mips/masm/parse.h

/* Copyright (c) 2024 Freya Murphy */

#ifndef __PARSE_H__
#define __PARSE_H__

#include "lex.h"

#include <mlimits.h>
#include <stdint.h>

/// the type to a direcive
enum expr_directive_type {
	EXPR_DIRECTIVE_ALIGN,
	EXPR_DIRECTIVE_SPACE,
	EXPR_DIRECTIVE_WORD,
	EXPR_DIRECTIVE_HALF,
	EXPR_DIRECTIVE_BYTE,
	EXPR_DIRECTIVE_SECTION,
	EXPR_DIRECTIVE_EXTERN,
	EXPR_DIRECTIVE_GLOBL,
	EXPR_DIRECTIVE_ASCII,
	EXPR_DIRECTIVE_ASCIIZ,
};

/// holds a directive
struct expr_directive {
	// the type of the directive
	enum expr_directive_type type;

	// lengh of .word, .half, or .byte directive
	uint32_t len;

	// directive data
	union {
		// e.g. align 2
		uint16_t align;
		// e.g. space 4096
		uint16_t space;
		// e.g. .word 0x1 0x2
		uint32_t words[MAX_ARG_LENGTH];
		uint16_t halfs[MAX_ARG_LENGTH];
		uint8_t  bytes[MAX_ARG_LENGTH];
		// e.g. .ascii "hello world!"
		struct string string;
		// e.g. .globl main
		struct string label;
		// e.g. .text
		struct string section;
	};
};

/// holds a constant expression
struct expr_const {
	// the name of the constant
	struct string name;

	// the value of the constant
	uint32_t num;
};

/// the type to a right
/// hand side argument to an
/// instruction
enum expr_ins_arg_type {
	// e.g. $ra
	EXPR_INS_ARG_REGISTER,

	// e.g. 0x80
	EXPR_INS_ARG_IMMEDIATE,

	// e.g. main
	EXPR_INS_ARG_LABEL,

	// e.g. 4($sp)
	EXPR_INS_ARG_OFFSET,
};

/// a right hand argument
/// to an instruction
struct expr_ins_arg {
	enum expr_ins_arg_type type;

	union {
		// register
		struct string reg;

		// immediate
		uint64_t immd;

		// label
		struct string label;

		// offset
		struct expr_ins_offset {
			// immediate
			uint64_t immd;
			// register
			struct string reg;
		} offset;
	};
};

/// holds a instruction
struct expr_ins {
	/// pesudo instructions can return
	/// more than one instruction
	struct string name;

	// the arguments of the instruction
	uint32_t args_len;
	struct expr_ins_arg args[MAX_ARG_LENGTH];
};

enum expr_type {
	// e.g. .align 2
	EXPR_DIRECTIVE,
	// e.g. SIZE = 8
	EXPR_CONSTANT,
	// e.g. li $t0, 17
	EXPR_INS,
	// e.g. _start:
	EXPR_LABEL,
};

struct expr {
	enum expr_type type;

	uint32_t line_no;
	uint32_t byte_start;
	uint32_t byte_end;

	union {
		// directive
		struct expr_directive directive;
		// constant
		struct expr_const constant;
		// instruction
		struct expr_ins instruction;
		// label
		struct string label;
	};
};

void expr_free(struct expr *expr);

struct parser {
	// the lexer
	// *weak* ponter, we do not own this
	struct lexer lexer;
	// the last token peeked
	struct token peek;
};

/* get the next expression in the parser */
int parser_next(struct parser *parser, struct expr *expr);

/* initalize the base parser */
int parser_init(const char *file, struct parser *parser);

/* free the base parser */
void parser_free(struct parser *parser);

#endif /* __PARSE_H__ */