wmlcparser.java

手机Wap浏览器源码

package wmlcparser;
import java.io.ByteArrayOutputStream;
import java.io.IOException;
import java.io.UnsupportedEncodingException;
import java.util.Stack;

/**
 * history:
 * 2007-11-23
 * - 将测试的类移出wmlcparser包, 打包的时候不打wmlcparser以外的类
 * - add method: getCharsetStr()
 * - 修改 getStrInCharset(), 自己处理Exception
 * 
 * 2007-11-20
 * - fix get string from table by index @see // 2007-11-20 fay: forget to reset it - -!
 * 
 * 2007-11-08
 * - from peter: change utf-16 to utf-8 manual 
 * 
 * 2007-11-07:
 * - 增加对mb_int32的处理(加多了buffer_4_mb_int32)
 * - 对charset进行了处理, 目前四种,
 * 	 "UTF-8", "UTF-16", "UTF-32", 和平台默认编码
 * 
 * 
 * - 两类TOKEN -> global token 和 application token
 * - application token 包括(Tag Token 和 Attribute Token)
 * - Attribute Start(<128) 和 Attribute Value(>128)
 * 
 * 
 * 
 * @author Fay
 *
 */
public class WmlcParser
{
	static final int STATUS_TAG = 1;
	static final int STATUS_ATTRIBUTE = 2;
	
	private Stack tagStack = new Stack();
	private int currentStatus = STATUS_TAG;
	
	private ByteArrayOutputStream baos = new ByteArrayOutputStream();	// buffer
	private ByteArrayOutputStream buffer_4_mb_int32 = new ByteArrayOutputStream();	// the buffer for handling the mb_int32 
	private int preGlobalToken = -1;
	private int preAttributeStart = -1;
	
	
	private byte wbxmlVersion = -1;
	private int xmlVersion = -1;
	private int charset = -1;
	private byte[] stringTable = null;
	private String charsetStr = null;
	
/*	public Tag parseWmlc(byte[] bs) throws IOException 
	{
		if(bs == null || bs.length == 0) { return null; }
    	//#ifdef debug
		//# System.out.println("parseWmlc, bs.length == " + bs.length);
		//#endif
		
		// 0. wbxml version
		// 1. wml version	(mb_u_int32)
		// 2. charset		(mb_u_int32)
		// 3. table length	(mb_u_int32)
		// 4. table(if table length != 0)
		
		wbxmlVersion = bs[0];			// 0. wbxml version
    	//#ifdef debug
		//# System.out.println("wbxmlVersion == " + wbxmlVersion);
		//#endif
		
		int pos = 1;
		
		// 1. wml version	(mb_u_int32)
		pos = parseWmlVersion(bs, pos);
    	//#ifdef debug
		//# System.out.println("xmlVersion == " + xmlVersion);
		//#endif
		
		// 2. charset		(mb_u_int32)
		pos = parseWmlCharset(bs, pos);
    	//#ifdef debug
		//# System.out.println("charset == " + charsetStr);
		//#endif
		
		// 3. table length	(mb_u_int32)
		// 4. table(if table length != 0)
		pos = parseWmlTableStr(bs, pos);
		
		// for debug!
    	//#ifdef debug
		//# System.out.println("tbl == " + (stringTable==null?0:stringTable.length));
		//# for(int i = 0 ; stringTable!= null && i < stringTable.length; i ++)
		//# {
			//# System.out.println("stringTable[" + i + "]=" + stringTable[i]);
		//# }
		//#endif

		int b = 0;
		for(int i = pos; i < bs.length ; i ++)
		{
			// 处理byte
			b = bs[i];
			if(b < 0)
			{
				b += 256;
			}
        	//#ifdef debug
			//# System.out.println("bs[" + i + "]=" + Integer.toHexString(b));
			//#endif
			
			if(preGlobalToken != -1)	// 有global token未处理
			{
				handleGlobalToken(b);
            	//#ifdef debug
				//# System.out.println("continue 0");
				//#endif
				continue;
			}
			
			// 是需要处理(有follow的)global token -> continue
			if(Token.isGlobalTokenNeedHandle(b))
			{
				preGlobalToken = b;
            	//#ifdef debug
				//# System.out.println("continue 1");
				//#endif
				continue;
			}
			
			// 是Global Token 而且不是END(1), 就Continue, END会由后面处理
			if(Token.isGlobalToken(b) && (b != Token.GLOBAL_END))
			{
            	//#ifdef debug
				//# System.out.println("continue 2");
				//#endif
				continue;
			}
			
			// 处理byte
			switch(currentStatus)
			{
			case STATUS_TAG:
            	//#ifdef debug
				//# System.out.println("handleTag -> " + b);
				//#endif
				handle_Tag_Status(b);
				break;
				
			case STATUS_ATTRIBUTE:
            	//#ifdef debug
				//# System.out.println("handle attribute -> " + b);
				//#endif
				handle_Attribute_Status(b);
				break;
			}
		}
		
		return rootTag;
	}
*/
	
	/**
	 * 2007-11-22 new parseWmlc, return the Tag.toString()
	 */
	public String parseWmlc(byte[] bs) throws Exception 
	{
		if(bs == null || bs.length == 0) { return null; }
    	//#ifdef debug
		//# System.out.println("parseWmlc, bs.length == " + bs.length);
		//#endif
		
		// 0. wbxml version
		// 1. wml version	(mb_u_int32)
		// 2. charset		(mb_u_int32)
		// 3. table length	(mb_u_int32)
		// 4. table(if table length != 0)
		
		wbxmlVersion = bs[0];			// 0. wbxml version
    	//#ifdef debug
		//# System.out.println("wbxmlVersion == " + wbxmlVersion);
		//#endif
		
		int pos = 1;
		
		// 1. wml version	(mb_u_int32)
		pos = parseWmlVersion(bs, pos);
    	//#ifdef debug
		//# System.out.println("xmlVersion == " + xmlVersion);
		//#endif
		
		// 2. charset		(mb_u_int32)
		pos = parseWmlCharset(bs, pos);
    	//#ifdef debug
		//# System.out.println("charset == " + charsetStr);
		//#endif
		
		// 3. table length	(mb_u_int32)
		// 4. table(if table length != 0)
		pos = parseWmlTableStr(bs, pos);
		
		// for debug!
    	//#ifdef debug
		//# System.out.println("tbl == " + (stringTable==null?0:stringTable.length));
		//# for(int i = 0 ; stringTable!= null && i < stringTable.length; i ++)
		//# {
			//# System.out.println("stringTable[" + i + "]=" + stringTable[i]);
		//# }
		//#endif

		int b = 0;
		for(int i = pos; i < bs.length ; i ++)
		{
			// 处理byte
			b = bs[i];
			if(b < 0)
			{
				b += 256;
			}
        	//#ifdef debug
			//# System.out.println("bs[" + i + "]=" + Integer.toHexString(b));
			//#endif
			
			if(preGlobalToken != -1)	// 有global token未处理
			{
				handleGlobalToken(b);
            	//#ifdef debug
				//# System.out.println("continue 0");
				//#endif
				continue;
			}
			
			// 是需要处理(有follow的)global token -> continue
			if(Token.isGlobalTokenNeedHandle(b))
			{
				preGlobalToken = b;
            	//#ifdef debug
				//# System.out.println("continue 1");
				//#endif
				continue;
			}
			
			// 是Global Token 而且不是END(1), 就Continue, END会由后面处理
			if(Token.isGlobalToken(b) && (b != Token.GLOBAL_END))
			{
            	//#ifdef debug
				//# System.out.println("continue 2");
				//#endif
				continue;
			}
			
			// 处理byte
			switch(currentStatus)
			{
			case STATUS_TAG:
            	//#ifdef debug
				//# System.out.println("handleTag -> " + b);
				//#endif
				handle_Tag_Status(b);
				break;
				
			case STATUS_ATTRIBUTE:
            	//#ifdef debug
				//# System.out.println("handle attribute -> " + b);
				//#endif
				handle_Attribute_Status(b);
				break;
			}
		}
		
		if(rootTag == null) 
		{ 
			return null; 
		}else
		{
			return rootTag.toString();	
		}
	}
	
	
	/**
	 * 将global token和后继转成相应的String, 存在buffer里.
	 * <br>
	 * 每一个global token后面跟的byte是一定的!
	 * <br>
	 * 记得清空preGlobalToken. preGlobalToken = -1;
	 * 
	 * @param b
	 * @throws IOException 
	 */
	private void handleGlobalToken(int b) throws Exception 
	{
    	//#ifdef debug
		//# System.out.println("handleGlobalToken -> " + b);
		//#endif
		switch(preGlobalToken)
		{
		case Token.GLOBAL_ENTITY:	// 未处理(mb_u_int32)
			// 2007-11-07, fay: 增加对Multi-byte Integers的判断
			if(isMB_U_INT32End(b))
			{
				preGlobalToken = -1;
			}
			break;
		case Token.GLOBAL_STRI_I:	// Followed by a termstr.	
			if(b != 0)	// b == 0 means string end
			{
				baos.write(b);
			}else
			{
            	//#ifdef debug
				//# System.out.println("STRI END -> " + getStrInCharset());
				//#endif
				preGlobalToken = -1;
			}
			break;
		case Token.GLOBAL_LITERAL:	// 未处理(mb_u_int32)
			// 2007-11-07, fay: 增加对Multi-byte Integers的判断
			if(isMB_U_INT32End(b))
			{
				preGlobalToken = -1;
			}
			break;
		case Token.GLOBAL_EXT_I_0:	// (termStr)
		case Token.GLOBAL_EXT_I_1:
		case Token.GLOBAL_EXT_I_2:
			if(b == 0)
			{
				String str = getStrInCharset();
				baos.reset();
				baos.write(("$(" + str + ")").getBytes());		// TODO handle the charset
				preGlobalToken = -1;
			}else
			{
				baos.write(b);
			}
			break;
		case Token.GLOBAL_EXT_T_0:	// (mb_u_int32)
		case Token.GLOBAL_EXT_T_1:	// str= table[b], buffer = "$(str)"
		case Token.GLOBAL_EXT_T_2:
			// 2007-11-07, fay: 增加对Multi-byte Integers的判断
			buffer_4_mb_int32.write(b);
			if(!isMB_U_INT32End(b))
			{
				break;			// write it to buffer, and return!
			}
			// 取出table的数据,
			// $(str)
			int tableIndex = mergeMB_U_INT32(buffer_4_mb_int32.toByteArray());
			String tableData = getStrFromStringTable(tableIndex);			// TODO check null!
        	//#ifdef debug
			//# System.out.println("tableData == " + tableData);
			//#endif
			
			baos.write(("$(" + tableData + ")").getBytes());				// TODO handle the charset
			
        	//#ifdef debug
			//# System.out.println("buffer  == " + getStrInCharset());
			//# 
			//#endif
			buffer_4_mb_int32.reset();		// reset the buffer, it's important!
			preGlobalToken = -1;
			break;
		case Token.GLOBAL_STR_T:	// (mb_u_int32)
			// 2007-11-07, fay: 增加对Multi-byte Integers的判断
			buffer_4_mb_int32.write(b);
			if(!isMB_U_INT32End(b))
			{
				break;			// write it to buffer, and return!
			}
			
			tableIndex = mergeMB_U_INT32(buffer_4_mb_int32.toByteArray());
			tableData = getStrFromStringTable(tableIndex);	// TODO check null!
			if(tableData == null)
			{
            	//#ifdef debug
				//# System.out.println("tableData == null");
				//#endif
			}else
			{
				baos.write(tableData.getBytes());			// TODO handle the charset
			}
			
			// 2007-11-20 fay: forget to reset it - -!
			buffer_4_mb_int32.reset();		
			preGlobalToken = -1;
			break;
			
			
		}
	}
	
	Tag rootTag = null;
	/**
	 * handle the b when it's TAG_STATUS
	 * @param b
	 * @throws UnsupportedEncodingException 
	 */
	private void handle_Tag_Status(int b) throws Exception
	{
		// Tag结束标记
		if(b == Token.GLOBAL_END)
		{
			if(tagStack.size() == 0)
			{
				// error, 没有tag, 但碰到了 TAG END
            	//#ifdef debug
				//# System.out.println("error, 没有tag, 但碰到了 TAG END");