favicons

pytorrent/services/tracker_cache.py

@@ -159,27 +159,20 @@ def summary(profile: dict, hashes: list[str], loader, scan_limit: int = TRACKER_
 
 
 
-def favicon_public_url(domain: str, enabled: bool = True, create: bool = False) -> str:
-    """Return the static URL for a cached tracker favicon, optionally creating it first."""
+def favicon_public_url(domain: str, enabled: bool = True, create: bool = False, force: bool = False) -> str:
+    """Return the static URL for a cached tracker favicon, optionally creating or refreshing it first."""
     # Note: Favicon files stay in data/tracker_favicons, but the browser loads them via the static/tracker_favicons symlink.
     clean = tracker_domain(domain)
     if not enabled or not clean:
         return ""
     if create:
-        favicon_path(clean, enabled=True)
+        favicon_path(clean, enabled=True, force=force)
     cached = _cached_favicon(clean)
     now = _now_epoch()
-    path = None
-    if cached and now - float(cached.get("updated_epoch") or 0) < FAVICON_CACHE_TTL_SECONDS:
-        cached_path = Path(str(cached.get("file_path") or ""))
-        if cached_path.exists() and cached_path.is_file():
-            path = cached_path
-    if path is None:
-        # Note: Existing symlinked .ico files are still linked directly even when the DB favicon row is missing or stale.
-        direct_path = FAVICON_DIR / f"{_safe_filename(clean)}.ico"
-        if direct_path.exists() and direct_path.is_file():
-            path = direct_path
-    if path is None:
+    if not cached or now - float(cached.get("updated_epoch") or 0) >= FAVICON_CACHE_TTL_SECONDS:
+        return ""
+    path = Path(str(cached.get("file_path") or ""))
+    if not path.exists() or not path.is_file():
         return ""
     try:
         rel = path.resolve().relative_to(FAVICON_DIR.resolve())
@@ -199,12 +192,32 @@ def _fetch(url: str, limit: int = 262144) -> tuple[bytes, str, str]:
 
 
 def _is_icon(data: bytes, content_type: str, url: str) -> bool:
-    if not data:
+    """Validate that downloaded bytes are a browser-readable image, not only an image-like HTTP header."""
+    # Note: Some trackers serve a broken /favicon.ico with image/vnd.microsoft.icon; pyTorrent now validates bytes before caching it.
+    if not data or len(data) < 16:
         return False
-    ctype = content_type.lower()
-    if ctype.startswith("image/") or ctype in {"application/octet-stream", "binary/octet-stream"}:
+    head = data[:32]
+    lower = data[:512].lstrip().lower()
+    if head.startswith(b"\x00\x00\x01\x00") or head.startswith(b"\x00\x00\x02\x00"):
+        try:
+            count = int.from_bytes(data[4:6], "little")
+        except Exception:
+            count = 0
+        return 0 < count <= 256 and len(data) >= 6 + (16 * count)
+    if head.startswith(b"\x89PNG\r\n\x1a\n"):
         return True
-    return urllib.parse.urlparse(url).path.lower().endswith((".ico", ".png", ".jpg", ".jpeg", ".svg", ".webp"))
+    if head.startswith(b"\xff\xd8\xff"):
+        return True
+    if head.startswith((b"GIF87a", b"GIF89a")):
+        return True
+    if head.startswith(b"RIFF") and data[8:12] == b"WEBP":
+        return True
+    if lower.startswith(b"<svg") or b"<svg" in lower[:256]:
+        return True
+    ctype = content_type.lower()
+    if ctype in {"image/svg+xml"}:
+        return b"<svg" in lower[:512]
+    return False
 
 
 def _favicon_candidates(domain: str) -> list[str]:
@@ -250,22 +263,28 @@ def _cached_favicon(domain: str):
         return conn.execute("SELECT * FROM tracker_favicon_cache WHERE domain=?", (clean,)).fetchone()
 
 
-def favicon_path(domain: str, enabled: bool = True) -> tuple[Path | None, str | None]:
+def favicon_path(domain: str, enabled: bool = True, force: bool = False) -> tuple[Path | None, str | None]:
     clean = tracker_domain(domain)
     if not enabled or not clean:
         return None, None
     cached = _cached_favicon(clean)
     now = _now_epoch()
-    if cached and now - float(cached.get("updated_epoch") or 0) < FAVICON_CACHE_TTL_SECONDS:
+    if cached and not force and now - float(cached.get("updated_epoch") or 0) < FAVICON_CACHE_TTL_SECONDS:
         path = Path(str(cached.get("file_path") or ""))
-        if path.exists():
-            return path, str(cached.get("mime_type") or mimetypes.guess_type(path.name)[0] or "image/x-icon")
+        mime = str(cached.get("mime_type") or mimetypes.guess_type(path.name)[0] or "image/x-icon")
+        if path.exists() and path.is_file():
+            try:
+                if _is_icon(path.read_bytes()[:524288], mime, str(cached.get("source_url") or path.name)):
+                    return path, mime
+            except Exception:
+                pass
         if cached.get("error"):
             return None, None
-    # Note: Favicon lookup tries tracker host, root domain, then HTML <link rel="icon"> and stores the result for a week.
+    # Note: Favicon lookup prefers HTML <link rel="icon"> over generic /favicon.ico, because some trackers serve a broken default icon there.
     FAVICON_DIR.mkdir(parents=True, exist_ok=True)
     errors = []
-    candidates = _favicon_candidates(clean)
+    candidates = _html_icon_candidates(clean) + _favicon_candidates(clean)
+    candidates = list(dict.fromkeys(candidates))
     checked_html = False
     idx = 0
     while idx < len(candidates):