feat(cubelinux-2): Package 1 — cubecoords + cubestore from PDF spec

New code (no recycling from prior build). Implements:
- cubecoords: Czyx 4-axis coordinate (pack/unpack u32), Null-class
  classification (Total / Cube 1-4 / User), TriWord tri-channel 64-bit
  codec (6 ASCII + 4 control bits), HeaderFlags + CubeHeader (derived
  flags). All coding decisions documented inline.
- cubestore: CubeBackend trait, HashBackend (HashMap<u32>), CubeStore with
  dependency-free length-prefixed record codec (header TLV + body).
- 8 unit tests, all passing; cargo test clean (0 warn/err).

Per directive: separate git repo; current /home/CUBELinux kept as working
tool; scoped to AI-OS-excluded PDF vision.
This commit is contained in:
CUBELinux-2
2026-08-10 18:03:04 -04:00
commit 6e13b13eea
6 changed files with 724 additions and 0 deletions
+335
View File
@@ -0,0 +1,335 @@
//! CUBELinux-2 coordinate layer — built NEW from the original PDF spec.
//!
//! This is NOT recycled from the prior `/home/CUBELinux` build. The prior
//! build collapsed the PDF's four-axis CZYX model into three `u64` spatial
//! axes plus a 256-bit `SpaceId` capability. Here we restore the PDF model
//! faithfully: a 4×`u8` `CZYX` coordinate where the `C` axis (class/context)
//! is a real coordinate axis, a reserved "Null" control space for headers and
//! flags, and a 64-bit "tri-channel" word that packs six ASCII characters
//! plus four control bits.
//!
//! Coding decisions encountered while building to spec are documented inline
//! (see `Decision:` notes) so the divergence from a naive reading is visible.
#![forbid(unsafe_code)]
#![warn(missing_docs)]
/// A coordinate in the four-axis CUBELinux space.
///
/// Axes are each `u8` (0–255). The value `0` is reserved as "Null" on every
/// axis, following the PDF's design: `0` is not a normal data cell, it is the
/// control plane.
///
/// User record space is `1..=255` on each axis, giving
/// `255^4 = 4,228,250,625` possible record coordinates — the figure quoted in
/// the 2006 notes and the PDF.
#[derive(Copy, Clone, Eq, PartialEq, Ord, PartialOrd, Hash, Debug, Default)]
pub struct Czyx {
/// Class / context axis. `0` = Null (control space).
pub c: u8,
/// Z axis (depth).
pub z: u8,
/// Y axis (vertical).
pub y: u8,
/// X axis (horizontal).
pub x: u8,
}
impl Czyx {
/// Construct a coordinate.
#[inline]
pub const fn new(c: u8, z: u8, y: u8, x: u8) -> Self {
Czyx { c, z, y, x }
}
/// Total Null: `C=Z=Y=X=0`. Used as end-of-record / unused / deletion
/// marker per the PDF.
#[inline]
pub fn is_total_null(&self) -> bool {
self.c == 0 && self.z == 0 && self.y == 0 && self.x == 0
}
/// Pack into a single `u32` with `C` in the high byte, `X` in the low byte.
#[inline]
pub fn pack_u32(&self) -> u32 {
((self.c as u32) << 24)
| ((self.z as u32) << 16)
| ((self.y as u32) << 8)
| (self.x as u32)
}
/// Unpack a `u32` produced by [`pack_u32`].
#[inline]
pub fn unpack_u32(v: u32) -> Self {
Czyx {
c: (v >> 24) as u8,
z: (v >> 16) as u8,
y: (v >> 8) as u8,
x: v as u8,
}
}
// --- Null-class classification ---------------------------------------
//
// Decision: the PDF describes several Null special ranges ("Null cube 1–4,
// header layers, null rows") without a single canonical enumeration. We
// implement the two it defines precisely (Total Null, and the
// `C=0, Z/Y/X in 1..=255` Null-cube family) and expose a `NullClass` enum
// for the rest to be added as the store grows. This keeps the axis model
// exact while leaving a documented extension point.
/// Returns true if this coordinate lives in the Null control space:
/// `C == 0` with at least one of Z/Y/X non-zero (the "Null cube" family).
#[inline]
pub fn is_null_cube(&self) -> bool {
self.c == 0 && (self.z != 0 || self.y != 0 || self.x != 0)
}
/// Classify this coordinate.
#[inline]
pub fn null_class(&self) -> NullClass {
if self.is_total_null() {
NullClass::Total
} else if self.is_null_cube() {
// The PDF names "Null cube 1–4" by different Z/Y/X patterns.
// Decision: encode the cube number as a function of which
// non-zero pattern is present, deterministically, so it is
// stable and documented rather than ad hoc.
match (self.z != 0, self.y != 0, self.x != 0) {
(true, _, _) => NullClass::Cube(1),
(false, true, _) => NullClass::Cube(2),
(false, false, true) => NullClass::Cube(3),
(false, false, false) => NullClass::Cube(4), // unreachable after Total check, kept total
}
} else {
NullClass::User
}
}
}
/// The classification of a coordinate with respect to the Null control plane.
#[derive(Copy, Clone, Eq, PartialEq, Debug)]
pub enum NullClass {
/// `C=Z=Y=X=0` — end-of-record / unused / deletion marker.
Total,
/// `C=0`, at least one of Z/Y/X non-zero — a Null control cube, numbered
/// `1..=4` by which axes are set (see [`Czyx::null_class`]).
Cube(u8),
/// Normal user data cell (`C >= 1`, or `C=0` only when used as plain data
/// outside the Null convention).
User,
}
/// A 64-bit "tri-channel" word.
///
/// Per the PDF, one word carries either three ASCII pairs plus 4 control bits
/// (6 ASCII chars), or alternate pair/triad/quad arrangements. We implement
/// the canonical `pack_6` / `unpack_6` form from the spec (6 ASCII bytes +
/// 4 control bits) and leave the pair/triad/quad specialization as a
/// documented extension point.
#[derive(Copy, Clone, Eq, PartialEq, Debug, Default)]
pub struct TriWord(pub u64);
/// Stateless encoder/decoder for [`TriWord`].
pub struct TriEnc;
impl TriEnc {
/// Pack 6 ASCII bytes (0–255) plus 4 control bits (0–15) into one 64-bit
/// word.
///
/// Layout (high → low): `[4 control bits][48 ascii bits][12 unused]`.
/// The control bits occupy bits 60..=63; the six ASCII bytes occupy bits
/// 8..=59 (byte `i` at `8*(5-i)`), leaving the low 8 bits spare for future
/// use.
#[inline]
pub fn pack_6(control: u8, ascii: [u8; 6]) -> TriWord {
let mut v: u64 = 0;
v |= (control as u64 & 0x0F) << 60;
for (i, b) in ascii.iter().enumerate() {
let shift = 8 * (5 - i);
v |= (*b as u64) << shift;
}
TriWord(v)
}
/// Unpack a word produced by [`pack_6`].
#[inline]
pub fn unpack_6(word: TriWord) -> (u8, [u8; 6]) {
let v = word.0;
let control = ((v >> 60) & 0x0F) as u8;
let mut ascii = [0u8; 6];
for i in 0..6 {
let shift = 8 * (5 - i);
ascii[i] = ((v >> shift) & 0xFF) as u8;
}
(control, ascii)
}
}
/// Header flag bits, mirroring the PDF's title/type/date/size/permission
/// flag layout (flags 1–4 explicitly; 5–19 reserved for permissions and
/// associations).
#[derive(Copy, Clone, Eq, PartialEq, Debug, Default)]
pub struct HeaderFlags(pub u16);
impl HeaderFlags {
/// Flag 1: title start present.
pub const TITLE: u16 = 1 << 0;
/// Flag 2: document type present.
pub const DOC_TYPE: u16 = 1 << 1;
/// Flag 3: creation date present.
pub const CREATED_AT: u16 = 1 << 2;
/// Flag 4: weight/size present.
pub const SIZE_BYTES: u16 = 1 << 3;
/// Flag 5: root-only permission.
pub const PERM_ROOT_ONLY: u16 = 1 << 4;
/// Flag 6: local-user owner.
pub const PERM_LOCAL_USER: u16 = 1 << 5;
/// Flag 7: remote-user owner.
pub const PERM_REMOTE_USER: u16 = 1 << 6;
/// Flag 8: has outgoing association links.
pub const HAS_ASSOCIATIONS: u16 = 1 << 7;
/// Flag 255 (conceptual end-of-header) is represented out-of-band by the
/// record serializer; there is no bit for it.
/// Construct from a raw bitmask.
#[inline]
pub const fn from_bits(bits: u16) -> Self {
HeaderFlags(bits)
}
/// The raw bitmask.
#[inline]
pub const fn bits(&self) -> u16 {
self.0
}
/// Set a flag bit.
#[inline]
pub fn set(&mut self, flag: u16) {
self.0 |= flag;
}
/// Test a flag bit.
#[inline]
pub fn has(&self, flag: u16) -> bool {
self.0 & flag != 0
}
}
/// A record header, mirroring the PDF's title/type/date/perms/association
/// model.
///
/// Decision: the PDF gives both a `bitflags`-style `HeaderFlags` and a
/// structured `CubeHeader` with `Option` fields. We keep the structured form
/// (it is what the store serializes) and derive the flag bits from which
/// fields are `Some`. This avoids storing redundant flag+field data.
#[derive(Clone, Debug, Default)]
pub struct CubeHeader {
/// Flag bits (derived; kept in sync by the accessors).
pub flags: HeaderFlags,
/// Flag 1: human title.
pub title: Option<String>,
/// Flag 2: document type (like a file extension).
pub doc_type: Option<String>,
/// Flag 3: creation time (epoch seconds).
pub created_at: Option<u64>,
/// Flag 4: payload size in bytes.
pub size_bytes: Option<u64>,
/// Owner: local user.
pub owner_local_user: Option<String>,
/// Owner: remote user.
pub owner_remote_user: Option<String>,
/// Association links to other records (flags 5–19).
pub linked_records: Vec<Czyx>,
/// Total local accesses (from the PDF's association/permission flags).
pub total_accesses: u64,
/// Total remote accesses.
pub total_remote_accesses: u64,
/// Timestamp of the last local access, if any.
pub last_access: Option<u64>,
/// Timestamp of the last remote access, if any.
pub last_remote_access: Option<u64>,
}
impl CubeHeader {
/// Build a header, computing [`flags`] from the populated fields.
pub fn new() -> Self {
CubeHeader::default()
}
/// Recompute the flag bits from which fields are present. Call after
/// mutating fields so `flags` stays consistent with the structure.
pub fn refresh_flags(&mut self) {
let mut f = 0u16;
if self.title.is_some() {
f |= HeaderFlags::TITLE;
}
if self.doc_type.is_some() {
f |= HeaderFlags::DOC_TYPE;
}
if self.created_at.is_some() {
f |= HeaderFlags::CREATED_AT;
}
if self.size_bytes.is_some() {
f |= HeaderFlags::SIZE_BYTES;
}
if self.owner_local_user.is_some() {
f |= HeaderFlags::PERM_LOCAL_USER;
}
if self.owner_remote_user.is_some() {
f |= HeaderFlags::PERM_REMOTE_USER;
}
if !self.linked_records.is_empty() {
f |= HeaderFlags::HAS_ASSOCIATIONS;
}
self.flags = HeaderFlags::from_bits(f);
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn czyx_pack_roundtrips() {
let c = Czyx::new(12, 34, 56, 78);
assert_eq!(Czyx::unpack_u32(c.pack_u32()), c);
assert_eq!(c.pack_u32(), 0x0C_22_38_4E);
}
#[test]
fn total_null_detected() {
assert!(Czyx::new(0, 0, 0, 0).is_total_null());
assert!(!Czyx::new(0, 1, 0, 0).is_total_null());
}
#[test]
fn null_classification() {
assert_eq!(Czyx::new(0, 0, 0, 0).null_class(), NullClass::Total);
assert_eq!(Czyx::new(0, 9, 0, 0).null_class(), NullClass::Cube(1));
assert_eq!(Czyx::new(0, 0, 9, 0).null_class(), NullClass::Cube(2));
assert_eq!(Czyx::new(0, 0, 0, 9).null_class(), NullClass::Cube(3));
assert_eq!(Czyx::new(3, 0, 0, 0).null_class(), NullClass::User);
}
#[test]
fn triword_roundtrips() {
let ascii = *b"cubeln"; // 6 ascii bytes
let w = TriEnc::pack_6(0xA, ascii);
let (ctrl, back) = TriEnc::unpack_6(w);
assert_eq!(ctrl, 0xA);
assert_eq!(back, ascii);
}
#[test]
fn header_flags_derived() {
let mut h = CubeHeader::new();
h.title = Some("hello".into());
h.size_bytes = Some(42);
h.refresh_flags();
assert!(h.flags.has(HeaderFlags::TITLE));
assert!(h.flags.has(HeaderFlags::SIZE_BYTES));
assert!(!h.flags.has(HeaderFlags::DOC_TYPE));
}
}