307 lines
11 KiB
C
307 lines
11 KiB
C
typedef enum {
|
|
|
|
#define AST_FLAG_XLIST\
|
|
X(null)\
|
|
X(string)\
|
|
X(integer)\
|
|
X(real)\
|
|
X(binary)\
|
|
X(enum)\
|
|
X(enum_member)\
|
|
X(struct)\
|
|
X(struct_member)\
|
|
X(var)\
|
|
X(type_name)\
|
|
X(type_pointer)\
|
|
X(type_array)\
|
|
|
|
#define X(NAME) ast_flag_##NAME,
|
|
AST_FLAG_XLIST
|
|
#undef X
|
|
} ast_flag_t;
|
|
|
|
typedef struct ast_t ast_t;
|
|
struct ast_t {
|
|
ast_flag_t flags;
|
|
|
|
lex_t *pos;
|
|
ast_t *next;
|
|
ast_t *first;
|
|
ast_t *last;
|
|
i32 len;
|
|
|
|
s8_t string;
|
|
f64 real;
|
|
i64 integer;
|
|
};
|
|
|
|
s8_t s8_serial_ast_flag_t(ma_arena_t *arena, ast_flag_t flag) {
|
|
sb8_t *sb = sb8_serial_begin(arena);
|
|
#define X(NAME) if (flag & set_bit(ast_flag_##NAME)) sb8_printf(sb, #NAME);
|
|
AST_FLAG_XLIST
|
|
#undef X
|
|
s8_t result = sb8_serial_end(sb);
|
|
return result;
|
|
}
|
|
|
|
ast_t *create_ast(parser_t *par, lex_t *pos, ast_flag_t flags) {
|
|
ast_t *result = ma_push_type(par->arena, ast_t);
|
|
memset(result, 0, sizeof(ast_t));
|
|
result->flags = flags;
|
|
result->pos = pos;
|
|
return result;
|
|
}
|
|
|
|
void ast_append(ast_t *parent, ast_t *node) {
|
|
SLLQ_APPEND(parent->first, parent->last, node);
|
|
parent->len += 1;
|
|
}
|
|
|
|
ast_t *create_ast_binary(parser_t *par, lex_t *pos, ast_t *left, lex_kind_t op, ast_t *right) {
|
|
ast_t *result = create_ast(par, pos, set_bit(ast_flag_string) | set_bit(ast_flag_binary) | set_bit(ast_flag_integer));
|
|
ast_append(result, left);
|
|
ast_append(result, right);
|
|
result->integer = op;
|
|
result->string = lex_kind_to_simple_s8(op);
|
|
return result;
|
|
}
|
|
|
|
ast_t *parse_expr(parser_t *par);
|
|
ast_t *parse_lit_expr(parser_t *par) {
|
|
lex_t *token = parser_next(par);
|
|
if (token->kind == lex_kind_integer) {
|
|
ast_t *result = create_ast(par, token, set_bit(ast_flag_integer) | set_bit(ast_flag_string));
|
|
result->integer = token->integer;
|
|
result->string = token->string;
|
|
return result;
|
|
} else if (token->kind == lex_kind_real) {
|
|
ast_t *result = create_ast(par, token, set_bit(ast_flag_real) | set_bit(ast_flag_string));
|
|
result->real = (double)token->real;
|
|
result->string = token->string;
|
|
return result;
|
|
} else if (token->kind == lex_kind_open_paren) {
|
|
ast_t *result = parse_expr(par);
|
|
parser_expect(par, lex_kind_close_paren);
|
|
return result;
|
|
} else {
|
|
lex_panicf(token, "got invalid token of kind: %S while parsing expression", lex_kind_to_s8(token->kind));
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
ast_t *parse_mul_expr(parser_t *par) {
|
|
ast_t *left = parse_lit_expr(par);
|
|
while (par->at->kind == lex_kind_multiply || par->at->kind == lex_kind_divide || par->at->kind == lex_kind_modulo) {
|
|
lex_t *op = parser_next(par);
|
|
left = create_ast_binary(par, op, left, op->kind, parse_lit_expr(par));
|
|
}
|
|
return left;
|
|
}
|
|
|
|
ast_t *parse_add_expr(parser_t *par) {
|
|
ast_t *left = parse_mul_expr(par);
|
|
while (par->at->kind == lex_kind_plus || par->at->kind == lex_kind_minus) {
|
|
lex_t *op = parser_next(par);
|
|
left = create_ast_binary(par, op, left, op->kind, parse_lit_expr(par));
|
|
}
|
|
return left;
|
|
}
|
|
|
|
ast_t *parse_logical_and_expr(parser_t *par) {
|
|
ast_t *left = parse_add_expr(par);
|
|
while (par->at->kind == lex_kind_or) {
|
|
lex_t *op = parser_next(par);
|
|
left = create_ast_binary(par, op, left, op->kind, parse_lit_expr(par));
|
|
}
|
|
return left;
|
|
}
|
|
|
|
ast_t *parse_logical_or_expr(parser_t *par) {
|
|
ast_t *left = parse_logical_and_expr(par);
|
|
while (par->at->kind == lex_kind_or) {
|
|
lex_t *op = parser_next(par);
|
|
left = create_ast_binary(par, op, left, op->kind, parse_lit_expr(par));
|
|
}
|
|
return left;
|
|
}
|
|
|
|
ast_t *parse_expr(parser_t *par) {
|
|
ast_t *expr = parse_logical_or_expr(par);
|
|
return expr;
|
|
}
|
|
|
|
ast_t *parse_expr_str(ma_arena_t *arena, char *file_name, char *stream) {
|
|
lex_array_t tokens = lex_tokens(arena, file_name, stream);
|
|
parser_t *par = parser_make(arena, tokens.data);
|
|
ast_t *result = parse_expr(par);
|
|
return result;
|
|
}
|
|
|
|
i64 eval_const_expr(ast_t *expr) {
|
|
if (expr->flags & set_bit(ast_flag_integer)) {
|
|
return expr->integer;
|
|
} else if (expr->flags & set_bit(ast_flag_binary)) {
|
|
assert(expr->first != expr->last);
|
|
i64 left = eval_const_expr(expr->first);
|
|
i64 right = eval_const_expr(expr->last);
|
|
|
|
switch(expr->integer) {
|
|
case lex_kind_plus: return left + right;
|
|
case lex_kind_minus: return left - right;
|
|
case lex_kind_multiply: return left * right;
|
|
case lex_kind_divide: return left / right;
|
|
case lex_kind_modulo: return left % right;
|
|
case lex_kind_and: return left && right;
|
|
case lex_kind_or: return left || right;
|
|
default: lex_panicf(expr->pos, "unhandled binary operator: %S", lex_kind_to_s8(expr->integer));
|
|
}
|
|
} else {
|
|
ma_temp_t scratch = ma_begin_scratch();
|
|
lex_panicf(expr->pos, "unhandled ast in const expression evaluation: %S", s8_serial_ast_flag_t(scratch.arena, expr->flags));
|
|
ma_end_scratch(scratch);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
#define test_expr(x) do {\
|
|
lex_array_t tokens = lex_tokens(scratch.arena, "parser_test", #x);\
|
|
parser_t *par = parser_make(scratch.arena, tokens.data);\
|
|
ast_t *expr = parse_expr(par);\
|
|
assert(expr != NULL);\
|
|
i64 value = eval_const_expr(expr);\
|
|
assert(value == x);\
|
|
} while (0)
|
|
|
|
void run_parser_test() {
|
|
ma_temp_t scratch = ma_begin_scratch();
|
|
test_expr(32 + 2 + 5 + 5);
|
|
test_expr(32 - 2 + 5 - 5);
|
|
test_expr(2 * 2 / 4 * 5 + 2 + 3);
|
|
test_expr(2 * 5 * 5 / 2 + 2 - 1 - 1);
|
|
test_expr(2 * (5 * 5) / 2 + (2 - 1 - 1));
|
|
test_expr((2 * (5 * 5) / (2)) + (2 - 1 - 1));
|
|
test_expr(10 % 3);
|
|
test_expr(10 % 3 + 4 || 2);
|
|
test_expr(10 % 3 + 4 || 2 && (4 && 2) || 3 && 1 || 0);
|
|
ma_end_scratch(scratch);
|
|
}
|
|
|
|
ast_t *parse_struct_mem(parser_t *par, s8_t *name) {
|
|
lex_t *type_name = parser_expect(par, lex_kind_ident);
|
|
ast_t *type = create_ast(par, type_name, set_bit(ast_flag_type_name) | set_bit(ast_flag_string));
|
|
type->string = type_name->string;
|
|
|
|
while (parser_match(par, lex_kind_multiply)) {
|
|
ast_t *pointer = create_ast(par, par->at, set_bit(ast_flag_type_pointer) | set_bit(ast_flag_string));
|
|
ast_append(pointer, type);
|
|
pointer->string = s8_fmt(par->arena, "%S*", type->string);
|
|
type = pointer;
|
|
}
|
|
|
|
*name = parser_expect(par, lex_kind_ident)->string;
|
|
|
|
while (parser_match(par, lex_kind_open_bracket)) {
|
|
ast_t *array = create_ast(par, par->at, set_bit(ast_flag_type_array) | set_bit(ast_flag_string));
|
|
ast_append(array, type);
|
|
lex_t *num = parser_match(par, lex_kind_integer);
|
|
if (num) {
|
|
array->flags |= set_bit(ast_flag_integer);
|
|
array->integer = (int)num->integer;
|
|
array->string = s8_fmt(par->arena, "%S[%d]", type->string, (int)array->integer);
|
|
} else {
|
|
array->string = s8_fmt(par->arena, "%S[]", type->string);
|
|
}
|
|
parser_expect(par, lex_kind_close_bracket);
|
|
type = array;
|
|
}
|
|
|
|
return type;
|
|
}
|
|
|
|
ast_t *parse_decls(ma_arena_t *arena, char *file, char *code) {
|
|
lex_array_t tokens = lex_tokens(arena, file, code);
|
|
parser_t *par = parser_make(arena, tokens.data);
|
|
ast_t *result = create_ast(par, par->at, set_bit(ast_flag_string));
|
|
result->string = s8_copy_char(arena, file);
|
|
for (;par->at->kind != lex_kind_eof;) {
|
|
lex_t *pos = par->at;
|
|
if (parser_matchi(par, s8_lit("enum"))) {
|
|
ast_t *n = create_ast(par, pos, set_bit(ast_flag_string) | set_bit(ast_flag_enum));
|
|
ast_append(result, n);
|
|
|
|
parser_expect(par, lex_kind_open_brace);
|
|
while (par->at->kind == lex_kind_ident) {
|
|
lex_t *val = parser_expect(par, lex_kind_ident);
|
|
ast_t *mem = create_ast(par, val, set_bit(ast_flag_enum_member) | set_bit(ast_flag_string));
|
|
mem->string = val->string;
|
|
ast_append(n, mem);
|
|
|
|
// if (parser_match(par, lex_kind_assign)) {
|
|
// parse_expr();
|
|
// }
|
|
|
|
if (!parser_match(par, lex_kind_comma)) break;
|
|
}
|
|
parser_expect(par, lex_kind_close_brace);
|
|
n->string = parser_expect(par, lex_kind_ident)->string;
|
|
parser_expect(par, lex_kind_semicolon);
|
|
} else if (parser_matchi(par, s8_lit("struct"))) {
|
|
ast_t *n = create_ast(par, pos, set_bit(ast_flag_string) | set_bit(ast_flag_struct));
|
|
ast_append(result, n);
|
|
n->string = parser_expect(par, lex_kind_ident)->string;
|
|
|
|
parser_expect(par, lex_kind_open_brace);
|
|
while (par->at->kind != lex_kind_close_brace) {
|
|
ast_t *mem = create_ast(par, par->at, set_bit(ast_flag_struct_member) | set_bit(ast_flag_var) | set_bit(ast_flag_string));
|
|
ast_append(n, mem);
|
|
|
|
ast_t *type = parse_struct_mem(par, &mem->string);
|
|
ast_append(mem, type);
|
|
|
|
parser_expect(par, lex_kind_semicolon);
|
|
}
|
|
parser_expect(par, lex_kind_close_brace);
|
|
parser_expect(par, lex_kind_semicolon);
|
|
|
|
} else {
|
|
parser_next(par);
|
|
}
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
ast_t *parse_table(ma_arena_t *arena, char *file, char *code) {
|
|
lex_array_t tokens = lex_tokens(arena, file, code);
|
|
parser_t *par = parser_make(arena, tokens.data);
|
|
ast_t *table = create_ast(par, par->at, 0);
|
|
while (par->at->kind != lex_kind_eof) {
|
|
ast_t *row = create_ast(par, par->at, 0);
|
|
ast_append(table, row);
|
|
while (par->at->kind != lex_kind_eof) {
|
|
parser_match(par, lex_kind_bit_or);
|
|
|
|
lex_t *token = par->at;
|
|
if (parser_match(par, lex_kind_ident) || parser_match(par, lex_kind_string)) {
|
|
ast_t *col = create_ast(par, par->at, set_bit(ast_flag_string));
|
|
ast_append(row, col);
|
|
col->string = token->string;
|
|
} else if (parser_match(par, lex_kind_integer)) {
|
|
ast_t *col = create_ast(par, par->at, set_bit(ast_flag_string) | set_bit(ast_flag_integer));
|
|
ast_append(row, col);
|
|
col->string = token->string;
|
|
col->integer = token->integer;
|
|
} else if (parser_match(par, lex_kind_real)) {
|
|
ast_t *col = create_ast(par, par->at, set_bit(ast_flag_string) | set_bit(ast_flag_real));
|
|
ast_append(row, col);
|
|
col->string = token->string;
|
|
col->real = token->real;
|
|
} else if (parser_match(par, lex_kind_bit_or) || parser_match(par, lex_kind_eof)) {
|
|
break;
|
|
} else {
|
|
lex_panicf(par->at, "invalid token: %S", lex_kind_to_s8(par->at->kind));
|
|
}
|
|
}
|
|
}
|
|
return table;
|
|
} |