lang: ast: implement parsing of complex messages

2024-12-02 10:48:16 +00:00
parent 1c5b23d968
commit 5ec3e8fcd4
1 changed files with 155 additions and 40 deletions
--- a/lang/ast/expr.c
+++ b/lang/ast/expr.c
@@ -16,6 +16,18 @@ enum expr_type {
 	EXPR_TYPE_ARITH,
 };

+enum expr_subtype {
+	EXPR_SUBTYPE_NONE = 0,
+	/* generic parenthesis-enclosed arithmetic expressions */
+	EXPR_SUBTYPE_PAREN,
+	/* keyword messages */
+	EXPR_SUBTYPE_KEYWORD_MSG,
+	/* expression delimited by labels */
+	EXPR_SUBTYPE_KEYWORD_ARG,
+	/* complex messages */
+	EXPR_SUBTYPE_COMPLEX_MSG,
+};
+
 enum expr_component {
 	EXPR_CMP_NONE = 0,
 	EXPR_CMP_OPERATOR,
@@ -43,15 +55,15 @@ struct expr_parser_state {
 	unsigned int s_terminator;

 	/* all sub-expressions (i.e. those conatained within brackets,
-	 * keyword-messages and keyword-message args, etc) will have this set
-	 * to true */
-	bool s_subexpr;
+	 * keyword-messages and keyword-message args, etc) will have this set/ */
+	enum expr_subtype s_subexpr;

 	b_queue s_output_queue;
 	b_queue s_operator_stack;

 	/* these variables are for keyword-message expressions */
 	struct ivy_ast_node *s_recipient;
+	struct ivy_ast_msg_node *s_msg;
 	b_queue s_labels;
 	b_queue s_args;
 };
@@ -139,7 +151,7 @@ static void push_operand(struct expr_parser_state *state, struct ivy_token *tok)
 	}
 }

-static const struct ivy_operator *get_operator(struct ivy_token *tok)
+static const struct ivy_operator *get_operator_from_token(struct ivy_token *tok)
 {
 	switch (tok->t_type) {
 	case IVY_TOK_SYMBOL:
@@ -151,6 +163,20 @@ static const struct ivy_operator *get_operator(struct ivy_token *tok)
 	}
 }

+static const struct ivy_operator *get_operator_from_node(struct ivy_ast_node *n)
+{
+	switch (n->n_type) {
+	case IVY_AST_OP: {
+		struct ivy_ast_op_node *op = (struct ivy_ast_op_node *)n;
+		return op->n_op;
+	}
+	case IVY_AST_MSG:
+		return ivy_operator_get_by_id(IVY_OP_MSG);
+	default:
+		return NULL;
+	}
+}
+
 static struct ivy_ast_selector_node *unary_selector_from_token(struct ivy_token *tok)
 {
 	struct ivy_ast_selector_node *sel
@@ -159,9 +185,32 @@ static struct ivy_ast_selector_node *unary_selector_from_token(struct ivy_token
 	return sel;
 }

-static void push_operator(struct expr_parser_state *state, struct ivy_token *tok)
+static struct ivy_ast_node *create_operator_node_from_token(struct ivy_token *tok)
 {
-	const struct ivy_operator *op = get_operator(tok);
+	const struct ivy_operator *op = get_operator_from_token(tok);
+	if (!op) {
+		return NULL;
+	}
+
+	if (op->op_id == IVY_OP_MSG) {
+		struct ivy_ast_msg_node *new_msg_node
+			= (struct ivy_ast_msg_node *)ast_node_create(IVY_AST_MSG);
+		new_msg_node->n_sel = unary_selector_from_token(tok);
+		return (struct ivy_ast_node *)new_msg_node;
+	}
+
+	struct ivy_ast_op_node *new_op_node
+		= (struct ivy_ast_op_node *)ast_node_create(IVY_AST_OP);
+	new_op_node->n_op = op;
+	return (struct ivy_ast_node *)new_op_node;
+}
+
+static void push_operator(struct expr_parser_state *state, struct ivy_ast_node *node)
+{
+	const struct ivy_operator *op = get_operator_from_node(node);
+	if (!op) {
+		return;
+	}

 	while (true) {
 		b_queue_entry *top = b_queue_last(&state->s_operator_stack);
@@ -200,19 +249,7 @@ static void push_operator(struct expr_parser_state *state, struct ivy_token *tok
 		b_queue_push_back(&state->s_output_queue, top);
 	}

-	if (op->op_id == IVY_OP_MSG) {
-		struct ivy_ast_msg_node *new_msg_node
-			= (struct ivy_ast_msg_node *)ast_node_create(IVY_AST_MSG);
-		new_msg_node->n_sel = unary_selector_from_token(tok);
-		b_queue_push_back(
-			&state->s_operator_stack, &new_msg_node->n_base.n_entry);
-	} else {
-		struct ivy_ast_op_node *new_op_node
-			= (struct ivy_ast_op_node *)ast_node_create(IVY_AST_OP);
-		new_op_node->n_op = op;
-		b_queue_push_back(
-			&state->s_operator_stack, &new_op_node->n_base.n_entry);
-	}
+	b_queue_push_back(&state->s_operator_stack, &node->n_entry);
 }

 static void print_expr_queues(struct expr_parser_state *state)
@@ -433,6 +470,7 @@ static struct token_parse_result parse_operand(
 	push_operand(state, tok);
 	// print_expr_queues(state);
 	state->s_prev_component = EXPR_CMP_OPERAND;
+	state->s_prev_token = tok->t_type;

 	return PARSE_RESULT(IVY_OK, 0);
 }
@@ -455,15 +493,16 @@ static struct token_parse_result parse_operator(
 		state->s_prev_token = tok->t_keyword;
 		break;
 	default:
+		state->s_prev_token = tok->t_type;
 		break;
 	}

-	const struct ivy_operator *op = get_operator(tok);
+	struct ivy_ast_node *op = create_operator_node_from_token(tok);
 	if (!op) {
 		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
 	}

-	push_operator(state, tok);
+	push_operator(state, op);
 	state->s_prev_component = EXPR_CMP_OPERATOR;

 	print_expr_queues(state);
@@ -502,6 +541,28 @@ static struct token_parse_result parse_left_paren(
 		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
 	}

+	if (state->s_prev_token == IVY_TOK_IDENT) {
+		/* this is the opening parenthesis for a complex message. */
+		b_queue_entry *msg_entry
+			= b_queue_pop_back(&state->s_operator_stack);
+		struct ivy_ast_node *msg
+			= b_unbox(struct ivy_ast_node, msg_entry, n_entry);
+
+		if (msg->n_type != IVY_AST_MSG) {
+			return PARSE_RESULT(IVY_ERR_INTERNAL_FAILURE, 0);
+		}
+
+		struct expr_parser_state *msg_expr
+			= (struct expr_parser_state *)parser_push_state(
+				ctx, IVY_AST_EXPR);
+
+		msg_expr->s_msg = (struct ivy_ast_msg_node *)msg;
+		msg_expr->s_subexpr = EXPR_SUBTYPE_COMPLEX_MSG;
+		msg_expr->s_type = EXPR_TYPE_ARITH;
+
+		return PARSE_RESULT(IVY_OK, 0);
+	}
+
 	/* a sub-expression surrounded by parentheses is parsed by creating
 	 * a sub-expression parser state */
 	struct expr_parser_state *sub_expr
@@ -551,6 +612,36 @@ static struct ivy_ast_msg_node *finalise_keyword_msg(struct expr_parser_state *s
 	return msg;
 }

+static struct ivy_ast_msg_node *finalise_complex_msg(struct expr_parser_state *state)
+{
+	struct ivy_ast_msg_node *msg = state->s_msg;
+	if (!msg) {
+		return NULL;
+	}
+
+	state->s_msg = NULL;
+
+	b_queue_iterator it = {0};
+	b_queue_iterator_begin(&state->s_labels, &it);
+	while (b_queue_iterator_is_valid(&it)) {
+		struct ivy_token *label
+			= b_unbox(struct ivy_token, it.entry, t_entry);
+		b_queue_iterator_erase(&it);
+		b_queue_push_back(&msg->n_sel->n_arg_labels, &label->t_entry);
+	}
+
+	b_queue_iterator_begin(&state->s_args, &it);
+	while (b_queue_iterator_is_valid(&it)) {
+		struct ivy_ast_node *arg
+			= b_unbox(struct ivy_ast_node, it.entry, n_entry);
+		b_queue_iterator_erase(&it);
+
+		b_queue_push_back(&msg->n_arg, &arg->n_entry);
+	}
+
+	return msg;
+}
+
 static struct token_parse_result parse_right_paren(
 	struct ivy_parser *ctx, struct ivy_token *tok)
 {
@@ -561,11 +652,11 @@ static struct token_parse_result parse_right_paren(
 		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
 	}

-	if (state->s_paren_depth == 0 && !state->s_subexpr) {
+	if (state->s_paren_depth == 0 && state->s_subexpr == EXPR_SUBTYPE_NONE) {
 		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, 0);
 	}

-	if (state->s_recipient) {
+	if (state->s_subexpr == EXPR_SUBTYPE_KEYWORD_MSG) {
 		/* this is the end of a keyword-message */
 		struct ivy_ast_msg_node *msg = finalise_keyword_msg(state);
 		parser_replace_current_node(ctx, (struct ivy_ast_node *)msg);
@@ -573,6 +664,14 @@ static struct token_parse_result parse_right_paren(
 		return PARSE_RESULT(IVY_OK, PARSE_REPEAT_TOKEN);
 	}

+	if (state->s_subexpr == EXPR_SUBTYPE_COMPLEX_MSG) {
+		/* this is the end of a keyword-message */
+		struct ivy_ast_msg_node *msg = finalise_complex_msg(state);
+		parser_replace_current_node(ctx, (struct ivy_ast_node *)msg);
+		parser_pop_state(ctx, STATE_ADD_NODE_TO_PARENT);
+		return PARSE_RESULT(IVY_OK, 0);
+	}
+
 	/* this is the end of a regular parentheses-surrounded sub-expression. */
 	struct ivy_ast_node *expr = NULL;
 	enum ivy_status status
@@ -584,7 +683,7 @@ static struct token_parse_result parse_right_paren(
 	parser_replace_current_node(ctx, expr);

 	int flags = 0;
-	if (state->s_paren_depth == 0 && state->s_subexpr) {
+	if (state->s_paren_depth == 0 && state->s_subexpr != EXPR_SUBTYPE_NONE) {
 		flags = PARSE_REPEAT_TOKEN;
 	}

@@ -626,7 +725,7 @@ static struct token_parse_result parse_dot(
 	parser_replace_current_node(ctx, expr);

 	int flags = 0;
-	if (state->s_subexpr) {
+	if (state->s_subexpr != EXPR_SUBTYPE_NONE) {
 		/* if we are currently parsing a subexpression (such as a
 		 * keyword-message argument), the dot needs to be propagated to
 		 * the parent expression context so that it can be terminated
@@ -649,15 +748,24 @@ static struct token_parse_result parse_label(
 	}

 	if (state->s_terminator == IVY_TOK_LABEL) {
-		/* we are currently parsing a keyword-message argument, and
-		 * have just encountered the label denoting the next argument.
-		 * terminate here and propagate this label to the parent
-		 * keyword-message parser context. */
+		/* we are currently parsing a keyword or complex message
+		 * argument, and have just encountered the label denoting the
+		 * next argument. terminate here and propagate this label to the
+		 * parent keyword-message parser context. */
+		struct ivy_ast_node *expr = NULL;
+		enum ivy_status status
+			= finalise_expr(state, &expr, IVY_PRECEDENCE_ASSIGN);
+		if (status != IVY_OK) {
+			return PARSE_RESULT(status, 0);
+		}
+
+		parser_replace_current_node(ctx, expr);
 		parser_pop_state(ctx, STATE_ADD_NODE_TO_PARENT);
 		return PARSE_RESULT(IVY_ERR_BAD_SYNTAX, PARSE_REPEAT_TOKEN);
 	}

-	if (!state->s_recipient) {
+	if (state->s_subexpr != EXPR_SUBTYPE_KEYWORD_MSG
+	    && state->s_subexpr != EXPR_SUBTYPE_COMPLEX_MSG) {
 		/* this is the beginning of a new keyword-message */
 		struct expr_parser_state *msg_expr
 			= (struct expr_parser_state *)parser_push_state(
@@ -675,20 +783,24 @@ static struct token_parse_result parse_label(
 		}

 		msg_expr->s_recipient = expr;
-		msg_expr->s_subexpr = true;
+		msg_expr->s_subexpr = EXPR_SUBTYPE_KEYWORD_MSG;
 		msg_expr->s_type = EXPR_TYPE_ARITH;
 		state = msg_expr;
 	}

-	/* we have just finished parsing a keyword-message argument, and this
-	 * label marks the start of a new one. store the label and create a
-	 * new argument parsing context. */
-	b_queue_push_back(&state->s_labels, &tok->t_entry);
+	if (state->s_subexpr == EXPR_SUBTYPE_KEYWORD_MSG
+	    || state->s_subexpr == EXPR_SUBTYPE_COMPLEX_MSG) {
+		/* we may have just finished parsing a keyword-message argument,
+		 * and this label marks the start of a new one. store the label
+		 * and create a new argument parsing context. */
+		b_queue_push_back(&state->s_labels, &tok->t_entry);

-	struct expr_parser_state *arg_expr
-		= (struct expr_parser_state *)parser_push_state(ctx, IVY_AST_EXPR);
-	arg_expr->s_terminator = IVY_TOK_LABEL;
-	arg_expr->s_subexpr = true;
+		struct expr_parser_state *arg_expr
+			= (struct expr_parser_state *)parser_push_state(
+				ctx, IVY_AST_EXPR);
+		arg_expr->s_terminator = IVY_TOK_LABEL;
+		arg_expr->s_subexpr = EXPR_SUBTYPE_KEYWORD_ARG;
+	}

 	return PARSE_RESULT(IVY_OK, 0);
 }
@@ -701,9 +813,12 @@ static enum ivy_status add_child(
 		return IVY_ERR_NOT_SUPPORTED;
 	}

-	if (state->s_recipient) {
+	if (state->s_subexpr == EXPR_SUBTYPE_KEYWORD_MSG
+	    || state->s_subexpr == EXPR_SUBTYPE_COMPLEX_MSG) {
 		/* treat the child node as a keyword-message argument */
 		b_queue_push_back(&state->s_args, &child->n_entry);
+	} else if (child->n_type == IVY_AST_MSG) {
+		push_operator(state, child);
 	} else {
 		/* treat the child node as a sub-expression enclosed in
 		 * parentheses (i.e. an operand). */