2023-11-17 00:30:48 -05:00
|
|
|
#pragma once
|
|
|
|
/*
|
|
|
|
* Created by Brett on 17/11/23.
|
|
|
|
* Licensed under GNU General Public License V3.0
|
|
|
|
* See LICENSE file for license detail
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef BRAINFUCK_MISC_BF_TOKENIZER_H
|
|
|
|
#define BRAINFUCK_MISC_BF_TOKENIZER_H
|
|
|
|
|
|
|
|
#include <vector>
|
|
|
|
#include <string>
|
2023-11-17 02:06:09 -05:00
|
|
|
#include <optional>
|
2023-11-17 00:30:48 -05:00
|
|
|
|
|
|
|
enum class bf_token
|
|
|
|
{
|
|
|
|
INC_DP,
|
|
|
|
DEC_DP,
|
|
|
|
INC_DV,
|
|
|
|
DEC_DV,
|
|
|
|
PRINT,
|
|
|
|
READ,
|
|
|
|
OPEN,
|
|
|
|
CLOSE
|
|
|
|
};
|
|
|
|
|
2023-11-24 14:41:45 -05:00
|
|
|
struct bf_token_t
|
2023-11-17 02:06:09 -05:00
|
|
|
{
|
2023-11-20 19:24:45 -05:00
|
|
|
bf_token type;
|
|
|
|
int64_t offset;
|
2023-11-20 17:26:37 -05:00
|
|
|
std::optional<uint64_t> name = {};
|
2023-11-17 02:06:09 -05:00
|
|
|
|
2023-11-24 14:41:45 -05:00
|
|
|
explicit bf_token_t(bf_token type, int64_t offset = 1): type(type), offset(offset)
|
2023-11-20 17:26:37 -05:00
|
|
|
{}
|
2023-11-17 02:06:09 -05:00
|
|
|
};
|
|
|
|
|
2023-11-22 17:13:31 -05:00
|
|
|
class bf_tokenizer
|
2023-11-20 17:26:37 -05:00
|
|
|
{
|
|
|
|
private:
|
2023-11-24 14:41:45 -05:00
|
|
|
std::vector<bf_token_t> tokens;
|
2023-11-20 17:26:37 -05:00
|
|
|
size_t conditionalCount = 0;
|
|
|
|
size_t currentIndex = 0;
|
|
|
|
|
|
|
|
void tokenize(const std::string& program);
|
|
|
|
|
|
|
|
void bf_name();
|
|
|
|
|
|
|
|
public:
|
2023-11-22 17:13:31 -05:00
|
|
|
explicit bf_tokenizer(const std::string& program)
|
2023-11-20 17:26:37 -05:00
|
|
|
{
|
|
|
|
tokenize(program);
|
|
|
|
bf_name();
|
2023-11-24 14:41:45 -05:00
|
|
|
if (hasNext())
|
|
|
|
{
|
2023-11-22 17:26:19 -05:00
|
|
|
// skip past comments
|
|
|
|
if (peek().type == bf_token::OPEN)
|
|
|
|
while (hasNext() && peek().type != bf_token::CLOSE)
|
|
|
|
advance();
|
|
|
|
}
|
2023-11-20 17:26:37 -05:00
|
|
|
}
|
|
|
|
|
2023-11-24 14:41:45 -05:00
|
|
|
inline bool hasNext()
|
|
|
|
{
|
2023-11-20 17:26:37 -05:00
|
|
|
return currentIndex < tokens.size();
|
|
|
|
}
|
|
|
|
|
2023-11-24 14:41:45 -05:00
|
|
|
inline size_t advance()
|
|
|
|
{
|
2023-11-20 19:24:45 -05:00
|
|
|
return currentIndex++;
|
2023-11-20 17:26:37 -05:00
|
|
|
}
|
|
|
|
|
2023-11-24 14:41:45 -05:00
|
|
|
inline const bf_token_t& next()
|
|
|
|
{
|
2023-11-20 19:24:45 -05:00
|
|
|
return tokens[advance()];
|
2023-11-20 17:26:37 -05:00
|
|
|
}
|
|
|
|
|
2023-11-24 14:41:45 -05:00
|
|
|
inline const bf_token_t& peek()
|
|
|
|
{
|
2023-11-20 17:26:37 -05:00
|
|
|
return tokens[currentIndex];
|
|
|
|
}
|
|
|
|
|
2023-11-24 14:41:45 -05:00
|
|
|
inline std::vector<bf_token_t>& data()
|
|
|
|
{
|
2023-11-22 17:26:19 -05:00
|
|
|
return tokens;
|
|
|
|
}
|
|
|
|
|
2023-11-20 19:24:45 -05:00
|
|
|
void print(size_t index);
|
2023-11-24 14:41:45 -05:00
|
|
|
|
|
|
|
inline void print()
|
|
|
|
{
|
2023-11-20 19:24:45 -05:00
|
|
|
print(currentIndex);
|
|
|
|
}
|
2023-11-20 17:26:37 -05:00
|
|
|
};
|
2023-11-17 00:30:48 -05:00
|
|
|
|
2023-11-24 14:41:45 -05:00
|
|
|
enum ast_token
|
|
|
|
{
|
|
|
|
INC_DP, DEC_DP, INC_VAL, DEC_VAL, PRINT, READ, WHILE
|
|
|
|
};
|
|
|
|
|
|
|
|
struct ast_token_t
|
|
|
|
{
|
|
|
|
ast_token type;
|
|
|
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
2023-11-17 00:30:48 -05:00
|
|
|
#endif //BRAINFUCK_MISC_BF_TOKENIZER_H
|