Indexed substring search

Question

Let's say I have lots of instances of the following class:   public class Bob {
      public final String name;
      public Bob(String name) {
         if (name==null) throw new ArgumentException();
         this.name=name;
   }The name of a bob can be pretty much anything, but usually it is 5-50 characters long and consists of ordinary words.
E.g.   List bobs=new ArrayList();
   bobs.add(new Bob("I am bob"));
   bobs.add(new Bob("foo bar"));
   bobs.add(new Bob("once upon a time there was a bob with a long name, and he's me"));How do I index the bobs so that I can search for the set of all bobs whose names contain a given substring. E.g., searching for "am" should result in a set containing the first ("I am bob") and the third ("...long name...") of the three bobs above.
If the substring is a whole word or at the beginning of one then it's easy; just keep a list of ⟨String word, Bob[] bobs⟩ tuples (each one listing all bobs whose name contains the word), sorted by word. Then you can do a binary search for the words you're interested in and put all matching bobs in a set that you'll then return.
However, if the substring is not at a word boundary it gets a lot harder!
The fact that there are so many bobs (perhaps a million) also places some constraints on the relative size of the indexes. After all, we don't want the index data structures to consume many times more memory than the bobs themselves do.
Any ideas?
- Marcus Sundman

Answer 1

I came up with this. I don't know if it's a good solution for you or at all but it seems to work and minimizes the nodes down to the size of the alphabet.
It can surely be optimized and cleaned up. Feel free to modify. Let me know what you think:
import java.util.*;
public class FancyTree
    public static void main(String[] args)
        FancyTree tree = new FancyTree();
        tree.add("amble");
        tree.add("same");
        tree.add("name");
        tree.add("mean");
        tree.add("earn");
        tree.add("glean");
        tree.add("same meaning");
        tree.add("meanie");
        System.out.println("a: " + tree.findStrings("a"));
        System.out.println("i: " + tree.findStrings("i"));
        System.out.println("am: " + tree.findStrings("am"));
        System.out.println("sa: " + tree.findStrings("sa"));
        System.out.println("me: " + tree.findStrings("me"));
        System.out.println("samean: " + tree.findStrings("samean"));
        System.out.println("mean: " + tree.findStrings("mean"));
        System.out.println("ean: " + tree.findStrings("ean"));
        System.out.println("ea: " + tree.findStrings("ea"));
    private Node[] nodes = new Node[255];
    public void add(String s)
        Link parent = null;
        for (int i = 0; i < s.length(); i++) {
            Link link = new Link(s, parent);
            getNode(s.charAt(i)).add(link);
            parent = link;
    private Node getNode(char c)
        if (c > 255) throw new IllegalArgumentException();
        Node node = nodes[c];
        if (node == null) {
            node = new Node();
            nodes[c] = node;
        return node;
    public Set findStrings(String subString)
        Collection links = getNode(subString.charAt(0)).getLinks();
        for (int i = 1; i < subString.length() && links.size() > 0; i++) {
            links = getNode(subString.charAt(i)).getLinks(links);
        Set strings = new HashSet();
        for (Iterator i = links.iterator(); i.hasNext(); ) {
            strings.add(((Link) i.next()).getStringParent());
        return strings;
class Node
    private Map links = new HashMap();
    public void add(Link link)
        links.put(link.getParent(), link);
    public Collection getLinks()
        return links.values();
    public Set getLinks(Collection parents)
        HashSet set = new HashSet();
        for (Iterator i = parents.iterator(); i.hasNext();) {
            Object link = links.get(i.next());
            if (link != null) set.add(link);
        return set;
class Link
    private String stringParent;
    private Link linkParent;
    Link(String stringParent, Link linkParent)
        this.stringParent = stringParent;
        this.linkParent = linkParent;
    public Object getParent()
          return linkParent == null ? (Object) stringParent : linkParent;
    public String getStringParent()
        return stringParent;
}