/** Test a configuration that behaves a lot like KeepWordFilter */
public void testKeep() throws Exception {
CharacterRunAutomaton keepWords =
new CharacterRunAutomaton(
BasicOperations.complement(
Automaton.union(
Arrays.asList(
BasicAutomata.makeString("foo"), BasicAutomata.makeString("bar")))));
Analyzer a = new MockAnalyzer(MockTokenizer.SIMPLE, true, keepWords, true);
assertAnalyzesTo(
a,
"quick foo brown bar bar fox foo",
new String[] {"foo", "bar", "bar", "foo"},
new int[] {2, 2, 1, 2});
}
/**
* Create a automaton for a given context query this automaton will be used to find the matching
* paths with the fst
*
* @param preserveSep set an additional char (<code>XAnalyzingSuggester.SEP_LABEL</code>)
* between each context query
* @param queries list of {@link ContextQuery} defining the lookup context
* @return Automaton matching the given Query
*/
public static Automaton toAutomaton(boolean preserveSep, Iterable<ContextQuery> queries) {
Automaton a = BasicAutomata.makeEmptyString();
Automaton gap = BasicAutomata.makeChar(ContextMapping.SEPARATOR);
if (preserveSep) {
// if separators are preserved the fst contains a SEP_LABEL
// behind each gap. To have a matching automaton, we need to
// include the SEP_LABEL in the query as well
gap =
BasicOperations.concatenate(gap, BasicAutomata.makeChar(XAnalyzingSuggester.SEP_LABEL));
}
for (ContextQuery query : queries) {
a = Automaton.concatenate(Arrays.asList(query.toAutomaton(), gap, a));
}
BasicOperations.determinize(a);
return a;
}
@Override
public String toString(String field) {
return field.toString() + automaton.toString();
}
