Add support for international domain names in PAC scripts. This converts non-ASCII hostnames in dnsResolve() and dnsResolveEx() to punycode.

BUG=47234
TEST=ProxyResolverV8Test.DNSResolutionOfInternationDomainName
Review URL: http://codereview.chromium.org/2842017

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@50624 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 65 insertions, 6 deletions

diff --git a/net/data/proxy_resolver_v8_unittest/international_domain_names.js b/net/data/proxy_resolver_v8_unittest/international_domain_names.js
new file mode 100755
index 0000000..546af13
--- /dev/null
+++ b/net/data/proxy_resolver_v8_unittest/international_domain_names.js
@@ -0,0 +1,16 @@
+// Try resolving hostnames containing non-ASCII characters.
+
+function FindProxyForURL(url, host) {
+  // This international hostname has a non-ASCII character. It is represented
+  // in punycode as 'xn--bcher-kva.ch'
+  var idn = 'B\u00fccher.ch';
+
+  // We disregard the actual return value -- all we care about is that on
+  // the C++ end the bindings were passed the punycode equivalent of this
+  // unicode hostname.
+  dnsResolve(idn);
+  dnsResolveEx(idn);
+
+  return "DIRECT";
+}
+
diff --git a/net/proxy/proxy_resolver_v8.cc b/net/proxy/proxy_resolver_v8.cc
index 05f16e9..4b54cb7 100644
--- a/net/proxy/proxy_resolver_v8.cc
+++ b/net/proxy/proxy_resolver_v8.cc
@@ -6,7 +6,9 @@
 
 #include "base/logging.h"
 #include "base/string_util.h"
+#include "base/utf_string_conversions.h"
 #include "googleurl/src/gurl.h"
+#include "googleurl/src/url_canon.h"
 #include "net/base/host_cache.h"
 #include "net/base/net_errors.h"
 #include "net/base/net_log.h"
@@ -109,13 +111,30 @@ bool GetHostnameArgument(const v8::Arguments& args, std::string* hostname) {
 
   const string16 hostname_utf16 = V8StringToUTF16(args[0]->ToString());
 
-  // TODO(eroman): Convert international domain names to punycode. Right
-  //               now we fail if the hostname isn't entered in ASCII.
-  //               crbug.com/47234
-  if (!IsStringASCII(hostname_utf16))
+  // If the hostname is already in ASCII, simply return it as is.
+  if (IsStringASCII(hostname_utf16)) {
+    *hostname = UTF16ToASCII(hostname_utf16);
+    return true;
+  }
+
+  // Otherwise try to convert it from IDN to punycode.
+  const int kInitialBufferSize = 256;
+  url_canon::RawCanonOutputT<char16, kInitialBufferSize> punycode_output;
+  if (!url_canon::IDNToASCII(hostname_utf16.data(),
+                             hostname_utf16.length(),
+                             &punycode_output)) {
     return false;
-  *hostname = UTF16ToASCII(hostname_utf16);
-  return true;
+  }
+
+  // |punycode_output| should now be ASCII; convert it to a std::string.
+  // (We could use UTF16ToASCII() instead, but that requires an extra string
+  // copy. Since ASCII is a subset of UTF8 the following is equivalent).
+  bool success = UTF16ToUTF8(punycode_output.data(),
+                             punycode_output.length(),
+                             hostname);
+  DCHECK(success);
+  DCHECK(IsStringASCII(*hostname));
+  return success;
 }
 
 }  // namespace
diff --git a/net/proxy/proxy_resolver_v8_unittest.cc b/net/proxy/proxy_resolver_v8_unittest.cc
index 5146fc6..9f9f50d 100644
--- a/net/proxy/proxy_resolver_v8_unittest.cc
+++ b/net/proxy/proxy_resolver_v8_unittest.cc
@@ -548,5 +548,29 @@ TEST(ProxyResolverV8Test, DNSResolutionFailure) {
   EXPECT_EQ("success:80", proxy_info.proxy_server().ToURI());
 }
 
+TEST(ProxyResolverV8Test, DNSResolutionOfInternationDomainName) {
+  ProxyResolverV8WithMockBindings resolver;
+  int result = resolver.SetPacScriptFromDisk("international_domain_names.js");
+  EXPECT_EQ(OK, result);
+
+  // Execute FindProxyForURL().
+  ProxyInfo proxy_info;
+  result = resolver.GetProxyForURL(kQueryUrl, &proxy_info, NULL, NULL,
+                                   BoundNetLog());
+
+  EXPECT_EQ(OK, result);
+  EXPECT_TRUE(proxy_info.is_direct());
+
+  // Check that the international domain name was converted to punycode
+  // before passing it onto the bindings layer.
+  MockJSBindings* bindings = resolver.mock_js_bindings();
+
+  ASSERT_EQ(1u, bindings->dns_resolves.size());
+  EXPECT_EQ("xn--bcher-kva.ch", bindings->dns_resolves[0]);
+
+  ASSERT_EQ(1u, bindings->dns_resolves_ex.size());
+  EXPECT_EQ("xn--bcher-kva.ch", bindings->dns_resolves_ex[0]);
+}
+
 }  // namespace
 }  // namespace net