74abbe936c1',
@@ -39,6 +37,18 @@ class CloudflareStreamIE(InfoExtractor):
         'url': 'https://customer-aw5py76sw8wyqzmh.cloudflarestream.com/2463f6d3e06fa29710a337f5f5389fd8/iframe',
         'only_matching': True,
     }]
+    _WEBPAGE_TESTS = [{
+        'url': 'https://upride.cc/incident/shoulder-pass-at-light/',
+        'info_dict': {
+            'id': 'eaef9dea5159cf968be84241b5cedfe7',
+            'ext': 'mp4',
+            'title': 'eaef9dea5159cf968be84241b5cedfe7',
+            'thumbnail': 'https://videodelivery.net/eaef9dea5159cf968be84241b5cedfe7/thumbnails/thumbnail.jpg',
+        },
+        'params': {
+            'skip_download': 'm3u8',
+        },
+    }]
 
     def _real_extract(self, url):
         video_id = self._match_id(url)