gistfile1.java

import java.io.File;
import java.util.HashMap;
import java.util.HashSet;
import java.util.Scanner;

/** create in-memory mappings from words to the files that contain them */
public class Indexer2 {
  public static HashMap<String, HashSet<String>> buildIndex(String dirName) throws java.io.IOException {
    HashMap<String,HashSet<String>> map = new HashMap<String,HashSet<String>>();
    for (File f : new File(dirName).listFiles()) {
      String fname = f.getName();
      Scanner s = new Scanner(f).useDelimiter("[^\\p{L}&&[^\\p{M}]]");
      while (s.hasNext()) {
        String key = s.next().toLowerCase();
        HashSet<String> set = map.get(key);
        if (set == null) {
          set = new HashSet<String>();
          map.put(key, set);
        }
        set.add(fname);
      }
    }
    return map;
  }

  //for testing
  public static void main(String args[]) throws java.io.IOException {
    long start = new java.util.Date().getTime();
    HashMap<String,HashSet<String>> index = Indexer2.buildIndex(args[0]);
    System.out.println(new java.util.Date().getTime() - start);
    System.out.println(index.get(args[1]));
  }
}