From 9730e0d3594240090a4aad83ba6f9afccf314e10 Mon Sep 17 00:00:00 2001
From: Jonathan Kamens <jik@kamens.brookline.ma.us>
Date: Tue, 21 Sep 2010 15:02:55 +0200
Subject: [PATCH] bug 564737 - plaintext serializer shouldn't take care of
 leading spaces in a block. tests by David :Bienvenu. r=laurentj sr+a=jst

---
 content/base/src/nsPlainTextSerializer.cpp    | 37 +++++++++----------
 content/base/test/TestPlainTextSerializer.cpp | 25 +++++++++++++
 content/base/test/test_copypaste.html         | 30 +++++++--------
 3 files changed, 58 insertions(+), 34 deletions(-)

diff --git a/content/base/src/nsPlainTextSerializer.cpp b/content/base/src/nsPlainTextSerializer.cpp
index aae3b11a535..df3a5b367aa 100644
--- a/content/base/src/nsPlainTextSerializer.cpp
+++ b/content/base/src/nsPlainTextSerializer.cpp
@@ -119,7 +119,7 @@ nsPlainTextSerializer::nsPlainTextSerializer()
 
   // Flow
   mEmptyLines = 1; // The start of the document is an "empty line" in itself,
-  mInWhitespace = PR_TRUE;
+  mInWhitespace = PR_FALSE;
   mPreFormatted = PR_FALSE;
   mStartedOutput = PR_FALSE;
 
@@ -638,6 +638,8 @@ nsPlainTextSerializer::DoOpenContainer(const nsIParserNode* aNode, PRInt32 aTag)
       }
     } 
     else {
+      /* See comment at end of function. */
+      mInWhitespace = PR_TRUE;
       mPreFormatted = PR_FALSE;
     }
 
@@ -807,6 +809,13 @@ nsPlainTextSerializer::DoOpenContainer(const nsIParserNode* aNode, PRInt32 aTag)
     Write(NS_LITERAL_STRING("_"));
   }
 
+  /* Container elements are always block elements, so we shouldn't
+     output any whitespace immediately after the container tag even if
+     there's extra whitespace there because the HTML is pretty-printed
+     or something. To ensure that happens, tell the serializer we're
+     already in whitespace so it won't output more. */
+  mInWhitespace = PR_TRUE;
+
   return NS_OK;
 }
 
@@ -1073,38 +1082,26 @@ nsPlainTextSerializer::DoAddLeaf(const nsIParserNode *aNode, PRInt32 aTag,
       EnsureVerticalSpace(mEmptyLines+1);
     }
   }
-  else if (type == eHTMLTag_whitespace) {
+  else if (type == eHTMLTag_whitespace || type == eHTMLTag_newline) {
     // The only times we want to pass along whitespace from the original
     // html source are if we're forced into preformatted mode via flags,
     // or if we're prettyprinting and we're inside a <pre>.
     // Otherwise, either we're collapsing to minimal text, or we're
     // prettyprinting to mimic the html format, and in neither case
     // does the formatting of the html source help us.
-    // One exception: at the very beginning of a selection,
-    // we want to preserve whitespace.
     if (mFlags & nsIDocumentEncoder::OutputPreformatted ||
         (mPreFormatted && !mWrapColumn) ||
         IsInPre()) {
-      Write(aText);
+      if (type == eHTMLTag_newline)
+        EnsureVerticalSpace(mEmptyLines+1);
+      else  
+        Write(aText);
     }
-    else if(!mInWhitespace ||
-            (!mStartedOutput
-             && mFlags | nsIDocumentEncoder::OutputSelectionOnly)) {
-      mInWhitespace = PR_FALSE;
+    else if(!mInWhitespace) {
       Write(kSpace);
       mInWhitespace = PR_TRUE;
     }
   }
-  else if (type == eHTMLTag_newline) {
-    if (mFlags & nsIDocumentEncoder::OutputPreformatted ||
-        (mPreFormatted && !mWrapColumn) ||
-        IsInPre()) {
-      EnsureVerticalSpace(mEmptyLines+1);
-    }
-    else {
-      Write(kSpace);
-    }
-  }
   else if (type == eHTMLTag_hr &&
            (mFlags & nsIDocumentEncoder::OutputFormatted)) {
     EnsureVerticalSpace(0);
@@ -1161,10 +1158,12 @@ nsPlainTextSerializer::EnsureVerticalSpace(PRInt32 noOfRows)
   // realize that we should start a new line.
   if(noOfRows >= 0 && !mInIndentString.IsEmpty()) {
     EndLine(PR_FALSE);
+    mInWhitespace = PR_TRUE;
   }
 
   while(mEmptyLines < noOfRows) {
     EndLine(PR_FALSE);
+    mInWhitespace = PR_TRUE;
   }
   mLineBreakDue = PR_FALSE;
   mFloatingLines = -1;
diff --git a/content/base/test/TestPlainTextSerializer.cpp b/content/base/test/TestPlainTextSerializer.cpp
index 21ad5d58a85..8b76aee95e9 100644
--- a/content/base/test/TestPlainTextSerializer.cpp
+++ b/content/base/test/TestPlainTextSerializer.cpp
@@ -46,6 +46,7 @@
 #include "nsStringGlue.h"
 #include "nsParserCIID.h"
 #include "nsIDocumentEncoder.h"
+#include "nsCRT.h"
 
 static NS_DEFINE_CID(kCParserCID, NS_PARSER_CID);
 
@@ -143,6 +144,27 @@ TestCJKWithFlowedDelSp()
   return NS_OK;
 }
 
+nsresult
+TestPrettyPrintedHtml()
+{
+  nsString test;
+  test.AppendLiteral(
+    "<html>" NS_LINEBREAK
+    "<body>" NS_LINEBREAK
+    "  first<br>" NS_LINEBREAK
+    "  second<br>" NS_LINEBREAK
+    "</body>" NS_LINEBREAK "</html>");
+
+  ConvertBufToPlainText(test, 0);
+  if (!test.EqualsLiteral("first" NS_LINEBREAK "second" NS_LINEBREAK)) {
+    fail("Wrong prettyprinted html to text serialization");
+    return NS_ERROR_FAILURE;
+  }
+
+  passed("prettyprinted HTML to text serialization test");
+  return NS_OK;
+}
+
 nsresult
 TestPlainTextSerializer()
 {
@@ -163,6 +185,9 @@ TestPlainTextSerializer()
   rv = TestCJKWithFlowedDelSp();
   NS_ENSURE_SUCCESS(rv, rv);
 
+  rv = TestPrettyPrintedHtml();
+  NS_ENSURE_SUCCESS(rv, rv);
+
   // Add new tests here...
   return NS_OK;
 }
diff --git a/content/base/test/test_copypaste.html b/content/base/test/test_copypaste.html
index 9a81b5bff6e..5b33f9b7907 100644
--- a/content/base/test/test_copypaste.html
+++ b/content/base/test/test_copypaste.html
@@ -112,36 +112,36 @@ function testCopyPaste () {
   testPasteText("This is a draggable bit of text.");
 
   copyChildrenToClipboard("alist");
-  testSelectionToString("bla\n\n    foo\n    bar\n\n");
-  testClipboardValue("text/unicode", " bla\n\n    foo\n    bar\n\n");
+  testSelectionToString(" bla\n\n    foo\n    bar\n\n");
+  testClipboardValue("text/unicode", "bla\n\n    foo\n    bar\n\n");
   testClipboardValue("text/html", "<div id=\"alist\">\n    bla\n    <ul>\n      <li>foo</li>\n      \n      <li>bar</li>\n    </ul>\n  </div>");
-  testPasteText(" bla\n\n    foo\n    bar\n\n");
+  testPasteText("bla\n\n    foo\n    bar\n\n");
 
   copyChildrenToClipboard("blist");
-  testSelectionToString("mozilla\n\n    foo\n    bar\n\n");
-  testClipboardValue("text/unicode", " mozilla\n\n    foo\n    bar\n\n");
+  testSelectionToString(" mozilla\n\n    foo\n    bar\n\n");
+  testClipboardValue("text/unicode", "mozilla\n\n    foo\n    bar\n\n");
   testClipboardValue("text/html", "<div id=\"blist\">\n    mozilla\n    <ol>\n      <li>foo</li>\n      \n      <li>bar</li>\n    </ol>\n  </div>");
-  testPasteText(" mozilla\n\n    foo\n    bar\n\n");
+  testPasteText("mozilla\n\n    foo\n    bar\n\n");
 
   copyChildrenToClipboard("clist");
-  testSelectionToString("mzla\n\n    foo\n        bazzinga!\n    bar\n\n");
-  testClipboardValue("text/unicode", " mzla\n\n    foo\n        bazzinga!\n    bar\n\n");
+  testSelectionToString(" mzla\n\n    foo\n        bazzinga!\n    bar\n\n");
+  testClipboardValue("text/unicode", "mzla\n\n    foo\n        bazzinga!\n    bar\n\n");
   testClipboardValue("text/html", "<div id=\"clist\">\n    mzla\n    <ul>\n      <li>foo<ul>\n        <li>bazzinga!</li>\n      </ul></li>\n      \n      <li>bar</li>\n    </ul>\n  </div>");
-  testPasteText(" mzla\n\n    foo\n        bazzinga!\n    bar\n\n");
+  testPasteText("mzla\n\n    foo\n        bazzinga!\n    bar\n\n");
 
   copyChildrenToClipboard("div4");
-  testSelectionToString("Tt t t ");
-  testClipboardValue("text/unicode", " Tt t t ");
+  testSelectionToString(" Tt t t ");
+  testClipboardValue("text/unicode", "Tt t t ");
   testClipboardValue("text/html", "<div id=\"div4\">\n  T<textarea>t t t</textarea>\n</div>");
   testInnerHTML("div4", "\n  T<textarea>t t t</textarea>\n");
-  testPasteText(" Tt t t ");
+  testPasteText("Tt t t ");
 
   copyChildrenToClipboard("div5");
-  testSelectionToString("T ");
-  testClipboardValue("text/unicode", " T ");
+  testSelectionToString(" T ");
+  testClipboardValue("text/unicode", "T ");
   testClipboardValue("text/html", "<div id=\"div5\">\n  T<textarea>     </textarea>\n</div>");
   testInnerHTML("div5", "\n  T<textarea>     </textarea>\n");
-  testPasteText(" T ");
+  testPasteText("T ");
 
   copyRangeToClipboard($("div6").childNodes[0],0, $("div6").childNodes[1],1);
   testSelectionToString("");