favicons

pytorrent/services/tracker_cache.py

@@ -58,7 +58,9 @@ def _root_domain(domain: str) -> str:
     parts = [p for p in str(domain or "").lower().strip(".").split(".") if p]
     if len(parts) <= 2:
         return ".".join(parts)
-    if len(parts[-1]) == 2 and len(parts[-2]) <= 3 and len(parts) >= 3:
+    # Note: Tracker favicon discovery needs the real main site first; for t.pte.nu that is pte.nu, not t.pte.nu.
+    known_second_level_suffixes = {"co", "com", "net", "org", "gov", "edu", "ac"}
+    if len(parts[-1]) == 2 and parts[-2] in known_second_level_suffixes and len(parts) >= 3:
         return ".".join(parts[-3:])
     return ".".join(parts[-2:])
 
@@ -314,10 +316,17 @@ def _extract_icon_hrefs(html: str) -> list[str]:
 def _tracker_icon_hosts(domain: str) -> list[str]:
     host = tracker_domain(domain)
     root = _root_domain(host)
-    # Note: Only probe the exact tracker host and the registrable root domain; CDN/static hosts are used only when HTML explicitly points to them.
+    # Note: Direct favicon fallback checks the tracker host first, then the main domain.
     return [h for h in dict.fromkeys([host, root]) if h]
 
 
+def _tracker_html_hosts(domain: str) -> list[str]:
+    host = tracker_domain(domain)
+    root = _root_domain(host)
+    # Note: HTML discovery checks the main site first, because tracker announce hosts often return text/plain.
+    return [h for h in dict.fromkeys([root, host]) if h]
+
+
 def _favicon_candidates(domain: str) -> list[str]:
     candidates = []
     for h in _tracker_icon_hosts(domain):
@@ -327,7 +336,7 @@ def _favicon_candidates(domain: str) -> list[str]:
 
 def _html_icon_candidates(domain: str, errors: list[str] | None = None) -> list[str]:
     urls = []
-    for h in _tracker_icon_hosts(domain):
+    for h in _tracker_html_hosts(domain):
         for scheme in ("https", "http"):
             base = f"{scheme}://{h}/"
             try:
@@ -377,7 +386,7 @@ def favicon_path(domain: str, enabled: bool = True, force: bool = False) -> tupl
                 pass
         if cached.get("error"):
             return None, None
-    # Note: Favicon lookup prefers HTML <link rel="icon"> over generic /favicon.ico, because some trackers serve a broken default icon there.
+    # Note: Favicon lookup checks the main-domain HTML first, then tracker HTML, then direct /favicon.ico fallbacks.
     FAVICON_DIR.mkdir(parents=True, exist_ok=True)
     errors = []
     candidates = _html_icon_candidates(clean, errors) + _favicon_candidates(clean)