Fixed a bug with final states in minimization

dfa.cpp

@@ -238,7 +238,26 @@ machine build_dfa(machine &org_dfa, std::vector<std::vector<sid_t> > sets, sid_t
     for (int i = 0 ; i < sets.size() ; i++) {
         if (i != starting_set) {
             token_type = get_token_type(sets[i], org_dfa, is_final);
-            dfa.add_new_state(token_type, is_final);
+            dfa.add_new_state(token_type, false, is_final);
+        }
+    }
+
+
+    for (int i = 0 ; i < sets.size() ; i++) {
+        std::vector<sid_t> set = sets[i];
+        for (char input : org_dfa.get_language()) {
+            for (sid_t state : set) {
+                std::vector<sid_t> transitions = org_dfa.get_transitions(state, input);
+                if (transitions.empty())
+                    continue;
+                int to = find_set(sets, transitions[0]);
+                if (to == -1) {
+                    std::cerr << "Internal Error: target can't be -1" << std::endl;
+                    return dfa;
+                }
+                dfa.add_new_transition(i + 1, to + 1, input);
+                break;
+            }
         }
     }
     return dfa;
@@ -253,6 +272,11 @@ machine dfa::minimize_dfa(machine& dfa) {
     for (sid_t s = 1 ; s <= dfa.get_states_count() ; s++) {
         std::string token_type = dfa.get_token_class(s);
         // std::cout << s << ": " << token_type << std::endl;
+        // TODO: remove this;
+        if (!dfa.is_accepting(s)) {
+            dfa.set_token_class(s, "notacc");
+            token_type = "notacc";
+        }
         if (tokenIdx.find(token_type) == tokenIdx.end()) {
             tokenIdx[token_type] = count++;
             std::vector<sid_t> temp;
@@ -273,6 +297,6 @@ machine dfa::minimize_dfa(machine& dfa) {
         // std::cout << "267: \n";
         // print_partitions(cur_set);
     }
-    print_partitions(cur_set);
+    // print_partitions(cur_set);
     return build_dfa(dfa, cur_set, starting_state);
 }

machine.cpp

@@ -43,6 +43,10 @@ void machine::state::set_key(std::string _key) {
     key = _key;
 }
 
+void machine::state::set_token_class(std::string new_token_class) {
+    token_class = new_token_class;
+}
+
 std::vector<int> machine::state::get_transitions_for(char input) {
     std::vector<int> t;
     for (machine::state::transition &trans : transitions[input]) {
@@ -150,6 +154,13 @@ void machine::set_key_for(sid_t id, std::string new_key) {
     states[id - 1].set_key(new_key);
 }
 
+bool machine::set_token_class(sid_t id, std::string new_token_class) {
+    if (id < 1 && id > states.size())
+        return false;
+    states[id - 1].set_token_class(new_token_class);    
+    return true;
+}
+
 std::set<char> machine::get_language() const {
     return language;
 }

machine.h

@@ -47,6 +47,8 @@ class machine {
 
         void set_key(std::string key);
 
+        void set_token_class(std::string token_class);
+
         std::string get_key() const;
 
         std::vector<sid_t> get_transitions_for(char input);
@@ -94,11 +96,14 @@ class machine {
 
     void set_key_for(sid_t id, std::string new_key);
 
+
     std::set<char> get_language() const;
 
     sid_t get_states_count() const;
 
     bool set_accepting(sid_t id);
+
+    bool set_token_class(sid_t id, std::string new_token_class);
 
     bool is_accepting(sid_t id);
 

main.cpp

@@ -16,16 +16,16 @@ int main ()
 
   rexparser rx;
 
-  // machine nfa = rx.rules2nfa ("letter = a-z | A-Z\ndigit = 0-9\nid: letter (letter|digit)*");
-  machine nfa = rx.rules2nfa("letter: a+");
+  machine nfa = rx.rules2nfa ("letter = a-z | A-Z\ndigit = 0-9\nid: letter (letter|digit)*");
+  // machine nfa = rx.rules2nfa("letter: a*");
   // for (int s = 1 ; s <= nfa.get_states_count() ; s++) {
   //   cout << s << ": " << nfa.get_token_class(s) << std::endl;
   // }
   // nfa.print_machine();
   machine dfa = dfa::to_dfa(nfa);
   // dfa.print_machine();
   machine min_dfa = dfa::minimize_dfa(dfa);
-  // min_dfa.print_machine();
+  min_dfa.print_machine();
   //rx.rules2nfa ("id: a b*").print_machine ();
 
   return 0;

make.sh

@@ -1,4 +1,4 @@
-g++ -ggdb -std=c++11 *.cpp
+g++ -ggdb -std=c++11 *.h *.cpp
 ./a.out > example.dot
 dot -Tpng example.dot -o example.png
 xdg-open example.png