From 6330d5a789d53b53540f167273f93d384206631d Mon Sep 17 00:00:00 2001
From: CREDO23 <bakerathierry@gmail.com>
Date: Thu, 23 Jul 2026 18:52:50 +0200
Subject: [PATCH] fix: reject spoofed non-YouTube URLs

---
 .../platforms/youtube/url_resolver.py         | 19 +++++++++++++------
 .../unit/platforms/youtube/test_parsers.py    | 18 ++++++++++++++++--
 2 files changed, 29 insertions(+), 8 deletions(-)

diff --git a/surfsense_backend/app/proprietary/platforms/youtube/url_resolver.py b/surfsense_backend/app/proprietary/platforms/youtube/url_resolver.py
index 0e68425a9..aa7df36e9 100644
--- a/surfsense_backend/app/proprietary/platforms/youtube/url_resolver.py
+++ b/surfsense_backend/app/proprietary/platforms/youtube/url_resolver.py
@@ -15,6 +15,9 @@ ResolvedKind = Literal["video", "channel", "playlist", "hashtag", "search"]
 
 _PLAYLIST_ID_RE = re.compile(r"[?&]list=([\w-]+)")
 
+_YOUTUBE_HOSTS = frozenset({"www.youtube.com", "youtube.com", "m.youtube.com"})
+"""Hosts whose page paths (/@, /channel, /playlist, ...) are trusted as YouTube."""
+
 
 @dataclass(frozen=True)
 class ResolvedUrl:
@@ -29,7 +32,7 @@ def get_youtube_video_id(url: str) -> str | None:
     hostname = parsed.hostname or ""
     if hostname == "youtu.be":
         return parsed.path[1:] or None
-    if hostname in ("www.youtube.com", "youtube.com", "m.youtube.com"):
+    if hostname in _YOUTUBE_HOSTS:
         if parsed.path == "/watch":
             return parse_qs(parsed.query).get("v", [None])[0]
         for prefix in ("/embed/", "/v/", "/shorts/"):
@@ -41,17 +44,21 @@ def get_youtube_video_id(url: str) -> str | None:
 def resolve_url(url: str) -> ResolvedUrl | None:
     """Classify a YouTube URL into a scrape job, or ``None`` if unrecognized."""
     parsed = urlparse(url)
+    hostname = (parsed.hostname or "").lower()
     path = parsed.path or ""
 
-    # Shorts are videos with their own path.
-    if "/shorts/" in path:
-        vid = path.split("/shorts/")[1].split("/")[0]
-        return ResolvedUrl("video", vid, url) if vid else None
-
+    # Videos: watch / youtu.be / embed / shorts. get_youtube_video_id validates
+    # the host itself, so this also covers youtu.be short links.
     video_id = get_youtube_video_id(url)
     if video_id:
         return ResolvedUrl("video", video_id, url)
 
+    # Every remaining shape is a youtube.com page keyed off its path. Require a
+    # YouTube host so a well-formed non-YouTube URL (e.g. https://evil.com/@x)
+    # is rejected instead of misclassified as a channel by its path alone.
+    if hostname not in _YOUTUBE_HOSTS:
+        return None
+
     # Playlist (either a /playlist page or any URL carrying ?list=).
     playlist_match = _PLAYLIST_ID_RE.search(url)
     if path.startswith("/playlist") and playlist_match:
diff --git a/surfsense_backend/tests/unit/platforms/youtube/test_parsers.py b/surfsense_backend/tests/unit/platforms/youtube/test_parsers.py
index 55f591c63..6b7b460a9 100644
--- a/surfsense_backend/tests/unit/platforms/youtube/test_parsers.py
+++ b/surfsense_backend/tests/unit/platforms/youtube/test_parsers.py
@@ -946,8 +946,22 @@ def test_resolve_url(url, kind, value):
     assert resolved.value == value
 
 
-def test_resolve_url_unrecognized():
-    assert resolve_url("https://example.com/foo") is None
+@pytest.mark.parametrize(
+    "url",
+    [
+        "https://example.com/foo",
+        # Well-formed non-YouTube URLs whose path mimics a YouTube page must be
+        # rejected, not misclassified by path alone (host-spoof guard).
+        "https://evil.com/@Apify",
+        "https://evil.com/channel/UC123456789abc",
+        "https://evil.com/shorts/abc123",
+        "https://evil.com/playlist?list=PL123",
+        "https://evil.com/hashtag/tech",
+        "https://evil.com/results?search_query=web+scraping",
+    ],
+)
+def test_resolve_url_unrecognized(url):
+    assert resolve_url(url) is None
 
 
 # --- optional: exercise captured real fixtures if present --------------------