Bug 553540 - need new flag to nsPlainTextSerializer to support delsp=yes (RFC 3676). r=laurent sr=jst

content/base/public/nsIDocumentEncoder.idl

@@ -61,7 +61,7 @@ interface nsIDocumentEncoderNodeFixup : nsISupports
   nsIDOMNode fixupNode(in nsIDOMNode aNode, out boolean aSerializeCloneKids);
 };
 
-[scriptable, uuid(196a3aee-006e-4f8f-a420-e1c1b0958a26)]
+[scriptable, uuid(794a81f6-bde6-4f76-9f5e-0ea0911a2d9f)]
 interface nsIDocumentEncoder : nsISupports
 {
   // Output methods flag bits. There are a frightening number of these,
@@ -215,6 +215,13 @@ interface nsIDocumentEncoder : nsISupports
    */
   const unsigned long OutputDontRewriteEncodingDeclaration = (1 << 18);
  
+  /**
+   * Output for delsp=yes (RFC 3676). This is used with OutputFormatFlowed
+   * when converting to text for mail sending.
+   * PlainText output only.
+   */
+  const unsigned long OutputFormatDelSp  = (1 << 19);
+ 
   /**
    * When using the HTML or XHTML serializer, skip elements that are not
    * visible when this flag is set.  Elements are not visible when they

content/base/src/nsPlainTextSerializer.cpp

@@ -1411,8 +1411,10 @@ nsPlainTextSerializer::AddToLine(const PRUnichar * aLineFragment,
         else {
           mCurrentLine.Right(restOfLine, linelength-goodSpace);
         }
+        // if breaker was U+0020, it has to consider for delsp=yes support
+        PRBool breakBySpace = mCurrentLine.CharAt(goodSpace) == ' ';
         mCurrentLine.Truncate(goodSpace); 
-        EndLine(PR_TRUE);
+        EndLine(PR_TRUE, breakBySpace);
         mCurrentLine.Truncate();
         // Space stuff new line?
         if(mFlags & nsIDocumentEncoder::OutputFormatFlowed) {
@@ -1450,7 +1452,7 @@ nsPlainTextSerializer::AddToLine(const PRUnichar * aLineFragment,
  * preformatted.
  */
 void
-nsPlainTextSerializer::EndLine(PRBool aSoftlinebreak)
+nsPlainTextSerializer::EndLine(PRBool aSoftlinebreak, PRBool aBreakBySpace)
 {
   PRUint32 currentlinelength = mCurrentLine.Length();
 
@@ -1482,6 +1484,12 @@ nsPlainTextSerializer::EndLine(PRBool aSoftlinebreak)
     // Add the soft part of the soft linebreak (RFC 2646 4.1)
     // We only do this when there is no indentation since format=flowed
     // lines and indentation doesn't work well together.
+
+    // If breaker character is ASCII space with RFC 3676 support (delsp=yes),
+    // add twice space.
+    if (mFlags & nsIDocumentEncoder::OutputFormatDelSp && aBreakBySpace)
+      mCurrentLine.Append(NS_LITERAL_STRING("  "));
+    else
       mCurrentLine.Append(PRUnichar(' '));
   }
 

content/base/src/nsPlainTextSerializer.h

@@ -129,7 +129,7 @@ public:
 protected:
   nsresult GetAttributeValue(const nsIParserNode* node, nsIAtom* aName, nsString& aValueRet);
   void AddToLine(const PRUnichar* aStringToAdd, PRInt32 aLength);
-  void EndLine(PRBool softlinebreak);
+  void EndLine(PRBool softlinebreak, PRBool aBreakBySpace = PR_FALSE);
   void EnsureVerticalSpace(PRInt32 noOfRows);
   void FlushLine();
   void OutputQuotesAndIndent(PRBool stripTrailingSpaces=PR_FALSE);

content/base/test/TestPlainTextSerializer.cpp

@@ -45,11 +45,12 @@
 #include "nsServiceManagerUtils.h"
 #include "nsStringGlue.h"
 #include "nsParserCIID.h"
+#include "nsIDocumentEncoder.h"
 
 static NS_DEFINE_CID(kCParserCID, NS_PARSER_CID);
 
 void
-ConvertBufToPlainText(nsString &aConBuf)
+ConvertBufToPlainText(nsString &aConBuf, int aFlag)
 {
   nsCOMPtr<nsIParser> parser = do_CreateInstance(kCParserCID);
   if (parser) {
@@ -59,7 +60,7 @@ ConvertBufToPlainText(nsString &aConBuf)
       nsCOMPtr<nsIHTMLToTextSink> textSink(do_QueryInterface(sink));
       if (textSink) {
         nsAutoString convertedText;
-        textSink->Initialize(&convertedText, 0, 72);
+        textSink->Initialize(&convertedText, aFlag, 72);
         parser->SetContentSink(sink);
         parser->Parse(aConBuf, 0, NS_LITERAL_CSTRING("text/html"), PR_TRUE);
         aConBuf = convertedText;
@@ -68,13 +69,87 @@ ConvertBufToPlainText(nsString &aConBuf)
   }
 }
 
+// Test for ASCII with format=flowed; delsp=yes
+nsresult
+TestASCIIWithFlowedDelSp()
+{
+  nsString test;
+  nsString result;
+
+  test.AssignLiteral("<html><body>"
+                     "Firefox Firefox Firefox Firefox "
+                     "Firefox Firefox Firefox Firefox "
+                     "Firefox Firefox Firefox Firefox"
+                     "</body></html>");
+
+  ConvertBufToPlainText(test, nsIDocumentEncoder::OutputFormatted |
+                              nsIDocumentEncoder::OutputCRLineBreak |
+                              nsIDocumentEncoder::OutputLFLineBreak |
+                              nsIDocumentEncoder::OutputFormatFlowed |
+                              nsIDocumentEncoder::OutputFormatDelSp);
+
+  // create result case
+  result.AssignLiteral("Firefox Firefox Firefox Firefox "
+                       "Firefox Firefox Firefox Firefox "
+                       "Firefox  \r\nFirefox Firefox Firefox\r\n");
+
+  if (!test.Equals(result)) {
+    fail("Wrong HTML to ASCII text serialization with format=flowed; delsp=yes");
+    return NS_ERROR_FAILURE;
+  }
+
+  passed("HTML to ASCII text serialization with format=flowed; delsp=yes");
+
+  return NS_OK;
+}
+
+// Test for CJK with format=flowed; delsp=yes
+nsresult
+TestCJKWithFlowedDelSp()
+{
+  nsString test;
+  nsString result;
+
+  test.AssignLiteral("<html><body>");
+  for (PRUint32 i = 0; i < 40; i++) {
+    // Insert Kanji (U+5341)
+    test.Append(0x5341);
+  }
+  test.AppendLiteral("</body></html>");
+
+  ConvertBufToPlainText(test, nsIDocumentEncoder::OutputFormatted |
+                              nsIDocumentEncoder::OutputCRLineBreak |
+                              nsIDocumentEncoder::OutputLFLineBreak |
+                              nsIDocumentEncoder::OutputFormatFlowed |
+                              nsIDocumentEncoder::OutputFormatDelSp);
+
+  // create result case
+  for (PRUint32 i = 0; i < 36; i++) {
+    result.Append(0x5341);
+  }
+  result.Append(NS_LITERAL_STRING(" \r\n"));
+  for (PRUint32 i = 0; i < 4; i++) {
+    result.Append(0x5341);
+  }
+  result.Append(NS_LITERAL_STRING("\r\n"));
+
+  if (!test.Equals(result)) {
+    fail("Wrong HTML to CJK text serialization with format=flowed; delsp=yes");
+    return NS_ERROR_FAILURE;
+  }
+
+  passed("HTML to CJK text serialization with format=flowed; delsp=yes");
+
+  return NS_OK;
+}
+
 nsresult
 TestPlainTextSerializer()
 {
   nsString test;
   test.AppendLiteral("<html><base>base</base><head><span>span</span></head>"
                      "<body>body</body></html>");
-  ConvertBufToPlainText(test);
+  ConvertBufToPlainText(test, 0);
   if (!test.EqualsLiteral("basespanbody")) {
     fail("Wrong html to text serialization");
     return NS_ERROR_FAILURE;
@@ -82,6 +157,12 @@ TestPlainTextSerializer()
 
   passed("HTML to text serialization test");
 
+  nsresult rv = TestASCIIWithFlowedDelSp();
+  NS_ENSURE_SUCCESS(rv, rv);
+
+  rv = TestCJKWithFlowedDelSp();
+  NS_ENSURE_SUCCESS(rv, rv);
+
   // Add new tests here...
   return NS_OK;
 }