Function TestIsVisibleCanonicalList

coderd/x/chatd/sanitize_test.go:279–327

(t *testing.T)

}

func TestIsVisibleCanonicalList(t *testing.T) {
	t.Parallel()

	// Canonical list — must match site/src/utils/invisibleUnicode.test.ts
	//
	// Every codepoint that isVisible returns false for is listed
	// here, with ranges expanded to individual values. If a
	// codepoint is added or removed, this test must be updated.
	stripped := []rune{
		0x00AD,
		0x034F,
		0x061C,
		0x180E,
		0x200B,
		// 0x200C (ZWNJ) deliberately NOT stripped.
		0x200D,
		0x200E,
		0x200F,
		0x202A, 0x202B, 0x202C, 0x202D, 0x202E,
		0x2060, 0x2061, 0x2062, 0x2063, 0x2064,
		0x2066, 0x2067, 0x2068, 0x2069,
		0x206A, 0x206B, 0x206C, 0x206D, 0x206E, 0x206F,
		0xFEFF,
		0xFFF9, 0xFFFA, 0xFFFB,
	}

	for _, r := range stripped {
		input := "a" + string(r) + "b"
		got := chatd.SanitizePromptText(input)
		require.Equalf(t, "ab", got, "U+%04X should be stripped", r)
	}

	// Codepoints that must NOT be stripped.
	preserved := []rune{
		'A',     // Normal ASCII.
		'z',     // Normal ASCII.
		'0',     // Digit.
		' ',     // Space.
		0x200C,  // ZWNJ — required for Persian/Urdu/Kurdish.
		0xE0067, // Tag character — used in subdivision flag emoji.
	}

	for _, r := range preserved {
		input := "a" + string(r) + "b"
		want := "a" + string(r) + "b"
		got := chatd.SanitizePromptText(input)
		require.Equalf(t, want, got, "U+%04X should be preserved", r)
	}
}