lex_store/store.rs
1//! `Store` — content-addressed code repository.
2//!
3//! The filesystem is the source of truth. All operations read/write JSON
4//! files under `<root>/stages/<SigId>/`. There is no SQLite cache: every
5//! query walks the directory and parses what's needed. `cargo test`
6//! runs aren't perf-critical and the §4.6 acceptance requires the
7//! rebuild-from-filesystem property anyway.
8
9use crate::branches::DEFAULT_BRANCH;
10use crate::model::*;
11use lex_ast::{sig_id, stage_id, Stage};
12use serde::de::DeserializeOwned;
13use serde::Serialize;
14use std::collections::BTreeMap;
15use std::fs;
16use std::path::{Path, PathBuf};
17use std::time::{SystemTime, UNIX_EPOCH};
18
19#[derive(Debug, thiserror::Error)]
20pub enum StoreError {
21 #[error("io error: {0}")]
22 Io(#[from] std::io::Error),
23 #[error("serialization error: {0}")]
24 Serde(#[from] serde_json::Error),
25 #[error("imports cannot be published as stages")]
26 CannotPublishImport,
27 #[error("unknown stage_id `{0}`")]
28 UnknownStage(String),
29 #[error("unknown sig_id `{0}`")]
30 UnknownSig(String),
31 #[error("invalid lifecycle transition: {0}")]
32 InvalidTransition(String),
33 #[error("unknown branch `{0}`")]
34 UnknownBranch(String),
35 #[error("unknown blob `{0}`")]
36 UnknownBlob(String),
37 #[error("unknown blob ref `{namespace}/{key}`")]
38 UnknownBlobRef { namespace: String, key: String },
39 #[error("unknown op_id `{0}`")]
40 UnknownOp(lex_vcs::OpId),
41 /// A typed AST transform (#280) — e.g. `ReplaceMatchArm` — was
42 /// asked to operate on a node it couldn't address (wrong kind,
43 /// out-of-range arm index, unknown NodeId, etc.). Distinct from
44 /// `TypeError` (which means the transform succeeded but its
45 /// output didn't typecheck) so callers can render the right
46 /// error message.
47 #[error("transform failed: {0}")]
48 TransformError(lex_ast::TransformError),
49 #[error(transparent)]
50 Apply(#[from] lex_vcs::ApplyError),
51 /// The candidate program — i.e. the source the caller is
52 /// publishing — doesn't typecheck. The branch head is unchanged
53 /// and no op records are persisted. Issue #130's "always-valid
54 /// HEAD" invariant: the gate runs before any side effect, so a
55 /// type-broken publish leaves no footprint.
56 #[error("type errors in published program: {} error(s)", .0.len())]
57 TypeError(Vec<lex_types::TypeError>),
58 /// The op was persisted but a `required_attestations` rule in
59 /// `policy.json` (#245) refused to advance the branch head past
60 /// it. The op record is durable — re-running with the missing
61 /// attestations recorded will succeed without re-persisting —
62 /// but the branch is unchanged. Surfaced as a structured JSON
63 /// envelope on the HTTP API.
64 #[error(
65 "branch advance blocked: op {} missing attestations: {}",
66 .0.op_id, .0.missing.join(", ")
67 )]
68 BranchAdvanceBlocked(crate::policy::BranchAdvanceBlocked),
69 /// All retry attempts of the CAS branch-head advance failed
70 /// because another writer kept advancing the same branch
71 /// (#262). The op record itself is durable in the op log
72 /// (orphaned), so re-running with backoff would eventually
73 /// land — return `503 Contention { retry_after }` from the
74 /// HTTP API and let the client back off.
75 #[error("branch advance contention on `{branch}`: {attempts} retries exhausted")]
76 Contention { branch: String, attempts: u32 },
77 /// The op was persisted but its stage carries an attestation
78 /// produced by a retroactively quarantined tool (#248). The
79 /// branch head is unchanged. The op record stays in the log
80 /// (audit trail intact); re-running with the producer
81 /// unblocked, or with un-contaminated attestations, succeeds
82 /// without re-persisting the op.
83 #[error(
84 "branch advance blocked: op {} touches stage {} with an attestation from \
85 quarantined producer `{}` (blocked at {}, attestation at {})",
86 .0.op_id, .0.stage_id, .0.tool_id, .0.blocked_at, .0.attestation_at
87 )]
88 ProducerBlocked(crate::policy::ProducerBlocked),
89 /// The op would push its session's monotonic budget over the
90 /// cap configured in `policy.session_budgets` (#292 slice 3).
91 /// The op is *not* persisted; the branch head is unchanged.
92 /// The caller should either start a new session, raise the
93 /// cap, or refactor to fit the budget. HTTP API maps to 503.
94 #[error("session `{session_id}` budget exceeded: spent_after={spent_after} > cap={cap}")]
95 BudgetExceeded {
96 session_id: String,
97 cap: u64,
98 spent_after: u64,
99 },
100}
101
102/// The outcome returned by [`Store::publish_program`].
103#[derive(Debug, Clone, serde::Serialize)]
104pub struct PublishOutcome {
105 pub ops: Vec<PublishOp>,
106 pub head_op: Option<lex_vcs::OpId>,
107}
108
109/// One applied operation within a [`PublishOutcome`].
110#[derive(Debug, Clone, serde::Serialize)]
111pub struct PublishOp {
112 pub op_id: lex_vcs::OpId,
113 pub kind: serde_json::Value,
114}
115
116/// One entry in the per-`SigId` stage history surfaced by
117/// `Store::sig_history`. Newest-first ordering is the responsibility
118/// of the producer.
119#[derive(Debug, Clone, serde::Serialize, PartialEq)]
120pub struct StageHistoryEntry {
121 pub stage_id: String,
122 pub status: StageStatus,
123 /// Wall-clock seconds of the most recent transition.
124 pub last_at: u64,
125 /// Wall-clock seconds when this stage was first written to the
126 /// store (its initial Draft transition). `None` for stages
127 /// whose lifecycle log doesn't include an explicit Draft entry
128 /// — shouldn't happen for stages published via `Store::publish`,
129 /// but the type allows hand-edited stores.
130 #[serde(skip_serializing_if = "Option::is_none")]
131 pub published_at: Option<u64>,
132}
133
134/// Per-candidate metadata surfaced by [`Store::list_candidates`]
135/// (#294). Returned sorted by `op_id` for deterministic output.
136#[derive(Debug, Clone, serde::Serialize, serde::Deserialize, PartialEq, Eq)]
137pub struct CandidateInfo {
138 pub op_id: lex_vcs::OpId,
139 pub stage_id: lex_vcs::StageId,
140 /// Author intent. Always set for `Candidate` ops emitted via
141 /// [`Store::propose_candidate`]; `None` only if a
142 /// hand-written raw op skipped the intent tag.
143 pub intent_id: Option<lex_vcs::IntentId>,
144}
145
146/// One line of `stage_index.jsonl`. See `Store::lookup_lifecycle`.
147#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
148struct StageIndexEntry {
149 stage_id: String,
150 sig_id: String,
151}
152
153/// Sentinel `sig_id` value recording "a full scan already established
154/// this stage_id exists nowhere in the store" (#825). Never a real
155/// sig — sig directory names are never empty.
156const MISSING_STAGE_MARKER: &str = "";
157
158pub struct Store {
159 root: PathBuf,
160}
161
162impl Store {
163 /// Open or create a store rooted at `root`.
164 pub fn open(root: impl AsRef<Path>) -> Result<Self, StoreError> {
165 let root = root.as_ref().to_path_buf();
166 fs::create_dir_all(root.join("stages"))?;
167 fs::create_dir_all(root.join("traces"))?;
168 let store = Self { root };
169 store.ensure_stage_index();
170 Ok(store)
171 }
172
173 /// One-time migration for a store that predates the reverse
174 /// index (#822), or whose previous rebuild pass never finished
175 /// (e.g. the process was killed or its client disconnected
176 /// mid-request — server-side work keeps running either way, but
177 /// a *restart* genuinely stops it): build the index in a single
178 /// pass instead of leaving every subsequent `lookup_lifecycle`
179 /// call to discover its own entry via the slow per-call scan-
180 /// and-backfill fallback.
181 ///
182 /// That per-call fallback is fine for the rare individual miss
183 /// it was designed for, but pathological as a *bulk* cold-start
184 /// strategy: on a tenant with a few thousand functions it means
185 /// redoing an O(total sigs) scan from scratch for *each* of a
186 /// few thousand cold entries — O(total sigs²) — which measured
187 /// as a near-stall (page-cache thrashing) on a memory-
188 /// constrained host. A single pass over `list_sigs()` is
189 /// O(total sigs) total.
190 ///
191 /// Gated on a dedicated completion marker
192 /// (`stage_index.complete`), NOT on `stage_index.jsonl`'s mere
193 /// existence — a partially-built index file (left behind by an
194 /// interrupted rebuild, lazy or bulk) must still trigger a
195 /// re-run so the remaining entries get backfilled in one more
196 /// cheap O(total sigs) pass, not silently be mistaken for
197 /// "already done" and fall back to the slow per-call path for
198 /// whatever's left. `rebuild_stage_index` already skips entries
199 /// it finds present, so re-running it against a partial index
200 /// only does the work that remains. The marker is written only
201 /// after a full pass returns `Ok`, so a failed pass (e.g. an I/O
202 /// error partway through `list_sigs`) is retried on the next
203 /// open rather than being marked done.
204 ///
205 /// Runs once per `Store::open` call — which, in a long-lived
206 /// server (lex-hub caches one `Store` per tenant for the life of
207 /// the process), means once per tenant per process lifetime, not
208 /// once per request. Once the marker exists (the steady state
209 /// after the first successful run on any given host) this is a
210 /// single cheap file-existence check. Best-effort like the rest
211 /// of the index: any failure here just leaves the slower per-call
212 /// fallback as the only path, never breaks correctness.
213 fn ensure_stage_index(&self) {
214 if self.stage_index_complete_marker_path().exists() {
215 return;
216 }
217 if self.rebuild_stage_index().is_ok() {
218 let _ = fs::write(self.stage_index_complete_marker_path(), "");
219 }
220 }
221
222 fn stage_index_complete_marker_path(&self) -> PathBuf {
223 self.root.join("stage_index.complete")
224 }
225
226 /// Build (or top up) the reverse index in one pass over every
227 /// SigId in the store, rather than relying on `lookup_lifecycle`
228 /// to discover entries one at a time. Safe to call at any time,
229 /// including on a partially-built index (e.g. one left behind by
230 /// an interrupted request that was populating it lazily): already-
231 /// indexed stage_ids are skipped, so this only does the work that
232 /// remains. Returns the number of newly-added entries.
233 pub fn rebuild_stage_index(&self) -> Result<usize, StoreError> {
234 // A sig's lifecycle can list the same stage_id more than once
235 // (Draft, then later Active, then Deprecated all carry the
236 // same stage_id with a different status) -- track newly-seen
237 // keys locally too, not just what was already on disk at the
238 // start, so a repeated stage_id within one sig's transitions
239 // doesn't get appended to the index more than once.
240 let mut existing = self.load_stage_index();
241 let mut added = 0usize;
242 for sig in self.list_sigs()? {
243 let Ok(life) = self.read_lifecycle(&sig) else { continue };
244 for t in &life.transitions {
245 if !existing.contains_key(&t.stage_id) {
246 self.append_stage_index_entry(&t.stage_id, &sig);
247 existing.insert(t.stage_id.clone(), sig.clone());
248 added += 1;
249 }
250 }
251 }
252 Ok(added)
253 }
254
255 pub fn root(&self) -> &Path {
256 &self.root
257 }
258
259 // ── Generic content-addressed blobs (#5 / M6.1) ──────────────────────────
260 //
261 // The stage store holds typed Lex ASTs; loom-style artifacts (generated
262 // code, JSON, prose) are opaque text. These blob methods give the store a
263 // generic content-addressed object alongside stages, plus a lightweight
264 // ref namespace so callers can bind names (e.g. a sprint's node ids) to
265 // blob shas without touching the operation-log branch machinery.
266 //
267 // The sha is the lowercase hex SHA-256 of the content's UTF-8 bytes —
268 // identical to Lex's `crypto.sha256_str`, so a blob written here and an
269 // artifact content-addressed in loom's SQLite store share the same id and
270 // are interchangeable by reference. Store-scoped, so under lex-hub each
271 // tenant store gets its own blob space for free.
272
273 fn blobs_dir(&self) -> PathBuf {
274 self.root.join("blobs")
275 }
276
277 fn blob_refs_dir(&self) -> PathBuf {
278 self.root.join("blobrefs")
279 }
280
281 /// Content-address `content` and persist it under `<root>/blobs/<sha>`.
282 /// Returns the sha. Idempotent: re-putting identical content is a no-op.
283 /// Concurrency-safe — writes to a unique temp file then atomically renames
284 /// onto the content-addressed path, so parallel writers of the same content
285 /// can't corrupt it.
286 pub fn put_blob(&self, content: &str) -> Result<String, StoreError> {
287 use sha2::{Digest, Sha256};
288 let sha = hex::encode(Sha256::digest(content.as_bytes()));
289 let dir = self.blobs_dir();
290 let path = dir.join(&sha);
291 if path.exists() {
292 return Ok(sha);
293 }
294 fs::create_dir_all(&dir)?;
295 static SEQ: std::sync::atomic::AtomicU64 = std::sync::atomic::AtomicU64::new(0);
296 let n = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
297 let tmp = dir.join(format!(".{sha}.{}.{n}.tmp", std::process::id()));
298 fs::write(&tmp, content.as_bytes())?;
299 // rename is atomic on the same filesystem; identical content makes a
300 // last-writer-wins race harmless.
301 fs::rename(&tmp, &path)?;
302 Ok(sha)
303 }
304
305 /// Read a blob by its sha. `UnknownBlob` if absent.
306 pub fn get_blob(&self, sha: &str) -> Result<String, StoreError> {
307 match fs::read_to_string(self.blobs_dir().join(sha)) {
308 Ok(s) => Ok(s),
309 Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
310 Err(StoreError::UnknownBlob(sha.to_string()))
311 }
312 Err(e) => Err(StoreError::Io(e)),
313 }
314 }
315
316 /// Whether a blob with this sha exists.
317 pub fn has_blob(&self, sha: &str) -> bool {
318 self.blobs_dir().join(sha).exists()
319 }
320
321 /// Bind `key` to a blob `sha` within `namespace` (e.g. namespace
322 /// `"loom/sprint-abc"`, key `"build-node"`). Overwrites an existing
323 /// binding. The namespace may contain `/`; neither namespace nor key may
324 /// contain a `..` path component.
325 pub fn set_blob_ref(&self, namespace: &str, key: &str, sha: &str) -> Result<(), StoreError> {
326 let dir = self.blob_ref_namespace_dir(namespace, key)?;
327 fs::create_dir_all(&dir)?;
328 fs::write(dir.join(key), sha.as_bytes())?;
329 Ok(())
330 }
331
332 /// Resolve `namespace`/`key` to a blob sha. `UnknownBlobRef` if unbound.
333 pub fn get_blob_ref(&self, namespace: &str, key: &str) -> Result<String, StoreError> {
334 let dir = self.blob_ref_namespace_dir(namespace, key)?;
335 match fs::read_to_string(dir.join(key)) {
336 Ok(s) => Ok(s.trim().to_string()),
337 Err(e) if e.kind() == std::io::ErrorKind::NotFound => Err(StoreError::UnknownBlobRef {
338 namespace: namespace.to_string(),
339 key: key.to_string(),
340 }),
341 Err(e) => Err(StoreError::Io(e)),
342 }
343 }
344
345 /// All `key → sha` bindings in a namespace (e.g. every artifact in a
346 /// sprint). Empty map if the namespace has no bindings yet.
347 pub fn list_blob_refs(
348 &self,
349 namespace: &str,
350 ) -> Result<std::collections::BTreeMap<String, String>, StoreError> {
351 let dir = self.blob_ref_namespace_dir(namespace, "x")?;
352 let mut out = std::collections::BTreeMap::new();
353 let entries = match fs::read_dir(&dir) {
354 Ok(e) => e,
355 Err(e) if e.kind() == std::io::ErrorKind::NotFound => return Ok(out),
356 Err(e) => return Err(StoreError::Io(e)),
357 };
358 for entry in entries {
359 let entry = entry?;
360 if entry.file_type()?.is_file() {
361 let key = entry.file_name().to_string_lossy().to_string();
362 let sha = fs::read_to_string(entry.path())?.trim().to_string();
363 out.insert(key, sha);
364 }
365 }
366 Ok(out)
367 }
368
369 // Resolve the on-disk dir for a (namespace, key), rejecting `..` traversal
370 // and `/` in the key. `key` is validated but not joined here (callers join
371 // it themselves so `list_blob_refs` can pass a dummy).
372 fn blob_ref_namespace_dir(&self, namespace: &str, key: &str) -> Result<PathBuf, StoreError> {
373 if key.contains('/') || key.contains('\\') || key.split('/').any(|c| c == "..") {
374 return Err(StoreError::UnknownBlobRef {
375 namespace: namespace.to_string(),
376 key: key.to_string(),
377 });
378 }
379 let mut dir = self.blob_refs_dir();
380 for comp in namespace.split('/') {
381 if comp == ".." || comp.contains('\\') {
382 return Err(StoreError::UnknownBlobRef {
383 namespace: namespace.to_string(),
384 key: key.to_string(),
385 });
386 }
387 if !comp.is_empty() {
388 dir.push(comp);
389 }
390 }
391 Ok(dir)
392 }
393
394 fn now() -> u64 {
395 SystemTime::now()
396 .duration_since(UNIX_EPOCH)
397 .map(|d| d.as_secs())
398 .unwrap_or(0)
399 }
400
401 fn sig_dir(&self, sig: &str) -> PathBuf {
402 self.root.join("stages").join(sig)
403 }
404 fn impl_dir(&self, sig: &str) -> PathBuf {
405 self.sig_dir(sig).join("implementations")
406 }
407 fn tests_dir(&self, sig: &str) -> PathBuf {
408 self.sig_dir(sig).join("tests")
409 }
410 fn specs_dir(&self, sig: &str) -> PathBuf {
411 self.sig_dir(sig).join("specs")
412 }
413 fn lifecycle_path(&self, sig: &str) -> PathBuf {
414 self.sig_dir(sig).join("lifecycle.json")
415 }
416
417 // ---- publish ----
418
419 /// Publish a stage as **Draft**. Returns the StageId.
420 /// Idempotent: republishing the same canonical AST returns the same
421 /// StageId without writing duplicates.
422 pub fn publish(&self, stage: &Stage) -> Result<String, StoreError> {
423 self.publish_signed(stage, None)
424 }
425
426 /// Like [`Self::publish`] but optionally attaches an Ed25519
427 /// signature over the StageId (#227). When `signer` is `Some`,
428 /// the persisted metadata gets a `signature` field that
429 /// downstream consumers can verify via
430 /// [`lex_vcs::verify_stage_id`].
431 ///
432 /// Idempotency: if a metadata file already exists the signature
433 /// is *not* re-written. This preserves "republishing is a no-op"
434 /// even across different signers — promoting a signed stage
435 /// requires a fresh stage hash anyway, so a metadata overwrite
436 /// would be the wrong primitive.
437 pub fn publish_signed(
438 &self,
439 stage: &Stage,
440 signer: Option<&lex_vcs::Keypair>,
441 ) -> Result<String, StoreError> {
442 let sig = sig_id(stage).ok_or(StoreError::CannotPublishImport)?;
443 let stage_id = stage_id(stage).ok_or(StoreError::CannotPublishImport)?;
444 let name = stage_name(stage).to_string();
445
446 fs::create_dir_all(self.impl_dir(&sig))?;
447 fs::create_dir_all(self.tests_dir(&sig))?;
448 fs::create_dir_all(self.specs_dir(&sig))?;
449
450 let ast_path = self.impl_dir(&sig).join(format!("{}.ast.json", stage_id));
451 let delta_path = self.impl_dir(&sig).join(format!("{}.delta.json", stage_id));
452 let meta_path = self
453 .impl_dir(&sig)
454 .join(format!("{}.metadata.json", stage_id));
455
456 // #261 slice 3: try delta encoding against the most recent
457 // prior stage in this sig's lifecycle. Falls back to a full
458 // snapshot when (a) no prior stage exists, (b) the diff
459 // ratio is over the threshold, or (c) the delta chain is
460 // already at its cap. The decision is internal — callers
461 // see the same `Stage` object on `get_ast` regardless.
462 if !ast_path.exists() && !delta_path.exists() {
463 self.persist_stage_bytes(&sig, &stage_id, stage, &ast_path, &delta_path)?;
464 }
465 if !meta_path.exists() {
466 let signature = signer.map(|kp| kp.sign_stage_id(&stage_id));
467 let metadata = Metadata {
468 stage_id: stage_id.clone(),
469 sig_id: sig.clone(),
470 name,
471 published_at: Self::now(),
472 note: None,
473 signature,
474 };
475 write_canonical_json(&meta_path, &metadata)?;
476 }
477
478 // Lifecycle: append a Draft transition for first publish.
479 let mut life = self.read_lifecycle(&sig).unwrap_or_else(|_| Lifecycle {
480 sig_id: sig.clone(),
481 ..Default::default()
482 });
483 if !life.transitions.iter().any(|t| t.stage_id == stage_id) {
484 life.transitions.push(Transition {
485 stage_id: stage_id.clone(),
486 from: StageStatus::Draft, // synthesized; "from" of first transition is itself
487 to: StageStatus::Draft,
488 at: Self::now(),
489 reason: None,
490 });
491 self.write_lifecycle(&sig, &life)?;
492 // Register the new stage_id's owning sig up front so a
493 // later `lookup_lifecycle` (e.g. `get_ast`) never needs
494 // to fall back to a full tenant-wide scan for it.
495 self.append_stage_index_entry(&stage_id, &sig);
496 }
497 Ok(stage_id)
498 }
499
500 // ---- lifecycle ----
501
502 pub fn activate(&self, stage_id: &str) -> Result<(), StoreError> {
503 let (sig, mut life) = self.lookup_lifecycle(stage_id)?;
504 // Demote any currently-Active impls for this SigId to Deprecated.
505 let active = life.current_active().map(|s| s.to_string());
506 if let Some(prev) = active {
507 if prev != stage_id {
508 life.transitions.push(Transition {
509 stage_id: prev,
510 from: StageStatus::Active,
511 to: StageStatus::Deprecated,
512 at: Self::now(),
513 reason: Some("superseded".into()),
514 });
515 }
516 }
517 let cur = life.status_of(stage_id);
518 if cur == Some(StageStatus::Tombstone) {
519 return Err(StoreError::InvalidTransition(
520 "tombstoned cannot be activated".into(),
521 ));
522 }
523 life.transitions.push(Transition {
524 stage_id: stage_id.into(),
525 from: cur.unwrap_or(StageStatus::Draft),
526 to: StageStatus::Active,
527 at: Self::now(),
528 reason: None,
529 });
530 self.write_lifecycle(&sig, &life)
531 }
532
533 pub fn deprecate(&self, stage_id: &str, reason: impl Into<String>) -> Result<(), StoreError> {
534 let (sig, mut life) = self.lookup_lifecycle(stage_id)?;
535 let cur = life
536 .status_of(stage_id)
537 .ok_or_else(|| StoreError::UnknownStage(stage_id.into()))?;
538 if cur != StageStatus::Active {
539 return Err(StoreError::InvalidTransition(format!(
540 "{cur:?} ⇒ Deprecated"
541 )));
542 }
543 life.transitions.push(Transition {
544 stage_id: stage_id.into(),
545 from: cur,
546 to: StageStatus::Deprecated,
547 at: Self::now(),
548 reason: Some(reason.into()),
549 });
550 self.write_lifecycle(&sig, &life)
551 }
552
553 pub fn tombstone(&self, stage_id: &str) -> Result<(), StoreError> {
554 let (sig, mut life) = self.lookup_lifecycle(stage_id)?;
555 let cur = life
556 .status_of(stage_id)
557 .ok_or_else(|| StoreError::UnknownStage(stage_id.into()))?;
558 if cur != StageStatus::Deprecated {
559 return Err(StoreError::InvalidTransition(format!(
560 "{cur:?} ⇒ Tombstone"
561 )));
562 }
563 life.transitions.push(Transition {
564 stage_id: stage_id.into(),
565 from: cur,
566 to: StageStatus::Tombstone,
567 at: Self::now(),
568 reason: None,
569 });
570 self.write_lifecycle(&sig, &life)
571 }
572
573 // ---- queries ----
574
575 /// The current Active StageId for a signature, or `None`.
576 pub fn resolve_sig(&self, sig: &str) -> Result<Option<String>, StoreError> {
577 let life = match self.read_lifecycle(sig) {
578 Ok(l) => l,
579 Err(_) => return Ok(None),
580 };
581 Ok(life.current_active().map(|s| s.to_string()))
582 }
583
584 /// Per-stage history for a SigId, ordered chronologically by
585 /// the *last* transition timestamp. Returns one entry per
586 /// distinct StageId that has ever been published under `sig`.
587 /// `Ok(vec![])` if the SigId doesn't exist in the store.
588 ///
589 /// Used by `lex blame` to render "where does this fn come from".
590 pub fn sig_history(&self, sig: &str) -> Result<Vec<StageHistoryEntry>, StoreError> {
591 let life = match self.read_lifecycle(sig) {
592 Ok(l) => l,
593 Err(_) => return Ok(Vec::new()),
594 };
595 // Collapse transitions: latest status + last_at per stage,
596 // plus the timestamp of the first Draft transition (≈ when
597 // the stage was published) when one exists.
598 let mut by_stage: indexmap::IndexMap<String, StageHistoryEntry> = indexmap::IndexMap::new();
599 for t in &life.transitions {
600 let entry = by_stage
601 .entry(t.stage_id.clone())
602 .or_insert(StageHistoryEntry {
603 stage_id: t.stage_id.clone(),
604 status: t.to,
605 last_at: t.at,
606 published_at: None,
607 });
608 entry.status = t.to;
609 entry.last_at = t.at;
610 if t.from == StageStatus::Draft && entry.published_at.is_none() {
611 entry.published_at = Some(t.at);
612 }
613 if t.to == StageStatus::Draft && entry.published_at.is_none() {
614 // Initial publication: Draft is the *destination*.
615 entry.published_at = Some(t.at);
616 }
617 }
618 let mut out: Vec<StageHistoryEntry> = by_stage.into_values().collect();
619 // Sort newest first so `lex blame` shows recent activity at top.
620 out.sort_by_key(|e| std::cmp::Reverse(e.last_at));
621 Ok(out)
622 }
623
624 pub fn get_ast(&self, stage_id: &str) -> Result<Stage, StoreError> {
625 let (sig, _) = self.lookup_lifecycle(stage_id)?;
626 let bytes = self.read_stage_canonical_bytes(&sig, stage_id)?;
627 Ok(serde_json::from_slice(&bytes)?)
628 }
629
630 /// Bulk AST fetch for callers that already know each stage's
631 /// **signature** — a branch head map, for instance, which is keyed
632 /// by SigId and whose values are the StageIds it points at.
633 ///
634 /// Prefer this over [`Self::get_asts_bulk`] whenever the SigId is in
635 /// hand, because resolving a StageId back to a SigId is not
636 /// reliable: a StageId hashes the structural signature plus the
637 /// implementation, deliberately *not* the name
638 /// (`docs/INVARIANTS.md`), so two functions that differ only in name
639 /// share one StageId while having two distinct SigIds — and two
640 /// separate ASTs, one under each sig directory. `stage_index` maps
641 /// each StageId to a single sig, so `get_ast`/`get_asts_bulk` return
642 /// whichever of those ASTs the index happens to name, i.e. the wrong
643 /// name half the time (#826). Reading straight from the sig the
644 /// caller already knows removes the ambiguity — and skips loading
645 /// the index at all.
646 ///
647 /// Returns results in the same order as `pairs`, `Err` for anything
648 /// that fails to resolve (mirroring `get_ast`'s error semantics).
649 pub fn get_asts_for_sigs_bulk(
650 &self,
651 pairs: &[(String, String)],
652 ) -> Vec<Result<Stage, StoreError>> {
653 pairs
654 .iter()
655 .map(|(sig_id, stage_id)| {
656 let bytes = self.read_stage_canonical_bytes(sig_id, stage_id)?;
657 Ok(serde_json::from_slice(&bytes)?)
658 })
659 .collect()
660 }
661
662 /// Bulk variant of [`Self::get_ast`] for callers resolving many
663 /// stage_ids at once (e.g. `pkg_publish_handler`'s `old_head`
664 /// scan over every live function in a tenant, once per publish
665 /// request). `get_ast` in a loop calls `lookup_lifecycle` once
666 /// per stage_id, and `lookup_lifecycle`'s index-hit path reads
667 /// and re-parses the *entire* `stage_index.jsonl` on every single
668 /// call — fine for one call, but O(index size × N) for N calls in
669 /// a row, which dominates once the index itself is large (#825's
670 /// follow-up: still correct and far better than the pre-index
671 /// full-tenant-scan-per-call behavior, but the per-call reparse
672 /// is itself a real, measured cost — 87.6s for 3,664 calls against
673 /// a ~14k-line index on the alpibrusl tenant).
674 ///
675 /// This loads the index once for the whole batch and keeps it in
676 /// memory across all `stage_ids`, only touching disk again to
677 /// append genuinely new entries (a positive backfill or a
678 /// negative "not found anywhere" cache, same as the single-call
679 /// path) — never to re-read what's already loaded.
680 ///
681 /// Returns results in the same order as `stage_ids`, `Err` for
682 /// anything that fails to resolve (mirroring `get_ast`'s error
683 /// semantics per call).
684 pub fn get_asts_bulk(&self, stage_ids: &[String]) -> Vec<Result<Stage, StoreError>> {
685 let mut index = self.load_stage_index();
686 let mut sigs_cache: BTreeMap<String, Option<Lifecycle>> = BTreeMap::new();
687 let mut all_sigs: Option<Vec<String>> = None;
688
689 stage_ids
690 .iter()
691 .map(|stage_id| {
692 self.lookup_lifecycle_bulk(stage_id, &mut index, &mut sigs_cache, &mut all_sigs)
693 .and_then(|sig| {
694 let bytes = self.read_stage_canonical_bytes(&sig, stage_id)?;
695 Ok(serde_json::from_slice(&bytes)?)
696 })
697 })
698 .collect()
699 }
700
701 /// Shared implementation behind [`Self::get_asts_bulk`]: identical
702 /// logic to [`Self::lookup_lifecycle`], but reads and writes the
703 /// caller-supplied `index` map instead of reloading it from disk
704 /// on every call, and memoizes `read_lifecycle` per sig and the
705 /// `list_sigs()` full-scan list across the whole batch. Disk
706 /// writes for newly-discovered entries (positive or negative)
707 /// still happen immediately, same as the single-call path — only
708 /// the repeated *reads* are batched away.
709 fn lookup_lifecycle_bulk(
710 &self,
711 stage_id: &str,
712 index: &mut BTreeMap<String, String>,
713 sigs_cache: &mut BTreeMap<String, Option<Lifecycle>>,
714 all_sigs: &mut Option<Vec<String>>,
715 ) -> Result<String, StoreError> {
716 if let Some(sig) = index.get(stage_id) {
717 if sig == MISSING_STAGE_MARKER {
718 return Err(StoreError::UnknownStage(stage_id.into()));
719 }
720 let life = sigs_cache
721 .entry(sig.clone())
722 .or_insert_with(|| self.read_lifecycle(sig).ok());
723 if let Some(life) = life {
724 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
725 return Ok(sig.clone());
726 }
727 }
728 }
729 if all_sigs.is_none() {
730 *all_sigs = Some(self.list_sigs()?);
731 }
732 for sig in all_sigs.as_ref().unwrap() {
733 let life = sigs_cache
734 .entry(sig.clone())
735 .or_insert_with(|| self.read_lifecycle(sig).ok());
736 if let Some(life) = life {
737 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
738 self.append_stage_index_entry(stage_id, sig);
739 index.insert(stage_id.to_string(), sig.clone());
740 return Ok(sig.clone());
741 }
742 }
743 }
744 self.append_stage_index_entry(stage_id, MISSING_STAGE_MARKER);
745 index.insert(stage_id.to_string(), MISSING_STAGE_MARKER.to_string());
746 Err(StoreError::UnknownStage(stage_id.into()))
747 }
748
749 /// Read the canonical bytes of a stage, walking back through
750 /// any delta chain (#261 slice 3). The recursion ends at a
751 /// `<stage_id>.ast.json` file (a full snapshot) or, in the
752 /// degenerate case of a missing chain, with `UnknownStage`.
753 fn read_stage_canonical_bytes(&self, sig: &str, stage_id: &str) -> Result<Vec<u8>, StoreError> {
754 let ast_path = self.impl_dir(sig).join(format!("{}.ast.json", stage_id));
755 if ast_path.exists() {
756 return Ok(fs::read(&ast_path)?);
757 }
758 let delta_path = self.impl_dir(sig).join(format!("{}.delta.json", stage_id));
759 if !delta_path.exists() {
760 return Err(StoreError::UnknownStage(stage_id.into()));
761 }
762 let delta_bytes = fs::read(&delta_path)?;
763 let delta: crate::delta::StageDelta = serde_json::from_slice(&delta_bytes)?;
764 let base_bytes = self.read_stage_canonical_bytes(sig, &delta.base_stage_id)?;
765 crate::delta::apply(&base_bytes, &delta).map_err(|e| {
766 StoreError::Io(std::io::Error::new(
767 std::io::ErrorKind::InvalidData,
768 format!("applying delta for {stage_id}: {e}"),
769 ))
770 })
771 }
772
773 /// Persist a freshly-published stage's canonical bytes (#261
774 /// slice 3). Tries delta encoding against the most recent
775 /// prior stage in the sig's lifecycle; falls back to a full
776 /// snapshot when no base exists, the diff ratio is too high,
777 /// or the delta chain is already at its cap.
778 fn persist_stage_bytes(
779 &self,
780 sig: &str,
781 stage_id: &str,
782 stage: &Stage,
783 ast_path: &Path,
784 delta_path: &Path,
785 ) -> Result<(), StoreError> {
786 let new_bytes = canonical_bytes(stage)?;
787 if let Some((base_stage_id, base_chain_length)) = self.pick_delta_base(sig, stage_id)? {
788 let base_bytes = self.read_stage_canonical_bytes(sig, &base_stage_id)?;
789 let (prefix, suffix, middle) = crate::delta::splice(&base_bytes, &new_bytes);
790 let chain_length = base_chain_length + 1;
791 if crate::delta::is_worth_encoding(middle.len(), new_bytes.len(), chain_length) {
792 let delta = crate::delta::StageDelta {
793 base_stage_id,
794 chain_length,
795 common_prefix: prefix,
796 common_suffix: suffix,
797 middle_hex: hex::encode(&middle),
798 };
799 write_canonical_json(delta_path, &delta)?;
800 return Ok(());
801 }
802 }
803 // Fall through: full snapshot.
804 if let Some(parent) = ast_path.parent() {
805 fs::create_dir_all(parent)?;
806 }
807 fs::write(ast_path, &new_bytes)?;
808 Ok(())
809 }
810
811 /// Pick a base stage for delta encoding from the given sig's
812 /// lifecycle. Returns `(base_stage_id, base_chain_length)` for
813 /// the most-recent non-tombstoned prior stage, or `None` when
814 /// there is no candidate. The chain length is read off the
815 /// base's `.delta.json` (if any) to enforce the cap.
816 fn pick_delta_base(
817 &self,
818 sig: &str,
819 new_stage_id: &str,
820 ) -> Result<Option<(String, usize)>, StoreError> {
821 let life = self.read_lifecycle(sig).ok();
822 let Some(life) = life else {
823 return Ok(None);
824 };
825 // Walk transitions newest-first; pick the first prior
826 // stage that isn't this one and isn't tombstoned.
827 let mut latest_per_stage: indexmap::IndexMap<&str, StageStatus> = indexmap::IndexMap::new();
828 for t in &life.transitions {
829 latest_per_stage.insert(&t.stage_id, t.to);
830 }
831 let mut candidates: Vec<&str> = latest_per_stage
832 .iter()
833 .filter(|(id, status)| **id != new_stage_id && **status != StageStatus::Tombstone)
834 .map(|(id, _)| *id)
835 .collect();
836 // Reverse to get newest-first (transitions are append-only,
837 // so latest_per_stage's iteration order matches insertion
838 // order, oldest-first).
839 candidates.reverse();
840 let Some(&base) = candidates.first() else {
841 return Ok(None);
842 };
843 let base_chain_length = self.delta_chain_length(sig, base)?;
844 Ok(Some((base.to_string(), base_chain_length)))
845 }
846
847 /// Length of the delta chain ending at `stage_id`. Zero when
848 /// the stage is a full snapshot (`.ast.json` present); the
849 /// stored `chain_length` from `.delta.json` otherwise.
850 fn delta_chain_length(&self, sig: &str, stage_id: &str) -> Result<usize, StoreError> {
851 let ast_path = self.impl_dir(sig).join(format!("{}.ast.json", stage_id));
852 if ast_path.exists() {
853 return Ok(0);
854 }
855 let delta_path = self.impl_dir(sig).join(format!("{}.delta.json", stage_id));
856 if !delta_path.exists() {
857 return Ok(0);
858 }
859 let bytes = fs::read(&delta_path)?;
860 let delta: crate::delta::StageDelta = serde_json::from_slice(&bytes)?;
861 Ok(delta.chain_length)
862 }
863
864 pub fn get_metadata(&self, stage_id: &str) -> Result<Metadata, StoreError> {
865 let (sig, _) = self.lookup_lifecycle(stage_id)?;
866 let path = self
867 .impl_dir(&sig)
868 .join(format!("{}.metadata.json", stage_id));
869 let bytes = fs::read(&path)?;
870 Ok(serde_json::from_slice(&bytes)?)
871 }
872
873 pub fn get_status(&self, stage_id: &str) -> Result<StageStatus, StoreError> {
874 let (_sig, life) = self.lookup_lifecycle(stage_id)?;
875 life.status_of(stage_id)
876 .ok_or_else(|| StoreError::UnknownStage(stage_id.into()))
877 }
878
879 pub fn list_stages_by_name(&self, name: &str) -> Result<Vec<String>, StoreError> {
880 // Walk every SigId → check metadata of any implementation; if its
881 // name matches, include the SigId.
882 let mut out = Vec::new();
883 let stages_dir = self.root.join("stages");
884 if !stages_dir.exists() {
885 return Ok(out);
886 }
887 for entry in fs::read_dir(&stages_dir)? {
888 let entry = entry?;
889 let sig_dir = entry.path();
890 if !sig_dir.is_dir() {
891 continue;
892 }
893 let sig = entry.file_name().to_string_lossy().to_string();
894 // Look at any one metadata file under this SigId.
895 let impls = self.impl_dir(&sig);
896 if !impls.exists() {
897 continue;
898 }
899 for f in fs::read_dir(impls)? {
900 let f = f?;
901 let p = f.path();
902 if p.extension().is_some_and(|e| e == "json")
903 && p.file_name()
904 .is_some_and(|n| n.to_string_lossy().ends_with(".metadata.json"))
905 {
906 if let Ok(bytes) = fs::read(&p) {
907 if let Ok(m) = serde_json::from_slice::<Metadata>(&bytes) {
908 if m.name == name {
909 if !out.contains(&sig) {
910 out.push(sig.clone());
911 }
912 break;
913 }
914 }
915 }
916 }
917 }
918 }
919 out.sort();
920 Ok(out)
921 }
922
923 pub fn list_sigs(&self) -> Result<Vec<String>, StoreError> {
924 let stages_dir = self.root.join("stages");
925 let mut out = Vec::new();
926 if !stages_dir.exists() {
927 return Ok(out);
928 }
929 for entry in fs::read_dir(stages_dir)? {
930 let entry = entry?;
931 if entry.file_type()?.is_dir() {
932 out.push(entry.file_name().to_string_lossy().to_string());
933 }
934 }
935 out.sort();
936 Ok(out)
937 }
938
939 // ---- tests/specs as metadata (§4.4) ----
940
941 pub fn attach_test(&self, sig: &str, test: &Test) -> Result<String, StoreError> {
942 if !self.sig_dir(sig).exists() {
943 return Err(StoreError::UnknownSig(sig.into()));
944 }
945 fs::create_dir_all(self.tests_dir(sig))?;
946 let path = self.tests_dir(sig).join(format!("{}.json", test.id));
947 write_canonical_json(&path, test)?;
948 Ok(test.id.clone())
949 }
950
951 pub fn list_tests(&self, sig: &str) -> Result<Vec<Test>, StoreError> {
952 let dir = self.tests_dir(sig);
953 if !dir.exists() {
954 return Ok(Vec::new());
955 }
956 let mut out = Vec::new();
957 for f in fs::read_dir(dir)? {
958 let f = f?;
959 if f.path().extension().is_some_and(|e| e == "json") {
960 let bytes = fs::read(f.path())?;
961 out.push(serde_json::from_slice(&bytes)?);
962 }
963 }
964 Ok(out)
965 }
966
967 pub fn attach_spec(&self, sig: &str, spec: &Spec) -> Result<String, StoreError> {
968 if !self.sig_dir(sig).exists() {
969 return Err(StoreError::UnknownSig(sig.into()));
970 }
971 fs::create_dir_all(self.specs_dir(sig))?;
972 let path = self.specs_dir(sig).join(format!("{}.json", spec.id));
973 write_canonical_json(&path, spec)?;
974 Ok(spec.id.clone())
975 }
976
977 pub fn list_specs(&self, sig: &str) -> Result<Vec<Spec>, StoreError> {
978 let dir = self.specs_dir(sig);
979 if !dir.exists() {
980 return Ok(Vec::new());
981 }
982 let mut out = Vec::new();
983 for f in fs::read_dir(dir)? {
984 let f = f?;
985 if f.path().extension().is_some_and(|e| e == "json") {
986 let bytes = fs::read(f.path())?;
987 out.push(serde_json::from_slice(&bytes)?);
988 }
989 }
990 Ok(out)
991 }
992
993 // ---- traces (§4.2 / M7) ----
994
995 // Native run-trace store — gated behind the `trace` feature (depends on
996 // lex-trace). Off when a lower crate (lex-runtime) depends on lex-store to
997 // avoid a dependency cycle; the blob/stage store below is unaffected.
998 #[cfg(feature = "trace")]
999 fn trace_path(&self, run_id: &str) -> PathBuf {
1000 self.root.join("traces").join(run_id).join("trace.json")
1001 }
1002
1003 #[cfg(feature = "trace")]
1004 pub fn save_trace(&self, tree: &lex_trace::TraceTree) -> Result<String, StoreError> {
1005 let path = self.trace_path(&tree.run_id);
1006 write_canonical_json(&path, tree)?;
1007 Ok(tree.run_id.clone())
1008 }
1009
1010 #[cfg(feature = "trace")]
1011 pub fn load_trace(&self, run_id: &str) -> Result<lex_trace::TraceTree, StoreError> {
1012 let bytes = fs::read(self.trace_path(run_id))?;
1013 Ok(serde_json::from_slice(&bytes)?)
1014 }
1015
1016 pub fn list_traces(&self) -> Result<Vec<String>, StoreError> {
1017 let dir = self.root.join("traces");
1018 if !dir.exists() {
1019 return Ok(Vec::new());
1020 }
1021 let mut out = Vec::new();
1022 for entry in fs::read_dir(dir)? {
1023 let entry = entry?;
1024 if entry.file_type()?.is_dir() {
1025 out.push(entry.file_name().to_string_lossy().to_string());
1026 }
1027 }
1028 out.sort();
1029 Ok(out)
1030 }
1031
1032 // ---- internals ----
1033
1034 /// `<root>/stage_index.jsonl` — an append-only, best-effort
1035 /// reverse index (`StageId` -> owning `SigId`), one JSON object
1036 /// per line. Backs `lookup_lifecycle`'s fast path; see its doc
1037 /// comment. Not a second source of truth: every entry is
1038 /// reconstructible from `stages/<sig>/lifecycle.json`, so a
1039 /// missing, truncated, or entirely absent index file only costs
1040 /// a slower lookup (the pre-existing full scan), never
1041 /// correctness — matching this module's "filesystem is the
1042 /// source of truth" stance (see the module doc comment) rather
1043 /// than introducing an actual second database.
1044 fn stage_index_path(&self) -> PathBuf {
1045 self.root.join("stage_index.jsonl")
1046 }
1047
1048 /// Best-effort load of the whole reverse index into memory.
1049 /// Tolerates a missing file (no index yet) and a corrupt or
1050 /// torn last line (a crash mid-append under the single-writer
1051 /// Tier-1 assumption) by skipping lines that don't parse,
1052 /// rather than failing the lookup that triggered the load.
1053 fn load_stage_index(&self) -> std::collections::BTreeMap<String, String> {
1054 let mut out = std::collections::BTreeMap::new();
1055 let Ok(raw) = fs::read_to_string(self.stage_index_path()) else {
1056 return out;
1057 };
1058 for line in raw.lines() {
1059 if let Ok(entry) = serde_json::from_str::<StageIndexEntry>(line) {
1060 out.insert(entry.stage_id, entry.sig_id);
1061 }
1062 }
1063 out
1064 }
1065
1066 /// Best-effort append of one new `(stage_id, sig_id)` pair.
1067 /// Failure (e.g. a read-only filesystem) only costs a future
1068 /// full scan for this stage_id, never correctness, so it's
1069 /// swallowed rather than propagated.
1070 fn append_stage_index_entry(&self, stage_id: &str, sig: &str) {
1071 use std::io::Write;
1072 let entry = StageIndexEntry { stage_id: stage_id.into(), sig_id: sig.into() };
1073 let Ok(line) = serde_json::to_string(&entry) else { return };
1074 if let Ok(mut f) = fs::OpenOptions::new()
1075 .create(true)
1076 .append(true)
1077 .open(self.stage_index_path())
1078 {
1079 let _ = writeln!(f, "{line}");
1080 }
1081 }
1082
1083 /// Find which SigId owns a StageId, and that sig's lifecycle.
1084 ///
1085 /// Before the reverse index (#822): a full scan over *every*
1086 /// SigId in the tenant (`list_sigs()`, not scoped to the
1087 /// package being looked at), reading and parsing each one's
1088 /// `lifecycle.json` until a match turned up. `get_ast` — called
1089 /// once per pre-existing function when building a publish
1090 /// request's `old_fns_by_name` (`lex-api/src/handlers.rs`) —
1091 /// calls this once per function, so a tenant with a few thousand
1092 /// published functions turned a single publish into millions of
1093 /// individual file reads; measured at roughly an hour on the
1094 /// `alpibrusl` tenant's ~2,400-function store.
1095 ///
1096 /// Now: check the persisted reverse index first (one sequential
1097 /// file read instead of up to N separate ones). A miss — the
1098 /// index doesn't exist yet, or this stage_id predates it — falls
1099 /// back to the full scan and backfills the index so the next
1100 /// lookup for the same stage_id is fast.
1101 fn lookup_lifecycle(&self, stage_id: &str) -> Result<(String, Lifecycle), StoreError> {
1102 let index = self.load_stage_index();
1103 if let Some(sig) = index.get(stage_id) {
1104 if sig == MISSING_STAGE_MARKER {
1105 // A previous full scan already established this
1106 // stage_id exists nowhere in the store. Re-scanning
1107 // would find nothing again -- see #825: a genuinely
1108 // orphaned reference (e.g. from data predating some
1109 // store migration) is looked up on *every* call that
1110 // needs it, forever, so without this negative cache
1111 // it silently costs a full O(total sigs) scan each
1112 // time, indistinguishable from the positive case at
1113 // the call site. Measured directly: on the alpibrusl
1114 // tenant, 988 of 3,664 branch-head entries are
1115 // orphaned this way, turning one `pkg publish`'s
1116 // old_fns_by_name build into ~16M wasted lifecycle
1117 // reads.
1118 return Err(StoreError::UnknownStage(stage_id.into()));
1119 }
1120 if let Ok(life) = self.read_lifecycle(sig) {
1121 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
1122 return Ok((sig.clone(), life));
1123 }
1124 }
1125 // Index entry is stale or wrong (shouldn't happen in
1126 // practice — sig ownership of a stage_id is permanent).
1127 // Fall through to the full scan below rather than trust it.
1128 }
1129 for sig in self.list_sigs()? {
1130 if let Ok(life) = self.read_lifecycle(&sig) {
1131 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
1132 self.append_stage_index_entry(stage_id, &sig);
1133 return Ok((sig, life));
1134 }
1135 }
1136 }
1137 // Genuinely not found anywhere: cache that fact so the next
1138 // lookup for this exact stage_id is an index hit, not another
1139 // full scan. Safe even if this stage_id somehow gets a real
1140 // sig later (content-addressed publish is idempotent, so
1141 // "later" only means "a byte-identical stage republished
1142 // under a real sig") — `append_stage_index_entry`'s later,
1143 // real entry is a later line in the file, and `load_stage_index`
1144 // folds duplicate keys last-write-wins, so the real entry wins.
1145 self.append_stage_index_entry(stage_id, MISSING_STAGE_MARKER);
1146 Err(StoreError::UnknownStage(stage_id.into()))
1147 }
1148
1149 fn read_lifecycle(&self, sig: &str) -> Result<Lifecycle, StoreError> {
1150 let path = self.lifecycle_path(sig);
1151 if !path.exists() {
1152 return Ok(Lifecycle {
1153 sig_id: sig.into(),
1154 transitions: Vec::new(),
1155 });
1156 }
1157 let bytes = fs::read(&path)?;
1158 Ok(serde_json::from_slice(&bytes)?)
1159 }
1160
1161 fn write_lifecycle(&self, sig: &str, life: &Lifecycle) -> Result<(), StoreError> {
1162 write_canonical_json(&self.lifecycle_path(sig), life)
1163 }
1164
1165 /// Apply a published program to a branch as a sequence of typed
1166 /// operations. Returns the ordered list of op_ids + the new
1167 /// head_op. The caller (`lex publish` CLI, `lex serve`'s HTTP
1168 /// handler) is responsible for computing the `DiffReport` against
1169 /// the current branch head — the diff infrastructure lives in
1170 /// `lex-vcs::compute_diff` (previously `lex-cli`) to keep this
1171 /// layer from owning diffing logic.
1172 ///
1173 /// On success: every op in the returned list is durable in the
1174 /// op log and the branch's head_op points at the last one.
1175 /// On a no-op (no diff): returns empty `ops` and the existing
1176 /// `head_op` unchanged.
1177 pub fn publish_program(
1178 &self,
1179 branch: &str,
1180 stages: &[lex_ast::Stage],
1181 diff: &lex_vcs::DiffReport,
1182 new_imports: &lex_vcs::ImportMap,
1183 activate: bool,
1184 ) -> Result<PublishOutcome, StoreError> {
1185 self.publish_program_signed(branch, stages, diff, new_imports, activate, None)
1186 }
1187
1188 /// Signed variant of [`Self::publish_program`] (#227). Every
1189 /// stage written under this batch gets the same signer; per-stage
1190 /// keys aren't supported because the agent identity model treats
1191 /// a publish as a single authorial act.
1192 pub fn publish_program_signed(
1193 &self,
1194 branch: &str,
1195 stages: &[lex_ast::Stage],
1196 diff: &lex_vcs::DiffReport,
1197 new_imports: &lex_vcs::ImportMap,
1198 activate: bool,
1199 signer: Option<&lex_vcs::Keypair>,
1200 ) -> Result<PublishOutcome, StoreError> {
1201 use std::collections::{BTreeMap, BTreeSet};
1202
1203 // #130's write-time gate: verify the candidate program
1204 // typechecks (and effects are correctly declared) before
1205 // any disk side-effect. If anything fails, return the
1206 // structured envelope and leave the branch head unchanged
1207 // — the store's "always-valid HEAD" invariant only holds
1208 // because this is the only batch-publish path that
1209 // advances heads. Single-op writes via the lower-level
1210 // `apply_operation` are not gated yet (#130 follow-up).
1211 if let Err(errors) = lex_types::check_program(stages) {
1212 return Err(StoreError::TypeError(errors));
1213 }
1214
1215 // Build old-side views from the current branch. There used to be
1216 // an `old_name_to_sig: BTreeMap<String, SigId>` built here too,
1217 // keyed by bare function name — but a bare name is not unique
1218 // across a package's files (#818: two files can legitimately
1219 // both declare a local `validate` helper with different
1220 // signatures), so a name-keyed map silently collapsed distinct
1221 // SigIds onto one. `diff` now carries each entry's own resolved
1222 // `old_sig_id` directly (see `diff_report`'s doc comments), so
1223 // `diff_to_ops` no longer needs this lookup at all.
1224 let old_head = self.branch_head(branch)?;
1225 // Read every live function's effects through the SigId the head
1226 // names, in one batch. Two reasons, both load-bearing:
1227 //
1228 // * Cost. This was a `get_ast` per live function, and
1229 // `get_ast`'s index-hit path re-reads and re-parses the whole
1230 // `stage_index.jsonl` on every call — O(index × live fns) per
1231 // publish, paid again for every `publish_program` call a
1232 // multi-file publish makes (#828; measured 34s for a no-op
1233 // republish of a real 21-file package against only 698 live
1234 // functions, nearly all of it here).
1235 // * Correctness. A StageId is name-independent, so two live
1236 // functions differing only in name share one and the index
1237 // maps it to a single sig — resolving by StageId therefore
1238 // attributed one function's effects to the *other* one's sig,
1239 // the same ambiguity #826 fixed in `pkg_publish_handler`.
1240 let head_pairs: Vec<(String, String)> = old_head
1241 .iter()
1242 .map(|(sig, stage)| (sig.clone(), stage.clone()))
1243 .collect();
1244 let old_effects: BTreeMap<String, BTreeSet<String>> = head_pairs
1245 .iter()
1246 .zip(self.get_asts_for_sigs_bulk(&head_pairs))
1247 .filter_map(|((sig, _), ast)| match ast.ok()? {
1248 lex_ast::Stage::FnDecl(fd) => {
1249 let s: BTreeSet<String> =
1250 fd.effects.iter().map(|e| e.name.clone()).collect();
1251 Some((sig.clone(), s))
1252 }
1253 _ => None,
1254 })
1255 .collect();
1256 let old_imports = self.derive_imports_from_oplog(branch)?;
1257
1258 let op_kinds = lex_vcs::diff_to_ops(lex_vcs::DiffInputs {
1259 old_head: &old_head,
1260 old_effects: &old_effects,
1261 old_imports: &old_imports,
1262 new_stages: stages,
1263 new_imports,
1264 diff,
1265 })
1266 .map_err(|e| StoreError::InvalidTransition(format!("diff_to_ops: {e}")))?;
1267
1268 let mut ops_out: Vec<PublishOp> = Vec::new();
1269 let mut last_op_id: Option<lex_vcs::OpId> = None;
1270 for kind in op_kinds {
1271 // Persist the underlying stage AST/metadata if this op
1272 // produces or replaces one.
1273 if let Some(stg) = stage_for_kind(&kind, stages) {
1274 if !matches!(stg, lex_ast::Stage::Import(_)) {
1275 self.publish_signed(stg, signer)?;
1276 if activate {
1277 if let Some(stage_id_str) = stage_id(stg) {
1278 let _ = self.activate(&stage_id_str);
1279 }
1280 }
1281 }
1282 }
1283 let transition = transition_for_kind(&kind);
1284 let attestable = attestable_stage_ids(&transition);
1285 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
1286 let op =
1287 lex_vcs::Operation::new(kind.clone(), head_now.into_iter().collect::<Vec<_>>());
1288 let op_id = self.apply_operation(branch, op, transition)?;
1289 self.record_typecheck_passed(&attestable, &op_id)?;
1290 ops_out.push(PublishOp {
1291 op_id: op_id.clone(),
1292 kind: serde_json::to_value(&kind).map_err(StoreError::Serde)?,
1293 });
1294 last_op_id = Some(op_id);
1295 }
1296
1297 let head_op = match last_op_id {
1298 Some(id) => Some(id),
1299 // No ops applied; return whatever the head was already.
1300 None => self.get_branch(branch)?.and_then(|b| b.head_op),
1301 };
1302
1303 Ok(PublishOutcome {
1304 ops: ops_out,
1305 head_op,
1306 })
1307 }
1308
1309 pub fn derive_imports_from_oplog(
1310 &self,
1311 branch: &str,
1312 ) -> Result<lex_vcs::ImportMap, StoreError> {
1313 use lex_vcs::OperationKind::*;
1314 let log = lex_vcs::OpLog::open(self.root())?;
1315 let head = match self.get_branch(branch)?.and_then(|b| b.head_op) {
1316 Some(h) => h,
1317 None => return Ok(Default::default()),
1318 };
1319 let mut out: lex_vcs::ImportMap = Default::default();
1320 for r in log.walk_forward(&head, None)? {
1321 match r.op.kind {
1322 AddImport { in_file, module } => {
1323 out.entry(in_file).or_default().insert(module);
1324 }
1325 RemoveImport { in_file, module } => {
1326 if let Some(set) = out.get_mut(&in_file) {
1327 set.remove(&module);
1328 }
1329 }
1330 _ => {}
1331 }
1332 }
1333 Ok(out)
1334 }
1335
1336 /// Apply an operation to a branch and advance its head_op.
1337 ///
1338 /// The single advance path. Validates parents via `lex_vcs::apply`,
1339 /// persists the operation via the op log, then atomically advances
1340 /// the branch file's head_op via `set_branch_head_op`.
1341 ///
1342 /// Errors:
1343 /// - `UnknownBranch`: branch does not exist (no op is persisted).
1344 /// - `Apply(ApplyError::StaleParent)`: the op's parents don't
1345 /// match the branch head — head is unchanged. Callers that
1346 /// want retry-on-stale (e.g. `lex publish` re-running against
1347 /// a moved head) match on this variant explicitly.
1348 /// - `Apply(ApplyError::UnknownMergeParent)`: a merge op's
1349 /// second parent isn't in the log.
1350 /// - `Io`: filesystem error during persist or branch advance.
1351 ///
1352 /// Crash recovery: between op persist and branch advance, a crash
1353 /// can leave an orphan op record in the log with no branch
1354 /// pointing at it. The op is content-addressed and cheap to
1355 /// re-derive from the same source. See
1356 /// Apply a single op against `branch`, gated on the candidate
1357 /// program typechecking. The per-op variant of #130's
1358 /// write-time gate — counterpart to [`Self::publish_program`]'s
1359 /// batch-mode check.
1360 ///
1361 /// `candidate` is the sequence of `Stage`s that *would* exist
1362 /// on this branch after the op is applied. Caller's
1363 /// responsibility: today neither `lex-store` nor `lex-vcs`
1364 /// reconstruct the candidate from the op + branch state on
1365 /// behalf of the caller. The natural callers (HTTP `POST
1366 /// /v1/publish` for a single op; agent harnesses driving
1367 /// merges via the future #134 API) already have the candidate
1368 /// in memory.
1369 ///
1370 /// On rejection: branch head unchanged, no op record persisted.
1371 /// Same atomicity guarantee as the publish path.
1372 ///
1373 /// # Why a separate method, not a flag on `apply_operation`
1374 ///
1375 /// `apply_operation` accepting `Option<&[Stage]>` and silently
1376 /// skipping the gate on `None` is exactly the kind of
1377 /// "secretly opt-out" path #130 is trying to remove. The honest
1378 /// split: `apply_operation` for the one caller that already
1379 /// typechecked its input up front (`publish_program`),
1380 /// `apply_operation_checked` for callers holding the candidate,
1381 /// [`Self::apply_operation_gated`] for single-parent callers
1382 /// that hold only the transition (`/v1/patch`), and
1383 /// [`Self::apply_merge_op_gated`] for merge commits (#833).
1384 pub fn apply_operation_checked(
1385 &self,
1386 branch: &str,
1387 op: lex_vcs::Operation,
1388 transition: lex_vcs::StageTransition,
1389 candidate: &[lex_ast::Stage],
1390 ) -> Result<lex_vcs::OpId, StoreError> {
1391 if let Err(errors) = lex_types::check_program(candidate) {
1392 // #281: emit a `RepairHint` attestation against each
1393 // candidate stage the transition was about to produce.
1394 // The op record itself isn't persisted (the gate is
1395 // pre-persistence), but the candidate stage IS — the
1396 // transform-flow methods publish before this call.
1397 // The attached hint lets `lex repair <op_id>` and
1398 // future LLM-assisted apply paths read the structured
1399 // errors without re-running the typecheck.
1400 let attestable = attestable_stage_ids(&transition);
1401 let failed_op_id = op.op_id();
1402 let _ = self.record_repair_hint(&attestable, &failed_op_id, &errors);
1403 return Err(StoreError::TypeError(errors));
1404 }
1405 // #292 slice 3: per-session budget gate. After typecheck
1406 // passes, refuse the op if it would push its session's
1407 // monotonic spend over the configured cap. Sessions
1408 // without an intent_id, or with an intent whose session
1409 // has no cap configured, sail through.
1410 self.check_session_budget(&op)?;
1411 let attestable = attestable_stage_ids(&transition);
1412 let op_effects = op_declared_effects(&op.kind);
1413 // #262: CAS retry loop. Single-parent ops can be safely
1414 // re-persisted under a new parent on contention (the kind
1415 // is invariant; only `parents` changes). Merge ops (already
1416 // 2-parent) come through the merge engine which has its own
1417 // coordination; we don't retry them here — we'll see the
1418 // first attempt's CAS fail and surface Contention.
1419 self.cas_retry_advance(branch, op, transition, |new_head| {
1420 self.record_typecheck_passed(&attestable, &new_head.op_id)?;
1421 self.run_required_attestations_gate(branch, &new_head.op_id, &attestable, &op_effects)
1422 })
1423 }
1424
1425 /// The program that would exist on `branch` after `transition`
1426 /// is applied: the branch head (snapshot-cached) with the
1427 /// transition replayed over it, every resulting `(sig, stage)`
1428 /// bulk-loaded. Exact for a **single-parent** transition — the
1429 /// candidate [`Self::apply_operation_gated`] wants. Not valid for
1430 /// a merge: a `StageTransition::Merge` records only the delta
1431 /// relative to dst, while the op-DAG replay that computes a
1432 /// merge's real head walks both parents (#833).
1433 pub fn candidate_program_for(
1434 &self,
1435 branch: &str,
1436 transition: &lex_vcs::StageTransition,
1437 ) -> Result<Vec<Stage>, StoreError> {
1438 let mut head = self.branch_head(branch)?;
1439 crate::branches::apply_transition(&mut head, transition);
1440 let pairs: Vec<(String, String)> = head.into_iter().collect();
1441 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect()
1442 }
1443
1444 /// [`Self::apply_operation_checked`] for a **single-parent** op
1445 /// where the caller holds only the transition: assembles the
1446 /// candidate via [`Self::candidate_program_for`] and runs the
1447 /// gate. Same rejection semantics — `TypeError`, a `RepairHint`
1448 /// attestation, head unchanged, nothing persisted. This is the
1449 /// write path for `/v1/patch` (#833). Merge ops must not use it
1450 /// (see `candidate_program_for`); they go through
1451 /// [`Self::apply_merge_op_gated`].
1452 pub fn apply_operation_gated(
1453 &self,
1454 branch: &str,
1455 op: lex_vcs::Operation,
1456 transition: lex_vcs::StageTransition,
1457 ) -> Result<lex_vcs::OpId, StoreError> {
1458 debug_assert!(
1459 op.parents.len() <= 1,
1460 "apply_operation_gated is single-parent only; merges use apply_merge_op_gated"
1461 );
1462 let candidate = self.candidate_program_for(branch, &transition)?;
1463 self.apply_operation_checked(branch, op, transition, &candidate)
1464 }
1465
1466 /// The gated write path for **merge** commits (`commit_merge`,
1467 /// `POST /v1/merge/<id>/commit`, `lex merge commit`).
1468 ///
1469 /// A `StageTransition::Merge` records only the delta relative to
1470 /// dst; the sig->stage map every consumer reads is recomputed by
1471 /// replaying the op DAG, which for a merge walks *both* parents
1472 /// and can surface sigs the delta never mentions. So the only way
1473 /// to know the true post-merge program is to replay it — land the
1474 /// op and read `branch_head`. This lands the merge op,
1475 /// type-checks the resulting head, and on a failure rolls the
1476 /// head back and returns `TypeError`.
1477 ///
1478 /// Before #833 the merge paths landed through the ungated
1479 /// `apply_operation`, so a merge whose result didn't compose
1480 /// (e.g. dst still calls `helper`, an agent-supplied resolution
1481 /// dropped it) advanced the head with nothing to catch it.
1482 ///
1483 /// Rollback leaves the rejected merge op as an unreachable record
1484 /// (reclaimed by `lex op gc`, the same orphan crash-recovery
1485 /// already tolerates). A stage the merge names that was never
1486 /// published surfaces as the underlying `StoreError` from the
1487 /// bulk read — the "never advance onto content that can't be
1488 /// loaded" invariant from the other side.
1489 pub fn apply_merge_op_gated(
1490 &self,
1491 branch: &str,
1492 op: lex_vcs::Operation,
1493 transition: lex_vcs::StageTransition,
1494 ) -> Result<lex_vcs::OpId, StoreError> {
1495 let head_before = self.get_branch(branch)?.and_then(|b| b.head_op);
1496 // Capture the stages this merge introduces before `transition`
1497 // is moved into `apply_operation`; used for the TypeCheck
1498 // attestation below.
1499 let attestable = attestable_stage_ids(&transition);
1500 let op_id = self.apply_operation(branch, op, transition)?;
1501
1502 let verdict = (|| -> Result<(), StoreError> {
1503 let head = self.branch_head(branch)?;
1504 let pairs: Vec<(String, String)> = head.into_iter().collect();
1505 let stages: Vec<Stage> =
1506 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
1507 if let Err(errors) = lex_types::check_program(&stages) {
1508 return Err(StoreError::TypeError(errors));
1509 }
1510 Ok(())
1511 })();
1512
1513 if let Err(e) = verdict {
1514 // Roll the head back. The empty-dst case never reaches
1515 // here (it fast-forwards without a merge op), so
1516 // `head_before` is always `Some` on this arm.
1517 if let Some(prev) = head_before {
1518 self.set_branch_head_op(branch, prev)?;
1519 }
1520 return Err(e);
1521 }
1522 // #835: the merge's post-merge head type-checked, but until now
1523 // that verdict left no trace in the attestation log — so a
1524 // merged stage looked un-type-checked to `lex blame
1525 // --with-evidence` and the attestation queries, unlike a
1526 // published or patched stage. Emit `TypeCheck::Passed` for the
1527 // stages the merge introduced, mirroring the publish / patch
1528 // paths (`record_typecheck_passed`). Emitted only after the
1529 // check passes and the head is committed, so a rolled-back
1530 // merge records nothing.
1531 self.record_typecheck_passed(&attestable, &op_id)?;
1532 Ok(op_id)
1533 }
1534
1535 /// Type-check the program that would result from overlaying a merge
1536 /// `delta` onto `branch`'s current head — **without moving the
1537 /// head** (#834). `delta` maps `sig_id -> Some(stage)` to set that
1538 /// sig to `stage`, or `sig_id -> None` to remove it, exactly the
1539 /// `entries` a `StageTransition::Merge` records.
1540 ///
1541 /// This is the read-only, resolve-time counterpart of
1542 /// `apply_merge_op_gated`'s commit-time gate: it lets a merge
1543 /// session tell an agent *which resolution broke type-checking* the
1544 /// moment it is submitted, instead of only after a failed commit.
1545 /// `Ok(())` means the projected program composes; a type failure is
1546 /// `Err(StoreError::TypeError(..))`; a read failure is the
1547 /// corresponding `StoreError` I/O variant.
1548 pub fn typecheck_merge_projection(
1549 &self,
1550 branch: &str,
1551 delta: &std::collections::BTreeMap<String, Option<String>>,
1552 ) -> Result<(), StoreError> {
1553 let mut head = self.branch_head(branch)?;
1554 for (sig, stage) in delta {
1555 match stage {
1556 Some(s) => { head.insert(sig.clone(), s.clone()); }
1557 None => { head.remove(sig); }
1558 }
1559 }
1560 let pairs: Vec<(String, String)> = head.into_iter().collect();
1561 let stages: Vec<Stage> =
1562 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
1563 if let Err(errors) = lex_types::check_program(&stages) {
1564 return Err(StoreError::TypeError(errors));
1565 }
1566 Ok(())
1567 }
1568
1569 /// #838: attempt a typed three-way merge of a single sig's body for
1570 /// a `ModifyModify` conflict — the intra-function, better-than-git
1571 /// case where two agents edited *disjoint* subtrees of the same
1572 /// function (different match arms, different let bindings).
1573 ///
1574 /// `base` / `ours` (the dst side) / `theirs` (the src side) are the
1575 /// three stage ids the merge engine surfaced for `sig_id`. Loads
1576 /// the three `FnDecl`s, structurally merges the bodies
1577 /// ([`lex_vcs::merge_bodies`]), and accepts the result *only if* the
1578 /// merged function also type-checks against `dst_branch`'s head — a
1579 /// body that composes syntactically but not by type is still a
1580 /// conflict (#838). On success the merged stage is published
1581 /// (content-addressed, idempotent; orphaned and GC-reclaimable if
1582 /// the merge is never committed) and its id returned; `None` means
1583 /// "fall back to a whole-function conflict."
1584 ///
1585 /// Deliberately narrow for this slice: only pure body divergence is
1586 /// merged. If the two sides disagree on anything but the body
1587 /// (examples, type params — the signature is identical by
1588 /// construction, since all three share `sig_id`), or either stage
1589 /// isn't a function, it falls back to a conflict.
1590 pub fn try_semantic_body_merge(
1591 &self,
1592 dst_branch: &str,
1593 sig_id: &str,
1594 base: &str,
1595 ours: &str,
1596 theirs: &str,
1597 ) -> Result<Option<String>, StoreError> {
1598 use lex_ast::Stage::FnDecl;
1599 let (base_fd, ours_fd, theirs_fd) =
1600 match (self.get_ast(base), self.get_ast(ours), self.get_ast(theirs)) {
1601 (Ok(FnDecl(b)), Ok(FnDecl(o)), Ok(FnDecl(t))) => (b, o, t),
1602 // A non-function stage (type decl / import) or a stage
1603 // that can't be loaded isn't an intra-body merge.
1604 _ => return Ok(None),
1605 };
1606
1607 // Only the body may diverge between the two sides.
1608 if !fndecl_same_except_body(&ours_fd, &theirs_fd) {
1609 return Ok(None);
1610 }
1611
1612 let merged_body =
1613 match lex_vcs::merge_bodies(&base_fd.body, &ours_fd.body, &theirs_fd.body) {
1614 lex_vcs::BodyMerge::Merged(b) => b,
1615 lex_vcs::BodyMerge::Conflict => return Ok(None),
1616 };
1617
1618 let mut merged_fd = ours_fd.clone();
1619 merged_fd.body = merged_body;
1620 let merged_stage = lex_ast::Stage::FnDecl(merged_fd);
1621 let new_stage_id = match stage_id(&merged_stage) {
1622 Some(id) => id,
1623 None => return Ok(None),
1624 };
1625
1626 // Type-check the merged fn in context: dst's head with this sig
1627 // swapped to the merged stage. Requires the merged stage to be
1628 // loadable, so publish first (idempotent, content-addressed).
1629 self.publish(&merged_stage)?;
1630 let mut delta = std::collections::BTreeMap::new();
1631 delta.insert(sig_id.to_string(), Some(new_stage_id.clone()));
1632 match self.typecheck_merge_projection(dst_branch, &delta) {
1633 Ok(()) => Ok(Some(new_stage_id)),
1634 // Composes syntactically, not by type → still a conflict.
1635 Err(StoreError::TypeError(_)) => Ok(None),
1636 Err(e) => Err(e),
1637 }
1638 }
1639
1640 /// Open the attestation log rooted at this store. The log lives
1641 /// under `<root>/attestations/`; opening is idempotent and cheap
1642 /// (`fs::create_dir_all`). Exposed publicly so consumers — `lex
1643 /// blame --with-evidence`, `GET /v1/stage/<id>/attestations` —
1644 /// can read what the store gate emitted without round-tripping
1645 /// through this crate's API surface.
1646 /// Recompute a producer's trust score from its recent
1647 /// attestation history and emit a fresh `ProducerTrust`
1648 /// attestation (#293). Score = `passed / (passed + failed
1649 /// + inconclusive)` over the last `window` attestations
1650 /// produced by `tool_id`, expressed in thousandths
1651 /// (`0..=1000`).
1652 ///
1653 /// Refuses to grant trust when the tool has an active
1654 /// `ProducerBlock` — the block wins as a hard veto. Returns
1655 /// `Ok(None)` for "no attestations to score" (a brand-new
1656 /// producer); the caller can choose how to handle it
1657 /// (typically: skip the publish until evidence accrues).
1658 ///
1659 /// `granted_by` is the identity of the actor running the
1660 /// recompute (typically the human admin, or "lex-ci-bot"
1661 /// for an automated nightly).
1662 pub fn recompute_producer_trust(
1663 &self,
1664 tool_id: &str,
1665 window: usize,
1666 granted_by: &str,
1667 ) -> Result<Option<lex_vcs::AttestationId>, StoreError> {
1668 let log = self.attestation_log()?;
1669 let all = log.list_all()?;
1670 // Hard veto: don't grant trust to a blocked tool.
1671 if lex_vcs::active_producer_block(&all, tool_id).is_some() {
1672 return Err(StoreError::InvalidTransition(format!(
1673 "cannot recompute trust for `{tool_id}` — \
1674 producer is currently blocked"
1675 )));
1676 }
1677 // Filter to attestations from this tool, newest-first by
1678 // timestamp, then take the window.
1679 let mut from_tool: Vec<&lex_vcs::Attestation> = all
1680 .iter()
1681 .filter(|a| a.produced_by.tool == tool_id)
1682 // Ignore self-referential trust attestations (we're
1683 // scoring evidence, not previous trust statements).
1684 .filter(|a| {
1685 !matches!(
1686 a.kind,
1687 lex_vcs::AttestationKind::ProducerTrust { .. }
1688 | lex_vcs::AttestationKind::TrustWaived { .. }
1689 )
1690 })
1691 .collect();
1692 from_tool.sort_by_key(|a| std::cmp::Reverse(a.timestamp));
1693 from_tool.truncate(window);
1694 if from_tool.is_empty() {
1695 return Ok(None);
1696 }
1697 let (mut passed, mut total) = (0u64, 0u64);
1698 for a in &from_tool {
1699 total += 1;
1700 if matches!(a.result, lex_vcs::AttestationResult::Passed) {
1701 passed += 1;
1702 }
1703 }
1704 let score = if total == 0 {
1705 0
1706 } else {
1707 let raw = (passed as f64) * 1000.0 / (total as f64);
1708 raw.round().clamp(0.0, 1000.0) as u32
1709 };
1710 let head_op = self
1711 .list_branches()?
1712 .into_iter()
1713 .find_map(|b| self.get_branch(&b).ok().flatten().and_then(|x| x.head_op))
1714 .unwrap_or_else(|| "fresh".into());
1715 let evidence = format!(
1716 "window={window}, sample={}, head_op={head_op:.16}",
1717 from_tool.len()
1718 );
1719 let attestation = lex_vcs::Attestation::new(
1720 tool_id.to_string(),
1721 None,
1722 None,
1723 lex_vcs::AttestationKind::ProducerTrust {
1724 tool_id: tool_id.into(),
1725 score_thousandths: score,
1726 evidence,
1727 granted_by: granted_by.into(),
1728 },
1729 lex_vcs::AttestationResult::Passed,
1730 producer_trust_producer(),
1731 None,
1732 );
1733 let id = attestation.attestation_id.clone();
1734 log.put(&attestation)?;
1735 Ok(Some(id))
1736 }
1737
1738 /// The latest live `ProducerTrust` score (thousandths, `0..=1000`) for
1739 /// every producer that currently has trust: the newest score per tool by
1740 /// timestamp, excluding any tool under an active `ProducerBlock` (a block
1741 /// is a hard veto over trust, matching `recompute_producer_trust`).
1742 ///
1743 /// Used to export a capsule trusted-keys keyring from *earned* trust — the
1744 /// producer id doubles as the publisher's signing key downstream, so this
1745 /// turns track record into the allowlist `capsule install` consumes.
1746 pub fn live_producer_trust_scores(
1747 &self,
1748 ) -> Result<std::collections::BTreeMap<String, u32>, StoreError> {
1749 let log = self.attestation_log()?;
1750 let all = log.list_all()?;
1751 // Newest score per tool.
1752 let mut latest: std::collections::BTreeMap<String, (u64, u32)> =
1753 std::collections::BTreeMap::new();
1754 for a in &all {
1755 if let lex_vcs::AttestationKind::ProducerTrust {
1756 tool_id,
1757 score_thousandths,
1758 ..
1759 } = &a.kind
1760 {
1761 let entry = latest.entry(tool_id.clone()).or_insert((0, 0));
1762 if a.timestamp >= entry.0 {
1763 *entry = (a.timestamp, *score_thousandths);
1764 }
1765 }
1766 }
1767 // Drop blocked producers; a block vetoes trust.
1768 let mut scores = std::collections::BTreeMap::new();
1769 for (tool, (_, score)) in latest {
1770 if lex_vcs::active_producer_block(&all, &tool).is_some() {
1771 continue;
1772 }
1773 scores.insert(tool, score);
1774 }
1775 Ok(scores)
1776 }
1777
1778 pub fn attestation_log(&self) -> Result<lex_vcs::AttestationLog, StoreError> {
1779 Ok(lex_vcs::AttestationLog::open(self.root())?)
1780 }
1781
1782 /// Emit one `TypeCheck::Passed` attestation per stage produced by
1783 /// a successful gated apply. Idempotent on `attestation_id` —
1784 /// re-running the same gate run dedups via content addressing.
1785 ///
1786 /// Failure modes: `io::Error` from the attestation log (disk
1787 /// full, perms). The op has already landed by the time this
1788 /// runs; an error here means the op is durable but the evidence
1789 /// is missing. We propagate so the caller sees the partial
1790 /// state rather than silently swallowing — re-attesting the
1791 /// same op against the same op_id is idempotent (content
1792 /// addressing) so a retry is safe once the underlying issue is
1793 /// fixed.
1794 fn record_typecheck_passed(
1795 &self,
1796 stage_ids: &[String],
1797 op_id: &lex_vcs::OpId,
1798 ) -> Result<(), StoreError> {
1799 if stage_ids.is_empty() {
1800 return Ok(());
1801 }
1802 let log = self.attestation_log()?;
1803 for stage_id in stage_ids {
1804 let attestation = lex_vcs::Attestation::new(
1805 stage_id.clone(),
1806 Some(op_id.clone()),
1807 None,
1808 lex_vcs::AttestationKind::TypeCheck,
1809 lex_vcs::AttestationResult::Passed,
1810 typecheck_producer(),
1811 None,
1812 );
1813 log.put(&attestation)?;
1814 }
1815 Ok(())
1816 }
1817
1818 /// Emit an `Examples::Passed` attestation for a published stage
1819 /// whose behavioral `examples {}` block was run and passed (#835,
1820 /// Tier 1). Mirrors [`Self::record_typecheck_passed`]. The
1821 /// behavioral run itself happens one layer up (lex-api / lex-cli)
1822 /// because it needs the bytecode compiler + VM, which this crate
1823 /// deliberately doesn't depend on; the store only records the
1824 /// verdict. `file_hash` uses the stage id — the stage fully
1825 /// determines its own examples.
1826 pub fn record_examples_passed(
1827 &self,
1828 stage_id: &str,
1829 op_id: &lex_vcs::OpId,
1830 count: usize,
1831 ) -> Result<(), StoreError> {
1832 let log = self.attestation_log()?;
1833 let attestation = lex_vcs::Attestation::new(
1834 stage_id.to_string(),
1835 Some(op_id.clone()),
1836 None,
1837 lex_vcs::AttestationKind::Examples { file_hash: stage_id.to_string(), count },
1838 lex_vcs::AttestationResult::Passed,
1839 examples_producer(),
1840 None,
1841 );
1842 log.put(&attestation)?;
1843 Ok(())
1844 }
1845
1846 /// Record a structured `Review` verdict on a stage (#836 G4).
1847 /// The verdict maps onto the attestation `result` so existing
1848 /// result-based tooling reads it: Approve->Passed,
1849 /// Reject->Failed, RequestChanges->Inconclusive.
1850 pub fn record_review(
1851 &self,
1852 stage_id: &str,
1853 op_id: Option<lex_vcs::OpId>,
1854 reviewer: &str,
1855 verdict: lex_vcs::ReviewVerdict,
1856 notes: Option<String>,
1857 ) -> Result<lex_vcs::AttestationId, StoreError> {
1858 let result = match verdict {
1859 lex_vcs::ReviewVerdict::Approve => lex_vcs::AttestationResult::Passed,
1860 lex_vcs::ReviewVerdict::Reject => lex_vcs::AttestationResult::Failed {
1861 detail: notes.clone().unwrap_or_else(|| "rejected".into()),
1862 },
1863 lex_vcs::ReviewVerdict::RequestChanges => lex_vcs::AttestationResult::Inconclusive {
1864 detail: notes.clone().unwrap_or_else(|| "changes requested".into()),
1865 },
1866 };
1867 let att = lex_vcs::Attestation::new(
1868 stage_id.to_string(),
1869 op_id,
1870 None,
1871 lex_vcs::AttestationKind::Review { reviewer: reviewer.to_string(), verdict, notes },
1872 result,
1873 review_producer(reviewer),
1874 None,
1875 );
1876 let id = att.attestation_id.clone();
1877 self.attestation_log()?.put(&att)?;
1878 Ok(id)
1879 }
1880
1881 /// The latest `Review` verdict recorded on a stage, if any
1882 /// (#836 G4). "Latest" is by attestation timestamp; ties keep the
1883 /// last one seen. Used by `promote_candidate` to honor a standing
1884 /// Reject.
1885 pub fn latest_review_verdict(
1886 &self,
1887 stage_id: &str,
1888 ) -> Result<Option<lex_vcs::ReviewVerdict>, StoreError> {
1889 let log = self.attestation_log()?;
1890 let mut latest: Option<(u64, lex_vcs::ReviewVerdict)> = None;
1891 for a in log.list_for_stage(&stage_id.to_string())? {
1892 if let lex_vcs::AttestationKind::Review { verdict, .. } = a.kind {
1893 if latest.as_ref().map(|(t, _)| a.timestamp >= *t).unwrap_or(true) {
1894 latest = Some((a.timestamp, verdict));
1895 }
1896 }
1897 }
1898 Ok(latest.map(|(_, v)| v))
1899 }
1900
1901 /// Consult `policy.session_budgets` for the op's session
1902 /// (resolved via `op.intent_id → Intent.session_id`) and
1903 /// refuse if applying would push the session's monotonic spend
1904 /// over the configured cap (#292 slice 3).
1905 ///
1906 /// Ops without an `intent_id`, or whose intent has no
1907 /// configured cap, return Ok without any disk read.
1908 fn check_session_budget(&self, op: &lex_vcs::Operation) -> Result<(), StoreError> {
1909 let Some(intent_id) = op.intent_id.as_deref() else {
1910 return Ok(());
1911 };
1912 let intent_log = lex_vcs::IntentLog::open(self.root())?;
1913 let Some(intent) = intent_log.get(&intent_id.to_string())? else {
1914 // Dangling intent — treat as "no session" and let it
1915 // sail through. Slice 1's ledger already documents
1916 // this as graceful-degradation semantics.
1917 return Ok(());
1918 };
1919 let policy = crate::policy::load(self.root())?.unwrap_or_default();
1920 let Some(cap) = policy.session_budgets.cap_for(&intent.session_id) else {
1921 return Ok(());
1922 };
1923 // Recompute the session's current spend + the contribution
1924 // from this op. Re-running the ledger walk on every gated
1925 // op is O(branch history); see #292 slice 1's note about
1926 // a future on-disk cache.
1927 let current = self.session_budget(&intent.session_id)?;
1928 let increment = crate::budget::monotonic_spend_of(&op.kind);
1929 let spent_after = current.spent.saturating_add(increment);
1930 if spent_after > cap {
1931 return Err(StoreError::BudgetExceeded {
1932 session_id: intent.session_id,
1933 cap,
1934 spent_after,
1935 });
1936 }
1937 Ok(())
1938 }
1939
1940 /// Emit `RepairHint` attestations for a TypeError-rejected op
1941 /// (#281). One per candidate stage in the transition. The hint
1942 /// records the *would-be* op_id (deterministic, content-
1943 /// addressed even though the op record was never persisted)
1944 /// and the structured errors.
1945 ///
1946 /// #306 slice 3: `suggested_transform` is populated from the
1947 /// static (rule_tag → likely_transform) table for the *first*
1948 /// error in the batch. The LLM-driven `lex repair --apply`
1949 /// flow can still overwrite this with a higher-quality
1950 /// suggestion; the static value is the floor, not the ceiling.
1951 ///
1952 /// Best-effort: a write failure here is swallowed by the
1953 /// caller (the original `TypeError` is the load-bearing
1954 /// signal; missing the hint is recoverable on a retry).
1955 fn record_repair_hint(
1956 &self,
1957 stage_ids: &[String],
1958 failed_op_id: &lex_vcs::OpId,
1959 errors: &[lex_types::TypeError],
1960 ) -> Result<(), StoreError> {
1961 if stage_ids.is_empty() {
1962 return Ok(());
1963 }
1964 let errors_json = serde_json::to_value(errors).map_err(StoreError::Serde)?;
1965 // #306 slice 3: look up the static suggested_transform for
1966 // the first error's rule_tag. Multiple errors per op are
1967 // possible — when they fire in lockstep (e.g. one bad let
1968 // binding propagates to several use sites), the first
1969 // error's rule_tag is usually the load-bearing one to fix.
1970 let suggested_transform = errors
1971 .first()
1972 .and_then(|e| lex_types::suggested_transform_for(e.rule_tag()));
1973 let log = self.attestation_log()?;
1974 for stage_id in stage_ids {
1975 let attestation = lex_vcs::Attestation::new(
1976 stage_id.clone(),
1977 None, // the failed op was never persisted; not the
1978 // attestation's op_id (which is for a
1979 // *successful* op).
1980 None,
1981 lex_vcs::AttestationKind::RepairHint {
1982 failed_op_id: failed_op_id.clone(),
1983 errors: errors_json.clone(),
1984 suggested_transform: suggested_transform.clone(),
1985 },
1986 lex_vcs::AttestationResult::Failed {
1987 detail: format!(
1988 "op {} rejected: {} type error(s)",
1989 failed_op_id,
1990 errors.len()
1991 ),
1992 },
1993 repair_hint_producer(),
1994 None,
1995 );
1996 log.put(&attestation)?;
1997 }
1998 Ok(())
1999 }
2000
2001 /// Emit `Trace` attestations linking an already-committed `op`
2002 /// to the run that produced it (#257). One attestation per
2003 /// produced stage (matching the `TypeCheck` emission contract
2004 /// — see [`Self::apply_operation_checked`]) with
2005 /// `op_id: Some(op_id)` set, so `lex trace --op <op_id>`
2006 /// surfaces the run.
2007 ///
2008 /// Returns the number of attestations emitted (zero for ops
2009 /// that produce no attestable stage, e.g. `Remove` /
2010 /// `ImportOnly`).
2011 ///
2012 /// Idempotent: re-emitting for the same
2013 /// `(run_id, root_target, op_id, stage_id, producer, result)`
2014 /// tuple dedups via content addressing.
2015 ///
2016 /// `op_id` must already exist in the op log — an unknown op
2017 /// surfaces as `StoreError::UnknownOp`.
2018 pub fn record_op_trace(
2019 &self,
2020 run_id: &str,
2021 root_target: &str,
2022 op_id: &lex_vcs::OpId,
2023 result: lex_vcs::AttestationResult,
2024 producer: lex_vcs::ProducerDescriptor,
2025 ) -> Result<usize, StoreError> {
2026 let log = lex_vcs::OpLog::open(self.root())?;
2027 let rec = log
2028 .get(op_id)?
2029 .ok_or_else(|| StoreError::UnknownOp(op_id.clone()))?;
2030 let stage_ids = attestable_stage_ids(&rec.produces);
2031 if stage_ids.is_empty() {
2032 return Ok(0);
2033 }
2034 let attlog = self.attestation_log()?;
2035 let mut emitted = 0;
2036 for stage_id in stage_ids {
2037 let attestation = lex_vcs::Attestation::new(
2038 stage_id,
2039 Some(op_id.clone()),
2040 None,
2041 lex_vcs::AttestationKind::Trace {
2042 run_id: run_id.into(),
2043 root_target: root_target.into(),
2044 },
2045 result.clone(),
2046 producer.clone(),
2047 None,
2048 );
2049 attlog.put(&attestation)?;
2050 emitted += 1;
2051 }
2052 Ok(emitted)
2053 }
2054
2055 /// Walk `ops_since(branch_head, base)` and emit per-stage
2056 /// `Trace` attestations for each new op, linking them to the
2057 /// run that produced them (#257). Used by `lex run --trace`
2058 /// after the VM exits: snapshot `base = branch_head` before
2059 /// the run, then call this with the post-run head.
2060 ///
2061 /// `base = None` means "every op currently reachable from the
2062 /// branch head" — generally not what you want for a single
2063 /// run; pass the pre-run head.
2064 ///
2065 /// Returns the total number of attestations emitted across
2066 /// every new op. Zero is the common case (the run committed no
2067 /// ops).
2068 ///
2069 /// Idempotent on the per-op level via [`Self::record_op_trace`].
2070 pub fn record_run_committed_ops_since(
2071 &self,
2072 run_id: &str,
2073 root_target: &str,
2074 branch: &str,
2075 base: Option<&lex_vcs::OpId>,
2076 result: lex_vcs::AttestationResult,
2077 producer: lex_vcs::ProducerDescriptor,
2078 ) -> Result<usize, StoreError> {
2079 let head = match self.get_branch(branch)?.and_then(|b| b.head_op) {
2080 Some(h) => h,
2081 None => return Ok(0),
2082 };
2083 let log = lex_vcs::OpLog::open(self.root())?;
2084 let new_ops = log.ops_since(&head, base)?;
2085 let mut total = 0;
2086 for rec in new_ops {
2087 total += self.record_op_trace(
2088 run_id,
2089 root_target,
2090 &rec.op_id,
2091 result.clone(),
2092 producer.clone(),
2093 )?;
2094 }
2095 Ok(total)
2096 }
2097
2098 /// Apply a typed `ReplaceMatchArm` transform (#280) and emit a
2099 /// `OperationKind::ReplaceMatchArm` op that records the
2100 /// semantic shape of the edit, not just the byte effect.
2101 ///
2102 /// Steps:
2103 /// 1. Load the source stage's canonical bytes (delta-aware).
2104 /// 2. Run [`lex_ast::replace_match_arm`] to produce the new
2105 /// `Stage`. Pure function, no I/O.
2106 /// 3. Publish the new stage. Idempotent on the
2107 /// content-addressed `to_stage_id`.
2108 /// 4. Assemble the candidate program (every active stage on
2109 /// the branch, with the rewritten one swapped in) and call
2110 /// [`Self::apply_operation_checked`] — re-typechecks and
2111 /// runs every existing gate (TypeCheck attestation,
2112 /// required_attestations, producer-block walk-back).
2113 ///
2114 /// Failure modes:
2115 /// * [`StoreError::TransformError`] — transform didn't apply.
2116 /// The branch is unchanged; no stage published.
2117 /// * [`StoreError::TypeError`] — transform produced an
2118 /// ill-typed program. The new stage is on disk (idempotent
2119 /// on its content hash) but the branch is unchanged. Same
2120 /// "publish without advance" semantics as #245.
2121 /// * Everything else from `apply_operation_checked`.
2122 pub fn apply_replace_match_arm(
2123 &self,
2124 branch: &str,
2125 from_stage_id: &str,
2126 match_node: &lex_ast::NodeId,
2127 arm_index: usize,
2128 new_body: lex_ast::CExpr,
2129 ) -> Result<lex_vcs::OpId, StoreError> {
2130 let from_stage = self.get_ast(from_stage_id)?;
2131 let new_stage = lex_ast::replace_match_arm(&from_stage, match_node, arm_index, new_body)
2132 .map_err(StoreError::TransformError)?;
2133 let sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2134 let to_stage_id = self.publish(&new_stage)?;
2135 if to_stage_id == from_stage_id {
2136 // No-op transform — the new body was structurally
2137 // identical to the old. Refuse rather than advancing
2138 // the branch with an empty edit.
2139 return Err(StoreError::InvalidTransition(format!(
2140 "replace_match_arm produced the same stage_id `{from_stage_id}`"
2141 )));
2142 }
2143
2144 // Assemble the candidate program: every active stage on
2145 // the branch, with `from_stage_id` swapped for `new_stage`.
2146 let head = self.branch_head(branch)?;
2147 let mut candidate: Vec<lex_ast::Stage> = Vec::with_capacity(head.len());
2148 for (other_sig, other_stage_id) in &head {
2149 if other_sig == &sig {
2150 candidate.push(new_stage.clone());
2151 } else {
2152 candidate.push(self.get_ast(other_stage_id)?);
2153 }
2154 }
2155 // If the source sig isn't on the current branch head, the
2156 // transform is operating on a stage that hasn't been added
2157 // yet — refuse rather than risking a candidate program
2158 // that doesn't reflect the branch's actual state.
2159 if !head.contains_key(&sig) {
2160 return Err(StoreError::InvalidTransition(format!(
2161 "sig `{sig}` not on branch `{branch}`'s head"
2162 )));
2163 }
2164
2165 // #247: budget delta captured for `lex op log --budget-drift`.
2166 let from_budget = budget_of_stage(&from_stage);
2167 let to_budget = budget_of_stage(&new_stage);
2168
2169 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2170 let kind = lex_vcs::OperationKind::ReplaceMatchArm {
2171 sig_id: sig.clone(),
2172 from_stage_id: from_stage_id.to_string(),
2173 to_stage_id: to_stage_id.clone(),
2174 match_node: match_node.as_str().to_string(),
2175 arm_index,
2176 from_budget,
2177 to_budget,
2178 };
2179 let transition = lex_vcs::StageTransition::Replace {
2180 sig_id: sig.clone(),
2181 from: from_stage_id.to_string(),
2182 to: to_stage_id.clone(),
2183 };
2184 let op = lex_vcs::Operation::new(kind, head_now.into_iter().collect::<Vec<_>>());
2185 self.apply_operation_checked(branch, op, transition, &candidate)
2186 }
2187
2188 /// Apply a typed `RenameLocal` transform (#280) — rename a
2189 /// `let`-bound local within a fn body and emit a matching
2190 /// `OperationKind::RenameLocal`. Same end-to-end shape as
2191 /// [`Self::apply_replace_match_arm`]; see that method for the
2192 /// failure-mode taxonomy.
2193 pub fn apply_rename_local(
2194 &self,
2195 branch: &str,
2196 from_stage_id: &str,
2197 let_node: &lex_ast::NodeId,
2198 new_name: &str,
2199 ) -> Result<lex_vcs::OpId, StoreError> {
2200 let from_stage = self.get_ast(from_stage_id)?;
2201 // Read the old name before running the transform, so the
2202 // op log records the rename target rather than just the
2203 // new value.
2204 let old_name = read_let_name(&from_stage, let_node).map_err(StoreError::TransformError)?;
2205 let new_stage = lex_ast::rename_local(&from_stage, let_node, new_name)
2206 .map_err(StoreError::TransformError)?;
2207 let sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2208 let to_stage_id = self.publish(&new_stage)?;
2209 if to_stage_id == from_stage_id {
2210 return Err(StoreError::InvalidTransition(format!(
2211 "rename_local produced the same stage_id `{from_stage_id}`"
2212 )));
2213 }
2214 let head = self.branch_head(branch)?;
2215 let mut candidate: Vec<lex_ast::Stage> = Vec::with_capacity(head.len());
2216 for (other_sig, other_stage_id) in &head {
2217 if other_sig == &sig {
2218 candidate.push(new_stage.clone());
2219 } else {
2220 candidate.push(self.get_ast(other_stage_id)?);
2221 }
2222 }
2223 if !head.contains_key(&sig) {
2224 return Err(StoreError::InvalidTransition(format!(
2225 "sig `{sig}` not on branch `{branch}`'s head"
2226 )));
2227 }
2228 let from_budget = budget_of_stage(&from_stage);
2229 let to_budget = budget_of_stage(&new_stage);
2230 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2231 let kind = lex_vcs::OperationKind::RenameLocal {
2232 sig_id: sig.clone(),
2233 from_stage_id: from_stage_id.to_string(),
2234 to_stage_id: to_stage_id.clone(),
2235 let_node: let_node.as_str().to_string(),
2236 old_name,
2237 new_name: new_name.to_string(),
2238 from_budget,
2239 to_budget,
2240 };
2241 let transition = lex_vcs::StageTransition::Replace {
2242 sig_id: sig.clone(),
2243 from: from_stage_id.to_string(),
2244 to: to_stage_id.clone(),
2245 };
2246 let op = lex_vcs::Operation::new(kind, head_now.into_iter().collect::<Vec<_>>());
2247 self.apply_operation_checked(branch, op, transition, &candidate)
2248 }
2249
2250 /// Apply a typed `InlineLet` transform (#280) — eliminate a
2251 /// `let x := v; body` by substituting `v` for every unshadowed
2252 /// `x` in `body`, then replacing the `Let` node with the
2253 /// substituted body. Same end-to-end shape as
2254 /// [`Self::apply_replace_match_arm`].
2255 pub fn apply_inline_let(
2256 &self,
2257 branch: &str,
2258 from_stage_id: &str,
2259 let_node: &lex_ast::NodeId,
2260 ) -> Result<lex_vcs::OpId, StoreError> {
2261 let from_stage = self.get_ast(from_stage_id)?;
2262 let binding_name =
2263 read_let_name(&from_stage, let_node).map_err(StoreError::TransformError)?;
2264 let new_stage =
2265 lex_ast::inline_let(&from_stage, let_node).map_err(StoreError::TransformError)?;
2266 let sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2267 let to_stage_id = self.publish(&new_stage)?;
2268 if to_stage_id == from_stage_id {
2269 return Err(StoreError::InvalidTransition(format!(
2270 "inline_let produced the same stage_id `{from_stage_id}`"
2271 )));
2272 }
2273 let head = self.branch_head(branch)?;
2274 let mut candidate: Vec<lex_ast::Stage> = Vec::with_capacity(head.len());
2275 for (other_sig, other_stage_id) in &head {
2276 if other_sig == &sig {
2277 candidate.push(new_stage.clone());
2278 } else {
2279 candidate.push(self.get_ast(other_stage_id)?);
2280 }
2281 }
2282 if !head.contains_key(&sig) {
2283 return Err(StoreError::InvalidTransition(format!(
2284 "sig `{sig}` not on branch `{branch}`'s head"
2285 )));
2286 }
2287 let from_budget = budget_of_stage(&from_stage);
2288 let to_budget = budget_of_stage(&new_stage);
2289 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2290 let kind = lex_vcs::OperationKind::InlineLet {
2291 sig_id: sig.clone(),
2292 from_stage_id: from_stage_id.to_string(),
2293 to_stage_id: to_stage_id.clone(),
2294 let_node: let_node.as_str().to_string(),
2295 binding_name,
2296 from_budget,
2297 to_budget,
2298 };
2299 let transition = lex_vcs::StageTransition::Replace {
2300 sig_id: sig.clone(),
2301 from: from_stage_id.to_string(),
2302 to: to_stage_id.clone(),
2303 };
2304 let op = lex_vcs::Operation::new(kind, head_now.into_iter().collect::<Vec<_>>());
2305 self.apply_operation_checked(branch, op, transition, &candidate)
2306 }
2307
2308 /// Apply a typed `ExtractFunction` transform (#280 slice 4) —
2309 /// extract a sub-expression of `from_stage_id`'s body into a
2310 /// new top-level fn defined by `spec`, and emit two ops tied
2311 /// together by a shared synthetic Intent so `lex op log
2312 /// --intent <id>` groups them.
2313 ///
2314 /// The two ops:
2315 /// 1. `AddFunction { sig_id: <new_fn_sig>, stage_id: <new_fn_stage> }`
2316 /// 2. `ModifyBody { sig_id: <source_sig>, from_stage_id, to_stage_id: <modified> }`
2317 ///
2318 /// The shared Intent's prompt is structured (`extract_function:
2319 /// <new_fn_name>` plus the source identity) so downstream
2320 /// tooling can recover the typed-transform shape from the
2321 /// op-log + intent-log join.
2322 ///
2323 /// Returns `(add_fn_op_id, modify_body_op_id)`.
2324 pub fn apply_extract_function(
2325 &self,
2326 branch: &str,
2327 from_stage_id: &str,
2328 expr_node: &lex_ast::NodeId,
2329 spec: lex_ast::ExtractFnSpec,
2330 ) -> Result<(lex_vcs::OpId, lex_vcs::OpId), StoreError> {
2331 let from_stage = self.get_ast(from_stage_id)?;
2332 let new_fn_name = spec.name.clone();
2333 let (modified_stage, new_fn_stage) =
2334 lex_ast::extract_function(&from_stage, expr_node, spec)
2335 .map_err(StoreError::TransformError)?;
2336
2337 let source_sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2338 let new_fn_sig = lex_ast::sig_id(&new_fn_stage).ok_or(StoreError::CannotPublishImport)?;
2339 if source_sig == new_fn_sig {
2340 return Err(StoreError::InvalidTransition(format!(
2341 "extract_function produced a sig matching the source `{source_sig}`"
2342 )));
2343 }
2344 let new_fn_stage_id = self.publish(&new_fn_stage)?;
2345 let modified_stage_id = self.publish(&modified_stage)?;
2346 if modified_stage_id == from_stage_id {
2347 return Err(StoreError::InvalidTransition(format!(
2348 "extract_function produced the same stage_id `{from_stage_id}` for the source"
2349 )));
2350 }
2351
2352 let head = self.branch_head(branch)?;
2353 if !head.contains_key(&source_sig) {
2354 return Err(StoreError::InvalidTransition(format!(
2355 "sig `{source_sig}` not on branch `{branch}`'s head"
2356 )));
2357 }
2358
2359 // Synthesize an Intent linking the two ops. The session_id
2360 // / model fields here are not load-bearing — they exist to
2361 // make the IntentId content-addressed; downstream tooling
2362 // reads `prompt` to reconstruct the typed-transform shape.
2363 let intent = lex_vcs::Intent::new(
2364 format!(
2365 "[lex.transform.extract_function]\nnew_fn={new_fn_name}\nsource_sig={source_sig}\nfrom_stage={from_stage_id}\nexpr_node={node}",
2366 node = expr_node.as_str(),
2367 ),
2368 "lex-store::apply_extract_function",
2369 lex_vcs::ModelDescriptor {
2370 provider: "lex-store".into(),
2371 name: env!("CARGO_PKG_VERSION").into(),
2372 version: None,
2373 },
2374 None,
2375 );
2376 let intent_id = intent.intent_id.clone();
2377 lex_vcs::IntentLog::open(self.root())?.put(&intent)?;
2378
2379 // Step 1 — emit the AddFunction op for the new fn. Build
2380 // the candidate program by appending the new fn to every
2381 // stage on the current branch head.
2382 let new_fn_effects: std::collections::BTreeSet<String> = match &new_fn_stage {
2383 lex_ast::Stage::FnDecl(fd) => fd.effects.iter().map(|e| e.name.clone()).collect(),
2384 _ => Default::default(),
2385 };
2386 let new_fn_budget = budget_of_stage(&new_fn_stage);
2387 let mut candidate_with_new_fn: Vec<lex_ast::Stage> = Vec::with_capacity(head.len() + 1);
2388 for stage_id in head.values() {
2389 candidate_with_new_fn.push(self.get_ast(stage_id)?);
2390 }
2391 candidate_with_new_fn.push(new_fn_stage.clone());
2392 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2393 let add_op = lex_vcs::Operation::new(
2394 lex_vcs::OperationKind::AddFunction {
2395 sig_id: new_fn_sig.clone(),
2396 stage_id: new_fn_stage_id.clone(),
2397 effects: new_fn_effects,
2398 budget_cost: new_fn_budget,
2399 },
2400 head_now.into_iter().collect::<Vec<_>>(),
2401 )
2402 .with_intent(intent_id.clone());
2403 let add_transition = lex_vcs::StageTransition::Create {
2404 sig_id: new_fn_sig.clone(),
2405 stage_id: new_fn_stage_id.clone(),
2406 };
2407 let add_op_id =
2408 self.apply_operation_checked(branch, add_op, add_transition, &candidate_with_new_fn)?;
2409
2410 // Step 2 — emit the ModifyBody op for the source. Build
2411 // the candidate program by replacing the source's stage
2412 // with `modified_stage` and keeping the new fn alongside.
2413 let from_budget = budget_of_stage(&from_stage);
2414 let to_budget = budget_of_stage(&modified_stage);
2415 let mut candidate_with_modified: Vec<lex_ast::Stage> = Vec::with_capacity(head.len() + 1);
2416 for (other_sig, other_stage_id) in &head {
2417 if other_sig == &source_sig {
2418 candidate_with_modified.push(modified_stage.clone());
2419 } else {
2420 candidate_with_modified.push(self.get_ast(other_stage_id)?);
2421 }
2422 }
2423 candidate_with_modified.push(new_fn_stage.clone());
2424 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2425 let modify_op = lex_vcs::Operation::new(
2426 lex_vcs::OperationKind::ModifyBody {
2427 sig_id: source_sig.clone(),
2428 from_stage_id: from_stage_id.to_string(),
2429 to_stage_id: modified_stage_id.clone(),
2430 from_budget,
2431 to_budget,
2432 },
2433 head_now.into_iter().collect::<Vec<_>>(),
2434 )
2435 .with_intent(intent_id);
2436 let modify_transition = lex_vcs::StageTransition::Replace {
2437 sig_id: source_sig,
2438 from: from_stage_id.to_string(),
2439 to: modified_stage_id,
2440 };
2441 let modify_op_id = self.apply_operation_checked(
2442 branch,
2443 modify_op,
2444 modify_transition,
2445 &candidate_with_modified,
2446 )?;
2447
2448 Ok((add_op_id, modify_op_id))
2449 }
2450
2451 /// Propose a stage for `sig_id` without advancing the branch
2452 /// head (#294). Multiple agents can call this concurrently
2453 /// for the same sig — every call lands a fresh `Candidate`
2454 /// op chained off the current head_op. The branch head stays
2455 /// where it was; a later [`Self::promote_candidate`] picks
2456 /// the winner.
2457 ///
2458 /// The caller is responsible for typechecking `new_stage`
2459 /// against whatever program context they consider valid —
2460 /// `propose_candidate` doesn't run the gate. Type errors
2461 /// surface at promotion time, where the candidate is
2462 /// composed back into a candidate program via the standard
2463 /// `apply_operation_checked` path.
2464 ///
2465 /// The stage is published (idempotent on content hash). The
2466 /// `intent_id` is required so downstream consumers can
2467 /// distinguish proposals by author.
2468 pub fn propose_candidate(
2469 &self,
2470 branch: &str,
2471 new_stage: &lex_ast::Stage,
2472 intent_id: &lex_vcs::IntentId,
2473 ) -> Result<lex_vcs::OpId, StoreError> {
2474 let sig = lex_ast::sig_id(new_stage).ok_or(StoreError::CannotPublishImport)?;
2475 let stage_id = self.publish(new_stage)?;
2476 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2477 let op = lex_vcs::Operation::new(
2478 lex_vcs::OperationKind::Candidate {
2479 sig_id: sig,
2480 stage_id,
2481 },
2482 head_now.into_iter().collect::<Vec<_>>(),
2483 )
2484 .with_intent(intent_id.clone());
2485 let transition = lex_vcs::StageTransition::ImportOnly;
2486 self.apply_operation(branch, op, transition)
2487 }
2488
2489 /// List every live `Candidate` op for `sig_id` — i.e. those
2490 /// not yet referenced by any `Promote` op (either as the
2491 /// winner or in the `supersedes` set). Used by `lex stage
2492 /// candidates`. Results are sorted by op_id for
2493 /// reproducibility.
2494 pub fn list_candidates(&self, sig_id: &str) -> Result<Vec<CandidateInfo>, StoreError> {
2495 let log = lex_vcs::OpLog::open(self.root())?;
2496 let all = log.list_all()?;
2497 // Collect the set of candidate op_ids referenced by any
2498 // Promote for this sig. Those candidates are no longer
2499 // live.
2500 let mut referenced: std::collections::BTreeSet<lex_vcs::OpId> = Default::default();
2501 for rec in &all {
2502 if let lex_vcs::OperationKind::Promote {
2503 sig_id: s,
2504 winner_candidate,
2505 supersedes,
2506 ..
2507 } = &rec.op.kind
2508 {
2509 if s != sig_id {
2510 continue;
2511 }
2512 referenced.insert(winner_candidate.clone());
2513 for sup in supersedes {
2514 referenced.insert(sup.clone());
2515 }
2516 }
2517 }
2518 let mut out: Vec<CandidateInfo> = Vec::new();
2519 for rec in all {
2520 let lex_vcs::OperationKind::Candidate {
2521 sig_id: s,
2522 stage_id,
2523 } = &rec.op.kind
2524 else {
2525 continue;
2526 };
2527 if s != sig_id {
2528 continue;
2529 }
2530 if referenced.contains(&rec.op_id) {
2531 continue;
2532 }
2533 out.push(CandidateInfo {
2534 op_id: rec.op_id.clone(),
2535 stage_id: stage_id.clone(),
2536 intent_id: rec.op.intent_id.clone(),
2537 });
2538 }
2539 out.sort_by(|a, b| a.op_id.cmp(&b.op_id));
2540 Ok(out)
2541 }
2542
2543 /// Promote a previously-landed `Candidate` op as the new
2544 /// branch head for its sig (#294). Emits a `Promote` op
2545 /// listing every other live `Candidate` for the same sig
2546 /// in its `supersedes` field. After this lands,
2547 /// [`Self::list_candidates`] returns an empty set for the
2548 /// sig.
2549 ///
2550 /// Re-typechecks the candidate program (winner stage + the
2551 /// rest of the branch) through `apply_operation_checked`, so
2552 /// a candidate that doesn't compose with the current branch
2553 /// state surfaces as `StoreError::TypeError`.
2554 pub fn promote_candidate(
2555 &self,
2556 branch: &str,
2557 candidate_op_id: &lex_vcs::OpId,
2558 ) -> Result<lex_vcs::OpId, StoreError> {
2559 let log = lex_vcs::OpLog::open(self.root())?;
2560 let candidate_rec = log
2561 .get(candidate_op_id)?
2562 .ok_or_else(|| StoreError::UnknownOp(candidate_op_id.clone()))?;
2563 let (sig, winner_stage_id) = match &candidate_rec.op.kind {
2564 lex_vcs::OperationKind::Candidate { sig_id, stage_id } => {
2565 (sig_id.clone(), stage_id.clone())
2566 }
2567 other => {
2568 return Err(StoreError::InvalidTransition(format!(
2569 "op `{candidate_op_id}` is a `{:?}`, not a Candidate",
2570 other
2571 )))
2572 }
2573 };
2574
2575 // #836 G4: a candidate carrying a standing `Reject` review must
2576 // not be promoted. "Standing" = the latest `Review` on the
2577 // winner's stage is a Reject; a later `Approve` (or
2578 // `RequestChanges`, which is advisory, not a veto) lifts it.
2579 // Safe by default: a candidate with no review, or an approved
2580 // one, promotes exactly as before.
2581 if let Some(lex_vcs::ReviewVerdict::Reject) = self.latest_review_verdict(&winner_stage_id)? {
2582 return Err(StoreError::InvalidTransition(format!(
2583 "candidate `{candidate_op_id}` has a standing Reject review on stage `{winner_stage_id}`; record an Approve review (or promote a different candidate) before promoting"
2584 )));
2585 }
2586
2587 // Gather every OTHER live candidate for this sig — the
2588 // ones this Promote will supersede.
2589 let live = self.list_candidates(&sig)?;
2590 let mut supersedes: Vec<lex_vcs::OpId> = live
2591 .iter()
2592 .filter(|c| &c.op_id != candidate_op_id)
2593 .map(|c| c.op_id.clone())
2594 .collect();
2595 supersedes.sort();
2596
2597 // Assemble candidate program: winner stage in place of
2598 // the sig's current head (if any), plus every other sig
2599 // unchanged.
2600 let head = self.branch_head(branch)?;
2601 let winner_stage = self.get_ast(&winner_stage_id)?;
2602 let mut candidate_program: Vec<lex_ast::Stage> = Vec::with_capacity(head.len() + 1);
2603 let mut found = false;
2604 for (other_sig, other_stage_id) in &head {
2605 if other_sig == &sig {
2606 candidate_program.push(winner_stage.clone());
2607 found = true;
2608 } else {
2609 candidate_program.push(self.get_ast(other_stage_id)?);
2610 }
2611 }
2612 if !found {
2613 // Sig doesn't have a head yet — append the winner
2614 // stage to make it a Create.
2615 candidate_program.push(winner_stage.clone());
2616 }
2617 let from_stage_id = head.get(&sig).cloned();
2618 // Budget delta from old head to winner — same shape as
2619 // ModifyBody.
2620 let from_budget = from_stage_id
2621 .as_deref()
2622 .and_then(|s| self.get_ast(s).ok())
2623 .and_then(|s| budget_of_stage(&s));
2624 let to_budget = budget_of_stage(&winner_stage);
2625
2626 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2627 let op = lex_vcs::Operation::new(
2628 lex_vcs::OperationKind::Promote {
2629 sig_id: sig.clone(),
2630 winner_candidate: candidate_op_id.clone(),
2631 winner_stage_id: winner_stage_id.clone(),
2632 supersedes,
2633 from_stage_id: from_stage_id.clone(),
2634 from_budget,
2635 to_budget,
2636 },
2637 head_now.into_iter().collect::<Vec<_>>(),
2638 );
2639 let transition = match &from_stage_id {
2640 Some(from) => lex_vcs::StageTransition::Replace {
2641 sig_id: sig,
2642 from: from.clone(),
2643 to: winner_stage_id,
2644 },
2645 None => lex_vcs::StageTransition::Create {
2646 sig_id: sig,
2647 stage_id: winner_stage_id,
2648 },
2649 };
2650 self.apply_operation_checked(branch, op, transition, &candidate_program)
2651 }
2652
2653 /// `set_branch_head_op` for the durability story on the branch
2654 /// file itself.
2655 pub fn apply_operation(
2656 &self,
2657 branch: &str,
2658 op: lex_vcs::Operation,
2659 transition: lex_vcs::StageTransition,
2660 ) -> Result<lex_vcs::OpId, StoreError> {
2661 let attestable = attestable_stage_ids(&transition);
2662 let op_effects = op_declared_effects(&op.kind);
2663 self.cas_retry_advance(branch, op, transition, |new_head| {
2664 self.run_required_attestations_gate(branch, &new_head.op_id, &attestable, &op_effects)
2665 })
2666 }
2667
2668 /// CAS retry loop for #262. Single-parent ops are rebuilt on
2669 /// each iteration with the current branch head as parent;
2670 /// the per-iteration callback runs the gate (and TypeCheck
2671 /// emission, for the checked path) between persist and CAS.
2672 /// Merge ops (with 2 parents already set) skip the rebuild —
2673 /// their parents are caller-supplied and meaningful — and get
2674 /// a single attempt; on CAS failure they surface `Contention`.
2675 fn cas_retry_advance<F>(
2676 &self,
2677 branch: &str,
2678 op: lex_vcs::Operation,
2679 transition: lex_vcs::StageTransition,
2680 mut between_persist_and_cas: F,
2681 ) -> Result<lex_vcs::OpId, StoreError>
2682 where
2683 F: FnMut(&lex_vcs::NewHead) -> Result<(), StoreError>,
2684 {
2685 // 32 retries handles up to ~32 concurrent writers racing on
2686 // the same branch tip. Beyond that, surfacing `Contention`
2687 // is the right signal — clients should back off or batch.
2688 const MAX_ATTEMPTS: u32 = 32;
2689 // Single-parent ops can be rebuilt on retry; merge ops
2690 // can't (their two parents are meaningful, supplied by the
2691 // merge engine). For merges, single attempt: if CAS
2692 // fails, surface Contention.
2693 let is_rebuildable = op.parents.len() <= 1;
2694 let kind = op.kind.clone();
2695 let intent_id = op.intent_id.clone();
2696
2697 let mut last_io_err: Option<StoreError> = None;
2698 let mut current_op = op;
2699 let current_transition = transition;
2700 // Only rebuild on retries — attempt 1 honors the caller's
2701 // exact op so a user-supplied bogus parent (parents =
2702 // ["someone-else"]) surfaces as `StaleParent` instead of
2703 // being silently corrected.
2704 //
2705 // Exception (#262 follow-up): an op with `parents = []`
2706 // means "I don't care; chain off whatever the current
2707 // head is." Under concurrent apply, attempt 1 can read
2708 // `head_op = Some(opA)` after a sibling writer landed,
2709 // and the persist's parent check fails StaleParent
2710 // unprompted. Rebuild attempt 1 for the empty-parents
2711 // case so the legitimate-race path retries cleanly.
2712 let mut rebuilt_already = false;
2713 for attempt in 1..=MAX_ATTEMPTS {
2714 // Read the current head BEFORE we persist — this is
2715 // the value we'll compare against in the CAS.
2716 let parent = self.get_branch(branch)?.and_then(|b| b.head_op);
2717
2718 // Rebuild the op against the current head, but only
2719 // on retries (not the caller's first attempt) and
2720 // only for single-parent operations. Multi-parent
2721 // (merge) ops are passed through unchanged.
2722 //
2723 // Empty-parents ops also rebuild on attempt 1 (see
2724 // the exception note above) so concurrent apply
2725 // doesn't false-positive on StaleParent.
2726 let should_rebuild = is_rebuildable
2727 && (rebuilt_already || (current_op.parents.is_empty() && parent.is_some()));
2728 if should_rebuild {
2729 current_op = lex_vcs::Operation {
2730 kind: kind.clone(),
2731 parents: parent.iter().cloned().collect(),
2732 intent_id: intent_id.clone(),
2733 };
2734 }
2735
2736 // Persist (idempotent). On `StaleParent` from a retry
2737 // attempt (where we already rebuilt), the head changed
2738 // between our `get_branch` and this `lex_vcs::apply`
2739 // — race; rebuild and continue. On `StaleParent` from
2740 // attempt 1 (caller's input), propagate.
2741 let new_head = match self.persist_op_only_with_parent(
2742 branch,
2743 parent.as_ref(),
2744 current_op.clone(),
2745 current_transition.clone(),
2746 ) {
2747 Ok(nh) => nh,
2748 Err(StoreError::Apply(lex_vcs::ApplyError::StaleParent { .. }))
2749 if is_rebuildable && rebuilt_already =>
2750 {
2751 rebuilt_already = true;
2752 continue;
2753 }
2754 Err(e) => return Err(e),
2755 };
2756
2757 // Run the caller's between-persist-and-cas hook
2758 // (TypeCheck emission + gate). If this fails, the op
2759 // record is durable but orphaned — same semantics as
2760 // pre-#262.
2761 between_persist_and_cas(&new_head)?;
2762
2763 // CAS the branch head. On success: done. On mismatch:
2764 // someone advanced in parallel; retry.
2765 match self.set_branch_head_op_cas(branch, parent, new_head.op_id.clone()) {
2766 Ok(()) => return Ok(new_head.op_id),
2767 Err(crate::branches::CasFailed::Mismatch { .. }) if is_rebuildable => {
2768 // Try again with the new head as parent.
2769 rebuilt_already = true;
2770 continue;
2771 }
2772 Err(crate::branches::CasFailed::Mismatch { .. }) => {
2773 // Merge op: surface immediately — we can't
2774 // rebuild without rerunning the merge engine.
2775 let _ = attempt;
2776 return Err(StoreError::Contention {
2777 branch: branch.into(),
2778 attempts: 1,
2779 });
2780 }
2781 Err(crate::branches::CasFailed::UnknownBranch(b)) => {
2782 return Err(StoreError::UnknownBranch(b));
2783 }
2784 Err(crate::branches::CasFailed::Io(e)) => {
2785 last_io_err = Some(StoreError::Io(std::io::Error::other(e)));
2786 continue;
2787 }
2788 }
2789 }
2790 // Retries exhausted. Prefer surfacing the most recent IO
2791 // error if we hit one; otherwise it's pure CAS contention.
2792 match last_io_err {
2793 Some(e) => Err(e),
2794 None => Err(StoreError::Contention {
2795 branch: branch.into(),
2796 attempts: MAX_ATTEMPTS,
2797 }),
2798 }
2799 }
2800
2801 /// Persist an op against an explicitly-supplied parent. Used
2802 /// by the CAS retry loop in `cas_retry_advance` so the
2803 /// `lex_vcs::apply` parent check matches what we read at the
2804 /// top of the loop iteration (avoids a TOCTOU race against
2805 /// `persist_op_only`'s second read).
2806 fn persist_op_only_with_parent(
2807 &self,
2808 branch: &str,
2809 parent: Option<&lex_vcs::OpId>,
2810 op: lex_vcs::Operation,
2811 transition: lex_vcs::StageTransition,
2812 ) -> Result<lex_vcs::NewHead, StoreError> {
2813 if branch != DEFAULT_BRANCH && self.get_branch(branch)?.is_none() {
2814 return Err(StoreError::UnknownBranch(branch.into()));
2815 }
2816 let log = lex_vcs::OpLog::open(self.root())?;
2817 lex_vcs::apply(&log, parent, op, transition).map_err(|e| match e {
2818 lex_vcs::ApplyError::Persist(io) => StoreError::Io(io),
2819 other => StoreError::Apply(other),
2820 })
2821 }
2822
2823 /// Run the `required_attestations` gate (#245) and the
2824 /// retroactive producer-block gate (#248) over a single op
2825 /// against the store's `policy.json` and attestation log.
2826 ///
2827 /// Failure modes (in order):
2828 ///
2829 /// 1. Producer-block first: if any attestation on the op's
2830 /// stage is from a quarantined tool, refuse with
2831 /// `ProducerBlocked` (#248). Surfaces *before* the
2832 /// required-attestations gate so a clearly-malicious record
2833 /// isn't masked by a missing-Spec error.
2834 /// 2. Required-attestations next: if any required attestation
2835 /// kind is missing, refuse with `BranchAdvanceBlocked`
2836 /// (#245).
2837 ///
2838 /// Loads the policy / attestation log lazily; with no policy
2839 /// file and no `ProducerBlock` attestations the gate is a no-op
2840 /// (default-permissive — matches pre-#245 stores).
2841 fn run_required_attestations_gate(
2842 &self,
2843 branch: &str,
2844 op_id: &lex_vcs::OpId,
2845 stage_ids: &[String],
2846 op_effects: &std::collections::BTreeSet<String>,
2847 ) -> Result<(), StoreError> {
2848 // Build the candidate slice for the new op. Ops with no
2849 // attestable stage (imports, empty merges) get a single
2850 // `None`-stage tuple; both gates skip those.
2851 let new_op_candidate: Vec<(
2852 lex_vcs::OpId,
2853 Option<String>,
2854 std::collections::BTreeSet<String>,
2855 )> = if stage_ids.is_empty() {
2856 vec![(op_id.clone(), None, op_effects.clone())]
2857 } else {
2858 stage_ids
2859 .iter()
2860 .map(|sid| (op_id.clone(), Some(sid.clone()), op_effects.clone()))
2861 .collect()
2862 };
2863 let attest_log = self.attestation_log()?;
2864
2865 // #248 + #256: producer-block gate, walk-back style.
2866 //
2867 // The naive #248 gate only checked the new op's stage. That
2868 // missed contamination on ancestors — once `lex attest
2869 // retro-block` lands, every previously-gated op stays in
2870 // the chain even though its attestations are now from a
2871 // quarantined producer.
2872 //
2873 // #256 fixes this by walking the chain from `head_op` back
2874 // to `last_gate_checkpoint` (or genesis when the checkpoint
2875 // is invalidated), collecting each ancestor's attestable
2876 // stages, and running `check_producer_block` on the
2877 // combined set. After a successful advance,
2878 // `set_branch_head_op` moves the checkpoint to the new
2879 // head (steady-state O(new ops) per advance).
2880 let walk_back_candidate = self.collect_ancestor_candidates(branch)?;
2881 let mut producer_block_candidate = walk_back_candidate;
2882 producer_block_candidate.extend(new_op_candidate.iter().cloned());
2883 crate::policy::check_producer_block(&attest_log, &producer_block_candidate)
2884 .map_err(StoreError::ProducerBlocked)?;
2885
2886 // #245: required-attestations gate. Forward-going only —
2887 // only the new op is checked. Walking back makes no sense
2888 // here: the policy is "this advance must carry these
2889 // attestations," not "every prior op must have."
2890 let policy = match crate::policy::load(self.root())? {
2891 Some(p) if !p.required_attestations.is_empty() => p,
2892 _ => return Ok(()),
2893 };
2894 let waivers =
2895 crate::policy::check_required_attestations(&attest_log, &new_op_candidate, &policy)
2896 .map_err(StoreError::BranchAdvanceBlocked)?;
2897 // #293: emit one `TrustWaived` attestation per waiver so
2898 // the audit trail records every skip. Idempotent on
2899 // attestation_id (content-addressed dedup) — re-running
2900 // the gate with the same state writes the same files.
2901 for w in waivers {
2902 let att = lex_vcs::Attestation::new(
2903 w.stage_id,
2904 Some(op_id.clone()),
2905 None,
2906 lex_vcs::AttestationKind::TrustWaived {
2907 producer: w.producer,
2908 score_thousandths: w.score_thousandths,
2909 threshold_thousandths: w.threshold_thousandths,
2910 kind_tag: w.kind_tag,
2911 },
2912 lex_vcs::AttestationResult::Passed,
2913 trust_waived_producer(),
2914 None,
2915 );
2916 attest_log.put(&att)?;
2917 }
2918 Ok(())
2919 }
2920
2921 /// Walk the branch from `head_op` back to `last_gate_checkpoint`
2922 /// (exclusive) and return the `(op_id, stage_id, op_effects)`
2923 /// tuples for every attestable stage touched by an ancestor
2924 /// (#256). Empty when the branch is fresh, when the checkpoint
2925 /// equals the head, or when the head is None.
2926 fn collect_ancestor_candidates(&self, branch: &str) -> Result<Vec<GateCandidate>, StoreError> {
2927 let b = match self.get_branch(branch)? {
2928 Some(b) => b,
2929 None => return Ok(Vec::new()),
2930 };
2931 let Some(head) = b.head_op else {
2932 return Ok(Vec::new());
2933 };
2934 if Some(&head) == b.last_gate_checkpoint.as_ref() {
2935 // Steady-state common case: previous advance left the
2936 // checkpoint at head. Nothing to re-walk.
2937 return Ok(Vec::new());
2938 }
2939
2940 let log = lex_vcs::OpLog::open(self.root())?;
2941 let walk = log.walk_back(&head, None)?;
2942 let stop_at = b.last_gate_checkpoint.clone();
2943 let mut out = Vec::new();
2944 for rec in walk {
2945 if Some(&rec.op_id) == stop_at.as_ref() {
2946 break;
2947 }
2948 let stages = attestable_stage_ids(&rec.produces);
2949 let effects = op_declared_effects(&rec.op.kind);
2950 if stages.is_empty() {
2951 out.push((rec.op_id.clone(), None, effects));
2952 } else {
2953 for sid in stages {
2954 out.push((rec.op_id.clone(), Some(sid), effects.clone()));
2955 }
2956 }
2957 }
2958 Ok(out)
2959 }
2960}
2961
2962fn stage_name(stage: &Stage) -> &str {
2963 match stage {
2964 Stage::FnDecl(fd) => &fd.name,
2965 Stage::TypeDecl(td) => &td.name,
2966 Stage::Import(i) => &i.alias,
2967 }
2968}
2969
2970fn stage_for_kind<'a>(
2971 kind: &lex_vcs::OperationKind,
2972 stages: &'a [lex_ast::Stage],
2973) -> Option<&'a lex_ast::Stage> {
2974 use lex_vcs::OperationKind::*;
2975 let target_sig = match kind {
2976 AddFunction { sig_id, .. }
2977 | ModifyBody { sig_id, .. }
2978 | ChangeEffectSig { sig_id, .. }
2979 | AddType { sig_id, .. }
2980 | ModifyType { sig_id, .. } => Some(sig_id.clone()),
2981 RenameSymbol { to, .. } => Some(to.clone()),
2982 _ => None,
2983 };
2984 let target_sig = target_sig?;
2985 stages
2986 .iter()
2987 .find(|s| sig_id(s).as_deref() == Some(target_sig.as_str()))
2988}
2989
2990fn transition_for_kind(kind: &lex_vcs::OperationKind) -> lex_vcs::StageTransition {
2991 use lex_vcs::OperationKind::*;
2992 use lex_vcs::StageTransition;
2993 match kind {
2994 AddFunction {
2995 sig_id, stage_id, ..
2996 }
2997 | AddType { sig_id, stage_id } => StageTransition::Create {
2998 sig_id: sig_id.clone(),
2999 stage_id: stage_id.clone(),
3000 },
3001 RemoveFunction {
3002 sig_id,
3003 last_stage_id,
3004 }
3005 | RemoveType {
3006 sig_id,
3007 last_stage_id,
3008 } => StageTransition::Remove {
3009 sig_id: sig_id.clone(),
3010 last: last_stage_id.clone(),
3011 },
3012 ModifyBody {
3013 sig_id,
3014 from_stage_id,
3015 to_stage_id,
3016 ..
3017 }
3018 | ChangeEffectSig {
3019 sig_id,
3020 from_stage_id,
3021 to_stage_id,
3022 ..
3023 }
3024 | ModifyType {
3025 sig_id,
3026 from_stage_id,
3027 to_stage_id,
3028 }
3029 | ReplaceMatchArm {
3030 sig_id,
3031 from_stage_id,
3032 to_stage_id,
3033 ..
3034 }
3035 | RenameLocal {
3036 sig_id,
3037 from_stage_id,
3038 to_stage_id,
3039 ..
3040 }
3041 | InlineLet {
3042 sig_id,
3043 from_stage_id,
3044 to_stage_id,
3045 ..
3046 } => StageTransition::Replace {
3047 sig_id: sig_id.clone(),
3048 from: from_stage_id.clone(),
3049 to: to_stage_id.clone(),
3050 },
3051 RenameSymbol {
3052 from,
3053 to,
3054 body_stage_id,
3055 } => StageTransition::Rename {
3056 from: from.clone(),
3057 to: to.clone(),
3058 body_stage_id: body_stage_id.clone(),
3059 },
3060 AddImport { .. } | RemoveImport { .. } => StageTransition::ImportOnly,
3061 Merge { .. } => StageTransition::Merge {
3062 entries: Default::default(),
3063 },
3064 // #294: a Candidate proposes a stage without advancing
3065 // the branch. ImportOnly keeps the branch head untouched
3066 // — the stage IS published on disk (Store::propose_candidate
3067 // calls publish before apply), but no head delta lands.
3068 Candidate { .. } => StageTransition::ImportOnly,
3069 // A Promote advances the head exactly like ModifyBody
3070 // (or Create when the sig had no head). The winner
3071 // stage is the new branch state for that sig.
3072 Promote {
3073 sig_id,
3074 winner_stage_id,
3075 from_stage_id,
3076 ..
3077 } => match from_stage_id {
3078 Some(from) => StageTransition::Replace {
3079 sig_id: sig_id.clone(),
3080 from: from.clone(),
3081 to: winner_stage_id.clone(),
3082 },
3083 None => StageTransition::Create {
3084 sig_id: sig_id.clone(),
3085 stage_id: winner_stage_id.clone(),
3086 },
3087 },
3088 }
3089}
3090
3091/// Producer identity for TypeCheck attestations emitted by the
3092/// store-write gate. Pinned to this crate's name + version so an
3093/// attestation produced by a different `lex-store` revision is
3094/// distinguishable (content-hashed `produced_by`).
3095fn typecheck_producer() -> lex_vcs::ProducerDescriptor {
3096 lex_vcs::ProducerDescriptor {
3097 tool: "lex-store".into(),
3098 version: env!("CARGO_PKG_VERSION").into(),
3099 model: None,
3100 }
3101}
3102
3103/// Producer identity for `Examples::Passed` attestations emitted by
3104/// [`Store::record_examples_passed`] (#835). Distinct tool name so
3105/// the activity feed can tell an auto-emitted publish-time examples
3106/// verdict apart from an `lex agent-tool --examples` one.
3107fn examples_producer() -> lex_vcs::ProducerDescriptor {
3108 lex_vcs::ProducerDescriptor {
3109 tool: "lex-store::examples".into(),
3110 version: env!("CARGO_PKG_VERSION").into(),
3111 model: None,
3112 }
3113}
3114
3115/// Producer identity for `Review` attestations (#836). The reviewer's
3116/// own id lives in the kind; this records which tool minted the record.
3117fn review_producer(reviewer: &str) -> lex_vcs::ProducerDescriptor {
3118 lex_vcs::ProducerDescriptor {
3119 tool: format!("lex-store::review:{reviewer}"),
3120 version: env!("CARGO_PKG_VERSION").into(),
3121 model: None,
3122 }
3123}
3124
3125/// Producer identity for `RepairHint` attestations emitted by
3126/// `apply_operation_checked` on TypeError (#281). Distinct tool
3127/// name from `typecheck_producer` so consumers can filter the
3128/// activity feed for repair hints without scanning kinds.
3129fn repair_hint_producer() -> lex_vcs::ProducerDescriptor {
3130 lex_vcs::ProducerDescriptor {
3131 tool: "lex-store::repair_hint".into(),
3132 version: env!("CARGO_PKG_VERSION").into(),
3133 model: None,
3134 }
3135}
3136
3137/// Producer identity for `TrustWaived` attestations emitted by
3138/// the `required_attestations` gate on a trust-driven waiver
3139/// (#293). Distinct from `typecheck_producer` and `repair_hint`
3140/// so the audit trail clearly shows "the gate let this advance
3141/// through because trust > threshold."
3142fn trust_waived_producer() -> lex_vcs::ProducerDescriptor {
3143 lex_vcs::ProducerDescriptor {
3144 tool: "lex-store::trust_waived".into(),
3145 version: env!("CARGO_PKG_VERSION").into(),
3146 model: None,
3147 }
3148}
3149
3150/// Producer identity for `ProducerTrust` attestations emitted by
3151/// [`Store::recompute_producer_trust`]. The score-derivation
3152/// recompute is its own machine-emittable kind, distinct from
3153/// the gate-side `TrustWaived` emit (#293).
3154fn producer_trust_producer() -> lex_vcs::ProducerDescriptor {
3155 lex_vcs::ProducerDescriptor {
3156 tool: "lex-store::producer_trust".into(),
3157 version: env!("CARGO_PKG_VERSION").into(),
3158 model: None,
3159 }
3160}
3161
3162/// The set of stage_ids a transition introduces. These are the
3163/// stages a successful TypeCheck pass attests *about* — the new
3164/// head produced by Create/Replace, the renamed body, or the per-
3165/// sig resolution of a Merge. Removes and ImportOnly produce no
3166/// attestable stage; the program typechecks but no specific stage
3167/// is the subject of the claim.
3168/// One row of input to the producer-block / required-attestations
3169/// gates: `(op_id, stage_id, op_effects)`. The `stage_id` is
3170/// `None` for ops that don't touch a stage (imports, empty
3171/// merges) — the gate skips those.
3172type GateCandidate = (
3173 lex_vcs::OpId,
3174 Option<String>,
3175 std::collections::BTreeSet<String>,
3176);
3177
3178/// Effect set declared *by the operation itself* (#245). Used by
3179/// the `required_attestations` gate's `EffectsIntersect` clause.
3180///
3181/// Only `AddFunction` and `ChangeEffectSig` carry an effect set in
3182/// their op payload; for everything else this returns the empty
3183/// set, which means `EffectsIntersect` rules don't fire on those
3184/// ops. `Always` rules continue to fire regardless. A future
3185/// improvement is to extract effects from the candidate `Stage`
3186/// for `ModifyBody` ops, but the typed-effects-on-ops path (#247)
3187/// is the cleaner solution and lands separately.
3188fn op_declared_effects(kind: &lex_vcs::OperationKind) -> std::collections::BTreeSet<String> {
3189 use lex_vcs::OperationKind::*;
3190 match kind {
3191 AddFunction { effects, .. } => effects.clone(),
3192 ChangeEffectSig { to_effects, .. } => to_effects.clone(),
3193 _ => std::collections::BTreeSet::new(),
3194 }
3195}
3196
3197fn attestable_stage_ids(transition: &lex_vcs::StageTransition) -> Vec<String> {
3198 use lex_vcs::StageTransition::*;
3199 match transition {
3200 Create { stage_id, .. } => vec![stage_id.clone()],
3201 Replace { to, .. } => vec![to.clone()],
3202 Rename { body_stage_id, .. } => vec![body_stage_id.clone()],
3203 Merge { entries } => entries.values().filter_map(|opt| opt.clone()).collect(),
3204 Remove { .. } | ImportOnly => Vec::new(),
3205 }
3206}
3207
3208/// True when two `FnDecl`s are identical except for their body — the
3209/// precondition for a pure intra-body three-way merge (#838). The
3210/// signature fields are equal by construction when both share a
3211/// `sig_id`; this also guards the non-signature fields (`type_params`,
3212/// `examples`) so a side that changed those isn't silently dropped.
3213fn fndecl_same_except_body(a: &lex_ast::FnDecl, b: &lex_ast::FnDecl) -> bool {
3214 a.name == b.name
3215 && a.type_params == b.type_params
3216 && a.params == b.params
3217 && a.effects == b.effects
3218 && a.effect_row_var == b.effect_row_var
3219 && a.return_type == b.return_type
3220 && a.examples == b.examples
3221}
3222
3223fn write_canonical_json<T: Serialize>(path: &Path, value: &T) -> Result<(), StoreError> {
3224 let v = serde_json::to_value(value)?;
3225 let s = lex_ast::canon_json::to_canonical_string(&v);
3226 if let Some(parent) = path.parent() {
3227 fs::create_dir_all(parent)?;
3228 }
3229 fs::write(path, s)?;
3230 Ok(())
3231}
3232
3233/// Read the `name` of the `Let` expression at `let_node` inside
3234/// `stage`'s body. Used by [`Store::apply_rename_local`] to record
3235/// the rename source. Returns the same `TransformError` shapes as
3236/// the transformer itself so callers see a consistent error
3237/// vocabulary.
3238fn read_let_name(
3239 stage: &Stage,
3240 let_node: &lex_ast::NodeId,
3241) -> Result<String, lex_ast::TransformError> {
3242 // The transformer is itself a pure function; ask it to perform
3243 // a rename to a sentinel value and read the resulting let's
3244 // original name from the output. Cheaper than duplicating the
3245 // node-walk here, and stays correct as the transform evolves.
3246 //
3247 // We use a sentinel that's invalid as a Lex identifier so even
3248 // if the rename somehow lands, downstream parsing would
3249 // surface it loudly. (The transform path discards the renamed
3250 // value — we only need the *original* name.)
3251 let probed = lex_ast::rename_local(stage, let_node, "__lex_rename_probe__")?;
3252 let Stage::FnDecl(fd) = probed else {
3253 return Err(lex_ast::TransformError::NonFnTarget {
3254 stage_kind: "non-FnDecl",
3255 });
3256 };
3257 // Walk back to the probed let to read its old name from the
3258 // *original* stage — the probed stage's let has already been
3259 // renamed.
3260 let Stage::FnDecl(orig_fd) = stage else {
3261 return Err(lex_ast::TransformError::NonFnTarget {
3262 stage_kind: "non-FnDecl",
3263 });
3264 };
3265 // Path-based lookup matches the transformer's navigation.
3266 let path = parse_let_node_path(let_node.as_str())?;
3267 if path.is_empty() {
3268 return Err(lex_ast::TransformError::NotALet {
3269 at: let_node.as_str().into(),
3270 found_kind: "stage_root",
3271 });
3272 }
3273 if path[0] != orig_fd.params.len() + 1 {
3274 return Err(lex_ast::TransformError::UnknownNode {
3275 at: let_node.as_str().into(),
3276 });
3277 }
3278 let inner = &path[1..];
3279 let target = navigate_to_let(&orig_fd.body, inner, let_node.as_str())?;
3280 let _ = fd; // probed stage discarded
3281 Ok(target.to_string())
3282}
3283
3284fn parse_let_node_path(id: &str) -> Result<Vec<usize>, lex_ast::TransformError> {
3285 let s = id
3286 .strip_prefix("n_")
3287 .ok_or_else(|| lex_ast::TransformError::BadNodeId(id.into()))?;
3288 let mut parts = s.split('.');
3289 let head = parts
3290 .next()
3291 .ok_or_else(|| lex_ast::TransformError::BadNodeId(id.into()))?;
3292 if head != "0" {
3293 return Err(lex_ast::TransformError::BadNodeId(id.into()));
3294 }
3295 let mut out = Vec::new();
3296 for p in parts {
3297 out.push(
3298 p.parse::<usize>()
3299 .map_err(|_| lex_ast::TransformError::BadNodeId(id.into()))?,
3300 );
3301 }
3302 Ok(out)
3303}
3304
3305fn navigate_to_let<'a>(
3306 root: &'a lex_ast::CExpr,
3307 path: &[usize],
3308 at: &str,
3309) -> Result<&'a str, lex_ast::TransformError> {
3310 use lex_ast::CExpr::*;
3311 let mut current = root;
3312 for &idx in path {
3313 current = match current {
3314 Call { callee, args } => {
3315 if idx == 0 {
3316 callee
3317 } else {
3318 args.get(idx - 1)
3319 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?
3320 }
3321 }
3322 Let { value, body, .. } => match idx {
3323 0 => value,
3324 1 => body,
3325 _ => return Err(lex_ast::TransformError::UnknownNode { at: at.into() }),
3326 },
3327 Match { scrutinee, arms } => {
3328 if idx == 0 {
3329 scrutinee
3330 } else {
3331 let arm_off = idx - 1;
3332 if arm_off % 2 != 1 {
3333 return Err(lex_ast::TransformError::UnknownNode { at: at.into() });
3334 }
3335 let arm_index = arm_off / 2;
3336 &arms
3337 .get(arm_index)
3338 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?
3339 .body
3340 }
3341 }
3342 Block { statements, result } => {
3343 if idx < statements.len() {
3344 &statements[idx]
3345 } else if idx == statements.len() {
3346 result
3347 } else {
3348 return Err(lex_ast::TransformError::UnknownNode { at: at.into() });
3349 }
3350 }
3351 Constructor { args, .. }
3352 | TupleLit { items: args, .. }
3353 | ListLit { items: args, .. } => args
3354 .get(idx)
3355 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?,
3356 RecordLit { fields } => {
3357 &fields
3358 .get(idx)
3359 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?
3360 .value
3361 }
3362 FieldAccess { value, .. } if idx == 0 => value,
3363 Lambda { body, .. } if idx == 0 => body,
3364 BinOp { lhs, rhs, .. } => match idx {
3365 0 => lhs,
3366 1 => rhs,
3367 _ => return Err(lex_ast::TransformError::UnknownNode { at: at.into() }),
3368 },
3369 UnaryOp { expr, .. } if idx == 0 => expr,
3370 Return { value } if idx == 0 => value,
3371 _ => return Err(lex_ast::TransformError::UnknownNode { at: at.into() }),
3372 };
3373 }
3374 let Let { name, .. } = current else {
3375 return Err(lex_ast::TransformError::NotALet {
3376 at: at.into(),
3377 found_kind: lex_cexpr_kind(current),
3378 });
3379 };
3380 Ok(name)
3381}
3382
3383fn lex_cexpr_kind(e: &lex_ast::CExpr) -> &'static str {
3384 use lex_ast::CExpr::*;
3385 match e {
3386 Literal { .. } => "Literal",
3387 Var { .. } => "Var",
3388 Call { .. } => "Call",
3389 Let { .. } => "Let",
3390 Match { .. } => "Match",
3391 Block { .. } => "Block",
3392 Constructor { .. } => "Constructor",
3393 RecordLit { .. } => "RecordLit",
3394 TupleLit { .. } => "TupleLit",
3395 ListLit { .. } => "ListLit",
3396 FieldAccess { .. } => "FieldAccess",
3397 Lambda { .. } => "Lambda",
3398 BinOp { .. } => "BinOp",
3399 UnaryOp { .. } => "UnaryOp",
3400 Return { .. } => "Return",
3401 }
3402}
3403
3404/// Extract the declared `[budget(N)]` integer from a stage's
3405/// effect set, if any (#280 + #247). Returns `None` for stages
3406/// that aren't `FnDecl` or don't carry a budget effect — same
3407/// shape as `lex_vcs::budget_from_effects`.
3408fn budget_of_stage(stage: &Stage) -> Option<u64> {
3409 let fd = match stage {
3410 Stage::FnDecl(fd) => fd,
3411 _ => return None,
3412 };
3413 let mut min_cost: Option<u64> = None;
3414 for eff in &fd.effects {
3415 if eff.name != "budget" {
3416 continue;
3417 }
3418 if let Some(lex_ast::EffectArg::Int { value }) = &eff.arg {
3419 let n = *value as u64;
3420 min_cost = Some(min_cost.map(|c| c.min(n)).unwrap_or(n));
3421 }
3422 }
3423 min_cost
3424}
3425
3426/// Serialize a stage to its canonical-JSON byte form. Used by
3427/// `publish_signed` for delta encoding (#261 slice 3) — both the
3428/// "compute the diff" path and the "write a full snapshot"
3429/// fallback need exactly the same bytes.
3430fn canonical_bytes(stage: &Stage) -> Result<Vec<u8>, StoreError> {
3431 let v = serde_json::to_value(stage)?;
3432 Ok(lex_ast::canon_json::to_canonical_string(&v).into_bytes())
3433}
3434
3435#[allow(dead_code)]
3436fn read_json<T: DeserializeOwned>(path: &Path) -> Result<T, StoreError> {
3437 let bytes = fs::read(path)?;
3438 Ok(serde_json::from_slice(&bytes)?)
3439}