[Htmlparser-cvs] htmlparser/src/org/htmlparser/beans StringBean.java,1.44,1.45

[Derrick O. <der...@us...>]

Update of //cvsroot/htmlparser/htmlparser/src/org/htmlparser/beans
In directory sc8-pr-cvs5.sourceforge.net:/tmp/cvs-serv32405/beans

Modified Files:
	StringBean.java 
Log Message:
fix bug#1496863 StringBean collapse() adds extra whitespace
Keep collapsing state machine state as member variable.

Index: StringBean.java
===================================================================
RCS file: //cvsroot/htmlparser/htmlparser/src/org/htmlparser/beans/StringBean.java,v
retrieving revision 1.44
retrieving revision 1.45
diff -C2 -d -r1.44 -r1.45
*** StringBean.java	15 May 2005 11:49:03 -0000	1.44
--- StringBean.java	30 May 2006 01:07:14 -0000	1.45
***************
*** 149,152 ****
--- 149,157 ----
  
      /**
+      * The state of the collapse processiung state machine.
+      */
+     protected int mCollapseState;
+ 
+     /**
       * The buffer text is stored in while traversing the HTML.
       */
***************
*** 189,192 ****
--- 194,198 ----
          mReplaceSpace = true;
          mCollapse = true;
+         mCollapseState = 0;
          mBuffer = new StringBuffer (4096);
          mIsScript = false;
***************
*** 213,216 ****
--- 219,223 ----
                  length - NEWLINE_SIZE, length).equals (NEWLINE))))
              mBuffer.append (NEWLINE);
+         mCollapseState = 0;
      }
  
***************
*** 238,243 ****
      {
          int chars;
-         int length;
-         int state;
          char character;
  
--- 245,248 ----
***************
*** 245,255 ****
          if (0 != chars)
          {
-             length = buffer.length ();
-             state = ((0 == length)
-                 || (buffer.charAt (length - 1) == ' ')
-                 || ((NEWLINE_SIZE <= length)
-                     && buffer.substring (
-                         length - NEWLINE_SIZE, length).equals (NEWLINE)))
-                 ? 0 : 1;
              for (int i = 0; i < chars; i++)
              {
--- 250,253 ----
***************
*** 265,275 ****
                      case '\r':
                      case '\n':
!                         if (0 != state)
!                             state = 1;
                          break;
                      default:
!                         if (1 == state)
                              buffer.append (' ');
!                         state = 2;
                          buffer.append (character);
                  }
--- 263,273 ----
                      case '\r':
                      case '\n':
!                         if (0 != mCollapseState)
!                             mCollapseState = 1;
                          break;
                      default:
!                         if (1 == mCollapseState)
                              buffer.append (' ');
!                         mCollapseState = 2;
                          buffer.append (character);
                  }
***************
*** 289,292 ****
--- 287,291 ----
          String ret;
  
+         mCollapseState = 0;
          mParser.visitAllNodesWith (this);
          ret = mBuffer.toString ();
***************
*** 319,322 ****
--- 318,322 ----
      protected void setStrings ()
      {
+         mCollapseState = 0;
          if (null != getURL ())
              try
***************
*** 341,344 ****
--- 341,345 ----
                      mParser.reset ();
                      mBuffer = new StringBuffer (4096);
+                     mCollapseState = 0;
                      mParser.visitAllNodesWith (this);
                      updateStrings (mBuffer.toString ());
***************
*** 558,561 ****
--- 559,565 ----
       * If the setting is changed after the URL has been set, the text from the
       * URL will be reacquired, which is possibly expensive.
+      * The internal state of the collapse state machine can be reset with
+      * code like this:
+      * <code>setCollapse (getCollapse ());</code>
       * @param collapse If <code>true</code>, sequences of whitespace
       * will be reduced to a single space.
***************
*** 563,566 ****
--- 567,571 ----
      public void setCollapse (boolean collapse)
      {
+         mCollapseState = 0;
          boolean oldValue = mCollapse;
          if (oldValue != collapse)