Bug 1908759 - Don't use UTS 46 ToUnicode from nsDocShell::AttemptURIFixup. r=necko-reviewers,jesup,valentin

Differential Revision: https://phabricator.services.mozilla.com/D217009

docshell/base/nsDocShell.cpp

@@ -5992,29 +5992,16 @@ already_AddRefed<nsIURI> nsDocShell::AttemptURIFixup(
           //
           // If this string was passed through nsStandardURL by
           // chance, then it may have been converted from UTF-8 to
-          // ACE, which would result in a completely bogus keyword
+          // Punycode, which would result in a completely bogus keyword
           // query.  Here we try to recover the original Unicode
           // value, but this is not 100% correct since the value may
           // have been normalized per the IDN normalization rules.
           //
           // Since we don't have access to the exact original string
           // that was entered by the user, this will just have to do.
-          //
-          // XXX: Since we are not trying to use the result as an
-          // actual domain name, label-wise Punycode decode would
-          // likely be more appropriate than the full ToUnicode
-          // operation.
-          bool isACE;
           nsAutoCString utf8Host;
-          nsCOMPtr<nsIIDNService> idnSrv =
-              do_GetService(NS_IDNSERVICE_CONTRACTID);
-          if (idnSrv && NS_SUCCEEDED(idnSrv->IsACE(host, &isACE)) && isACE &&
-              NS_SUCCEEDED(idnSrv->ConvertACEtoUTF8(host, utf8Host))) {
-            info = KeywordToURI(utf8Host, aUsePrivateBrowsing);
-
-          } else {
-            info = KeywordToURI(host, aUsePrivateBrowsing);
-          }
+          mozilla_net_recover_keyword_from_punycode(&host, &utf8Host);
+          info = KeywordToURI(utf8Host, aUsePrivateBrowsing);
         }
         if (info) {
           info->GetPreferredURI(getter_AddRefs(newURI));

netwerk/base/idna_glue/src/lib.rs

@@ -228,3 +228,34 @@ fn process<OutputUnicode: FnMut(&[char], &[char], bool) -> bool>(
     dst.assign(src);
     nserror::NS_OK
 }
+
+/// Not general-purpose! Only to be used from `nsDocShell::AttemptURIFixup`.
+#[no_mangle]
+pub unsafe extern "C" fn mozilla_net_recover_keyword_from_punycode(
+    src: *const nsACString,
+    dst: *mut nsACString,
+) {
+    let sink = &mut (*dst);
+    let mut seen_label = false;
+    for label in (*src).split(|b| *b == b'.') {
+        if seen_label {
+            sink.append(".");
+        }
+        seen_label = true;
+        // We know the Punycode prefix is in lower case if we got it from
+        // our own IDNA conversion code.
+        if let Some(punycode) = label.strip_prefix(b"xn--") {
+            // Not bothering to optimize this.
+            // Just unwrap, since we know our IDNA conversion code gives
+            // us ASCII here.
+            let utf8 = std::str::from_utf8(punycode).unwrap();
+            if let Some(decoded) = idna::punycode::decode_to_string(utf8) {
+                sink.append(&decoded);
+            } else {
+                sink.append(label);
+            }
+        } else {
+            sink.append(label);
+        }
+    }
+}

netwerk/dns/nsIDNService.cpp

@@ -183,32 +183,6 @@ NS_IMETHODIMP nsIDNService::ConvertACEtoUTF8(const nsACString& input,
   return NS_DomainToUnicodeAllowAnyGlyphfulASCII(input, _retval);
 }
 
-NS_IMETHODIMP nsIDNService::IsACE(const nsACString& input, bool* _retval) {
-  // look for the ACE prefix in the input string.  it may occur
-  // at the beginning of any segment in the domain name.  for
-  // example: "www.xn--ENCODED.com"
-
-  if (!IsAscii(input)) {
-    *_retval = false;
-    return NS_OK;
-  }
-
-  auto stringContains = [](const nsACString& haystack,
-                           const nsACString& needle) {
-    return std::search(haystack.BeginReading(), haystack.EndReading(),
-                       needle.BeginReading(), needle.EndReading(),
-                       [](unsigned char ch1, unsigned char ch2) {
-                         return tolower(ch1) == tolower(ch2);
-                       }) != haystack.EndReading();
-  };
-
-  *_retval =
-      StringBeginsWith(input, "xn--"_ns, nsCaseInsensitiveCStringComparator) ||
-      (!input.IsEmpty() && input[0] != '.' &&
-       stringContains(input, ".xn--"_ns));
-  return NS_OK;
-}
-
 NS_IMETHODIMP nsIDNService::DomainToDisplay(const nsACString& input,
                                             nsACString& _retval) {
   nsresult rv = NS_DomainToDisplay(input, _retval);

netwerk/dns/nsIIDNService.idl

@@ -77,9 +77,4 @@ interface nsIIDNService : nsISupports
      * off using the `idna` crate directly.
      */
     AUTF8String convertACEtoUTF8(in ACString input);
-
-    /**
-     * DO NOT USE! Checks if the input string is ACE encoded or not.
-     */
-    boolean isACE(in ACString input);
 };

netwerk/test/unit/test_idnservice.js

@@ -7,6 +7,10 @@ const idnService = Cc["@mozilla.org/network/idn-service;1"].getService(
 );
 
 add_task(async function test_simple() {
+  function isACE(domain) {
+    return domain.startsWith("xn--") || domain.indexOf(".xn--") > -1;
+  }
+
   let reference = [
     // The 3rd element indicates whether the second element
     // is ACE-encoded
@@ -23,6 +27,6 @@ add_task(async function test_simple() {
     Assert.equal(idnService.convertUTF8toACE(reference[i][0]), reference[i][1]);
     Assert.equal(idnService.convertUTF8toACE(reference[i][1]), reference[i][1]);
     Assert.equal(idnService.convertACEtoUTF8(reference[i][1]), reference[i][0]);
-    Assert.equal(idnService.isACE(reference[i][1]), reference[i][2]);
+    Assert.equal(isACE(reference[i][1]), reference[i][2]);
   }
 });