valise 0.1.4

Single-file retrieval capsule for local-first AI: text and vector search over one portable file
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
// This Source Code Form is subject to the terms of the Mozilla Public
// License, v. 2.0. If a copy of the MPL was not distributed with this
// file, You can obtain one at https://mozilla.org/MPL/2.0/.

//! `TimeIndexSegment` payload codec, spec §17 (magic `VLTI`).
//!
//! Naming: `_segment` (not `_codec`) because this file owns the
//! `TimeIndexSegment` payload *type* plus its wire codec. The per-entry
//! record `TimeIndexEntry` lives in the sibling `time_index.rs`. See the
//! `format/` file vocabulary rule in `SKILL.md`.
//!
//! v2 wire form (columnar):
//!
//!   MAGIC (4)
//!   VERSION (2) = 2
//!   PARTITION_FLAG (1) + PADDING_OR_CID (4)
//!   COUNT (u32, little-endian)
//!   PREVIOUS_ROOT (variable, optional SegmentRef)
//!   -- column blobs (only when COUNT > 0) --
//!   FRAME_ID_BYTES (u32 len + bytes)     unsigned-LEB128 deltas
//!   CREATED_BYTES  (u32 len + bytes)     zigzag-LEB128 delta-of-deltas, prefixed by raw i64 base
//!   UPDATED_BYTES  (u32 len + bytes)     zigzag-LEB128 of (updated_at - created_at) per entry
//!   COLLECTION_RUNS (u32 run_count + (collection_id u32, run_len varint) × runs)
//!
//! Rationale: in append-only ingest each `frame_id` is +1 from the
//! previous (delta = 1 → 1 byte LEB128). For ingest bursts within the
//! same second, `created_at` deltas are 0 → delta-of-delta is 0 → 1
//! byte LEB128. `updated_at == created_at` is the common case → 1 byte
//! zigzag-LEB128 zero. `collection_id` is typically one-hot per build,
//! so RLE collapses 522 k 4-byte fields to a single run record.
//!
//! v1 was a row-by-row dump of `(u64, u32, i64, i64) = 28 B/entry`. v2
//! is rejected by v1 readers (version mismatch) and v1 is rejected by
//! v2 readers — there is no on-disk compatibility, consistent with the
//! v0.1-alpha policy that lets us iterate the format freely.

use crate::{
    error::{Error, Result},
    format::{CollectionId, FrameId, segment::SegmentRef},
};

const MAGIC: [u8; 4] = *b"VLTI";
const VERSION: u16 = 2;

#[derive(Clone, Debug, Eq, PartialEq)]
pub(crate) struct TimeIndexSegment {
    /// Optional partition key. v1 always emits `None` (global index);
    /// readers must accept `Some(collection_id)` for v2 forward-compat.
    pub(crate) collection_id: Option<CollectionId>,
    pub(crate) entries: Vec<TimeIndexEntry>,
    pub(crate) previous_root: Option<SegmentRef>,
}

#[derive(Clone, Copy, Debug, Eq, PartialEq)]
pub(crate) struct TimeIndexEntry {
    pub(crate) frame_id: FrameId,
    pub(crate) collection_id: CollectionId,
    pub(crate) created_at: i64,
    pub(crate) updated_at: i64,
}

pub(crate) fn encode_time_index_segment(seg: &TimeIndexSegment) -> Result<Vec<u8>> {
    if !seg
        .entries
        .windows(2)
        .all(|w| w[0].created_at <= w[1].created_at)
    {
        return Err(Error::Format(
            "TimeIndexSegment entries must be sorted ascending by created_at".into(),
        ));
    }
    let mut buf = Vec::new();
    buf.extend_from_slice(&MAGIC);
    buf.extend_from_slice(&VERSION.to_le_bytes());
    match seg.collection_id {
        None => {
            buf.push(0u8);
            buf.extend_from_slice(&[0u8; 4]);
        }
        Some(cid) => {
            buf.push(1u8);
            buf.extend_from_slice(&cid.0.to_le_bytes());
        }
    }
    let count = u32::try_from(seg.entries.len())
        .map_err(|_| Error::Format("TimeIndexSegment entry count overflow".into()))?;
    buf.extend_from_slice(&count.to_le_bytes());
    write_optional_segment_ref(&mut buf, seg.previous_root);

    if seg.entries.is_empty() {
        return Ok(buf);
    }

    // --- frame_id column (deltas, unsigned LEB128) ---
    // The first frame_id is stored raw (8 bytes prepended inside the
    // blob); subsequent entries store (cur - prev) as a varint.
    let mut frame_bytes: Vec<u8> = Vec::with_capacity(seg.entries.len() + 8);
    frame_bytes.extend_from_slice(&seg.entries[0].frame_id.0.to_le_bytes());
    for w in seg.entries.windows(2) {
        let prev = w[0].frame_id.0;
        let cur = w[1].frame_id.0;
        if cur < prev {
            return Err(Error::Format(
                "TimeIndexSegment frame_ids must be monotone within the same created_at".into(),
            ));
        }
        write_uvarint(&mut frame_bytes, cur - prev);
    }
    write_blob(&mut buf, &frame_bytes)?;

    // --- created_at column (delta-of-delta, zigzag LEB128) ---
    // Layout: i64 first_created_at, varint first_delta (signed
    // 0-against), then zigzag deltas-of-deltas.
    let mut created_bytes: Vec<u8> = Vec::with_capacity(seg.entries.len() + 16);
    let first_created = seg.entries[0].created_at;
    created_bytes.extend_from_slice(&first_created.to_le_bytes());
    let mut prev_delta: i64 = 0;
    let mut prev_value: i64 = first_created;
    for entry in &seg.entries[1..] {
        let delta = entry
            .created_at
            .checked_sub(prev_value)
            .ok_or_else(|| Error::Format("TimeIndexSegment created_at delta overflow".into()))?;
        let dod = delta.checked_sub(prev_delta).ok_or_else(|| {
            Error::Format("TimeIndexSegment created_at delta-of-delta overflow".into())
        })?;
        write_ivarint(&mut created_bytes, dod);
        prev_value = entry.created_at;
        prev_delta = delta;
    }
    write_blob(&mut buf, &created_bytes)?;

    // --- updated_at column (zigzag LEB128 of (updated_at - created_at)) ---
    let mut updated_bytes: Vec<u8> = Vec::with_capacity(seg.entries.len());
    for entry in &seg.entries {
        let offset = entry
            .updated_at
            .checked_sub(entry.created_at)
            .ok_or_else(|| Error::Format("TimeIndexSegment updated/created overflow".into()))?;
        write_ivarint(&mut updated_bytes, offset);
    }
    write_blob(&mut buf, &updated_bytes)?;

    // --- collection_id column (RLE) ---
    let mut runs: Vec<(u32, u32)> = Vec::new();
    let mut iter = seg.entries.iter();
    if let Some(first) = iter.next() {
        let mut cur_cid = first.collection_id.0;
        let mut run_len: u32 = 1;
        for entry in iter {
            if entry.collection_id.0 == cur_cid {
                run_len = run_len
                    .checked_add(1)
                    .ok_or_else(|| Error::Format("TimeIndexSegment RLE run too long".into()))?;
            } else {
                runs.push((cur_cid, run_len));
                cur_cid = entry.collection_id.0;
                run_len = 1;
            }
        }
        runs.push((cur_cid, run_len));
    }
    let run_count = u32::try_from(runs.len())
        .map_err(|_| Error::Format("TimeIndexSegment run count overflow".into()))?;
    let mut coll_blob: Vec<u8> = Vec::with_capacity(runs.len() * 5);
    coll_blob.extend_from_slice(&run_count.to_le_bytes());
    for (cid, run_len) in &runs {
        coll_blob.extend_from_slice(&cid.to_le_bytes());
        write_uvarint(&mut coll_blob, u64::from(*run_len));
    }
    write_blob(&mut buf, &coll_blob)?;

    Ok(buf)
}

pub(crate) fn decode_time_index_segment(bytes: &[u8]) -> Result<TimeIndexSegment> {
    let mut cur = Cursor::new(bytes);
    let magic = cur.read_array::<4>()?;
    if magic != MAGIC {
        return Err(Error::Format(format!(
            "TimeIndexSegment magic mismatch: expected VLTI, got {magic:?}"
        )));
    }
    let version = cur.read_u16()?;
    if version != VERSION {
        return Err(Error::Format(format!(
            "TimeIndexSegment version mismatch: expected {VERSION}, got {version}"
        )));
    }
    let partition_flag = cur.read_u8()?;
    let collection_id = match partition_flag {
        0 => {
            let _ = cur.read_array::<4>()?;
            None
        }
        1 => Some(CollectionId(cur.read_u32()?)),
        other => {
            return Err(Error::Format(format!(
                "TimeIndexSegment partition flag must be 0 or 1, got {other}"
            )));
        }
    };
    let count = cur.read_u32()? as usize;
    let previous_root = read_optional_segment_ref(&mut cur)?;

    if count == 0 {
        if !cur.is_empty() {
            return Err(Error::Format(format!(
                "TimeIndexSegment: {} trailing bytes (count=0)",
                cur.remaining()
            )));
        }
        return Ok(TimeIndexSegment {
            collection_id,
            entries: Vec::new(),
            previous_root,
        });
    }

    // --- frame_id column ---
    let frame_blob = cur.read_blob()?;
    let mut fc = Cursor::new(frame_blob);
    let first_frame = fc.read_u64()?;
    let mut frame_ids: Vec<u64> = Vec::with_capacity(count);
    frame_ids.push(first_frame);
    let mut prev = first_frame;
    for _ in 1..count {
        let delta = fc.read_uvarint()?;
        let next = prev
            .checked_add(delta)
            .ok_or_else(|| Error::Format("TimeIndexSegment frame_id delta overflow".into()))?;
        frame_ids.push(next);
        prev = next;
    }
    if !fc.is_empty() {
        return Err(Error::Format(format!(
            "TimeIndexSegment: {} trailing bytes in frame_id column",
            fc.remaining()
        )));
    }

    // --- created_at column ---
    let created_blob = cur.read_blob()?;
    let mut cc = Cursor::new(created_blob);
    let first_created = cc.read_i64()?;
    let mut created: Vec<i64> = Vec::with_capacity(count);
    created.push(first_created);
    let mut prev_delta: i64 = 0;
    let mut prev_value: i64 = first_created;
    for _ in 1..count {
        let dod = cc.read_ivarint()?;
        let delta = prev_delta
            .checked_add(dod)
            .ok_or_else(|| Error::Format("TimeIndexSegment created delta overflow".into()))?;
        let value = prev_value
            .checked_add(delta)
            .ok_or_else(|| Error::Format("TimeIndexSegment created value overflow".into()))?;
        created.push(value);
        prev_value = value;
        prev_delta = delta;
    }
    if !cc.is_empty() {
        return Err(Error::Format(format!(
            "TimeIndexSegment: {} trailing bytes in created_at column",
            cc.remaining()
        )));
    }

    // --- updated_at column ---
    let updated_blob = cur.read_blob()?;
    let mut uc = Cursor::new(updated_blob);
    let mut updated: Vec<i64> = Vec::with_capacity(count);
    for c in &created {
        let offset = uc.read_ivarint()?;
        let value = c
            .checked_add(offset)
            .ok_or_else(|| Error::Format("TimeIndexSegment updated overflow".into()))?;
        updated.push(value);
    }
    if !uc.is_empty() {
        return Err(Error::Format(format!(
            "TimeIndexSegment: {} trailing bytes in updated_at column",
            uc.remaining()
        )));
    }

    // --- collection_id column (RLE) ---
    let coll_blob = cur.read_blob()?;
    let mut rc = Cursor::new(coll_blob);
    let run_count = rc.read_u32()? as usize;
    let mut collections: Vec<u32> = Vec::with_capacity(count);
    let mut total: u64 = 0;
    for _ in 0..run_count {
        let cid = rc.read_u32()?;
        let run_len = rc.read_uvarint()?;
        total = total.checked_add(run_len).ok_or_else(|| {
            Error::Format("TimeIndexSegment collection RLE total overflow".into())
        })?;
        for _ in 0..run_len {
            collections.push(cid);
        }
    }
    if total != count as u64 {
        return Err(Error::Format(format!(
            "TimeIndexSegment collection RLE covers {total} entries but count={count}"
        )));
    }
    if !rc.is_empty() {
        return Err(Error::Format(format!(
            "TimeIndexSegment: {} trailing bytes in collection column",
            rc.remaining()
        )));
    }

    if !cur.is_empty() {
        return Err(Error::Format(format!(
            "TimeIndexSegment: {} trailing bytes",
            cur.remaining()
        )));
    }

    let mut entries = Vec::with_capacity(count);
    for i in 0..count {
        entries.push(TimeIndexEntry {
            frame_id: FrameId(frame_ids[i]),
            collection_id: CollectionId(collections[i]),
            created_at: created[i],
            updated_at: updated[i],
        });
    }

    Ok(TimeIndexSegment {
        collection_id,
        entries,
        previous_root,
    })
}

fn write_optional_segment_ref(buf: &mut Vec<u8>, maybe_ref: Option<SegmentRef>) {
    if let Some(r) = maybe_ref {
        buf.push(1u8);
        buf.extend_from_slice(&r.segment_id.0.to_le_bytes());
        buf.extend_from_slice(&r.offset.to_le_bytes());
        buf.extend_from_slice(&r.length.to_le_bytes());
        buf.extend_from_slice(&r.checksum);
    } else {
        buf.push(0u8);
    }
}

fn read_optional_segment_ref(cur: &mut Cursor<'_>) -> Result<Option<SegmentRef>> {
    let flag = cur.read_u8()?;
    match flag {
        0 => Ok(None),
        1 => {
            let segment_id = crate::format::SegmentId(cur.read_u64()?);
            let offset = cur.read_u64()?;
            let length = cur.read_u64()?;
            let checksum = cur.read_array::<32>()?;
            Ok(Some(SegmentRef {
                segment_id,
                offset,
                length,
                checksum,
            }))
        }
        other => Err(Error::Format(format!(
            "TimeIndexSegment previous_root flag must be 0 or 1, got {other}"
        ))),
    }
}

fn write_blob(buf: &mut Vec<u8>, blob: &[u8]) -> Result<()> {
    let len = u32::try_from(blob.len())
        .map_err(|_| Error::Format("TimeIndexSegment blob too large".into()))?;
    buf.extend_from_slice(&len.to_le_bytes());
    buf.extend_from_slice(blob);
    Ok(())
}

fn write_uvarint(buf: &mut Vec<u8>, mut value: u64) {
    while value >= 0x80 {
        buf.push(((value as u8) & 0x7F) | 0x80);
        value >>= 7;
    }
    buf.push(value as u8);
}

fn write_ivarint(buf: &mut Vec<u8>, value: i64) {
    let zz = ((value << 1) ^ (value >> 63)) as u64;
    write_uvarint(buf, zz);
}

struct Cursor<'a> {
    bytes: &'a [u8],
    pos: usize,
}
impl<'a> Cursor<'a> {
    fn new(bytes: &'a [u8]) -> Self {
        Self { bytes, pos: 0 }
    }
    fn remaining(&self) -> usize {
        self.bytes.len() - self.pos
    }
    fn is_empty(&self) -> bool {
        self.pos == self.bytes.len()
    }
    fn read_array<const N: usize>(&mut self) -> Result<[u8; N]> {
        if self.pos + N > self.bytes.len() {
            return Err(Error::Format(format!(
                "TimeIndexSegment truncated: need {N} bytes at offset {}",
                self.pos
            )));
        }
        let mut out = [0u8; N];
        out.copy_from_slice(&self.bytes[self.pos..self.pos + N]);
        self.pos += N;
        Ok(out)
    }
    fn read_u8(&mut self) -> Result<u8> {
        Ok(self.read_array::<1>()?[0])
    }
    fn read_u16(&mut self) -> Result<u16> {
        Ok(u16::from_le_bytes(self.read_array::<2>()?))
    }
    fn read_u32(&mut self) -> Result<u32> {
        Ok(u32::from_le_bytes(self.read_array::<4>()?))
    }
    fn read_u64(&mut self) -> Result<u64> {
        Ok(u64::from_le_bytes(self.read_array::<8>()?))
    }
    fn read_i64(&mut self) -> Result<i64> {
        Ok(i64::from_le_bytes(self.read_array::<8>()?))
    }
    fn read_blob(&mut self) -> Result<&'a [u8]> {
        let len = self.read_u32()? as usize;
        if self.pos + len > self.bytes.len() {
            return Err(Error::Format(format!(
                "TimeIndexSegment blob truncated: declared {len} but only {} bytes left",
                self.bytes.len() - self.pos
            )));
        }
        let out = &self.bytes[self.pos..self.pos + len];
        self.pos += len;
        Ok(out)
    }
    fn read_uvarint(&mut self) -> Result<u64> {
        let mut result: u64 = 0;
        let mut shift: u32 = 0;
        loop {
            let b = self.read_u8()?;
            let chunk = u64::from(b & 0x7F);
            result |= chunk
                .checked_shl(shift)
                .ok_or_else(|| Error::Format("TimeIndexSegment varint overflow".into()))?;
            if b & 0x80 == 0 {
                return Ok(result);
            }
            shift += 7;
            if shift >= 64 {
                return Err(Error::Format("TimeIndexSegment varint too long".into()));
            }
        }
    }
    fn read_ivarint(&mut self) -> Result<i64> {
        let zz = self.read_uvarint()?;
        Ok(((zz >> 1) as i64) ^ -((zz & 1) as i64))
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    fn entry(frame: u64, coll: u32, created: i64, updated: i64) -> TimeIndexEntry {
        TimeIndexEntry {
            frame_id: FrameId(frame),
            collection_id: CollectionId(coll),
            created_at: created,
            updated_at: updated,
        }
    }

    #[test]
    fn roundtrip_global_empty() {
        let seg = TimeIndexSegment {
            collection_id: None,
            entries: Vec::new(),
            previous_root: None,
        };
        let bytes = encode_time_index_segment(&seg).unwrap();
        let decoded = decode_time_index_segment(&bytes).unwrap();
        assert_eq!(decoded, seg);
    }

    #[test]
    fn roundtrip_global_with_entries() {
        let seg = TimeIndexSegment {
            collection_id: None,
            entries: vec![
                entry(1, 1, 1_000, 1_000),
                entry(2, 1, 1_500, 2_000),
                entry(3, 2, 1_500, 1_500),
                entry(4, 1, 2_000, 2_500),
            ],
            previous_root: None,
        };
        let bytes = encode_time_index_segment(&seg).unwrap();
        let decoded = decode_time_index_segment(&bytes).unwrap();
        assert_eq!(decoded, seg);
    }

    #[test]
    fn roundtrip_partitioned() {
        let seg = TimeIndexSegment {
            collection_id: Some(CollectionId(7)),
            entries: vec![entry(1, 7, 100, 100), entry(2, 7, 200, 250)],
            previous_root: None,
        };
        let bytes = encode_time_index_segment(&seg).unwrap();
        let decoded = decode_time_index_segment(&bytes).unwrap();
        assert_eq!(decoded, seg);
    }

    #[test]
    fn rejects_unsorted_created_at() {
        let seg = TimeIndexSegment {
            collection_id: None,
            entries: vec![entry(1, 1, 200, 200), entry(2, 1, 100, 100)],
            previous_root: None,
        };
        let err = encode_time_index_segment(&seg).expect_err("unsorted");
        assert!(err.to_string().contains("sorted ascending"));
    }

    #[test]
    fn rejects_magic() {
        let seg = TimeIndexSegment {
            collection_id: None,
            entries: vec![entry(1, 1, 100, 100)],
            previous_root: None,
        };
        let mut bytes = encode_time_index_segment(&seg).unwrap();
        bytes[0] = b'X';
        let err = decode_time_index_segment(&bytes).expect_err("magic");
        assert!(err.to_string().contains("magic"));
    }

    #[test]
    fn rejects_trailing() {
        let seg = TimeIndexSegment {
            collection_id: None,
            entries: vec![entry(1, 1, 100, 100)],
            previous_root: None,
        };
        let mut bytes = encode_time_index_segment(&seg).unwrap();
        bytes.push(0xAB);
        let err = decode_time_index_segment(&bytes).expect_err("trailing");
        assert!(err.to_string().contains("trailing"));
    }

    /// On a synthetic Quora-shaped load (522 k entries, sequential
    /// frame_ids, single collection, same created_at second, no
    /// updates), v2 must produce ≤ 4 B/entry — a >7× reduction from
    /// the 28 B/entry v1 layout.
    #[test]
    fn columnar_compresses_quora_shape() {
        let now: i64 = 1_700_000_000;
        let entries: Vec<TimeIndexEntry> = (1..=522_931u64)
            .map(|fid| entry(fid, 1, now, now))
            .collect();
        let seg = TimeIndexSegment {
            collection_id: None,
            entries,
            previous_root: None,
        };
        let bytes = encode_time_index_segment(&seg).unwrap();
        let per_entry = bytes.len() as f64 / 522_931.0;
        assert!(
            per_entry < 4.0,
            "Quora-shaped v2 encode is {per_entry:.2} B/entry, expected < 4",
        );
        let decoded = decode_time_index_segment(&bytes).unwrap();
        assert_eq!(decoded.entries.len(), 522_931);
        assert_eq!(decoded.entries[0].frame_id.0, 1);
        assert_eq!(decoded.entries[522_930].frame_id.0, 522_931);
    }
}