Correctly escape non-unicode regex

[sebastienros]

No description provided.

[lahma]

Something seems to break with these changes.

• I replaced NuGet reference in Jint.csproj with local reference to this branch
• Enabeld unicode tests in Tests262Test.cs (search for unicode)
• Enabled unicode tests in Jint.Tests.Tests262\test\skipped.json (search for unicode)

Observations:

• Jint.Tests.CommonScripts has now failing test in babel testing:
  • babel-standalone.js - Line 79828: Invalid regular expression
• Jint.Tests.Ecma now has 8 failing tests
  • one example: Error: #0: var arr = /\u0000/.exec(\u0000); arr[0] === "\u0000". Actual. null
• Jint.Tests.Tests262 had 20 failing tests with unicode enabled, now it has 22

[lahma]

Looks better now, I still see two test cases failing for old Ecma tests:

• Ecma Chapter 7(sourceFile: ch07/7.8/7.8.5/S7.8.5_A1.1_T1.js)
• Ecma Chapter 7(sourceFile: ch07/7.8/7.8.5/S7.8.5_A2.1_T1.js)

And then the failing ones in new test suite when unicode enabled (the shouldn't be blockers as new tests):

• built-ins\RegExp(sourceFile: built-ins/RegExp/dotall/without-dotall-unicode.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/dotall/without-dotall.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/property-escapes/character-class.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/prototype/Symbol.match/builtin-infer-unicode.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/prototype/Symbol.match/builtin-success-u-return-val-groups.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/prototype/Symbol.match/u-advance-after-empty.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/prototype/Symbol.replace/u-advance-after-empty.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/prototype/Symbol.search/u-lastindex-advance.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/unicode_restricted_brackets.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/unicode_restricted_character_class_escape.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/unicode_restricted_identity_escape.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/unicode_restricted_identity_escape_alpha.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/unicode_restricted_identity_escape_c.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/unicode_restricted_identity_escape_u.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/unicode_restricted_incomple_quantifier.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/unicode_restricted_octal_escape.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/unicode_restricted_quantifiable_assertion.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/character-class-escape-non-whitespace.js)

• built-ins\RegExp(sourceFile: built-ins/RegExp/dotall/with-dotall-unicode.js)

• language\white-space(sourceFile: language/white-space/mongolian-vowel-separator-eval.js)

  • I think this has some changed spec behind it, if I recall right

[sebastienros]

I marked some of these explicitly in Jint for now. Some rules are not implemented in esprima, like verifying that []{}() are balances when the u flag is set. Will attempt to fix these too. Haven't checked the old Ecma ones yet, thanks.

[lahma]

I'll trust that you know what you are doing ;)