From 1ce80e005a374488c186d0f545af33096f6523d5 Mon Sep 17 00:00:00 2001
From: EuAndreh <eu@euandre.org>
Date: Tue, 15 Jul 2025 21:37:16 -0300
Subject: Only tolerate escaping of special chars

* src/paca.mjs
  (escapingStateStep): Return an error when escaping non-metacharacters.
    This way cases like \d, which is syntax for [0-9] which will
    eventually be recognized, will not change its behaviour from a noop
    escape of "d" to matching digits.
  (operatorChars, isOperator): Hoist both of these up before their usage
    in `escapingStateStep()`.
* tests/paca.mjs
  (test_isOperator): Hoist its definition and position inside the
    `runTests([...])` array to match src/paca.mjs.
  (test_escapingStateStep): Adjust existing cases and add test case for
    good/bad escapes.
  (test_tokenizeRegexStep): Fix bad starting escape, that broke because
    it was escaping a non-metacharacter.
---
 src/paca.mjs | 34 ++++++++++++++++++++++------------
 1 file changed, 22 insertions(+), 12 deletions(-)

(limited to 'src')

diff --git a/src/paca.mjs b/src/paca.mjs
index 5d11b05..7a01407 100644
--- a/src/paca.mjs
+++ b/src/paca.mjs
@@ -24,19 +24,33 @@ const shouldConcat = (char, next) =>
 	char !== "{" &&
 	!nonConcatOperators.has(next);
 
+const operatorChars = new Set([...nonConcatOperators, "(", "."]);
+const isOperator = char =>
+	operatorChars.has(char);
+
 const numFromDigits = digits =>
 	digits.length === 0
 		? -1
 		: Number(digits.join(""));
 
-const escapingStateStep = ({ out, _state, context }, char, _index, next) => ({
-	out: out.concat(
-		char,
-		shouldConcat(null, next) ? [{ operator: "concat" }] : [],
-	),
-	state: ConcatStep.ACCEPTING,
-	context,
-});
+const escapingStateStep = ({ out, state, context }, char, _index, next) => 
+	!(isOperator(char) || char === "\\")
+		? reduced({
+			out,
+			state,
+			context,
+			error: new SyntaxError(
+				"unknown escape sequence: \\" + char,
+			),
+		})
+		: {
+			out: out.concat(
+				char,
+				shouldConcat(null, next) ? [{ operator: "concat" }] : [],
+			),
+			state: ConcatStep.ACCEPTING,
+			context,
+		};
 
 const rangeStateStep = ({ out, state, context }, char, _index, _next) => {
 	if (char === "}") {
@@ -313,10 +327,6 @@ const transitionChars = new Set(Object.keys(TRANSITION_FNS));
 const isTransition = char =>
 	transitionChars.has(char);
 
-const operatorChars = new Set([...nonConcatOperators, "(", "."]);
-const isOperator = char =>
-	operatorChars.has(char);
-
 const tokenizeRegexStep = chars => ({ out, state, context }, char, index) => {
 	const next = chars[index + 1];
 
-- 
cgit v1.2.3