Unescape BiDi control chars while parsing data urls

net/base/data_url_unittest.cc

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "base/basictypes.h"
 #include "net/base/data_url.h"
 #include "testing/gtest/include/gtest/gtest.h"
 #include "url/gurl.h"
 
 namespace {
@@ skipping 166 matching lines @@
       "US-ASCII",
       "hello world" },
 
     // Bad encoding (truncated).
     { "data:;base64,aGVsbG8gd29yb",
       false,
       "",
       "",
       "" },
 
+    // BiDi control characters should be unescaped and preserved as is, and
+    // should not be replaced with % versions. In the below case, \xE2\x80\x8F
+    // is the RTL mark and the parsed text should preserve it as is.
+    {
+      "data:text/html;charset=utf-8,<b>\xE2\x80\x8Ftest</b>",

[mmenke, 2014/10/17 15:24:43]

Any reason not to just use text/plain?  Don't think using HTML gets us a whole
lot, not that it really hurts, either.

[meacer, 2014/10/17 20:41:54]

Done.

+      true,
+      "text/html",
+      "utf-8",
+      "<b>\xE2\x80\x8Ftest</b>"},
+
+    // Same as above but with Arabic text after RTL mark.
+    {
+      "data:text/html;charset=utf-8,"
+          "<b>\xE2\x80\x8F\xD8\xA7\xD8\xAE\xD8\xAA\xD8\xA8\xD8\xA7\xD8\xB1</b>",
+      true,
+      "text/html",
+      "utf-8",
+      "<b>\xE2\x80\x8F\xD8\xA7\xD8\xAE\xD8\xAA\xD8\xA8\xD8\xA7\xD8\xB1</b>"},
+
+    // RTL mark encoded as %E2%80%8F should be unescaped too.
+    {
+      "data:text/html;charset=utf-8,<b>%E2%80%8Ftest</b>",

[mmenke, 2014/10/17 15:24:44]

When this is wrapped in a GURL, this becomes the same thing as the first test,
right?  Fine with testing it both ways, just want to make sure I'm
understanding.  If so, think it's worth noting in a comment.

[meacer, 2014/10/17 20:41:54]

Added a note.

+      true,
+      "text/html",
+      "utf-8",
+      "<b>\xE2\x80\x8Ftest</b>"},
+
+    // Same as above but with Arabic text after RTL mark.
+    {
+      "data:text/html;charset=utf-8,"
+          "<b>%E2%80%8F\xD8\xA7\xD8\xAE\xD8\xAA\xD8\xA8\xD8\xA7\xD8\xB1</b>",
+      true,
+      "text/html",
+      "utf-8",
+      "<b>\xE2\x80\x8F\xD8\xA7\xD8\xAE\xD8\xAA\xD8\xA8\xD8\xA7\xD8\xB1</b>"}
+
     // TODO(darin): add more interesting tests
   };
 
   for (size_t i = 0; i < arraysize(tests); ++i) {
     std::string mime_type;
     std::string charset;
     std::string data;
     bool ok =
         net::DataURL::Parse(GURL(tests[i].url), &mime_type, &charset, &data);
     EXPECT_EQ(ok, tests[i].is_valid);
     if (tests[i].is_valid) {
       EXPECT_EQ(tests[i].mime_type, mime_type);
       EXPECT_EQ(tests[i].charset, charset);
       EXPECT_EQ(tests[i].data, data);
     }
   }
 }