`nx_uxf_cid.nx` source

1// nx_uxf_cid.nx -- UXF arc R1c: the SELF-DESCRIBING multicodec/profile CID, built ON TOP
2// of nx_canon_cid (do NOT touch the foundation -- Rule 19 additive, Rule 3 no patch cascade).
3//
4// The substrate spec (knowledge/specs/2026-06-09-tutoring-storage-substrate-rung1.md) defines
5// CID = multihash + a multicodec(record-type) tag, but the shipped cid_of() omits the
6// multicodec.  This adds it.  The profiled CID carries a PROFILE/codec id that is:
7//   (a) FOLDED INTO the hash  -> same bytes under different profiles => different CID
8//       (no cross-profile collision; still fully content-addressed within a profile), AND
9//   (b) READABLE from the CID string -> an endpoint can pick the right view WITHOUT any
10//       out-of-band schema  = Law-1 endpoint-adaptive ; new profiles = new codec ids with
11//       NO format change = Law-4 extensible (Rule 25: add an id, never a new format).
12//
13// Profiled CID: "nxc1-" <2hex codec> "-" <64hex sha256(codec_byte || canon_bytes)> NUL  (len 72)
14// No hardware/persistent writes (Rule 26).  license_tier: ORIGINAL
15import "nx_syscalls.nx"
16import "nx_sha256.nx"
17import "nx_canon_cid.nx"
18
19// UXF profile/codec ids (the multicodec tag) -- the data-driven extension point.
20const UXF_DATA:    i64 = 1
21const UXF_DOC:     i64 = 2
22const UXF_MEDIA:   i64 = 3
23const UXF_CONFIG:  i64 = 4
24const UXF_ARCHIVE: i64 = 5
25const UXF_SEARCH:  i64 = 6   // a search answer (search rung E8, 2026-09-14): query, corpus, hits -- JSON and text are its projections
26const UXF_CID_LEN: i64 = 72  // the profiled CID's width, derived: 5 (nxc1-) + 2 (codec hex) + 1 (-) + 64 (sha256 hex); callers size cid buffers UXF_CID_LEN + 1
27
28func uxf_nib_hex(v: i64) -> i64 {
29    if v > 9 { return 87 + v }   // a-f
30    return 48 + v                 // 0-9
31}
32
33func uxf_hex_nib(b: i64) -> i64 {
34    if b >= 48 { if b <= 57 { return b - 48 } }       // 0-9
35    if b >= 97 { if b <= 102 { return b - 87 } }      // a-f
36    if b >= 65 { if b <= 70 { return b - 55 } }       // A-F
37    return 0 - 1
38}
39
40// profiled CID = "nxc1-" + 2hex(codec) + "-" + 64hex(sha256(codec || canon)); writes NUL; returns 72.
41func uxf_cid_profiled(codec: i64, canon: *u8, n: i64, cid: *u8) -> i64 {
42    let tmp: *u8 = sys_mmap(n + 16)
43    tmp[0] = (codec & 0xff) as u8
44    var i: i64 = 0
45    while i < n { tmp[1 + i] = canon[i]; i = i + 1 }
46    let dg: *u8 = sys_mmap(40)
47    sha256_digest(tmp, n + 1, dg)
48    cid[0] = 110 as u8; cid[1] = 120 as u8; cid[2] = 99 as u8; cid[3] = 49 as u8; cid[4] = 45 as u8
49    cid[5] = uxf_nib_hex((codec >> 4) & 15) as u8
50    cid[6] = uxf_nib_hex(codec & 15) as u8
51    cid[7] = 45 as u8
52    i = 0
53    while i < 32 {
54        let b: i64 = dg[i] & 0xff
55        cid[8 + i * 2] = uxf_nib_hex((b >> 4) & 15) as u8
56        cid[9 + i * 2] = uxf_nib_hex(b & 15) as u8
57        i = i + 1
58    }
59    cid[UXF_CID_LEN] = 0 as u8
60    return UXF_CID_LEN
61}
62
63// READ the profile/codec back out of a profiled CID (self-describing); (0-1) if malformed.
64func uxf_codec_of_cid(cid: *u8) -> i64 {
65    let hi: i64 = uxf_hex_nib(cid[5] as i64)
66    let lo: i64 = uxf_hex_nib(cid[6] as i64)
67    if hi < 0 { return 0 - 1 }
68    if lo < 0 { return 0 - 1 }
69    return (hi * 16) + lo
70}
nx_uxf_cid.nx source

`nx_uxf_cid.nx` source
