[Htmlparser-cvs] htmlparser/src/org/htmlparser/parserHelper AttributeParser.java,1.25,1.26

[<so...@us...>]

Update of /cvsroot/htmlparser/htmlparser/src/org/htmlparser/parserHelper
In directory sc8-pr-cvs1:/tmp/cvs-serv6124/src/org/htmlparser/parserHelper

Modified Files:
	AttributeParser.java 
Log Message:
refactored AttributeParser to be more readable

Index: AttributeParser.java
===================================================================
RCS file: /cvsroot/htmlparser/htmlparser/src/org/htmlparser/parserHelper/AttributeParser.java,v
retrieving revision 1.25
retrieving revision 1.26
diff -C2 -d -r1.25 -r1.26
*** AttributeParser.java	1 Jun 2003 20:50:10 -0000	1.25
--- AttributeParser.java	17 Jun 2003 01:35:04 -0000	1.26
***************
*** 44,55 ****
   */
  public class AttributeParser {
!     private final String delima = " \t\r\n\f=\"'>";
!     private final String delimb = " \t\r\n\f\"'>";
!     private final char doubleQuote = '\"';
!     private final char singleQuote = '\'';
      private String delim;
  
  
  
  	/**
  	* Method to break the tag into pieces.
--- 44,68 ----
   */
  public class AttributeParser {
!     private final String DELIMETERS = " \t\r\n\f=\"'>";
!     private final String DELIMETERS_WITHOUT_EQUALS = " \t\r\n\f\"'>";
!     private final char DOUBLE_QUOTE = '\"';
!     private final char SINGLE_QUOTE = '\'';
      private String delim;
  
  
  
+ 	private Hashtable attributeTable;
+ 	private String element;
+ 	private String name;
+ 	private String value;
+ 	private String part;
+ 	private String empty;
+ 	private boolean equal;
+ 	private StringTokenizer tokenizer;
+ 	private boolean doubleQuote;
+ 	private boolean singleQuote;
+ 	private boolean ready;
+ 	private String currentToken;
+ 	private String tokenAccumulator;
  	/**
  	* Method to break the tag into pieces.
***************
*** 91,149 ****
  	*
  	*/
!    public Hashtable parseAttributes(Tag tag){
!         Hashtable h = new Hashtable();
!         String element,name,value,nextPart=null;
!         String empty=null;
          name=null;
          value=null;
          element=null;
!         boolean waitingForEqual=false;
!         delim=delima;
!         StringTokenizer tokenizer = new StringTokenizer(tag.getText(),delim,true);
          while (true) {
!             nextPart=getNextPart(tokenizer,delim);
!             delim=delima;
!             if (element==null && nextPart != null && !nextPart.equals("=")){
!                 element = nextPart;
!                 putDataIntoTable(h,element,null,true);
              }
              else {
!                 if (nextPart != null && (0 < nextPart.length ())) {
!                     if (name == null) {
!                         if (!nextPart.substring(0,1).equals(" ")) {
!                             name = nextPart;
!                             waitingForEqual=true;
!                         }
!                     }
!                     else {
!                         if (waitingForEqual){
!                             if (nextPart.equals("=")) {
!                                 waitingForEqual=false;
!                                 delim=delimb;
!                             }
!                             else {
!                                  putDataIntoTable(h,name,"",false);
!                                  name=nextPart;
!                                  value=null;
!                             }
!                         }
!                         if (!waitingForEqual && !nextPart.equals("=")) {
!                             value=nextPart;
!                             putDataIntoTable(h,name,value,false);
!                             name=null;
!                             value=null;
!                         }
!                     }
                  }
                  else {
!                     if (name != null) {
!                       if (name.equals("/")) {
!                         putDataIntoTable(h,Tag.EMPTYTAG,"",false);
!                       } else {
!                         putDataIntoTable(h,name,"",false);
!                       }
!                       name=null;
!                       value=null;
!                     }
                      break;
                  }
--- 104,130 ----
  	*
  	*/
!    public Hashtable parseAttributes(Tag tag) {
! 		attributeTable = new Hashtable();
! 		part = null;
! 		empty = null;
          name=null;
          value=null;
          element=null;
! 		equal = false;
!         delim=DELIMETERS;
! 		tokenizer = new StringTokenizer(tag.getText(),delim,true);
          while (true) {
!             part=getNextPartUsing(delim);
!             delim=DELIMETERS;
!             if (element==null && part != null && !part.equals("=")){
!                 element = part;
!                 putDataIntoTable(attributeTable,element,null,true);
              }
              else {
!                 if (isValid(part)) {
!                     process(part);
                  }
                  else {
!                    	processInvalidPart();
                      break;
                  }
***************
*** 151,207 ****
          }
          if (null == element) // handle no tag contents
!             putDataIntoTable(h,"",null,true);
!         return h;
      }
  
!     private String getNextPart(StringTokenizer tokenizer,String deli){
!         String tokenAccumulator=null;
!         boolean isDoubleQuote=false;
!         boolean isSingleQuote=false;
!         boolean isDataReady=false;
!         String currentToken;
!         while (isDataReady == false && tokenizer.hasMoreTokens()) {
!             currentToken = tokenizer.nextToken(deli);
!             //
!             // First let's combine tokens that are inside "" or ''
!             //
!             if (isDoubleQuote || isSingleQuote) {
!                 if (isDoubleQuote && currentToken.charAt(0)==doubleQuote){
!                     isDoubleQuote= false;
!                     isDataReady=true;
!                 } else if (isSingleQuote && currentToken.charAt(0)==singleQuote) {
!                     isSingleQuote=false;
!                     isDataReady=true;
!                 }else {
!                     tokenAccumulator += currentToken;
!                     continue;
!                 }
!             } else if (currentToken.charAt(0)==doubleQuote){
!                 isDoubleQuote= true;
!                 tokenAccumulator = "";
!                 continue;
!             } else if (currentToken.charAt(0)==singleQuote){
!                 isSingleQuote=true;
!                 tokenAccumulator="";
!                 continue;
!             } else tokenAccumulator = currentToken;
  
!             if (tokenAccumulator.equals(currentToken)) {
  
!                 if (delim.indexOf(tokenAccumulator)>=0) {
!                     if (tokenAccumulator.equals("=")){
!                         isDataReady=true;
!                     }
!                 }
!                 else {
  
!                     isDataReady=true;
!                 }
!             }
!             else isDataReady=true;
  
          }
          return tokenAccumulator;
      }
  
  
--- 132,243 ----
          }
          if (null == element) // handle no tag contents
!             putDataIntoTable(attributeTable,"",null,true);
!         return attributeTable;
      }
  
! 	private void processInvalidPart() {
! 		 if (name != null) {
! 		      if (name.equals("/")) {
! 		        putDataIntoTable(attributeTable,Tag.EMPTYTAG,"",false);
! 		      } else {
! 		        putDataIntoTable(attributeTable,name,"",false);
! 		      }
! 		      name=null;
! 		      value=null;
! 		    }
! 	}
  
! 	private boolean isValid(String part) {
! 		return part != null && (0 < part.length ());
! 	}
  
! 	private void process(String part) {
! 		if (name == null) {
! 		    if (!part.substring(0,1).equals(" ")) {
! 		        name = part;
! 		        equal=true;
! 		    }
! 		}
! 		else {
! 		    if (equal){
! 		        if (part.equals("=")) {
! 		            equal=false;
! 		            delim=DELIMETERS_WITHOUT_EQUALS;
! 		        }
! 		        else {
! 		             putDataIntoTable(attributeTable,name,"",false);
! 		             name=part;
! 		             value=null;
! 		        }
! 		    }
! 		    if (!equal && !part.equals("=")) {
! 		        value=part;
! 		        putDataIntoTable(attributeTable,name,value,false);
! 		        name=null;
! 		        value=null;
! 		    }
! 		}
! 	}
  
!     private String getNextPartUsing(String delimiter) {
! 		tokenAccumulator = null;
! 		doubleQuote = false;
! 		singleQuote = false;
! 		ready = false;
!         while (ready == false && tokenizer.hasMoreTokens()) {
!             currentToken = tokenizer.nextToken(delimiter);
  
+             if (doubleQuote || singleQuote) {
+                 combineTokensInsideSingleOrDoubleQuotes();
+             } else if (isCurrentTokenDoubleQuote()){
+                 doubleQuote= true;
+                 tokenAccumulator = "";
+             } else if (isCurrentTokenSingleQuote()){
+                 singleQuote=true;
+                 tokenAccumulator="";
+             } else {
+             	tokenAccumulator = currentToken;
+                 ready = isReadyWithNextPart(currentToken);
+             }
          }
          return tokenAccumulator;
      }
+ 
+ 	private boolean isReadyWithNextPart(String currentToken) {
+ 		boolean ready = false;
+ 		if (isDelimeter(currentToken)) {
+ 		    if (currentToken.equals("=")){
+ 		        ready=true;
+ 		    }
+ 		}
+ 		else {
+ 		    ready=true;
+ 		}
+ 		return ready;
+ 	}
+ 
+ 	private boolean isDelimeter(String token) {
+ 		return delim.indexOf(tokenAccumulator)>=0;
+ 	}
+ 	
+ 	private boolean isCurrentTokenSingleQuote() {
+ 		return currentToken.charAt(0)==SINGLE_QUOTE;
+ 	}
+ 
+ 	private boolean isCurrentTokenDoubleQuote() {
+ 		return currentToken.charAt(0)==DOUBLE_QUOTE;
+ 	}
+ 
+ 	private void combineTokensInsideSingleOrDoubleQuotes() {
+ 		if (doubleQuote && currentToken.charAt(0)==DOUBLE_QUOTE){
+ 		    doubleQuote= false;
+ 		    ready=true;
+ 		} else if (singleQuote && currentToken.charAt(0)==SINGLE_QUOTE) {
+ 		    singleQuote=false;
+ 		    ready=true;
+ 		}else {
+ 		    tokenAccumulator += currentToken;
+ 		}
+ 	}