Bug 1408782 - Force punycode display for IDNs with a <dotless-i, combining mark above> sequence. r=valentin

netwerk/dns/nsIDNService.cpp

@@ -750,6 +750,7 @@ bool nsIDNService::isLabelSafe(const nsAString &label)
 
   Script lastScript = Script::INVALID;
   uint32_t previousChar = 0;
+  uint32_t baseChar = 0; // last non-diacritic seen (base char for marks)
   uint32_t savedNumberingSystem = 0;
 // Simplified/Traditional Chinese check temporarily disabled -- bug 857481
 #if 0
@@ -830,6 +831,14 @@ bool nsIDNService::isLabelSafe(const nsAString &label)
           }
         }
       }
+      // Check for diacritics on dotless-i, which would be indistinguishable
+      // from normal accented letter i.
+      if (baseChar == 0x0131 &&
+          ((ch >= 0x0300 && ch <= 0x0314) || ch == 0x031a)) {
+        return false;
+      }
+    } else {
+      baseChar = ch;
     }
 
     if (script != Script::COMMON && script != Script::INHERITED) {

netwerk/test/unit/test_idn_urls.js

@@ -303,6 +303,13 @@ const testcases = [
 
     // Hebrew diacritic also not allowed in Latin text (bug 1404349)
     ["goo\u05b4gle", "xn--google-rvh", false, false, false],
+
+    // Accents above dotless-i are not allowed
+    ["na\u0131\u0308ve", "xn--nave-mza04z", false, false, false],
+    ["d\u0131\u0302ner", "xn--dner-lza40z", false, false, false],
+    // but the corresponding accented-i (based on dotted i) is OK
+    ["na\u00efve.com", "xn--nave-6pa.com", false, true, true],
+    ["d\u00eener.com", "xn--dner-0pa.com", false, true, true],
 ];
 
 const profiles = ["ASCII", "high", "moderate"];