Class NFA
Contains algorithms RegExp → NFA.
- Version:
- JFlex 1.9.1
-
Field Summary
FieldsModifier and TypeFieldDescriptionprivate Action[]action[current_state]: the action associated with the state current_state (null, if there is no action for the state)private CharClassesprivate StateSet[]epsilon[current_state] is the set of states that can be reached from current_state via epsilon edgesprivate final intestimated size of the NFA (before actual construction)private boolean[]isFinal[state] == true invalid input: '<'=> state is a final state of the NFAprivate final intthe current maximum number of input charactersprivate intthe number of lexical States.private intthe number of states in this NFAprivate RegExpsprivate LexScanprivate final StateSetEnumeratorprivate StateSet[][]table[current_state][next_char] is the set of states that can be reached from current_state with an input next_charprivate final StateSet -
Constructor Summary
Constructors -
Method Summary
Modifier and TypeMethodDescriptionvoidaddEpsilonTransition(int start, int dest) voidaddRegExp(int regExpNum) Add a regexp to this NFA.voidAdd a standalone rule that has minimum priority, fires a transition on all single input characters and has a "print yytext" action.voidaddTransition(int start, int input, int dest) private StateSetclosure(int startState) Calculates the epsilon closure for a specified set of states.private IntPaircomplement(IntPair nfa) Constructs an NFA accepting the complement of the language of a given NFA.booleancontainsFinal(StateSet set) Returnstrue, iff the specified set of states contains a final state.private StateSetCalculates the set of states that can be reached from another set of statesstartwith an specified input characterinputvoidprivate voidensureCapacity(int newNumStates) Make sure the NFA can contain at least newNumStates states.epsilon(int i) voidReturns the action with highest priority in the specified set of states.private voidinsertCCLNFA(RegExp regExp, int start, int end) Constructs a two state NFA for char class regexps, such that the NFA has exactly one start state, exactly one end state, no transitions leading out of the end state, no transitions leading into the start state.private voidinsertClassNFA(IntCharSet set, int start, int end) private voidinsertLetterNFA(boolean caseless, int ch, int start, int end) private voidinsertLookAheadChoices(int baseEnd, Action a, RegExp lookAhead) Insert NFAs for the (finitely many) fixed length lookahead choices.Constructs an NFA for regExp such that the NFA hasprivate IntPairinsertStringNFA(boolean caseless, String str) intintnumInput()intintreachableStates(int currentState, int nextChar) Returns the set of states that can be reached from currentState with an input nextChar.private voidremoveDead(int start, int end) Find all states from (numerically)startto @endthat (transitively) cannot reach reachend, and remove the transitions leading to those states.states()toString()void
-
Field Details
-
table
table[current_state][next_char] is the set of states that can be reached from current_state with an input next_char -
epsilon
epsilon[current_state] is the set of states that can be reached from current_state via epsilon edges -
isFinal
private boolean[] isFinalisFinal[state] == true invalid input: '<'=> state is a final state of the NFA -
action
action[current_state]: the action associated with the state current_state (null, if there is no action for the state) -
numStates
private int numStatesthe number of states in this NFA -
numInput
private final int numInputthe current maximum number of input characters -
numLexStates
private int numLexStatesthe number of lexical States. Lexical states have the indices 0..numLexStates-1 in the transition table -
estSize
private final int estSizeestimated size of the NFA (before actual construction) -
classes
-
scanner
-
regExps
-
states
-
tempStateSet
-
-
Constructor Details
-
NFA
public NFA(int numInput, int estSize) Constructor for NFA. -
NFA
Construct new NFA.Assumes that lookahead cases and numbers are already resolved in RegExps.
- Parameters:
numInput- a int.scanner- aLexScanobject.regExps- aRegExpsobject.macros- aMacrosobject.classes- aCharClassesobject.- See Also:
-
-
Method Details
-
epsilon
-
numEntryStates
public int numEntryStates() -
numInput
public int numInput() -
numLexStates
public int numLexStates() -
numStates
public int numStates() -
reachableStates
Returns the set of states that can be reached from currentState with an input nextChar. -
states
-
tempStateSet
-
addStandaloneRule
public void addStandaloneRule()Add a standalone rule that has minimum priority, fires a transition on all single input characters and has a "print yytext" action. -
addRegExp
public void addRegExp(int regExpNum) Add a regexp to this NFA.- Parameters:
regExpNum- the number of the regexp to add.
-
insertLookAheadChoices
Insert NFAs for the (finitely many) fixed length lookahead choices.- Parameters:
baseEnd- the end state of the base expression NFAa- the action of the expressionlookAhead- a lookahead of which isFiniteChoice is true- See Also:
-
ensureCapacity
private void ensureCapacity(int newNumStates) Make sure the NFA can contain at least newNumStates states.- Parameters:
newNumStates- the minimum number of states.
-
addTransition
public void addTransition(int start, int input, int dest) -
addEpsilonTransition
public void addEpsilonTransition(int start, int dest) -
containsFinal
Returnstrue, iff the specified set of states contains a final state.- Parameters:
set- the set of states that is tested for final states.
-
getAction
Returns the action with highest priority in the specified set of states.- Parameters:
set- the set of states for which to determine the action
-
closure
Calculates the epsilon closure for a specified set of states.The epsilon closure for set a is the set of states that can be reached by epsilon edges from a.
- Parameters:
startState- the start state for the set of states to calculate the epsilon closure for- Returns:
- the epsilon closure of the specified set of states in this NFA
-
epsilonFill
public void epsilonFill() -
DFAEdge
Calculates the set of states that can be reached from another set of statesstartwith an specified input characterinput- Parameters:
start- the set of states to start frominput- the input character for which to search the next states- Returns:
- the set of states that are reached from
start</code> via <code>input
-
dumpTable
public void dumpTable() -
toString
-
writeDot
-
dotFormat
-
insertLetterNFA
private void insertLetterNFA(boolean caseless, int ch, int start, int end) -
insertStringNFA
-
insertClassNFA
-
complement
Constructs an NFA accepting the complement of the language of a given NFA.Converts the NFA into a DFA, then negates that DFA. Exponential state blowup possible and common.
- Parameters:
nfa- the NFA to construct the complement for.- Returns:
- a pair of integers denoting the index of start and end state of the complement NFA.
-
removeDead
private void removeDead(int start, int end) Find all states from (numerically)startto @endthat (transitively) cannot reach reachend, and remove the transitions leading to those states.After a complement operation, there may be dead states left over in the NFA, which could lead the scanning engine into a situation where it is trying to perform lookahead even though no final state can ever be reached.
Precondition: all states that potentially lead to
endare within the interval @{code [start,end]}. This is satisfied by DFA generation in the complement operation.Precondition: end state has no outgoing transitions
- Parameters:
start- the first state from which to compute live statesend- the state that if it can be reached makes a state live- See Also:
-
insertCCLNFA
Constructs a two state NFA for char class regexps, such that the NFA has- exactly one start state,
- exactly one end state,
- no transitions leading out of the end state,
- no transitions leading into the start state.
Assumes that regExp.isCharClass(macros) == true
- Parameters:
regExp- the regular expression to construct the NFA for
-
insertNFA
Constructs an NFA for regExp such that the NFA hasexactly one start state, exactly one end state, no transitions leading out of the end state no transitions leading into the start state
- Parameters:
regExp- the regular expression to construct the NFA for- Returns:
- a pair of integers denoting the index of start and end state of the NFA.
-