Support IDNA 2008

net/base/net_util.cc

Index: net/base/net_util.cc
===================================================================
--- net/base/net_util.cc	(revision 215936)
+++ net/base/net_util.cc	(working copy)
@@ -322,7 +322,7 @@
   UErrorCode status = U_ZERO_ERROR;
 #ifdef U_WCHAR_IS_UTF16
   icu::UnicodeSet dangerous_characters(icu::UnicodeString(
-      L"[[\\ \u00bc\u00bd\u01c3\u0337\u0338"
+      L"[[\\ \u00ad\u00bc\u00bd\u01c3\u0337\u0338"
       L"\u05c3\u05f4\u06d4\u0702\u115f\u1160][\u2000-\u200b]"
       L"[\u2024\u2027\u2028\u2029\u2039\u203a\u2044\u205f]"
       L"[\u2154-\u2156][\u2159-\u215b][\u215f\u2215\u23ae"
@@ -339,7 +339,7 @@
       0, status);
 #else
   icu::UnicodeSet dangerous_characters(icu::UnicodeString(
-      "[[\\u0020\\u00bc\\u00bd\\u01c3\\u0337\\u0338"
+      "[[\\u0020\\u00ad\\u00bc\\u00bd\\u01c3\\u0337\\u0338"
       "\\u05c3\\u05f4\\u06d4\\u0702\\u115f\\u1160][\\u2000-\\u200b]"
       "[\\u2024\\u2027\\u2028\\u2029\\u2039\\u203a\\u2044\\u205f]"
       "[\\u2154-\\u2156][\\u2159-\\u215b][\\u215f\\u2215\\u23ae"
@@ -396,6 +396,9 @@
   return false;
 }
 
+static base::LazyInstance<base::Lock>::Leaky
+    g_uidna_lock = LAZY_INSTANCE_INITIALIZER;
+
 // Converts one component of a host (between dots) to IDN if safe. The result
 // will be APPENDED to the given output string and will be the same as the input
 // if it is not IDN or the IDN is unsafe to display.  Returns whether any
@@ -408,30 +411,48 @@
   if (comp_len == 0)
     return false;
 
+
   // Only transform if the input can be an IDN component.
   static const base::char16 kIdnPrefix[] = {'x', 'n', '-', '-'};
   if ((comp_len > arraysize(kIdnPrefix)) &&
       !memcmp(comp, kIdnPrefix, arraysize(kIdnPrefix) * sizeof(base::char16))) {
-    // Repeatedly expand the output string until it's big enough.  It looks like
-    // ICU will return the required size of the buffer, but that's not
-    // documented, so we'll just grow by 2x. This should be rare and is not on a
-    // critical path.
+    static UIDNA* uidna = NULL; // will be leaked.
+    {
+      UErrorCode err = U_ZERO_ERROR;
+      base::AutoLock lock(g_uidna_lock.Get());
+      if (uidna == NULL) {
+        // This is the option closest to what we had in the past with IDNA 2003
+        // API and matches what IE 10 does except for BiDi check.
+        // IDNA 2003 always checks BiDi. We used to allow unassigned code
+        // points. However, with our Unicode DB pretty up to date, we'd not
+        // need to turn this on.
+        // We didn't use STD3 rules and we continue not to.
+        // TODO(jungshik) : Review and change options as different
+        // parties (browsers, registrars, search engines) converge toward
+        // a consensus.
+        int32_t options = UIDNA_CHECK_BIDI;
+        uidna = uidna_openUTS46(options, &err);
+        if (U_FAILURE(err))
+          return false;
+      }
+    }
     size_t original_length = out->length();
-    for (int extra_space = 64; ; extra_space *= 2) {
+    for (int output_length = 64; ; ) {
       UErrorCode status = U_ZERO_ERROR;
-      out->resize(out->length() + extra_space);
-      int output_chars = uidna_IDNToUnicode(comp,
-          static_cast<int32_t>(comp_len), &(*out)[original_length], extra_space,
-          UIDNA_DEFAULT, NULL, &status);
-      if (status == U_ZERO_ERROR) {
+      UIDNAInfo info = UIDNA_INFO_INITIALIZER;
+      out->resize(original_length + output_length);
+      output_length = uidna_labelToUnicode(uidna, comp,
+          static_cast<int32_t>(comp_len), &(*out)[original_length],
+          output_length, &info, &status);
+      if (U_SUCCESS(status) && info.errors == 0) {
         // Converted successfully.
-        out->resize(original_length + output_chars);
-        if (IsIDNComponentSafe(out->data() + original_length, output_chars,
+        out->resize(original_length + output_length);
+        if (IsIDNComponentSafe(out->data() + original_length, output_length,
                                languages))
           return true;
       }
 
-      if (status != U_BUFFER_OVERFLOW_ERROR)
+      if (status != U_BUFFER_OVERFLOW_ERROR || info.errors != 0)
         break;
     }
     // Failed, revert back to original string.