Description

The . metacharacter incorrectly matches individual bytes instead of UTF-8 codepoints on multibyte input.

Reproduction

package main

import (
    "fmt"
    "regexp"
    "github.com/coregx/coregex"
)

func main() {
    input := "日本語"  // 9 bytes, 3 codepoints
    
    stdRe := regexp.MustCompile(`.`)
    cgRe := coregex.MustCompile(`.`)
    
    fmt.Printf("stdlib:  %d matches %v\n", len(stdRe.FindAllString(input, -1)), stdRe.FindAllString(input, -1))
    fmt.Printf("coregex: %d matches %v\n", len(cgRe.FindAllString(input, -1)), cgRe.FindAllString(input, -1))
}

Expected (stdlib)

stdlib:  3 matches [日 本 語]

Actual (coregex)

coregex: 9 matches [� � � � � � � � �]

Impact

Critical - Affects ALL patterns using . on non-ASCII text:

• .+, .*, .?
• (.) capture groups
• Any pattern with dot metacharacter

Affected Patterns

• . - single dot
• .+, .*, .? - dot with quantifiers
• (.) - dot in capture group
• \D, \W, \S - may also be affected (negated classes)

Root Cause

Likely in NFA/DFA handling of OpAnyCharNotNL - matching byte-by-byte instead of rune-by-rune.

Priority

🔴 Critical - Breaks Unicode compatibility, blocking v1.0.0