Improved Character Range and Special Sequence Support

[cdwmhcc]

🆒 Character Range Issues

Description	Current	Generated Pattern	Expected Pattern
Number Range	charIn('1-9')	/[1\-9]/	/[1-9]/
Alternatives	charIn('123456789')	/[123456789]/	/[1-9]/
Ideal API	charIn('1-9')	n/a	/[1-9]/

Whitespace Character Class Issues

Description	Current	Generated Pattern	Expected Pattern
Escaped \s in String	charIn('abc\\s')	/[abc\\s]/	/[abc\s]/
Alternatives	charIn('abc').or(whitespace)	/(?:[abc]|\s)/	/[abc\s]/
Ideal API Option 1	charIn('abc\\s')	n/a	/[abc\s]/
Ideal API Option 2	charIn('abc${whitespace}')	n/a	/[abc\s]/

Complex Lookbehind or lookahead Structure Issues

Description	Current	Generated Pattern	Expected Pattern
Lookbehind	exactly('').after(anyOf(exactly('').at.lineStart(), charIn('-_(:'))	/(?<=(?:^|[\-_(:]))/	/(?<=(?:^|[-_(:]))/
Ideal API	after(anyOf(lineStart, charIn('-_(:'))	n/a	/(?<=(?:^|[-_(:]))/

ℹ️ Additional info

1. Character Range Interpretation:

   • The library interprets '1-9' literally as the characters "1", "-", and "9" instead of the range from 1 to 9
   • Proper character ranges need to be enumerated manually

2. Escaped Character Handling:

   • Escape sequences like \\s in strings are not correctly translated to regex character classes
   • The library creates unnecessary alternation when combining regular characters with special classes

Suggested Improvements

1. Implement proper character range parsing in charIn(): - between two characters should create a range
2. Support proper escape sequence handling in character classes
3. Introduce more concise helper functions for common patterns (e.g., lineStart, after)

[danielroe]

what do you think of the implementation in https://github.com/unjs/magic-regexp/pull/399?

[cdwmhcc]

what do you think of the implementation in #399?

Thanks for pointing me to PR #399. I initially misunderstood issue #397, thinking it was language-specific. After taking a closer look, I see that it addresses the same Character Range Issues I mentioned in my issue. Having reviewed PR #399, I can confirm that its implementation would indeed solve the Character Range Issues portion of my issue. I appreciate you making this connection.