/** 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]);
}
}
}
@Test
public void testLookupPropertiesDefault() throws IOException {
ExactComparator comp = new ExactComparator();
List<Property> props = new ArrayList();
props.add(new PropertyImpl("ID"));
Property name = new PropertyImpl("NAME", comp, 0.3, 0.8);
props.add(name);
Property email = new PropertyImpl("EMAIL", comp, 0.3, 0.8);
email.setLookupBehaviour(Property.Lookup.DEFAULT);
props.add(email);
ConfigurationImpl config = new ConfigurationImpl();
config.setThreshold(0.85);
config.setProperties(props);
config.validate();
Collection<Property> lookups = config.getLookupProperties();
assertEquals(2, lookups.size());
assertTrue(lookups.contains(name));
assertTrue(lookups.contains(email));
}
/** 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;
}
