nl.inl.blacklab.search.fimatch

Class NfaState

java.lang.Object

nl.inl.blacklab.search.fimatch.NfaState

Direct Known Subclasses:

NfaStateAnd, NfaStateAndAcyclic, NfaStateMatch, NfaStateMultiTermPattern, NfaStateNot, NfaStateOr, NfaStateOrAcyclic, NfaStateToken

public abstract class NfaState
extends Object

Represents both a state in an NFA, and a complete NFA with this as the starting state. (Note that the "matching state" is simply represented as null in the NFAs we build, which is convenient when possibly appending other NFAs to it later)

Constructor Summary

Constructors

Constructor and Description
NfaState()

Method Summary

Modifier and Type	Method and Description
static NfaState	and(boolean clausesMayLoopBack, List<NfaState> nextStates) Build an AND state.
static NfaState	anyToken(String luceneField, NfaState nextState)
String	dump(Map<NfaState,Integer> stateNrs)
static String	dump(NfaState state, Map<NfaState,Integer> stateNrs)
protected abstract String	dumpInternal(Map<NfaState,Integer> stateNrs)
static Set<NfaState>	emptySet()
NavigableSet<Integer>	findMatches(ForwardIndexDocument fiDoc, int pos, int direction) Find all matches for this NFA in the token source.
void	finish(Set<NfaState> visited) Visit each node and replace dangling arrows (nulls) with the match state.
protected abstract void	finishInternal(Set<NfaState> visited) Visit each node and replace dangling arrows (nulls) with the match state.
abstract boolean	hitsAllSameLength(Set<NfaState> statesVisited) Are all hits from this NFA the same length?
abstract int	hitsLengthMax(Set<NfaState> statesVisited) What's the maximum hit length?
abstract int	hitsLengthMin(Set<NfaState> statesVisited) What's the minimum hit length?
void	lookupPropertyNumbers(ForwardIndexAccessor fiAccessor, Map<NfaState,Boolean> statesVisited)
static NfaState	match()
boolean	matches(ForwardIndexDocument fiDoc, int pos, int direction) Does the token source match this NFA?
abstract boolean	matchesEmptySequence(Set<NfaState> statesVisited) Does this NFA match the empty sequence?
static NfaState	or(boolean clausesMayLoopBack, List<NfaState> nextStates, boolean clausesAllSameLength) Build am OR state.
static NfaState	regex(String luceneField, String pattern, NfaState nextState)
abstract void	setNextState(int input, NfaState state) Set the next state for a given input.
static NfaState	token(String luceneField, Set<String> inputTokens, NfaState nextState) Build a token state.
static NfaState	token(String luceneField, String inputToken, NfaState nextState) Build a token state.
String	toString()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

NfaState

public NfaState()

Method Detail

token

public static NfaState token(String luceneField,
                             String inputToken,
                             NfaState nextState)

Build a token state.

Parameters:

luceneField - what annotation to match

inputToken - what token to match

nextState - what state to go to after a succesful match

Returns:

the state object

token

public static NfaState token(String luceneField,
                             Set<String> inputTokens,
                             NfaState nextState)

Build a token state.

Parameters:

luceneField - what annotation to match

inputTokens - what tokens to match

nextState - what state to go to after a succesful match

Returns:

the state object

regex

public static NfaState regex(String luceneField,
                             String pattern,
                             NfaState nextState)

anyToken

public static NfaState anyToken(String luceneField,
                                NfaState nextState)

or

public static NfaState or(boolean clausesMayLoopBack,
                          List<NfaState> nextStates,
                          boolean clausesAllSameLength)

Build am OR state.

Parameters:

clausesMayLoopBack - if false, no clauses loop back to earlier states, (a more efficient way of matching can be used in this case)

nextStates - states to try

clausesAllSameLength - are all hits for all clauses the same length? (used to optimize matching)

Returns:

the state object

and

public static NfaState and(boolean clausesMayLoopBack,
                           List<NfaState> nextStates)

Build an AND state.

Parameters:

clausesMayLoopBack - if false, no clauses loop back to earlier states, (a more efficient way of matching can be used in this case)

nextStates - NFAs that must match

Returns:

the state object

match

public static NfaState match()

findMatches

public NavigableSet<Integer> findMatches(ForwardIndexDocument fiDoc,
                                         int pos,
                                         int direction)

Find all matches for this NFA in the token source.

Parameters:

fiDoc - where to read tokens from

pos - current matching position

direction - matching direction

Returns:

the matches found, if any

matches

public boolean matches(ForwardIndexDocument fiDoc,
                       int pos,
                       int direction)

Does the token source match this NFA?

Parameters:

fiDoc - where to read tokens from

pos - current matching position

direction - matching direction

Returns:

true if fiDoc matches, false if not

setNextState

public abstract void setNextState(int input,
                                  NfaState state)

Set the next state for a given input.

Parameters:

input - input

state - next state

toString

public String toString()

Overrides:

toString in class Object

finish

public void finish(Set<NfaState> visited)

Visit each node and replace dangling arrows (nulls) with the match state.

Parameters:

visited - nodes visited so far, so we don't visit nodes multiple times

finishInternal

protected abstract void finishInternal(Set<NfaState> visited)

Visit each node and replace dangling arrows (nulls) with the match state.

Parameters:

visited - nodes visited so far, this one included. finish() uses this to make sure we don't visit the same node twice, so always call finish() in your implementations, not finishInternal().

dump

public static String dump(NfaState state,
                          Map<NfaState,Integer> stateNrs)

dump

public String dump(Map<NfaState,Integer> stateNrs)

dumpInternal

protected abstract String dumpInternal(Map<NfaState,Integer> stateNrs)

matchesEmptySequence

public abstract boolean matchesEmptySequence(Set<NfaState> statesVisited)

Does this NFA match the empty sequence?

Parameters:

statesVisited - states we've already visited, so we can deal with cycles

Returns:

true if it matches the empty sequence, false if not

hitsAllSameLength

public abstract boolean hitsAllSameLength(Set<NfaState> statesVisited)

Are all hits from this NFA the same length?

Parameters:

statesVisited - states we've already visited, so we can deal with cycles

Returns:

true if all hits are the same length, false if not

hitsLengthMin

public abstract int hitsLengthMin(Set<NfaState> statesVisited)

What's the minimum hit length?

Parameters:

statesVisited - states we've already visited, so we can deal with cycles

Returns:

minimum hit length

hitsLengthMax

public abstract int hitsLengthMax(Set<NfaState> statesVisited)

What's the maximum hit length?

Parameters:

statesVisited - states we've already visited, so we can deal with cycles

Returns:

maximum hit length

emptySet

public static Set<NfaState> emptySet()

lookupPropertyNumbers

public final void lookupPropertyNumbers(ForwardIndexAccessor fiAccessor,
                                        Map<NfaState,Boolean> statesVisited)