treat ipv4 addresses as a single lookup key in the url classifier. b=395376, r=tony, a=mconnor

toolkit/components/url-classifier/src/nsUrlClassifierDBService.cpp

@@ -489,9 +489,12 @@ private:
   nsresult GetLookupFragments(const nsCSubstring& spec,
                               nsTArray<nsUrlClassifierHash>& fragments);
 
+  // Check for a canonicalized IP address.
+  PRBool IsCanonicalizedIP(const nsACString& host);
+
   // Get the database key for a given URI.  This is the top three
   // domain components if they exist, otherwise the top two.
-  //  hostname.com/foo/bar -> hostname
+  //  hostname.com/foo/bar -> hostname.com
   //  mail.hostname.com/foo/bar -> mail.hostname.com
   //  www.mail.hostname.com/foo/bar -> mail.hostname.com
   nsresult GetKey(const nsACString& spec, nsUrlClassifierHash& hash);
@@ -812,6 +815,15 @@ nsUrlClassifierDBServiceWorker::DoLookup(const nsACString& spec,
   }
 
   const nsCSubstring& host = Substring(begin, iter++);
+
+  nsTArray<PRUint32> resultTables;
+  nsUrlClassifierHash hash;
+
+  if (IsCanonicalizedIP(host)) {
+    // Don't break up the host into components
+    hash.FromPlaintext(host, mCryptoHash);
+    CheckKey(spec, hash, resultTables);
+  } else {
     nsCStringArray hostComponents;
     hostComponents.ParseString(PromiseFlatCString(host).get(), ".");
 
@@ -828,12 +840,10 @@ nsUrlClassifierDBServiceWorker::DoLookup(const nsACString& spec,
     lookupHost.Append(".");
     lookupHost.Append(*hostComponents[last]);
     lookupHost.Append("/");
-  nsUrlClassifierHash hash;
     hash.FromPlaintext(lookupHost, mCryptoHash);
 
     // we ignore failures from CheckKey because we'd rather try to find
     // more results than fail.
-  nsTArray<PRUint32> resultTables;
     CheckKey(spec, hash, resultTables);
 
     // Now check with three domain components
@@ -846,6 +856,7 @@ nsUrlClassifierDBServiceWorker::DoLookup(const nsACString& spec,
 
       CheckKey(spec, hash, resultTables);
     }
+  }
 
   nsCAutoString result;
   for (PRUint32 i = 0; i < resultTables.Length(); i++) {
@@ -1151,6 +1162,21 @@ nsUrlClassifierDBServiceWorker::WriteEntry(nsUrlClassifierEntry& entry)
   return NS_OK;
 }
 
+PRBool
+nsUrlClassifierDBServiceWorker::IsCanonicalizedIP(const nsACString& host)
+{
+  // The canonicalization process will have left IP addresses in dotted
+  // decimal with no surprises.
+  PRUint32 i1, i2, i3, i4;
+  char c;
+  if (PR_sscanf(PromiseFlatCString(host).get(), "%u.%u.%u.%u%c",
+                &i1, &i2, &i3, &i4, &c) == 4) {
+    return (i1 <= 0xFF && i1 <= 0xFF && i1 <= 0xFF && i1 <= 0xFF);
+  }
+
+  return PR_FALSE;
+}
+
 nsresult
 nsUrlClassifierDBServiceWorker::GetKey(const nsACString& spec,
                                        nsUrlClassifierHash& hash)
@@ -1164,7 +1190,12 @@ nsUrlClassifierDBServiceWorker::GetKey(const nsACString& spec,
     return NS_OK;
   }
 
-  const nsCSubstring& host = Substring(begin, iter++);
+  const nsCSubstring& host = Substring(begin, iter);
+
+  if (IsCanonicalizedIP(host)) {
+    return hash.FromPlaintext(host, mCryptoHash);
+  }
+
   nsCStringArray hostComponents;
   hostComponents.ParseString(PromiseFlatCString(host).get(), ".");