@Override
public Query rewrite(IndexReader reader) throws IOException {
Query rewritten = super.rewrite(reader);
if (rewritten != this) {
return rewritten;
}
if (termArrays.isEmpty()) {
return new MatchNoDocsQuery();
}
MultiPhraseQuery.Builder query = new MultiPhraseQuery.Builder();
query.setSlop(slop);
int sizeMinus1 = termArrays.size() - 1;
for (int i = 0; i < sizeMinus1; i++) {
query.add(termArrays.get(i), positions.get(i));
}
Term[] suffixTerms = termArrays.get(sizeMinus1);
int position = positions.get(sizeMinus1);
ObjectHashSet<Term> terms = new ObjectHashSet<>();
for (Term term : suffixTerms) {
getPrefixTerms(terms, term, reader);
if (terms.size() > maxExpansions) {
break;
}
}
if (terms.isEmpty()) {
return Queries.newMatchNoDocsQuery();
}
query.add(terms.toArray(Term.class), position);
return query.build();
}
// TODO: Move to QueryParserTestBase once standard flexible parser gets this capability
public void testMultiWordSynonyms() throws Exception {
QueryParser dumb = new QueryParser("field", new Analyzer1());
dumb.setSplitOnWhitespace(false);
// A multi-word synonym source will form a synonym query for the same-starting-position tokens
BooleanQuery.Builder multiWordExpandedBqBuilder = new BooleanQuery.Builder();
Query multiWordSynonymQuery =
new SynonymQuery(new Term("field", "guinea"), new Term("field", "cavy"));
multiWordExpandedBqBuilder.add(multiWordSynonymQuery, BooleanClause.Occur.SHOULD);
multiWordExpandedBqBuilder.add(
new TermQuery(new Term("field", "pig")), BooleanClause.Occur.SHOULD);
Query multiWordExpandedBq = multiWordExpandedBqBuilder.build();
assertEquals(multiWordExpandedBq, dumb.parse("guinea pig"));
// With the phrase operator, a multi-word synonym source will form a multiphrase query.
// When the number of expanded term(s) is different from that of the original term(s), this is
// not good.
MultiPhraseQuery.Builder multiWordExpandedMpqBuilder = new MultiPhraseQuery.Builder();
multiWordExpandedMpqBuilder.add(
new Term[] {new Term("field", "guinea"), new Term("field", "cavy")});
multiWordExpandedMpqBuilder.add(new Term("field", "pig"));
Query multiWordExpandedMPQ = multiWordExpandedMpqBuilder.build();
assertEquals(multiWordExpandedMPQ, dumb.parse("\"guinea pig\""));
// custom behavior, the synonyms are expanded, unless you use quote operator
QueryParser smart = new SmartQueryParser();
smart.setSplitOnWhitespace(false);
assertEquals(multiWordExpandedBq, smart.parse("guinea pig"));
PhraseQuery.Builder multiWordUnexpandedPqBuilder = new PhraseQuery.Builder();
multiWordUnexpandedPqBuilder.add(new Term("field", "guinea"));
multiWordUnexpandedPqBuilder.add(new Term("field", "pig"));
Query multiWordUnexpandedPq = multiWordUnexpandedPqBuilder.build();
assertEquals(multiWordUnexpandedPq, smart.parse("\"guinea pig\""));
}
/** forms multiphrase query */
public void testSynonymsPhrase() throws Exception {
MultiPhraseQuery.Builder expectedBuilder = new MultiPhraseQuery.Builder();
expectedBuilder.add(new Term("field", "old"));
expectedBuilder.add(new Term[] {new Term("field", "dogs"), new Term("field", "dog")});
QueryBuilder builder = new QueryBuilder(new MockSynonymAnalyzer());
assertEquals(expectedBuilder.build(), builder.createPhraseQuery("field", "old dogs"));
}
/** forms multiphrase query */
public void testCJKSynonymsPhrase() throws Exception {
MultiPhraseQuery.Builder expectedBuilder = new MultiPhraseQuery.Builder();
expectedBuilder.add(new Term("field", "中"));
expectedBuilder.add(new Term[] {new Term("field", "国"), new Term("field", "國")});
QueryBuilder builder = new QueryBuilder(new MockCJKSynonymAnalyzer());
assertEquals(expectedBuilder.build(), builder.createPhraseQuery("field", "中国"));
expectedBuilder.setSlop(3);
assertEquals(expectedBuilder.build(), builder.createPhraseQuery("field", "中国", 3));
}
/** forms multiphrase query */
public void testCJKSynonymsPhrase() throws Exception {
MultiPhraseQuery.Builder expectedQBuilder = new MultiPhraseQuery.Builder();
expectedQBuilder.add(new Term("field", "中"));
expectedQBuilder.add(new Term[] {new Term("field", "国"), new Term("field", "國")});
QueryParser qp = new QueryParser("field", new MockCJKSynonymAnalyzer());
qp.setDefaultOperator(Operator.AND);
assertEquals(expectedQBuilder.build(), qp.parse("\"中国\""));
Query expected = new BoostQuery(expectedQBuilder.build(), 2f);
assertEquals(expected, qp.parse("\"中国\"^2"));
expectedQBuilder.setSlop(3);
expected = new BoostQuery(expectedQBuilder.build(), 2f);
assertEquals(expected, qp.parse("\"中国\"~3^2"));
}
/** forms multiphrase query */
public void testSynonymsPhrase() throws Exception {
MultiPhraseQuery.Builder expectedQBuilder = new MultiPhraseQuery.Builder();
expectedQBuilder.add(new Term(FIELD, "old"));
expectedQBuilder.add(new Term[] {new Term(FIELD, "dogs"), new Term(FIELD, "dog")});
QueryParser qp = new QueryParser(FIELD, new MockSynonymAnalyzer());
assertEquals(expectedQBuilder.build(), qp.parse("\"old dogs\""));
qp.setDefaultOperator(Operator.AND);
assertEquals(expectedQBuilder.build(), qp.parse("\"old dogs\""));
BoostQuery expected = new BoostQuery(expectedQBuilder.build(), 2f);
assertEquals(expected, qp.parse("\"old dogs\"^2"));
expectedQBuilder.setSlop(3);
expected = new BoostQuery(expectedQBuilder.build(), 2f);
assertEquals(expected, qp.parse("\"old dogs\"~3^2"));
}