internal Automaton ToLevenshteinAutomata(Automaton automaton)
{
var @ref = SpecialOperations.GetFiniteStrings(automaton, -1);
Automaton[] subs = new Automaton[@ref.Count];
int upto = 0;
foreach (IntsRef path in @ref)
{
if (path.Length <= nonFuzzyPrefix || path.Length < minFuzzyLength)
{
subs[upto] = BasicAutomata.MakeString(path.Ints, path.Offset, path.Length);
upto++;
}
else
{
Automaton prefix = BasicAutomata.MakeString(path.Ints, path.Offset, nonFuzzyPrefix);
int[] ints = new int[path.Length - nonFuzzyPrefix];
Array.Copy(path.Ints, path.Offset + nonFuzzyPrefix, ints, 0, ints.Length);
// TODO: maybe add alphaMin to LevenshteinAutomata,
// and pass 1 instead of 0? We probably don't want
// to allow the trailing dedup bytes to be
// edited... but then 0 byte is "in general" allowed
// on input (but not in UTF8).
LevenshteinAutomata lev = new LevenshteinAutomata(ints, unicodeAware ? char.MAX_CODE_POINT : 255, transpositions);
Automaton levAutomaton = lev.ToAutomaton(maxEdits);
Automaton combined = BasicOperations.Concatenate(Arrays.AsList(prefix, levAutomaton));
combined.Deterministic = true; // its like the special case in concatenate itself, except we cloneExpanded already
subs[upto] = combined;
upto++;
}
}
if (subs.Length == 0)
{
// automaton is empty, there is no accepted paths through it
return(BasicAutomata.MakeEmpty()); // matches nothing
}
else if (subs.Length == 1)
{
// no synonyms or anything: just a single path through the tokenstream
return(subs[0]);
}
else
{
// multiple paths: this is really scary! is it slow?
// maybe we should not do this and throw UOE?
Automaton a = BasicOperations.Union(Arrays.AsList(subs));
// TODO: we could call toLevenshteinAutomata() before det?
// this only happens if you have multiple paths anyway (e.g. synonyms)
BasicOperations.Determinize(a);
return(a);
}
}
/// <summary>
/// Initialize levenshtein DFAs up to maxDistance, if possible </summary>
private IList <CompiledAutomaton> InitAutomata(int maxDistance)
{
IList <CompiledAutomaton> runAutomata = dfaAtt.Automata;
//System.out.println("cached automata size: " + runAutomata.size());
if (runAutomata.Count <= maxDistance && maxDistance <= LevenshteinAutomata.MAXIMUM_SUPPORTED_DISTANCE)
{
LevenshteinAutomata builder = new LevenshteinAutomata(UnicodeUtil.NewString(m_termText, m_realPrefixLength, m_termText.Length - m_realPrefixLength), transpositions);
for (int i = runAutomata.Count; i <= maxDistance; i++)
{
Automaton a = builder.ToAutomaton(i);
//System.out.println("compute automaton n=" + i);
// constant prefix
if (m_realPrefixLength > 0)
{
Automaton prefix = BasicAutomata.MakeString(UnicodeUtil.NewString(m_termText, 0, m_realPrefixLength));
a = BasicOperations.Concatenate(prefix, a);
}
runAutomata.Add(new CompiledAutomaton(a, true, false));
}
}
return(runAutomata);
}
/// <summary>
/// initialize levenshtein DFAs up to maxDistance, if possible </summary>
private IList<CompiledAutomaton> InitAutomata(int maxDistance)
{
IList<CompiledAutomaton> runAutomata = DfaAtt.Automata();
//System.out.println("cached automata size: " + runAutomata.size());
if (runAutomata.Count <= maxDistance && maxDistance <= LevenshteinAutomata.MAXIMUM_SUPPORTED_DISTANCE)
{
LevenshteinAutomata builder = new LevenshteinAutomata(UnicodeUtil.NewString(TermText, RealPrefixLength, TermText.Length - RealPrefixLength), Transpositions);
for (int i = runAutomata.Count; i <= maxDistance; i++)
{
Automaton a = builder.ToAutomaton(i);
//System.out.println("compute automaton n=" + i);
// constant prefix
if (RealPrefixLength > 0)
{
Automaton prefix = BasicAutomata.MakeString(UnicodeUtil.NewString(TermText, 0, RealPrefixLength));
a = BasicOperations.Concatenate(prefix, a);
}
runAutomata.Add(new CompiledAutomaton(a, true, false));
}
}
return runAutomata;
}
/// <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()]*/));
}