lexer.h

// This file if part of the llir-opt project.
// Licensing information can be found in the LICENSE file.
// (C) 2018 Nandor Licker. All rights reserved.
 
#pragma once
 
#include <llvm/ADT/StringRef.h>
 
#include <string>
#include <fstream>
 
#include "core/adt/sexp.h"
#include "core/constant.h"
 
class Prog;
class Func;
class Block;
 
 
 
class Lexer final {
public:
  enum class Token {
    // '\n'
    NEWLINE,
    // End of stream
    END,
    // '['
    LBRACKET,
    // ']'
    RBRACKET,
    // '(',
    LPAREN,
    // ')',
    RPAREN,
    // ','
    COMMA,
    // '$[a-z]+'
    REG,
    // '$[0-9]+'
    VREG,
    // [a-zA-Z_.][a-zA-Z_0-9.]*
    IDENT,
    // [IDENT]:
    COLON,
    // [0-9]+
    NUMBER,
    // @[a-zA-Z0-9_]+
    ANNOT,
    // Quoted string
    STRING,
    // Plus sign.
    PLUS,
    // Minus sign.
    MINUS,
  };
 
public:
  Lexer(llvm::StringRef buf);
 
  ~Lexer();
 
  Token GetToken() const { return tk_; }
  Token NextToken();
  void Expect(Token type);
  void Check(Token type);
  bool AtEnd() const { return tk_ == Token::END; }
 
  std::string_view String() const;
  int64_t Int() const;
  Register Reg() const;
  uint64_t VReg() const;
 
  SExp ParseSExp();
 
  [[noreturn]] void Error(const std::string &msg);
  [[noreturn]] void Error(Func *f, const std::string &msg);
  [[noreturn]] void Error(Func *f, Block *b, const std::string &msg);
 
private:
  char NextChar();
 
private:
  llvm::StringRef buf_;
  const char *ptr_;
  char char_;
  Token tk_;
  unsigned row_;
  unsigned col_;
  std::string str_;
  Register reg_;
  uint64_t vreg_;
  int64_t int_;
};