1use serde::{Deserialize, Serialize};
26use sha2::{Digest, Sha256};
27use std::path::Path;
28
29const MAGIC: &[u8; 6] = b"LCIB1\n";
30const ZSTD_LEVEL: i32 = 3;
31const NONCE_LEN: usize = 24;
32const MAX_DECODED_BYTES: usize = 512 * 1024 * 1024;
35
36const BUNDLE_FILES: [&str; 2] = ["bm25_index.bin.zst", "embeddings.json"];
38
39#[derive(Debug, Clone, Serialize, Deserialize)]
40pub struct BundleManifest {
41 pub version: u32,
43 pub project_hash: String,
45 pub created_at: String,
47 pub engine_version: String,
49 pub files: Vec<BundleFileEntry>,
50}
51
52#[derive(Debug, Clone, Serialize, Deserialize)]
53pub struct BundleFileEntry {
54 pub name: String,
55 pub size: u64,
56 pub sha256: String,
57}
58
59#[derive(Debug, thiserror::Error)]
60pub enum BundleError {
61 #[error("no index artifacts found in {0} — run a search once (or `lean-ctx index`) to build the local index first")]
62 NothingToBundle(String),
63 #[error("not a lean-ctx index bundle (bad magic)")]
64 BadMagic,
65 #[error("corrupt bundle: {0}")]
66 Corrupt(String),
67 #[error("decryption failed — bundle was encrypted with a different account key (re-push from a logged-in device)")]
68 Decrypt,
69 #[error("io: {0}")]
70 Io(#[from] std::io::Error),
71}
72
73fn sha256_hex(data: &[u8]) -> String {
74 let mut h = Sha256::new();
75 h.update(data);
76 let digest = h.finalize();
77 let mut out = String::with_capacity(digest.len() * 2);
78 for b in digest {
79 use std::fmt::Write;
80 let _ = write!(out, "{b:02x}");
81 }
82 out
83}
84
85#[must_use]
93pub fn local_index_present(project_root: &Path) -> bool {
94 let dir = crate::core::index_namespace::vectors_dir(project_root);
95 BUNDLE_FILES.iter().any(|name| dir.join(name).is_file())
96}
97
98pub fn pack(project_root: &Path) -> Result<(Vec<u8>, BundleManifest), BundleError> {
101 let dir = crate::core::index_namespace::vectors_dir(project_root);
102 let mut files = Vec::new();
103 let mut payload = Vec::new();
104
105 for name in BUNDLE_FILES {
106 let path = dir.join(name);
107 let Ok(data) = std::fs::read(&path) else {
108 continue;
109 };
110 files.push(BundleFileEntry {
111 name: name.to_string(),
112 size: data.len() as u64,
113 sha256: sha256_hex(&data),
114 });
115 payload.extend_from_slice(&data);
116 }
117
118 if files.is_empty() {
119 return Err(BundleError::NothingToBundle(dir.display().to_string()));
120 }
121
122 let manifest = BundleManifest {
123 version: 1,
124 project_hash: crate::core::index_namespace::namespace_hash(project_root),
128 created_at: chrono::Utc::now().to_rfc3339(),
129 engine_version: env!("CARGO_PKG_VERSION").to_string(),
130 files,
131 };
132
133 let manifest_json =
134 serde_json::to_vec(&manifest).map_err(|e| BundleError::Corrupt(e.to_string()))?;
135 let compressed = zstd::encode_all(payload.as_slice(), ZSTD_LEVEL)
136 .map_err(|e| BundleError::Corrupt(format!("zstd: {e}")))?;
137
138 let mut out = Vec::with_capacity(MAGIC.len() + 4 + manifest_json.len() + compressed.len());
139 out.extend_from_slice(MAGIC);
140 out.extend_from_slice(
141 &u32::try_from(manifest_json.len())
142 .map_err(|_| BundleError::Corrupt("manifest too large".into()))?
143 .to_le_bytes(),
144 );
145 out.extend_from_slice(&manifest_json);
146 out.extend_from_slice(&compressed);
147 Ok((out, manifest))
148}
149
150pub fn read_manifest(container: &[u8]) -> Result<BundleManifest, BundleError> {
152 let (manifest, _payload) = split_container(container)?;
153 Ok(manifest)
154}
155
156fn split_container(container: &[u8]) -> Result<(BundleManifest, Vec<u8>), BundleError> {
157 if container.len() < MAGIC.len() + 4 || &container[..MAGIC.len()] != MAGIC {
158 return Err(BundleError::BadMagic);
159 }
160 let len_start = MAGIC.len();
161 let manifest_len = u32::from_le_bytes(
162 container[len_start..len_start + 4]
163 .try_into()
164 .map_err(|_| BundleError::Corrupt("truncated header".into()))?,
165 ) as usize;
166 let manifest_end = len_start + 4 + manifest_len;
167 if container.len() < manifest_end {
168 return Err(BundleError::Corrupt("truncated manifest".into()));
169 }
170 let manifest: BundleManifest = serde_json::from_slice(&container[len_start + 4..manifest_end])
171 .map_err(|e| BundleError::Corrupt(format!("manifest: {e}")))?;
172
173 let declared: u64 = manifest.files.iter().map(|f| f.size).sum();
174 if declared > MAX_DECODED_BYTES as u64 {
175 return Err(BundleError::Corrupt(format!(
176 "declared payload {declared} bytes exceeds the {MAX_DECODED_BYTES} byte ceiling"
177 )));
178 }
179
180 let payload = zstd::decode_all(&container[manifest_end..])
181 .map_err(|e| BundleError::Corrupt(format!("zstd: {e}")))?;
182 if payload.len() as u64 != declared {
183 return Err(BundleError::Corrupt(format!(
184 "payload size mismatch: got {}, manifest declares {declared}",
185 payload.len()
186 )));
187 }
188 Ok((manifest, payload))
189}
190
191pub fn unpack(project_root: &Path, container: &[u8]) -> Result<BundleManifest, BundleError> {
195 let (manifest, payload) = split_container(container)?;
196
197 let dir = crate::core::index_namespace::vectors_dir(project_root);
198 std::fs::create_dir_all(&dir)?;
199
200 let mut offset = 0usize;
202 let mut verified: Vec<(&BundleFileEntry, &[u8])> = Vec::with_capacity(manifest.files.len());
203 for entry in &manifest.files {
204 let size = usize::try_from(entry.size)
205 .map_err(|_| BundleError::Corrupt("file size overflow".into()))?;
206 let end = offset
207 .checked_add(size)
208 .filter(|&e| e <= payload.len())
209 .ok_or_else(|| BundleError::Corrupt("file extends past payload".into()))?;
210 let data = &payload[offset..end];
211 if sha256_hex(data) != entry.sha256 {
212 return Err(BundleError::Corrupt(format!(
213 "sha256 mismatch for {}",
214 entry.name
215 )));
216 }
217 if !BUNDLE_FILES.contains(&entry.name.as_str()) {
219 return Err(BundleError::Corrupt(format!(
220 "unexpected file in bundle: {}",
221 entry.name
222 )));
223 }
224 verified.push((entry, data));
225 offset = end;
226 }
227
228 for (entry, data) in verified {
229 let target = dir.join(&entry.name);
230 let tmp = dir.join(format!(".{}.pull.tmp", entry.name));
231 std::fs::write(&tmp, data)?;
232 std::fs::rename(&tmp, &target)?;
233 }
234 Ok(manifest)
235}
236
237#[must_use]
244pub fn derive_key(api_key: &str) -> [u8; 32] {
245 let hk = hkdf::Hkdf::<Sha256>::new(Some(b"leanctx"), api_key.as_bytes());
246 let mut okm = [0u8; 32];
247 hk.expand(b"index-bundle-v1", &mut okm)
248 .expect("32 bytes is a valid HKDF-SHA256 output length");
249 okm
250}
251
252pub fn encrypt(container: &[u8], key: &[u8; 32]) -> Result<Vec<u8>, BundleError> {
254 use chacha20poly1305::aead::{Aead, KeyInit};
255 use chacha20poly1305::{XChaCha20Poly1305, XNonce};
256
257 let mut nonce_bytes = [0u8; NONCE_LEN];
258 getrandom::fill(&mut nonce_bytes)
259 .map_err(|e| BundleError::Corrupt(format!("nonce generation: {e}")))?;
260 let cipher = XChaCha20Poly1305::new(key.into());
261 let ciphertext = cipher
262 .encrypt(XNonce::from_slice(&nonce_bytes), container)
263 .map_err(|_| BundleError::Corrupt("encryption failed".into()))?;
264
265 let mut out = Vec::with_capacity(NONCE_LEN + ciphertext.len());
266 out.extend_from_slice(&nonce_bytes);
267 out.extend_from_slice(&ciphertext);
268 Ok(out)
269}
270
271pub fn decrypt(blob: &[u8], key: &[u8; 32]) -> Result<Vec<u8>, BundleError> {
273 use chacha20poly1305::aead::{Aead, KeyInit};
274 use chacha20poly1305::{XChaCha20Poly1305, XNonce};
275
276 if blob.len() <= NONCE_LEN {
277 return Err(BundleError::Corrupt("blob shorter than nonce".into()));
278 }
279 let (nonce, ciphertext) = blob.split_at(NONCE_LEN);
280 let cipher = XChaCha20Poly1305::new(key.into());
281 cipher
282 .decrypt(XNonce::from_slice(nonce), ciphertext)
283 .map_err(|_| BundleError::Decrypt)
284}
285
286#[cfg(test)]
287mod tests {
288 use super::*;
289
290 fn project_with_index() -> (tempfile::TempDir, std::path::PathBuf) {
292 let data_dir = tempfile::tempdir().unwrap();
293 let root = data_dir.path().join("proj");
294 std::fs::create_dir_all(root.join(".git")).unwrap();
295 std::env::set_var("LEAN_CTX_DATA_DIR", data_dir.path());
296
297 let vectors = crate::core::index_namespace::vectors_dir(&root);
298 std::fs::create_dir_all(&vectors).unwrap();
299 std::fs::write(vectors.join("bm25_index.bin.zst"), b"fake-bm25-bytes").unwrap();
300 std::fs::write(vectors.join("embeddings.json"), br#"{"entries":[]}"#).unwrap();
301 (data_dir, root)
302 }
303
304 #[test]
305 fn pack_unpack_roundtrip_preserves_artifacts() {
306 let _env = crate::core::data_dir::test_env_lock();
307 let (data_dir, root) = project_with_index();
308
309 let (container, manifest) = pack(&root).unwrap();
310 assert_eq!(manifest.version, 1);
311 assert_eq!(manifest.files.len(), 2);
312
313 let vectors = crate::core::index_namespace::vectors_dir(&root);
315 std::fs::remove_file(vectors.join("bm25_index.bin.zst")).unwrap();
316 std::fs::remove_file(vectors.join("embeddings.json")).unwrap();
317
318 let restored = unpack(&root, &container).unwrap();
319 assert_eq!(restored.project_hash, manifest.project_hash);
320 assert_eq!(
321 std::fs::read(vectors.join("bm25_index.bin.zst")).unwrap(),
322 b"fake-bm25-bytes"
323 );
324 assert_eq!(
325 std::fs::read(vectors.join("embeddings.json")).unwrap(),
326 br#"{"entries":[]}"#
327 );
328 drop(data_dir);
329 }
330
331 #[test]
332 fn pack_without_artifacts_is_a_clear_error() {
333 let _env = crate::core::data_dir::test_env_lock();
334 let data_dir = tempfile::tempdir().unwrap();
335 let root = data_dir.path().join("empty");
336 std::fs::create_dir_all(root.join(".git")).unwrap();
337 std::env::set_var("LEAN_CTX_DATA_DIR", data_dir.path());
338
339 match pack(&root) {
340 Err(BundleError::NothingToBundle(_)) => {}
341 other => panic!("expected NothingToBundle, got {other:?}"),
342 }
343 }
344
345 #[test]
346 fn tampered_payload_is_rejected() {
347 let _env = crate::core::data_dir::test_env_lock();
348 let (_data_dir, root) = project_with_index();
349 let (container, _) = pack(&root).unwrap();
350
351 let (manifest, mut payload) = split_container(&container).unwrap();
354 payload[0] ^= 0xFF;
355 let manifest_json = serde_json::to_vec(&manifest).unwrap();
356 let mut forged = Vec::new();
357 forged.extend_from_slice(MAGIC);
358 forged.extend_from_slice(&(u32::try_from(manifest_json.len()).unwrap()).to_le_bytes());
359 forged.extend_from_slice(&manifest_json);
360 forged.extend_from_slice(&zstd::encode_all(payload.as_slice(), 3).unwrap());
361
362 match unpack(&root, &forged) {
363 Err(BundleError::Corrupt(msg)) => assert!(msg.contains("sha256"), "{msg}"),
364 other => panic!("expected Corrupt(sha256), got {other:?}"),
365 }
366 }
367
368 #[test]
369 fn bad_magic_is_rejected() {
370 match read_manifest(b"not-a-bundle") {
371 Err(BundleError::BadMagic) => {}
372 other => panic!("expected BadMagic, got {other:?}"),
373 }
374 }
375
376 #[test]
377 fn encrypt_decrypt_roundtrip_and_wrong_key_fails() {
378 let key = derive_key("test-api-key-1");
379 let plaintext = b"LCIB1\n-fake-container-bytes".to_vec();
380
381 let blob = encrypt(&plaintext, &key).unwrap();
382 assert_ne!(&blob[NONCE_LEN..], plaintext.as_slice());
383 assert_eq!(decrypt(&blob, &key).unwrap(), plaintext);
384
385 let wrong = derive_key("test-api-key-2");
386 match decrypt(&blob, &wrong) {
387 Err(BundleError::Decrypt) => {}
388 other => panic!("expected Decrypt error, got {other:?}"),
389 }
390 }
391
392 #[test]
393 fn key_derivation_is_stable_and_key_separated() {
394 assert_eq!(derive_key("k"), derive_key("k"));
396 assert_ne!(derive_key("k"), derive_key("k2"));
397 assert_ne!(derive_key("k").as_slice(), b"k".as_slice());
399 }
400
401 #[test]
402 fn nonces_are_unique_per_encryption() {
403 let key = derive_key("k");
404 let a = encrypt(b"same", &key).unwrap();
405 let b = encrypt(b"same", &key).unwrap();
406 assert_ne!(a[..NONCE_LEN], b[..NONCE_LEN]);
407 }
408}