xmlparser.java

一个即时消息系统的源码（J2ME编程部分）。

package org.kxml.parser;

import java.io.*;
import java.util.*;

import org.kxml.*;
import org.kxml.io.*;


/** A simple, pull based "Common XML" parser. Attention: This class has
    been renamed from DefaultParser for consitency with the org.kxml.io
    package. */

public class XmlParser extends AbstractXmlParser {

    LookAheadReader reader;
    boolean relaxed;

    Vector qNames = new Vector ();

    boolean immediateClose = false;
    StartTag current;
    
    /** creates a new Parser based on the give reader */

    class DefaultParserException extends ParseException {
	DefaultParserException (String msg, Exception chained) {
	    super (msg, chained, 
		   XmlParser.this.reader.getLineNumber (),
		   XmlParser.this.reader.getColumnNumber ());
	}
    }


    public XmlParser (Reader reader) throws IOException {

	this.reader = new LookAheadReader (reader); 
    }



    public String resolveCharacterEntity (String name) throws IOException {

	throw new DefaultParserException 
	    ("Unknown character entity: &"+name+";", null);
    }

    
    /* precondition: <!- consumed */
    
    LegacyEvent parseComment () throws IOException {
	   
	StringBuffer buf = new StringBuffer ();
	
	if (reader.read () != '-') 
	    throw new DefaultParserException ("'-' expected", null);

	int cnt;
	int lst;

	while (true) {
	    buf.append (reader.readTo ('-'));
	
	    if (reader.read () == -1) 
		throw new DefaultParserException 
		    ("Unexpected end of file while reading comment", null); 

	    cnt = 0;

	    do {
		lst = reader.read ();
		cnt++;  // adds one more, but first is not cnted
	    }
	    while (lst == '-');

	    if (lst == '>' && cnt >= 2) break;

	    while (cnt-- > 0) 
		buf.append ('-');

	    buf.append ((char) lst);
	}

	while (cnt-- > 2) 
	    buf.append ('-');
	       
	return new LegacyEvent (Xml.COMMENT, buf.toString ());
    }


    /* precondition: &lt! consumed */ 

    
    LegacyEvent parseDoctype () throws IOException {

	StringBuffer buf = new StringBuffer ();
	int nesting = 1;

	while (true) {
	    int i = reader.read ();
	    switch (i) {
	    case -1:
		throw new DefaultParserException ("unexpected eof", null);
	    case '<': 
		nesting++;
		break;
	    case '>':
		if ((--nesting) == 0)
		    return new LegacyEvent (Xml.DOCTYPE, buf.toString ());
		break;
	    }
	    buf.append ((char) i);
	}
    }


    TextEvent parseCData () throws IOException {
	
	if (!reader.readTo ('[').equals ("CDATA")) 
	    throw new DefaultParserException ("Invalid CDATA start!", null);

	reader.read (); // skip '['

	StringBuffer buf = new StringBuffer ();

	int c0 = reader.read ();
	int c1 = reader.read ();

	while (true) {
	    int c2 = reader.read ();

	    if (c2 == -1) 
		throw new DefaultParserException ("unexpected eof", null);

	    if (c0 == ']' && c1 == ']' && c2 == '>') 
		break;

	    buf.append (c0);
	    c0 = c1;
	    c1 = c2;
	}
	    
	return new TextEvent (Xml.TEXT, buf.toString ());
    }



    /* precondition: &lt;/ consumed */

    ParseEvent parseEndTag () throws IOException {

	String name = reader.readTo ('>').trim ();

	if (reader.read () != '>') 
	    throw new DefaultParserException 
		("EndTag </"+name
		 +"> not terminated by '>'", null);

	
	int last = qNames.size ();
	while (true) {
	    if (last == 0) {
		if (relaxed) return new EndDocument ();
		throw new DefaultParserException 
		    ("tagstack empty parsing </"+name+">", null);
	    }
	    String qName = (String) qNames.elementAt (--last);
	    qNames.removeElementAt (last);

	    if (qName.equals (name)) break; 
	    if (!relaxed) throw new DefaultParserException 
		("StartTag <"+qName
		 +"> does not match end tag </" + name + ">", null);
	    if (qName.toLowerCase ().equals (name.toLowerCase ())) break;
	    current = current.parent;
	}

	EndTag result = new EndTag (current);

	current = current.parent;

	return result;
    }

    /** precondition: <? consumed */

    ParseEvent parsePI () throws IOException {
	StringBuffer buf = new StringBuffer (reader.readTo ('?'));
	reader.read (); // ?
	
	while (reader.peek () != '>') {
	    buf.append ('?');
	    
	    int r = reader.read ();
	    if (r == -1) throw new DefaultParserException 
		("Unexpected eof while reading PI", null);

	    buf.append ((char) r);
	    buf.append (reader.readTo ('?'));
	    reader.read ();
	}
	
	reader.read ();  // consume >

	return new LegacyEvent 
	    (Xml.PROCESSING_INSTRUCTION, buf.toString ());
    }


    StartTag parseStartTag () throws IOException {
	current = new StartTag (current, reader, relaxed);
	String qName = current.name;
	
	if (processNamespaces) {
	    try {
		current.fixNamespaces ();
	    }
	    catch (Exception e) {
		throw new DefaultParserException (null, e);
	    }
	}
	//System.out.println ("tag: ("+next+")");
	
	immediateClose = current.degenerated;
	if (!immediateClose)
	    qNames.addElement (qName);
	
	return current;
    }

    
    TextEvent parseText () throws IOException {
	
	StringBuffer buf = new StringBuffer ();
	int nextChar;
	int type = Xml.WHITESPACE;

	do {
	    nextChar = reader.read ();
	    
	    if (nextChar == '&') {
		String code = reader.readTo (';');
		reader.read ();
	
		if (code.charAt (0) == '#') 
		    buf.append (code.charAt (1) == 'x' 
			? (char) Integer.parseInt (code.substring (2), 16) 
			: (char) Integer.parseInt (code.substring (1)));
		else if (code.equals ("lt")) buf.append ('<');
		else if (code.equals ("gt")) buf.append ('>');
		else if (code.equals ("apos")) buf.append ('\'');
		else if (code.equals ("quot")) buf.append ('"');
		else if (code.equals ("amp")) buf.append ('&');
		else buf.append (resolveCharacterEntity (code));
	    }
	    else {
		if (nextChar > ' ') 
		    type = Xml.TEXT;
		buf.append ((char) nextChar);
	    }
	    nextChar = reader.peek ();
	}
	while (nextChar != '<' && nextChar != -1);
	
	return new TextEvent (type, buf.toString ());
    }    


    /** precondition: &lt; consumed */

    ParseEvent parseSpecial () throws IOException {

	switch  (reader.peek ()) {
	case -1: throw new DefaultParserException 
	    ("Unexpected end of file after reading <", null); 

	case '!':
	    reader.read ();
	    switch (reader.peek ()) {
	    case '-': 
		reader.read ();
		return parseComment ();

	    case '[':
		reader.read ();
		return parseCData ();

	    default: 
	        return parseDoctype ();
	    }
	    
	case '?':
	    reader.read ();
	    return parsePI ();

	case '/': 
	    reader.read ();
	    return parseEndTag ();

	default: 
	    return parseStartTag ();
	}
    }




    protected ParseEvent readImpl () throws IOException {
	
	if (immediateClose) {
	    ParseEvent result = new EndTag (current);
	    current = current.getParent ();
	    immediateClose = false;
	    return result;
	}
	
	switch (reader.peek ()) {
		
	case '<': 
	    reader.read ();
	    return parseSpecial ();
	    
	case -1: 
	    if (current != null && !relaxed) 
		throw new DefaultParserException 
		    ("End tag missing for: "+current, null);
	    return new EndDocument (); 

	default: 
	    return parseText ();
	}
    }
    
    /** default is false. Setting relaxed true
	allows CHTML parsing */

    public void setRelaxed (boolean relaxed) {
	this.relaxed = relaxed;
    }

    public int getLineNumber () {
	return reader.getLineNumber ();
    }
}