Thread: [Htmlparser-cvs] htmlparser/src/org/htmlparser/util ParserUtils.java,1.45,1.46

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Update of /cvsroot/htmlparser/htmlparser/src/org/htmlparser/util
In directory sc8-pr-cvs1.sourceforge.net:/tmp/cvs-serv26947

Modified Files:
	ParserUtils.java 
Log Message:

Bug fixing and trimAllTags method added.

Index: ParserUtils.java
===================================================================
RCS file: /cvsroot/htmlparser/htmlparser/src/org/htmlparser/util/ParserUtils.java,v
retrieving revision 1.45
retrieving revision 1.46
diff -C2 -d -r1.45 -r1.46
*** ParserUtils.java	31 Jul 2004 16:42:34 -0000	1.45
--- ParserUtils.java	27 Aug 2004 09:54:27 -0000	1.46
***************
*** 93,97 ****
      /**
       * Split the input string considering as string separator
!      * all the non numerical characters
       * with the only exception of the characters specified in charsDoNotBeRemoved param.
       * <BR>For example if you call splitButDigits(&quot;&lt;DIV&gt;  +12.5, +3.4 &lt;/DIV&gt;&quot;, &quot;+.&quot;),
--- 93,97 ----
      /**
       * Split the input string considering as string separator
!      * all the not numerical characters
       * with the only exception of the characters specified in charsDoNotBeRemoved param.
       * <BR>For example if you call splitButDigits(&quot;&lt;DIV&gt;  +12.5, +3.4 &lt;/DIV&gt;&quot;, &quot;+.&quot;),
***************
*** 154,158 ****

      /**
!      * Remove from the input string all the non numerical characters
       * with the only exception of the characters specified in charsDoNotBeRemoved param.
       * <BR>For example if you call trimButDigits(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;+.&quot;),
--- 154,158 ----

      /**
!      * Remove from the input string all the not numerical characters
       * with the only exception of the characters specified in charsDoNotBeRemoved param.
       * <BR>For example if you call trimButDigits(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;+.&quot;),
***************
*** 185,189 ****

      /**
!      * Remove from the input string all the non numerical characters
       * with the only exception of the characters specified in charsDoNotBeRemoved param.
       * <BR>The removal process removes only chars at the beginning and at the end of the string.
--- 185,189 ----

      /**
!      * Remove from the beginning and the end of the input string all the not numerical characters
       * with the only exception of the characters specified in charsDoNotBeRemoved param.
       * <BR>The removal process removes only chars at the beginning and at the end of the string.
***************
*** 191,195 ****
       * <BR>you obtain a string &quot;+12.5&quot; as output (1,2 and 5 are digits and +,. are chars that do not be removed).
       * <BR>For example if you call trimButDigitsBeginEnd(&quot;&lt;DIV&gt;  +1 2 . 5 &lt;/DIV&gt;&quot;, &quot;+.&quot;),
!      * <BR>you obtain a string &quot;+1 2 . 5&quot; as output (the spaces inside the string are not removed).
       * @param input - The string in input.
       * @param charsDoNotBeRemoved - The chars that do not be removed.
--- 191,195 ----
       * <BR>you obtain a string &quot;+12.5&quot; as output (1,2 and 5 are digits and +,. are chars that do not be removed).
       * <BR>For example if you call trimButDigitsBeginEnd(&quot;&lt;DIV&gt;  +1 2 . 5 &lt;/DIV&gt;&quot;, &quot;+.&quot;),
!      * <BR>you obtain a string &quot;+1 2 . 5&quot; as output (the spacess inside the string are not removed).
       * @param input - The string in input.
       * @param charsDoNotBeRemoved - The chars that do not be removed.
***************
*** 238,242 ****
      /**
       * Split the input string considering as string separator
!      * all the space and tabs like chars and
       * the chars specified in the input variable charsToBeRemoved.
       * <BR>For example if you call splitSpaces(&quot;&lt;DIV&gt;  +12.5, +3.4 &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/,&quot;),
--- 238,242 ----
      /**
       * Split the input string considering as string separator
!      * all the spaces and tabs like chars and
       * the chars specified in the input variable charsToBeRemoved.
       * <BR>For example if you call splitSpaces(&quot;&lt;DIV&gt;  +12.5, +3.4 &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/,&quot;),
***************
*** 299,308 ****

      /**
!      * Remove from the input string all the space and tabs like chars.
!      * <BR>Remove also the chars specified in the input variable charsToBeRemoved.
       * <BR>For example if you call trimSpaces(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
       * <BR>you obtain a string &quot;+12.5&quot; as output (space chars and &lt;,&gt;,D,I,V,/ are chars that must be removed).
       * <BR>For example if you call trimSpaces(&quot;&lt;DIV&gt;  Trim All Spaces Also The Ones Inside The String &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
!      * <BR>you obtain a string &quot;TrimAllSpacesAlsoTheOnesInsideTheString&quot; as output (all the space inside the string are removed).
       * @param input The string in input.
       * @param charsToBeRemoved The chars to be removed.
--- 299,308 ----

      /**
!      * Remove from the input string all the spaces and tabs like chars.
!      * Remove also the chars specified in the input variable charsToBeRemoved.
       * <BR>For example if you call trimSpaces(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
       * <BR>you obtain a string &quot;+12.5&quot; as output (space chars and &lt;,&gt;,D,I,V,/ are chars that must be removed).
       * <BR>For example if you call trimSpaces(&quot;&lt;DIV&gt;  Trim All Spaces Also The Ones Inside The String &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
!      * <BR>you obtain a string &quot;TrimAllSpacesAlsoTheOnesInsideTheString&quot; as output (all the spaces inside the string are removed).
       * @param input The string in input.
       * @param charsToBeRemoved The chars to be removed.
***************
*** 330,340 ****

      /**
!      * Remove from the input string all the space and tabs like chars.
!      * <BR>Remove also the chars specified in the input variable charsToBeRemoved.
       * <BR>The removal process removes only chars at the beginning and at the end of the string.
       * <BR>For example if you call trimSpacesBeginEnd(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
       * <BR>you obtain a string &quot;+12.5&quot; as output (space chars and &lt;,&gt;,D,I,V,/ are chars that must be removed).
       * <BR>For example if you call trimSpacesBeginEnd(&quot;&lt;DIV&gt;  Trim all spaces but not the ones inside the string &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
!      * <BR>you obtain a string &quot;Trim all spaces but not the ones inside the string&quot; as output (all the space inside the string are preserved).
       * @param input The string in input.
       * @param charsToBeRemoved The chars to be removed.
--- 330,340 ----

      /**
!      * Remove from the beginning and the end of the input string all the spaces and tabs like chars.
!      * Remove also the chars specified in the input variable charsToBeRemoved.
       * <BR>The removal process removes only chars at the beginning and at the end of the string.
       * <BR>For example if you call trimSpacesBeginEnd(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
       * <BR>you obtain a string &quot;+12.5&quot; as output (space chars and &lt;,&gt;,D,I,V,/ are chars that must be removed).
       * <BR>For example if you call trimSpacesBeginEnd(&quot;&lt;DIV&gt;  Trim all spaces but not the ones inside the string &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
!      * <BR>you obtain a string &quot;Trim all spaces but not the ones inside the string&quot; as output (all the spaces inside the string are preserved).
       * @param input The string in input.
       * @param charsToBeRemoved The chars to be removed.
***************
*** 369,373 ****
                  if (charsToBeRemoved.charAt(charsCount)==input.charAt(index))
                      charFound=true;
!             if (!( (Character.isWhitespace(input.charAt(index))) || (Character.isSpaceChar(input.charAt(index-1))) || (charFound) ))
              {
                  end=index;
--- 369,373 ----
                  if (charsToBeRemoved.charAt(charsCount)==input.charAt(index))
                      charFound=true;
!             if (!( (Character.isWhitespace(input.charAt(index))) || (Character.isSpaceChar(input.charAt(index))) || (charFound) ))
              {
                  end=index;
***************
*** 475,479 ****

      /**
!      * Remove from the input string all the characters
       * with the only exception of the characters specified in charsDoNotBeRemoved param.
       * <BR>The removal process removes only chars at the beginning and at the end of the string.
--- 475,479 ----

      /**
!      * Remove from the beginning and the end of the input string all the characters
       * with the only exception of the characters specified in charsDoNotBeRemoved param.
       * <BR>The removal process removes only chars at the beginning and at the end of the string.
***************
*** 592,596 ****
       * <BR>you obtain a string &quot;+12.5&quot; as output (&lt;,&gt;,D,I,V,/ and space char are chars that must be removed).
       * <BR>For example if you call trimChars(&quot;&lt;DIV&gt;  Trim All Chars Also The Ones Inside The String &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),
!      * <BR>you obtain a string &quot;TrimAllCharsAlsoTheOnesInsideTheString&quot; as output (all the space inside the string are removed).
       * @param input The string in input.
       * @param charsToBeRemoved The chars to be removed.
--- 592,596 ----
       * <BR>you obtain a string &quot;+12.5&quot; as output (&lt;,&gt;,D,I,V,/ and space char are chars that must be removed).
       * <BR>For example if you call trimChars(&quot;&lt;DIV&gt;  Trim All Chars Also The Ones Inside The String &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),
!      * <BR>you obtain a string &quot;TrimAllCharsAlsoTheOnesInsideTheString&quot; as output (all the spaces inside the string are removed).
       * @param input The string in input.
       * @param charsToBeRemoved The chars to be removed.
***************
*** 618,627 ****

      /**
!      * Remove from the input string all the chars specified in the input variable charsToBeRemoved.
       * <BR>The removal process removes only chars at the beginning and at the end of the string.
       * <BR>For example if you call trimCharsBeginEnd(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),
       * <BR>you obtain a string &quot;+12.5&quot; as output (' ' is a space char and &lt;,&gt;,D,I,V,/ are chars that must be removed).
       * <BR>For example if you call trimCharsBeginEnd(&quot;&lt;DIV&gt;  Trim all spaces but not the ones inside the string &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),
!      * <BR>you obtain a string &quot;Trim all spaces but not the ones inside the string&quot; as output (all the space inside the string are preserved).
       * @param input The string in input.
       * @param charsToBeRemoved The chars to be removed.
--- 618,627 ----

      /**
!      * Remove from the beginning and the end of the input string all the chars specified in the input variable charsToBeRemoved.
       * <BR>The removal process removes only chars at the beginning and at the end of the string.
       * <BR>For example if you call trimCharsBeginEnd(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),
       * <BR>you obtain a string &quot;+12.5&quot; as output (' ' is a space char and &lt;,&gt;,D,I,V,/ are chars that must be removed).
       * <BR>For example if you call trimCharsBeginEnd(&quot;&lt;DIV&gt;  Trim all spaces but not the ones inside the string &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),
!      * <BR>you obtain a string &quot;Trim all spaces but not the ones inside the string&quot; as output (all the spaces inside the string are preserved).
       * @param input The string in input.
       * @param charsToBeRemoved The chars to be removed.
***************
*** 900,905 ****

      /**
!      * Trim the input string in a string array,
!      * considering the tags as delimiter for splitting.
       * @see ParserUtils#trimTags (String input, String[] tags, boolean recursive, boolean insideTag).
       */
--- 900,945 ----

      /**
!      * Trim the input string, removing all the tags in the input string.
!      * <BR>The method trims all the substrings included in the input string of the following type:
!      * &quot;&lt;XXX&gt;&quot;, where XXX could be a string of any type.
!      * <BR>If you set to true the inside parameter, the method deletes also the YYY string in the following input string:
!      * &quot;&lt;XXX&gt;YYY&lt;ZZZ&gt;&quot;, note that ZZZ is not necessary the closing tag of XXX.
!      * @param input The string in input.
!      * @param inside If true, it forces the method to delete also what is inside the tags.
!      * @return The string without tags.
!      */
!     public static String trimAllTags (String input, boolean inside)
!     {
!  	
!         StringBuffer output = new StringBuffer();
! 
!         if (inside) {
!             if ((input.indexOf('<')==-1) || (input.lastIndexOf('>')==-1) || (input.lastIndexOf('>')<input.indexOf('<'))) {
!                 output.append(input);
!             } else {
!                 output.append(input.substring(0, input.indexOf('<')));
!                 output.append(input.substring(input.lastIndexOf('>')+1, input.length()));
!             }
!         } else {
!             boolean write = true;
!             for (int index=0; index<input.length(); index++)
!             {    
!                 if (input.charAt(index)=='<' && write)
!                     write = false;
!                 if (write)
!                     output.append(input.charAt(index));
!                 if (input.charAt(index)=='>' && (!write))
!                     write = true;
!             }
!         }
! 
!         return output.toString();
!     }
!     
! 
!     /**
!      * Trim all tags in the input string and
!      * return a string like the input one
!      * without the tags and their content.
       * @see ParserUtils#trimTags (String input, String[] tags, boolean recursive, boolean insideTag).
       */
***************
*** 988,993 ****

      /**
!      * Trim the input string in a string array,
!      * considering the tags as delimiter for splitting.
       * <BR>Use Class class as input parameter
       * instead of tags[] string array.
--- 1028,1034 ----

      /**
!      * Trim all tags in the input string and
!      * return a string like the input one
!      * without the tags and their content.
       * <BR>Use Class class as input parameter
       * instead of tags[] string array.
***************
*** 1001,1006 ****

      /**
!      * Trim the input string in a string array,
!      * considering the tags as delimiter for splitting.
       * <BR>Use Class class as input parameter
       * instead of tags[] string array.
--- 1042,1048 ----

      /**
!      * Trim all tags in the input string and
!      * return a string like the input one
!      * without the tags and their content (optional).
       * <BR>Use Class class as input parameter
       * instead of tags[] string array.
***************
*** 1014,1019 ****

      /**
!      * Trim the input string in a string array,
!      * considering the tags as delimiter for splitting.
       * <BR>Use NodeFilter class as input parameter
       * instead of tags[] string array.
--- 1056,1062 ----

      /**
!      * Trim all tags in the input string and
!      * return a string like the input one
!      * without the tags and their content.
       * <BR>Use NodeFilter class as input parameter
       * instead of tags[] string array.
***************
*** 1027,1032 ****

      /**
!      * Trim the input string in a string array,
!      * considering the tags as delimiter for splitting.
       * <BR>Use NodeFilter class as input parameter
       * instead of tags[] string array.
--- 1070,1076 ----

      /**
!      * Trim all tags in the input string and
!      * return a string like the input one
!      * without the tags and their content (optional).
       * <BR>Use NodeFilter class as input parameter
       * instead of tags[] string array.

Thread: [Htmlparser-cvs] htmlparser/src/org/htmlparser/util ParserUtils.java,1.45,1.46

htmlparser-cvs