gecko/content/base/src/nsPlainTextSerializer.h
Michael Wu d8e503c38b Bug 675553 - Switch from PRBool to bool on a CLOSED TREE , r=bsmedberg,khuey,bz,cjones
--HG--
rename : tools/trace-malloc/bloatblame.c => tools/trace-malloc/bloatblame.cpp
2011-09-28 23:19:26 -07:00

280 lines
11 KiB
C++

/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* ***** BEGIN LICENSE BLOCK *****
* Version: MPL 1.1/GPL 2.0/LGPL 2.1
*
* The contents of this file are subject to the Mozilla Public License Version
* 1.1 (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
* http://www.mozilla.org/MPL/
*
* Software distributed under the License is distributed on an "AS IS" basis,
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
* for the specific language governing rights and limitations under the
* License.
*
* The Original Code is mozilla.org code.
*
* The Initial Developer of the Original Code is
* Netscape Communications Corporation.
* Portions created by the Initial Developer are Copyright (C) 1998
* the Initial Developer. All Rights Reserved.
*
* Contributor(s):
*
* Alternatively, the contents of this file may be used under the terms of
* either of the GNU General Public License Version 2 or later (the "GPL"),
* or the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
* in which case the provisions of the GPL or the LGPL are applicable instead
* of those above. If you wish to allow use of your version of this file only
* under the terms of either the GPL or the LGPL, and not to allow others to
* use your version of this file under the terms of the MPL, indicate your
* decision by deleting the provisions above and replace them with the notice
* and other provisions required by the GPL or the LGPL. If you do not delete
* the provisions above, a recipient may use your version of this file under
* the terms of any one of the MPL, the GPL or the LGPL.
*
* ***** END LICENSE BLOCK ***** */
/*
* nsIContentSerializer implementation that can be used with an
* nsIDocumentEncoder to convert a DOM into plaintext in a nice way
* (eg for copy/paste as plaintext).
*/
#ifndef nsPlainTextSerializer_h__
#define nsPlainTextSerializer_h__
#include "nsIContentSerializer.h"
#include "nsIHTMLContentSink.h"
#include "nsHTMLTags.h"
#include "nsCOMPtr.h"
#include "nsString.h"
#include "nsILineBreaker.h"
#include "nsIContent.h"
#include "nsIAtom.h"
#include "nsIHTMLToTextSink.h"
#include "nsIDocumentEncoder.h"
#include "nsTArray.h"
namespace mozilla {
namespace dom {
class Element;
} // namespace dom
} // namespace mozilla
class nsPlainTextSerializer : public nsIContentSerializer,
public nsIHTMLContentSink,
public nsIHTMLToTextSink
{
public:
nsPlainTextSerializer();
virtual ~nsPlainTextSerializer();
NS_DECL_ISUPPORTS
// nsIContentSerializer
NS_IMETHOD Init(PRUint32 flags, PRUint32 aWrapColumn,
const char* aCharSet, bool aIsCopying,
bool aIsWholeDocument);
NS_IMETHOD AppendText(nsIContent* aText, PRInt32 aStartOffset,
PRInt32 aEndOffset, nsAString& aStr);
NS_IMETHOD AppendCDATASection(nsIContent* aCDATASection,
PRInt32 aStartOffset, PRInt32 aEndOffset,
nsAString& aStr);
NS_IMETHOD AppendProcessingInstruction(nsIContent* aPI,
PRInt32 aStartOffset,
PRInt32 aEndOffset,
nsAString& aStr) { return NS_OK; }
NS_IMETHOD AppendComment(nsIContent* aComment, PRInt32 aStartOffset,
PRInt32 aEndOffset, nsAString& aStr) { return NS_OK; }
NS_IMETHOD AppendDoctype(nsIContent *aDoctype,
nsAString& aStr) { return NS_OK; }
NS_IMETHOD AppendElementStart(mozilla::dom::Element* aElement,
mozilla::dom::Element* aOriginalElement,
nsAString& aStr);
NS_IMETHOD AppendElementEnd(mozilla::dom::Element* aElement,
nsAString& aStr);
NS_IMETHOD Flush(nsAString& aStr);
NS_IMETHOD AppendDocumentStart(nsIDocument *aDocument,
nsAString& aStr);
// nsIContentSink
NS_IMETHOD WillParse(void) { return NS_OK; }
NS_IMETHOD WillInterrupt(void) { return NS_OK; }
NS_IMETHOD WillResume(void) { return NS_OK; }
NS_IMETHOD SetParser(nsIParser* aParser) { return NS_OK; }
NS_IMETHOD OpenContainer(const nsIParserNode& aNode);
NS_IMETHOD CloseContainer(const nsHTMLTag aTag);
NS_IMETHOD AddLeaf(const nsIParserNode& aNode);
NS_IMETHOD AddComment(const nsIParserNode& aNode) { return NS_OK; }
NS_IMETHOD AddProcessingInstruction(const nsIParserNode& aNode) { return NS_OK; }
NS_IMETHOD AddDocTypeDecl(const nsIParserNode& aNode) { return NS_OK; }
virtual void FlushPendingNotifications(mozFlushType aType) { }
NS_IMETHOD SetDocumentCharset(nsACString& aCharset) { return NS_OK; }
virtual nsISupports *GetTarget() { return nsnull; }
// nsIHTMLContentSink
NS_IMETHOD OpenHead();
NS_IMETHOD IsEnabled(PRInt32 aTag, bool* aReturn);
NS_IMETHOD NotifyTagObservers(nsIParserNode* aNode) { return NS_OK; }
NS_IMETHOD_(bool) IsFormOnStack() { return false; }
NS_IMETHOD BeginContext(PRInt32 aPosition) { return NS_OK; }
NS_IMETHOD EndContext(PRInt32 aPosition) { return NS_OK; }
NS_IMETHOD DidProcessTokens(void) { return NS_OK; }
NS_IMETHOD WillProcessAToken(void) { return NS_OK; }
NS_IMETHOD DidProcessAToken(void) { return NS_OK; }
// nsIHTMLToTextSink
NS_IMETHOD Initialize(nsAString* aOutString,
PRUint32 aFlags, PRUint32 aWrapCol);
protected:
nsresult GetAttributeValue(const nsIParserNode* node, nsIAtom* aName, nsString& aValueRet);
void AddToLine(const PRUnichar* aStringToAdd, PRInt32 aLength);
void EndLine(bool softlinebreak, bool aBreakBySpace = false);
void EnsureVerticalSpace(PRInt32 noOfRows);
void FlushLine();
void OutputQuotesAndIndent(bool stripTrailingSpaces=false);
void Output(nsString& aString);
void Write(const nsAString& aString);
bool IsBlockLevel(PRInt32 aId);
bool IsContainer(PRInt32 aId);
bool IsInPre();
bool IsInOL();
bool IsCurrentNodeConverted(const nsIParserNode* aNode);
static PRInt32 GetIdForContent(nsIContent* aContent);
nsresult DoOpenContainer(const nsIParserNode* aNode, PRInt32 aTag);
nsresult DoCloseContainer(PRInt32 aTag);
nsresult DoAddLeaf(const nsIParserNode* aNode,
PRInt32 aTag,
const nsAString& aText);
// Inlined functions
inline bool MayWrap()
{
return mWrapColumn &&
((mFlags & nsIDocumentEncoder::OutputFormatted) ||
(mFlags & nsIDocumentEncoder::OutputWrap));
}
inline bool DoOutput()
{
return mHeadLevel == 0;
}
// Stack handling functions
bool GetLastBool(const nsTArray<bool>& aStack);
void SetLastBool(nsTArray<bool>& aStack, bool aValue);
void PushBool(nsTArray<bool>& aStack, bool aValue);
bool PopBool(nsTArray<bool>& aStack);
protected:
nsString mCurrentLine;
PRUint32 mHeadLevel;
bool mAtFirstColumn;
// Handling of quoted text (for mail):
// Quotes need to be wrapped differently from non-quoted text,
// because quoted text has a few extra characters (e.g. ">> ")
// which makes the line length longer.
// Mail can represent quotes in different ways: it can wrap
// quotes in a <pre> (if editor.quotesPreformatted is set),
// or not wrapped in any special tag (if mail.compose.wrap_to_window_width)
// or in a <span> (if neither of the above are set).
bool mQuotesPreformatted; // expect quotes wrapped in <pre>
bool mDontWrapAnyQuotes; // no special quote markers
bool mStructs; // Output structs (pref)
// If we've just written out a cite blockquote, we need to remember it
// so we don't duplicate spaces before a <pre wrap> (which mail uses to quote
// old messages).
bool mHasWrittenCiteBlockquote;
PRInt32 mIndent;
// mInIndentString keeps a header that has to be written in the indent.
// That could be, for instance, the bullet in a bulleted list.
nsString mInIndentString;
PRInt32 mCiteQuoteLevel;
PRInt32 mFlags;
PRInt32 mFloatingLines; // To store the number of lazy line breaks
// The wrap column is how many standard sized chars (western languages)
// should be allowed on a line. There could be less chars if the chars
// are wider than latin chars of more if the chars are more narrow.
PRUint32 mWrapColumn;
// The width of the line as it will appear on the screen (approx.)
PRUint32 mCurrentLineWidth;
// Treat quoted text as though it's preformatted -- don't wrap it.
// Having it on a pref is a temporary measure, See bug 69638.
PRInt32 mSpanLevel;
PRInt32 mEmptyLines; // Will be the number of empty lines before
// the current. 0 if we are starting a new
// line and -1 if we are in a line.
bool mInWhitespace;
bool mPreFormatted;
bool mStartedOutput; // we've produced at least a character
// While handling a new tag, this variable should remind if any line break
// is due because of a closing tag. Setting it to "TRUE" while closing the tags.
// Hence opening tags are guaranteed to start with appropriate line breaks.
bool mLineBreakDue;
nsString mURL;
PRInt32 mHeaderStrategy; /* Header strategy (pref)
0 = no indention
1 = indention, increased with
header level (default)
2 = numbering and slight indention */
PRInt32 mHeaderCounter[7]; /* For header-numbering:
Number of previous headers of
the same depth and in the same
section.
mHeaderCounter[1] for <h1> etc. */
nsRefPtr<mozilla::dom::Element> mElement;
// For handling table rows
nsAutoTArray<bool, 8> mHasWrittenCellsForRow;
// Values gotten in OpenContainer that is (also) needed in CloseContainer
nsAutoTArray<bool, 8> mCurrentNodeIsConverted;
nsAutoTArray<bool, 8> mIsInCiteBlockquote;
// The output data
nsAString* mOutputString;
// The tag stack: the stack of tags we're operating on, so we can nest:
nsHTMLTag *mTagStack;
PRUint32 mTagStackIndex;
// Content in the stack above this index should be ignored:
PRUint32 mIgnoreAboveIndex;
// The stack for ordered lists
PRInt32 *mOLStack;
PRUint32 mOLStackIndex;
PRUint32 mULCount;
nsString mLineBreak;
nsCOMPtr<nsILineBreaker> mLineBreaker;
// Conveniance constant. It would be nice to have it as a const static
// variable, but that causes issues with OpenBSD and module unloading.
const nsString kSpace;
};
nsresult
NS_NewPlainTextSerializer(nsIContentSerializer** aSerializer);
#endif