New lexer

1 files changed, 29 insertions, 3 deletions

diff --git a/lexer.h b/lexer.h
index f9363ef..eb41e3d 100644
--- a/lexer.h
+++ b/lexer.h
@@ -3,16 +3,42 @@
 #include "minicc.h"
 #include "bnf.h"
 
+#include <string>
+#include <unordered_map>
+#include <vector>
+
 namespace Lex {
 
 class Lexer
 {
+ // constructor input
+ const BNF& m_bnf;
+ const std::string& m_top;
+
+ // Graph
+ 
+ size_t states{}; // start, ...
+ std::unordered_map<size_t, std::vector<std::pair<size_t, char>>> transitions; //transitions: state -> {state,character}, ...; empty transition is marked by \0
+ std::vector<std::string> m_state_types;
+ size_t m_startState;
+ size_t m_endState;
+
+ // Graph manipulation
+ 
+ size_t newState(std::string state_type = "");
+ void addTransition(size_t state0, size_t state1, char c);
+ std::vector<std::pair<size_t, char>> getSuccessors(size_t state);
+
+ // Build up automaton, recursively
+ void addPath(size_t state0, size_t state1, std::string s, std::string type);
+ void addPathOrTransition(size_t state0, size_t state1, std::string symbol, std::string type);
+ void addRule(const std::vector<std::string>& list, size_t list_index_from, size_t list_index_to, size_t state0, size_t state1, const std::string& rule_symbol, std::string type);
 
- //states; // start, ...
- //transitions; // state, state, character
+ Token getToken(const std::string& s, Location& location);
+ void skipWhitespace(const std::string& s, Location& location);
 
 public:
- Lexer(const BNF& bnf, const std::string& Top);
+ Lexer(const BNF& bnf, const std::string& top);
  std::vector<Token> Lex(const std::string& s);
 
 };