codlet_core/code/
normalize.rs

1//! Code normalization (RFC-003 FR-2, INV-4).
2//!
3//! Normalization must be **identical** on the issue path and the redeem path,
4//! and **idempotent** (`normalize(normalize(x)) == normalize(x)`), or valid
5//! codes fail to match their stored lookup key.
6//!
7//! ## Compatibility note
8//!
9//! Normalization strips ASCII whitespace and
10//! hyphens and uppercases ASCII letters — and **nothing else**. In particular
11//! it does *not* drop the visually ambiguous characters `0 1 O I L` (those are
12//! merely excluded from the generation *alphabet*, RFC-003 §4). codlet
13//! reproduces that exact behavior so existing service codes keep matching. The
14//! ambiguity handling lives in [`super::alphabet`], not here.
15
16/// Normalize raw code input into its canonical form: strip ASCII whitespace and
17/// `-`, uppercase ASCII letters, leave everything else untouched.
18///
19/// This never panics on arbitrary Unicode input. Non-ASCII characters are
20/// preserved here; validation (RFC-003 FR-2) is responsible for rejecting them.
21#[must_use]
22pub fn normalize(raw: &str) -> String {
23    raw.chars()
24        .filter(|c| !c.is_whitespace() && *c != '-')
25        .map(|c| c.to_ascii_uppercase())
26        .collect()
27}
28
29#[cfg(test)]
30mod tests {
31    use super::*;
32
33    #[test]
34    fn strips_separators_and_uppercases() {
35        assert_eq!(normalize("x7-y9 z2"), "X7Y9Z2");
36        assert_eq!(normalize("X7Y9Z2"), "X7Y9Z2");
37        assert_eq!(normalize("  a b - c "), "ABC");
38    }
39
40    #[test]
41    fn does_not_drop_ambiguous_characters() {
42        // Compatibility guard: normalization must NOT remove 0/1/O/I/L
43        // (contrast with the generation alphabet, which excludes them).
44        assert_eq!(normalize("o1il0"), "O1IL0");
45    }
46
47    #[test]
48    fn idempotent() {
49        for s in ["X7-Y9 Z2", "abc", "  ", "Ünïcödé", "a-b-c-1-2-3", ""] {
50            assert_eq!(
51                normalize(&normalize(s)),
52                normalize(s),
53                "not idempotent for {s:?}"
54            );
55        }
56    }
57
58    #[test]
59    fn empty_and_separator_only_become_empty() {
60        assert_eq!(normalize(""), "");
61        assert_eq!(normalize("  --  "), "");
62    }
63
64    #[test]
65    fn no_panic_on_arbitrary_unicode() {
66        // Spot-check a range of scalar values; full coverage is in the
67        // property test in the crate test suite.
68        for cp in [
69            0u32, 0x09, 0x20, 0x2d, 0x41, 0x7f, 0x80, 0xa0, 0x1f600, 0x10ffff,
70        ] {
71            if let Some(ch) = char::from_u32(cp) {
72                let s: String = core::iter::once(ch).collect();
73                let _ = normalize(&s);
74            }
75        }
76    }
77}
codlet_core/code/normalize.rs

codlet_core/code/
normalize.rs
