proof-format 1.0.0

Canonical CBOR encode/decode, schema validation, and versioned parsers for Proof Engine
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
// Copyright 2026 Proof Engine Contributors
// SPDX-License-Identifier: MIT OR Apache-2.0
//! Minimal deterministic CBOR codec for the V0.1 subset.
//! Encoder always emits preferred serialization + lexicographic map order.
//! Decoder rejects everything outside the subset (fail closed).

use proof_core::{ErrorCode, Limits, ProofError};

/// Closed CBOR value set for V0.1.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum CborValue {
    Uint(u64),
    Nint(i64), // -2^64..=-1
    Text(String),
    Bytes(Vec<u8>),
    Array(Vec<CborValue>),
    Map(Vec<(CborValue, CborValue)>),
    Bool(bool),
    Null,
}

impl CborValue {
    fn sort_maps(&mut self) {
        match self {
            Self::Array(items) => {
                for i in items.iter_mut() {
                    i.sort_maps();
                }
            }
            Self::Map(pairs) => {
                for (k, v) in pairs.iter_mut() {
                    k.sort_maps();
                    v.sort_maps();
                }
                pairs.sort_by_key(|a| encode_canonical(&a.0));
            }
            _ => {}
        }
    }
}

// ---------- encoder ----------

fn encode_head(major: u8, n: u64, out: &mut Vec<u8>) {
    debug_assert!(major < 8);
    let prefix = major << 5;
    if n < 24 {
        out.push(prefix | n as u8);
    } else if n <= u8::MAX as u64 {
        out.push(prefix | 24);
        out.push(n as u8);
    } else if n <= u16::MAX as u64 {
        out.push(prefix | 25);
        out.extend_from_slice(&(n as u16).to_be_bytes());
    } else if n <= u32::MAX as u64 {
        out.push(prefix | 26);
        out.extend_from_slice(&(n as u32).to_be_bytes());
    } else {
        out.push(prefix | 27);
        out.extend_from_slice(&n.to_be_bytes());
    }
}

// PE-FMT-001 (lexicographic maps) · PE-FMT-002 (shortest ints).
fn encode_value(v: &CborValue, out: &mut Vec<u8>) {
    match v {
        CborValue::Uint(n) => encode_head(0, *n, out),
        CborValue::Nint(n) => {
            // Invariant: CBOR nint is always <= -1. Callers must uphold it
            // (schema layer never constructs non-negative Nint).
            debug_assert!(*n < 0, "CborValue::Nint must be negative");
            // value = -1 - n  =>  n_arg = (-1 - value) as u64
            let arg = (-1i128 - (*n as i128)) as u64;
            encode_head(1, arg, out);
        }
        CborValue::Bytes(b) => {
            encode_head(2, b.len() as u64, out);
            out.extend_from_slice(b);
        }
        CborValue::Text(s) => {
            encode_head(3, s.len() as u64, out);
            out.extend_from_slice(s.as_bytes());
        }
        CborValue::Array(items) => {
            encode_head(4, items.len() as u64, out);
            for i in items {
                encode_value(i, out);
            }
        }
        CborValue::Map(pairs) => {
            // Deterministic: sort by encoded key bytes.
            let mut enc_keys: Vec<(Vec<u8>, &CborValue, &CborValue)> = pairs
                .iter()
                .map(|(k, val)| {
                    let mut kb = Vec::new();
                    encode_value(k, &mut kb);
                    (kb, k, val)
                })
                .collect();
            enc_keys.sort_by(|a, b| a.0.cmp(&b.0));
            encode_head(5, pairs.len() as u64, out);
            for (kb, _, val) in enc_keys {
                out.extend_from_slice(&kb);
                encode_value(val, out);
            }
        }
        CborValue::Bool(true) => out.push(0xf5),
        CborValue::Bool(false) => out.push(0xf4),
        CborValue::Null => out.push(0xf6),
    }
}

/// Encode deterministically (preferred serialization + lexicographic maps).
pub fn encode_canonical(v: &CborValue) -> Vec<u8> {
    let mut out = Vec::new();
    encode_value(v, &mut out);
    out
}

// ---------- decoder ----------

struct Decoder<'a> {
    buf: &'a [u8],
    pos: usize,
    depth: usize,
    limits: Limits,
}

impl<'a> Decoder<'a> {
    fn eof(&self) -> bool {
        self.pos >= self.buf.len()
    }

    fn read(&mut self, n: usize) -> Result<&'a [u8], ProofError> {
        if self.pos.checked_add(n).is_none_or(|e| e > self.buf.len()) {
            return Err(ErrorCode::Malformed.err("truncated CBOR input"));
        }
        let s = &self.buf[self.pos..self.pos + n];
        self.pos += n;
        Ok(s)
    }

    fn read_arg(&mut self, ai: u8) -> Result<u64, ProofError> {
        match ai {
            n @ 0..=23 => Ok(n as u64),
            24 => {
                let b = self.read(1)?[0];
                // Shortest-form: values < 24 MUST use single-byte form.
                if b < 24 {
                    return Err(ErrorCode::NonCanonical.err("non-shortest int encoding (ai=24)"));
                }
                Ok(b as u64)
            }
            25 => {
                let b = self.read(2)?;
                let v = u16::from_be_bytes([b[0], b[1]]) as u64;
                if v <= u8::MAX as u64 {
                    return Err(ErrorCode::NonCanonical.err("non-shortest int encoding (ai=25)"));
                }
                Ok(v)
            }
            26 => {
                let b = self.read(4)?;
                let v = u32::from_be_bytes([b[0], b[1], b[2], b[3]]) as u64;
                if v <= u16::MAX as u64 {
                    return Err(ErrorCode::NonCanonical.err("non-shortest int encoding (ai=26)"));
                }
                Ok(v)
            }
            27 => {
                let b = self.read(8)?;
                let v = u64::from_be_bytes([b[0], b[1], b[2], b[3], b[4], b[5], b[6], b[7]]);
                if v <= u32::MAX as u64 {
                    return Err(ErrorCode::NonCanonical.err("non-shortest int encoding (ai=27)"));
                }
                Ok(v)
            }
            28..=30 => Err(ErrorCode::ForbiddenCborConstruct.err("reserved additional info")),
            31 => Err(ErrorCode::ForbiddenCborConstruct.err("indefinite length forbidden")),
            _ => unreachable!(),
        }
    }

    // PE-FMT-003 (dup keys) · PE-FMT-004 (forbidden) · PE-FMT-005 (trailing/UTF-8) · PE-FMT-006 (limits).
    fn decode_value(&mut self) -> Result<CborValue, ProofError> {
        if self.depth > self.limits.max_depth {
            return Err(ErrorCode::LimitExceeded.err("max CBOR depth exceeded"));
        }
        if self.eof() {
            return Err(ErrorCode::Malformed.err("empty CBOR input"));
        }
        let ib = self.read(1)?[0];
        let major = ib >> 5;
        let ai = ib & 0x1f;
        match major {
            0 => {
                let n = self.read_arg(ai)?;
                Ok(CborValue::Uint(n))
            }
            1 => {
                let n = self.read_arg(ai)?;
                // nint range check: arg <= u64::MAX always fits i64 -1-n? -1-2^64 < i64::MIN.
                // CBOR nint covers -2^64..-1; we restrict to i64 (-2^63..-1) for V0.1? No:
                // allow full range but store as i64 only if fits, else reject (no bignum path).
                let v = -1i128 - (n as i128);
                if v < i64::MIN as i128 {
                    return Err(ErrorCode::ForbiddenCborConstruct
                        .err("nint out of i64 range (bignum not supported in V0.1)"));
                }
                Ok(CborValue::Nint(v as i64))
            }
            2 => {
                let len = self.read_arg(ai)? as usize;
                if len > self.limits.max_field_size {
                    return Err(ErrorCode::LimitExceeded.err("bytes field too large"));
                }
                Ok(CborValue::Bytes(self.read(len)?.to_vec()))
            }
            3 => {
                let len = self.read_arg(ai)? as usize;
                if len > self.limits.max_field_size {
                    return Err(ErrorCode::LimitExceeded.err("text field too large"));
                }
                let b = self.read(len)?;
                let s = std::str::from_utf8(b)
                    .map_err(|_| ErrorCode::Malformed.err("invalid UTF-8 in text string"))?;
                Ok(CborValue::Text(s.to_owned()))
            }
            4 => {
                let len = self.read_arg(ai)? as usize;
                if len > self.limits.max_array_items {
                    return Err(ErrorCode::LimitExceeded.err("array too large"));
                }
                self.depth += 1;
                let mut items = Vec::with_capacity(len.min(32));
                for _ in 0..len {
                    items.push(self.decode_value()?);
                }
                self.depth -= 1;
                Ok(CborValue::Array(items))
            }
            5 => {
                let len = self.read_arg(ai)? as usize;
                if len > self.limits.max_map_entries {
                    return Err(ErrorCode::LimitExceeded.err("map too large"));
                }
                self.depth += 1;
                let mut pairs = Vec::with_capacity(len.min(32));
                let mut prev_key_bytes: Option<Vec<u8>> = None;
                for _ in 0..len {
                    let k = self.decode_value()?;
                    let val = self.decode_value()?;
                    // Keys must be valid deterministic encodings; re-encode key for ordering check.
                    let kb = encode_canonical(&k);
                    if let Some(prev) = &prev_key_bytes {
                        if kb.as_slice() == prev.as_slice() {
                            return Err(ErrorCode::DuplicateMapKey.err("duplicate map key"));
                        }
                        if kb.as_slice() < prev.as_slice() {
                            return Err(
                                ErrorCode::NonCanonical.err("map keys not in lexicographic order")
                            );
                        }
                    }
                    prev_key_bytes = Some(kb);
                    pairs.push((k, val));
                }
                self.depth -= 1;
                Ok(CborValue::Map(pairs))
            }
            6 => Err(ErrorCode::ForbiddenCborConstruct.err("tags forbidden in V0.1")),
            7 => match ai {
                20 => Ok(CborValue::Bool(false)),
                21 => Ok(CborValue::Bool(true)),
                22 => Ok(CborValue::Null),
                24 => Err(ErrorCode::ForbiddenCborConstruct.err("2-byte simple values forbidden")),
                25..=27 => Err(ErrorCode::ForbiddenCborConstruct.err("floats forbidden in V0.1")),
                31 => {
                    Err(ErrorCode::ForbiddenCborConstruct.err("break stop code outside indefinite"))
                }
                _ => {
                    Err(ErrorCode::ForbiddenCborConstruct.err("simple/undefined values forbidden"))
                }
            },
            _ => unreachable!(),
        }
    }
}

/// Strict decode: full input consumed, limits enforced, canonical order checked.
pub fn decode_strict(bytes: &[u8], limits: &Limits) -> Result<CborValue, ProofError> {
    if bytes.len() > limits.max_proof_size {
        return Err(ErrorCode::LimitExceeded.err("proof bytes exceed max_proof_size"));
    }
    let mut d = Decoder {
        buf: bytes,
        pos: 0,
        depth: 0,
        limits: *limits,
    };
    let v = d.decode_value()?;
    if d.pos != bytes.len() {
        return Err(ErrorCode::Malformed.err("trailing bytes after CBOR item"));
    }
    Ok(v)
}

/// True iff `bytes` is exactly the deterministic encoding of its decoded value.
pub fn is_canonical(bytes: &[u8], limits: &Limits) -> Result<bool, ProofError> {
    let v = decode_strict(bytes, limits)?;
    Ok(encode_canonical(&v) == bytes)
}

/// Decode + re-encode check. Returns value or NON_CANONICAL.
// PE-FMT-007 (decode->re-encode identity).
pub fn decode_and_check_canonical(bytes: &[u8], limits: &Limits) -> Result<CborValue, ProofError> {
    let v = decode_strict(bytes, limits)?;
    if encode_canonical(&v) != bytes {
        return Err(ErrorCode::NonCanonical.err("bytes are not deterministic encoding"));
    }
    Ok(v)
}

#[allow(dead_code)]
pub fn sorted_map(mut pairs: Vec<(CborValue, CborValue)>) -> CborValue {
    let mut v = CborValue::Map(std::mem::take(&mut pairs));
    v.sort_maps();
    v
}

#[cfg(test)]
// PE-FMT-008 · PE-FMT-009 · PE-CRYPTO-009: ids bind the exact UTF-8 bytes as
/// given — no Unicode normalization/folding anywhere (FORMAT.md §3.1). Tests:
/// `text_bytes_as_given_no_normalization`, `nfc_nfd_are_distinct_ids`,
/// `golden_20_text_byte_identity`.
mod tests {
    use super::*;
    use proof_core::{ErrorCode, Limits};

    fn lim() -> Limits {
        Limits::default()
    }

    #[test]
    fn round_trip_primitives() {
        let vs = vec![
            CborValue::Uint(0),
            CborValue::Uint(23),
            CborValue::Uint(24),
            CborValue::Uint(1000),
            CborValue::Uint(u64::MAX),
            CborValue::Nint(-1),
            CborValue::Nint(-1000),
            CborValue::Text("hello".into()),
            CborValue::Text("".into()),
            CborValue::Bytes(vec![0, 1, 2, 255]),
            CborValue::Bool(true),
            CborValue::Bool(false),
            CborValue::Null,
            CborValue::Array(vec![CborValue::Uint(1), CborValue::Text("a".into())]),
        ];
        for v in vs {
            let b = encode_canonical(&v);
            let d = decode_strict(&b, &lim()).expect("decode");
            assert_eq!(d, v);
            assert_eq!(encode_canonical(&d), b);
        }
    }

    #[test]
    fn map_keys_sorted_lexicographically() {
        // Insert out of order; encoder must sort by encoded bytes.
        let v = CborValue::Map(vec![
            (CborValue::Text("z".into()), CborValue::Uint(1)),
            (CborValue::Text("aa".into()), CborValue::Uint(2)),
            (CborValue::Uint(10), CborValue::Uint(3)),
        ]);
        let b = encode_canonical(&v);
        let d = decode_strict(&b, &lim()).expect("decode");
        // Expected lexicographic byte order: 0x0a (10) < 0x617a ("z") < 0x626161 ("aa").
        match d {
            CborValue::Map(p) => {
                assert_eq!(p[0].0, CborValue::Uint(10));
                assert_eq!(p[1].0, CborValue::Text("z".into()));
                assert_eq!(p[2].0, CborValue::Text("aa".into()));
            }
            _ => panic!("expected map"),
        }
    }

    #[test]
    fn duplicate_map_key_rejected() {
        // {"a":1,"a":2} hand-encoded: a2 61 61 01 61 61 02.
        let raw = hex::decode("a2616101616102").unwrap();
        let e = decode_strict(&raw, &lim()).unwrap_err();
        assert_eq!(e.code, ErrorCode::DuplicateMapKey);
    }

    #[test]
    fn non_shortest_int_rejected() {
        // 0 encoded as 0x1800 (two bytes) instead of 0x00.
        let e = decode_strict(&[0x18, 0x00], &lim()).unwrap_err();
        assert_eq!(e.code, ErrorCode::NonCanonical);
        // 24 encoded as 0x190018 instead of 0x1818.
        let e = decode_strict(&[0x19, 0x00, 0x18], &lim()).unwrap_err();
        assert_eq!(e.code, ErrorCode::NonCanonical);
    }

    #[test]
    fn forbidden_constructs_rejected() {
        // tag 0 (0xc0 + "2020-01-01"), float16 (0xf9), indefinite array (0x9f), break (0xff),
        // undefined (0xf7), reserved ai 28 (0x1c).
        let cases: &[&[u8]] = &[
            &[
                0xc0, 0x6a, 0x32, 0x30, 0x32, 0x30, 0x2d, 0x30, 0x31, 0x2d, 0x30, 0x31,
            ],
            &[0xf9, 0x3c, 0x00],
            &[0x9f, 0x01, 0xff],
            &[0xff],
            &[0xf7],
            &[0x1c, 0x00],
            &[0xfa, 0x47, 0xc3, 0x50, 0x00], // float32
        ];
        for c in cases {
            let e = decode_strict(c, &lim()).unwrap_err();
            assert_eq!(
                e.code,
                ErrorCode::ForbiddenCborConstruct,
                "input {:02x?}",
                c
            );
        }
    }

    #[test]
    fn trailing_bytes_and_empty_rejected() {
        assert_eq!(
            decode_strict(&[0x01, 0x02], &lim()).unwrap_err().code,
            ErrorCode::Malformed
        );
        assert_eq!(
            decode_strict(&[], &lim()).unwrap_err().code,
            ErrorCode::Malformed
        );
    }

    #[test]
    fn invalid_utf8_rejected() {
        // text(1) with 0xff.
        let e = decode_strict(&[0x61, 0xff], &lim()).unwrap_err();
        assert_eq!(e.code, ErrorCode::Malformed);
    }

    // PE-FMT-009 (FORMAT.md §3.1): ids bind exact bytes; no normalization.
    #[test]
    fn text_bytes_as_given_no_normalization() {
        // NFC spelling must round-trip byte-exactly: no folding, no trimming.
        let v = decode_strict(
            &encode_canonical(&CborValue::Text("caf\u{e9}".into())),
            &lim(),
        )
        .unwrap();
        assert_eq!(v, CborValue::Text("caf\u{e9}".into()));
        // Whitespace and case are preserved verbatim too.
        let v = decode_strict(
            &encode_canonical(&CborValue::Text("  MixedCase ".into())),
            &lim(),
        )
        .unwrap();
        assert_eq!(v, CborValue::Text("  MixedCase ".into()));
    }

    #[test]
    fn nfc_nfd_are_distinct_ids() {
        // The two byte spellings of the same logical string are different
        // canonical texts -> different ids (FORMAT.md §3.1; golden-20).
        let nfc_bytes = "caf\u{e9}".as_bytes(); // 63 61 66 C3 A9
        let nfd_bytes = "cafe\u{301}".as_bytes(); // 63 61 66 65 CC 81
        assert_ne!(nfc_bytes, nfd_bytes);
        let nfc = encode_canonical(&CborValue::Text("caf\u{e9}".into()));
        let nfd = encode_canonical(&CborValue::Text("cafe\u{301}".into()));
        assert_ne!(nfc, nfd, "engine never folds; distinct bytes stay distinct");
        // Distinct canonical bytes must yield distinct ids (byte-binding).
        let id_nfc = proof_crypto_id_shim(&nfc);
        let id_nfd = proof_crypto_id_shim(&nfd);
        assert_ne!(id_nfc, id_nfd);
        // But each id verifies against exactly its own bytes.
        assert!(verify_spelling(&id_nfc, &nfc));
        assert!(verify_spelling(&id_nfd, &nfd));
    }

    /// Local stand-ins so the format crate does not depend on proof-crypto:
    /// replicate the id construction `<prefix>:v1:<b64u(sha256(bytes))>` and
    /// its byte-binding check to assert the id-level consequence.
    fn proof_crypto_id_shim(bytes: &[u8]) -> String {
        use base64::engine::general_purpose::URL_SAFE_NO_PAD;
        use base64::Engine as _;
        use sha2::{Digest, Sha256};
        format!("evt:v1:{}", URL_SAFE_NO_PAD.encode(Sha256::digest(bytes)))
    }

    fn verify_spelling(id: &str, bytes: &[u8]) -> bool {
        proof_crypto_id_shim(bytes) == id
    }

    #[test]
    fn golden_20_text_byte_identity() {
        // Interop vector: exact canonical bytes and derived id for NFC input.
        // Golden-20 (fixtures/golden-20.json) pins these bytes; re-derive the
        // id from the parsed text to prove byte fidelity end to end.
        let fixture =
            std::path::Path::new(env!("CARGO_MANIFEST_DIR")).join("../../fixtures/golden-20.json");
        let Ok(text) = std::fs::read_to_string(&fixture) else {
            // Fixture not present in this checkout: in-repo assertion above
            // (nfc_nfd_are_distinct_ids) still covers the property.
            return;
        };
        let v: serde_json::Value = serde_json::from_str(&text).unwrap();
        let bytes = hex::decode(v["text_canonical_hex"].as_str().unwrap()).unwrap();
        let nfd_bytes = hex::decode(v["text_canonical_hex_nfd"].as_str().unwrap()).unwrap();
        let decoded = decode_strict(&bytes, &lim()).unwrap();
        match decoded {
            CborValue::Text(s) => {
                assert_eq!(proof_crypto_id_shim(&bytes), v["evt_id"].as_str().unwrap());
                assert_eq!(encode_canonical(&CborValue::Text(s)), bytes);
                // The NFD spelling of the same word must never share the id.
                assert_ne!(
                    proof_crypto_id_shim(&nfd_bytes),
                    v["evt_id"].as_str().unwrap()
                );
            }
            other => panic!("golden-20 must be a text item, got {other:?}"),
        }
    }

    #[test]
    fn is_canonical_detects_reordered_map() {
        let v = CborValue::Map(vec![
            (CborValue::Uint(1), CborValue::Uint(1)),
            (CborValue::Uint(2), CborValue::Uint(2)),
        ]);
        let good = encode_canonical(&v);
        assert!(is_canonical(&good, &lim()).unwrap());
        // Manually swap pairs: still valid CBOR but non-canonical order.
        let bad = vec![0xa2, 0x02, 0x02, 0x01, 0x01];
        assert_eq!(
            decode_strict(&bad, &lim()).unwrap_err().code,
            ErrorCode::NonCanonical
        );
    }

    // Property: any decoded value re-encodes identically (idempotent canonical form),
    // and any encoded value decodes back to itself.
    #[test]
    fn prop_encode_decode_round_trip() {
        use proptest::prelude::*;
        use proptest::strategy::ValueTree;
        fn arb_value() -> impl Strategy<Value = CborValue> {
            let leaf = prop_oneof![
                any::<u64>().prop_map(CborValue::Uint),
                // CBOR nint is always <= -1; positive values are unrepresentable.
                (-1_000_000i64..0i64).prop_map(CborValue::Nint),
                "[a-z0-9]{0,8}".prop_map(CborValue::Text),
                proptest::collection::vec(any::<u8>(), 0..16).prop_map(CborValue::Bytes),
                proptest::bool::ANY.prop_map(CborValue::Bool),
                Just(CborValue::Null),
            ];
            leaf.prop_recursive(3, 16, 4, |inner| {
                prop_oneof![
                    proptest::collection::vec(inner.clone(), 0..4).prop_map(CborValue::Array),
                    proptest::collection::vec((inner.clone(), inner.clone()), 0..4).prop_map(
                        |pairs| {
                            // Deduplicate by encoded key to keep maps valid.
                            let mut seen = std::collections::HashSet::new();
                            let mut out = vec![];
                            for (k, v) in pairs {
                                let kb = encode_canonical(&k);
                                if seen.insert(kb) {
                                    out.push((k, v));
                                }
                            }
                            CborValue::Map(out)
                        }
                    ),
                ]
            })
        }
        let mut runner = proptest::test_runner::TestRunner::deterministic();
        for _ in 0..256 {
            let v = arb_value().new_tree(&mut runner).expect("tree").current();
            let b = encode_canonical(&v);
            let d = decode_strict(&b, &lim()).expect("must decode own encoding");
            assert_eq!(d, {
                let mut norm = v.clone();
                norm.sort_maps();
                // Encoder sorts maps; decoded form equals sorted original.
                norm
            });
            assert_eq!(encode_canonical(&d), b);
        }
    }

    #[test]
    fn prop_raw_bytes_reject_or_round_trip() {
        use proptest::prelude::*;
        use proptest::strategy::ValueTree;
        let mut runner = proptest::test_runner::TestRunner::deterministic();
        let strat = proptest::collection::vec(any::<u8>(), 0..48);
        for _ in 0..512 {
            let bytes: Vec<u8> = strat.new_tree(&mut runner).expect("tree").current();
            if let Ok(v) = decode_strict(&bytes, &lim()) {
                assert!(encode_canonical(&v).len() <= 1 << 20);
            }
        }
    }
}