Bug 479959 - Reimplement text/plain loading using the HTML5 parser. r=Olli.Pettay.

content/base/test/chrome/test_bug430050.xul

@@ -36,7 +36,7 @@ https://bugzilla.mozilla.org/show_bug.cgi?id=430050
           document.getElementById('b').loadURI('data:text/plain,succeeded',
                                                null,
                                                'UTF-8');
-          setTimeout(endTest, 0);
+          document.getElementById('b').addEventListener("load", endTest);
         }
       }, true);
     document.documentElement.setAttribute("foo", "bar");

content/html/document/src/nsHTMLDocument.cpp

@@ -141,6 +141,7 @@
 #include "prprf.h"
 #include "mozilla/dom/Element.h"
 #include "mozilla/Preferences.h"
+#include "nsMimeTypes.h"
 
 using namespace mozilla;
 using namespace mozilla::dom;
@@ -162,6 +163,7 @@ const PRInt32 kBackward = 1;
 //#define DEBUG_charset
 
 #define NS_USE_NEW_VIEW_SOURCE 1
+#define NS_USE_NEW_PLAIN_TEXT 1
 
 static NS_DEFINE_CID(kCParserCID, NS_PARSER_CID);
 
@@ -652,17 +654,30 @@ nsHTMLDocument::StartDocumentLoad(const char* aCommand,
                                   bool aReset,
                                   nsIContentSink* aSink)
 {
+  nsCAutoString contentType;
+  aChannel->GetContentType(contentType);
+
   bool viewSource = aCommand && !nsCRT::strcmp(aCommand, "view-source") &&
     NS_USE_NEW_VIEW_SOURCE;
-  bool loadAsHtml5 = nsHtml5Module::sEnabled || viewSource;
+  bool plainText = (contentType.EqualsLiteral(TEXT_PLAIN) ||
+    contentType.EqualsLiteral(TEXT_CSS) ||
+    contentType.EqualsLiteral(APPLICATION_JAVASCRIPT) ||
+    contentType.EqualsLiteral(APPLICATION_XJAVASCRIPT) ||
+    contentType.EqualsLiteral(TEXT_ECMASCRIPT) ||
+    contentType.EqualsLiteral(APPLICATION_ECMASCRIPT) ||
+    contentType.EqualsLiteral(TEXT_JAVASCRIPT));
+  bool loadAsHtml5 = nsHtml5Module::sEnabled || viewSource || plainText;
+  if (!NS_USE_NEW_PLAIN_TEXT && !viewSource) {
+    plainText = false;
+  }
+
+  NS_ASSERTION(!(plainText && aSink),
+               "Someone tries to load plain text into a custom sink.");
 
   if (aSink) {
     loadAsHtml5 = false;
   }
 
-  nsCAutoString contentType;
-  aChannel->GetContentType(contentType);
-
   if (contentType.Equals("application/xhtml+xml") && !viewSource) {
     // We're parsing XHTML as XML, remember that.
 
@@ -677,7 +692,8 @@ nsHTMLDocument::StartDocumentLoad(const char* aCommand,
   }
 #endif
   
-  if (loadAsHtml5 && !viewSource && !(contentType.EqualsLiteral("text/html") &&
+  if (loadAsHtml5 && !viewSource &&
+      (!(contentType.EqualsLiteral("text/html") || plainText) &&
       aCommand && !nsCRT::strcmp(aCommand, "view"))) {
     loadAsHtml5 = false;
   }
@@ -731,9 +747,13 @@ nsHTMLDocument::StartDocumentLoad(const char* aCommand,
   if (needsParser) {
     if (loadAsHtml5) {
       mParser = nsHtml5Module::NewHtml5Parser();
-      mParser->MarkAsNotScriptCreated((viewSource &&
-        !contentType.EqualsLiteral("text/html")) ?
-        "view-source-xml": aCommand);
+      if (plainText) {
+        mParser->MarkAsNotScriptCreated("plain-text");
+      } else if (viewSource && !contentType.EqualsLiteral("text/html")) {
+        mParser->MarkAsNotScriptCreated("view-source-xml");
+      } else {
+        mParser->MarkAsNotScriptCreated(aCommand);
+      }
     } else {
       mParser = do_CreateInstance(kCParserCID, &rv);
       NS_ENSURE_SUCCESS(rv, rv);

layout/build/nsContentDLF.cpp

@@ -87,7 +87,8 @@ nsresult
 NS_NewContentViewer(nsIContentViewer** aResult);
 
 // XXXbz if you change the MIME types here, be sure to update
-// nsIParser.h and DetermineParseMode in nsParser.cpp accordingly.
+// nsIParser.h and DetermineParseMode in nsParser.cpp and
+// nsHTMLDocument::StartDocumentLoad accordingly.
 static const char* const gHTMLTypes[] = {
   TEXT_HTML,
   TEXT_PLAIN,

parser/html/nsHtml5Parser.cpp

@@ -707,6 +707,8 @@ nsHtml5Parser::MarkAsNotScriptCreated(const char* aCommand)
     mode = VIEW_SOURCE_HTML;
   } else if (!nsCRT::strcmp(aCommand, "view-source-xml")) {
     mode = VIEW_SOURCE_XML;
+  } else if (!nsCRT::strcmp(aCommand, "plain-text")) {
+    mode = PLAIN_TEXT;
   }
 #ifdef DEBUG
   else {

parser/html/nsHtml5StreamParser.cpp

@@ -866,6 +866,12 @@ nsHtml5StreamParser::OnStartRequest(nsIRequest* aRequest, nsISupports* aContext)
   mTokenizer->start();
   mExecutor->Start();
   mExecutor->StartReadingFromStage();
+
+  if (mMode == PLAIN_TEXT) {
+    mTreeBuilder->StartPlainText();
+    mTokenizer->StartPlainText();
+  }
+
   /*
    * If you move the following line, be very careful not to cause 
    * WillBuildModel to be called before the document has had its 
@@ -891,7 +897,7 @@ nsHtml5StreamParser::OnStartRequest(nsIRequest* aRequest, nsISupports* aContext)
   // The line below means that the encoding can end up being wrong if
   // a view-source URL is loaded without having the encoding hint from a
   // previous normal load in the history.
-  mReparseForbidden = !(mMode == NORMAL);
+  mReparseForbidden = !(mMode == NORMAL || mMode == PLAIN_TEXT);
 
   nsCOMPtr<nsIHttpChannel> httpChannel(do_QueryInterface(mRequest, &rv));
   if (NS_SUCCEEDED(rv)) {

parser/html/nsHtml5TokenizerCppSupplement.h

@@ -35,6 +35,12 @@
  *
  * ***** END LICENSE BLOCK ***** */
 
+void
+nsHtml5Tokenizer::StartPlainText()
+{
+  stateSave = NS_HTML5TOKENIZER_PLAINTEXT;
+}
+
 void
 nsHtml5Tokenizer::EnableViewSource(nsHtml5Highlighter* aHighlighter)
 {

parser/html/nsHtml5TokenizerHSupplement.h

@@ -37,6 +37,12 @@
 
 nsAutoPtr<nsHtml5Highlighter> mViewSource;
 
+/**
+ * Starts handling text/plain. This is a one-way initialization. There is
+ * no corresponding EndPlainText() call.
+ */
+void StartPlainText();
+
 void EnableViewSource(nsHtml5Highlighter* aHighlighter);
 
 bool FlushViewSource();

parser/html/nsHtml5TreeBuilderCppSupplement.h

@@ -700,6 +700,15 @@ nsHtml5TreeBuilder::MarkAsBroken()
   mOpQueue.AppendElement()->Init(eTreeOpMarkAsBroken);
 }
 
+void
+nsHtml5TreeBuilder::StartPlainText()
+{
+  startTag(nsHtml5ElementName::ELT_PRE,
+           nsHtml5HtmlAttributes::EMPTY_ATTRIBUTES,
+           false);
+  needToDropLF = false;
+}
+
 // DocumentModeHandler
 void
 nsHtml5TreeBuilder::documentMode(nsHtml5DocumentMode m)

parser/html/nsHtml5TreeBuilderHSupplement.h

@@ -95,6 +95,8 @@
 
     ~nsHtml5TreeBuilder();
     
+    void StartPlainText();
+
     bool HasScript();
     
     void SetOpSink(nsAHtml5TreeOpSink* aOpSink) {