Unpunycode search keywords and short names.

components/search_engines/template_url_parser.cc

Index: components/search_engines/template_url_parser.cc
diff --git a/components/search_engines/template_url_parser.cc b/components/search_engines/template_url_parser.cc
index 51644b16d2b9e64578998a5fa5cc2528a3104caf..5455f82a77529fa71927d4becb3f73d6385c74e3 100644
--- a/components/search_engines/template_url_parser.cc
+++ b/components/search_engines/template_url_parser.cc
@@ -16,6 +16,7 @@
 #include "components/search_engines/template_url.h"
 #include "libxml/parser.h"
 #include "libxml/xmlwriter.h"
+#include "net/base/net_util.h"
 #include "ui/gfx/favicon_size.h"
 #include "url/gurl.h"
 #include "url/url_constants.h"
@@ -251,7 +252,11 @@ void TemplateURLParsingContext::EndElementImpl(void* ctx, const xmlChar* name) {
       context->ProcessURLParams();
       break;
     case TemplateURLParsingContext::SHORT_NAME:
-      context->data_.SetShortName(context->string_);
+      // If someone gives us ShortName in punycode, decode it.
+      context->data_.SetShortName(
+          base::IsStringASCII(context->string_)

[Peter Kasting, 2015/07/14 18:05:33]

Why do an ASCII check here?  Why not just unconditionally go the IDNToUnicode()
route as you have elsewhere?

If there's an efficiency issue, perhaps IDNToUnicode() should do an ASCII check
and early-out itself?  Or maybe we need to insert these sorts of conditionals in
other places too?

[alshabalin, 2015/07/15 14:18:01]

IDNToUnicode requires ASCII because it calls ASCIIToUTF16 on its argument. Here
it's the data from OpenSearch xml which could contain unicode. In other places
it is guaranteed to be in ASCII: in JS API it previously called ASCIIToUTF16, in
TemplateURL::GenerateKeyword it uses hostname.

But it might indeed make sense to move the ASCII check to IDNToUnicode.

[Peter Kasting, 2015/07/15 21:19:12]

At the least you need a comment in all these places noting why you do or do not
need an ASCII check.

+              ? net::IDNToUnicode(base::UTF16ToASCII(context->string_), "")
+              : context->string_);
       break;
     case TemplateURLParsingContext::IMAGE: {
       GURL image_url(base::UTF16ToUTF8(context->string_));