org.apache.xml.serialize.IndentPrinter Maven / Gradle / Ivy
Go to download
Xerces2 provides high performance, fully compliant XML parsers in the Apache Xerces family. This new version of Xerces continues to build upon the Xerces Native Interface (XNI), a complete framework for building parser components and configurations that is extremely modular and easy to program.
The Apache Xerces2 parser is the reference implementation of XNI but other parser components, configurations, and parsers can be written using the Xerces Native Interface. For complete design and implementation documents, refer to the XNI Manual.
Xerces2 provides fully conforming XML Schema 1.0 and 1.1 processors. An experimental implementation of the "XML Schema Definition Language (XSD): Component Designators (SCD) Candidate Recommendation (January 2010)" is also provided for evaluation. For more information, refer to the XML Schema page.
Xerces2 also provides a complete implementation of the Document Object Model Level 3 Core and Load/Save W3C Recommendations and provides a complete implementation of the XML Inclusions (XInclude) W3C Recommendation. It also provides support for OASIS XML Catalogs v1.1.
Xerces2 is able to parse documents written according to the XML 1.1 Recommendation, except that it does not yet provide an option to enable normalization checking as described in section 2.13 of this specification. It also handles namespaces according to the XML Namespaces 1.1 Recommendation, and will correctly serialize XML 1.1 documents if the DOM level 3 load/save APIs are in use.
/*
* The Apache Software License, Version 1.1
*
*
* Copyright (c) 1999-2002 The Apache Software Foundation. All rights
* reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
*
* 3. The end-user documentation included with the redistribution,
* if any, must include the following acknowledgment:
* "This product includes software developed by the
* Apache Software Foundation (http://www.apache.org/)."
* Alternately, this acknowledgment may appear in the software itself,
* if and wherever such third-party acknowledgments normally appear.
*
* 4. The names "Xerces" and "Apache Software Foundation" must
* not be used to endorse or promote products derived from this
* software without prior written permission. For written
* permission, please contact [email protected].
*
* 5. Products derived from this software may not be called "Apache",
* nor may "Apache" appear in their name, without prior written
* permission of the Apache Software Foundation.
*
* THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESSED OR IMPLIED
* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE APACHE SOFTWARE FOUNDATION OR
* ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
* USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
* ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
* OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
* ====================================================================
*
* This software consists of voluntary contributions made by many
* individuals on behalf of the Apache Software Foundation and was
* originally based on software copyright (c) 1999, International
* Business Machines, Inc., http://www.apache.org. For more
* information on the Apache Software Foundation, please see
* .
*/
package org.apache.xml.serialize;
import java.io.Writer;
import java.io.StringWriter;
import java.io.IOException;
/**
* Extends {@link Printer} and adds support for indentation and line
* wrapping.
*
* @version $Revision: 1.8 $ $Date: 2003/01/13 15:59:09 $
* @author Assaf Arkin
*/
public class IndentPrinter
extends Printer
{
/**
* Holds the currently accumulating text line. This buffer will constantly
* be reused by deleting its contents instead of reallocating it.
*/
private StringBuffer _line;
/**
* Holds the currently accumulating text that follows {@link #_line}.
* When the end of the part is identified by a call to {@link #printSpace}
* or {@link #breakLine}, this part is added to the accumulated line.
*/
private StringBuffer _text;
/**
* Counts how many white spaces come between the accumulated line and the
* current accumulated text. Multiple spaces at the end of the a line
* will not be printed.
*/
private int _spaces;
/**
* Holds the indentation for the current line that is now accumulating in
* memory and will be sent for printing shortly.
*/
private int _thisIndent;
/**
* Holds the indentation for the next line to be printed. After this line is
* printed, {@link #_nextIndent} is assigned to {@link #_thisIndent}.
*/
private int _nextIndent;
public IndentPrinter( Writer writer, OutputFormat format)
{
super( writer, format );
// Initialize everything for a first/second run.
_line = new StringBuffer( 80 );
_text = new StringBuffer( 20 );
_spaces = 0;
_thisIndent = _nextIndent = 0;
}
/**
* Called by any of the DTD handlers to enter DTD mode.
* Once entered, all output will be accumulated in a string
* that can be printed as part of the document's DTD.
* This method may be called any number of time but will only
* have affect the first time it's called. To exist DTD state
* and get the accumulated DTD, call {@link #leaveDTD}.
*/
public void enterDTD()
{
// Can only enter DTD state once. Once we're out of DTD
// state, can no longer re-enter it.
if ( _dtdWriter == null ) {
_line.append( _text );
_text = new StringBuffer( 20 );
flushLine( false );
_dtdWriter = new StringWriter();
_docWriter = _writer;
_writer = _dtdWriter;
}
}
/**
* Called by the root element to leave DTD mode and if any
* DTD parts were printer, will return a string with their
* textual content.
*/
public String leaveDTD()
{
// Only works if we're going out of DTD mode.
if ( _writer == _dtdWriter ) {
_line.append( _text );
_text = new StringBuffer( 20 );
flushLine( false );
_writer = _docWriter;
return _dtdWriter.toString();
} else
return null;
}
/**
* Called to print additional text. Each time this method is called
* it accumulates more text. When a space is printed ({@link
* #printSpace}) all the accumulated text becomes one part and is
* added to the accumulate line. When a line is long enough, it can
* be broken at its text boundary.
*
* @param text The text to print
*/
public void printText( String text )
{
_text.append( text );
}
public void printText( StringBuffer text )
{
_text.append( text.toString() );
}
public void printText( char ch )
{
_text.append( ch );
}
public void printText( char[] chars, int start, int length )
{
_text.append( chars, start, length );
}
/**
* Called to print a single space between text parts that may be
* broken into separate lines. Must not be called to print a space
* when preserving spaces. The text accumulated so far with {@link
* #printText} will be added to the accumulated line, and a space
* separator will be counted. If the line accumulated so far is
* long enough, it will be printed.
*/
public void printSpace()
{
// The line consists of the text accumulated in _line,
// followed by one or more spaces as counted by _spaces,
// followed by more space accumulated in _text:
// - Text is printed and accumulated into _text.
// - A space is printed, so _text is added to _line and
// a space is counted.
// - More text is printed and accumulated into _text.
// - A space is printed, the previous spaces are added
// to _line, the _text is added to _line, and a new
// space is counted.
// If text was accumulated with printText(), then the space
// means we have to move that text into the line and
// start accumulating new text with printText().
if ( _text.length() > 0 ) {
// If the text breaks a line bounary, wrap to the next line.
// The printed line size consists of the indentation we're going
// to use next, the accumulated line so far, some spaces and the
// accumulated text so far.
if ( _format.getLineWidth() > 0 &&
_thisIndent + _line.length() + _spaces + _text.length() > _format.getLineWidth() ) {
flushLine( false );
try {
// Print line and new line, then zero the line contents.
_writer.write( _format.getLineSeparator() );
} catch ( IOException except ) {
// We don't throw an exception, but hold it
// until the end of the document.
if ( _exception == null )
_exception = except;
}
}
// Add as many spaces as we accumulaed before.
// At the end of this loop, _spaces is zero.
while ( _spaces > 0 ) {
_line.append( ' ' );
--_spaces;
}
_line.append( _text );
_text = new StringBuffer( 20 );
}
// Starting a new word: accumulate the text between the line
// and this new word; not a new word: just add another space.
++_spaces;
}
/**
* Called to print a line consisting of the text accumulated so
* far. This is equivalent to calling {@link #printSpace} but
* forcing the line to print and starting a new line ({@link
* #printSpace} will only start a new line if the current line
* is long enough).
*/
public void breakLine()
{
breakLine( false );
}
public void breakLine( boolean preserveSpace )
{
// Equivalent to calling printSpace and forcing a flushLine.
if ( _text.length() > 0 ) {
while ( _spaces > 0 ) {
_line.append( ' ' );
--_spaces;
}
_line.append( _text );
_text = new StringBuffer( 20 );
}
flushLine( preserveSpace );
try {
// Print line and new line, then zero the line contents.
_writer.write( _format.getLineSeparator() );
} catch ( IOException except ) {
// We don't throw an exception, but hold it
// until the end of the document.
if ( _exception == null )
_exception = except;
}
}
/**
* Flushes the line accumulated so far to the writer and get ready
* to accumulate the next line. This method is called by {@link
* #printText} and {@link #printSpace} when the accumulated line plus
* accumulated text are two long to fit on a given line. At the end of
* this method _line is empty and _spaces is zero.
*/
public void flushLine( boolean preserveSpace )
{
int indent;
if ( _line.length() > 0 ) {
try {
if ( _format.getIndenting() && ! preserveSpace ) {
// Make sure the indentation does not blow us away.
indent = _thisIndent;
if ( ( 2 * indent ) > _format.getLineWidth() && _format.getLineWidth() > 0 )
indent = _format.getLineWidth() / 2;
// Print the indentation as spaces and set the current
// indentation to the next expected indentation.
while ( indent > 0 ) {
_writer.write( ' ' );
--indent;
}
}
_thisIndent = _nextIndent;
// There is no need to print the spaces at the end of the line,
// they are simply stripped and replaced with a single line
// separator.
_spaces = 0;
_writer.write( _line.toString() );
_line = new StringBuffer( 40 );
} catch ( IOException except ) {
// We don't throw an exception, but hold it
// until the end of the document.
if ( _exception == null )
_exception = except;
}
}
}
/**
* Flush the output stream. Must be called when done printing
* the document, otherwise some text might be buffered.
*/
public void flush()
{
if ( _line.length() > 0 || _text.length() > 0 )
breakLine();
try {
_writer.flush();
} catch ( IOException except ) {
// We don't throw an exception, but hold it
// until the end of the document.
if ( _exception == null )
_exception = except;
}
}
/**
* Increment the indentation for the next line.
*/
public void indent()
{
_nextIndent += _format.getIndent();
}
/**
* Decrement the indentation for the next line.
*/
public void unindent()
{
_nextIndent -= _format.getIndent();
if ( _nextIndent < 0 )
_nextIndent = 0;
// If there is no current line and we're de-identing then
// this indentation level is actually the next level.
if ( ( _line.length() + _spaces + _text.length() ) == 0 )
_thisIndent = _nextIndent;
}
public int getNextIndent()
{
return _nextIndent;
}
public void setNextIndent( int indent )
{
_nextIndent = indent;
}
public void setThisIndent( int indent )
{
_thisIndent = indent;
}
}