public virtual void TestRewritePrefix()
{
Automaton pfx = BasicAutomata.MakeString("do");
pfx.ExpandSingleton(); // expand singleton representation for testing
Automaton prefixAutomaton = BasicOperations.Concatenate(pfx, BasicAutomata.MakeAnyString());
AutomatonQuery aq = new AutomatonQuery(NewTerm("bogus"), prefixAutomaton);
Terms terms = MultiFields.GetTerms(Searcher.IndexReader, FN);
Assert.IsTrue(aq.GetTermsEnum(terms) is PrefixTermsEnum);
Assert.AreEqual(3, AutomatonQueryNrHits(aq));
}
public virtual void TestBasicAutomata()
{
AssertAutomatonHits(0, BasicAutomata.MakeEmpty());
AssertAutomatonHits(0, BasicAutomata.MakeEmptyString());
AssertAutomatonHits(2, BasicAutomata.MakeAnyChar());
AssertAutomatonHits(3, BasicAutomata.MakeAnyString());
AssertAutomatonHits(2, BasicAutomata.MakeString("doc"));
AssertAutomatonHits(1, BasicAutomata.MakeChar('a'));
AssertAutomatonHits(2, BasicAutomata.MakeCharRange('a', 'b'));
AssertAutomatonHits(2, BasicAutomata.MakeInterval(1233, 2346, 0));
AssertAutomatonHits(1, BasicAutomata.MakeInterval(0, 2000, 0));
AssertAutomatonHits(2, BasicOperations.Union(BasicAutomata.MakeChar('a'), BasicAutomata.MakeChar('b')));
AssertAutomatonHits(0, BasicOperations.Intersection(BasicAutomata.MakeChar('a'), BasicAutomata.MakeChar('b')));
AssertAutomatonHits(1, BasicOperations.Minus(BasicAutomata.MakeCharRange('a', 'b'), BasicAutomata.MakeChar('a')));
}
/// <summary>
/// Convert Lucene wildcard syntax into an automaton.
/// <para/>
/// @lucene.internal
/// </summary>
public static Automaton ToAutomaton(Term wildcardquery)
{
IList <Automaton> automata = new JCG.List <Automaton>();
string wildcardText = wildcardquery.Text;
for (int i = 0; i < wildcardText.Length;)
{
int c = Character.CodePointAt(wildcardText, i);
int length = Character.CharCount(c);
switch (c)
{
case WILDCARD_STRING:
automata.Add(BasicAutomata.MakeAnyString());
break;
case WILDCARD_CHAR:
automata.Add(BasicAutomata.MakeAnyChar());
break;
case WILDCARD_ESCAPE:
// add the next codepoint instead, if it exists
if (i + length < wildcardText.Length)
{
int nextChar = Character.CodePointAt(wildcardText, i + length);
length += Character.CharCount(nextChar);
automata.Add(BasicAutomata.MakeChar(nextChar));
break;
} // else fallthru, lenient parsing with a trailing \
goto default;
default:
automata.Add(BasicAutomata.MakeChar(c));
break;
}
i += length;
}
return(BasicOperations.Concatenate(automata));
}
/// <summary>
/// Extracts all <see cref="MultiTermQuery"/>s for <paramref name="field"/>, and returns equivalent
/// automata that will match terms.
/// </summary>
internal static CharacterRunAutomaton[] ExtractAutomata(Query query, string field)
{
List <CharacterRunAutomaton> list = new List <CharacterRunAutomaton>();
if (query is BooleanQuery)
{
BooleanClause[] clauses = ((BooleanQuery)query).GetClauses();
foreach (BooleanClause clause in clauses)
{
if (!clause.IsProhibited)
{
list.AddAll(Arrays.AsList(ExtractAutomata(clause.Query, field)));
}
}
}
else if (query is DisjunctionMaxQuery)
{
foreach (Query sub in ((DisjunctionMaxQuery)query).Disjuncts)
{
list.AddAll(Arrays.AsList(ExtractAutomata(sub, field)));
}
}
else if (query is SpanOrQuery)
{
foreach (Query sub in ((SpanOrQuery)query).GetClauses())
{
list.AddAll(Arrays.AsList(ExtractAutomata(sub, field)));
}
}
else if (query is SpanNearQuery)
{
foreach (Query sub in ((SpanNearQuery)query).GetClauses())
{
list.AddAll(Arrays.AsList(ExtractAutomata(sub, field)));
}
}
else if (query is SpanNotQuery)
{
list.AddAll(Arrays.AsList(ExtractAutomata(((SpanNotQuery)query).Include, field)));
}
else if (query is SpanPositionCheckQuery)
{
list.AddAll(Arrays.AsList(ExtractAutomata(((SpanPositionCheckQuery)query).Match, field)));
}
else if (query is ISpanMultiTermQueryWrapper)
{
list.AddAll(Arrays.AsList(ExtractAutomata(((ISpanMultiTermQueryWrapper)query).WrappedQuery, field)));
}
else if (query is AutomatonQuery)
{
AutomatonQuery aq = (AutomatonQuery)query;
if (aq.Field.Equals(field, StringComparison.Ordinal))
{
list.Add(new CharacterRunAutomatonToStringAnonymousHelper(aq.Automaton, () => aq.ToString()));
}
}
else if (query is PrefixQuery)
{
PrefixQuery pq = (PrefixQuery)query;
Term prefix = pq.Prefix;
if (prefix.Field.Equals(field, StringComparison.Ordinal))
{
list.Add(new CharacterRunAutomatonToStringAnonymousHelper(
BasicOperations.Concatenate(BasicAutomata.MakeString(prefix.Text()), BasicAutomata.MakeAnyString()),
() => pq.ToString()));
}
}
else if (query is FuzzyQuery)
{
FuzzyQuery fq = (FuzzyQuery)query;
if (fq.Field.Equals(field, StringComparison.Ordinal))
{
string utf16 = fq.Term.Text();
int[] termText = new int[utf16.CodePointCount(0, utf16.Length)];
for (int cp, i = 0, j = 0; i < utf16.Length; i += Character.CharCount(cp))
{
termText[j++] = cp = utf16.CodePointAt(i);
}
int termLength = termText.Length;
int prefixLength = Math.Min(fq.PrefixLength, termLength);
string suffix = UnicodeUtil.NewString(termText, prefixLength, termText.Length - prefixLength);
LevenshteinAutomata builder = new LevenshteinAutomata(suffix, fq.Transpositions);
Automaton automaton = builder.ToAutomaton(fq.MaxEdits);
if (prefixLength > 0)
{
Automaton prefix = BasicAutomata.MakeString(UnicodeUtil.NewString(termText, 0, prefixLength));
automaton = BasicOperations.Concatenate(prefix, automaton);
}
list.Add(new CharacterRunAutomatonToStringAnonymousHelper(automaton, () => fq.ToString()));
}
}
else if (query is TermRangeQuery)
{
TermRangeQuery tq = (TermRangeQuery)query;
if (tq.Field.Equals(field, StringComparison.Ordinal))
{
// this is *not* an automaton, but its very simple
list.Add(new SimpleCharacterRunAutomatonAnonymousHelper(BasicAutomata.MakeEmpty(), tq));
}
}
return(list.ToArray(/*new CharacterRunAutomaton[list.size()]*/));
}