Need a quicker way to strip most html tag while retaining a few.

Question

I have to strip the majority of non plane text (html, javascript, css) from a file whilst retaining the html tags I want.
I have found many solutions for stripping ALL of the tags but not for stripping most whilst retaining a few.
I first looked at regular expressions but I could not find a solution that didnt involve specifying all the tags I want removed insted of just listing all the tags I want to keep.
I had something like this> str.replaceAll("\\<+?(^p|^th|^tr|^td|^h2|^h3|^h4|^li).+?\\>","") but it does not work.
A regular expression would be great but I could not find one so I coded my own solution.
The code below goes through each tag (both start and end tags) and strips it from the string if it is not a wanted tag.
private void stripTags(int startIndex){
          int startArrow = 0;
          int endArrow = 0;
          while(true){
               startArrow = text.indexOf("<", startIndex);
               endArrow = text.indexOf(">", startArrow+1);
               //reached EOF?
               if(startArrow == -1 || endArrow == -1)   return; // return -1; 
               if(text.substring(startArrow+1, startArrow+2).equals("/")){
                    //deal with the end tag
                    if(isWantedTag(text.substring(startArrow+2, startArrow+4))) {
                         startIndex = endArrow+1;
                    } else {
                         //remove the tag
                         text = text.substring(0, startArrow).concat(text.substring(endArrow+1, text.length()));
                         startIndex = startArrow;
               } else {
                    //deal with the start tag
                    if(isWantedTag(text.substring(startArrow+1, startArrow+3))){
                         //remove tag parameters
                         if(endArrow-(startArrow+2) > 1) text = text.substring(0, startArrow+3).concat(text.substring(endArrow, text.length()));
                         startIndex = text.indexOf(">", startArrow+1)+1;
                    } else {
                         //remove the tag
                         text = text.substring(0, startArrow).concat(text.substring(endArrow+1, text.length()));
                         startIndex = startArrow;
private boolean isWantedTag(String tag){
          return (tag.equals("p>") || tag.equals("p ") || tag.equals("th") || tag.equals("tr") || tag.equals("td") || tag.equals("h2") || tag.equals("h3") || tag.equals("h4") || tag.equals("li"));
     }The problem is that this code is taking ages to run. It takes about 10 seconds to run this method on a very large html file. I need a solution that takes only a 10th of that time.
Does anybody have a regExp that could do the job or a more efficient version on my tag stripper above?
Thanks in advance

Answer 1

I suggest the following template:
int copyFrom = 0;
StringBuilder sb = new StringBuilder();
while (copyFrom < text.length()) {
    int copyTo = findStartTag(text, copyFrom);
    sb.append(text, copyFrom, copyTo);
    copyFrom = findEndTag(text, copyTo + 1);
text = sb.toString();where findStartTag(String s, int offset) returns the index of the next start tag in s, starting search at offset. It should return s.length() rather than -1 if nothing is found. findEndTag is similar.
Sorry, that's obviously wrong since you don't want to remove everything between the tags, just the tags themselves. The basic idea is the same, though:
int copyFrom = 0;
StringBuilder sb = new StringBuilder();
while (copyFrom < text.length()) {
    int copyTo = findStartOfTagToRemove(text, copyFrom);
    sb.append(text, copyFrom, copyTo);
    copyFrom = findEndOfTagToRemove(text, copyTo) + 1;
text = sb.toString();Here, findStartOfTagToRemove finds the position of the '<' character of the next tag to remove.
And findEndOfTagToRemove finds the position of the '>' character of that tag.
Edited by: nygaard on Sep 30, 2008 10:39 AM