Bug 801402 - Use FindEncodingForLabel from HTML parser. r=hsivonen

dom/locales/en-US/chrome/layout/htmlparser.properties

@@ -14,7 +14,6 @@ EncMetaUnsupported=An unsupported character encoding was declared for the HTML d
 EncProtocolUnsupported=An unsupported character encoding was declared on the transfer protocol level. The declaration was ignored.
 EncBomlessUtf16=Detected UTF-16-encoded Basic Latin-only text without a byte order mark and without a transfer protocol-level declaration. Encoding this content in UTF-16 is inefficient and the character encoding should have been declared in any case.
 EncMetaUtf16=A meta tag was used to declare the character encoding as UTF-16. This was interpreted as an UTF-8 declaration instead.
-EncMetaNonRoughSuperset=A meta tag was used to declare a character encoding the does not encode the Basic Latin range roughly like US-ASCII. The declaration was ignored.
 
 # The bulk of the messages below are derived from 
 # http://hg.mozilla.org/projects/htmlparser/file/1f633cef7de7/src/nu/validator/htmlparser/impl/ErrorReportingTokenizer.java

parser/html/nsHtml5MetaScannerCppSupplement.h

@@ -4,10 +4,12 @@
  
 #include "nsICharsetConverterManager.h"
 #include "nsServiceManagerUtils.h"
-#include "nsCharsetAlias.h"
 #include "nsEncoderDecoderUtils.h"
 #include "nsTraceRefcnt.h"
 
+#include "mozilla/dom/EncodingUtils.h"
+
+using mozilla::dom::EncodingUtils;
 
 void
 nsHtml5MetaScanner::sniff(nsHtml5ByteReadable* bytes, nsIUnicodeDecoder** decoder, nsACString& charset)
@@ -48,8 +50,7 @@ nsHtml5MetaScanner::tryCharset(nsString* charset)
     return true;
   }
   nsAutoCString preferred;
-  res = nsCharsetAlias::GetPreferred(encoding, preferred);
-  if (NS_FAILED(res)) {
+  if (!EncodingUtils::FindEncodingForLabel(encoding, preferred)) {
     return false;
   }
   if (preferred.LowerCaseEqualsLiteral("utf-16") ||

parser/html/nsHtml5StreamParser.cpp

@@ -6,7 +6,6 @@
 
 #include "nsHtml5StreamParser.h"
 #include "nsICharsetConverterManager.h"
-#include "nsCharsetAlias.h"
 #include "nsServiceManagerUtils.h"
 #include "nsEncoderDecoderUtils.h"
 #include "nsContentUtils.h"
@@ -26,8 +25,10 @@
 #include "nsCharsetSource.h"
 #include "nsIWyciwygChannel.h"
 
-using namespace mozilla;
+#include "mozilla/dom/EncodingUtils.h"
 
+using namespace mozilla;
+using mozilla::dom::EncodingUtils;
 
 int32_t nsHtml5StreamParser::sTimerInitialDelay = 120;
 int32_t nsHtml5StreamParser::sTimerSubsequentDelay = 120;
@@ -1193,28 +1194,25 @@ nsHtml5StreamParser::OnDataAvailable(nsIRequest* aRequest,
 bool
 nsHtml5StreamParser::PreferredForInternalEncodingDecl(nsACString& aEncoding)
 {
-  nsAutoCString newEncoding(aEncoding);
-  newEncoding.Trim(" \t\r\n\f");
-  if (newEncoding.LowerCaseEqualsLiteral("utf-16") ||
-      newEncoding.LowerCaseEqualsLiteral("utf-16be") ||
-      newEncoding.LowerCaseEqualsLiteral("utf-16le")) {
-    mTreeBuilder->MaybeComplainAboutCharset("EncMetaUtf16",
-                                            true,
-                                            mTokenizer->getLineNumber());
-    newEncoding.Assign("UTF-8");
-  }
-
-  nsresult rv = NS_OK;
-  bool eq;
-  rv = nsCharsetAlias::Equals(newEncoding, mCharset, &eq);
-  if (NS_FAILED(rv)) {
+  nsAutoCString newEncoding;
+  if (!EncodingUtils::FindEncodingForLabel(aEncoding, newEncoding)) {
     // the encoding name is bogus
     mTreeBuilder->MaybeComplainAboutCharset("EncMetaUnsupported",
                                             true,
                                             mTokenizer->getLineNumber());
     return false;
   }
-  if (eq) {
+
+  if (newEncoding.EqualsLiteral("UTF-16") ||
+      newEncoding.EqualsLiteral("UTF-16BE") ||
+      newEncoding.EqualsLiteral("UTF-16LE")) {
+    mTreeBuilder->MaybeComplainAboutCharset("EncMetaUtf16",
+                                            true,
+                                            mTokenizer->getLineNumber());
+    newEncoding.Assign("UTF-8");
+  }
+
+  if (newEncoding.Equals(mCharset)) {
     if (mCharsetSource < kCharsetFromMetaPrescan) {
       if (mInitialEncodingWasFromParentFrame) {
         mTreeBuilder->MaybeComplainAboutCharset("EncLateMetaFrame",
@@ -1231,36 +1229,7 @@ nsHtml5StreamParser::PreferredForInternalEncodingDecl(nsACString& aEncoding)
     return false;
   }
 
-  // XXX check HTML5 non-IANA aliases here
-
-  nsAutoCString preferred;
-  rv = nsCharsetAlias::GetPreferred(newEncoding, preferred);
-  if (NS_FAILED(rv)) {
-    // This charset has been blacklisted for permitting XSS smuggling.
-    // EncMetaNonRoughSuperset is a reasonable approximation to the
-    // right error message.
-    mTreeBuilder->MaybeComplainAboutCharset("EncMetaNonRoughSuperset",
-                                            true,
-                                            mTokenizer->getLineNumber());
-    return false;
-  }
-
-  // ??? Explicit further blacklist of character sets that are not
-  // "rough supersets" of ASCII.  Some of these are handled above (utf-16),
-  // some by the XSS smuggling blacklist in charsetData.properties,
-  // maybe all of the remainder should also be blacklisted there.
-  if (preferred.LowerCaseEqualsLiteral("utf-16") ||
-      preferred.LowerCaseEqualsLiteral("utf-16be") ||
-      preferred.LowerCaseEqualsLiteral("utf-16le") ||
-      preferred.LowerCaseEqualsLiteral("utf-7") ||
-      preferred.LowerCaseEqualsLiteral("x-imap4-modified-utf7")) {
-    // Not a rough ASCII superset
-    mTreeBuilder->MaybeComplainAboutCharset("EncMetaNonRoughSuperset",
-                                            true,
-                                            mTokenizer->getLineNumber());
-    return false;
-  }
-  aEncoding.Assign(preferred);
+  aEncoding.Assign(newEncoding);
   return true;
 }
 

parser/htmlparser/src/nsParser.cpp

@@ -14,7 +14,6 @@
 #include "nsIChannel.h"
 #include "nsICachingChannel.h"
 #include "nsICacheEntryDescriptor.h"
-#include "nsCharsetAlias.h"
 #include "nsICharsetConverterManager.h"
 #include "nsIInputStream.h"
 #include "CNavDTD.h"
@@ -43,7 +42,10 @@
 #include "nsCharsetSource.h"
 #include "nsContentUtils.h"
 
+#include "mozilla/dom/EncodingUtils.h"
+
 using namespace mozilla;
+using mozilla::dom::EncodingUtils;
 
 #define NS_PARSER_FLAG_PARSER_ENABLED         0x00000002
 #define NS_PARSER_FLAG_OBSERVERS_ENABLED      0x00000004
@@ -1840,8 +1842,7 @@ ParserWriteFunc(nsIInputStream* in,
       nsAutoCString declCharset;
 
       if (ExtractCharsetFromXmlDeclaration(buf, count, declCharset)) {
-        nsresult rv = nsCharsetAlias::GetPreferred(declCharset, maybePrefer);
-        if (NS_SUCCEEDED(rv)) {
+        if (EncodingUtils::FindEncodingForLabel(declCharset, maybePrefer)) {
           preferred.Assign(maybePrefer);
           source = kCharsetFromMetaTag;
         }

parser/htmlparser/src/nsScanner.cpp

@@ -10,7 +10,6 @@
 #include "nsDebug.h"
 #include "nsIServiceManager.h"
 #include "nsICharsetConverterManager.h"
-#include "nsCharsetAlias.h"
 #include "nsReadableUtils.h"
 #include "nsIInputStream.h"
 #include "nsIFile.h"
@@ -20,6 +19,10 @@
 #include "nsParser.h"
 #include "nsCharsetSource.h"
 
+#include "mozilla/dom/EncodingUtils.h"
+
+using mozilla::dom::EncodingUtils;
+
 // We replace NUL characters with this character.
 static PRUnichar sInvalid = UCS2_REPLACEMENT_CHAR;
 
@@ -118,12 +121,12 @@ nsresult nsScanner::SetDocumentCharset(const nsACString& aCharset , int32_t aSou
   if (aSource < mCharsetSource) // priority is lower the the current one , just
     return NS_OK;
 
-  nsresult res = NS_OK;
+  nsCString charsetName;
+  bool valid = EncodingUtils::FindEncodingForLabel(aCharset, charsetName);
+  MOZ_ASSERT(valid, "Should never call with a bogus aCharset.");
   if (!mCharset.IsEmpty())
   {
-    bool same;
-    res = nsCharsetAlias::Equals(aCharset, mCharset, &same);
-    if(NS_SUCCEEDED(res) && same)
+    if (charsetName.Equals(mCharset))
     {
       mCharsetSource = aSource;
       return NS_OK; // no difference, don't change it
@@ -131,9 +134,6 @@ nsresult nsScanner::SetDocumentCharset(const nsACString& aCharset , int32_t aSou
   }
 
   // different, need to change it
-  nsCString charsetName;
-  res = nsCharsetAlias::GetPreferred(aCharset, charsetName);
-  MOZ_ASSERT(NS_SUCCEEDED(res), "Should never call with a bogus aCharset.");
 
   mCharset.Assign(charsetName);
 
@@ -142,7 +142,7 @@ nsresult nsScanner::SetDocumentCharset(const nsACString& aCharset , int32_t aSou
   NS_ASSERTION(nsParser::GetCharsetConverterManager(),
                "Must have the charset converter manager!");
 
-  res = nsParser::GetCharsetConverterManager()->
+  nsresult res = nsParser::GetCharsetConverterManager()->
     GetUnicodeDecoderRaw(mCharset.get(), getter_AddRefs(mUnicodeDecoder));
   if (NS_SUCCEEDED(res) && mUnicodeDecoder)
   {