Unescape BiDi control chars while parsing data urls

net/base/data_url_unittest.cc

Index: net/base/data_url_unittest.cc
diff --git a/net/base/data_url_unittest.cc b/net/base/data_url_unittest.cc
index 3876301cad5ecb441630d1669bbbc8a6fcac8118..2ea7cb3fb5ede283fc57cd7e8b5fba3dfd4609d3 100644
--- a/net/base/data_url_unittest.cc
+++ b/net/base/data_url_unittest.cc
@@ -184,6 +184,42 @@ TEST(DataURLTest, Parse) {
       "",
       "" },
 
+    // 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
   };