Function TestStdlibCompat_UnicodePatterns

stdlib_compat_test.go:1127–1172

(t *testing.T)

}

func TestStdlibCompat_UnicodePatterns(t *testing.T) {
	// Known differences: negated Unicode property classes may have different
	// boundary handling or empty match behavior
	unicodeKnownDiffs := map[string]string{
		`\P{Han}+`: "negated Unicode property class boundary handling differs",
	}

	patterns := []struct {
		pattern string
		input   string
	}{
		{`[日本語]+`, "日本語test日本語"},
		{`\p{Han}+`, "test中文test"},
		{`\P{Han}+`, "abc中文def"},
		{`[\x{4e00}-\x{9fff}]+`, "test中文test"},
	}

	for _, tc := range patterns {
		t.Run(tc.pattern, func(t *testing.T) {
			if reason, ok := unicodeKnownDiffs[tc.pattern]; ok {
				t.Skipf("Known difference: %s", reason)
			}

			stdRe, stdErr := regexp.Compile(tc.pattern)
			cgRe, cgErr := Compile(tc.pattern)

			// Both should succeed or both should fail
			if (stdErr == nil) != (cgErr == nil) {
				t.Errorf("Compile error mismatch:\n  stdlib: %v\n  coregex: %v", stdErr, cgErr)
				return
			}

			if stdErr != nil {
				return // Both failed, that's consistent
			}

			stdResult := stdRe.FindString(tc.input)
			cgResult := cgRe.FindString(tc.input)

			if stdResult != cgResult {
				t.Errorf("FindString(%q, %q):\n  stdlib: %q\n  coregex: %q",
					tc.pattern, tc.input, stdResult, cgResult)
			}
		})
	}
}

func TestStdlibCompat_LimitedFindAll(t *testing.T) {
	pattern := `\d+`