harec

commit f9eff2c2abe9d87037320f7310e16f1d902a472e
parent 641f60a61b7fd59d640fa2c35fbf3b4dbd1b8e12
Author: Sebastian <sebastian@sebsite.pw>
Date:   Sun, 12 Jun 2022 19:30:55 -0400

Parse labels in parse instead of lex

As per the change to the spec which allows whitespace characters between
the colon and name of a label, labels are now handled during parse
instead of lex. The lexer only sees a colon and a name, and the label is
constructed by the parser from this.

Signed-off-by: Sebastian <sebastian@sebsite.pw>

Diffstat:
M
include/lex.h
 | 
2
--
M
src/lex.c
 | 
26
--------------------------
M
src/main.c
 | 
7
-------
M
src/parse.c
 | 
15
+++++++++------

4 files changed, 9 insertions(+), 41 deletions(-)
diff --git a/include/lex.h b/include/lex.h
@@ -129,7 +129,6 @@ enum lexical_token {
 	T_LAST_OPERATOR = T_BXOREQ,
 
 	// Tokens with additional information
-	T_LABEL,
 	T_LITERAL,
 	T_NAME,
 
@@ -167,7 +166,6 @@ struct lexer {
 	uint32_t c[2];
 	struct token un;
 	struct location loc;
-	bool disable_labels;
 	bool require_int;
 };
 
diff --git a/src/lex.c b/src/lex.c
@@ -787,22 +787,6 @@ lex3(struct lexer *lexer, struct token *out, uint32_t c)
 	return out->token;
 }
 
-static enum lexical_token
-lex_label(struct lexer *lexer, struct token *out)
-{
-	uint32_t c;
-	while ((c = next(lexer, NULL, true)) != UTF8_INVALID) {
-		if (c > 0x7F || (!isalnum(c) && c != '_')) {
-			push(lexer, c, true);
-			break;
-		}
-	}
-	out->token = T_LABEL;
-	out->name = strdup(lexer->buf);
-	consume(lexer, -1);
-	return out->token;
-}
-
 static enum lexical_token _lex(struct lexer *lexer, struct token *out);
 
 static enum lexical_token
@@ -876,10 +860,6 @@ lex2(struct lexer *lexer, struct token *out, uint32_t c)
 			break;
 		default:
 			push(lexer, c, false);
-			if (!lexer->disable_labels && c <= 0x7F
-					&& (isalpha(c) || c == '_')) {
-				return lex_label(lexer, out);
-			}
 			out->token = T_COLON;
 			break;
 		}
@@ -1018,7 +998,6 @@ token_finish(struct token *tok)
 {
 	switch (tok->token) {
 	case T_NAME:
-	case T_LABEL:
 		free(tok->name);
 		break;
 	case T_LITERAL:
@@ -1046,8 +1025,6 @@ lexical_token_str(enum lexical_token tok)
 	switch (tok) {
 	case T_NAME:
 		return "name";
-	case T_LABEL:
-		return "label";
 	case T_LITERAL:
 		return "literal";
 	case T_EOF:
@@ -1142,9 +1119,6 @@ token_str(const struct token *tok)
 	case T_NAME:
 		snprintf(buf, sizeof(buf), "name %s", tok->name);
 		return buf;
-	case T_LABEL:
-		snprintf(buf, sizeof(buf), ":%s", tok->name);
-		return buf;
 	case T_LITERAL:
 		switch (tok->storage) {
 		case STORAGE_U8:
diff --git a/src/main.c b/src/main.c
@@ -64,13 +64,6 @@ parse_define(const char *argv_0, const char *in)
 	sources = &d;
 	lex_init(&lexer, f, 0);
 
-	// The syntax for this parameter is:
-	//
-	// -D ident:type=value
-	//
-	// :type is lexed as a label unless we disable it here.
-	lexer.disable_labels = true;
-
 	parse_identifier(&lexer, &def->ident, false);
 	if (lex(&lexer, &tok) != T_COLON) {
 		lex_finish(&lexer);
diff --git a/src/parse.c b/src/parse.c
@@ -2164,7 +2164,8 @@ parse_control_expression(struct lexer *lexer)
 		exp->type = tok.token == T_BREAK ? EXPR_BREAK : EXPR_CONTINUE;
 		exp->control.label = NULL;
 		switch (lex(lexer, &tok)) {
-		case T_LABEL:
+		case T_COLON:
+			want(lexer, T_NAME, &tok);
 			exp->control.label = tok.name;
 			break;
 		default:
@@ -2193,7 +2194,8 @@ parse_control_expression(struct lexer *lexer)
 		case T_SEMICOLON:
 			unlex(lexer, &tok);
 			break;
-		case T_LABEL:
+		case T_COLON:
+			want(lexer, T_NAME, &tok);
 			exp->control.label = tok.name;
 			switch (lex(lexer, &tok)) {
 			case T_COMMA:
@@ -2228,14 +2230,15 @@ parse_compound_expression(struct lexer *lexer)
 
 	struct token tok = {0};
 	switch (lex(lexer, &tok)) {
-	case T_LABEL:
+	case T_COLON:
+		want(lexer, T_NAME, &tok);
 		exp->compound.label = tok.name;
 		want(lexer, T_LBRACE, &tok);
 		break;
 	case T_LBRACE:
 		break; // no-op
 	default:
-		synerr(&tok, T_LBRACE, T_LABEL, T_EOF);
+		synerr(&tok, T_LBRACE, T_COLON, T_EOF);
 		break;
 	};
 
@@ -2291,11 +2294,11 @@ parse_expression(struct lexer *lexer)
 	case T_YIELD:
 	case T_DEFER:
 	case T_FOR:
-	case T_LABEL:
 	case T_IF:
 	case T_LBRACE:
 	case T_MATCH:
 	case T_SWITCH:
+	case T_COLON:
 		switch (tok.token) {
 		case T_BREAK:
 		case T_CONTINUE:
@@ -2315,7 +2318,7 @@ parse_expression(struct lexer *lexer)
 			value = parse_if_expression(lexer);
 			break;
 		case T_LBRACE:
-		case T_LABEL:
+		case T_COLON:
 			unlex(lexer, &tok);
 			value = parse_compound_expression(lexer);
 			value = parse_cast_expression(lexer, value);