Fix Node.js without ICU support (#1144)

ehmicky · web-flow · commit 074ea4ab50a9 · 2024-08-14T14:54:56.000+02:00
diff --git a/lib/arguments/escape.js b/lib/arguments/escape.js
@@ -38,7 +38,23 @@ const escapeControlCharacter = character => {
 // Some shells do not even have a way to print those characters in an escaped fashion.
 // Therefore, we prioritize printing those safely, instead of allowing those to be copy-pasted.
 // List of Unicode character categories: https://www.fileformat.info/info/unicode/category/index.htm
-const SPECIAL_CHAR_REGEXP = /\p{Separator}|\p{Other}/gu;
+const getSpecialCharRegExp = () => {
+	try {
+		// This throws when using Node.js without ICU support.
+		// When using a RegExp literal, this would throw at parsing-time, instead of runtime.
+		// eslint-disable-next-line prefer-regex-literals
+		return new RegExp('\\p{Separator}|\\p{Other}', 'gu');
+	} catch {
+		// Similar to the above RegExp, but works even when Node.js has been built without ICU support.
+		// Unlike the above RegExp, it only covers whitespaces and C0/C1 control characters.
+		// It does not cover some edge cases, such as Unicode reserved characters.
+		// See https://github.com/sindresorhus/execa/issues/1143
+		// eslint-disable-next-line no-control-regex
+		return /[\s\u0000-\u001F\u007F-\u009F\u00AD]/g;
+	}
+};
+
+const SPECIAL_CHAR_REGEXP = getSpecialCharRegExp();
 
 // Accepted by $'...' in Bash.
 // Exclude \a \e \v which are accepted in Bash but not in JavaScript (except \v) and JSON.
diff --git a/test/arguments/escape-no-icu.js b/test/arguments/escape-no-icu.js
@@ -0,0 +1,16 @@
+// Mimics Node.js when built without ICU support
+// See https://github.com/sindresorhus/execa/issues/1143
+globalThis.RegExp = class extends RegExp {
+	constructor(regExpString, flags) {
+		if (flags?.includes('u') && regExpString.includes('\\p{')) {
+			throw new Error('Invalid property name');
+		}
+
+		super(regExpString, flags);
+	}
+
+	static isMocked = true;
+};
+
+// Execa computes the RegExp when first loaded, so we must delay this import
+await import('./escape.js');
diff --git a/test/arguments/escape.js b/test/arguments/escape.js
@@ -21,8 +21,11 @@ test(testResultCommand, ' foo bar', 'foo', 'bar');
 test(testResultCommand, ' baz quz', 'baz', 'quz');
 test(testResultCommand, '');
 
-const testEscapedCommand = async (t, commandArguments, expectedUnix, expectedWindows) => {
-	const expected = isWindows ? expectedWindows : expectedUnix;
+// eslint-disable-next-line max-params
+const testEscapedCommand = async (t, commandArguments, expectedUnix, expectedWindows, expectedUnixNoIcu = expectedUnix, expectedWindowsNoIcu = expectedWindows) => {
+	const expected = RegExp.isMocked
+		? (isWindows ? expectedWindowsNoIcu : expectedUnixNoIcu)
+		: (isWindows ? expectedWindows : expectedUnix);
 
 	t.like(
 		await t.throwsAsync(execa('fail.js', commandArguments)),
@@ -89,12 +92,12 @@ test('result.escapedCommand - \\x01', testEscapedCommand, ['\u0001'], '\'\\u0001
 test('result.escapedCommand - \\x7f', testEscapedCommand, ['\u007F'], '\'\\u007f\'', '"\\u007f"');
 test('result.escapedCommand - \\u0085', testEscapedCommand, ['\u0085'], '\'\\u0085\'', '"\\u0085"');
 test('result.escapedCommand - \\u2000', testEscapedCommand, ['\u2000'], '\'\\u2000\'', '"\\u2000"');
-test('result.escapedCommand - \\u200E', testEscapedCommand, ['\u200E'], '\'\\u200e\'', '"\\u200e"');
+test('result.escapedCommand - \\u200E', testEscapedCommand, ['\u200E'], '\'\\u200e\'', '"\\u200e"', '\'\u200E\'', '"\u200E"');
 test('result.escapedCommand - \\u2028', testEscapedCommand, ['\u2028'], '\'\\u2028\'', '"\\u2028"');
 test('result.escapedCommand - \\u2029', testEscapedCommand, ['\u2029'], '\'\\u2029\'', '"\\u2029"');
 test('result.escapedCommand - \\u5555', testEscapedCommand, ['\u5555'], '\'\u5555\'', '"\u5555"');
-test('result.escapedCommand - \\uD800', testEscapedCommand, ['\uD800'], '\'\\ud800\'', '"\\ud800"');
-test('result.escapedCommand - \\uE000', testEscapedCommand, ['\uE000'], '\'\\ue000\'', '"\\ue000"');
+test('result.escapedCommand - \\uD800', testEscapedCommand, ['\uD800'], '\'\\ud800\'', '"\\ud800"', '\'\uD800\'', '"\uD800"');
+test('result.escapedCommand - \\uE000', testEscapedCommand, ['\uE000'], '\'\\ue000\'', '"\\ue000"', '\'\uE000\'', '"\uE000"');
 test('result.escapedCommand - \\U1D172', testEscapedCommand, ['\u{1D172}'], '\'\u{1D172}\'', '"\u{1D172}"');
-test('result.escapedCommand - \\U1D173', testEscapedCommand, ['\u{1D173}'], '\'\\U1d173\'', '"\\U1d173"');
-test('result.escapedCommand - \\U10FFFD', testEscapedCommand, ['\u{10FFFD}'], '\'\\U10fffd\'', '"\\U10fffd"');
+test('result.escapedCommand - \\U1D173', testEscapedCommand, ['\u{1D173}'], '\'\\U1d173\'', '"\\U1d173"', '\'\u{1D173}\'', '"\u{1D173}"');
+test('result.escapedCommand - \\U10FFFD', testEscapedCommand, ['\u{10FFFD}'], '\'\\U10fffd\'', '"\\U10fffd"', '\'\u{10FFFD}\'', '"\u{10FFFD}"');
