Show deprecation warnings for header values in XHR according to RFC 7230

Source/platform/network/HTTPParsers.cpp

Index: Source/platform/network/HTTPParsers.cpp
diff --git a/Source/platform/network/HTTPParsers.cpp b/Source/platform/network/HTTPParsers.cpp
index 7e344e63987720c6f3eae4b6f1d67a57122eaf00..ebc54f0639c1fc950a079f54cc80f842fa6de119 100644
--- a/Source/platform/network/HTTPParsers.cpp
+++ b/Source/platform/network/HTTPParsers.cpp
@@ -115,7 +115,32 @@ bool isValidHTTPHeaderValue(const String& name)
     return name.containsOnlyLatin1() && !name.contains('\r') && !name.contains('\n') && !name.contains(static_cast<UChar>('\0'));
 }
 
-// See RFC 2616, Section 2.2.
+// See RFC 7230, Section 3.2.
+// Checks whether |value| matches field-content in RFC 7230.
+// link: http://tools.ietf.org/html/rfc7230#section-3.2
+bool isValidHTTPFieldContentRFC7230(const String& value)
+{
+    if (value.isEmpty())
+        return false;
+
+    UChar firstCharacter = value[0];
+    if (firstCharacter == ' ' || firstCharacter == '\t')
+        return false;
+
+    UChar lastCharacter = value[value.length() - 1];
+    if (lastCharacter == ' ' || lastCharacter == '\t')
+        return false;

[Mike West, 2015/09/02 06:59:14]

Why not just strip leading and trailing whitespace rather than yelling at
developers about them? :)

[shiva.jm, 2015/09/02 09:37:12]

striping of leading and trailing whitespace will be applied in another separate
CL after we measure the usage form these patch, as discussed in
https://codereview.chromium.org/1018903002/#msg31 and 
https://codereview.chromium.org/1018903002/#msg33

+
+    for (unsigned i = 0; i < value.length(); ++i) {

[Mike West, 2015/09/02 06:59:14]

Can you use the hot new C++11 loop syntax here? Not sure it works with strings,
so this is an honest question. :)

[shiva.jm, 2015/09/02 09:37:12]

tried using for ( UChar c : value), but got error:
invalid range expression of type 'const WTF::String'; no viable 'begin' function
available

+        UChar c = value[i];
+        if (c == 0x7F || c > 0xFF || (c < 0x20 && c != '\t'))

[Mike West, 2015/09/02 06:59:14]

This seems like it ought to already exist somewhere. Can you do a quick `git gs
'0x20'` to see where else we do this check? I suspect we can extract a character
class out of this.

[shiva.jm, 2015/09/02 09:37:12]

In these files same kind of check is done, WTFString.cpp GlyphPageTreeNode.cpp
Response.cpp Document.cpp 
CSSMarkup.cpp CSSPropertyParser.cpp

+            return false;
+    }
+
+    return true;
+}
+
+// See RFC 7230, Section 3.2.6.
 bool isValidHTTPToken(const String& characters)
 {
     if (characters.isEmpty())