df

ProviderCore/registry.py

@@ -224,11 +224,19 @@ def match_provider_name_for_url(url: str) -> Optional[str]:
     #
     # This keeps direct downloads and item pages routed to `internetarchive`, while
     # preserving OpenLibrary's scripted borrow pipeline for loan/reader URLs.
-    if host:
-        if host == "openlibrary.org" or host.endswith(".openlibrary.org"):
+    def _norm_host(h: str) -> str:
+        h_norm = str(h or "").strip().lower()
+        if h_norm.startswith("www."):
+            h_norm = h_norm[4:]
+        return h_norm
+
+    host_norm = _norm_host(host)
+
+    if host_norm:
+        if host_norm == "openlibrary.org" or host_norm.endswith(".openlibrary.org"):
             return "openlibrary" if "openlibrary" in _PROVIDERS else None
 
-        if host == "archive.org" or host.endswith(".archive.org"):
+        if host_norm == "archive.org" or host_norm.endswith(".archive.org"):
             low_path = str(path or "").lower()
             is_borrowish = (
                 low_path.startswith("/borrow/") or low_path.startswith("/stream/")
@@ -243,16 +251,20 @@ def match_provider_name_for_url(url: str) -> Optional[str]:
         if not domains:
             continue
         for d in domains:
-            dom = str(d or "").strip().lower()
+            dom_raw = str(d or "").strip()
+            dom = dom_raw.lower()
             if not dom:
                 continue
-            if raw_url_lower.startswith(dom):
-                return name
-        for d in domains:
-            dom = str(d or "").strip().lower()
-            if not dom or not host:
+            # Scheme-like patterns (magnet:, http://example) still use prefix match.
+            if dom.startswith("magnet:") or dom.startswith("http://") or dom.startswith("https://"):
+                if raw_url_lower.startswith(dom):
+                    return name
                 continue
-            if host == dom or host.endswith("." + dom):
+
+            dom_norm = _norm_host(dom)
+            if not dom_norm or not host_norm:
+                continue
+            if host_norm == dom_norm or host_norm.endswith("." + dom_norm):
                 return name
 
     return None