Creative GUI element assistance needed

Hi ASCII_Man,

I found this, I think is does the thing you want: having a sort of match score between two strings.

 This is a match that I find quite useful. Its based on correlation . It is flawed in that characters that are not
int the ASCII range (0-255) will break it but this can be fixed.

The method returns a score for all possible matches between the data and the match string. Larger
value is better match. It is quite flexible in that you can assign values (int the table) for different kinds of
match to tailor it to your needs. At the moment it performs a straight case insensitive match (does not
handle numbers). If you want to make it better able to handle typoes on a qwerty keyboard then you
could modify the values in the table to be inversly proportional to the distance between two keys. A
Gaussian type function may work better.


/**
 * this method creates the weight table for the correlation. Its just sets up a simple case insensitive
 * match. All characters that do not match have a zero entry. All characters that do match have a 1.0f entry
 * Depnding on your domain you may wish to refine these weights. (to perhaps account for typoes on a  
 * keyboard)
 *
 */
public static float[][] generateWeights() {


   // build case insensitive cost matrix
   float[][] tmp = new float[256][256];
   int r = (int)'A';
   for(int i=(int)'a';i<=(int)'z';i++) {
     tmp[i][i] = 1.0f;
     tmp[i][r] = 1.0f;
     tmp[r][i] = 1.0f;
     tmp[r][r] = 1.0f;
     //System.out.println("Processing "+(char)i+" "+(char)r);
     r++;
   }  

   return tmp;
}


/**
 * performs a correlation match on the two strings. The weighting table is used to hold the
 * measure of similarity between two characters. Returns an array of floats which are in the range
 * 0 to 1 inclusive. The number of elements in the returned array is data.length()+match.length(). To
 * find the index into the data array from a position in the result array you subtract match.length().
 * if this results in a value less then zero then it indicates that a partial match exists between
 * match.substring(Math.abs(pos-match.length()),match.length()) and the start of the data String.
 */
public static float[] softMatch(String data, String match, float[][] weighting) {
  if((data==null)||(match==null)) throw new IllegalArgumentException("Arguements must not be null");
   
  // pad the input string to make the correlation easier.
  // idealy this would not happen but not doing it makes the correlation loop more complex
  StringBuffer b = new StringBuffer(data.length()+(match.length()));
  for(int i=0;i<match.length();i++) b.append(" ");
  b.append(data);
  float[] results = new float[b.length()];
  for(int i=0;i<match.length();i++) b.append(" ");

 
 
  for(int n=0;n<(b.length()-match.length());n++){
     float res = 0.0f;
   
     for(int m=0;m<match.length();m++)   {
       // System.out.println("Processing "+b.charAt(n+m)+" "+match.charAt(m));
        res+=weighting[(int)b.charAt(n+m)][(int)match.charAt(m)];
       
     }
     results[n] = res/match.length();
  }
  return results;
 
}

Regards,
Gerben Ernst