BlitzLLVM/code_compiler/source/lexer.hpp

/// AUTOGENERATED COPYRIGHT HEADER START
// Copyright (C) 2017-2025 Michael Fabian 'Xaymar' Dirks <info@xaymar.com>
// AUTOGENERATED COPYRIGHT HEADER END
#pragma once
#include <cinttypes>
#include <filesystem>
#include <fstream>
#include <iostream>
#include <istream>
#include <memory>
#include <string>
#include <utility>

#include "error.hpp"

// ToDo:
// - Figure out a way to let the lexer output line and character information?

namespace blitz {
	struct token {
		std::pair<uint64_t, uint64_t> location;
		std::string                   text;
		enum class variant : uint64_t {
			NONE, // There is no token here.
			UNKNOWN, // We have absolutely no fucking clue.
			ENDOFFILE, // End of the file.
			NEWLINE, // New Line.
			//SEPARATOR, // Command Separator.
			CONTROL, // All kinds of control signals
			SYMBOL, // All kinds of symbols.
			COMMENT, // ; Whatever
			TEXT, // HelloWorld
			STRING, // "HelloWorld"
			INTEGER, // 1, 1% (without the %)
			REAL, // 1.0, 1# (without the #)
		} type;

		std::string to_string();

		bool operator==(blitz::token::variant rhs);
		bool operator==(std::string const& rhs);
	};

	class lexer {
		std::filesystem::path _file;
		std::ifstream         _stream;

		// Current location in the file.
		std::pair<uint64_t, uint64_t> _location;

		blitz::token _current;
		blitz::token _next;

		public:
		~lexer();
		lexer(std::filesystem::path file);

		/** Retrieve the current token information.
		 */
		blitz::token current();

		/** Retrieve the next token in the given stream.
		 *
		 * This will replace the current token.
		 */
		blitz::token next();

		/** Peek at the next token in the given stream.
		 *
		 * The current token will remain in-tact.
		 */
		blitz::token peek();

		public:
		std::filesystem::path file();
	};
} // namespace blitz