michael@0: /*
michael@0:  * $Id: JSONParser.java,v 1.1 2006/04/15 14:10:48 platform Exp $
michael@0:  * Created on 2006-4-15
michael@0:  */
michael@0: package org.json.simple.parser;
michael@0: 
michael@0: import java.io.IOException;
michael@0: import java.io.Reader;
michael@0: import java.io.StringReader;
michael@0: import java.util.LinkedList;
michael@0: import java.util.List;
michael@0: import java.util.Map;
michael@0: 
michael@0: import org.json.simple.JSONArray;
michael@0: import org.json.simple.JSONObject;
michael@0: 
michael@0: 
michael@0: /**
michael@0:  * Parser for JSON text. Please note that JSONParser is NOT thread-safe.
michael@0:  * 
michael@0:  * @author FangYidong<fangyidong@yahoo.com.cn>
michael@0:  */
michael@0: public class JSONParser {
michael@0: 	public static final int S_INIT=0;
michael@0: 	public static final int S_IN_FINISHED_VALUE=1;//string,number,boolean,null,object,array
michael@0: 	public static final int S_IN_OBJECT=2;
michael@0: 	public static final int S_IN_ARRAY=3;
michael@0: 	public static final int S_PASSED_PAIR_KEY=4;
michael@0: 	public static final int S_IN_PAIR_VALUE=5;
michael@0: 	public static final int S_END=6;
michael@0: 	public static final int S_IN_ERROR=-1;
michael@0: 	
michael@0: 	private LinkedList handlerStatusStack;
michael@0: 	private Yylex lexer = new Yylex((Reader)null);
michael@0: 	private Yytoken token = null;
michael@0: 	private int status = S_INIT;
michael@0: 	
michael@0: 	private int peekStatus(LinkedList statusStack){
michael@0: 		if(statusStack.size()==0)
michael@0: 			return -1;
michael@0: 		Integer status=(Integer)statusStack.getFirst();
michael@0: 		return status.intValue();
michael@0: 	}
michael@0: 	
michael@0:     /**
michael@0:      *  Reset the parser to the initial state without resetting the underlying reader.
michael@0:      *
michael@0:      */
michael@0:     public void reset(){
michael@0:         token = null;
michael@0:         status = S_INIT;
michael@0:         handlerStatusStack = null;
michael@0:     }
michael@0:     
michael@0:     /**
michael@0:      * Reset the parser to the initial state with a new character reader.
michael@0:      * 
michael@0:      * @param in - The new character reader.
michael@0:      * @throws IOException
michael@0:      * @throws ParseException
michael@0:      */
michael@0: 	public void reset(Reader in){
michael@0: 		lexer.yyreset(in);
michael@0: 		reset();
michael@0: 	}
michael@0: 	
michael@0: 	/**
michael@0: 	 * @return The position of the beginning of the current token.
michael@0: 	 */
michael@0: 	public int getPosition(){
michael@0: 		return lexer.getPosition();
michael@0: 	}
michael@0: 	
michael@0: 	public Object parse(String s) throws ParseException{
michael@0: 		return parse(s, (ContainerFactory)null);
michael@0: 	}
michael@0: 	
michael@0: 	public Object parse(String s, ContainerFactory containerFactory) throws ParseException{
michael@0: 		StringReader in=new StringReader(s);
michael@0: 		try{
michael@0: 			return parse(in, containerFactory);
michael@0: 		}
michael@0: 		catch(IOException ie){
michael@0: 			/*
michael@0: 			 * Actually it will never happen.
michael@0: 			 */
michael@0: 			throw new ParseException(-1, ParseException.ERROR_UNEXPECTED_EXCEPTION, ie);
michael@0: 		}
michael@0: 	}
michael@0: 	
michael@0: 	public Object parse(Reader in) throws IOException, ParseException{
michael@0: 		return parse(in, (ContainerFactory)null);
michael@0: 	}
michael@0: 	
michael@0: 	/**
michael@0: 	 * Parse JSON text into java object from the input source.
michael@0: 	 * 	
michael@0: 	 * @param in
michael@0:      * @param containerFactory - Use this factory to createyour own JSON object and JSON array containers.
michael@0: 	 * @return Instance of the following:
michael@0: 	 *  org.json.simple.JSONObject,
michael@0: 	 * 	org.json.simple.JSONArray,
michael@0: 	 * 	java.lang.String,
michael@0: 	 * 	java.lang.Number,
michael@0: 	 * 	java.lang.Boolean,
michael@0: 	 * 	null
michael@0: 	 * 
michael@0: 	 * @throws IOException
michael@0: 	 * @throws ParseException
michael@0: 	 */
michael@0: 	public Object parse(Reader in, ContainerFactory containerFactory) throws IOException, ParseException{
michael@0: 		reset(in);
michael@0: 		LinkedList statusStack = new LinkedList();
michael@0: 		LinkedList valueStack = new LinkedList();
michael@0: 		
michael@0: 		try{
michael@0: 			do{
michael@0: 				nextToken();
michael@0: 				switch(status){
michael@0: 				case S_INIT:
michael@0: 					switch(token.type){
michael@0: 					case Yytoken.TYPE_VALUE:
michael@0: 						status=S_IN_FINISHED_VALUE;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						valueStack.addFirst(token.value);
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_BRACE:
michael@0: 						status=S_IN_OBJECT;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						valueStack.addFirst(createObjectContainer(containerFactory));
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_SQUARE:
michael@0: 						status=S_IN_ARRAY;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						valueStack.addFirst(createArrayContainer(containerFactory));
michael@0: 						break;
michael@0: 					default:
michael@0: 						status=S_IN_ERROR;
michael@0: 					}//inner switch
michael@0: 					break;
michael@0: 					
michael@0: 				case S_IN_FINISHED_VALUE:
michael@0: 					if(token.type==Yytoken.TYPE_EOF)
michael@0: 						return valueStack.removeFirst();
michael@0: 					else
michael@0: 						throw new ParseException(getPosition(), ParseException.ERROR_UNEXPECTED_TOKEN, token);
michael@0: 					
michael@0: 				case S_IN_OBJECT:
michael@0: 					switch(token.type){
michael@0: 					case Yytoken.TYPE_COMMA:
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_VALUE:
michael@0: 						if(token.value instanceof String){
michael@0: 							String key=(String)token.value;
michael@0: 							valueStack.addFirst(key);
michael@0: 							status=S_PASSED_PAIR_KEY;
michael@0: 							statusStack.addFirst(new Integer(status));
michael@0: 						}
michael@0: 						else{
michael@0: 							status=S_IN_ERROR;
michael@0: 						}
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_RIGHT_BRACE:
michael@0: 						if(valueStack.size()>1){
michael@0: 							statusStack.removeFirst();
michael@0: 							valueStack.removeFirst();
michael@0: 							status=peekStatus(statusStack);
michael@0: 						}
michael@0: 						else{
michael@0: 							status=S_IN_FINISHED_VALUE;
michael@0: 						}
michael@0: 						break;
michael@0: 					default:
michael@0: 						status=S_IN_ERROR;
michael@0: 						break;
michael@0: 					}//inner switch
michael@0: 					break;
michael@0: 					
michael@0: 				case S_PASSED_PAIR_KEY:
michael@0: 					switch(token.type){
michael@0: 					case Yytoken.TYPE_COLON:
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_VALUE:
michael@0: 						statusStack.removeFirst();
michael@0: 						String key=(String)valueStack.removeFirst();
michael@0: 						Map parent=(Map)valueStack.getFirst();
michael@0: 						parent.put(key,token.value);
michael@0: 						status=peekStatus(statusStack);
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_SQUARE:
michael@0: 						statusStack.removeFirst();
michael@0: 						key=(String)valueStack.removeFirst();
michael@0: 						parent=(Map)valueStack.getFirst();
michael@0: 						List newArray=createArrayContainer(containerFactory);
michael@0: 						parent.put(key,newArray);
michael@0: 						status=S_IN_ARRAY;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						valueStack.addFirst(newArray);
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_BRACE:
michael@0: 						statusStack.removeFirst();
michael@0: 						key=(String)valueStack.removeFirst();
michael@0: 						parent=(Map)valueStack.getFirst();
michael@0: 						Map newObject=createObjectContainer(containerFactory);
michael@0: 						parent.put(key,newObject);
michael@0: 						status=S_IN_OBJECT;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						valueStack.addFirst(newObject);
michael@0: 						break;
michael@0: 					default:
michael@0: 						status=S_IN_ERROR;
michael@0: 					}
michael@0: 					break;
michael@0: 					
michael@0: 				case S_IN_ARRAY:
michael@0: 					switch(token.type){
michael@0: 					case Yytoken.TYPE_COMMA:
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_VALUE:
michael@0: 						List val=(List)valueStack.getFirst();
michael@0: 						val.add(token.value);
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_RIGHT_SQUARE:
michael@0: 						if(valueStack.size()>1){
michael@0: 							statusStack.removeFirst();
michael@0: 							valueStack.removeFirst();
michael@0: 							status=peekStatus(statusStack);
michael@0: 						}
michael@0: 						else{
michael@0: 							status=S_IN_FINISHED_VALUE;
michael@0: 						}
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_BRACE:
michael@0: 						val=(List)valueStack.getFirst();
michael@0: 						Map newObject=createObjectContainer(containerFactory);
michael@0: 						val.add(newObject);
michael@0: 						status=S_IN_OBJECT;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						valueStack.addFirst(newObject);
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_SQUARE:
michael@0: 						val=(List)valueStack.getFirst();
michael@0: 						List newArray=createArrayContainer(containerFactory);
michael@0: 						val.add(newArray);
michael@0: 						status=S_IN_ARRAY;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						valueStack.addFirst(newArray);
michael@0: 						break;
michael@0: 					default:
michael@0: 						status=S_IN_ERROR;
michael@0: 					}//inner switch
michael@0: 					break;
michael@0: 				case S_IN_ERROR:
michael@0: 					throw new ParseException(getPosition(), ParseException.ERROR_UNEXPECTED_TOKEN, token);
michael@0: 				}//switch
michael@0: 				if(status==S_IN_ERROR){
michael@0: 					throw new ParseException(getPosition(), ParseException.ERROR_UNEXPECTED_TOKEN, token);
michael@0: 				}
michael@0: 			}while(token.type!=Yytoken.TYPE_EOF);
michael@0: 		}
michael@0: 		catch(IOException ie){
michael@0: 			throw ie;
michael@0: 		}
michael@0: 		
michael@0: 		throw new ParseException(getPosition(), ParseException.ERROR_UNEXPECTED_TOKEN, token);
michael@0: 	}
michael@0: 	
michael@0: 	private void nextToken() throws ParseException, IOException{
michael@0: 		token = lexer.yylex();
michael@0: 		if(token == null)
michael@0: 			token = new Yytoken(Yytoken.TYPE_EOF, null);
michael@0: 	}
michael@0: 	
michael@0: 	private Map createObjectContainer(ContainerFactory containerFactory){
michael@0: 		if(containerFactory == null)
michael@0: 			return new JSONObject();
michael@0: 		Map m = containerFactory.createObjectContainer();
michael@0: 		
michael@0: 		if(m == null)
michael@0: 			return new JSONObject();
michael@0: 		return m;
michael@0: 	}
michael@0: 	
michael@0: 	private List createArrayContainer(ContainerFactory containerFactory){
michael@0: 		if(containerFactory == null)
michael@0: 			return new JSONArray();
michael@0: 		List l = containerFactory.creatArrayContainer();
michael@0: 		
michael@0: 		if(l == null)
michael@0: 			return new JSONArray();
michael@0: 		return l;
michael@0: 	}
michael@0: 	
michael@0: 	public void parse(String s, ContentHandler contentHandler) throws ParseException{
michael@0: 		parse(s, contentHandler, false);
michael@0: 	}
michael@0: 	
michael@0: 	public void parse(String s, ContentHandler contentHandler, boolean isResume) throws ParseException{
michael@0: 		StringReader in=new StringReader(s);
michael@0: 		try{
michael@0: 			parse(in, contentHandler, isResume);
michael@0: 		}
michael@0: 		catch(IOException ie){
michael@0: 			/*
michael@0: 			 * Actually it will never happen.
michael@0: 			 */
michael@0: 			throw new ParseException(-1, ParseException.ERROR_UNEXPECTED_EXCEPTION, ie);
michael@0: 		}
michael@0: 	}
michael@0: 	
michael@0: 	public void parse(Reader in, ContentHandler contentHandler) throws IOException, ParseException{
michael@0: 		parse(in, contentHandler, false);
michael@0: 	}
michael@0: 	
michael@0: 	/**
michael@0: 	 * Stream processing of JSON text.
michael@0: 	 * 
michael@0: 	 * @see ContentHandler
michael@0: 	 * 
michael@0: 	 * @param in
michael@0: 	 * @param contentHandler
michael@0: 	 * @param isResume - Indicates if it continues previous parsing operation.
michael@0:      *                   If set to true, resume parsing the old stream, and parameter 'in' will be ignored. 
michael@0: 	 *                   If this method is called for the first time in this instance, isResume will be ignored.
michael@0: 	 * 
michael@0: 	 * @throws IOException
michael@0: 	 * @throws ParseException
michael@0: 	 */
michael@0: 	public void parse(Reader in, ContentHandler contentHandler, boolean isResume) throws IOException, ParseException{
michael@0: 		if(!isResume){
michael@0: 			reset(in);
michael@0: 			handlerStatusStack = new LinkedList();
michael@0: 		}
michael@0: 		else{
michael@0: 			if(handlerStatusStack == null){
michael@0: 				isResume = false;
michael@0: 				reset(in);
michael@0: 				handlerStatusStack = new LinkedList();
michael@0: 			}
michael@0: 		}
michael@0: 		
michael@0: 		LinkedList statusStack = handlerStatusStack;	
michael@0: 		
michael@0: 		try{
michael@0: 			do{
michael@0: 				switch(status){
michael@0: 				case S_INIT:
michael@0: 					contentHandler.startJSON();
michael@0: 					nextToken();
michael@0: 					switch(token.type){
michael@0: 					case Yytoken.TYPE_VALUE:
michael@0: 						status=S_IN_FINISHED_VALUE;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						if(!contentHandler.primitive(token.value))
michael@0: 							return;
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_BRACE:
michael@0: 						status=S_IN_OBJECT;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						if(!contentHandler.startObject())
michael@0: 							return;
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_SQUARE:
michael@0: 						status=S_IN_ARRAY;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						if(!contentHandler.startArray())
michael@0: 							return;
michael@0: 						break;
michael@0: 					default:
michael@0: 						status=S_IN_ERROR;
michael@0: 					}//inner switch
michael@0: 					break;
michael@0: 					
michael@0: 				case S_IN_FINISHED_VALUE:
michael@0: 					nextToken();
michael@0: 					if(token.type==Yytoken.TYPE_EOF){
michael@0: 						contentHandler.endJSON();
michael@0: 						status = S_END;
michael@0: 						return;
michael@0: 					}
michael@0: 					else{
michael@0: 						status = S_IN_ERROR;
michael@0: 						throw new ParseException(getPosition(), ParseException.ERROR_UNEXPECTED_TOKEN, token);
michael@0: 					}
michael@0: 			
michael@0: 				case S_IN_OBJECT:
michael@0: 					nextToken();
michael@0: 					switch(token.type){
michael@0: 					case Yytoken.TYPE_COMMA:
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_VALUE:
michael@0: 						if(token.value instanceof String){
michael@0: 							String key=(String)token.value;
michael@0: 							status=S_PASSED_PAIR_KEY;
michael@0: 							statusStack.addFirst(new Integer(status));
michael@0: 							if(!contentHandler.startObjectEntry(key))
michael@0: 								return;
michael@0: 						}
michael@0: 						else{
michael@0: 							status=S_IN_ERROR;
michael@0: 						}
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_RIGHT_BRACE:
michael@0: 						if(statusStack.size()>1){
michael@0: 							statusStack.removeFirst();
michael@0: 							status=peekStatus(statusStack);
michael@0: 						}
michael@0: 						else{
michael@0: 							status=S_IN_FINISHED_VALUE;
michael@0: 						}
michael@0: 						if(!contentHandler.endObject())
michael@0: 							return;
michael@0: 						break;
michael@0: 					default:
michael@0: 						status=S_IN_ERROR;
michael@0: 						break;
michael@0: 					}//inner switch
michael@0: 					break;
michael@0: 					
michael@0: 				case S_PASSED_PAIR_KEY:
michael@0: 					nextToken();
michael@0: 					switch(token.type){
michael@0: 					case Yytoken.TYPE_COLON:
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_VALUE:
michael@0: 						statusStack.removeFirst();
michael@0: 						status=peekStatus(statusStack);
michael@0: 						if(!contentHandler.primitive(token.value))
michael@0: 							return;
michael@0: 						if(!contentHandler.endObjectEntry())
michael@0: 							return;
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_SQUARE:
michael@0: 						statusStack.removeFirst();
michael@0: 						statusStack.addFirst(new Integer(S_IN_PAIR_VALUE));
michael@0: 						status=S_IN_ARRAY;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						if(!contentHandler.startArray())
michael@0: 							return;
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_BRACE:
michael@0: 						statusStack.removeFirst();
michael@0: 						statusStack.addFirst(new Integer(S_IN_PAIR_VALUE));
michael@0: 						status=S_IN_OBJECT;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						if(!contentHandler.startObject())
michael@0: 							return;
michael@0: 						break;
michael@0: 					default:
michael@0: 						status=S_IN_ERROR;
michael@0: 					}
michael@0: 					break;
michael@0: 				
michael@0: 				case S_IN_PAIR_VALUE:
michael@0: 					/*
michael@0: 					 * S_IN_PAIR_VALUE is just a marker to indicate the end of an object entry, it doesn't proccess any token,
michael@0: 					 * therefore delay consuming token until next round.
michael@0: 					 */
michael@0: 					statusStack.removeFirst();
michael@0: 					status = peekStatus(statusStack);
michael@0: 					if(!contentHandler.endObjectEntry())
michael@0: 						return;
michael@0: 					break;
michael@0: 					
michael@0: 				case S_IN_ARRAY:
michael@0: 					nextToken();
michael@0: 					switch(token.type){
michael@0: 					case Yytoken.TYPE_COMMA:
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_VALUE:
michael@0: 						if(!contentHandler.primitive(token.value))
michael@0: 							return;
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_RIGHT_SQUARE:
michael@0: 						if(statusStack.size()>1){
michael@0: 							statusStack.removeFirst();
michael@0: 							status=peekStatus(statusStack);
michael@0: 						}
michael@0: 						else{
michael@0: 							status=S_IN_FINISHED_VALUE;
michael@0: 						}
michael@0: 						if(!contentHandler.endArray())
michael@0: 							return;
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_BRACE:
michael@0: 						status=S_IN_OBJECT;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						if(!contentHandler.startObject())
michael@0: 							return;
michael@0: 						break;
michael@0: 					case Yytoken.TYPE_LEFT_SQUARE:
michael@0: 						status=S_IN_ARRAY;
michael@0: 						statusStack.addFirst(new Integer(status));
michael@0: 						if(!contentHandler.startArray())
michael@0: 							return;
michael@0: 						break;
michael@0: 					default:
michael@0: 						status=S_IN_ERROR;
michael@0: 					}//inner switch
michael@0: 					break;
michael@0: 					
michael@0: 				case S_END:
michael@0: 					return;
michael@0: 					
michael@0: 				case S_IN_ERROR:
michael@0: 					throw new ParseException(getPosition(), ParseException.ERROR_UNEXPECTED_TOKEN, token);
michael@0: 				}//switch
michael@0: 				if(status==S_IN_ERROR){
michael@0: 					throw new ParseException(getPosition(), ParseException.ERROR_UNEXPECTED_TOKEN, token);
michael@0: 				}
michael@0: 			}while(token.type!=Yytoken.TYPE_EOF);
michael@0: 		}
michael@0: 		catch(IOException ie){
michael@0: 			status = S_IN_ERROR;
michael@0: 			throw ie;
michael@0: 		}
michael@0: 		catch(ParseException pe){
michael@0: 			status = S_IN_ERROR;
michael@0: 			throw pe;
michael@0: 		}
michael@0: 		catch(RuntimeException re){
michael@0: 			status = S_IN_ERROR;
michael@0: 			throw re;
michael@0: 		}
michael@0: 		catch(Error e){
michael@0: 			status = S_IN_ERROR;
michael@0: 			throw e;
michael@0: 		}
michael@0: 		
michael@0: 		status = S_IN_ERROR;
michael@0: 		throw new ParseException(getPosition(), ParseException.ERROR_UNEXPECTED_TOKEN, token);
michael@0: 	}
michael@0: }