HELP with Verity searching

Question

I have a Verity collection and I am searching it
successfully. HOWEVER, I have three custom fields that I am using
as part of the index for each file. I am noticing that by default
Verity is NOT searching these custom fields. I figured out that for
the criteria parameter of cfsearch I can do this:
criteria="#strSearch# OR cf_custom1 <contains>
#strSearch# OR cf_custom2 <contains> #strSearch# etc...."
This will indeed search the custom fields, but only using
simple search terms like "web" or "web development" or "web AND
development".
As soon as I try to use more of the other Verity searches
like "web <near> development" or "<case> CIA", it
throws an error. How can I make it so I can search the custom
fields but also allow the user to use Verity's more powerful search
terms?

Answer 1

public class StringSearchTest {
  public static void main(String[] args){
    Random rnd = new Random();
    String alpha = "     abcdefghijklmnopqrstuvwxyz";
    String s = "";
    StringBuffer sb = new StringBuffer(3000100);
    for(int i = 0; i< 1500000;i++){
      sb.append(alpha.charAt(rnd.nextInt(alpha.length())));
      //if((i % 1000)==0) System.out.println(i);
    sb.append("j_a_v_a");
    for(int i = 0; i< 1500000;i++){
      sb.append(alpha.charAt(rnd.nextInt(alpha.length())));
    s = sb.toString();
    // now s is about 3,000,000 characters long or about 500,000 5 letter words
    for(int i = 0; i<1000; i++){
      System.out.println(i + s.indexOf("j_a_v_a"));
}I generate a string of about 3 million random characters in is. Stuff the string "j_a_v_a" into the middle and then use good old indexOf() to find it. I repeat the search a thousand times, printing crap out to the console each time through the loop and this process easily runs in 10 seconds. This means it takes less than a hundredth of a second to do a single find (on my box - millage may vary).
Now I know that that a hundredth of a second is WAY slower than it could be. But exactly why do you need it to be way faster than that? Do you need it to be faster so bad that you need to actually bring in the consultants, ask about algorithms, write a bunch o' code, test it out and document it?
Remember the quote about premature optimization being the root of all evil?
Well, it is. You should not be writing the fast algorithm until you have first written the butt simple code to see how slow it is, and if you did write that code, you should have posted it, or clocked it and told us how slow it was and given us some indication of why and how much faster it needed to be.
Sorry, nothing personal, I'm in a bad mood today.