Finite state tool (FST)

Finite state tool (FST) is a tool for working with different **finite-state machines** and many algorithmic constructions. The projects consists of four components:

1. Finite state automation
2. Pushdown automation
3. Context-free grammars
4. Minimal acyclic subsequential transducer

1-3 are developed for the Languages, Automata and computability course at the Faculty of Mathematics and Informatics, Sofia university. All algorithms are direct implementations of the shown constructions here and may not have optimal time and space complexity

4 is developed for an advanced course of the same topic and it's tested with real-world data.

1-3 and 4 should be compiled separatly!

FST interface and commands

Syntax:	Effect:	Example:
load [config_file]	loads configuration from a file. Example files in Example_config_files	load myfsa.fst
environment fsa	prints information for all registered fsa-s	environment fsa
regex [fsa1]	returns a regex for the automation.	regex a1
fsa [name]	registers an automation with one state with this name.	fsa test1
fsa [name] [regex]	registers an automation with this name for this regex.	fsa test2 a*(a+b)*b
print [name]	prints the automation.	print test1
vis [name]	generates an html file with an image of the automation.	vis test1
accepts [fsa] [string]	returns true if the fsa accepts the string and false otherwise	accepts a1 aabbb
add_state [fsa]	adds a new state to the automation	add_state a1
make_final [fsa] [state]	makes the state final	make_final a1 0
arc [fsa] [start] [end] [symbol]	adds a new transition from state start to state end with this symbol	arc a1 0 1 a
det [fsa]	fsa becomes deterministc	det a1
[fsa1] = det [fsa]	registers an automation fsa1 which is the deterministic fsa2	a1 = det a
min [fsa]	fsa becomes deterministic, total and minimal	min a1
[fsa1] = min [fsa]	registers an automation fsa1 which is the deterministic and minimal fsa2	a1 = min a2
[fsa1] = union [fsa2] [fsa3]	registers an automation fsa1 which is the union of fsa2 and fsa3	a1 = union a2 a3
[fsa1] = concat [fsa2] [fsa3]	registers an automation fsa1 which is the concatenation of fsa2 and fsa3	a1 = concat a2 a3
[fsa1] = intersect [fsa2] [fsa3]	registers an automation fsa1 which is the intersection of fsa2 and fsa3	a1 = intersect a2 a3
star [fsa1]	fsa1 becomes an new automation for the Kleene star of the old fsa1 automation	star a1
[fsa1] = star [fsa2]	registers an automation fsa1 which is the Kleene star of fsa2	a1 = star a2
compl [fsa1]	fsa1 becomes an new automation for the complement of the old fsa1 automation	compl a1
[fsa1] = compl [fsa2]	registers an automation fsa1 which is the complement of fsa2	a1 = compl a2
reverse [fsa1]	fsa1 becomes an new automation for the reverse of the old fsa1 automation	reverse a1
[fsa1] = reverse [fsa2]	registers an automation fsa1 which is the reverse of fsa2	a1 = reverse a2
regex [fsa1]	returns a regex for the automation.	regex a1
environment npda	prints information for all registered npda-s	environment npda
npda [name]	registers an Pushdown automation with one state with this name.	npda test1
add_state [npda]	adds a new state to the npda	add_state test1
make_final [npda] [state]	makes the state final	make_final test1 0
arc [npda] [start] [symbol] [stack_top] [end] [replace_stack]	adds a new transition from state start to state end	arc test1 0 a # 0 A#
accepts [npda] [string]	returns true if the npda accepts the string and false otherwise	accepts test1 aabbb
cfg [name] [rules]	Creates a context-free grammer. The rules are separated with a space.	cfg test1 S->aSb|$

! Epsilon in the regex is denoted with by the $ !

Visualization tool

The FST system supports automata visualization. The command is called vis.

This commands generates a html file (with the name of the automation). You can find it in the main folder.

The html file looks like this:

1. Finite State Automation

Finite state automation is A = <Q,Σ,s,F,δ>, where
Q - is a finite, non-empty set of states
Σ- is the input alphabet (a finite, non-empty set of symbols).
s - is an initial state, an element of Q
F- is the set of final states, a subset of Q
δ- is the state-transition function

Creation

Diffrent ways of creating an automation:

int main()
{

	// Automation for a(a+b)*
	FiniteStateAutomation A("a(a+b)*");
	
	std::string computation;
	cout << A.accepts("abbb", computation) << endl; //true;
	cout << A.accepts("bbba", computation) << endl; //false

	FiniteStateAutomation A2; //Only one state with index 0
	
	A2.addState(); //Adds state with index 1
	A2.addTransition({"0", "1", "a"});
	A2.addTransition({"1", "1", "a"});
	A2.addTransition({"1", "1", "b"});


	return 0;
}

Control

addState()	Adds a new state. Returns the index of the new state.
addTransition(start,end,ch)	Creates a new transition from state start to state end with the letter ch. May become from DFA to NFA.
changeStartState(newStart)	Changes the start state of the automation. Returns false if the given new start state doesn’t exist and true if it was successful.
makeStateFinal(state)	Marks the given state as final. Returns false if the state doesn’t exist.
accepts(word, computation, shouldReturnComputation)	Returns true if the word is accepted by the DFA/NFA. If the third parameter is true, then the computation will contain the successful (if such exists) path
isEmptyLanguage()	Returns true if the FSA doesn't accept any strings.
removeNotReachableStates()	Removes all states that are not reachable from the start state..

Properties

makeTotal()	Makes the automation complete. It shoud define a transition for each state and each input symbol. We create an error state for every non-existing transition.
makeDeterministic()	Makes the Non-deterministic finite automaton(NFA) to Deterministic finite Automation(DFA).
minimize()	Transforms the given DETERMINISTIC finite automaton into an equivalent DFA that has a minimum number of states.
getRegEx()	Gets a regular expression for the language accepted by the NFA. The regular expression may be non-practical as it becomes long very fast.

Operations

All basic operations with NFA-s are available:

Union(A1,A2)	Concat(A1,A2)	KleeneStar(A1)	Complement(A1)	Intersect(A1,A2)	Reverse(A1)

Example of creating and minimizing a FSA

#include "FiniteStateAutomation.h"

int main() 
{
	//Regex: a(a+b)*b + b(a+b)*a
	FiniteStateAutomation A("a(a+b)*b+b(a+b)*a");

	A.print(); //Image 1 and 2

	A.minimize();

	A.print(); // Image 3 and 4.

   	return 0;
}

The first print:

Without the unreachable states it looks like:

And after minimizing the automation, the second print:

It looks like this:

Example of getting a regex from FSA

int main() 
{	
	// FSA for ab(a+b)*
	FiniteStateAutomation A("ab(a+b)*");
		
	A.minimize(); //better to be minimized, so the regex would be simple.

	A.print();
	
	cout << A.getRegEx();

   	return 0;
}

Here is the FSA:

And the result:

ab+(ab(a+b)*(e+a+b))

= ab + (ab(a+b)*) (since e,a and b are in (a+b)*)

= ab(a+b)* (since ab is in ab(a+b)*)

2. Non-deterministic Pushdown Automation

.Non-deterministic Pushdown Automation is P = <Q,Σ,G,#, s,F,δ>, where
Q - is a finite, non-empty set of states
Σ- is the input alphabet (a finite, non-empty set of symbols).
G- finite set which is called the stack alphabet.
# - initial stack symbol.
s - is an initial state, an element of Q
F- is the set of final states, a subset of Q
δ- is the transition function

! The empty string/symbol is $ !

Each rule(transition) looks like this:

<current state, read symbol from the tape, top of the stack, destination state, string to replace the top of stack>

• Example 1: <0,'a', 'A', 1, "BA"> From state 0 to state 1, we push B to the stack
• Example 2: <0,'b', 'A', 3, "$"> From state 0 to state 3, we pop A from the stack (since $ is the empty string)

Example for NPDA for { ww^rev | w in {a,b} }*

// Example for Nondeterministic pushdown automata for { ww^rev | w in {a,b}* }
int main()
{
	NPDA PA(3); //3 initial states

	PA.makeStateFinal(2);

	PA.addTransition({ "0", "a", "#", "0", "A#" });

	PA.addTransition({ "0", "b", "#", "0", "B#" });
	PA.addTransition({ "0", "$", "#", "2", "$" });

	PA.addTransition({ "0", "a", "A", "0", "AA" });
	PA.addTransition({ "0", "a", "A", "1", "$" });

	PA.addTransition({ "0", "b", "B", "0", "BB" });
	PA.addTransition({ "0", "b", "B", "1", "$" });

	PA.addTransition({ "0", "b", "A", "0", "BA" });
	PA.addTransition({ "0", "a", "B", "0", "AB" });

	PA.addTransition({ "1", "a", "A", "1", "$" });
	PA.addTransition({ "1", "b", "B", "1", "$" });

	PA.addTransition({ "1", "$", "#", "2", "$" });

	string computation;
	std::cout << PA.accepts("abba", computation) << std::endl; //true
	std::cout << PA.accepts("abbb", computation) << std::endl; //false
	std::cout << PA.accepts("aaabbbbbbaaa", computation) << std::endl; //true
}

If you are looking for simple PDA simulation program, you can use: this

3. Context-Free Grammar

We can input out context-free grammar (CFG) and it will simulate it using NPDA.

Example for simulating the CFG:

S->aSc|B

B->bB|$

int main()
{
	//Example Nondeterministic pushdown automata  from a context-free grammar
	// S->aSc | B
	// B->bB | $
	// L(S) = { a^n b^k c^n | n,k \in N}
	ContextFreeGrammar cfg;
	cfg.grammarRules.push_back("S->aSc|B");
	cfg.grammarRules.push_back("B->bB|$");

	NPDA PA2(cfg);

	std::cout << PA2.accepts("abc", true) << std::endl; //true
	std::cout << PA2.accepts("aaaaaabbbbcccccc") << std::endl; //true
	std::cout << PA2.accepts("abcc") << std::endl; //false
}

4. Minimal acyclic subsequential transducer

The project contains algorithms for creating and maintaining a minimal acyclic subsequential transducer for a finite function f : Σ* -> N

(MAST for [a, 11], [ab, 3], [ac, 1])

The project consist of three functions:

• Building and MAST from a given sorted list of pairs <key, value>. It follows the aproach in [Mihov and Maurel, 2001] Mihov, S. and Maurel, D. (2001). Direct construction of minimal acyclic subsequential transducers.
• Inserting words to the MAST from a given list of pairs
• Removing words to the MAST from a given list of pairs

All the algorithms can be easy modified for a function of the type f : Σ* -> Σ*

The check for equivalent states is done with a proper hash. Each state is hashed using it's transitions, output and finality. More information about the hash on SubsequentialTransducer.h:18