Initial commit

include/lexer.hpp

@@ -0,0 +1,92 @@
+#pragma once
+
+#include <list>
+#include <functional>
+#include <string>
+
+namespace fsh {
+
+class Lexer {
+public:
+    enum TokenType {
+        WHITESPACE,
+        WORD,
+        STRING_LITERAL,
+        OPT,
+        FLAG,
+        LREDIRECTION,
+        RREDIRECTION,
+        PIPE,
+        END_OF_STREAM
+    };
+
+    using Token = std::pair<Lexer::TokenType, std::string>;
+
+    int peek();   
+    int consume();   
+    Token next_token();
+
+    static std::list<Token> process(std::string line);
+
+private:
+    std::string workspace;
+    unsigned int idx;
+
+    Lexer(std::string str) : workspace(str), idx(0) {}
+
+};
+using Token = Lexer::Token;
+
+
+
+class LexerStateMachine {
+    friend class Lexer;
+private:
+    enum State {
+        START,
+        REDIRECT,
+        FLAG_OPT,
+        POTENTIAL_WORD,
+        STRING_LITERAL,
+        STATE_COUNT,
+
+        END
+    };
+
+    using StateHandler = std::function<State(void)>;
+
+    Lexer& parser;
+    std::string token;
+    Lexer::TokenType type;
+    std::vector<StateHandler> state_handlers;
+
+    LexerStateMachine(Lexer& parser) : parser(parser), token(""), type(Lexer::TokenType::WHITESPACE), state_handlers(STATE_COUNT) {
+        state_handlers[State::START]          = std::bind(&LexerStateMachine::handle_start         , this);
+        state_handlers[State::REDIRECT      ] = std::bind(&LexerStateMachine::handle_redirect      , this);
+        state_handlers[State::FLAG_OPT      ] = std::bind(&LexerStateMachine::handle_flag_opt      , this);
+        state_handlers[State::POTENTIAL_WORD] = std::bind(&LexerStateMachine::handle_potential_word, this);
+        state_handlers[State::STRING_LITERAL] = std::bind(&LexerStateMachine::handle_string_literal, this);
+    }
+
+    State set_type(Lexer::TokenType type, State state);
+    
+    Token run();
+
+    State handle_start();
+    State handle_redirect();
+    State handle_flag_opt();  
+    State handle_potential_word();  
+    State handle_string_literal();
+
+    State word_like_handler(Lexer::TokenType type, State state);
+}; 
+
+    inline LexerStateMachine::State LexerStateMachine::set_type(
+        Lexer::TokenType type = Lexer::TokenType::WHITESPACE, 
+        State state = State::END
+    ){
+        this->type = type;
+        return state;
+    }
+
+}