def testSpanNearQuery(self):
quick_brown_dog = [self.quick, self.brown, self.dog]
snq = SpanNearQuery(quick_brown_dog, 0, True)
self.assertNoMatches(snq)
self.dumpSpans(snq)
snq = SpanNearQuery(quick_brown_dog, 4, True)
self.assertNoMatches(snq)
self.dumpSpans(snq)
snq = SpanNearQuery(quick_brown_dog, 5, True)
self.assertOnlyBrownFox(snq)
self.dumpSpans(snq)
# interesting - even a sloppy phrase query would require
# more slop to match
snq = SpanNearQuery([self.lazy, self.fox], 3, False)
self.assertOnlyBrownFox(snq)
self.dumpSpans(snq)
pq = PhraseQuery()
pq.add(Term("f", "lazy"))
pq.add(Term("f", "fox"))
pq.setSlop(4)
self.assertNoMatches(pq)
pq.setSlop(5)
self.assertOnlyBrownFox(pq)
def testSpanNearQuery(self):
quick_brown_dog = [self.quick, self.brown, self.dog]
snq = SpanNearQuery(quick_brown_dog, 0, True)
self.assertNoMatches(snq)
self.dumpSpans(snq)
snq = SpanNearQuery(quick_brown_dog, 4, True)
self.assertNoMatches(snq)
self.dumpSpans(snq)
snq = SpanNearQuery(quick_brown_dog, 5, True)
self.assertOnlyBrownFox(snq)
self.dumpSpans(snq)
# interesting - even a sloppy phrase query would require
# more slop to match
snq = SpanNearQuery([self.lazy, self.fox], 3, False)
self.assertOnlyBrownFox(snq)
self.dumpSpans(snq)
pq = PhraseQuery()
pq.add(Term("f", "lazy"))
pq.add(Term("f", "fox"))
pq.setSlop(4)
self.assertNoMatches(pq)
pq.setSlop(5)
self.assertOnlyBrownFox(pq)
def matched(self, phrase, slop):
query = PhraseQuery()
query.setSlop(slop)
for word in phrase:
query.add(Term("field", word))
topDocs = self.searcher.search(query, 50)
return topDocs.totalHits > 0
def testSearchByAPI(self):
tq = TermQuery(Term("content", "hops"))
topDocs = self.searcher.search(tq, 50)
self.assertEqual(1, topDocs.totalHits)
pq = PhraseQuery()
pq.add(Term("content", "fox"))
pq.add(Term("content", "hops"))
topDocs = self.searcher.search(pq, 50)
self.assertEquals(1, topDocs.totalHits)
def getFieldQuery_slop(self, field, queryText, slop):
orig = super(CustomQueryParser,
self).getFieldQuery_slop_super(field, queryText, slop)
if not PhraseQuery.instance_(orig):
return orig
pq = PhraseQuery.cast_(orig)
clauses = [SpanTermQuery(term) for term in pq.getTerms()]
return SpanNearQuery(clauses, slop, True);
def matched(self, phrase, slop):
query = PhraseQuery()
query.setSlop(slop)
for word in phrase:
query.add(Term("field", word))
topDocs = self.searcher.search(query, 50)
return topDocs.totalHits > 0
def testSearchByAPI(self):
tq = TermQuery(Term("content", "hops"))
topDocs = self.searcher.search(tq, 50)
self.assertEqual(1, topDocs.totalHits)
pq = PhraseQuery()
pq.add(Term("content", "fox"))
pq.add(Term("content", "hops"))
topDocs = self.searcher.search(pq, 50)
self.assertEquals(1, topDocs.totalHits)
def getFieldQuery(self, field, queryText, slop=None):
if slop is None:
return super(CustomQueryParser,
self).getFieldQuery(field, queryText)
# let QueryParser's implementation do the analysis
orig = super(CustomQueryParser,
self).getFieldQuery(field, queryText, slop)
if "PhraseQuery" != orig.getClassName():
return orig
pq = PhraseQuery.cast_(orig)
clauses = [SpanTermQuery(term) for term in pq.getTerms()]
return SpanNearQuery(clauses, slop, True);
def testAgainstOR(self):
quickFox = PhraseQuery()
quickFox.setSlop(1)
quickFox.add(Term("field", "quick"))
quickFox.add(Term("field", "fox"))
fastFox = PhraseQuery()
fastFox.add(Term("field", "fast"))
fastFox.add(Term("field", "fox"))
query = BooleanQuery()
query.add(quickFox, BooleanClause.Occur.SHOULD)
query.add(fastFox, BooleanClause.Occur.SHOULD)
topDocs = self.searcher.search(query, 10)
self.assertEqual(2, topDocs.totalHits)
def testAgainstOR(self):
quickFox = PhraseQuery()
quickFox.setSlop(1)
quickFox.add(Term("field", "quick"))
quickFox.add(Term("field", "fox"))
fastFox = PhraseQuery()
fastFox.add(Term("field", "fast"))
fastFox.add(Term("field", "fox"))
query = BooleanQuery()
query.add(quickFox, BooleanClause.Occur.SHOULD)
query.add(fastFox, BooleanClause.Occur.SHOULD)
topDocs = self.searcher.search(query, 10)
self.assertEqual(2, topDocs.totalHits)