hare

commit 3e156307174d18e17a60a65a18e7ec40cf7a40fb
parent 8da4b7361169870e902eaa1ce4d98e84d22270a9
Author: Adnan Maolood <adnan@maolood.com>
Date:   Tue, 14 Mar 2023 12:56:15 -0400

regex: Handle escaped characters in brackets

Signed-off-by: Adnan Maolood <adnan@maolood.com>

Diffstat:
M
regex/+test.ha
 | 
3
+++
M
regex/regex.ha
 | 
20
+++++++++++++-------

2 files changed, 16 insertions(+), 7 deletions(-)
diff --git a/regex/+test.ha b/regex/+test.ha
@@ -333,6 +333,9 @@ fn run_findall_case(
 		(`^a\\b$`, "a\\b", matchres::MATCH, 0, -1),
 		(`^x(abc)\{,2\}$`, "xabc{,2}", matchres::MATCH, 0, -1),
 		(`^x(abc)\{,2\}$`, "xabcabc{,2}", matchres::NOMATCH, 0, -1),
+		(`^[\\]+$`, "\\", matchres::MATCH, 0, -1),
+		(`^[\]]+$`, "]", matchres::MATCH, 0, -1),
+		(`^[A-Za-z\[\]]+$`, "foo[bar]baz", matchres::MATCH, 0, -1),
 		// {m,n}
 		(`^x(abc){2}$`, "xabcabc", matchres::MATCH, 0, -1),
 		(`^x(abc){3}$`, "xabcabc", matchres::NOMATCH, 0, -1),
diff --git a/regex/regex.ha b/regex/regex.ha
@@ -146,7 +146,17 @@ fn handle_bracket(
 	const range_end = peek2;
 	const is_first_char = *bracket_idx == 0 || *bracket_idx == 1
 		&& !*is_charset_positive;
-	if (r == ']' && !is_first_char) {
+
+	if (r == '\\') {
+		if (peek1 is void) {
+			return `Trailing backslash '\'`: error;
+		} else {
+			append(charsets[len(charsets) - 1],
+				peek1: charset_lit_item);
+			strings::next(iter);
+			*r_idx += 1;
+		};
+	} else if (r == ']' && !is_first_char) {
 		const newinst = inst_charset {
 			idx = len(charsets) - 1,
 			is_positive = *is_charset_positive,
@@ -212,7 +222,7 @@ export fn compile(expr: str) (regex | error) = {
 		const next = strings::next(&iter);
 
 		if (r_idx == 0 && next is rune && next: rune != '^') {
-				append(insts, void: inst_skip);
+			append(insts, void: inst_skip);
 		};
 
 		if (in_bracket) {
@@ -256,11 +266,7 @@ export fn compile(expr: str) (regex | error) = {
 		case '[' =>
 			in_bracket = true;
 		case ']' =>
-			if (in_bracket) {
-				in_bracket = false;
-			} else {
-				append(insts, r: inst_lit);
-			};
+			append(insts, r: inst_lit);
 		case '(' =>
 			if (n_groupstarts > 0) {
 				return `Nested capture groups are unsupported`: error;