From 7aa322c02cec54eb77154a89da7e400194f0bd03 Mon Sep 17 00:00:00 2001 From: bashonly <88596187+bashonly@users.noreply.github.com> Date: Thu, 20 Jun 2024 17:05:25 -0500 Subject: [PATCH] [ie/cloudflarestream] Fix `_VALID_URL` and embed extraction (#10215) Authored by: bashonly --- yt_dlp/extractor/cloudflarestream.py | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) diff --git a/yt_dlp/extractor/cloudflarestream.py b/yt_dlp/extractor/cloudflarestream.py index f902daacf..8a409461a 100644 --- a/yt_dlp/extractor/cloudflarestream.py +++ b/yt_dlp/extractor/cloudflarestream.py @@ -6,11 +6,11 @@ from .common import InfoExtractor class CloudflareStreamIE(InfoExtractor): _SUBDOMAIN_RE = r'(?:(?:watch|iframe|customer-\w+)\.)?' _DOMAIN_RE = r'(?:cloudflarestream\.com|(?:videodelivery|bytehighway)\.net)' - _EMBED_RE = rf'embed\.{_DOMAIN_RE}/embed/[^/]+\.js\?.*?\bvideo=' - _ID_RE = r'[\da-f]{32}|[\w-]+\.[\w-]+\.[\w-]+' + _EMBED_RE = rf'(?:embed\.|{_SUBDOMAIN_RE}){_DOMAIN_RE}/embed/[^/?#]+\.js\?(?:[^#]+&)?video=' + _ID_RE = r'[\da-f]{32}|eyJ[\w-]+\.[\w-]+\.[\w-]+' _VALID_URL = rf'https?://(?:{_SUBDOMAIN_RE}{_DOMAIN_RE}/|{_EMBED_RE})(?P{_ID_RE})' _EMBED_REGEX = [ - rf']+\bsrc=(["\'])(?P(?:https?:)?//{_EMBED_RE}(?:{_ID_RE}).*?)\1', + rf']+\bsrc=(["\'])(?P(?:https?:)?//{_EMBED_RE}(?:{_ID_RE})(?:(?!\1).)*)\1', rf']+\bsrc=["\'](?Phttps?://{_SUBDOMAIN_RE}{_DOMAIN_RE}/[\da-f]{{32}})', ] _TESTS = [{ @@ -24,6 +24,14 @@ class CloudflareStreamIE(InfoExtractor): 'params': { 'skip_download': 'm3u8', }, + }, { + 'url': 'https://watch.cloudflarestream.com/embed/sdk-iframe-integration.fla9.latest.js?video=0e8e040aec776862e1d632a699edf59e', + 'info_dict': { + 'id': '0e8e040aec776862e1d632a699edf59e', + 'ext': 'mp4', + 'title': '0e8e040aec776862e1d632a699edf59e', + 'thumbnail': 'https://videodelivery.net/0e8e040aec776862e1d632a699edf59e/thumbnails/thumbnail.jpg', + }, }, { 'url': 'https://watch.cloudflarestream.com/9df17203414fd1db3e3ed74abbe936c1', 'only_matching': True, @@ -36,6 +44,9 @@ class CloudflareStreamIE(InfoExtractor): }, { 'url': 'https://customer-aw5py76sw8wyqzmh.cloudflarestream.com/2463f6d3e06fa29710a337f5f5389fd8/iframe', 'only_matching': True, + }, { + 'url': 'https://watch.cloudflarestream.com/eyJhbGciOiJSUzI1NiIsInR5cCI6IkpXVCJ9.eyJraWQiOiJmYTA0YjViMzQ2NDkwYTM5NWJiNzQ1NWFhZTA2YzYwZSIsInN1YiI6Ijg4ZDQxMDhhMzY0MjA3M2VhYmFhZjg3ZGExODJkMjYzIiwiZXhwIjoxNjAwNjA5MzE5fQ.xkRJwLGkt0nZ%5F0BlPiwU7iW4pqb4lKkznbKfAhGg0tGcxSS6ZBA3lcTUwu7W%2DyCFbnAl%2Dhqk3Fn%5FqeQS%5FQydP27qTHpB9iIFFsMtk1tqzGZV5v4yrYDnwLSKzEKvVd6QwJnfABtxH2JdpSNuWlMUiVXFxGWgjOw6QeTNDDklTQYXV%5FNLV7sErSn5CeOPeRRkdXb%2D8ip%5FVOcfk1nDsFoOo4fctFtGP0wYMyY5ae8nhhatydHwevuvJCcEvEfh%2D4qjq9mCZOodevmtSQ4YWmggf4BxtWnDWYrGW8Otp6oqezrR8oY4%2DbKdV6PaqBj49aJdcls6xK7PmM8%5Fvjy3xfm0Mg', + 'only_matching': True, }] _WEBPAGE_TESTS = [{ 'url': 'https://upride.cc/incident/shoulder-pass-at-light/',