[RegExp] Match updated spec for `/\w/iu` and `/\W/iu`

[mathiasbynens]

Now that tc39/ecma262#525 has landed, \u017F (LATIN SMALL LETTER LONG S) and \u212A (KELVIN SIGN) are word characters for /iu patterns.

Expected vs. actual behavior:

/\w/iu.test('\u017F') // true
/\w/iu.test('\u212A') // true
/\W/iu.test('\u017F') // false, actual: true
/\W/iu.test('\u212A') // false, actual: true
/\W/iu.test('s')      // false
/\W/iu.test('S')      // false
/\W/iu.test('K')      // false
/\W/iu.test('k')      // false

See also

• V8: https://bugs.chromium.org/p/v8/issues/detail?id=5145
• JavaScriptCore: https://bugs.webkit.org/show_bug.cgi?id=159055
• SpiderMonkey: https://bugzilla.mozilla.org/show_bug.cgi?id=1281739
• regexpu-core: Match updated spec for /\w/iu and /\W/iu  mathiasbynens/regexpu-core#8

[dilijev]

Related to #517

[dilijev]

Just to clarify, I'm seeing in the following in ChakraCore for actual behavior.

/\w/iu.test('\u017F') // expected: true,  actual: false ***
/\w/iu.test('\u212A') // expected: true,  actual: false ***
/\W/iu.test('\u017F') // expected: false, actual: true
/\W/iu.test('\u212A') // expected: false, actual: true
/\W/iu.test('s')      // expected: false
/\W/iu.test('S')      // expected: false
/\W/iu.test('K')      // expected: false
/\W/iu.test('k')      // expected: false

Since sharp s and kelvin are word characters, \w should match those two characters (first two lines above).
Also since \W should be the inverse set of \w, \W does not match those characters. Correct?

[mathiasbynens]

The expected behavior you’ve outlined is correct.

One minor correction, though:

Since sharp s and kelvin are word characters

They’re not (as far as \w goes in JavaScript), but with the i flag set they canonicalize to S and K which are.

[dilijev]

@mathiasbynens Oh, I see. That makes sense. I still think the simplest / most efficient way to get this behavior is to add them to the char set we use for \w when both /i and /u flags are set.