Matching lone surrogates *only*

[mathiasbynens]

See mathiasbynens/regexpu#16 and https://gist.github.com/mathiasbynens/bbe7f870208abcfec860.

var set = regenerate()
  .addRange(0xD800, 0xDBFF) // lone high surrogates
  .addRange(0xDC00, 0xDFFF); // lone low surrogates

var match = '𝌆'.match(RegExp('(' + set.toString() + ')'));
console.log(match == null);
// expected: true
// actual: false, since the surrogate halves are matched

Instead, it would make more sense to match lone surrogates only in such cases.

[mathiasbynens]

This doesn’t fully fix the issue. Doing so is hard, since JS doesn’t support lookbehind.

var set = regenerate().addRange(0xD800, 0xDBFF).addRange(0xDC00, 0xDFFF);

var regex = RegExp('^a(?:' + set.toString() + ')b$');
// currently, this results in `/^a(?:[\uD800-\uDBFF](?![\uDC00-\uDFFF])|(?:[^\uD800-\uDBFF]|^)[\uDC00-\uDFFF])b$/`
console.log(regex.test('a\uD834b')); // expected: true; actual: true
console.log(regex.test('a\uDC00b')); // expected: true; actual: false

[mathiasbynens]

Given the above code:

var set = regenerate().addRange(0xD800, 0xDBFF).addRange(0xDC00, 0xDFFF);

var regex = RegExp('^a(?:' + set.toString() + ')b$');
// currently, this results in `/^a(?:[\uD800-\uDBFF](?![\uDC00-\uDFFF])|(?:[^\uD800-\uDBFF]|^)[\uDC00-\uDFFF])b$/`

Consider these three tests:

console.log(regex.test('a\uD834b')); // expected: true
console.log(regex.test('a\uDC00b')); // expected: true
console.log(regex.test('a\uD834\uDF06b')); // expected: false

There are two options:

a. Either we pass test 1 and 3 but fail test 2 (i.e. lone low surrogates aren’t matched accurately). (As in the current implementation in v1.2.1.)
b. We pass test 1 and 2 but fail test 3 (i.e., surrogate halves in pairs are matched as if they are lone surrogates).

Which is the lesser evil — a or b?

[mathiasbynens]

As Marja said:

Paired surrogates are the normal case, so I think cases involving only them should have priority.

Let’s go with a, i.e. the current implementation.