public double compare(String s1, String s2) {
if (s1.equals(s2)) return 1.0;
// tokenize
String[] t1 = StringUtils.split(s1);
String[] t2 = StringUtils.split(s2);
// ensure that t1 is shorter than or same length as t2
if (t1.length > t2.length) {
String[] tmp = t2;
t2 = t1;
t1 = tmp;
}
// find best matches for each token in t1
double sum = 0;
for (int ix1 = 0; ix1 < t1.length; ix1++) {
double highest = 0;
for (int ix2 = 0; ix2 < t2.length; ix2++)
highest = Math.max(highest, subcomp.compare(t1[ix1], t2[ix2]));
sum += highest;
}
return (sum * 2) / (t1.length + t2.length);
}
private Cleaner makeCleaner(String value) {
if (value == null) return null;
String[] names = StringUtils.split(value);
Cleaner[] cleaners = new Cleaner[names.length];
for (int ix = 0; ix < cleaners.length; ix++) cleaners[ix] = _makeCleaner(names[ix]);
if (cleaners.length == 1) return cleaners[0];
else return new ChainedCleaner(cleaners);
}
/** Add the record to the index. */
public void index(Record record) {
// FIXME: check if record is already indexed
// allocate an ID for this record
long id = store.makeNewRecordId();
store.registerRecord(id, record);
// go through ID properties and register them
for (Property p : config.getIdentityProperties())
for (String extid : record.getValues(p.getName())) store.registerId(id, extid);
// go through lookup properties and register those
for (Property p : config.getLookupProperties()) {
String propname = p.getName();
for (String value : record.getValues(propname)) {
String[] tokens = StringUtils.split(value);
for (int ix = 0; ix < tokens.length; ix++) store.registerToken(id, propname, tokens[ix]);
}
}
}
/** Tokenizes lookup fields and returns all matching buckets in the index. */
private List<Bucket> lookup(Record record) {
List<Bucket> buckets = new ArrayList();
for (Property p : config.getLookupProperties()) {
String propname = p.getName();
Collection<String> values = record.getValues(propname);
if (values == null) continue;
for (String value : values) {
String[] tokens = StringUtils.split(value);
for (int ix = 0; ix < tokens.length; ix++) {
Bucket b = store.lookupToken(propname, tokens[ix]);
if (b == null || b.records == null) continue;
long[] ids = b.records;
if (DEBUG)
System.out.println(
propname + ", " + tokens[ix] + ": " + b.nextfree + " (" + b.getScore() + ")");
buckets.add(b);
}
}
}
return buckets;
}
