ivy/lang/ast/for.c

#include "block.h"
#include "expr/expr.h"
#include "iterate.h"

struct for_parser_state {
	struct parser_state s_base;
	bool s_inline;
	unsigned int s_prev_token;

	struct ivy_ast_node *s_body, *s_iterator, *s_iterable;

	struct ivy_ast_node *s_prev_node;
};

static void init_state(struct ivy_parser *ctx, struct parser_state *sp, uintptr_t arg)
{
	struct for_parser_state *state = (struct for_parser_state *)sp;
	state->s_prev_node = (struct ivy_ast_node *)arg;
}

struct token_parse_result parse_for(struct ivy_parser *ctx, struct ivy_token *tok)
{
	struct for_parser_state *state
		= parser_get_state(ctx, struct for_parser_state);

	if (state->s_prev_node) {
		/* this is an inline for-loop */
		state->s_inline = true;
		state->s_body = state->s_prev_node;
		state->s_prev_node = NULL;
	} else {
		state->s_inline = false;
	}

	struct expr_parser_state *expr
		= (struct expr_parser_state *)parser_push_state(
			ctx, IVY_AST_EXPR, 0);

	state->s_prev_token = IVY_KW_FOR;
	expr_add_terminator(expr, IVY_KW_IN);
	expr->s_subexpr_depth = 1;

	return PARSE_RESULT(IVY_OK, 0);
}

static struct token_parse_result parse_in(
	struct ivy_parser *ctx, struct ivy_token *tok)
{
	struct for_parser_state *state
		= parser_get_state(ctx, struct for_parser_state);

	if (state->s_prev_token != IVY_KW_FOR) {
		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
	}

	/* previous component was the for iterator */
	if (!state->s_prev_node) {
		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
	}

	state->s_iterator = state->s_prev_node;
	state->s_prev_node = NULL;

	/* next component will be the for iterable. */
	struct expr_parser_state *expr
		= (struct expr_parser_state *)parser_push_state(
			ctx, IVY_AST_EXPR, 0);

	/* set the sub-expression depth to be non-zero so the expression parser doesn't consume the expression separator. */
	expr->s_subexpr_depth = 1;
	expr_add_terminator(expr, IVY_KW_DO);
	expr_add_terminator(expr, IVY_SYM_RIGHT_PAREN);

	state->s_prev_token = IVY_KW_IN;
	return PARSE_RESULT(IVY_OK, 0);
}

static struct token_parse_result parse_do(
	struct ivy_parser *ctx, struct ivy_token *tok)
{
	struct for_parser_state *state
		= parser_get_state(ctx, struct for_parser_state);

	if (state->s_inline) {
		struct ivy_diag *diag = parser_push_diag(
			ctx, IVY_LANG_E_INCORRECT_INLINE_FOR_LOOP,
			IVY_LANG_MSG_DO_UNEXPECTED_IN_INLINE_FOR, tok);
		ivy_diag_push_msg(
			diag, IVY_LANG_MSG_PREVIOUS_EXPRESSION_IS_FOR_BODY);
		const struct ivy_char_cell *x = &state->s_body->n_end;
		const struct ivy_char_cell *z = &tok->t_end;

		const struct ivy_diag_amendment a[] = {
			IVY_DIAG_ADD(x->c_row, x->c_col + 1, "."),
		};
		const size_t nr_a = sizeof a / sizeof a[0];

		const struct ivy_diag_highlight hl[] = {
			IVY_DIAG_HL(
				HINT, x->c_row, x->c_col + 1, x->c_row,
				x->c_col + 1),
		};
		const size_t nr_hl = sizeof hl / sizeof hl[0];

		ivy_diag_push_snippet(diag, x->c_row, z->c_row, a, nr_a, hl, nr_hl);
		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
	}

	if (state->s_prev_token != IVY_KW_IN) {
		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
	}

	/* previous component was the for iterable. */
	if (!state->s_prev_node) {
		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
	}

	state->s_iterable = state->s_prev_node;
	state->s_prev_node = NULL;

	/* next component will be a block. */
	struct block_parser_state *block
		= (struct block_parser_state *)parser_push_state(
			ctx, IVY_AST_BLOCK, 0);

	block_add_terminator(block, IVY_KW_END);
	state->s_prev_token = IVY_KW_DO;
	return PARSE_RESULT(IVY_OK, 0);
}

static struct token_parse_result parse_expr_begin(
	struct ivy_parser *ctx, struct ivy_token *tok)
{
	struct for_parser_state *state
		= parser_get_state(ctx, struct for_parser_state);

	if (state->s_prev_token != IVY_KW_DO) {
		/* expression can only follow do keyword. */
		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
	}

	struct block_parser_state *block
		= (struct block_parser_state *)parser_push_state(
			ctx, IVY_AST_BLOCK, 0);
	block_add_terminator(block, IVY_KW_END);
	return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}

static enum ivy_status finalise_for_loop(struct for_parser_state *state)
{
	struct ivy_ast_for_loop_node *loop
		= (struct ivy_ast_for_loop_node *)state->s_base.s_node;

	if (state->s_inline) {
		/* we have just reached the end of the for iterable. */
		if (!state->s_prev_node) {
			/* for condition is empty. */
			return IVY_ERR_BAD_SYNTAX;
		}

		state->s_iterable = state->s_prev_node;
		state->s_prev_node = NULL;

		loop->n_iterator = state->s_iterator;
		loop->n_iterable = state->s_iterable;
		loop->n_body = state->s_body;

		state->s_iterator = NULL;
		state->s_iterable = NULL;
		state->s_body = NULL;

		return IVY_OK;
	}

	if (!state->s_prev_node) {
		/* for body is empty. */
		return IVY_ERR_BAD_SYNTAX;
	}

	state->s_body = state->s_prev_node;
	state->s_prev_node = NULL;

	loop->n_iterator = state->s_iterator;
	loop->n_iterable = state->s_iterable;
	loop->n_body = state->s_body;

	state->s_iterator = NULL;
	state->s_iterable = NULL;
	state->s_body = NULL;

	return IVY_OK;
}

static struct token_parse_result parse_punct_terminator(
	struct ivy_parser *ctx, struct ivy_token *tok)
{
	struct for_parser_state *state
		= parser_get_state(ctx, struct for_parser_state);

	if (!state->s_inline) {
		/* only inline for loop can be ended with punctuation. */
		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
	}

	enum ivy_status status = finalise_for_loop(state);
	if (status != IVY_OK) {
		return PARSE_RESULT(status, 0);
	}

	parser_pop_state(ctx, STATE_ADD_NODE_TO_PARENT);
	return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
}

static struct token_parse_result parse_end(
	struct ivy_parser *ctx, struct ivy_token *tok)
{
	struct for_parser_state *state
		= parser_get_state(ctx, struct for_parser_state);

	if (state->s_inline) {
		/* inline for loop must be terminated with punctuation. */
		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
	}

	enum ivy_status status = finalise_for_loop(state);
	if (status != IVY_OK) {
		return PARSE_RESULT(status, 0);
	}

	parser_pop_state(ctx, STATE_ADD_NODE_TO_PARENT);
	return PARSE_RESULT(IVY_OK, 0);
}

static enum ivy_status add_child(
	struct parser_state *parent, struct ivy_ast_node *child)
{
	struct for_parser_state *state = (struct for_parser_state *)parent;

	if (state->s_prev_node) {
		return IVY_ERR_BAD_SYNTAX;
	}

	state->s_prev_node = child;
	return IVY_OK;
}

static void collect_children(
	struct ivy_ast_node *node, struct ivy_ast_node_iterator *iterator)
{
	struct ivy_ast_for_loop_node *loop = (struct ivy_ast_for_loop_node *)node;

	if (loop->n_iterator) {
		ast_node_iterator_enqueue_node(iterator, node, loop->n_iterator);
	}

	if (loop->n_iterable) {
		ast_node_iterator_enqueue_node(iterator, node, loop->n_iterable);
	}

	if (loop->n_body) {
		ast_node_iterator_enqueue_node(iterator, node, loop->n_body);
	}
}

struct ast_node_type for_loop_node_ops = {
	.n_init_state = init_state,
	.n_add_child = add_child,
	.n_collect_children = collect_children,
	.n_state_size = sizeof(struct for_parser_state),
	.n_node_size = sizeof(struct ivy_ast_for_loop_node),
	.n_keyword_parsers = {
		KW_PARSER(FOR, parse_for),
		KW_PARSER(IN, parse_in),
		KW_PARSER(DO, parse_do),
		KW_PARSER(END, parse_end),
	},
	.n_symbol_parsers = {
		SYM_PARSER(DOT, parse_punct_terminator),
		SYM_PARSER(COMMA, parse_punct_terminator),
		SYM_PARSER(RIGHT_PAREN, parse_punct_terminator),
		SYM_PARSER(BANG, parse_punct_terminator),
	},
	.n_expr_parser = {
		.expr_begin = parse_expr_begin,
	},
};