Find named groups with letters that are not ASCII in --js to close #816

src/engine.ts

@@ -835,15 +835,20 @@ function countFile(config, text, result) {
 // The code uses a named group of the pattern, like rr_year for (?<year>…)
 function usesNamedGroups(config) {
 	return namedGroups(config.regex).some(name =>
-		new RegExp('(^|[^\\w$])rr_' + name.replace(/\$/g, '\\$') + '(?![\\w$])').test(
-			config.replacementOri
-		)
+		new RegExp(
+			// rr_year_ is the group with a space at the end, so the code uses the group then too
+			'(^|[^\\p{ID_Continue}$])rr_' + name.replace(/\$/g, '\\$') + '_?(?![\\p{ID_Continue}$])',
+			'u'
+		).test(config.replacementOri)
 	);
 }
 
-// The names of the named groups of a regex, like year for (?<year>…)
+// The names of the named groups of a regex, like year for (?<year>…). A name can have any letter, like año.
+// They come from the regex itself, so text that only looks like a group, like \(?<a> or [(?<a>], is not one.
+// With | at the end the regex matches an empty text, and the match has all the groups.
 function namedGroups(regex) {
-	return [...regex.source.matchAll(/\(\?<([A-Za-z_$][\w$]*)>/g)].map(([, name]) => name);
+	const match = new RegExp(regex.source + '|', regex.flags.replace(/[gy]/g, '')).exec('');
+	return Object.keys(match?.groups ?? {});
 }
 
 // The numbers of --js and their cousins: $n0 is $n with zeros in front to 2 digits, $n00 to 3 and so on to 10,

test/fix/816.test.js

@@ -0,0 +1,41 @@
+// https://github.com/mathiasrw/rexreplace/issues/816
+// A named group with a letter that is not ASCII was not seen in the code, so the code ran once per file, where the group was not there.
+import {describe, test, expect} from 'bun:test';
+import {rr} from '../helper.js';
+
+describe('#816 named groups with letters that are not ASCII', () => {
+	test.each([
+		['año', '(?<año>\\d)', '"[" + rr_año + "]"', 'a[1]b[2]'],
+		['日', '(?<日>\\d)', 'rr_日 * 2', 'a2b4'],
+		['año with a space at the end', '(?<año>\\d)', 'rr_año_ + "|"', 'a1 |b2 |'],
+		['𝒜, outside the BMP', '(?<𝒜>\\d)', 'rr_𝒜 * 2', 'a2b4'],
+		['$é, with $', '(?<$é>\\d)', 'rr_$é * 2', 'a2b4'],
+		['Æble, in uppercase', '(?<Æble>\\d)', 'rr_Æble * 2', 'a2b4'],
+		['x٣, with a digit that is not ASCII', '(?<x٣>\\d)', 'rr_x٣ * 2', 'a2b4'],
+	])('can be used, like %s', async (name, pattern, code, out) => {
+		expect((await rr(pattern, code, {pipedData: 'a1b2', replacementJs: true})).out).toBe(out);
+	});
+
+	test('can not end with _ either', async () => {
+		const result = await rr('(?<año_>x)', 'rr_año', {pipedData: 'x', replacementJs: true});
+		expect(result.err).toStartWith(' ❌ A named group can not end with _: año_');
+		expect(result.exit).toBe(3);
+	});
+
+	test('are read from the regex, so text that only looks like a group is not one', async () => {
+		const config = {replacementJs: true};
+		expect((await rr('\\(?<a_>b', '"z"', {...config, pipedData: 'x(?<a_>b'})).out).toBe('x(?z');
+		expect((await rr('[(?<a_>]', '"z"', {...config, pipedData: 'ab'})).out).toBe('zb');
+	});
+
+	test('are only seen as the whole name, so rr_rr_year is not the group year', async () => {
+		const result = await rr('(?<year>\\d)', 'rr_rr_year', {pipedData: '1', replacementJs: true});
+		expect(result.err).toContain('ReferenceError: rr_rr_year is not defined');
+		expect(result.exit).toBe(3);
+	});
+
+	test('can be used with a space at the end on its own, like rr_year_', async () => {
+		const config = {pipedData: '2024', replacementJs: true};
+		expect((await rr('(?<year>\\d+)', 'rr_year_ + "|"', config)).out).toBe('2024 |');
+	});
+});