lang: lex: implement state stack; make lexer structure opaque

the state stack is used to track whether a string is currently being read, what kind of string it is, and whether or not we are in an interpolation within that string.
2024-11-16 23:03:43 +00:00
parent 412e9abf05
commit f14c387a6e
2 changed files with 157 additions and 35 deletions
--- a/lang/lex.c
+++ b/lang/lex.c
@@ -15,6 +15,24 @@
 		.id = (i), .name = (n)                                         \
 	}

+struct ivy_lexer {
+	struct ivy_lexer_symbol_node *lex_sym_tree;
+	struct ivy_line_source *lex_source;
+
+	enum ivy_status lex_status;
+
+	struct ivy_token *lex_queue;
+	enum ivy_token_type lex_prev_token;
+
+	b_queue lex_state;
+	unsigned int lex_brace_depth;
+
+	char *lex_linebuf;
+	size_t lex_linebuf_len;
+	size_t lex_linebuf_cap;
+	size_t lex_linebuf_ptr;
+};
+
 enum lexer_state_type {
 	STATE_NORMAL,
 	STATE_STRING,
@@ -24,6 +42,8 @@ enum lexer_state_type {

 struct lexer_state {
 	enum lexer_state_type s_type;
+	unsigned int s_brace_depth;
+	b_queue_entry s_entry;
 };

 struct ivy_lexer_symbol_node {
@@ -112,6 +132,57 @@ static struct lex_token_def symbols[] = {
 };
 static const size_t nr_symbols = sizeof symbols / sizeof symbols[0];

+static struct lexer_state *push_lexer_state(
+	struct ivy_lexer *lex, enum lexer_state_type state_type)
+{
+	struct lexer_state *state = malloc(sizeof *state);
+	if (!state) {
+		return NULL;
+	}
+
+	memset(state, 0x0, sizeof *state);
+
+	state->s_type = state_type;
+	state->s_brace_depth = lex->lex_brace_depth;
+	b_queue_push_back(&lex->lex_state, &state->s_entry);
+
+	return state;
+}
+
+static void pop_lexer_state(struct ivy_lexer *lex)
+{
+	b_queue_entry *entry = b_queue_pop_back(&lex->lex_state);
+	if (!entry) {
+		return;
+	}
+
+	struct lexer_state *state = b_unbox(struct lexer_state, entry, s_entry);
+	free(state);
+}
+
+static struct lexer_state *get_lexer_state(struct ivy_lexer *lex)
+{
+	b_queue_entry *entry = b_queue_last(&lex->lex_state);
+	if (!entry) {
+		return NULL;
+	}
+
+	return b_unbox(struct lexer_state, entry, s_entry);
+}
+
+static void destroy_state_stack(b_queue *state)
+{
+	b_queue_iterator it;
+	b_queue_iterator_begin(state, &it);
+	while (b_queue_iterator_is_valid(&it)) {
+		struct lexer_state *node
+			= b_unbox(struct lexer_state, it.entry, s_entry);
+		b_queue_iterator_erase(&it);
+
+		free(node);
+	}
+}
+
 static struct ivy_lexer_symbol_node *get_symbol_node(
 	struct ivy_lexer_symbol_node *node, char c)
 {
@@ -245,8 +316,13 @@ static enum ivy_keyword find_keyword_by_name(const char *s)
 	return IVY_KW_NONE;
 }

-enum ivy_status ivy_lexer_init(struct ivy_lexer *lex)
+enum ivy_status ivy_lexer_create(struct ivy_lexer **lexp)
 {
+	struct ivy_lexer *lex = malloc(sizeof *lex);
+	if (!lex) {
+		return IVY_ERR_NO_MEMORY;
+	}
+
 	memset(lex, 0x0, sizeof *lex);

 	lex->lex_status = IVY_OK;
@@ -256,15 +332,26 @@ enum ivy_status ivy_lexer_init(struct ivy_lexer *lex)
 	lex->lex_linebuf_cap = LINEBUF_DEFAULT_CAPACITY;

 	lex->lex_sym_tree = build_symbol_tree();
+	if (!lex->lex_sym_tree) {
+		ivy_lexer_destroy(lex);
+		return IVY_ERR_NO_MEMORY;
+	}
+
+	if (!push_lexer_state(lex, STATE_NORMAL)) {
+		ivy_lexer_destroy(lex);
+		return IVY_ERR_NO_MEMORY;
+	}
+
 	print_symbol_node(lex->lex_sym_tree, 0);

 	/* TODO only do keyword initialisation once */
 	init_keywords();
+	*lexp = lex;

 	return IVY_OK;
 }

-void ivy_lexer_finish(struct ivy_lexer *lex)
+void ivy_lexer_destroy(struct ivy_lexer *lex)
 {
 	while (lex->lex_queue) {
 		struct ivy_token *next = lex->lex_queue->t_next;
@@ -280,7 +367,19 @@ void ivy_lexer_finish(struct ivy_lexer *lex)
 		destroy_symbol_tree(lex->lex_sym_tree);
 	}

-	memset(lex, 0x0, sizeof *lex);
+	destroy_state_stack(&lex->lex_state);
+
+	free(lex);
+}
+
+void ivy_lexer_set_source(struct ivy_lexer *lex, struct ivy_line_source *src)
+{
+	lex->lex_source = src;
+}
+
+enum ivy_status ivy_lexer_get_status(struct ivy_lexer *lex)
+{
+	return lex->lex_status;
 }

 static enum ivy_status refill_linebuf(struct ivy_lexer *lex)
@@ -538,27 +637,38 @@ static enum ivy_status read_block_comment(struct ivy_lexer *lex)
 static enum ivy_status read_squote_marker(struct ivy_lexer *lex)
 {
 	enum ivy_status status = IVY_OK;
+	struct lexer_state *state = get_lexer_state(lex);

-	if (lex->lex_state & STATE_FSTRING) {
+	if (state->s_type == STATE_FSTRING) {
 		/* already within an fstring */
-		lex->lex_state &= ~STATE_FSTRING;
-		return push_string_end(lex);	
-	} else {
-		/* start of a new fstring */
-		status = push_string_start(lex);
-		lex->lex_state |= STATE_FSTRING;
+		pop_lexer_state(lex);
+		return push_string_end(lex);
 	}
+
+	/* start of a new fstring */
+	status = push_string_start(lex);
+
+	if (status != IVY_OK) {
+		return status;
+	}
+
+	if (!push_lexer_state(lex, STATE_FSTRING)) {
+		return IVY_ERR_NO_MEMORY;
+	}
+
+	return IVY_OK;
 }

 static enum ivy_status read_dquote_marker(struct ivy_lexer *lex)
 {
-
+	return IVY_OK;
 }

 static enum ivy_status read_string_content(struct ivy_lexer *lex)
 {
 	int c;
 	b_string *str = b_string_create();
+	struct lexer_state *state = get_lexer_state(lex);

 	if (!str) {
 		return IVY_ERR_NO_MEMORY;
@@ -567,20 +677,17 @@ static enum ivy_status read_string_content(struct ivy_lexer *lex)
 	while (true) {
 		c = peek(lex);

-		if (c == '{') {
+		if (state->s_type == STATE_FSTRING && (c == '\'' || c == '{')) {
 			break;
 		}

-		if ((lex->lex_state & STATE_FSTRING) && c == '\'') {
-			break;
-		}
-
-		if ((lex->lex_state & STATE_STRING) && c == '"') {
+		if (state->s_type == STATE_STRING && c == '"') {
 			break;
 		}

 		char s[2] = {c, 0};
 		b_string_append_cstr(str, s);
+		advance(lex);
 	}

 	if (b_string_get_size(str, B_STRLEN_NORMAL) == 0) {
@@ -602,6 +709,7 @@ static enum ivy_status read_string_content(struct ivy_lexer *lex)
 static enum ivy_status read_symbol(struct ivy_lexer *lex)
 {
 	struct ivy_lexer_symbol_node *node = lex->lex_sym_tree;
+	struct lexer_state *state = get_lexer_state(lex);

 	while (true) {
 		int c = peek(lex);
@@ -628,6 +736,23 @@ static enum ivy_status read_symbol(struct ivy_lexer *lex)
 		return read_block_comment(lex);
 	case IVY_SYM_DOUBLE_HYPHEN:
 		return read_line_comment(lex);
+	case IVY_SYM_LEFT_BRACE:
+		push_symbol(lex, node->s_id);
+		lex->lex_brace_depth++;
+
+		if (state->s_type == STATE_FSTRING) {
+			push_lexer_state(lex, STATE_INTERPOLATION);
+		}
+		return IVY_OK;
+	case IVY_SYM_RIGHT_BRACE:
+		push_symbol(lex, node->s_id);
+		lex->lex_brace_depth--;
+
+		if (state->s_type == STATE_INTERPOLATION
+		    && lex->lex_brace_depth < state->s_brace_depth) {
+			pop_lexer_state(lex);
+		}
+		return IVY_OK;
 	default:
 		push_symbol(lex, node->s_id);
 		return IVY_OK;
@@ -684,20 +809,26 @@ static enum ivy_status read_ident(struct ivy_lexer *lex)
 static enum ivy_status pump_tokens(struct ivy_lexer *lex)
 {
 	enum ivy_status status;
+	struct lexer_state *state = get_lexer_state(lex);
+
 	int c = peek(lex);

 	if (c < 0) {
 		return c;
 	}

-	if (lex->lex_state & STATE_STRING && c != '"') {
+	if (state->s_type == STATE_STRING && c != '"') {
 		return read_string_content(lex);
 	}

-	if ((lex->lex_state & STATE_FSTRING) && !(lex->lex_state & STATE_INTERPOLATION)) {
+	if (state->s_type == STATE_FSTRING && c != '\'' && c != '{') {
 		return read_string_content(lex);
 	}

+	/* `state` is invalid past this point, as the read_* functions may
+	 * perform state transitions. */
+	state = NULL;
+
 	if (c == '\n') {
 		while (c == '\n') {
 			advance(lex);