rtfparser.java

源码包含生成 PDF 和 HTML 的类库

	 * Total control words processed.
	 * 
	 * Contains both known and unknown.
	 * 
	 * <code>ctrlWordCount</code> should equal 
	 * <code>ctrlWrodHandlecCount</code> + <code>ctrlWordNotHandledCount</code + <code>ctrlWordSkippedCount</code>
	 */
	private long ctrlWordCount = 0;
	/**
	 * Total { encountered as an open group token.
	 */
	private long openGroupCount = 0;
	/**
	 * Total } encountered as a close group token.
	 */
	private long closeGroupCount = 0;
	/**
	 * Total clear text characters processed.
	 */
	private long characterCount = 0;
	/**
	 * Total control words recognized.
	 */
	private long ctrlWordHandledCount = 0;
	/**
	 * Total control words not handled.
	 */
	private long ctrlWordNotHandledCount = 0;
	/**
	 * Total control words skipped.
	 */
	private long ctrlWordSkippedCount = 0;
	/**
	 * Total groups skipped. Includes { and } as a group.
	 */
	private long groupSkippedCount = 0;
	/**
	 * Start time as a long.
	 */
	private long startTime = 0;
	/**
	 * Stop time as a long.
	 */
	private long endTime = 0;
	/**
	 * Start date as a date.
	 */
	private Date startDate = null;
	/**
	 * End date as a date.
	 */
	private Date endDate = null;
	//////////////////////////////////// STATS VARIABLES ///////////////////
	/**
	 * Last control word and parameter processed.
	 */
	private RtfCtrlWordData lastCtrlWordParam = null;
	
	/** The <code>RtfCtrlWordListener</code>. */
    private ArrayList listeners = new ArrayList();
    
	/**
	 * Constructor 
	 * @param doc
	 * @since 2.1.3
	 */
    public RtfParser(Document doc) {
    	this.document = doc;
    }
	/* *********
	 *  READER *
	 ***********/
	/**
	 * Imports a complete RTF document.
	 * 
	 * @param readerIn 
	 * 		The Reader to read the RTF document from.
	 * @param rtfDoc 
	 * 		The RtfDocument to add the imported document to.
	 * @throws IOException On I/O errors.
	 *  @since 2.1.3
	 */
	public void importRtfDocument(InputStream readerIn, RtfDocument rtfDoc) throws IOException {
		if(readerIn == null || rtfDoc == null) return;
		this.init(TYPE_IMPORT_FULL, rtfDoc, readerIn, this.document, null);
		this.setCurrentDestination(RtfDestinationMgr.DESTINATION_NULL);
		startDate = new Date();
		startTime = System.currentTimeMillis();
		this.groupLevel = 0;
		try {
			this.tokenise();
		} catch (RuntimeException e) {
			// TODO Auto-generated catch block
			e.printStackTrace();
		}
		catch (Exception e) {
			// TODO Auto-generated catch block
			e.printStackTrace();
		}
		endTime = System.currentTimeMillis();
		endDate = new Date();
	}
	/**
	 * Imports a complete RTF document into an Element, i.e. Chapter, section, Table Cell, etc.
	 * 
	 * @param elem The Element the document is to be imported into.
	 * @param readerIn 
	 * 		The Reader to read the RTF document from.
	 * @param rtfDoc 
	 * 		The RtfDocument to add the imported document to.
	 * @throws IOException On I/O errors.
	 * @since 2.1.4
	 */
	public void importRtfDocumentIntoElement(Element elem, InputStream readerIn, RtfDocument rtfDoc) throws IOException {
		if(readerIn == null || rtfDoc == null || elem == null) return;
		this.init(TYPE_IMPORT_INTO_ELEMENT, rtfDoc, readerIn, this.document, elem);
		this.setCurrentDestination(RtfDestinationMgr.DESTINATION_NULL);
		startDate = new Date();
		startTime = System.currentTimeMillis();
		this.groupLevel = 0;
		try {
			this.tokenise();
		} catch (RuntimeException e) {
			// TODO Auto-generated catch block
			e.printStackTrace();
		}
		catch (Exception e) {
			// TODO Auto-generated catch block
			e.printStackTrace();
		}
		endTime = System.currentTimeMillis();
		endDate = new Date();
	}
	/**
	 * Converts an RTF document to an iText document.
	 * 
	 * Usage: Create a parser object and call this method with the input stream and the iText Document object
	 * 
	 * @param readerIn 
	 * 		The Reader to read the RTF file from.
	 * @param doc 
	 * 		The iText document that the RTF file is to be added to.
	 * @throws IOException 
	 * 		On I/O errors.
	 *  @since 2.1.3
	 */
	public void convertRtfDocument(InputStream readerIn, Document doc) throws IOException {
		if(readerIn == null || doc == null) return;
		this.init(TYPE_CONVERT, null, readerIn, doc, null);
		this.setCurrentDestination(RtfDestinationMgr.DESTINATION_DOCUMENT);
		startDate = new Date();
		startTime = System.currentTimeMillis();
		this.groupLevel = 0;
		this.tokenise();
		endTime = System.currentTimeMillis();
		endDate = new Date();
	}

	/**
	 * Imports an RTF fragment.
	 * 
	 * @param readerIn 
	 * 		The Reader to read the RTF fragment from.
	 * @param rtfDoc 
	 * 		The RTF document to add the RTF fragment to.
	 * @param importMappings 
	 * 		The RtfImportMappings defining font and color mappings for the fragment.
	 * @throws IOException 
	 * 		On I/O errors.
	 *   @since 2.1.3
	 */
	public void importRtfFragment(InputStream readerIn, RtfDocument rtfDoc, RtfImportMappings importMappings) throws IOException {
	//public void importRtfFragment2(Reader readerIn, RtfDocument rtfDoc, RtfImportMappings importMappings) throws IOException {
		if(readerIn == null || rtfDoc == null || importMappings==null) return;
		this.init(TYPE_IMPORT_FRAGMENT, rtfDoc, readerIn, null, null);
		this.handleImportMappings(importMappings);
		this.setCurrentDestination(RtfDestinationMgr.DESTINATION_DOCUMENT);
		this.groupLevel = 1;
		setParserState(RtfParser.PARSER_IN_DOCUMENT);
		startDate = new Date();
		startTime = System.currentTimeMillis();
		this.tokenise();
		endTime = System.currentTimeMillis();
		endDate = new Date();
	}
	
    // listener methods

	/**
	 * Adds a <CODE>EventListener</CODE> to the <CODE>RtfCtrlWordMgr</CODE>.
	 *
	 * @param listener
	 *            the new EventListener.
	 * @since 2.1.3
	 */
	public void addListener(EventListener listener) {
		listeners.add(listener);
	}

	/**
	 * Removes a <CODE>EventListener</CODE> from the <CODE>RtfCtrlWordMgr</CODE>.
	 *
	 * @param listener
	 *            the EventListener that has to be removed.
	 *  @since 2.1.3
	 */
	public void removeListener(EventListener listener) {
		listeners.remove(listener);
	}

	/**
	 * Initialize the parser object values. 
	 * 
	 * @param type Type of conversion or import
	 * @param rtfDoc The <code>RtfDocument</code>
	 * @param readerIn The input stream
	 * @param doc The iText <code>Document</code>
	 *   @since 2.1.3
	 */
	private void init(int type, RtfDocument rtfDoc, InputStream readerIn, Document doc, Element elem) {

		init_stats();
		// initialize reader to a PushbackReader
		this.pbReader = init_Reader(readerIn);
		
		this.conversionType = type;
		this.rtfDoc = rtfDoc;
		this.document = doc;
		this.elem = elem;
		this.currentState = new RtfParserState();
		this.stackState = new Stack();
		this.setParserState(PARSER_STARTSTOP);
		this.importMgr = new RtfImportMgr(this.rtfDoc, this.document);

		// get destination Mgr
		this.destinationMgr = RtfDestinationMgr.getInstance(this);
		// set the parser
		RtfDestinationMgr.setParser(this);


		// DEBUG INFO for timing and memory usage of RtfCtrlWordMgr object
		// create multiple new RtfCtrlWordMgr objects to check timing and memory usage
//		System.gc();
//		long endTime = 0;
//		Date endDate = null;		
//		long endFree = 0;
//		DecimalFormat df = new DecimalFormat("#,##0");
//		Date startDate = new Date();
//		long startTime = System.currentTimeMillis();
//		long startFree = Runtime.getRuntime().freeMemory();
//		System.out.println("1:");
		
		this.rtfKeywordMgr = new RtfCtrlWordMgr(this, this.pbReader);/////////DO NOT COMMENT OUT THIS LINE ///////////
		
		Object listener;
		for (Iterator iterator = listeners.iterator(); iterator.hasNext();) {
            listener = iterator.next();
            if(listener instanceof RtfCtrlWordListener) {
                this.rtfKeywordMgr.addRtfCtrlWordListener((RtfCtrlWordListener)listener);    
            }
        }
//		endFree = Runtime.getRuntime().freeMemory();
//		endTime = System.currentTimeMillis();
//		endDate = new Date();
//		System.out.println("RtfCtrlWordMgr start date: " + startDate.toLocaleString());
//		System.out.println("RtfCtrlWordMgr end date  : " + endDate.toLocaleString());
//		System.out.println("  Elapsed time    : " + Long.toString(endTime - startTime) + " milliseconds.");
//		System.out.println("Begin Constructor RtfCtrlWordMgr , free mem is " + df.format(startFree / 1024) + "k");
//		System.out.println("End Constructor RtfCtrlWordMgr , free mem is " + df.format(endFree / 1024) + "k");
//		System.out.println("RtfCtrlWordMgr used approximately " + df.format((startFree - endFree) / 1024) + "k");
//		
//		System.gc();
//		System.out.println("2:");
//		startDate = new Date();
//		startTime = System.currentTimeMillis();
//		startFree = Runtime.getRuntime().freeMemory();
//		RtfCtrlWordMgr rtfKeywordMgr2 = new RtfCtrlWordMgr(this, this.pbReader);		
//		endFree = Runtime.getRuntime().freeMemory();
//		endTime = System.currentTimeMillis();
//		endDate = new Date();
//		System.out.println("RtfCtrlWordMgr start date: " + startDate.toLocaleString());
//		System.out.println("RtfCtrlWordMgr end date  : " + endDate.toLocaleString());
//		System.out.println("  Elapsed time    : " + Long.toString(endTime - startTime) + " milliseconds.");
//		System.out.println("Begin Constructor RtfCtrlWordMgr , free mem is " + df.format(startFree / 1024) + "k");
//		System.out.println("End Constructor RtfCtrlWordMgr , free mem is " + df.format(endFree / 1024) + "k");
//		System.out.println("RtfCtrlWordMgr used approximately " + df.format((startFree - endFree) / 1024) + "k");
//		
//		System.gc();
//		System.out.println("3:");
//		startDate = new Date();
//		startTime = System.currentTimeMillis();
//		startFree = Runtime.getRuntime().freeMemory();
//		RtfCtrlWordMgr rtfKeywordMgr3 = new RtfCtrlWordMgr(this, this.pbReader);	
//		endFree = Runtime.getRuntime().freeMemory();
//		endTime = System.currentTimeMillis();
//		endDate = new Date();
//		System.out.println("RtfCtrlWordMgr start date: " + startDate.toLocaleString());
//		System.out.println("RtfCtrlWordMgr end date  : " + endDate.toLocaleString());
//		System.out.println("  Elapsed time    : " + Long.toString(endTime - startTime) + " milliseconds.");
//		System.out.println("Begin Constructor RtfCtrlWordMgr , free mem is " + df.format(startFree / 1024) + "k");
//		System.out.println("End Constructor RtfCtrlWordMgr , free mem is " + df.format(endFree / 1024) + "k");
//		System.out.println("RtfCtrlWordMgr used approximately " + df.format((startFree - endFree) / 1024) + "k");
//
//		System.gc();
//		System.out.println("4:");
//		startDate = new Date();
//		startTime = System.currentTimeMillis();
//		startFree = Runtime.getRuntime().freeMemory();
//		RtfCtrlWordMgr rtfKeywordMgr4 = new RtfCtrlWordMgr(this, this.pbReader);	
//		endFree = Runtime.getRuntime().freeMemory();
//		endTime = System.currentTimeMillis();
//		endDate = new Date();
//		System.out.println("RtfCtrlWordMgr start date: " + startDate.toLocaleString());
//		System.out.println("RtfCtrlWordMgr end date  : " + endDate.toLocaleString());
//		System.out.println("  Elapsed time    : " + Long.toString(endTime - startTime) + " milliseconds.");
//		System.out.println("Begin Constructor RtfCtrlWordMgr , free mem is " + df.format(startFree / 1024) + "k");
//		System.out.println("End Constructor RtfCtrlWordMgr , free mem is " + df.format(endFree / 1024) + "k");
//		System.out.println("RtfCtrlWordMgr used approximately " + df.format((startFree - endFree) / 1024) + "k");
//
//		System.gc();
//		System.out.println("5:");
//		startDate = new Date();
//		startTime = System.currentTimeMillis();
//		startFree = Runtime.getRuntime().freeMemory();
//		RtfCtrlWordMgr rtfKeywordMgr5 = new RtfCtrlWordMgr(this, this.pbReader);	
//		endFree = Runtime.getRuntime().freeMemory();
//		endTime = System.currentTimeMillis();
//		endDate = new Date();
//		System.out.println("RtfCtrlWordMgr start date: " + startDate.toLocaleString());
//		System.out.println("RtfCtrlWordMgr end date  : " + endDate.toLocaleString());
//		System.out.println("  Elapsed time    : " + Long.toString(endTime - startTime) + " milliseconds.");
//		System.out.println("Begin Constructor RtfCtrlWordMgr , free mem is " + df.format(startFree / 1024) + "k");
//		System.out.println("End Constructor RtfCtrlWordMgr , free mem is " + df.format(endFree / 1024) + "k");
//		System.out.println("RtfCtrlWordMgr used approximately " + df.format((startFree - endFree) / 1024) + "k");
//		System.gc();
//		System.out.println("At ed:");
//		startDate = new Date();
//		startTime = System.currentTimeMillis();
//		startFree = Runtime.getRuntime().freeMemory();
//		//RtfCtrlWordMgr rtfKeywordMgr6 = new RtfCtrlWordMgr(this, this.pbReader);	
//		endFree = Runtime.getRuntime().freeMemory();
//		endTime = System.currentTimeMillis();
//		endDate = new Date();
//		System.out.println("RtfCtrlWordMgr start date: " + startDate.toLocaleString());
//		System.out.println("RtfCtrlWordMgr end date  : " + endDate.toLocaleString());
//		System.out.println("  Elapsed time    : " + Long.toString(endTime - startTime) + " milliseconds.");
//		System.out.println("Begin Constructor RtfCtrlWordMgr , free mem is " + df.format(startFree / 1024) + "k");
//		System.out.println("End Constructor RtfCtrlWordMgr , free mem is " + df.format(endFree / 1024) + "k");
//		System.out.println("RtfCtrlWordMgr used approximately " + df.format((startFree - endFree) / 1024) + "k");
	}
	/**
	 * Initialize the statistics values.
	 * @since 2.1.3
	 */
	protected void init_stats() {
		byteCount = 0;
		ctrlWordCount = 0;
		openGroupCount = 0;
		closeGroupCount = 0;
		characterCount = 0;
		ctrlWordHandledCount = 0;
		ctrlWordNotHandledCount = 0;
		ctrlWordSkippedCount = 0;
		groupSkippedCount = 0;
		startTime = 0;
		endTime = 0;
		startDate = null;
		endDate = null;
	}
	
	/**
	 * Casts the input reader to a PushbackReader or 
	 * creates a new PushbackReader from the Reader passed in.
	 * The reader is also transformed into a BufferedReader if necessary.
	 * 
	 * @param readerIn
	 * 		The Reader object for the input file.
	 * @return
	 * 		PushbackReader object
	 * @since 2.1.3
	 */
	private PushbackInputStream init_Reader(InputStream readerIn) {
//		Reader newReader = readerIn;
//		// Initializing the reader as a BufferedReader 
//		// cut test processing time by approximately 50%
//		// default uses 8192 character buffer
//		if(!(newReader instanceof BufferedReader)) {
//			newReader = new BufferedReader(newReader);	// Since JDK1.1
//		}
//		// Initializing the reader as a PushbackReader is
//		// a requirement of the parser to be able to put back
//		// read ahead characters.
//		if(!(newReader instanceof PushbackReader)) {
//			newReader = new PushbackReader(newReader);	// Since JDK1.1
//		}
		
		if(!(readerIn instanceof BufferedInputStream)) {
			readerIn = new BufferedInputStream(readerIn);
		}
		if(!(readerIn instanceof PushbackInputStream)) {
			readerIn = new PushbackInputStream(readerIn);
		}