[Htmlparser-cvs] htmlparser/src/org/htmlparser/parserHelper TagParser.java,1.28,1.29

[<po...@us...>]

Update of /cvsroot/htmlparser/htmlparser/src/org/htmlparser/parserHelper
In directory sc8-pr-cvs1:/tmp/cvs-serv7363/src/org/htmlparser/parserHelper

Modified Files:
	TagParser.java 
Log Message:
- Tag now remembers (in tagLines) *all* the lines spanned by the tag (not just the last one).
- Tag now remembers line number on which tag starts (from TagData).
- Tag now has new public methods:
      int getTagStartLine()
      int getTagEndLine()
      String[] getTagLines()
- TagParser now gives Tag (via constructor and setTagLine) sufficient info to support the above.

Index: TagParser.java
===================================================================
RCS file: /cvsroot/htmlparser/htmlparser/src/org/htmlparser/parserHelper/TagParser.java,v
retrieving revision 1.28
retrieving revision 1.29
diff -C2 -d -r1.28 -r1.29
*** TagParser.java	19 May 2003 02:49:57 -0000	1.28
--- TagParser.java	22 May 2003 00:36:15 -0000	1.29
***************
*** 39,44 ****
  public class TagParser {
  	public final static int TAG_BEFORE_PARSING_STATE=1;
!     public final static int TAG_BEGIN_PARSING_STATE=1<<2;
!     public final static int TAG_FINISHED_PARSING_STATE=1<<3;
  	public final static int TAG_ILLEGAL_STATE=1<<4;
  	public final static int TAG_IGNORE_DATA_STATE=1<<5;	    
--- 39,44 ----
  public class TagParser {
  	public final static int TAG_BEFORE_PARSING_STATE=1;
! 	public final static int TAG_BEGIN_PARSING_STATE=1<<2;
! 	public final static int TAG_FINISHED_PARSING_STATE=1<<3;
  	public final static int TAG_ILLEGAL_STATE=1<<4;
  	public final static int TAG_IGNORE_DATA_STATE=1<<5;	    
***************
*** 58,63 ****
  		int i=position;
  		char ch;
!         char[] ignorechar = new char[1]; // holds the character we're looking for when in TAG_IGNORE_DATA_STATE
! 		Tag tag = new Tag(new TagData(0,0,"",input));
  		Bool encounteredQuery = new Bool(false);
  		while (i<tag.getTagLine().length() && 
--- 58,64 ----
  		int i=position;
  		char ch;
! 		char[] ignorechar = new char[1]; // holds the character we're looking for when in TAG_IGNORE_DATA_STATE
! 		Tag tag = new Tag(new TagData(position, 0, reader.getLastLineNumber(), 0, "", input, "", false));
! 
  		Bool encounteredQuery = new Bool(false);
  		while (i<tag.getTagLine().length() && 
***************
*** 79,83 ****
  			return tag;
  		} else
! 		return null;	
  	}
  
--- 80,84 ----
  			return tag;
  		} else
! 			return null;
  	}
  
***************
*** 87,92 ****
  		state = toggleIgnoringState(state, ch, ignorechar);
  		if (state==TAG_BEFORE_PARSING_STATE && ch!='<') {
!             state= TAG_ILLEGAL_STATE;
!         }
  		if (state==TAG_IGNORE_DATA_STATE && ch=='<') {
  			// If the next tag char is is close tag, then
--- 88,93 ----
  		state = toggleIgnoringState(state, ch, ignorechar);
  		if (state==TAG_BEFORE_PARSING_STATE && ch!='<') {
! 			state= TAG_ILLEGAL_STATE;
! 		}
  		if (state==TAG_IGNORE_DATA_STATE && ch=='<') {
  			// If the next tag char is is close tag, then
***************
*** 96,100 ****
  		}
  		if (state==TAG_IGNORE_BEGIN_TAG_STATE && ch=='>') {
! 				state = TAG_IGNORE_DATA_STATE;
  		}
  		checkIfAppendable(encounteredQuery, state, ch, tag);
--- 97,101 ----
  		}
  		if (state==TAG_IGNORE_BEGIN_TAG_STATE && ch=='>') {
! 			state = TAG_IGNORE_DATA_STATE;
  		}
  		checkIfAppendable(encounteredQuery, state, ch, tag);
***************
*** 277,282 ****
  			// Annette Doyle - see testcase HTMLImageScannerTest.testImageTagOnMultipleLines()
  			// Further modified by Somik Raha, to remove bug - HTMLTagTest.testBrokenTag
  			do {
! 				nextLine = reader.getNextLine();		
  			}
  			while (nextLine!=null && nextLine.length()==0);
--- 278,285 ----
  			// Annette Doyle - see testcase HTMLImageScannerTest.testImageTagOnMultipleLines()
  			// Further modified by Somik Raha, to remove bug - HTMLTagTest.testBrokenTag
+ 			int numLinesAdvanced = 0;
  			do {
! 				nextLine = reader.getNextLine();
! 				numLinesAdvanced++;
  			}
  			while (nextLine!=null && nextLine.length()==0);
***************
*** 288,291 ****
--- 291,298 ----
  				tag.append(Node.getLineSeparator());
  			}
+ 			
+ 			// Ensure blank lines are included in tag's 'tagLines'
+ 			while (--numLinesAdvanced > 0)
+ 				tag.setTagLine("");
  
  			// We need to continue parsing to the next line