1use std::collections::HashSet;
17
18use limnifs_core::{
19 compute_merkle_root, hash_empty_section, hash_section, parse_feature_flags_section,
20 parse_manifest_header, parse_metadata_blob, parse_metadata_reference, parse_slab_index,
21 ContentHandle, CoreError, ManifestCursor, SectionHashes, HISTORY_SECTION_VERSION,
22 SLAB_INDEX_SECTION_VERSION,
23};
24use limnifs_format::{ManifestRoot, SlabId};
25
26#[derive(Clone, Debug)]
28pub struct CompactionResult {
29 pub manifest_bytes: Vec<u8>,
30 pub merkle_root: ManifestRoot,
31 pub slab_bytes: Option<Vec<u8>>,
32 pub original_drop_count: usize,
33 pub compacted_drop_count: usize,
34 pub reclaimed_drops: usize,
35}
36
37#[derive(Debug)]
39pub enum CompactionError {
40 Core(CoreError),
41 Io(std::io::Error),
42}
43
44impl std::fmt::Display for CompactionError {
45 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
46 match self {
47 Self::Core(e) => write!(f, "{e}"),
48 Self::Io(e) => write!(f, "I/O: {e}"),
49 }
50 }
51}
52
53impl std::error::Error for CompactionError {}
54
55impl From<CoreError> for CompactionError {
56 fn from(e: CoreError) -> Self {
57 Self::Core(e)
58 }
59}
60
61impl From<std::io::Error> for CompactionError {
62 fn from(e: std::io::Error) -> Self {
63 Self::Io(e)
64 }
65}
66
67struct ExtractedDrop {
69 id: [u8; 32],
70 compressed: Vec<u8>,
71 codec: u8,
72 plaintext_len: u32,
73 flags: u8,
76}
77
78pub fn compact_image(
86 manifest_bytes: &[u8],
87 slab_bytes: &[u8],
88) -> Result<CompactionResult, CompactionError> {
89 let mut cursor = ManifestCursor::new(manifest_bytes);
91 let _header = parse_manifest_header(&mut cursor)?;
92
93 let prefix_end = {
96 let flags_start = cursor.position();
97 let _ = parse_feature_flags_section(&mut cursor)?;
98 let flags_end = cursor.position();
99 let meta_ref_start = cursor.position();
100 let meta_ref = parse_metadata_reference(&mut cursor)?;
101 let _meta_ref_end = cursor.position();
102 let _ = (flags_start, flags_end, meta_ref_start);
103
104 let referenced = find_referenced_drops(&meta_ref)?;
106
107 let slab_index_start = cursor.position();
109 let slab_index = parse_slab_index(&mut cursor)?;
110 let slab_index_end = cursor.position();
111 let _ = (slab_index_start, slab_index_end, &slab_index);
112
113 referenced
114 };
115
116 let view = limnifs_core::parse_slab(slab_bytes)?;
119 let original_count = view.drop_records().len();
120 let win_start = view.solid_window_offset();
121 let mut kept_drops: Vec<ExtractedDrop> = Vec::new();
122
123 for record in view.drop_records() {
124 if prefix_end.contains(record.drop_id.as_bytes()) {
125 let offset = usize::try_from(record.offset_in_window).unwrap_or(0);
126 let len = usize::try_from(record.len_in_window).unwrap_or(0);
127 let start = win_start + offset;
128 let end = start + len;
129 if end > slab_bytes.len() {
130 continue;
131 }
132 kept_drops.push(ExtractedDrop {
133 id: *record.drop_id.as_bytes(),
134 compressed: slab_bytes[start..end].to_vec(),
135 codec: record.representation.codec,
136 plaintext_len: record.plaintext_len,
137 flags: record.flags,
138 });
139 }
140 }
141
142 let compacted_count = kept_drops.len();
143 let reclaimed = original_count.saturating_sub(compacted_count);
144
145 let (new_slab_bytes, new_slab_id) = encode_compacted_slab(&kept_drops);
147
148 let new_manifest = reassemble_manifest(manifest_bytes, &new_slab_bytes, &new_slab_id)?;
150
151 Ok(CompactionResult {
152 manifest_bytes: new_manifest.bytes,
153 merkle_root: new_manifest.merkle_root,
154 slab_bytes: Some(new_slab_bytes),
155 original_drop_count: original_count,
156 compacted_drop_count: compacted_count,
157 reclaimed_drops: reclaimed,
158 })
159}
160
161fn find_referenced_drops(
163 meta_ref: &limnifs_core::MetadataReference,
164) -> Result<HashSet<[u8; 32]>, CompactionError> {
165 let mut referenced = HashSet::new();
166 if let Some(blob_bytes) = &meta_ref.inline_metadata {
167 let mut blob_cursor = ManifestCursor::new(blob_bytes);
168 let blob = parse_metadata_blob(&mut blob_cursor)?;
169 for inode in &blob.inodes {
170 if let ContentHandle::SliceMap(slices) = &inode.content_handle {
171 for slice in slices {
172 referenced.insert(*slice.drop_id.as_bytes());
173 }
174 }
175 }
176 }
177 Ok(referenced)
178}
179
180fn encode_compacted_slab(drops: &[ExtractedDrop]) -> (Vec<u8>, SlabId) {
182 let mut drop_records = Vec::new();
183 let mut solid_window = Vec::new();
184
185 for drop in drops {
186 let win_len = u32::try_from(drop.compressed.len()).unwrap_or(0);
187 let offset = u32::try_from(solid_window.len()).unwrap_or(0);
188 drop_records.extend_from_slice(&drop.id);
189 drop_records.extend_from_slice(&drop.plaintext_len.to_le_bytes());
190 drop_records.extend_from_slice(&[drop.codec, 0x00, 0x00]); drop_records.push(0x00); drop_records.extend_from_slice(&offset.to_le_bytes());
193 drop_records.extend_from_slice(&win_len.to_le_bytes());
194 drop_records.push(limnifs_core::drop_record::NO_DICT); drop_records.push(drop.flags); solid_window.extend_from_slice(&drop.compressed);
197 }
198
199 let slab_content = [&drop_records[..], &solid_window[..]].concat();
200 let slab_hash = hash_section(&slab_content);
201 let slab_id = SlabId::new(0, slab_hash);
202
203 let total_length = 56 + slab_content.len();
204 let mut slab_bytes = Vec::with_capacity(total_length);
205 slab_bytes.extend_from_slice(b"LIM1");
206 slab_bytes.extend_from_slice(&1u16.to_le_bytes()); slab_bytes.extend_from_slice(&slab_id.to_bytes());
208 slab_bytes.extend_from_slice(&(total_length as u64).to_le_bytes());
209 slab_bytes.push(0x00); slab_bytes.push(0x00); slab_bytes.extend_from_slice(&slab_content);
212
213 (slab_bytes, slab_id)
214}
215
216struct ReassembledManifest {
217 bytes: Vec<u8>,
218 merkle_root: ManifestRoot,
219}
220
221fn reassemble_manifest(
223 source: &[u8],
224 _slab_bytes: &[u8],
225 slab_id: &SlabId,
226) -> Result<ReassembledManifest, CompactionError> {
227 let mut cursor = ManifestCursor::new(source);
228
229 let header_start = cursor.position();
231 let _ = parse_manifest_header(&mut cursor)?;
232 let header_end = cursor.position();
233
234 let flags_start = cursor.position();
235 let _ = parse_feature_flags_section(&mut cursor)?;
236 let flags_end = cursor.position();
237
238 let meta_ref_start = cursor.position();
239 let meta_ref = parse_metadata_reference(&mut cursor)?;
240 let meta_ref_end = cursor.position();
241
242 let _ = parse_slab_index(&mut cursor)?;
244 let _ = limnifs_core::parse_history(&mut cursor)?;
245
246 let mut manifest = Vec::new();
248
249 manifest.extend_from_slice(&source[header_start..meta_ref_end]);
251
252 let slab_index_start_new = manifest.len();
254 manifest.push(SLAB_INDEX_SECTION_VERSION);
255 manifest.extend_from_slice(&1u32.to_le_bytes());
256 manifest.extend_from_slice(&slab_id.to_bytes());
257 manifest.extend_from_slice(&1u32.to_le_bytes());
258 let locator = "file:slab-0.bin";
259 manifest.extend_from_slice(&u32::try_from(locator.len()).unwrap_or(0).to_le_bytes());
260 manifest.extend_from_slice(locator.as_bytes());
261 let slab_index_end_new = manifest.len();
262
263 let history_start_new = manifest.len();
265 manifest.push(HISTORY_SECTION_VERSION);
266 manifest.extend_from_slice(&1u32.to_le_bytes());
267 manifest.push(0x04); manifest.extend_from_slice(&0u64.to_le_bytes());
269 manifest.extend_from_slice(&0u32.to_le_bytes());
270 manifest.extend_from_slice(&0u32.to_le_bytes());
271 let history_end_new = manifest.len();
272
273 let hashes = SectionHashes {
274 metadata: meta_ref.metadata_hash,
275 format_header: hash_section(&source[header_start..header_end]),
276 feature_flags: hash_section(&source[flags_start..flags_end]),
277 metadata_reference: hash_section(&source[meta_ref_start..meta_ref_end]),
278 slab_index: hash_section(&manifest[slab_index_start_new..slab_index_end_new]),
279 crypto_params: hash_empty_section(),
280 ec_params: hash_empty_section(),
281 dms_policy: hash_empty_section(),
282 delta_linkage: hash_empty_section(),
283 history: hash_section(&manifest[history_start_new..history_end_new]),
284 };
285 let merkle_root = compute_merkle_root(&hashes);
286
287 Ok(ReassembledManifest {
288 bytes: manifest,
289 merkle_root,
290 })
291}
292
293#[cfg(test)]
294mod tests {
295 use super::*;
296
297 #[test]
298 fn compact_preserves_referenced_drops() {
299 let temp =
300 std::env::temp_dir().join(format!("limnifs-compaction-test-{}", std::process::id()));
301 std::fs::create_dir_all(&temp).expect("create temp");
302 let data = vec![0xABu8; crate::INLINE_THRESHOLD + 100];
304 std::fs::write(temp.join("big.bin"), &data).expect("write big");
305
306 let artifact = crate::write_directory(&temp).expect("write");
307 std::fs::remove_dir_all(&temp).ok();
308
309 let slab_bytes = artifact.slab_bytes().map(Vec::from).unwrap_or_default();
310 if slab_bytes.is_empty() {
311 return;
312 }
313
314 let result = compact_image(&artifact.bytes, &slab_bytes).expect("compact");
315 assert_eq!(result.original_drop_count, result.compacted_drop_count);
316 assert_eq!(result.reclaimed_drops, 0);
317
318 let new_slab = result.slab_bytes.as_ref().expect("slab exists");
320 let view = limnifs_core::parse_slab(new_slab).expect("compacted slab parses");
321 assert_eq!(view.drop_records().len(), result.compacted_drop_count);
322 }
323
324 #[test]
325 fn compact_preserves_drop_plaintext() {
326 let temp =
327 std::env::temp_dir().join(format!("limnifs-compaction-pt-{}", std::process::id()));
328 std::fs::create_dir_all(&temp).expect("create temp");
329 let data = vec![0xCDu8; crate::INLINE_THRESHOLD + 200];
330 std::fs::write(temp.join("data.bin"), &data).expect("write");
331
332 let artifact = crate::write_directory(&temp).expect("write");
333 std::fs::remove_dir_all(&temp).ok();
334
335 let slab_bytes = artifact.slab_bytes().map(Vec::from).unwrap_or_default();
336 if slab_bytes.is_empty() {
337 return;
338 }
339
340 let result = compact_image(&artifact.bytes, &slab_bytes).expect("compact");
341 let new_slab = result.slab_bytes.as_ref().expect("slab");
342
343 let old_view = limnifs_core::parse_slab(&slab_bytes).expect("old slab parses");
345 let new_view = limnifs_core::parse_slab(new_slab).expect("new slab parses");
346
347 for old_record in old_view.drop_records() {
348 let old_pt = old_view
349 .plaintext_for(old_record.drop_id.as_bytes())
350 .expect("old drop exists")
351 .expect("decompress ok");
352 let new_pt = new_view
353 .plaintext_for(old_record.drop_id.as_bytes())
354 .expect("new drop exists")
355 .expect("decompress ok");
356 assert_eq!(old_pt, new_pt, "plaintext must match after compaction");
357 }
358 }
359
360 #[test]
361 fn compact_manifest_parses_correctly() {
362 let temp =
363 std::env::temp_dir().join(format!("limnifs-compaction-mp-{}", std::process::id()));
364 std::fs::create_dir_all(&temp).expect("create temp");
365 let data = vec![0xEFu8; crate::INLINE_THRESHOLD + 50];
366 std::fs::write(temp.join("file.bin"), &data).expect("write");
367
368 let artifact = crate::write_directory(&temp).expect("write");
369 std::fs::remove_dir_all(&temp).ok();
370
371 let slab_bytes = artifact.slab_bytes().map(Vec::from).unwrap_or_default();
372 if slab_bytes.is_empty() {
373 return;
374 }
375
376 let result = compact_image(&artifact.bytes, &slab_bytes).expect("compact");
377
378 let mut cursor = limnifs_core::ManifestCursor::new(&result.manifest_bytes);
380 limnifs_core::parse_manifest_header(&mut cursor).expect("header parses");
381 limnifs_core::parse_feature_flags_section(&mut cursor).expect("flags parse");
382 let meta_ref =
383 limnifs_core::parse_metadata_reference(&mut cursor).expect("metadata ref parses");
384 assert!(meta_ref.is_inlined());
385 limnifs_core::parse_slab_index(&mut cursor).expect("slab index parses");
386 limnifs_core::parse_history(&mut cursor).expect("history parses");
387 }
388}