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 /// A branch-head advance (e.g. the ref half of `op push`) was
36 /// asked to move `branch` to `attempted`, but `attempted` is not a
37 /// descendant of the branch's `current` head — a non-fast-forward
38 /// that would orphan history. Refused, git-style, so a disjoint or
39 /// diverged push can't silently clobber a shared branch. The op
40 /// objects may already be present; only the ref is left unchanged.
41 #[error("non-fast-forward on `{branch}`: {attempted} is not a descendant of current head {current}")]
42 NonFastForward { branch: String, current: lex_vcs::OpId, attempted: lex_vcs::OpId },
43 #[error("unknown blob `{0}`")]
44 UnknownBlob(String),
45 #[error("unknown blob ref `{namespace}/{key}`")]
46 UnknownBlobRef { namespace: String, key: String },
47 #[error("unknown op_id `{0}`")]
48 UnknownOp(lex_vcs::OpId),
49 /// A typed AST transform (#280) — e.g. `ReplaceMatchArm` — was
50 /// asked to operate on a node it couldn't address (wrong kind,
51 /// out-of-range arm index, unknown NodeId, etc.). Distinct from
52 /// `TypeError` (which means the transform succeeded but its
53 /// output didn't typecheck) so callers can render the right
54 /// error message.
55 #[error("transform failed: {0}")]
56 TransformError(lex_ast::TransformError),
57 #[error(transparent)]
58 Apply(#[from] lex_vcs::ApplyError),
59 /// The candidate program — i.e. the source the caller is
60 /// publishing — doesn't typecheck. The branch head is unchanged
61 /// and no op records are persisted. Issue #130's "always-valid
62 /// HEAD" invariant: the gate runs before any side effect, so a
63 /// type-broken publish leaves no footprint.
64 #[error("type errors in published program: {} error(s)", .0.len())]
65 TypeError(Vec<lex_types::TypeError>),
66 /// A dependency being resolved for the write-time gate (#930) is a
67 /// multi-module package. Per-module signature extraction (picking the
68 /// imported module's file out of the de-flattened tree) is not yet
69 /// implemented; single-module (leaf) dependencies resolve today. A
70 /// caller can treat this as "cannot resolve here" rather than a hard
71 /// failure.
72 #[error("multi-module dependency resolution is not yet supported")]
73 UnsupportedMultiModuleDependency,
74 /// A typed issue's example targets a function that isn't declared at
75 /// the head being evaluated (#949) — the example cannot be attached, so
76 /// the issue cannot be judged there.
77 #[error("issue example targets `{0}`, which is not declared at this head")]
78 IssueTarget(String),
79 /// The op was persisted but a `required_attestations` rule in
80 /// `policy.json` (#245) refused to advance the branch head past
81 /// it. The op record is durable — re-running with the missing
82 /// attestations recorded will succeed without re-persisting —
83 /// but the branch is unchanged. Surfaced as a structured JSON
84 /// envelope on the HTTP API.
85 #[error(
86 "branch advance blocked: op {} missing attestations: {}",
87 .0.op_id, .0.missing.join(", ")
88 )]
89 BranchAdvanceBlocked(crate::policy::BranchAdvanceBlocked),
90 /// All retry attempts of the CAS branch-head advance failed
91 /// because another writer kept advancing the same branch
92 /// (#262). The op record itself is durable in the op log
93 /// (orphaned), so re-running with backoff would eventually
94 /// land — return `503 Contention { retry_after }` from the
95 /// HTTP API and let the client back off.
96 #[error("branch advance contention on `{branch}`: {attempts} retries exhausted")]
97 Contention { branch: String, attempts: u32 },
98 /// The op was persisted but its stage carries an attestation
99 /// produced by a retroactively quarantined tool (#248). The
100 /// branch head is unchanged. The op record stays in the log
101 /// (audit trail intact); re-running with the producer
102 /// unblocked, or with un-contaminated attestations, succeeds
103 /// without re-persisting the op.
104 #[error(
105 "branch advance blocked: op {} touches stage {} with an attestation from \
106 quarantined producer `{}` (blocked at {}, attestation at {})",
107 .0.op_id, .0.stage_id, .0.tool_id, .0.blocked_at, .0.attestation_at
108 )]
109 ProducerBlocked(crate::policy::ProducerBlocked),
110 /// The op would push its session's monotonic budget over the
111 /// cap configured in `policy.session_budgets` (#292 slice 3).
112 /// The op is *not* persisted; the branch head is unchanged.
113 /// The caller should either start a new session, raise the
114 /// cap, or refactor to fit the budget. HTTP API maps to 503.
115 #[error("session `{session_id}` budget exceeded: spent_after={spent_after} > cap={cap}")]
116 BudgetExceeded {
117 session_id: String,
118 cap: u64,
119 spent_after: u64,
120 },
121}
122
123/// The outcome returned by [`Store::publish_program`].
124#[derive(Debug, Clone, serde::Serialize)]
125pub struct PublishOutcome {
126 pub ops: Vec<PublishOp>,
127 pub head_op: Option<lex_vcs::OpId>,
128}
129
130/// Everything a regenerator needs to *replay* an op (#836 G3), produced
131/// by [`Store::replay_request`]. The model call is external: a harness
132/// feeds `prompt` + `parent_program` to `model`, then hands the
133/// regenerated stage to [`Store::replay_compare`].
134#[derive(Debug, Clone, serde::Serialize)]
135pub struct ReplayRequest {
136 pub op_id: String,
137 /// The sig the op changed — the function to regenerate.
138 pub target_sig: String,
139 /// The target function's name (the recorded stage is a function
140 /// for every replayable op).
141 #[serde(default, skip_serializing_if = "Option::is_none")]
142 pub target_name: Option<String>,
143 /// The target function's rendered signature (`fn name(...) -> T`),
144 /// so a regenerator knows the interface to implement without
145 /// re-deriving it from the sig hash.
146 #[serde(default, skip_serializing_if = "Option::is_none")]
147 pub target_signature: Option<String>,
148 /// The stage id a faithful regeneration should reproduce.
149 pub expected_stage_id: String,
150 /// The recorded intent prompt (`None` if the op carried no intent).
151 pub prompt: Option<String>,
152 /// The recorded model (`provider/name[@version]`), if any.
153 pub model: Option<String>,
154 /// The recorded session id, if any.
155 pub session_id: Option<String>,
156 /// The program the change was made against — the parent state
157 /// rendered to source — the context a regenerator needs.
158 pub parent_program: String,
159}
160
161/// The result of comparing a regenerated candidate against an op's
162/// recorded output (#836 G3), returned by [`Store::replay_compare`]
163/// after it emits the `Replay` attestation.
164#[derive(Debug, Clone, serde::Serialize)]
165pub struct ReplayOutcome {
166 pub op_id: String,
167 pub expected_stage_id: String,
168 /// The candidate's stage id when it regenerated the same sig, else
169 /// `None`.
170 pub produced_stage_id: Option<String>,
171 /// Whether the regeneration reproduced the recorded change (exact or
172 /// behavioral).
173 pub reproduced: bool,
174 /// Set when reproduction was behavioral (same values over N sampled
175 /// inputs) rather than an exact stage-id match. `None` for an exact
176 /// match or a genuine miss.
177 #[serde(default, skip_serializing_if = "Option::is_none")]
178 pub behavioral_samples: Option<usize>,
179 /// The id of the `Replay` attestation this comparison emitted.
180 pub attestation_id: String,
181}
182
183/// One applied operation within a [`PublishOutcome`].
184#[derive(Debug, Clone, serde::Serialize)]
185pub struct PublishOp {
186 pub op_id: lex_vcs::OpId,
187 pub kind: serde_json::Value,
188}
189
190/// One entry in the per-`SigId` stage history surfaced by
191/// `Store::sig_history`. Newest-first ordering is the responsibility
192/// of the producer.
193#[derive(Debug, Clone, serde::Serialize, PartialEq)]
194pub struct StageHistoryEntry {
195 pub stage_id: String,
196 pub status: StageStatus,
197 /// Wall-clock seconds of the most recent transition.
198 pub last_at: u64,
199 /// Wall-clock seconds when this stage was first written to the
200 /// store (its initial Draft transition). `None` for stages
201 /// whose lifecycle log doesn't include an explicit Draft entry
202 /// — shouldn't happen for stages published via `Store::publish`,
203 /// but the type allows hand-edited stores.
204 #[serde(skip_serializing_if = "Option::is_none")]
205 pub published_at: Option<u64>,
206}
207
208/// Per-candidate metadata surfaced by [`Store::list_candidates`]
209/// (#294). Returned sorted by `op_id` for deterministic output.
210#[derive(Debug, Clone, serde::Serialize, serde::Deserialize, PartialEq, Eq)]
211pub struct CandidateInfo {
212 pub op_id: lex_vcs::OpId,
213 pub stage_id: lex_vcs::StageId,
214 /// Author intent. Always set for `Candidate` ops emitted via
215 /// [`Store::propose_candidate`]; `None` only if a
216 /// hand-written raw op skipped the intent tag.
217 pub intent_id: Option<lex_vcs::IntentId>,
218}
219
220/// One line of `stage_index.jsonl`. See `Store::lookup_lifecycle`.
221#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
222struct StageIndexEntry {
223 stage_id: String,
224 sig_id: String,
225}
226
227/// Sentinel `sig_id` value recording "a full scan already established
228/// this stage_id exists nowhere in the store" (#825). Never a real
229/// sig — sig directory names are never empty.
230const MISSING_STAGE_MARKER: &str = "";
231
232/// Resolves a head's external (registry/git) dependencies to their public
233/// module signatures, so the write-time gate can type-check a head that keeps
234/// `import "<pkg>/mod" as <alias>` edges instead of inlining the dependency
235/// (#930). The returned map is keyed by import *reference* (`"lex-nt/lib"`) →
236/// that dependency module's record type (as [`crate::render::module_record_at_op`]
237/// or a source-based equivalent produces), exactly the shape
238/// [`lex_types::check_program_with_modules`] consumes.
239///
240/// Implementations differ by context and live in the crate that has the
241/// resolution machinery: the client (`lex publish`) resolves from the
242/// working-copy `lex.lock` + local package cache; the hub resolves from the
243/// committed lock + its own hosted stores (cross-tenant), which a single
244/// [`Store`] cannot reach on its own. When no resolver is installed the gate
245/// resolves an empty map — only stdlib binds and any external reference is an
246/// unbound-name error, exactly as before #930 (so inlined heads, which carry
247/// no external edges, are unaffected).
248pub trait DepResolver: Send + Sync {
249 /// `head_op` is the op the head is known by when the gate has one (merge,
250 /// patch and hub-verify reconstruct a committed head); `None` for a
251 /// candidate not yet committed (`publish`), where the client resolver
252 /// falls back to the working-copy lock.
253 fn resolve_modules(
254 &self,
255 stages: &[Stage],
256 head_op: Option<&str>,
257 ) -> BTreeMap<String, lex_types::Ty>;
258}
259
260pub struct Store {
261 root: PathBuf,
262 /// Optional dependency resolver (#930). Injected by the client or the hub;
263 /// `None` in a bare store means the gate resolves an empty module map.
264 resolver: Option<std::sync::Arc<dyn DepResolver>>,
265}
266
267impl Store {
268 /// Open or create a store rooted at `root`.
269 pub fn open(root: impl AsRef<Path>) -> Result<Self, StoreError> {
270 let root = root.as_ref().to_path_buf();
271 fs::create_dir_all(root.join("stages"))?;
272 fs::create_dir_all(root.join("traces"))?;
273 let store = Self { root, resolver: None };
274 store.ensure_stage_index();
275 Ok(store)
276 }
277
278 /// Install the dependency resolver the write-time gate uses to type-check
279 /// heads that keep external `import` edges (#930). Builder-style so a
280 /// caller can write `Store::open(p)?.with_dep_resolver(r)`.
281 pub fn with_dep_resolver(mut self, resolver: std::sync::Arc<dyn DepResolver>) -> Self {
282 self.resolver = Some(resolver);
283 self
284 }
285
286 /// Set the dependency resolver in place (for a `Store` already owned, e.g.
287 /// behind a `Mutex` in the HTTP `State`).
288 pub fn set_dep_resolver(&mut self, resolver: std::sync::Arc<dyn DepResolver>) {
289 self.resolver = Some(resolver);
290 }
291
292 /// The resolved dependency module map for a head being gated — the
293 /// installed resolver's answer, or empty when none is installed (#930).
294 fn resolved_modules(
295 &self,
296 stages: &[Stage],
297 head_op: Option<&str>,
298 ) -> BTreeMap<String, lex_types::Ty> {
299 match &self.resolver {
300 Some(r) => r.resolve_modules(stages, head_op),
301 None => BTreeMap::new(),
302 }
303 }
304
305 /// One-time migration for a store that predates the reverse
306 /// index (#822), or whose previous rebuild pass never finished
307 /// (e.g. the process was killed or its client disconnected
308 /// mid-request — server-side work keeps running either way, but
309 /// a *restart* genuinely stops it): build the index in a single
310 /// pass instead of leaving every subsequent `lookup_lifecycle`
311 /// call to discover its own entry via the slow per-call scan-
312 /// and-backfill fallback.
313 ///
314 /// That per-call fallback is fine for the rare individual miss
315 /// it was designed for, but pathological as a *bulk* cold-start
316 /// strategy: on a tenant with a few thousand functions it means
317 /// redoing an O(total sigs) scan from scratch for *each* of a
318 /// few thousand cold entries — O(total sigs²) — which measured
319 /// as a near-stall (page-cache thrashing) on a memory-
320 /// constrained host. A single pass over `list_sigs()` is
321 /// O(total sigs) total.
322 ///
323 /// Gated on a dedicated completion marker
324 /// (`stage_index.complete`), NOT on `stage_index.jsonl`'s mere
325 /// existence — a partially-built index file (left behind by an
326 /// interrupted rebuild, lazy or bulk) must still trigger a
327 /// re-run so the remaining entries get backfilled in one more
328 /// cheap O(total sigs) pass, not silently be mistaken for
329 /// "already done" and fall back to the slow per-call path for
330 /// whatever's left. `rebuild_stage_index` already skips entries
331 /// it finds present, so re-running it against a partial index
332 /// only does the work that remains. The marker is written only
333 /// after a full pass returns `Ok`, so a failed pass (e.g. an I/O
334 /// error partway through `list_sigs`) is retried on the next
335 /// open rather than being marked done.
336 ///
337 /// Runs once per `Store::open` call — which, in a long-lived
338 /// server (lex-hub caches one `Store` per tenant for the life of
339 /// the process), means once per tenant per process lifetime, not
340 /// once per request. Once the marker exists (the steady state
341 /// after the first successful run on any given host) this is a
342 /// single cheap file-existence check. Best-effort like the rest
343 /// of the index: any failure here just leaves the slower per-call
344 /// fallback as the only path, never breaks correctness.
345 fn ensure_stage_index(&self) {
346 if self.stage_index_complete_marker_path().exists() {
347 return;
348 }
349 if self.rebuild_stage_index().is_ok() {
350 let _ = fs::write(self.stage_index_complete_marker_path(), "");
351 }
352 }
353
354 fn stage_index_complete_marker_path(&self) -> PathBuf {
355 self.root.join("stage_index.complete")
356 }
357
358 /// Build (or top up) the reverse index in one pass over every
359 /// SigId in the store, rather than relying on `lookup_lifecycle`
360 /// to discover entries one at a time. Safe to call at any time,
361 /// including on a partially-built index (e.g. one left behind by
362 /// an interrupted request that was populating it lazily): already-
363 /// indexed stage_ids are skipped, so this only does the work that
364 /// remains. Returns the number of newly-added entries.
365 pub fn rebuild_stage_index(&self) -> Result<usize, StoreError> {
366 // A sig's lifecycle can list the same stage_id more than once
367 // (Draft, then later Active, then Deprecated all carry the
368 // same stage_id with a different status) -- track newly-seen
369 // keys locally too, not just what was already on disk at the
370 // start, so a repeated stage_id within one sig's transitions
371 // doesn't get appended to the index more than once.
372 let mut existing = self.load_stage_index();
373 let mut added = 0usize;
374 for sig in self.list_sigs()? {
375 let Ok(life) = self.read_lifecycle(&sig) else { continue };
376 for t in &life.transitions {
377 if !existing.contains_key(&t.stage_id) {
378 self.append_stage_index_entry(&t.stage_id, &sig);
379 existing.insert(t.stage_id.clone(), sig.clone());
380 added += 1;
381 }
382 }
383 }
384 Ok(added)
385 }
386
387 pub fn root(&self) -> &Path {
388 &self.root
389 }
390
391 // ── Generic content-addressed blobs (#5 / M6.1) ──────────────────────────
392 //
393 // The stage store holds typed Lex ASTs; loom-style artifacts (generated
394 // code, JSON, prose) are opaque text. These blob methods give the store a
395 // generic content-addressed object alongside stages, plus a lightweight
396 // ref namespace so callers can bind names (e.g. a sprint's node ids) to
397 // blob shas without touching the operation-log branch machinery.
398 //
399 // The sha is the lowercase hex SHA-256 of the content's UTF-8 bytes —
400 // identical to Lex's `crypto.sha256_str`, so a blob written here and an
401 // artifact content-addressed in loom's SQLite store share the same id and
402 // are interchangeable by reference. Store-scoped, so under lex-hub each
403 // tenant store gets its own blob space for free.
404
405 fn blobs_dir(&self) -> PathBuf {
406 self.root.join("blobs")
407 }
408
409 fn blob_refs_dir(&self) -> PathBuf {
410 self.root.join("blobrefs")
411 }
412
413 /// Content-address `content` and persist it under `<root>/blobs/<sha>`.
414 /// Returns the sha. Idempotent: re-putting identical content is a no-op.
415 /// Concurrency-safe — writes to a unique temp file then atomically renames
416 /// onto the content-addressed path, so parallel writers of the same content
417 /// can't corrupt it.
418 pub fn put_blob(&self, content: &str) -> Result<String, StoreError> {
419 use sha2::{Digest, Sha256};
420 let sha = hex::encode(Sha256::digest(content.as_bytes()));
421 let dir = self.blobs_dir();
422 let path = dir.join(&sha);
423 if path.exists() {
424 return Ok(sha);
425 }
426 fs::create_dir_all(&dir)?;
427 static SEQ: std::sync::atomic::AtomicU64 = std::sync::atomic::AtomicU64::new(0);
428 let n = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
429 let tmp = dir.join(format!(".{sha}.{}.{n}.tmp", std::process::id()));
430 fs::write(&tmp, content.as_bytes())?;
431 // rename is atomic on the same filesystem; identical content makes a
432 // last-writer-wins race harmless.
433 fs::rename(&tmp, &path)?;
434 Ok(sha)
435 }
436
437 /// Read a blob by its sha. `UnknownBlob` if absent.
438 pub fn get_blob(&self, sha: &str) -> Result<String, StoreError> {
439 match fs::read_to_string(self.blobs_dir().join(sha)) {
440 Ok(s) => Ok(s),
441 Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
442 Err(StoreError::UnknownBlob(sha.to_string()))
443 }
444 Err(e) => Err(StoreError::Io(e)),
445 }
446 }
447
448 /// Whether a blob with this sha exists.
449 pub fn has_blob(&self, sha: &str) -> bool {
450 self.blobs_dir().join(sha).exists()
451 }
452
453 /// Bind `key` to a blob `sha` within `namespace` (e.g. namespace
454 /// `"loom/sprint-abc"`, key `"build-node"`). Overwrites an existing
455 /// binding. The namespace may contain `/`; neither namespace nor key may
456 /// contain a `..` path component.
457 pub fn set_blob_ref(&self, namespace: &str, key: &str, sha: &str) -> Result<(), StoreError> {
458 let dir = self.blob_ref_namespace_dir(namespace, key)?;
459 fs::create_dir_all(&dir)?;
460 fs::write(dir.join(key), sha.as_bytes())?;
461 Ok(())
462 }
463
464 /// Resolve `namespace`/`key` to a blob sha. `UnknownBlobRef` if unbound.
465 pub fn get_blob_ref(&self, namespace: &str, key: &str) -> Result<String, StoreError> {
466 let dir = self.blob_ref_namespace_dir(namespace, key)?;
467 match fs::read_to_string(dir.join(key)) {
468 Ok(s) => Ok(s.trim().to_string()),
469 Err(e) if e.kind() == std::io::ErrorKind::NotFound => Err(StoreError::UnknownBlobRef {
470 namespace: namespace.to_string(),
471 key: key.to_string(),
472 }),
473 Err(e) => Err(StoreError::Io(e)),
474 }
475 }
476
477 /// Blob-ref namespace for committed lockfiles (#930 phase 2b-1).
478 const LOCK_NS: &'static str = "lock";
479
480 /// Record the `lex.lock` committed with the package head `head_op` — the
481 /// exact dependency versions and op-log heads that head was built and
482 /// type-checks against (#930 phase 2b-1: "HEAD + its committed lex.lock
483 /// always type-checks"). Content-addressed via [`Self::put_blob`] and
484 /// bound under the `lock` namespace keyed by the head op, so it is
485 /// idempotent (a re-push converges) and travels with the package through
486 /// the same object-sync path as stages and intents. Keyed by head op
487 /// rather than by branch so re-verifying a *historical* head resolves it
488 /// against the lock that head actually committed, not whatever the branch
489 /// points at now.
490 pub fn set_committed_lock(&self, head_op: &str, lock_toml: &str) -> Result<(), StoreError> {
491 let sha = self.put_blob(lock_toml)?;
492 self.set_blob_ref(Self::LOCK_NS, head_op, &sha)
493 }
494
495 /// The `lex.lock` committed with `head_op`, or `None` when the head
496 /// carries no committed lock — a dependency-free package, or one
497 /// published before locks were committed (the write-time gate then has no
498 /// registry/git dependencies to resolve, exactly as today).
499 pub fn committed_lock(&self, head_op: &str) -> Result<Option<String>, StoreError> {
500 match self.get_blob_ref(Self::LOCK_NS, head_op) {
501 Ok(sha) => Ok(Some(self.get_blob(&sha)?)),
502 Err(StoreError::UnknownBlobRef { .. }) => Ok(None),
503 Err(e) => Err(e),
504 }
505 }
506
507 /// All `key → sha` bindings in a namespace (e.g. every artifact in a
508 /// sprint). Empty map if the namespace has no bindings yet.
509 pub fn list_blob_refs(
510 &self,
511 namespace: &str,
512 ) -> Result<std::collections::BTreeMap<String, String>, StoreError> {
513 let dir = self.blob_ref_namespace_dir(namespace, "x")?;
514 let mut out = std::collections::BTreeMap::new();
515 let entries = match fs::read_dir(&dir) {
516 Ok(e) => e,
517 Err(e) if e.kind() == std::io::ErrorKind::NotFound => return Ok(out),
518 Err(e) => return Err(StoreError::Io(e)),
519 };
520 for entry in entries {
521 let entry = entry?;
522 if entry.file_type()?.is_file() {
523 let key = entry.file_name().to_string_lossy().to_string();
524 let sha = fs::read_to_string(entry.path())?.trim().to_string();
525 out.insert(key, sha);
526 }
527 }
528 Ok(out)
529 }
530
531 // Resolve the on-disk dir for a (namespace, key), rejecting `..` traversal
532 // and `/` in the key. `key` is validated but not joined here (callers join
533 // it themselves so `list_blob_refs` can pass a dummy).
534 fn blob_ref_namespace_dir(&self, namespace: &str, key: &str) -> Result<PathBuf, StoreError> {
535 if key.contains('/') || key.contains('\\') || key.split('/').any(|c| c == "..") {
536 return Err(StoreError::UnknownBlobRef {
537 namespace: namespace.to_string(),
538 key: key.to_string(),
539 });
540 }
541 let mut dir = self.blob_refs_dir();
542 for comp in namespace.split('/') {
543 if comp == ".." || comp.contains('\\') {
544 return Err(StoreError::UnknownBlobRef {
545 namespace: namespace.to_string(),
546 key: key.to_string(),
547 });
548 }
549 if !comp.is_empty() {
550 dir.push(comp);
551 }
552 }
553 Ok(dir)
554 }
555
556 fn now() -> u64 {
557 SystemTime::now()
558 .duration_since(UNIX_EPOCH)
559 .map(|d| d.as_secs())
560 .unwrap_or(0)
561 }
562
563 fn sig_dir(&self, sig: &str) -> PathBuf {
564 self.root.join("stages").join(sig)
565 }
566 fn impl_dir(&self, sig: &str) -> PathBuf {
567 self.sig_dir(sig).join("implementations")
568 }
569 fn tests_dir(&self, sig: &str) -> PathBuf {
570 self.sig_dir(sig).join("tests")
571 }
572 fn specs_dir(&self, sig: &str) -> PathBuf {
573 self.sig_dir(sig).join("specs")
574 }
575 fn lifecycle_path(&self, sig: &str) -> PathBuf {
576 self.sig_dir(sig).join("lifecycle.json")
577 }
578
579 // ---- publish ----
580
581 /// Publish a stage as **Draft**. Returns the StageId.
582 /// Idempotent: republishing the same canonical AST returns the same
583 /// StageId without writing duplicates.
584 pub fn publish(&self, stage: &Stage) -> Result<String, StoreError> {
585 self.publish_signed(stage, None)
586 }
587
588 /// Like [`Self::publish`] but optionally attaches an Ed25519
589 /// signature over the StageId (#227). When `signer` is `Some`,
590 /// the persisted metadata gets a `signature` field that
591 /// downstream consumers can verify via
592 /// [`lex_vcs::verify_stage_id`].
593 ///
594 /// Idempotency: if a metadata file already exists the signature
595 /// is *not* re-written. This preserves "republishing is a no-op"
596 /// even across different signers — promoting a signed stage
597 /// requires a fresh stage hash anyway, so a metadata overwrite
598 /// would be the wrong primitive.
599 pub fn publish_signed(
600 &self,
601 stage: &Stage,
602 signer: Option<&lex_vcs::Keypair>,
603 ) -> Result<String, StoreError> {
604 let sig = sig_id(stage).ok_or(StoreError::CannotPublishImport)?;
605 let stage_id = stage_id(stage).ok_or(StoreError::CannotPublishImport)?;
606 let name = stage_name(stage).to_string();
607
608 fs::create_dir_all(self.impl_dir(&sig))?;
609 fs::create_dir_all(self.tests_dir(&sig))?;
610 fs::create_dir_all(self.specs_dir(&sig))?;
611
612 let ast_path = self.impl_dir(&sig).join(format!("{}.ast.json", stage_id));
613 let delta_path = self.impl_dir(&sig).join(format!("{}.delta.json", stage_id));
614 let meta_path = self
615 .impl_dir(&sig)
616 .join(format!("{}.metadata.json", stage_id));
617
618 // #261 slice 3: try delta encoding against the most recent
619 // prior stage in this sig's lifecycle. Falls back to a full
620 // snapshot when (a) no prior stage exists, (b) the diff
621 // ratio is over the threshold, or (c) the delta chain is
622 // already at its cap. The decision is internal — callers
623 // see the same `Stage` object on `get_ast` regardless.
624 if !ast_path.exists() && !delta_path.exists() {
625 self.persist_stage_bytes(&sig, &stage_id, stage, &ast_path, &delta_path)?;
626 }
627 if !meta_path.exists() {
628 let signature = signer.map(|kp| kp.sign_stage_id(&stage_id));
629 let metadata = Metadata {
630 stage_id: stage_id.clone(),
631 sig_id: sig.clone(),
632 name,
633 published_at: Self::now(),
634 note: None,
635 signature,
636 };
637 write_canonical_json(&meta_path, &metadata)?;
638 }
639
640 // Lifecycle: append a Draft transition for first publish.
641 let mut life = self.read_lifecycle(&sig).unwrap_or_else(|_| Lifecycle {
642 sig_id: sig.clone(),
643 ..Default::default()
644 });
645 if !life.transitions.iter().any(|t| t.stage_id == stage_id) {
646 life.transitions.push(Transition {
647 stage_id: stage_id.clone(),
648 from: StageStatus::Draft, // synthesized; "from" of first transition is itself
649 to: StageStatus::Draft,
650 at: Self::now(),
651 reason: None,
652 });
653 self.write_lifecycle(&sig, &life)?;
654 // Register the new stage_id's owning sig up front so a
655 // later `lookup_lifecycle` (e.g. `get_ast`) never needs
656 // to fall back to a full tenant-wide scan for it.
657 self.append_stage_index_entry(&stage_id, &sig);
658 }
659 Ok(stage_id)
660 }
661
662 // ---- lifecycle ----
663
664 pub fn activate(&self, stage_id: &str) -> Result<(), StoreError> {
665 let (sig, mut life) = self.lookup_lifecycle(stage_id)?;
666 // Demote any currently-Active impls for this SigId to Deprecated.
667 let active = life.current_active().map(|s| s.to_string());
668 if let Some(prev) = active {
669 if prev != stage_id {
670 life.transitions.push(Transition {
671 stage_id: prev,
672 from: StageStatus::Active,
673 to: StageStatus::Deprecated,
674 at: Self::now(),
675 reason: Some("superseded".into()),
676 });
677 }
678 }
679 let cur = life.status_of(stage_id);
680 if cur == Some(StageStatus::Tombstone) {
681 return Err(StoreError::InvalidTransition(
682 "tombstoned cannot be activated".into(),
683 ));
684 }
685 life.transitions.push(Transition {
686 stage_id: stage_id.into(),
687 from: cur.unwrap_or(StageStatus::Draft),
688 to: StageStatus::Active,
689 at: Self::now(),
690 reason: None,
691 });
692 self.write_lifecycle(&sig, &life)
693 }
694
695 pub fn deprecate(&self, stage_id: &str, reason: impl Into<String>) -> Result<(), StoreError> {
696 let (sig, mut life) = self.lookup_lifecycle(stage_id)?;
697 let cur = life
698 .status_of(stage_id)
699 .ok_or_else(|| StoreError::UnknownStage(stage_id.into()))?;
700 if cur != StageStatus::Active {
701 return Err(StoreError::InvalidTransition(format!(
702 "{cur:?} ⇒ Deprecated"
703 )));
704 }
705 life.transitions.push(Transition {
706 stage_id: stage_id.into(),
707 from: cur,
708 to: StageStatus::Deprecated,
709 at: Self::now(),
710 reason: Some(reason.into()),
711 });
712 self.write_lifecycle(&sig, &life)
713 }
714
715 pub fn tombstone(&self, stage_id: &str) -> Result<(), StoreError> {
716 let (sig, mut life) = self.lookup_lifecycle(stage_id)?;
717 let cur = life
718 .status_of(stage_id)
719 .ok_or_else(|| StoreError::UnknownStage(stage_id.into()))?;
720 if cur != StageStatus::Deprecated {
721 return Err(StoreError::InvalidTransition(format!(
722 "{cur:?} ⇒ Tombstone"
723 )));
724 }
725 life.transitions.push(Transition {
726 stage_id: stage_id.into(),
727 from: cur,
728 to: StageStatus::Tombstone,
729 at: Self::now(),
730 reason: None,
731 });
732 self.write_lifecycle(&sig, &life)
733 }
734
735 // ---- queries ----
736
737 /// The current Active StageId for a signature, or `None`.
738 pub fn resolve_sig(&self, sig: &str) -> Result<Option<String>, StoreError> {
739 let life = match self.read_lifecycle(sig) {
740 Ok(l) => l,
741 Err(_) => return Ok(None),
742 };
743 Ok(life.current_active().map(|s| s.to_string()))
744 }
745
746 /// Per-stage history for a SigId, ordered chronologically by
747 /// the *last* transition timestamp. Returns one entry per
748 /// distinct StageId that has ever been published under `sig`.
749 /// `Ok(vec![])` if the SigId doesn't exist in the store.
750 ///
751 /// Used by `lex blame` to render "where does this fn come from".
752 pub fn sig_history(&self, sig: &str) -> Result<Vec<StageHistoryEntry>, StoreError> {
753 let life = match self.read_lifecycle(sig) {
754 Ok(l) => l,
755 Err(_) => return Ok(Vec::new()),
756 };
757 // Collapse transitions: latest status + last_at per stage,
758 // plus the timestamp of the first Draft transition (≈ when
759 // the stage was published) when one exists.
760 let mut by_stage: indexmap::IndexMap<String, StageHistoryEntry> = indexmap::IndexMap::new();
761 for t in &life.transitions {
762 let entry = by_stage
763 .entry(t.stage_id.clone())
764 .or_insert(StageHistoryEntry {
765 stage_id: t.stage_id.clone(),
766 status: t.to,
767 last_at: t.at,
768 published_at: None,
769 });
770 entry.status = t.to;
771 entry.last_at = t.at;
772 if t.from == StageStatus::Draft && entry.published_at.is_none() {
773 entry.published_at = Some(t.at);
774 }
775 if t.to == StageStatus::Draft && entry.published_at.is_none() {
776 // Initial publication: Draft is the *destination*.
777 entry.published_at = Some(t.at);
778 }
779 }
780 let mut out: Vec<StageHistoryEntry> = by_stage.into_values().collect();
781 // Sort newest first so `lex blame` shows recent activity at top.
782 out.sort_by_key(|e| std::cmp::Reverse(e.last_at));
783 Ok(out)
784 }
785
786 pub fn get_ast(&self, stage_id: &str) -> Result<Stage, StoreError> {
787 let (sig, _) = self.lookup_lifecycle(stage_id)?;
788 let bytes = self.read_stage_canonical_bytes(&sig, stage_id)?;
789 Ok(serde_json::from_slice(&bytes)?)
790 }
791
792 /// Bulk AST fetch for callers that already know each stage's
793 /// **signature** — a branch head map, for instance, which is keyed
794 /// by SigId and whose values are the StageIds it points at.
795 ///
796 /// Prefer this over [`Self::get_asts_bulk`] whenever the SigId is in
797 /// hand, because resolving a StageId back to a SigId is not
798 /// reliable: a StageId hashes the structural signature plus the
799 /// implementation, deliberately *not* the name
800 /// (`docs/INVARIANTS.md`), so two functions that differ only in name
801 /// share one StageId while having two distinct SigIds — and two
802 /// separate ASTs, one under each sig directory. `stage_index` maps
803 /// each StageId to a single sig, so `get_ast`/`get_asts_bulk` return
804 /// whichever of those ASTs the index happens to name, i.e. the wrong
805 /// name half the time (#826). Reading straight from the sig the
806 /// caller already knows removes the ambiguity — and skips loading
807 /// the index at all.
808 ///
809 /// Returns results in the same order as `pairs`, `Err` for anything
810 /// that fails to resolve (mirroring `get_ast`'s error semantics).
811 pub fn get_asts_for_sigs_bulk(
812 &self,
813 pairs: &[(String, String)],
814 ) -> Vec<Result<Stage, StoreError>> {
815 pairs
816 .iter()
817 .map(|(sig_id, stage_id)| {
818 let bytes = self.read_stage_canonical_bytes(sig_id, stage_id)?;
819 Ok(serde_json::from_slice(&bytes)?)
820 })
821 .collect()
822 }
823
824 /// Bulk variant of [`Self::get_ast`] for callers resolving many
825 /// stage_ids at once (e.g. `pkg_publish_handler`'s `old_head`
826 /// scan over every live function in a tenant, once per publish
827 /// request). `get_ast` in a loop calls `lookup_lifecycle` once
828 /// per stage_id, and `lookup_lifecycle`'s index-hit path reads
829 /// and re-parses the *entire* `stage_index.jsonl` on every single
830 /// call — fine for one call, but O(index size × N) for N calls in
831 /// a row, which dominates once the index itself is large (#825's
832 /// follow-up: still correct and far better than the pre-index
833 /// full-tenant-scan-per-call behavior, but the per-call reparse
834 /// is itself a real, measured cost — 87.6s for 3,664 calls against
835 /// a ~14k-line index on the alpibrusl tenant).
836 ///
837 /// This loads the index once for the whole batch and keeps it in
838 /// memory across all `stage_ids`, only touching disk again to
839 /// append genuinely new entries (a positive backfill or a
840 /// negative "not found anywhere" cache, same as the single-call
841 /// path) — never to re-read what's already loaded.
842 ///
843 /// Returns results in the same order as `stage_ids`, `Err` for
844 /// anything that fails to resolve (mirroring `get_ast`'s error
845 /// semantics per call).
846 pub fn get_asts_bulk(&self, stage_ids: &[String]) -> Vec<Result<Stage, StoreError>> {
847 let mut index = self.load_stage_index();
848 let mut sigs_cache: BTreeMap<String, Option<Lifecycle>> = BTreeMap::new();
849 let mut all_sigs: Option<Vec<String>> = None;
850
851 stage_ids
852 .iter()
853 .map(|stage_id| {
854 self.lookup_lifecycle_bulk(stage_id, &mut index, &mut sigs_cache, &mut all_sigs)
855 .and_then(|sig| {
856 let bytes = self.read_stage_canonical_bytes(&sig, stage_id)?;
857 Ok(serde_json::from_slice(&bytes)?)
858 })
859 })
860 .collect()
861 }
862
863 /// Shared implementation behind [`Self::get_asts_bulk`]: identical
864 /// logic to [`Self::lookup_lifecycle`], but reads and writes the
865 /// caller-supplied `index` map instead of reloading it from disk
866 /// on every call, and memoizes `read_lifecycle` per sig and the
867 /// `list_sigs()` full-scan list across the whole batch. Disk
868 /// writes for newly-discovered entries (positive or negative)
869 /// still happen immediately, same as the single-call path — only
870 /// the repeated *reads* are batched away.
871 fn lookup_lifecycle_bulk(
872 &self,
873 stage_id: &str,
874 index: &mut BTreeMap<String, String>,
875 sigs_cache: &mut BTreeMap<String, Option<Lifecycle>>,
876 all_sigs: &mut Option<Vec<String>>,
877 ) -> Result<String, StoreError> {
878 if let Some(sig) = index.get(stage_id) {
879 if sig == MISSING_STAGE_MARKER {
880 return Err(StoreError::UnknownStage(stage_id.into()));
881 }
882 let life = sigs_cache
883 .entry(sig.clone())
884 .or_insert_with(|| self.read_lifecycle(sig).ok());
885 if let Some(life) = life {
886 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
887 return Ok(sig.clone());
888 }
889 }
890 }
891 if all_sigs.is_none() {
892 *all_sigs = Some(self.list_sigs()?);
893 }
894 for sig in all_sigs.as_ref().unwrap() {
895 let life = sigs_cache
896 .entry(sig.clone())
897 .or_insert_with(|| self.read_lifecycle(sig).ok());
898 if let Some(life) = life {
899 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
900 self.append_stage_index_entry(stage_id, sig);
901 index.insert(stage_id.to_string(), sig.clone());
902 return Ok(sig.clone());
903 }
904 }
905 }
906 self.append_stage_index_entry(stage_id, MISSING_STAGE_MARKER);
907 index.insert(stage_id.to_string(), MISSING_STAGE_MARKER.to_string());
908 Err(StoreError::UnknownStage(stage_id.into()))
909 }
910
911 /// Read the canonical bytes of a stage, walking back through
912 /// any delta chain (#261 slice 3). The recursion ends at a
913 /// `<stage_id>.ast.json` file (a full snapshot) or, in the
914 /// degenerate case of a missing chain, with `UnknownStage`.
915 fn read_stage_canonical_bytes(&self, sig: &str, stage_id: &str) -> Result<Vec<u8>, StoreError> {
916 let ast_path = self.impl_dir(sig).join(format!("{}.ast.json", stage_id));
917 if ast_path.exists() {
918 return Ok(fs::read(&ast_path)?);
919 }
920 let delta_path = self.impl_dir(sig).join(format!("{}.delta.json", stage_id));
921 if !delta_path.exists() {
922 return Err(StoreError::UnknownStage(stage_id.into()));
923 }
924 let delta_bytes = fs::read(&delta_path)?;
925 let delta: crate::delta::StageDelta = serde_json::from_slice(&delta_bytes)?;
926 let base_bytes = self.read_stage_canonical_bytes(sig, &delta.base_stage_id)?;
927 crate::delta::apply(&base_bytes, &delta).map_err(|e| {
928 StoreError::Io(std::io::Error::new(
929 std::io::ErrorKind::InvalidData,
930 format!("applying delta for {stage_id}: {e}"),
931 ))
932 })
933 }
934
935 /// Persist a freshly-published stage's canonical bytes (#261
936 /// slice 3). Tries delta encoding against the most recent
937 /// prior stage in the sig's lifecycle; falls back to a full
938 /// snapshot when no base exists, the diff ratio is too high,
939 /// or the delta chain is already at its cap.
940 fn persist_stage_bytes(
941 &self,
942 sig: &str,
943 stage_id: &str,
944 stage: &Stage,
945 ast_path: &Path,
946 delta_path: &Path,
947 ) -> Result<(), StoreError> {
948 let new_bytes = canonical_bytes(stage)?;
949 if let Some((base_stage_id, base_chain_length)) = self.pick_delta_base(sig, stage_id)? {
950 let base_bytes = self.read_stage_canonical_bytes(sig, &base_stage_id)?;
951 let (prefix, suffix, middle) = crate::delta::splice(&base_bytes, &new_bytes);
952 let chain_length = base_chain_length + 1;
953 if crate::delta::is_worth_encoding(middle.len(), new_bytes.len(), chain_length) {
954 let delta = crate::delta::StageDelta {
955 base_stage_id,
956 chain_length,
957 common_prefix: prefix,
958 common_suffix: suffix,
959 middle_hex: hex::encode(&middle),
960 };
961 write_canonical_json(delta_path, &delta)?;
962 return Ok(());
963 }
964 }
965 // Fall through: full snapshot.
966 if let Some(parent) = ast_path.parent() {
967 fs::create_dir_all(parent)?;
968 }
969 fs::write(ast_path, &new_bytes)?;
970 Ok(())
971 }
972
973 /// Pick a base stage for delta encoding from the given sig's
974 /// lifecycle. Returns `(base_stage_id, base_chain_length)` for
975 /// the most-recent non-tombstoned prior stage, or `None` when
976 /// there is no candidate. The chain length is read off the
977 /// base's `.delta.json` (if any) to enforce the cap.
978 fn pick_delta_base(
979 &self,
980 sig: &str,
981 new_stage_id: &str,
982 ) -> Result<Option<(String, usize)>, StoreError> {
983 let life = self.read_lifecycle(sig).ok();
984 let Some(life) = life else {
985 return Ok(None);
986 };
987 // Walk transitions newest-first; pick the first prior
988 // stage that isn't this one and isn't tombstoned.
989 let mut latest_per_stage: indexmap::IndexMap<&str, StageStatus> = indexmap::IndexMap::new();
990 for t in &life.transitions {
991 latest_per_stage.insert(&t.stage_id, t.to);
992 }
993 let mut candidates: Vec<&str> = latest_per_stage
994 .iter()
995 .filter(|(id, status)| **id != new_stage_id && **status != StageStatus::Tombstone)
996 .map(|(id, _)| *id)
997 .collect();
998 // Reverse to get newest-first (transitions are append-only,
999 // so latest_per_stage's iteration order matches insertion
1000 // order, oldest-first).
1001 candidates.reverse();
1002 let Some(&base) = candidates.first() else {
1003 return Ok(None);
1004 };
1005 let base_chain_length = self.delta_chain_length(sig, base)?;
1006 Ok(Some((base.to_string(), base_chain_length)))
1007 }
1008
1009 /// Length of the delta chain ending at `stage_id`. Zero when
1010 /// the stage is a full snapshot (`.ast.json` present); the
1011 /// stored `chain_length` from `.delta.json` otherwise.
1012 fn delta_chain_length(&self, sig: &str, stage_id: &str) -> Result<usize, StoreError> {
1013 let ast_path = self.impl_dir(sig).join(format!("{}.ast.json", stage_id));
1014 if ast_path.exists() {
1015 return Ok(0);
1016 }
1017 let delta_path = self.impl_dir(sig).join(format!("{}.delta.json", stage_id));
1018 if !delta_path.exists() {
1019 return Ok(0);
1020 }
1021 let bytes = fs::read(&delta_path)?;
1022 let delta: crate::delta::StageDelta = serde_json::from_slice(&bytes)?;
1023 Ok(delta.chain_length)
1024 }
1025
1026 pub fn get_metadata(&self, stage_id: &str) -> Result<Metadata, StoreError> {
1027 let (sig, _) = self.lookup_lifecycle(stage_id)?;
1028 let path = self
1029 .impl_dir(&sig)
1030 .join(format!("{}.metadata.json", stage_id));
1031 let bytes = fs::read(&path)?;
1032 Ok(serde_json::from_slice(&bytes)?)
1033 }
1034
1035 pub fn get_status(&self, stage_id: &str) -> Result<StageStatus, StoreError> {
1036 let (_sig, life) = self.lookup_lifecycle(stage_id)?;
1037 life.status_of(stage_id)
1038 .ok_or_else(|| StoreError::UnknownStage(stage_id.into()))
1039 }
1040
1041 pub fn list_stages_by_name(&self, name: &str) -> Result<Vec<String>, StoreError> {
1042 // Walk every SigId → check metadata of any implementation; if its
1043 // name matches, include the SigId.
1044 let mut out = Vec::new();
1045 let stages_dir = self.root.join("stages");
1046 if !stages_dir.exists() {
1047 return Ok(out);
1048 }
1049 for entry in fs::read_dir(&stages_dir)? {
1050 let entry = entry?;
1051 let sig_dir = entry.path();
1052 if !sig_dir.is_dir() {
1053 continue;
1054 }
1055 let sig = entry.file_name().to_string_lossy().to_string();
1056 // Look at any one metadata file under this SigId.
1057 let impls = self.impl_dir(&sig);
1058 if !impls.exists() {
1059 continue;
1060 }
1061 for f in fs::read_dir(impls)? {
1062 let f = f?;
1063 let p = f.path();
1064 if p.extension().is_some_and(|e| e == "json")
1065 && p.file_name()
1066 .is_some_and(|n| n.to_string_lossy().ends_with(".metadata.json"))
1067 {
1068 if let Ok(bytes) = fs::read(&p) {
1069 if let Ok(m) = serde_json::from_slice::<Metadata>(&bytes) {
1070 if m.name == name {
1071 if !out.contains(&sig) {
1072 out.push(sig.clone());
1073 }
1074 break;
1075 }
1076 }
1077 }
1078 }
1079 }
1080 }
1081 out.sort();
1082 Ok(out)
1083 }
1084
1085 pub fn list_sigs(&self) -> Result<Vec<String>, StoreError> {
1086 let stages_dir = self.root.join("stages");
1087 let mut out = Vec::new();
1088 if !stages_dir.exists() {
1089 return Ok(out);
1090 }
1091 for entry in fs::read_dir(stages_dir)? {
1092 let entry = entry?;
1093 if entry.file_type()?.is_dir() {
1094 out.push(entry.file_name().to_string_lossy().to_string());
1095 }
1096 }
1097 out.sort();
1098 Ok(out)
1099 }
1100
1101 // ---- tests/specs as metadata (§4.4) ----
1102
1103 pub fn attach_test(&self, sig: &str, test: &Test) -> Result<String, StoreError> {
1104 if !self.sig_dir(sig).exists() {
1105 return Err(StoreError::UnknownSig(sig.into()));
1106 }
1107 fs::create_dir_all(self.tests_dir(sig))?;
1108 let path = self.tests_dir(sig).join(format!("{}.json", test.id));
1109 write_canonical_json(&path, test)?;
1110 Ok(test.id.clone())
1111 }
1112
1113 pub fn list_tests(&self, sig: &str) -> Result<Vec<Test>, StoreError> {
1114 let dir = self.tests_dir(sig);
1115 if !dir.exists() {
1116 return Ok(Vec::new());
1117 }
1118 let mut out = Vec::new();
1119 for f in fs::read_dir(dir)? {
1120 let f = f?;
1121 if f.path().extension().is_some_and(|e| e == "json") {
1122 let bytes = fs::read(f.path())?;
1123 out.push(serde_json::from_slice(&bytes)?);
1124 }
1125 }
1126 Ok(out)
1127 }
1128
1129 pub fn attach_spec(&self, sig: &str, spec: &Spec) -> Result<String, StoreError> {
1130 if !self.sig_dir(sig).exists() {
1131 return Err(StoreError::UnknownSig(sig.into()));
1132 }
1133 fs::create_dir_all(self.specs_dir(sig))?;
1134 let path = self.specs_dir(sig).join(format!("{}.json", spec.id));
1135 write_canonical_json(&path, spec)?;
1136 Ok(spec.id.clone())
1137 }
1138
1139 pub fn list_specs(&self, sig: &str) -> Result<Vec<Spec>, StoreError> {
1140 let dir = self.specs_dir(sig);
1141 if !dir.exists() {
1142 return Ok(Vec::new());
1143 }
1144 let mut out = Vec::new();
1145 for f in fs::read_dir(dir)? {
1146 let f = f?;
1147 if f.path().extension().is_some_and(|e| e == "json") {
1148 let bytes = fs::read(f.path())?;
1149 out.push(serde_json::from_slice(&bytes)?);
1150 }
1151 }
1152 Ok(out)
1153 }
1154
1155 // ---- traces (§4.2 / M7) ----
1156
1157 // Native run-trace store — gated behind the `trace` feature (depends on
1158 // lex-trace). Off when a lower crate (lex-runtime) depends on lex-store to
1159 // avoid a dependency cycle; the blob/stage store below is unaffected.
1160 #[cfg(feature = "trace")]
1161 fn trace_path(&self, run_id: &str) -> PathBuf {
1162 self.root.join("traces").join(run_id).join("trace.json")
1163 }
1164
1165 #[cfg(feature = "trace")]
1166 pub fn save_trace(&self, tree: &lex_trace::TraceTree) -> Result<String, StoreError> {
1167 let path = self.trace_path(&tree.run_id);
1168 write_canonical_json(&path, tree)?;
1169 Ok(tree.run_id.clone())
1170 }
1171
1172 #[cfg(feature = "trace")]
1173 pub fn load_trace(&self, run_id: &str) -> Result<lex_trace::TraceTree, StoreError> {
1174 let bytes = fs::read(self.trace_path(run_id))?;
1175 Ok(serde_json::from_slice(&bytes)?)
1176 }
1177
1178 pub fn list_traces(&self) -> Result<Vec<String>, StoreError> {
1179 let dir = self.root.join("traces");
1180 if !dir.exists() {
1181 return Ok(Vec::new());
1182 }
1183 let mut out = Vec::new();
1184 for entry in fs::read_dir(dir)? {
1185 let entry = entry?;
1186 if entry.file_type()?.is_dir() {
1187 out.push(entry.file_name().to_string_lossy().to_string());
1188 }
1189 }
1190 out.sort();
1191 Ok(out)
1192 }
1193
1194 // ---- internals ----
1195
1196 /// `<root>/stage_index.jsonl` — an append-only, best-effort
1197 /// reverse index (`StageId` -> owning `SigId`), one JSON object
1198 /// per line. Backs `lookup_lifecycle`'s fast path; see its doc
1199 /// comment. Not a second source of truth: every entry is
1200 /// reconstructible from `stages/<sig>/lifecycle.json`, so a
1201 /// missing, truncated, or entirely absent index file only costs
1202 /// a slower lookup (the pre-existing full scan), never
1203 /// correctness — matching this module's "filesystem is the
1204 /// source of truth" stance (see the module doc comment) rather
1205 /// than introducing an actual second database.
1206 fn stage_index_path(&self) -> PathBuf {
1207 self.root.join("stage_index.jsonl")
1208 }
1209
1210 /// Best-effort load of the whole reverse index into memory.
1211 /// Tolerates a missing file (no index yet) and a corrupt or
1212 /// torn last line (a crash mid-append under the single-writer
1213 /// Tier-1 assumption) by skipping lines that don't parse,
1214 /// rather than failing the lookup that triggered the load.
1215 fn load_stage_index(&self) -> std::collections::BTreeMap<String, String> {
1216 let mut out = std::collections::BTreeMap::new();
1217 let Ok(raw) = fs::read_to_string(self.stage_index_path()) else {
1218 return out;
1219 };
1220 for line in raw.lines() {
1221 if let Ok(entry) = serde_json::from_str::<StageIndexEntry>(line) {
1222 out.insert(entry.stage_id, entry.sig_id);
1223 }
1224 }
1225 out
1226 }
1227
1228 /// Best-effort append of one new `(stage_id, sig_id)` pair.
1229 /// Failure (e.g. a read-only filesystem) only costs a future
1230 /// full scan for this stage_id, never correctness, so it's
1231 /// swallowed rather than propagated.
1232 fn append_stage_index_entry(&self, stage_id: &str, sig: &str) {
1233 use std::io::Write;
1234 let entry = StageIndexEntry { stage_id: stage_id.into(), sig_id: sig.into() };
1235 let Ok(line) = serde_json::to_string(&entry) else { return };
1236 if let Ok(mut f) = fs::OpenOptions::new()
1237 .create(true)
1238 .append(true)
1239 .open(self.stage_index_path())
1240 {
1241 let _ = writeln!(f, "{line}");
1242 }
1243 }
1244
1245 /// Find which SigId owns a StageId, and that sig's lifecycle.
1246 ///
1247 /// Before the reverse index (#822): a full scan over *every*
1248 /// SigId in the tenant (`list_sigs()`, not scoped to the
1249 /// package being looked at), reading and parsing each one's
1250 /// `lifecycle.json` until a match turned up. `get_ast` — called
1251 /// once per pre-existing function when building a publish
1252 /// request's `old_fns_by_name` (`lex-api/src/handlers.rs`) —
1253 /// calls this once per function, so a tenant with a few thousand
1254 /// published functions turned a single publish into millions of
1255 /// individual file reads; measured at roughly an hour on the
1256 /// `alpibrusl` tenant's ~2,400-function store.
1257 ///
1258 /// Now: check the persisted reverse index first (one sequential
1259 /// file read instead of up to N separate ones). A miss — the
1260 /// index doesn't exist yet, or this stage_id predates it — falls
1261 /// back to the full scan and backfills the index so the next
1262 /// lookup for the same stage_id is fast.
1263 fn lookup_lifecycle(&self, stage_id: &str) -> Result<(String, Lifecycle), StoreError> {
1264 let index = self.load_stage_index();
1265 if let Some(sig) = index.get(stage_id) {
1266 if sig == MISSING_STAGE_MARKER {
1267 // A previous full scan already established this
1268 // stage_id exists nowhere in the store. Re-scanning
1269 // would find nothing again -- see #825: a genuinely
1270 // orphaned reference (e.g. from data predating some
1271 // store migration) is looked up on *every* call that
1272 // needs it, forever, so without this negative cache
1273 // it silently costs a full O(total sigs) scan each
1274 // time, indistinguishable from the positive case at
1275 // the call site. Measured directly: on the alpibrusl
1276 // tenant, 988 of 3,664 branch-head entries are
1277 // orphaned this way, turning one `pkg publish`'s
1278 // old_fns_by_name build into ~16M wasted lifecycle
1279 // reads.
1280 return Err(StoreError::UnknownStage(stage_id.into()));
1281 }
1282 if let Ok(life) = self.read_lifecycle(sig) {
1283 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
1284 return Ok((sig.clone(), life));
1285 }
1286 }
1287 // Index entry is stale or wrong (shouldn't happen in
1288 // practice — sig ownership of a stage_id is permanent).
1289 // Fall through to the full scan below rather than trust it.
1290 }
1291 for sig in self.list_sigs()? {
1292 if let Ok(life) = self.read_lifecycle(&sig) {
1293 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
1294 self.append_stage_index_entry(stage_id, &sig);
1295 return Ok((sig, life));
1296 }
1297 }
1298 }
1299 // Genuinely not found anywhere: cache that fact so the next
1300 // lookup for this exact stage_id is an index hit, not another
1301 // full scan. Safe even if this stage_id somehow gets a real
1302 // sig later (content-addressed publish is idempotent, so
1303 // "later" only means "a byte-identical stage republished
1304 // under a real sig") — `append_stage_index_entry`'s later,
1305 // real entry is a later line in the file, and `load_stage_index`
1306 // folds duplicate keys last-write-wins, so the real entry wins.
1307 self.append_stage_index_entry(stage_id, MISSING_STAGE_MARKER);
1308 Err(StoreError::UnknownStage(stage_id.into()))
1309 }
1310
1311 fn read_lifecycle(&self, sig: &str) -> Result<Lifecycle, StoreError> {
1312 let path = self.lifecycle_path(sig);
1313 if !path.exists() {
1314 return Ok(Lifecycle {
1315 sig_id: sig.into(),
1316 transitions: Vec::new(),
1317 });
1318 }
1319 let bytes = fs::read(&path)?;
1320 Ok(serde_json::from_slice(&bytes)?)
1321 }
1322
1323 fn write_lifecycle(&self, sig: &str, life: &Lifecycle) -> Result<(), StoreError> {
1324 write_canonical_json(&self.lifecycle_path(sig), life)
1325 }
1326
1327 /// Apply a published program to a branch as a sequence of typed
1328 /// operations. Returns the ordered list of op_ids + the new
1329 /// head_op. The caller (`lex publish` CLI, `lex serve`'s HTTP
1330 /// handler) is responsible for computing the `DiffReport` against
1331 /// the current branch head — the diff infrastructure lives in
1332 /// `lex-vcs::compute_diff` (previously `lex-cli`) to keep this
1333 /// layer from owning diffing logic.
1334 ///
1335 /// On success: every op in the returned list is durable in the
1336 /// op log and the branch's head_op points at the last one.
1337 /// On a no-op (no diff): returns empty `ops` and the existing
1338 /// `head_op` unchanged.
1339 pub fn publish_program(
1340 &self,
1341 branch: &str,
1342 stages: &[lex_ast::Stage],
1343 diff: &lex_vcs::DiffReport,
1344 new_imports: &lex_vcs::ImportMap,
1345 activate: bool,
1346 ) -> Result<PublishOutcome, StoreError> {
1347 self.publish_program_signed(branch, stages, diff, new_imports, activate, None)
1348 }
1349
1350 /// Signed variant of [`Self::publish_program`] (#227). Every
1351 /// stage written under this batch gets the same signer; per-stage
1352 /// keys aren't supported because the agent identity model treats
1353 /// a publish as a single authorial act.
1354 pub fn publish_program_signed(
1355 &self,
1356 branch: &str,
1357 stages: &[lex_ast::Stage],
1358 diff: &lex_vcs::DiffReport,
1359 new_imports: &lex_vcs::ImportMap,
1360 activate: bool,
1361 signer: Option<&lex_vcs::Keypair>,
1362 ) -> Result<PublishOutcome, StoreError> {
1363 // Single-file / test callers don't publish a mangled package, so
1364 // there are no module prefixes to record (`in_file` stays `None`).
1365 self.publish_program_with_intent(
1366 branch,
1367 stages,
1368 diff,
1369 new_imports,
1370 activate,
1371 signer,
1372 None,
1373 &std::collections::BTreeMap::new(),
1374 )
1375 }
1376
1377 /// [`Self::publish_program_signed`] plus an optional `intent_id`
1378 /// (#131 / #839): when given, every op this publish emits is stamped
1379 /// with it, so the op log records *why* the change happened — the
1380 /// prompt / model / session an agent was acting under — not only
1381 /// what it was. `lex recall --intent <id>` and `lex op replay` read
1382 /// it back. The caller records the [`lex_vcs::Intent`] in the
1383 /// [`lex_vcs::IntentLog`] beforehand; this only links ops to it.
1384 /// `None` is the existing (intent-less) behavior, so op ids for
1385 /// intent-less publishes are unchanged.
1386 // A batch publish legitimately takes the branch, program, diff,
1387 // imports, activate flag, signer, and now the intent — bundling
1388 // them into a struct for one optional field would obscure more
1389 // than it clarifies.
1390 #[allow(clippy::too_many_arguments)]
1391 pub fn publish_program_with_intent(
1392 &self,
1393 branch: &str,
1394 stages: &[lex_ast::Stage],
1395 diff: &lex_vcs::DiffReport,
1396 new_imports: &lex_vcs::ImportMap,
1397 activate: bool,
1398 signer: Option<&lex_vcs::Keypair>,
1399 intent_id: Option<lex_vcs::IntentId>,
1400 // Mangling prefix → package source file, for a multi-module
1401 // package publish; empty for a single file. Recorded as each
1402 // `AddFunction`/`AddType`'s `in_file` so `export-git` can
1403 // de-flatten the package (#894).
1404 module_prefixes: &std::collections::BTreeMap<String, String>,
1405 ) -> Result<PublishOutcome, StoreError> {
1406 use std::collections::{BTreeMap, BTreeSet};
1407
1408 // #130's write-time gate: verify the candidate program
1409 // typechecks (and effects are correctly declared) before
1410 // any disk side-effect. If anything fails, return the
1411 // structured envelope and leave the branch head unchanged
1412 // — the store's "always-valid HEAD" invariant only holds
1413 // because this is the only batch-publish path that
1414 // advances heads. Single-op writes via the lower-level
1415 // `apply_operation` are not gated yet (#130 follow-up).
1416 // #930: resolve any external dependency edges the head keeps
1417 // (empty when no resolver is installed or the head is inlined).
1418 let modules = self.resolved_modules(stages, None);
1419 if let Err(errors) = lex_types::check_program_with_modules(stages, &modules) {
1420 return Err(StoreError::TypeError(errors));
1421 }
1422
1423 // Build old-side views from the current branch. There used to be
1424 // an `old_name_to_sig: BTreeMap<String, SigId>` built here too,
1425 // keyed by bare function name — but a bare name is not unique
1426 // across a package's files (#818: two files can legitimately
1427 // both declare a local `validate` helper with different
1428 // signatures), so a name-keyed map silently collapsed distinct
1429 // SigIds onto one. `diff` now carries each entry's own resolved
1430 // `old_sig_id` directly (see `diff_report`'s doc comments), so
1431 // `diff_to_ops` no longer needs this lookup at all.
1432 let old_head = self.branch_head(branch)?;
1433 // Read every live function's effects through the SigId the head
1434 // names, in one batch. Two reasons, both load-bearing:
1435 //
1436 // * Cost. This was a `get_ast` per live function, and
1437 // `get_ast`'s index-hit path re-reads and re-parses the whole
1438 // `stage_index.jsonl` on every call — O(index × live fns) per
1439 // publish, paid again for every `publish_program` call a
1440 // multi-file publish makes (#828; measured 34s for a no-op
1441 // republish of a real 21-file package against only 698 live
1442 // functions, nearly all of it here).
1443 // * Correctness. A StageId is name-independent, so two live
1444 // functions differing only in name share one and the index
1445 // maps it to a single sig — resolving by StageId therefore
1446 // attributed one function's effects to the *other* one's sig,
1447 // the same ambiguity #826 fixed in `pkg_publish_handler`.
1448 let head_pairs: Vec<(String, String)> = old_head
1449 .iter()
1450 .map(|(sig, stage)| (sig.clone(), stage.clone()))
1451 .collect();
1452 let old_effects: BTreeMap<String, BTreeSet<String>> = head_pairs
1453 .iter()
1454 .zip(self.get_asts_for_sigs_bulk(&head_pairs))
1455 .filter_map(|((sig, _), ast)| match ast.ok()? {
1456 lex_ast::Stage::FnDecl(fd) => {
1457 let s: BTreeSet<String> =
1458 fd.effects.iter().map(|e| e.name.clone()).collect();
1459 Some((sig.clone(), s))
1460 }
1461 _ => None,
1462 })
1463 .collect();
1464 let old_imports = self.derive_imports_from_oplog(branch)?;
1465
1466 let op_kinds = lex_vcs::diff_to_ops(lex_vcs::DiffInputs {
1467 old_head: &old_head,
1468 old_effects: &old_effects,
1469 old_imports: &old_imports,
1470 new_stages: stages,
1471 new_imports,
1472 diff,
1473 module_prefixes,
1474 })
1475 .map_err(|e| StoreError::InvalidTransition(format!("diff_to_ops: {e}")))?;
1476
1477 let mut ops_out: Vec<PublishOp> = Vec::new();
1478 let mut last_op_id: Option<lex_vcs::OpId> = None;
1479 for kind in op_kinds {
1480 // Persist the underlying stage AST/metadata if this op
1481 // produces or replaces one.
1482 if let Some(stg) = stage_for_kind(&kind, stages) {
1483 if !matches!(stg, lex_ast::Stage::Import(_)) {
1484 self.publish_signed(stg, signer)?;
1485 if activate {
1486 if let Some(stage_id_str) = stage_id(stg) {
1487 let _ = self.activate(&stage_id_str);
1488 }
1489 }
1490 }
1491 }
1492 let transition = transition_for_kind(&kind);
1493 let attestable = attestable_stage_ids(&transition);
1494 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
1495 let op =
1496 lex_vcs::Operation::new(kind.clone(), head_now.into_iter().collect::<Vec<_>>());
1497 // #131 / #839: stamp the caller's intent so the op log records
1498 // why this change happened, not just what it was. The CAS
1499 // retry path preserves `intent_id` when it rebuilds the op.
1500 let op = match &intent_id {
1501 Some(id) => op.with_intent(id.clone()),
1502 None => op,
1503 };
1504 let op_id = self.apply_operation(branch, op, transition)?;
1505 self.record_typecheck_passed(&attestable, &op_id)?;
1506 ops_out.push(PublishOp {
1507 op_id: op_id.clone(),
1508 kind: serde_json::to_value(&kind).map_err(StoreError::Serde)?,
1509 });
1510 last_op_id = Some(op_id);
1511 }
1512
1513 let head_op = match last_op_id {
1514 Some(id) => Some(id),
1515 // No ops applied; return whatever the head was already.
1516 None => self.get_branch(branch)?.and_then(|b| b.head_op),
1517 };
1518
1519 Ok(PublishOutcome {
1520 ops: ops_out,
1521 head_op,
1522 })
1523 }
1524
1525 pub fn derive_imports_from_oplog(
1526 &self,
1527 branch: &str,
1528 ) -> Result<lex_vcs::ImportMap, StoreError> {
1529 use lex_vcs::OperationKind::*;
1530 let log = lex_vcs::OpLog::open(self.root())?;
1531 let head = match self.get_branch(branch)?.and_then(|b| b.head_op) {
1532 Some(h) => h,
1533 None => return Ok(Default::default()),
1534 };
1535 let mut out: lex_vcs::ImportMap = Default::default();
1536 for r in log.walk_forward(&head, None)? {
1537 match r.op.kind {
1538 AddImport { in_file, module, alias } => {
1539 // The op omits the alias when it's the module's
1540 // default (last path segment) to keep its OpId
1541 // stable; rebuild it the same way on the way out.
1542 let alias =
1543 alias.unwrap_or_else(|| lex_vcs::default_import_alias(&module));
1544 out.entry(in_file)
1545 .or_default()
1546 .insert(lex_vcs::ImportRef { reference: module, alias });
1547 }
1548 RemoveImport { in_file, module } => {
1549 // Removal is keyed by reference (the op carries no
1550 // alias), so drop any binding of that module.
1551 if let Some(set) = out.get_mut(&in_file) {
1552 set.retain(|ir| ir.reference != module);
1553 }
1554 }
1555 _ => {}
1556 }
1557 }
1558 Ok(out)
1559 }
1560
1561 /// Apply an operation to a branch and advance its head_op.
1562 ///
1563 /// The single advance path. Validates parents via `lex_vcs::apply`,
1564 /// persists the operation via the op log, then atomically advances
1565 /// the branch file's head_op via `set_branch_head_op`.
1566 ///
1567 /// Errors:
1568 /// - `UnknownBranch`: branch does not exist (no op is persisted).
1569 /// - `Apply(ApplyError::StaleParent)`: the op's parents don't
1570 /// match the branch head — head is unchanged. Callers that
1571 /// want retry-on-stale (e.g. `lex publish` re-running against
1572 /// a moved head) match on this variant explicitly.
1573 /// - `Apply(ApplyError::UnknownMergeParent)`: a merge op's
1574 /// second parent isn't in the log.
1575 /// - `Io`: filesystem error during persist or branch advance.
1576 ///
1577 /// Crash recovery: between op persist and branch advance, a crash
1578 /// can leave an orphan op record in the log with no branch
1579 /// pointing at it. The op is content-addressed and cheap to
1580 /// re-derive from the same source. See
1581 /// Apply a single op against `branch`, gated on the candidate
1582 /// program typechecking. The per-op variant of #130's
1583 /// write-time gate — counterpart to [`Self::publish_program`]'s
1584 /// batch-mode check.
1585 ///
1586 /// `candidate` is the sequence of `Stage`s that *would* exist
1587 /// on this branch after the op is applied. Caller's
1588 /// responsibility: today neither `lex-store` nor `lex-vcs`
1589 /// reconstruct the candidate from the op + branch state on
1590 /// behalf of the caller. The natural callers (HTTP `POST
1591 /// /v1/publish` for a single op; agent harnesses driving
1592 /// merges via the future #134 API) already have the candidate
1593 /// in memory.
1594 ///
1595 /// On rejection: branch head unchanged, no op record persisted.
1596 /// Same atomicity guarantee as the publish path.
1597 ///
1598 /// # Why a separate method, not a flag on `apply_operation`
1599 ///
1600 /// `apply_operation` accepting `Option<&[Stage]>` and silently
1601 /// skipping the gate on `None` is exactly the kind of
1602 /// "secretly opt-out" path #130 is trying to remove. The honest
1603 /// split: `apply_operation` for the one caller that already
1604 /// typechecked its input up front (`publish_program`),
1605 /// `apply_operation_checked` for callers holding the candidate,
1606 /// [`Self::apply_operation_gated`] for single-parent callers
1607 /// that hold only the transition (`/v1/patch`), and
1608 /// [`Self::apply_merge_op_gated`] for merge commits (#833).
1609 pub fn apply_operation_checked(
1610 &self,
1611 branch: &str,
1612 op: lex_vcs::Operation,
1613 transition: lex_vcs::StageTransition,
1614 candidate: &[lex_ast::Stage],
1615 ) -> Result<lex_vcs::OpId, StoreError> {
1616 let modules = self.resolved_modules(candidate, None); // #930
1617 if let Err(errors) = lex_types::check_program_with_modules(candidate, &modules) {
1618 // #281: emit a `RepairHint` attestation against each
1619 // candidate stage the transition was about to produce.
1620 // The op record itself isn't persisted (the gate is
1621 // pre-persistence), but the candidate stage IS — the
1622 // transform-flow methods publish before this call.
1623 // The attached hint lets `lex repair <op_id>` and
1624 // future LLM-assisted apply paths read the structured
1625 // errors without re-running the typecheck.
1626 let attestable = attestable_stage_ids(&transition);
1627 let failed_op_id = op.op_id();
1628 let _ = self.record_repair_hint(&attestable, &failed_op_id, &errors);
1629 return Err(StoreError::TypeError(errors));
1630 }
1631 // #292 slice 3: per-session budget gate. After typecheck
1632 // passes, refuse the op if it would push its session's
1633 // monotonic spend over the configured cap. Sessions
1634 // without an intent_id, or with an intent whose session
1635 // has no cap configured, sail through.
1636 self.check_session_budget(&op)?;
1637 let attestable = attestable_stage_ids(&transition);
1638 let op_effects = op_declared_effects(&op.kind);
1639 // #262: CAS retry loop. Single-parent ops can be safely
1640 // re-persisted under a new parent on contention (the kind
1641 // is invariant; only `parents` changes). Merge ops (already
1642 // 2-parent) come through the merge engine which has its own
1643 // coordination; we don't retry them here — we'll see the
1644 // first attempt's CAS fail and surface Contention.
1645 self.cas_retry_advance(branch, op, transition, |new_head| {
1646 self.record_typecheck_passed(&attestable, &new_head.op_id)?;
1647 self.run_required_attestations_gate(branch, &new_head.op_id, &attestable, &op_effects)
1648 })
1649 }
1650
1651 /// The program that would exist on `branch` after `transition`
1652 /// is applied: the branch head (snapshot-cached) with the
1653 /// transition replayed over it, every resulting `(sig, stage)`
1654 /// bulk-loaded. Exact for a **single-parent** transition — the
1655 /// candidate [`Self::apply_operation_gated`] wants. Not valid for
1656 /// a merge: a `StageTransition::Merge` records only the delta
1657 /// relative to dst, while the op-DAG replay that computes a
1658 /// merge's real head walks both parents (#833).
1659 pub fn candidate_program_for(
1660 &self,
1661 branch: &str,
1662 transition: &lex_vcs::StageTransition,
1663 ) -> Result<Vec<Stage>, StoreError> {
1664 let mut head = self.branch_head(branch)?;
1665 crate::branches::apply_transition(&mut head, transition);
1666 let pairs: Vec<(String, String)> = head.into_iter().collect();
1667 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect()
1668 }
1669
1670 /// [`Self::apply_operation_checked`] for a **single-parent** op
1671 /// where the caller holds only the transition: assembles the
1672 /// candidate via [`Self::candidate_program_for`] and runs the
1673 /// gate. Same rejection semantics — `TypeError`, a `RepairHint`
1674 /// attestation, head unchanged, nothing persisted. This is the
1675 /// write path for `/v1/patch` (#833). Merge ops must not use it
1676 /// (see `candidate_program_for`); they go through
1677 /// [`Self::apply_merge_op_gated`].
1678 pub fn apply_operation_gated(
1679 &self,
1680 branch: &str,
1681 op: lex_vcs::Operation,
1682 transition: lex_vcs::StageTransition,
1683 ) -> Result<lex_vcs::OpId, StoreError> {
1684 debug_assert!(
1685 op.parents.len() <= 1,
1686 "apply_operation_gated is single-parent only; merges use apply_merge_op_gated"
1687 );
1688 let candidate = self.candidate_program_for(branch, &transition)?;
1689 self.apply_operation_checked(branch, op, transition, &candidate)
1690 }
1691
1692 /// The gated write path for **merge** commits (`commit_merge`,
1693 /// `POST /v1/merge/<id>/commit`, `lex merge commit`).
1694 ///
1695 /// A `StageTransition::Merge` records only the delta relative to
1696 /// dst; the sig->stage map every consumer reads is recomputed by
1697 /// replaying the op DAG, which for a merge walks *both* parents
1698 /// and can surface sigs the delta never mentions. So the only way
1699 /// to know the true post-merge program is to replay it — land the
1700 /// op and read `branch_head`. This lands the merge op,
1701 /// type-checks the resulting head, and on a failure rolls the
1702 /// head back and returns `TypeError`.
1703 ///
1704 /// Before #833 the merge paths landed through the ungated
1705 /// `apply_operation`, so a merge whose result didn't compose
1706 /// (e.g. dst still calls `helper`, an agent-supplied resolution
1707 /// dropped it) advanced the head with nothing to catch it.
1708 ///
1709 /// Rollback leaves the rejected merge op as an unreachable record
1710 /// (reclaimed by `lex op gc`, the same orphan crash-recovery
1711 /// already tolerates). A stage the merge names that was never
1712 /// published surfaces as the underlying `StoreError` from the
1713 /// bulk read — the "never advance onto content that can't be
1714 /// loaded" invariant from the other side.
1715 pub fn apply_merge_op_gated(
1716 &self,
1717 branch: &str,
1718 op: lex_vcs::Operation,
1719 transition: lex_vcs::StageTransition,
1720 ) -> Result<lex_vcs::OpId, StoreError> {
1721 let head_before = self.get_branch(branch)?.and_then(|b| b.head_op);
1722 // Capture the stages this merge introduces before `transition`
1723 // is moved into `apply_operation`; used for the TypeCheck
1724 // attestation below.
1725 let attestable = attestable_stage_ids(&transition);
1726 let op_id = self.apply_operation(branch, op, transition)?;
1727
1728 let verdict = (|| -> Result<(), StoreError> {
1729 let head = self.branch_head(branch)?;
1730 let pairs: Vec<(String, String)> = head.into_iter().collect();
1731 let stages: Vec<Stage> =
1732 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
1733 // #930: per-head dep resolution for the merge path is a follow-up;
1734 // None lets a client resolver use its working-copy lock.
1735 let modules = self.resolved_modules(&stages, None);
1736 if let Err(errors) = lex_types::check_program_with_modules(&stages, &modules) {
1737 return Err(StoreError::TypeError(errors));
1738 }
1739 Ok(())
1740 })();
1741
1742 if let Err(e) = verdict {
1743 // Roll the head back. The empty-dst case never reaches
1744 // here (it fast-forwards without a merge op), so
1745 // `head_before` is always `Some` on this arm.
1746 if let Some(prev) = head_before {
1747 self.set_branch_head_op(branch, prev)?;
1748 }
1749 return Err(e);
1750 }
1751 // #835: the merge's post-merge head type-checked, but until now
1752 // that verdict left no trace in the attestation log — so a
1753 // merged stage looked un-type-checked to `lex blame
1754 // --with-evidence` and the attestation queries, unlike a
1755 // published or patched stage. Emit `TypeCheck::Passed` for the
1756 // stages the merge introduced, mirroring the publish / patch
1757 // paths (`record_typecheck_passed`). Emitted only after the
1758 // check passes and the head is committed, so a rolled-back
1759 // merge records nothing.
1760 self.record_typecheck_passed(&attestable, &op_id)?;
1761 Ok(op_id)
1762 }
1763
1764 /// Type-check the program that would result from overlaying a merge
1765 /// `delta` onto `branch`'s current head — **without moving the
1766 /// head** (#834). `delta` maps `sig_id -> Some(stage)` to set that
1767 /// sig to `stage`, or `sig_id -> None` to remove it, exactly the
1768 /// `entries` a `StageTransition::Merge` records.
1769 ///
1770 /// This is the read-only, resolve-time counterpart of
1771 /// `apply_merge_op_gated`'s commit-time gate: it lets a merge
1772 /// session tell an agent *which resolution broke type-checking* the
1773 /// moment it is submitted, instead of only after a failed commit.
1774 /// `Ok(())` means the projected program composes; a type failure is
1775 /// `Err(StoreError::TypeError(..))`; a read failure is the
1776 /// corresponding `StoreError` I/O variant.
1777 pub fn typecheck_merge_projection(
1778 &self,
1779 branch: &str,
1780 delta: &std::collections::BTreeMap<String, Option<String>>,
1781 ) -> Result<(), StoreError> {
1782 let mut head = self.branch_head(branch)?;
1783 for (sig, stage) in delta {
1784 match stage {
1785 Some(s) => { head.insert(sig.clone(), s.clone()); }
1786 None => { head.remove(sig); }
1787 }
1788 }
1789 let pairs: Vec<(String, String)> = head.into_iter().collect();
1790 let stages: Vec<Stage> =
1791 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
1792 let modules = self.resolved_modules(&stages, None); // #930 (patch path)
1793 if let Err(errors) = lex_types::check_program_with_modules(&stages, &modules) {
1794 return Err(StoreError::TypeError(errors));
1795 }
1796 Ok(())
1797 }
1798
1799 /// #838: attempt a typed three-way merge of a single sig's body for
1800 /// a `ModifyModify` conflict — the intra-function, better-than-git
1801 /// case where two agents edited *disjoint* subtrees of the same
1802 /// function (different match arms, different let bindings).
1803 ///
1804 /// `base` / `ours` (the dst side) / `theirs` (the src side) are the
1805 /// three stage ids the merge engine surfaced for `sig_id`. Loads
1806 /// the three `FnDecl`s, structurally merges the bodies
1807 /// ([`lex_vcs::merge_bodies`]), and accepts the result *only if* the
1808 /// merged function also type-checks against `dst_branch`'s head — a
1809 /// body that composes syntactically but not by type is still a
1810 /// conflict (#838). On success the merged stage is published
1811 /// (content-addressed, idempotent; orphaned and GC-reclaimable if
1812 /// the merge is never committed) and its id returned; `None` means
1813 /// "fall back to a whole-function conflict."
1814 ///
1815 /// Deliberately narrow for this slice: only pure body divergence is
1816 /// merged. If the two sides disagree on anything but the body
1817 /// (examples, type params — the signature is identical by
1818 /// construction, since all three share `sig_id`), or either stage
1819 /// isn't a function, it falls back to a conflict.
1820 pub fn try_semantic_body_merge(
1821 &self,
1822 dst_branch: &str,
1823 sig_id: &str,
1824 base: &str,
1825 ours: &str,
1826 theirs: &str,
1827 ) -> Result<Option<String>, StoreError> {
1828 use lex_ast::Stage::FnDecl;
1829 let (base_fd, ours_fd, theirs_fd) =
1830 match (self.get_ast(base), self.get_ast(ours), self.get_ast(theirs)) {
1831 (Ok(FnDecl(b)), Ok(FnDecl(o)), Ok(FnDecl(t))) => (b, o, t),
1832 // A non-function stage (type decl / import) or a stage
1833 // that can't be loaded isn't an intra-body merge.
1834 _ => return Ok(None),
1835 };
1836
1837 // Only the body may diverge between the two sides.
1838 if !fndecl_same_except_body(&ours_fd, &theirs_fd) {
1839 return Ok(None);
1840 }
1841
1842 let merged_body =
1843 match lex_vcs::merge_bodies(&base_fd.body, &ours_fd.body, &theirs_fd.body) {
1844 lex_vcs::BodyMerge::Merged(b) => b,
1845 lex_vcs::BodyMerge::Conflict => return Ok(None),
1846 };
1847
1848 let mut merged_fd = ours_fd.clone();
1849 merged_fd.body = merged_body;
1850 let merged_stage = lex_ast::Stage::FnDecl(merged_fd);
1851 let new_stage_id = match stage_id(&merged_stage) {
1852 Some(id) => id,
1853 None => return Ok(None),
1854 };
1855
1856 // Type-check the merged fn in context: dst's head with this sig
1857 // swapped to the merged stage. Requires the merged stage to be
1858 // loadable, so publish first (idempotent, content-addressed).
1859 self.publish(&merged_stage)?;
1860 let mut delta = std::collections::BTreeMap::new();
1861 delta.insert(sig_id.to_string(), Some(new_stage_id.clone()));
1862 match self.typecheck_merge_projection(dst_branch, &delta) {
1863 Ok(()) => Ok(Some(new_stage_id)),
1864 // Composes syntactically, not by type → still a conflict.
1865 Err(StoreError::TypeError(_)) => Ok(None),
1866 Err(e) => Err(e),
1867 }
1868 }
1869
1870 /// #836 G3: assemble everything a regenerator needs to *replay* an
1871 /// op — re-derive the change from its recorded cause. Returns the
1872 /// op's recorded intent (prompt / model / session), the target sig
1873 /// and the stage id it produced, and the program the change was
1874 /// made against (the parent state, rendered to source). An external
1875 /// harness feeds the prompt + parent program to the recorded model,
1876 /// then hands the regenerated stage back to [`Self::replay_compare`]
1877 /// (lex owns the deterministic comparison; the model call is the
1878 /// harness's, matching the rest of the architecture).
1879 ///
1880 /// Errors with `UnknownOp` if the op_id is unknown, or
1881 /// `InvalidTransition` if the op didn't produce a stage (a removal /
1882 /// import / merge has nothing to regenerate).
1883 pub fn replay_request(&self, op_id: &str) -> Result<ReplayRequest, StoreError> {
1884 let log = lex_vcs::OpLog::open(self.root())?;
1885 let record = log
1886 .get(&op_id.to_string())?
1887 .ok_or_else(|| StoreError::UnknownOp(op_id.to_string()))?;
1888 let (target_sig, expected_stage_id) = produced_sig_stage(&record.produces)
1889 .ok_or_else(|| StoreError::InvalidTransition(format!("op {op_id} produced no stage to replay")))?;
1890
1891 let (prompt, model, session_id) = match &record.op.intent_id {
1892 Some(id) => {
1893 let intents = lex_vcs::IntentLog::open(self.root())?;
1894 match intents.get(id)? {
1895 Some(i) => (Some(i.prompt), Some(model_label(&i.model)), Some(i.session_id)),
1896 None => (None, None, None),
1897 }
1898 }
1899 None => (None, None, None),
1900 };
1901
1902 // The program the op was applied against: the head state at its
1903 // (first) parent, rendered with the canonical printer. A root
1904 // op has no parent → empty program.
1905 let parent_program = match record.op.parents.first() {
1906 Some(parent) => self.program_source_at_op(parent)?,
1907 None => String::new(),
1908 };
1909
1910 // The target function's name + signature, from the recorded
1911 // stage — a regenerator needs the interface, not just the hash.
1912 let (target_name, target_signature) = match self.get_ast(&expected_stage_id) {
1913 Ok(lex_ast::Stage::FnDecl(fd)) => {
1914 (Some(fd.name.clone()), Some(lex_vcs::render_signature(&fd)))
1915 }
1916 _ => (None, None),
1917 };
1918
1919 Ok(ReplayRequest {
1920 op_id: op_id.to_string(),
1921 target_sig,
1922 target_name,
1923 target_signature,
1924 expected_stage_id,
1925 prompt,
1926 model,
1927 session_id,
1928 parent_program,
1929 })
1930 }
1931
1932 /// #836 G3: compare a regenerated `candidate` against what the op
1933 /// recorded producing, and emit the `Replay` attestation. The
1934 /// reproducibility claim made concrete — a faithful regeneration of
1935 /// the same function from the same cause yields the same
1936 /// content-addressed stage id.
1937 ///
1938 /// `reproduced` is true iff the candidate is the same sig *and* the
1939 /// same stage id the op recorded. A candidate for a different sig
1940 /// counts as "not reproduced" (`produced_stage_id: None`) rather
1941 /// than an error — it's a legitimate, if negative, replay result.
1942 /// The attestation is addressed to the op's recorded stage, so
1943 /// `list_for_stage` surfaces it alongside the TypeCheck/Examples
1944 /// evidence.
1945 pub fn replay_compare(
1946 &self,
1947 op_id: &str,
1948 candidate: &Stage,
1949 ) -> Result<ReplayOutcome, StoreError> {
1950 let (target_sig, expected_stage_id) = self.replay_target(op_id)?;
1951 let cand_sig = lex_ast::sig_id(candidate);
1952 let cand_stage = stage_id(candidate);
1953 let produced_stage_id = match (cand_sig.as_deref(), &cand_stage) {
1954 // Same function regenerated: the produced stage is
1955 // whatever it content-addresses to.
1956 (Some(s), Some(st)) if s == target_sig => Some(st.clone()),
1957 // A different sig (or an unhashable stage) isn't a
1958 // regeneration of this op's change.
1959 _ => None,
1960 };
1961 let reproduced = produced_stage_id.as_deref() == Some(expected_stage_id.as_str());
1962 let detail = if reproduced {
1963 None
1964 } else {
1965 Some("regeneration did not reproduce the recorded stage".to_string())
1966 };
1967 self.emit_replay(op_id, &expected_stage_id, produced_stage_id, reproduced, None, detail)
1968 }
1969
1970 /// Record a *negative* replay result for a regeneration that never
1971 /// yielded a comparable stage — the output didn't parse, or didn't
1972 /// define the target sig (#836 G3). Emits a `Replay { reproduced:
1973 /// false, produced_stage_id: None }` attestation with `reason` in
1974 /// its `Failed` detail, so an automated `lex op replay` run always
1975 /// records a verdict rather than aborting. `reason` is caller-supplied
1976 /// (e.g. "regenerated source did not parse").
1977 pub fn replay_record_miss(&self, op_id: &str, reason: &str) -> Result<ReplayOutcome, StoreError> {
1978 let (_target_sig, expected_stage_id) = self.replay_target(op_id)?;
1979 self.emit_replay(op_id, &expected_stage_id, None, false, None, Some(reason.to_string()))
1980 }
1981
1982 /// `(target_sig, expected_stage_id)` for a replayable op, or an
1983 /// error if the op is unknown or produced no stage.
1984 fn replay_target(&self, op_id: &str) -> Result<(String, String), StoreError> {
1985 let log = lex_vcs::OpLog::open(self.root())?;
1986 let record = log
1987 .get(&op_id.to_string())?
1988 .ok_or_else(|| StoreError::UnknownOp(op_id.to_string()))?;
1989 produced_sig_stage(&record.produces)
1990 .ok_or_else(|| StoreError::InvalidTransition(format!("op {op_id} produced no stage to replay")))
1991 }
1992
1993 /// Record a replay verdict the caller has already decided — used by
1994 /// the CLI's behavioral tier, which does the (VM-backed) equivalence
1995 /// check the store deliberately can't. `expected_stage_id` is looked
1996 /// up from the op. Set `behavioral_samples` to `Some(n)` when the
1997 /// candidate reproduced *behaviorally* over `n` sampled inputs rather
1998 /// than by exact stage-id match; the attestation then records that
1999 /// weaker-but-real claim distinctly.
2000 pub fn replay_record(
2001 &self,
2002 op_id: &str,
2003 produced_stage_id: Option<String>,
2004 reproduced: bool,
2005 behavioral_samples: Option<usize>,
2006 fail_detail: Option<String>,
2007 ) -> Result<ReplayOutcome, StoreError> {
2008 let (_target_sig, expected_stage_id) = self.replay_target(op_id)?;
2009 self.emit_replay(op_id, &expected_stage_id, produced_stage_id, reproduced, behavioral_samples, fail_detail)
2010 }
2011
2012 /// Compute the exact-match verdict for a candidate *without* emitting
2013 /// an attestation — `(expected_stage_id, produced_stage_id, exact)`.
2014 /// Lets a caller (the CLI) fall back to a behavioral check on a valid
2015 /// but non-identical candidate and emit a single verdict, instead of
2016 /// [`Self::replay_compare`]'s emit-immediately shape.
2017 pub fn replay_stage_of(
2018 &self,
2019 op_id: &str,
2020 candidate: &Stage,
2021 ) -> Result<(String, Option<String>, bool), StoreError> {
2022 let (target_sig, expected_stage_id) = self.replay_target(op_id)?;
2023 let cand_sig = lex_ast::sig_id(candidate);
2024 let cand_stage = stage_id(candidate);
2025 let produced_stage_id = match (cand_sig.as_deref(), &cand_stage) {
2026 (Some(s), Some(st)) if s == target_sig => Some(st.clone()),
2027 _ => None,
2028 };
2029 let exact = produced_stage_id.as_deref() == Some(expected_stage_id.as_str());
2030 Ok((expected_stage_id, produced_stage_id, exact))
2031 }
2032
2033 /// Emit the `Replay` attestation and build the outcome. Shared by
2034 /// [`Self::replay_compare`], [`Self::replay_record_miss`], and
2035 /// [`Self::replay_record`].
2036 fn emit_replay(
2037 &self,
2038 op_id: &str,
2039 expected_stage_id: &str,
2040 produced_stage_id: Option<String>,
2041 reproduced: bool,
2042 behavioral_samples: Option<usize>,
2043 fail_detail: Option<String>,
2044 ) -> Result<ReplayOutcome, StoreError> {
2045 let model = {
2046 let log = lex_vcs::OpLog::open(self.root())?;
2047 match log.get(&op_id.to_string())?.and_then(|r| r.op.intent_id) {
2048 Some(id) => lex_vcs::IntentLog::open(self.root())?
2049 .get(&id)?
2050 .map(|i| model_label(&i.model)),
2051 None => None,
2052 }
2053 };
2054 let result = if reproduced {
2055 lex_vcs::AttestationResult::Passed
2056 } else {
2057 lex_vcs::AttestationResult::Failed {
2058 detail: fail_detail.unwrap_or_else(|| "not reproduced".into()),
2059 }
2060 };
2061 let attestation = lex_vcs::Attestation::new(
2062 expected_stage_id.to_string(),
2063 Some(op_id.to_string()),
2064 None,
2065 lex_vcs::AttestationKind::Replay {
2066 expected_stage_id: expected_stage_id.to_string(),
2067 produced_stage_id: produced_stage_id.clone(),
2068 reproduced,
2069 behavioral_samples,
2070 model,
2071 },
2072 result,
2073 replay_producer(),
2074 None,
2075 );
2076 let attestation_id = attestation.attestation_id.clone();
2077 self.attestation_log()?.put(&attestation)?;
2078 Ok(ReplayOutcome {
2079 op_id: op_id.to_string(),
2080 expected_stage_id: expected_stage_id.to_string(),
2081 produced_stage_id,
2082 reproduced,
2083 behavioral_samples,
2084 attestation_id,
2085 })
2086 }
2087
2088 /// The program at an op (that op and all its ancestors applied), as
2089 /// canonical stages. The behavioral replay tier needs the whole
2090 /// program — a regenerated function may call helpers from its parent
2091 /// state, so it can only be run in context. Exposed for the CLI's
2092 /// equivalence check; `op_id` may be any op in the log.
2093 pub fn program_stages_at_op(&self, op_id: &str) -> Result<Vec<Stage>, StoreError> {
2094 let oid: lex_vcs::OpId = op_id.to_string();
2095 let log = lex_vcs::OpLog::open(self.root())?;
2096 let mut map: std::collections::BTreeMap<String, String> = std::collections::BTreeMap::new();
2097 for rec in log.walk_forward(&oid, None)? {
2098 crate::branches::apply_transition(&mut map, &rec.produces);
2099 }
2100 let pairs: Vec<(String, String)> = map.into_iter().collect();
2101 let stages: Vec<Stage> =
2102 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
2103 Ok(stages)
2104 }
2105
2106 /// The program at an op, rendered to source. Used to give a replay
2107 /// regenerator the context the change was made against.
2108 fn program_source_at_op(&self, op_id: &lex_vcs::OpId) -> Result<String, StoreError> {
2109 Ok(lex_ast::print_stages(&self.program_stages_at_op(op_id)?))
2110 }
2111
2112 /// Open the attestation log rooted at this store. The log lives
2113 /// under `<root>/attestations/`; opening is idempotent and cheap
2114 /// (`fs::create_dir_all`). Exposed publicly so consumers — `lex
2115 /// blame --with-evidence`, `GET /v1/stage/<id>/attestations` —
2116 /// can read what the store gate emitted without round-tripping
2117 /// through this crate's API surface.
2118 /// Recompute a producer's trust score from its recent
2119 /// attestation history and emit a fresh `ProducerTrust`
2120 /// attestation (#293). Score = `passed / (passed + failed
2121 /// + inconclusive)` over the last `window` attestations
2122 /// produced by `tool_id`, expressed in thousandths
2123 /// (`0..=1000`).
2124 ///
2125 /// Refuses to grant trust when the tool has an active
2126 /// `ProducerBlock` — the block wins as a hard veto. Returns
2127 /// `Ok(None)` for "no attestations to score" (a brand-new
2128 /// producer); the caller can choose how to handle it
2129 /// (typically: skip the publish until evidence accrues).
2130 ///
2131 /// `granted_by` is the identity of the actor running the
2132 /// recompute (typically the human admin, or "lex-ci-bot"
2133 /// for an automated nightly).
2134 pub fn recompute_producer_trust(
2135 &self,
2136 tool_id: &str,
2137 window: usize,
2138 granted_by: &str,
2139 ) -> Result<Option<lex_vcs::AttestationId>, StoreError> {
2140 let log = self.attestation_log()?;
2141 let all = log.list_all()?;
2142 // Hard veto: don't grant trust to a blocked tool.
2143 if lex_vcs::active_producer_block(&all, tool_id).is_some() {
2144 return Err(StoreError::InvalidTransition(format!(
2145 "cannot recompute trust for `{tool_id}` — \
2146 producer is currently blocked"
2147 )));
2148 }
2149 // Filter to attestations from this tool, newest-first by
2150 // timestamp, then take the window.
2151 let mut from_tool: Vec<&lex_vcs::Attestation> = all
2152 .iter()
2153 .filter(|a| a.produced_by.tool == tool_id)
2154 // Ignore self-referential trust attestations (we're
2155 // scoring evidence, not previous trust statements).
2156 .filter(|a| {
2157 !matches!(
2158 a.kind,
2159 lex_vcs::AttestationKind::ProducerTrust { .. }
2160 | lex_vcs::AttestationKind::TrustWaived { .. }
2161 )
2162 })
2163 .collect();
2164 from_tool.sort_by_key(|a| std::cmp::Reverse(a.timestamp));
2165 from_tool.truncate(window);
2166 if from_tool.is_empty() {
2167 return Ok(None);
2168 }
2169 let (mut passed, mut total) = (0u64, 0u64);
2170 for a in &from_tool {
2171 total += 1;
2172 if matches!(a.result, lex_vcs::AttestationResult::Passed) {
2173 passed += 1;
2174 }
2175 }
2176 let score = if total == 0 {
2177 0
2178 } else {
2179 let raw = (passed as f64) * 1000.0 / (total as f64);
2180 raw.round().clamp(0.0, 1000.0) as u32
2181 };
2182 let head_op = self
2183 .list_branches()?
2184 .into_iter()
2185 .find_map(|b| self.get_branch(&b).ok().flatten().and_then(|x| x.head_op))
2186 .unwrap_or_else(|| "fresh".into());
2187 let evidence = format!(
2188 "window={window}, sample={}, head_op={head_op:.16}",
2189 from_tool.len()
2190 );
2191 let attestation = lex_vcs::Attestation::new(
2192 tool_id.to_string(),
2193 None,
2194 None,
2195 lex_vcs::AttestationKind::ProducerTrust {
2196 tool_id: tool_id.into(),
2197 score_thousandths: score,
2198 evidence,
2199 granted_by: granted_by.into(),
2200 },
2201 lex_vcs::AttestationResult::Passed,
2202 producer_trust_producer(),
2203 None,
2204 );
2205 let id = attestation.attestation_id.clone();
2206 log.put(&attestation)?;
2207 Ok(Some(id))
2208 }
2209
2210 /// The latest live `ProducerTrust` score (thousandths, `0..=1000`) for
2211 /// every producer that currently has trust: the newest score per tool by
2212 /// timestamp, excluding any tool under an active `ProducerBlock` (a block
2213 /// is a hard veto over trust, matching `recompute_producer_trust`).
2214 ///
2215 /// Used to export a capsule trusted-keys keyring from *earned* trust — the
2216 /// producer id doubles as the publisher's signing key downstream, so this
2217 /// turns track record into the allowlist `capsule install` consumes.
2218 pub fn live_producer_trust_scores(
2219 &self,
2220 ) -> Result<std::collections::BTreeMap<String, u32>, StoreError> {
2221 let log = self.attestation_log()?;
2222 let all = log.list_all()?;
2223 // Newest score per tool.
2224 let mut latest: std::collections::BTreeMap<String, (u64, u32)> =
2225 std::collections::BTreeMap::new();
2226 for a in &all {
2227 if let lex_vcs::AttestationKind::ProducerTrust {
2228 tool_id,
2229 score_thousandths,
2230 ..
2231 } = &a.kind
2232 {
2233 let entry = latest.entry(tool_id.clone()).or_insert((0, 0));
2234 if a.timestamp >= entry.0 {
2235 *entry = (a.timestamp, *score_thousandths);
2236 }
2237 }
2238 }
2239 // Drop blocked producers; a block vetoes trust.
2240 let mut scores = std::collections::BTreeMap::new();
2241 for (tool, (_, score)) in latest {
2242 if lex_vcs::active_producer_block(&all, &tool).is_some() {
2243 continue;
2244 }
2245 scores.insert(tool, score);
2246 }
2247 Ok(scores)
2248 }
2249
2250 pub fn attestation_log(&self) -> Result<lex_vcs::AttestationLog, StoreError> {
2251 Ok(lex_vcs::AttestationLog::open(self.root())?)
2252 }
2253
2254 /// Emit one `TypeCheck::Passed` attestation per stage produced by
2255 /// a successful gated apply. Idempotent on `attestation_id` —
2256 /// re-running the same gate run dedups via content addressing.
2257 ///
2258 /// Failure modes: `io::Error` from the attestation log (disk
2259 /// full, perms). The op has already landed by the time this
2260 /// runs; an error here means the op is durable but the evidence
2261 /// is missing. We propagate so the caller sees the partial
2262 /// state rather than silently swallowing — re-attesting the
2263 /// same op against the same op_id is idempotent (content
2264 /// addressing) so a retry is safe once the underlying issue is
2265 /// fixed.
2266 fn record_typecheck_passed(
2267 &self,
2268 stage_ids: &[String],
2269 op_id: &lex_vcs::OpId,
2270 ) -> Result<(), StoreError> {
2271 if stage_ids.is_empty() {
2272 return Ok(());
2273 }
2274 let log = self.attestation_log()?;
2275 for stage_id in stage_ids {
2276 let attestation = lex_vcs::Attestation::new(
2277 stage_id.clone(),
2278 Some(op_id.clone()),
2279 None,
2280 lex_vcs::AttestationKind::TypeCheck,
2281 lex_vcs::AttestationResult::Passed,
2282 typecheck_producer(),
2283 None,
2284 );
2285 log.put(&attestation)?;
2286 }
2287 Ok(())
2288 }
2289
2290 /// The hosted CI runner (#93): independently re-run the write-time
2291 /// type-check gate on a branch head and record the verdict as a
2292 /// `lex-hub-ci`-produced `TypeCheck` attestation for the stages the
2293 /// advance introduced. Called after an `op push` fast-forwards the
2294 /// head, so `require-attestation type_check` gates are backed by a
2295 /// producer that actually verified the code server-side, not by
2296 /// whatever attestation a client chose to attach. Does NOT move or
2297 /// roll back the head — the client's own always-valid-HEAD gate is
2298 /// what refuses a bad publish; this produces the trusted verdict on
2299 /// top of an already-committed advance (so a client that bypassed
2300 /// its gate is caught by a `TypeCheck::Failed` from `lex-hub-ci`).
2301 ///
2302 /// `from_head` is the branch head *before* the advance; the ops
2303 /// between it and `to_head` are the ones whose stages get attested.
2304 /// Idempotent: attestations are content-addressed, so re-verifying
2305 /// the same head is a no-op.
2306 pub fn verify_head_and_attest(
2307 &self,
2308 branch: &str,
2309 from_head: Option<&str>,
2310 to_head: &str,
2311 ) -> Result<HubCiVerdict, StoreError> {
2312 // Reconstruct the program at the new head and re-check it.
2313 let head = self.branch_head(branch)?;
2314 let pairs: Vec<(String, String)> = head.into_iter().collect();
2315 let decls: Vec<Stage> =
2316 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
2317 let checked_stages = decls.len();
2318 // #930: the SigId→stage map holds only fn/type declarations — the
2319 // head's `import` edges are AddImport ops, absent here. Reconstruct
2320 // them so a non-inlined head's `<alias>.name` references bind: the
2321 // resolver scans these imports to resolve each dependency, and the
2322 // checker's Pass 1 binds the alias to the resolved module. (Without
2323 // this the alias is unbound and the head fails as `unknown_identifier`,
2324 // even with the dependency correctly resolved.)
2325 let head_imports = crate::render::package_head_at_op(self, to_head)
2326 .map(|ph| {
2327 ph.flat_imports
2328 .into_iter()
2329 .map(|(reference, alias)| {
2330 Stage::Import(lex_ast::Import { reference, alias })
2331 })
2332 .collect::<Vec<_>>()
2333 })
2334 .unwrap_or_default();
2335 let mut stages = head_imports;
2336 stages.extend(decls);
2337 // #930: the hub gate resolves this head's external dependencies from
2338 // the lock committed with `to_head` (via the installed cross-store
2339 // resolver); empty when none is installed or the head is inlined.
2340 let modules = self.resolved_modules(&stages, Some(to_head));
2341 let result = match lex_types::check_program_with_modules(&stages, &modules) {
2342 Ok(_) => lex_vcs::AttestationResult::Passed,
2343 Err(errors) => lex_vcs::AttestationResult::Failed {
2344 detail: serde_json::to_string(&errors).unwrap_or_else(|_| "type errors".into()),
2345 },
2346 };
2347 let passed = matches!(result, lex_vcs::AttestationResult::Passed);
2348
2349 // Stages introduced by THIS advance (from_head exclusive → to_head).
2350 let log = lex_vcs::OpLog::open(self.root())?;
2351 let to = to_head.to_string();
2352 let records = match from_head {
2353 Some(f) => log
2354 .walk_forward_since(&to, &f.to_string())?
2355 .unwrap_or_else(|| log.walk_forward(&to, None).unwrap_or_default()),
2356 None => log.walk_forward(&to, None)?,
2357 };
2358 let mut introduced: Vec<String> = Vec::new();
2359 for rec in &records {
2360 introduced.extend(attestable_stage_ids(&rec.produces));
2361 }
2362
2363 let alog = self.attestation_log()?;
2364 for sid in &introduced {
2365 let att = lex_vcs::Attestation::new(
2366 sid.clone(),
2367 Some(to_head.to_string()),
2368 None,
2369 lex_vcs::AttestationKind::TypeCheck,
2370 result.clone(),
2371 hub_ci_producer(),
2372 None,
2373 );
2374 alog.put(&att)?;
2375 }
2376
2377 let detail = match &result {
2378 lex_vcs::AttestationResult::Failed { detail } => Some(detail.clone()),
2379 _ => None,
2380 };
2381 Ok(HubCiVerdict { passed, checked_stages, attested_stages: introduced.len(), detail })
2382 }
2383
2384 /// Emit an `Examples::Passed` attestation for a published stage
2385 /// whose behavioral `examples {}` block was run and passed (#835,
2386 /// Tier 1). Mirrors [`Self::record_typecheck_passed`]. The
2387 /// behavioral run itself happens one layer up (lex-api / lex-cli)
2388 /// because it needs the bytecode compiler + VM, which this crate
2389 /// deliberately doesn't depend on; the store only records the
2390 /// verdict. `file_hash` uses the stage id — the stage fully
2391 /// determines its own examples.
2392 pub fn record_examples_passed(
2393 &self,
2394 stage_id: &str,
2395 op_id: &lex_vcs::OpId,
2396 count: usize,
2397 ) -> Result<(), StoreError> {
2398 let log = self.attestation_log()?;
2399 let attestation = lex_vcs::Attestation::new(
2400 stage_id.to_string(),
2401 Some(op_id.clone()),
2402 None,
2403 lex_vcs::AttestationKind::Examples { file_hash: stage_id.to_string(), count },
2404 lex_vcs::AttestationResult::Passed,
2405 examples_producer(),
2406 None,
2407 );
2408 log.put(&attestation)?;
2409 Ok(())
2410 }
2411
2412 /// Record a structured `Review` verdict on a stage (#836 G4).
2413 /// The verdict maps onto the attestation `result` so existing
2414 /// result-based tooling reads it: Approve->Passed,
2415 /// Reject->Failed, RequestChanges->Inconclusive.
2416 pub fn record_review(
2417 &self,
2418 stage_id: &str,
2419 op_id: Option<lex_vcs::OpId>,
2420 reviewer: &str,
2421 verdict: lex_vcs::ReviewVerdict,
2422 notes: Option<String>,
2423 ) -> Result<lex_vcs::AttestationId, StoreError> {
2424 let result = match verdict {
2425 lex_vcs::ReviewVerdict::Approve => lex_vcs::AttestationResult::Passed,
2426 lex_vcs::ReviewVerdict::Reject => lex_vcs::AttestationResult::Failed {
2427 detail: notes.clone().unwrap_or_else(|| "rejected".into()),
2428 },
2429 lex_vcs::ReviewVerdict::RequestChanges => lex_vcs::AttestationResult::Inconclusive {
2430 detail: notes.clone().unwrap_or_else(|| "changes requested".into()),
2431 },
2432 };
2433 let att = lex_vcs::Attestation::new(
2434 stage_id.to_string(),
2435 op_id,
2436 None,
2437 lex_vcs::AttestationKind::Review { reviewer: reviewer.to_string(), verdict, notes },
2438 result,
2439 review_producer(reviewer),
2440 None,
2441 );
2442 let id = att.attestation_id.clone();
2443 self.attestation_log()?.put(&att)?;
2444 Ok(id)
2445 }
2446
2447 /// The latest `Review` verdict recorded on a stage, if any
2448 /// (#836 G4). "Latest" is by attestation timestamp; ties keep the
2449 /// last one seen. Used by `promote_candidate` to honor a standing
2450 /// Reject.
2451 pub fn latest_review_verdict(
2452 &self,
2453 stage_id: &str,
2454 ) -> Result<Option<lex_vcs::ReviewVerdict>, StoreError> {
2455 let log = self.attestation_log()?;
2456 let mut latest: Option<(u64, lex_vcs::ReviewVerdict)> = None;
2457 for a in log.list_for_stage(&stage_id.to_string())? {
2458 if let lex_vcs::AttestationKind::Review { verdict, .. } = a.kind {
2459 if latest.as_ref().map(|(t, _)| a.timestamp >= *t).unwrap_or(true) {
2460 latest = Some((a.timestamp, verdict));
2461 }
2462 }
2463 }
2464 Ok(latest.map(|(_, v)| v))
2465 }
2466
2467 /// Consult `policy.session_budgets` for the op's session
2468 /// (resolved via `op.intent_id → Intent.session_id`) and
2469 /// refuse if applying would push the session's monotonic spend
2470 /// over the configured cap (#292 slice 3).
2471 ///
2472 /// Ops without an `intent_id`, or whose intent has no
2473 /// configured cap, return Ok without any disk read.
2474 fn check_session_budget(&self, op: &lex_vcs::Operation) -> Result<(), StoreError> {
2475 let Some(intent_id) = op.intent_id.as_deref() else {
2476 return Ok(());
2477 };
2478 let intent_log = lex_vcs::IntentLog::open(self.root())?;
2479 let Some(intent) = intent_log.get(&intent_id.to_string())? else {
2480 // Dangling intent — treat as "no session" and let it
2481 // sail through. Slice 1's ledger already documents
2482 // this as graceful-degradation semantics.
2483 return Ok(());
2484 };
2485 let policy = crate::policy::load(self.root())?.unwrap_or_default();
2486 let Some(cap) = policy.session_budgets.cap_for(&intent.session_id) else {
2487 return Ok(());
2488 };
2489 // Recompute the session's current spend + the contribution
2490 // from this op. Re-running the ledger walk on every gated
2491 // op is O(branch history); see #292 slice 1's note about
2492 // a future on-disk cache.
2493 let current = self.session_budget(&intent.session_id)?;
2494 let increment = crate::budget::monotonic_spend_of(&op.kind);
2495 let spent_after = current.spent.saturating_add(increment);
2496 if spent_after > cap {
2497 return Err(StoreError::BudgetExceeded {
2498 session_id: intent.session_id,
2499 cap,
2500 spent_after,
2501 });
2502 }
2503 Ok(())
2504 }
2505
2506 /// Emit `RepairHint` attestations for a TypeError-rejected op
2507 /// (#281). One per candidate stage in the transition. The hint
2508 /// records the *would-be* op_id (deterministic, content-
2509 /// addressed even though the op record was never persisted)
2510 /// and the structured errors.
2511 ///
2512 /// #306 slice 3: `suggested_transform` is populated from the
2513 /// static (rule_tag → likely_transform) table for the *first*
2514 /// error in the batch. The LLM-driven `lex repair --apply`
2515 /// flow can still overwrite this with a higher-quality
2516 /// suggestion; the static value is the floor, not the ceiling.
2517 ///
2518 /// Best-effort: a write failure here is swallowed by the
2519 /// caller (the original `TypeError` is the load-bearing
2520 /// signal; missing the hint is recoverable on a retry).
2521 fn record_repair_hint(
2522 &self,
2523 stage_ids: &[String],
2524 failed_op_id: &lex_vcs::OpId,
2525 errors: &[lex_types::TypeError],
2526 ) -> Result<(), StoreError> {
2527 if stage_ids.is_empty() {
2528 return Ok(());
2529 }
2530 let errors_json = serde_json::to_value(errors).map_err(StoreError::Serde)?;
2531 // #306 slice 3: look up the static suggested_transform for
2532 // the first error's rule_tag. Multiple errors per op are
2533 // possible — when they fire in lockstep (e.g. one bad let
2534 // binding propagates to several use sites), the first
2535 // error's rule_tag is usually the load-bearing one to fix.
2536 let suggested_transform = errors
2537 .first()
2538 .and_then(|e| lex_types::suggested_transform_for(e.rule_tag()));
2539 let log = self.attestation_log()?;
2540 for stage_id in stage_ids {
2541 let attestation = lex_vcs::Attestation::new(
2542 stage_id.clone(),
2543 None, // the failed op was never persisted; not the
2544 // attestation's op_id (which is for a
2545 // *successful* op).
2546 None,
2547 lex_vcs::AttestationKind::RepairHint {
2548 failed_op_id: failed_op_id.clone(),
2549 errors: errors_json.clone(),
2550 suggested_transform: suggested_transform.clone(),
2551 },
2552 lex_vcs::AttestationResult::Failed {
2553 detail: format!(
2554 "op {} rejected: {} type error(s)",
2555 failed_op_id,
2556 errors.len()
2557 ),
2558 },
2559 repair_hint_producer(),
2560 None,
2561 );
2562 log.put(&attestation)?;
2563 }
2564 Ok(())
2565 }
2566
2567 /// Emit `Trace` attestations linking an already-committed `op`
2568 /// to the run that produced it (#257). One attestation per
2569 /// produced stage (matching the `TypeCheck` emission contract
2570 /// — see [`Self::apply_operation_checked`]) with
2571 /// `op_id: Some(op_id)` set, so `lex trace --op <op_id>`
2572 /// surfaces the run.
2573 ///
2574 /// Returns the number of attestations emitted (zero for ops
2575 /// that produce no attestable stage, e.g. `Remove` /
2576 /// `ImportOnly`).
2577 ///
2578 /// Idempotent: re-emitting for the same
2579 /// `(run_id, root_target, op_id, stage_id, producer, result)`
2580 /// tuple dedups via content addressing.
2581 ///
2582 /// `op_id` must already exist in the op log — an unknown op
2583 /// surfaces as `StoreError::UnknownOp`.
2584 pub fn record_op_trace(
2585 &self,
2586 run_id: &str,
2587 root_target: &str,
2588 op_id: &lex_vcs::OpId,
2589 result: lex_vcs::AttestationResult,
2590 producer: lex_vcs::ProducerDescriptor,
2591 ) -> Result<usize, StoreError> {
2592 let log = lex_vcs::OpLog::open(self.root())?;
2593 let rec = log
2594 .get(op_id)?
2595 .ok_or_else(|| StoreError::UnknownOp(op_id.clone()))?;
2596 let stage_ids = attestable_stage_ids(&rec.produces);
2597 if stage_ids.is_empty() {
2598 return Ok(0);
2599 }
2600 let attlog = self.attestation_log()?;
2601 let mut emitted = 0;
2602 for stage_id in stage_ids {
2603 let attestation = lex_vcs::Attestation::new(
2604 stage_id,
2605 Some(op_id.clone()),
2606 None,
2607 lex_vcs::AttestationKind::Trace {
2608 run_id: run_id.into(),
2609 root_target: root_target.into(),
2610 },
2611 result.clone(),
2612 producer.clone(),
2613 None,
2614 );
2615 attlog.put(&attestation)?;
2616 emitted += 1;
2617 }
2618 Ok(emitted)
2619 }
2620
2621 /// Walk `ops_since(branch_head, base)` and emit per-stage
2622 /// `Trace` attestations for each new op, linking them to the
2623 /// run that produced them (#257). Used by `lex run --trace`
2624 /// after the VM exits: snapshot `base = branch_head` before
2625 /// the run, then call this with the post-run head.
2626 ///
2627 /// `base = None` means "every op currently reachable from the
2628 /// branch head" — generally not what you want for a single
2629 /// run; pass the pre-run head.
2630 ///
2631 /// Returns the total number of attestations emitted across
2632 /// every new op. Zero is the common case (the run committed no
2633 /// ops).
2634 ///
2635 /// Idempotent on the per-op level via [`Self::record_op_trace`].
2636 pub fn record_run_committed_ops_since(
2637 &self,
2638 run_id: &str,
2639 root_target: &str,
2640 branch: &str,
2641 base: Option<&lex_vcs::OpId>,
2642 result: lex_vcs::AttestationResult,
2643 producer: lex_vcs::ProducerDescriptor,
2644 ) -> Result<usize, StoreError> {
2645 let head = match self.get_branch(branch)?.and_then(|b| b.head_op) {
2646 Some(h) => h,
2647 None => return Ok(0),
2648 };
2649 let log = lex_vcs::OpLog::open(self.root())?;
2650 let new_ops = log.ops_since(&head, base)?;
2651 let mut total = 0;
2652 for rec in new_ops {
2653 total += self.record_op_trace(
2654 run_id,
2655 root_target,
2656 &rec.op_id,
2657 result.clone(),
2658 producer.clone(),
2659 )?;
2660 }
2661 Ok(total)
2662 }
2663
2664 /// Apply a typed `ReplaceMatchArm` transform (#280) and emit a
2665 /// `OperationKind::ReplaceMatchArm` op that records the
2666 /// semantic shape of the edit, not just the byte effect.
2667 ///
2668 /// Steps:
2669 /// 1. Load the source stage's canonical bytes (delta-aware).
2670 /// 2. Run [`lex_ast::replace_match_arm`] to produce the new
2671 /// `Stage`. Pure function, no I/O.
2672 /// 3. Publish the new stage. Idempotent on the
2673 /// content-addressed `to_stage_id`.
2674 /// 4. Assemble the candidate program (every active stage on
2675 /// the branch, with the rewritten one swapped in) and call
2676 /// [`Self::apply_operation_checked`] — re-typechecks and
2677 /// runs every existing gate (TypeCheck attestation,
2678 /// required_attestations, producer-block walk-back).
2679 ///
2680 /// Failure modes:
2681 /// * [`StoreError::TransformError`] — transform didn't apply.
2682 /// The branch is unchanged; no stage published.
2683 /// * [`StoreError::TypeError`] — transform produced an
2684 /// ill-typed program. The new stage is on disk (idempotent
2685 /// on its content hash) but the branch is unchanged. Same
2686 /// "publish without advance" semantics as #245.
2687 /// * Everything else from `apply_operation_checked`.
2688 pub fn apply_replace_match_arm(
2689 &self,
2690 branch: &str,
2691 from_stage_id: &str,
2692 match_node: &lex_ast::NodeId,
2693 arm_index: usize,
2694 new_body: lex_ast::CExpr,
2695 ) -> Result<lex_vcs::OpId, StoreError> {
2696 let from_stage = self.get_ast(from_stage_id)?;
2697 let new_stage = lex_ast::replace_match_arm(&from_stage, match_node, arm_index, new_body)
2698 .map_err(StoreError::TransformError)?;
2699 let sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2700 let to_stage_id = self.publish(&new_stage)?;
2701 if to_stage_id == from_stage_id {
2702 // No-op transform — the new body was structurally
2703 // identical to the old. Refuse rather than advancing
2704 // the branch with an empty edit.
2705 return Err(StoreError::InvalidTransition(format!(
2706 "replace_match_arm produced the same stage_id `{from_stage_id}`"
2707 )));
2708 }
2709
2710 // Assemble the candidate program: every active stage on
2711 // the branch, with `from_stage_id` swapped for `new_stage`.
2712 let head = self.branch_head(branch)?;
2713 let mut candidate: Vec<lex_ast::Stage> = Vec::with_capacity(head.len());
2714 for (other_sig, other_stage_id) in &head {
2715 if other_sig == &sig {
2716 candidate.push(new_stage.clone());
2717 } else {
2718 candidate.push(self.get_ast(other_stage_id)?);
2719 }
2720 }
2721 // If the source sig isn't on the current branch head, the
2722 // transform is operating on a stage that hasn't been added
2723 // yet — refuse rather than risking a candidate program
2724 // that doesn't reflect the branch's actual state.
2725 if !head.contains_key(&sig) {
2726 return Err(StoreError::InvalidTransition(format!(
2727 "sig `{sig}` not on branch `{branch}`'s head"
2728 )));
2729 }
2730
2731 // #247: budget delta captured for `lex op log --budget-drift`.
2732 let from_budget = budget_of_stage(&from_stage);
2733 let to_budget = budget_of_stage(&new_stage);
2734
2735 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2736 let kind = lex_vcs::OperationKind::ReplaceMatchArm {
2737 sig_id: sig.clone(),
2738 from_stage_id: from_stage_id.to_string(),
2739 to_stage_id: to_stage_id.clone(),
2740 match_node: match_node.as_str().to_string(),
2741 arm_index,
2742 from_budget,
2743 to_budget,
2744 };
2745 let transition = lex_vcs::StageTransition::Replace {
2746 sig_id: sig.clone(),
2747 from: from_stage_id.to_string(),
2748 to: to_stage_id.clone(),
2749 };
2750 let op = lex_vcs::Operation::new(kind, head_now.into_iter().collect::<Vec<_>>());
2751 self.apply_operation_checked(branch, op, transition, &candidate)
2752 }
2753
2754 /// Apply a typed `RenameLocal` transform (#280) — rename a
2755 /// `let`-bound local within a fn body and emit a matching
2756 /// `OperationKind::RenameLocal`. Same end-to-end shape as
2757 /// [`Self::apply_replace_match_arm`]; see that method for the
2758 /// failure-mode taxonomy.
2759 pub fn apply_rename_local(
2760 &self,
2761 branch: &str,
2762 from_stage_id: &str,
2763 let_node: &lex_ast::NodeId,
2764 new_name: &str,
2765 ) -> Result<lex_vcs::OpId, StoreError> {
2766 let from_stage = self.get_ast(from_stage_id)?;
2767 // Read the old name before running the transform, so the
2768 // op log records the rename target rather than just the
2769 // new value.
2770 let old_name = read_let_name(&from_stage, let_node).map_err(StoreError::TransformError)?;
2771 let new_stage = lex_ast::rename_local(&from_stage, let_node, new_name)
2772 .map_err(StoreError::TransformError)?;
2773 let sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2774 let to_stage_id = self.publish(&new_stage)?;
2775 if to_stage_id == from_stage_id {
2776 return Err(StoreError::InvalidTransition(format!(
2777 "rename_local produced the same stage_id `{from_stage_id}`"
2778 )));
2779 }
2780 let head = self.branch_head(branch)?;
2781 let mut candidate: Vec<lex_ast::Stage> = Vec::with_capacity(head.len());
2782 for (other_sig, other_stage_id) in &head {
2783 if other_sig == &sig {
2784 candidate.push(new_stage.clone());
2785 } else {
2786 candidate.push(self.get_ast(other_stage_id)?);
2787 }
2788 }
2789 if !head.contains_key(&sig) {
2790 return Err(StoreError::InvalidTransition(format!(
2791 "sig `{sig}` not on branch `{branch}`'s head"
2792 )));
2793 }
2794 let from_budget = budget_of_stage(&from_stage);
2795 let to_budget = budget_of_stage(&new_stage);
2796 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2797 let kind = lex_vcs::OperationKind::RenameLocal {
2798 sig_id: sig.clone(),
2799 from_stage_id: from_stage_id.to_string(),
2800 to_stage_id: to_stage_id.clone(),
2801 let_node: let_node.as_str().to_string(),
2802 old_name,
2803 new_name: new_name.to_string(),
2804 from_budget,
2805 to_budget,
2806 };
2807 let transition = lex_vcs::StageTransition::Replace {
2808 sig_id: sig.clone(),
2809 from: from_stage_id.to_string(),
2810 to: to_stage_id.clone(),
2811 };
2812 let op = lex_vcs::Operation::new(kind, head_now.into_iter().collect::<Vec<_>>());
2813 self.apply_operation_checked(branch, op, transition, &candidate)
2814 }
2815
2816 /// Apply a typed `InlineLet` transform (#280) — eliminate a
2817 /// `let x := v; body` by substituting `v` for every unshadowed
2818 /// `x` in `body`, then replacing the `Let` node with the
2819 /// substituted body. Same end-to-end shape as
2820 /// [`Self::apply_replace_match_arm`].
2821 pub fn apply_inline_let(
2822 &self,
2823 branch: &str,
2824 from_stage_id: &str,
2825 let_node: &lex_ast::NodeId,
2826 ) -> Result<lex_vcs::OpId, StoreError> {
2827 let from_stage = self.get_ast(from_stage_id)?;
2828 let binding_name =
2829 read_let_name(&from_stage, let_node).map_err(StoreError::TransformError)?;
2830 let new_stage =
2831 lex_ast::inline_let(&from_stage, let_node).map_err(StoreError::TransformError)?;
2832 let sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2833 let to_stage_id = self.publish(&new_stage)?;
2834 if to_stage_id == from_stage_id {
2835 return Err(StoreError::InvalidTransition(format!(
2836 "inline_let produced the same stage_id `{from_stage_id}`"
2837 )));
2838 }
2839 let head = self.branch_head(branch)?;
2840 let mut candidate: Vec<lex_ast::Stage> = Vec::with_capacity(head.len());
2841 for (other_sig, other_stage_id) in &head {
2842 if other_sig == &sig {
2843 candidate.push(new_stage.clone());
2844 } else {
2845 candidate.push(self.get_ast(other_stage_id)?);
2846 }
2847 }
2848 if !head.contains_key(&sig) {
2849 return Err(StoreError::InvalidTransition(format!(
2850 "sig `{sig}` not on branch `{branch}`'s head"
2851 )));
2852 }
2853 let from_budget = budget_of_stage(&from_stage);
2854 let to_budget = budget_of_stage(&new_stage);
2855 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2856 let kind = lex_vcs::OperationKind::InlineLet {
2857 sig_id: sig.clone(),
2858 from_stage_id: from_stage_id.to_string(),
2859 to_stage_id: to_stage_id.clone(),
2860 let_node: let_node.as_str().to_string(),
2861 binding_name,
2862 from_budget,
2863 to_budget,
2864 };
2865 let transition = lex_vcs::StageTransition::Replace {
2866 sig_id: sig.clone(),
2867 from: from_stage_id.to_string(),
2868 to: to_stage_id.clone(),
2869 };
2870 let op = lex_vcs::Operation::new(kind, head_now.into_iter().collect::<Vec<_>>());
2871 self.apply_operation_checked(branch, op, transition, &candidate)
2872 }
2873
2874 /// Apply a typed `ExtractFunction` transform (#280 slice 4) —
2875 /// extract a sub-expression of `from_stage_id`'s body into a
2876 /// new top-level fn defined by `spec`, and emit two ops tied
2877 /// together by a shared synthetic Intent so `lex op log
2878 /// --intent <id>` groups them.
2879 ///
2880 /// The two ops:
2881 /// 1. `AddFunction { sig_id: <new_fn_sig>, stage_id: <new_fn_stage> }`
2882 /// 2. `ModifyBody { sig_id: <source_sig>, from_stage_id, to_stage_id: <modified> }`
2883 ///
2884 /// The shared Intent's prompt is structured (`extract_function:
2885 /// <new_fn_name>` plus the source identity) so downstream
2886 /// tooling can recover the typed-transform shape from the
2887 /// op-log + intent-log join.
2888 ///
2889 /// Returns `(add_fn_op_id, modify_body_op_id)`.
2890 pub fn apply_extract_function(
2891 &self,
2892 branch: &str,
2893 from_stage_id: &str,
2894 expr_node: &lex_ast::NodeId,
2895 spec: lex_ast::ExtractFnSpec,
2896 ) -> Result<(lex_vcs::OpId, lex_vcs::OpId), StoreError> {
2897 let from_stage = self.get_ast(from_stage_id)?;
2898 let new_fn_name = spec.name.clone();
2899 let (modified_stage, new_fn_stage) =
2900 lex_ast::extract_function(&from_stage, expr_node, spec)
2901 .map_err(StoreError::TransformError)?;
2902
2903 let source_sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2904 let new_fn_sig = lex_ast::sig_id(&new_fn_stage).ok_or(StoreError::CannotPublishImport)?;
2905 if source_sig == new_fn_sig {
2906 return Err(StoreError::InvalidTransition(format!(
2907 "extract_function produced a sig matching the source `{source_sig}`"
2908 )));
2909 }
2910 let new_fn_stage_id = self.publish(&new_fn_stage)?;
2911 let modified_stage_id = self.publish(&modified_stage)?;
2912 if modified_stage_id == from_stage_id {
2913 return Err(StoreError::InvalidTransition(format!(
2914 "extract_function produced the same stage_id `{from_stage_id}` for the source"
2915 )));
2916 }
2917
2918 let head = self.branch_head(branch)?;
2919 if !head.contains_key(&source_sig) {
2920 return Err(StoreError::InvalidTransition(format!(
2921 "sig `{source_sig}` not on branch `{branch}`'s head"
2922 )));
2923 }
2924
2925 // Synthesize an Intent linking the two ops. The session_id
2926 // / model fields here are not load-bearing — they exist to
2927 // make the IntentId content-addressed; downstream tooling
2928 // reads `prompt` to reconstruct the typed-transform shape.
2929 let intent = lex_vcs::Intent::new(
2930 format!(
2931 "[lex.transform.extract_function]\nnew_fn={new_fn_name}\nsource_sig={source_sig}\nfrom_stage={from_stage_id}\nexpr_node={node}",
2932 node = expr_node.as_str(),
2933 ),
2934 "lex-store::apply_extract_function",
2935 lex_vcs::ModelDescriptor {
2936 provider: "lex-store".into(),
2937 name: env!("CARGO_PKG_VERSION").into(),
2938 version: None,
2939 },
2940 None,
2941 );
2942 let intent_id = intent.intent_id.clone();
2943 lex_vcs::IntentLog::open(self.root())?.put(&intent)?;
2944
2945 // Step 1 — emit the AddFunction op for the new fn. Build
2946 // the candidate program by appending the new fn to every
2947 // stage on the current branch head.
2948 let new_fn_effects: std::collections::BTreeSet<String> = match &new_fn_stage {
2949 lex_ast::Stage::FnDecl(fd) => fd.effects.iter().map(|e| e.name.clone()).collect(),
2950 _ => Default::default(),
2951 };
2952 let new_fn_budget = budget_of_stage(&new_fn_stage);
2953 let mut candidate_with_new_fn: Vec<lex_ast::Stage> = Vec::with_capacity(head.len() + 1);
2954 for stage_id in head.values() {
2955 candidate_with_new_fn.push(self.get_ast(stage_id)?);
2956 }
2957 candidate_with_new_fn.push(new_fn_stage.clone());
2958 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2959 let add_op = lex_vcs::Operation::new(
2960 lex_vcs::OperationKind::AddFunction {
2961 sig_id: new_fn_sig.clone(),
2962 stage_id: new_fn_stage_id.clone(),
2963 effects: new_fn_effects,
2964 budget_cost: new_fn_budget,
2965 // Single-op apply path — no package context here.
2966 in_file: None,
2967 },
2968 head_now.into_iter().collect::<Vec<_>>(),
2969 )
2970 .with_intent(intent_id.clone());
2971 let add_transition = lex_vcs::StageTransition::Create {
2972 sig_id: new_fn_sig.clone(),
2973 stage_id: new_fn_stage_id.clone(),
2974 };
2975 let add_op_id =
2976 self.apply_operation_checked(branch, add_op, add_transition, &candidate_with_new_fn)?;
2977
2978 // Step 2 — emit the ModifyBody op for the source. Build
2979 // the candidate program by replacing the source's stage
2980 // with `modified_stage` and keeping the new fn alongside.
2981 let from_budget = budget_of_stage(&from_stage);
2982 let to_budget = budget_of_stage(&modified_stage);
2983 let mut candidate_with_modified: Vec<lex_ast::Stage> = Vec::with_capacity(head.len() + 1);
2984 for (other_sig, other_stage_id) in &head {
2985 if other_sig == &source_sig {
2986 candidate_with_modified.push(modified_stage.clone());
2987 } else {
2988 candidate_with_modified.push(self.get_ast(other_stage_id)?);
2989 }
2990 }
2991 candidate_with_modified.push(new_fn_stage.clone());
2992 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2993 let modify_op = lex_vcs::Operation::new(
2994 lex_vcs::OperationKind::ModifyBody {
2995 sig_id: source_sig.clone(),
2996 from_stage_id: from_stage_id.to_string(),
2997 to_stage_id: modified_stage_id.clone(),
2998 from_budget,
2999 to_budget,
3000 },
3001 head_now.into_iter().collect::<Vec<_>>(),
3002 )
3003 .with_intent(intent_id);
3004 let modify_transition = lex_vcs::StageTransition::Replace {
3005 sig_id: source_sig,
3006 from: from_stage_id.to_string(),
3007 to: modified_stage_id,
3008 };
3009 let modify_op_id = self.apply_operation_checked(
3010 branch,
3011 modify_op,
3012 modify_transition,
3013 &candidate_with_modified,
3014 )?;
3015
3016 Ok((add_op_id, modify_op_id))
3017 }
3018
3019 /// Propose a stage for `sig_id` without advancing the branch
3020 /// head (#294). Multiple agents can call this concurrently
3021 /// for the same sig — every call lands a fresh `Candidate`
3022 /// op chained off the current head_op. The branch head stays
3023 /// where it was; a later [`Self::promote_candidate`] picks
3024 /// the winner.
3025 ///
3026 /// The caller is responsible for typechecking `new_stage`
3027 /// against whatever program context they consider valid —
3028 /// `propose_candidate` doesn't run the gate. Type errors
3029 /// surface at promotion time, where the candidate is
3030 /// composed back into a candidate program via the standard
3031 /// `apply_operation_checked` path.
3032 ///
3033 /// The stage is published (idempotent on content hash). The
3034 /// `intent_id` is required so downstream consumers can
3035 /// distinguish proposals by author.
3036 pub fn propose_candidate(
3037 &self,
3038 branch: &str,
3039 new_stage: &lex_ast::Stage,
3040 intent_id: &lex_vcs::IntentId,
3041 ) -> Result<lex_vcs::OpId, StoreError> {
3042 let sig = lex_ast::sig_id(new_stage).ok_or(StoreError::CannotPublishImport)?;
3043 let stage_id = self.publish(new_stage)?;
3044 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
3045 let op = lex_vcs::Operation::new(
3046 lex_vcs::OperationKind::Candidate {
3047 sig_id: sig,
3048 stage_id,
3049 },
3050 head_now.into_iter().collect::<Vec<_>>(),
3051 )
3052 .with_intent(intent_id.clone());
3053 let transition = lex_vcs::StageTransition::ImportOnly;
3054 self.apply_operation(branch, op, transition)
3055 }
3056
3057 /// List every live `Candidate` op for `sig_id` — i.e. those
3058 /// not yet referenced by any `Promote` op (either as the
3059 /// winner or in the `supersedes` set). Used by `lex stage
3060 /// candidates`. Results are sorted by op_id for
3061 /// reproducibility.
3062 pub fn list_candidates(&self, sig_id: &str) -> Result<Vec<CandidateInfo>, StoreError> {
3063 let log = lex_vcs::OpLog::open(self.root())?;
3064 let all = log.list_all()?;
3065 // Collect the set of candidate op_ids referenced by any
3066 // Promote for this sig. Those candidates are no longer
3067 // live.
3068 let mut referenced: std::collections::BTreeSet<lex_vcs::OpId> = Default::default();
3069 for rec in &all {
3070 if let lex_vcs::OperationKind::Promote {
3071 sig_id: s,
3072 winner_candidate,
3073 supersedes,
3074 ..
3075 } = &rec.op.kind
3076 {
3077 if s != sig_id {
3078 continue;
3079 }
3080 referenced.insert(winner_candidate.clone());
3081 for sup in supersedes {
3082 referenced.insert(sup.clone());
3083 }
3084 }
3085 }
3086 let mut out: Vec<CandidateInfo> = Vec::new();
3087 for rec in all {
3088 let lex_vcs::OperationKind::Candidate {
3089 sig_id: s,
3090 stage_id,
3091 } = &rec.op.kind
3092 else {
3093 continue;
3094 };
3095 if s != sig_id {
3096 continue;
3097 }
3098 if referenced.contains(&rec.op_id) {
3099 continue;
3100 }
3101 out.push(CandidateInfo {
3102 op_id: rec.op_id.clone(),
3103 stage_id: stage_id.clone(),
3104 intent_id: rec.op.intent_id.clone(),
3105 });
3106 }
3107 out.sort_by(|a, b| a.op_id.cmp(&b.op_id));
3108 Ok(out)
3109 }
3110
3111 /// Promote a previously-landed `Candidate` op as the new
3112 /// branch head for its sig (#294). Emits a `Promote` op
3113 /// listing every other live `Candidate` for the same sig
3114 /// in its `supersedes` field. After this lands,
3115 /// [`Self::list_candidates`] returns an empty set for the
3116 /// sig.
3117 ///
3118 /// Re-typechecks the candidate program (winner stage + the
3119 /// rest of the branch) through `apply_operation_checked`, so
3120 /// a candidate that doesn't compose with the current branch
3121 /// state surfaces as `StoreError::TypeError`.
3122 pub fn promote_candidate(
3123 &self,
3124 branch: &str,
3125 candidate_op_id: &lex_vcs::OpId,
3126 ) -> Result<lex_vcs::OpId, StoreError> {
3127 let log = lex_vcs::OpLog::open(self.root())?;
3128 let candidate_rec = log
3129 .get(candidate_op_id)?
3130 .ok_or_else(|| StoreError::UnknownOp(candidate_op_id.clone()))?;
3131 let (sig, winner_stage_id) = match &candidate_rec.op.kind {
3132 lex_vcs::OperationKind::Candidate { sig_id, stage_id } => {
3133 (sig_id.clone(), stage_id.clone())
3134 }
3135 other => {
3136 return Err(StoreError::InvalidTransition(format!(
3137 "op `{candidate_op_id}` is a `{:?}`, not a Candidate",
3138 other
3139 )))
3140 }
3141 };
3142
3143 // #836 G4: a candidate carrying a standing `Reject` review must
3144 // not be promoted. "Standing" = the latest `Review` on the
3145 // winner's stage is a Reject; a later `Approve` (or
3146 // `RequestChanges`, which is advisory, not a veto) lifts it.
3147 // Safe by default: a candidate with no review, or an approved
3148 // one, promotes exactly as before.
3149 if let Some(lex_vcs::ReviewVerdict::Reject) = self.latest_review_verdict(&winner_stage_id)? {
3150 return Err(StoreError::InvalidTransition(format!(
3151 "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"
3152 )));
3153 }
3154
3155 // Gather every OTHER live candidate for this sig — the
3156 // ones this Promote will supersede.
3157 let live = self.list_candidates(&sig)?;
3158 let mut supersedes: Vec<lex_vcs::OpId> = live
3159 .iter()
3160 .filter(|c| &c.op_id != candidate_op_id)
3161 .map(|c| c.op_id.clone())
3162 .collect();
3163 supersedes.sort();
3164
3165 // Assemble candidate program: winner stage in place of
3166 // the sig's current head (if any), plus every other sig
3167 // unchanged.
3168 let head = self.branch_head(branch)?;
3169 let winner_stage = self.get_ast(&winner_stage_id)?;
3170 let mut candidate_program: Vec<lex_ast::Stage> = Vec::with_capacity(head.len() + 1);
3171 let mut found = false;
3172 for (other_sig, other_stage_id) in &head {
3173 if other_sig == &sig {
3174 candidate_program.push(winner_stage.clone());
3175 found = true;
3176 } else {
3177 candidate_program.push(self.get_ast(other_stage_id)?);
3178 }
3179 }
3180 if !found {
3181 // Sig doesn't have a head yet — append the winner
3182 // stage to make it a Create.
3183 candidate_program.push(winner_stage.clone());
3184 }
3185 let from_stage_id = head.get(&sig).cloned();
3186 // Budget delta from old head to winner — same shape as
3187 // ModifyBody.
3188 let from_budget = from_stage_id
3189 .as_deref()
3190 .and_then(|s| self.get_ast(s).ok())
3191 .and_then(|s| budget_of_stage(&s));
3192 let to_budget = budget_of_stage(&winner_stage);
3193
3194 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
3195 let op = lex_vcs::Operation::new(
3196 lex_vcs::OperationKind::Promote {
3197 sig_id: sig.clone(),
3198 winner_candidate: candidate_op_id.clone(),
3199 winner_stage_id: winner_stage_id.clone(),
3200 supersedes,
3201 from_stage_id: from_stage_id.clone(),
3202 from_budget,
3203 to_budget,
3204 },
3205 head_now.into_iter().collect::<Vec<_>>(),
3206 );
3207 let transition = match &from_stage_id {
3208 Some(from) => lex_vcs::StageTransition::Replace {
3209 sig_id: sig,
3210 from: from.clone(),
3211 to: winner_stage_id,
3212 },
3213 None => lex_vcs::StageTransition::Create {
3214 sig_id: sig,
3215 stage_id: winner_stage_id,
3216 },
3217 };
3218 self.apply_operation_checked(branch, op, transition, &candidate_program)
3219 }
3220
3221 /// `set_branch_head_op` for the durability story on the branch
3222 /// file itself.
3223 pub fn apply_operation(
3224 &self,
3225 branch: &str,
3226 op: lex_vcs::Operation,
3227 transition: lex_vcs::StageTransition,
3228 ) -> Result<lex_vcs::OpId, StoreError> {
3229 let attestable = attestable_stage_ids(&transition);
3230 let op_effects = op_declared_effects(&op.kind);
3231 self.cas_retry_advance(branch, op, transition, |new_head| {
3232 self.run_required_attestations_gate(branch, &new_head.op_id, &attestable, &op_effects)
3233 })
3234 }
3235
3236 /// CAS retry loop for #262. Single-parent ops are rebuilt on
3237 /// each iteration with the current branch head as parent;
3238 /// the per-iteration callback runs the gate (and TypeCheck
3239 /// emission, for the checked path) between persist and CAS.
3240 /// Merge ops (with 2 parents already set) skip the rebuild —
3241 /// their parents are caller-supplied and meaningful — and get
3242 /// a single attempt; on CAS failure they surface `Contention`.
3243 fn cas_retry_advance<F>(
3244 &self,
3245 branch: &str,
3246 op: lex_vcs::Operation,
3247 transition: lex_vcs::StageTransition,
3248 mut between_persist_and_cas: F,
3249 ) -> Result<lex_vcs::OpId, StoreError>
3250 where
3251 F: FnMut(&lex_vcs::NewHead) -> Result<(), StoreError>,
3252 {
3253 // 32 retries handles up to ~32 concurrent writers racing on
3254 // the same branch tip. Beyond that, surfacing `Contention`
3255 // is the right signal — clients should back off or batch.
3256 const MAX_ATTEMPTS: u32 = 32;
3257 // Single-parent ops can be rebuilt on retry; merge ops
3258 // can't (their two parents are meaningful, supplied by the
3259 // merge engine). For merges, single attempt: if CAS
3260 // fails, surface Contention.
3261 let is_rebuildable = op.parents.len() <= 1;
3262 let kind = op.kind.clone();
3263 let intent_id = op.intent_id.clone();
3264
3265 let mut last_io_err: Option<StoreError> = None;
3266 let mut current_op = op;
3267 let current_transition = transition;
3268 // Only rebuild on retries — attempt 1 honors the caller's
3269 // exact op so a user-supplied bogus parent (parents =
3270 // ["someone-else"]) surfaces as `StaleParent` instead of
3271 // being silently corrected.
3272 //
3273 // Exception (#262 follow-up): an op with `parents = []`
3274 // means "I don't care; chain off whatever the current
3275 // head is." Under concurrent apply, attempt 1 can read
3276 // `head_op = Some(opA)` after a sibling writer landed,
3277 // and the persist's parent check fails StaleParent
3278 // unprompted. Rebuild attempt 1 for the empty-parents
3279 // case so the legitimate-race path retries cleanly.
3280 let mut rebuilt_already = false;
3281 for attempt in 1..=MAX_ATTEMPTS {
3282 // Read the current head BEFORE we persist — this is
3283 // the value we'll compare against in the CAS.
3284 let parent = self.get_branch(branch)?.and_then(|b| b.head_op);
3285
3286 // Rebuild the op against the current head, but only
3287 // on retries (not the caller's first attempt) and
3288 // only for single-parent operations. Multi-parent
3289 // (merge) ops are passed through unchanged.
3290 //
3291 // Empty-parents ops also rebuild on attempt 1 (see
3292 // the exception note above) so concurrent apply
3293 // doesn't false-positive on StaleParent.
3294 let should_rebuild = is_rebuildable
3295 && (rebuilt_already || (current_op.parents.is_empty() && parent.is_some()));
3296 if should_rebuild {
3297 current_op = lex_vcs::Operation {
3298 kind: kind.clone(),
3299 parents: parent.iter().cloned().collect(),
3300 intent_id: intent_id.clone(),
3301 };
3302 }
3303
3304 // Persist (idempotent). On `StaleParent` from a retry
3305 // attempt (where we already rebuilt), the head changed
3306 // between our `get_branch` and this `lex_vcs::apply`
3307 // — race; rebuild and continue. On `StaleParent` from
3308 // attempt 1 (caller's input), propagate.
3309 let new_head = match self.persist_op_only_with_parent(
3310 branch,
3311 parent.as_ref(),
3312 current_op.clone(),
3313 current_transition.clone(),
3314 ) {
3315 Ok(nh) => nh,
3316 Err(StoreError::Apply(lex_vcs::ApplyError::StaleParent { .. }))
3317 if is_rebuildable && rebuilt_already =>
3318 {
3319 rebuilt_already = true;
3320 continue;
3321 }
3322 Err(e) => return Err(e),
3323 };
3324
3325 // Run the caller's between-persist-and-cas hook
3326 // (TypeCheck emission + gate). If this fails, the op
3327 // record is durable but orphaned — same semantics as
3328 // pre-#262.
3329 between_persist_and_cas(&new_head)?;
3330
3331 // CAS the branch head. On success: done. On mismatch:
3332 // someone advanced in parallel; retry.
3333 match self.set_branch_head_op_cas(branch, parent, new_head.op_id.clone()) {
3334 Ok(()) => return Ok(new_head.op_id),
3335 Err(crate::branches::CasFailed::Mismatch { .. }) if is_rebuildable => {
3336 // Try again with the new head as parent.
3337 rebuilt_already = true;
3338 continue;
3339 }
3340 Err(crate::branches::CasFailed::Mismatch { .. }) => {
3341 // Merge op: surface immediately — we can't
3342 // rebuild without rerunning the merge engine.
3343 let _ = attempt;
3344 return Err(StoreError::Contention {
3345 branch: branch.into(),
3346 attempts: 1,
3347 });
3348 }
3349 Err(crate::branches::CasFailed::UnknownBranch(b)) => {
3350 return Err(StoreError::UnknownBranch(b));
3351 }
3352 Err(crate::branches::CasFailed::Io(e)) => {
3353 last_io_err = Some(StoreError::Io(std::io::Error::other(e)));
3354 continue;
3355 }
3356 }
3357 }
3358 // Retries exhausted. Prefer surfacing the most recent IO
3359 // error if we hit one; otherwise it's pure CAS contention.
3360 match last_io_err {
3361 Some(e) => Err(e),
3362 None => Err(StoreError::Contention {
3363 branch: branch.into(),
3364 attempts: MAX_ATTEMPTS,
3365 }),
3366 }
3367 }
3368
3369 /// Persist an op against an explicitly-supplied parent. Used
3370 /// by the CAS retry loop in `cas_retry_advance` so the
3371 /// `lex_vcs::apply` parent check matches what we read at the
3372 /// top of the loop iteration (avoids a TOCTOU race against
3373 /// `persist_op_only`'s second read).
3374 fn persist_op_only_with_parent(
3375 &self,
3376 branch: &str,
3377 parent: Option<&lex_vcs::OpId>,
3378 op: lex_vcs::Operation,
3379 transition: lex_vcs::StageTransition,
3380 ) -> Result<lex_vcs::NewHead, StoreError> {
3381 if branch != DEFAULT_BRANCH && self.get_branch(branch)?.is_none() {
3382 return Err(StoreError::UnknownBranch(branch.into()));
3383 }
3384 let log = lex_vcs::OpLog::open(self.root())?;
3385 lex_vcs::apply(&log, parent, op, transition).map_err(|e| match e {
3386 lex_vcs::ApplyError::Persist(io) => StoreError::Io(io),
3387 other => StoreError::Apply(other),
3388 })
3389 }
3390
3391 /// Run the `required_attestations` gate (#245) and the
3392 /// retroactive producer-block gate (#248) over a single op
3393 /// against the store's `policy.json` and attestation log.
3394 ///
3395 /// Failure modes (in order):
3396 ///
3397 /// 1. Producer-block first: if any attestation on the op's
3398 /// stage is from a quarantined tool, refuse with
3399 /// `ProducerBlocked` (#248). Surfaces *before* the
3400 /// required-attestations gate so a clearly-malicious record
3401 /// isn't masked by a missing-Spec error.
3402 /// 2. Required-attestations next: if any required attestation
3403 /// kind is missing, refuse with `BranchAdvanceBlocked`
3404 /// (#245).
3405 ///
3406 /// Loads the policy / attestation log lazily; with no policy
3407 /// file and no `ProducerBlock` attestations the gate is a no-op
3408 /// (default-permissive — matches pre-#245 stores).
3409 fn run_required_attestations_gate(
3410 &self,
3411 branch: &str,
3412 op_id: &lex_vcs::OpId,
3413 stage_ids: &[String],
3414 op_effects: &std::collections::BTreeSet<String>,
3415 ) -> Result<(), StoreError> {
3416 // Build the candidate slice for the new op. Ops with no
3417 // attestable stage (imports, empty merges) get a single
3418 // `None`-stage tuple; both gates skip those.
3419 let new_op_candidate: Vec<(
3420 lex_vcs::OpId,
3421 Option<String>,
3422 std::collections::BTreeSet<String>,
3423 )> = if stage_ids.is_empty() {
3424 vec![(op_id.clone(), None, op_effects.clone())]
3425 } else {
3426 stage_ids
3427 .iter()
3428 .map(|sid| (op_id.clone(), Some(sid.clone()), op_effects.clone()))
3429 .collect()
3430 };
3431 let attest_log = self.attestation_log()?;
3432
3433 // #248 + #256: producer-block gate, walk-back style.
3434 //
3435 // The naive #248 gate only checked the new op's stage. That
3436 // missed contamination on ancestors — once `lex attest
3437 // retro-block` lands, every previously-gated op stays in
3438 // the chain even though its attestations are now from a
3439 // quarantined producer.
3440 //
3441 // #256 fixes this by walking the chain from `head_op` back
3442 // to `last_gate_checkpoint` (or genesis when the checkpoint
3443 // is invalidated), collecting each ancestor's attestable
3444 // stages, and running `check_producer_block` on the
3445 // combined set. After a successful advance,
3446 // `set_branch_head_op` moves the checkpoint to the new
3447 // head (steady-state O(new ops) per advance).
3448 let walk_back_candidate = self.collect_ancestor_candidates(branch)?;
3449 let mut producer_block_candidate = walk_back_candidate;
3450 producer_block_candidate.extend(new_op_candidate.iter().cloned());
3451 crate::policy::check_producer_block(&attest_log, &producer_block_candidate)
3452 .map_err(StoreError::ProducerBlocked)?;
3453
3454 // #245: required-attestations gate. Forward-going only —
3455 // only the new op is checked. Walking back makes no sense
3456 // here: the policy is "this advance must carry these
3457 // attestations," not "every prior op must have."
3458 let policy = match crate::policy::load(self.root())? {
3459 Some(p) if !p.required_attestations.is_empty() => p,
3460 _ => return Ok(()),
3461 };
3462 let waivers =
3463 crate::policy::check_required_attestations(&attest_log, &new_op_candidate, &policy)
3464 .map_err(StoreError::BranchAdvanceBlocked)?;
3465 // #293: emit one `TrustWaived` attestation per waiver so
3466 // the audit trail records every skip. Idempotent on
3467 // attestation_id (content-addressed dedup) — re-running
3468 // the gate with the same state writes the same files.
3469 for w in waivers {
3470 let att = lex_vcs::Attestation::new(
3471 w.stage_id,
3472 Some(op_id.clone()),
3473 None,
3474 lex_vcs::AttestationKind::TrustWaived {
3475 producer: w.producer,
3476 score_thousandths: w.score_thousandths,
3477 threshold_thousandths: w.threshold_thousandths,
3478 kind_tag: w.kind_tag,
3479 },
3480 lex_vcs::AttestationResult::Passed,
3481 trust_waived_producer(),
3482 None,
3483 );
3484 attest_log.put(&att)?;
3485 }
3486 Ok(())
3487 }
3488
3489 /// Walk the branch from `head_op` back to `last_gate_checkpoint`
3490 /// (exclusive) and return the `(op_id, stage_id, op_effects)`
3491 /// tuples for every attestable stage touched by an ancestor
3492 /// (#256). Empty when the branch is fresh, when the checkpoint
3493 /// equals the head, or when the head is None.
3494 fn collect_ancestor_candidates(&self, branch: &str) -> Result<Vec<GateCandidate>, StoreError> {
3495 let b = match self.get_branch(branch)? {
3496 Some(b) => b,
3497 None => return Ok(Vec::new()),
3498 };
3499 let Some(head) = b.head_op else {
3500 return Ok(Vec::new());
3501 };
3502 if Some(&head) == b.last_gate_checkpoint.as_ref() {
3503 // Steady-state common case: previous advance left the
3504 // checkpoint at head. Nothing to re-walk.
3505 return Ok(Vec::new());
3506 }
3507
3508 let log = lex_vcs::OpLog::open(self.root())?;
3509 let walk = log.walk_back(&head, None)?;
3510 let stop_at = b.last_gate_checkpoint.clone();
3511 let mut out = Vec::new();
3512 for rec in walk {
3513 if Some(&rec.op_id) == stop_at.as_ref() {
3514 break;
3515 }
3516 let stages = attestable_stage_ids(&rec.produces);
3517 let effects = op_declared_effects(&rec.op.kind);
3518 if stages.is_empty() {
3519 out.push((rec.op_id.clone(), None, effects));
3520 } else {
3521 for sid in stages {
3522 out.push((rec.op_id.clone(), Some(sid), effects.clone()));
3523 }
3524 }
3525 }
3526 Ok(out)
3527 }
3528}
3529
3530fn stage_name(stage: &Stage) -> &str {
3531 match stage {
3532 Stage::FnDecl(fd) => &fd.name,
3533 Stage::TypeDecl(td) => &td.name,
3534 Stage::Import(i) => &i.alias,
3535 }
3536}
3537
3538fn stage_for_kind<'a>(
3539 kind: &lex_vcs::OperationKind,
3540 stages: &'a [lex_ast::Stage],
3541) -> Option<&'a lex_ast::Stage> {
3542 use lex_vcs::OperationKind::*;
3543 let target_sig = match kind {
3544 AddFunction { sig_id, .. }
3545 | ModifyBody { sig_id, .. }
3546 | ChangeEffectSig { sig_id, .. }
3547 | AddType { sig_id, .. }
3548 | ModifyType { sig_id, .. } => Some(sig_id.clone()),
3549 RenameSymbol { to, .. } => Some(to.clone()),
3550 _ => None,
3551 };
3552 let target_sig = target_sig?;
3553 stages
3554 .iter()
3555 .find(|s| sig_id(s).as_deref() == Some(target_sig.as_str()))
3556}
3557
3558fn transition_for_kind(kind: &lex_vcs::OperationKind) -> lex_vcs::StageTransition {
3559 use lex_vcs::OperationKind::*;
3560 use lex_vcs::StageTransition;
3561 match kind {
3562 AddFunction {
3563 sig_id, stage_id, ..
3564 }
3565 | AddType { sig_id, stage_id, .. } => StageTransition::Create {
3566 sig_id: sig_id.clone(),
3567 stage_id: stage_id.clone(),
3568 },
3569 RemoveFunction {
3570 sig_id,
3571 last_stage_id,
3572 }
3573 | RemoveType {
3574 sig_id,
3575 last_stage_id,
3576 } => StageTransition::Remove {
3577 sig_id: sig_id.clone(),
3578 last: last_stage_id.clone(),
3579 },
3580 ModifyBody {
3581 sig_id,
3582 from_stage_id,
3583 to_stage_id,
3584 ..
3585 }
3586 | ChangeEffectSig {
3587 sig_id,
3588 from_stage_id,
3589 to_stage_id,
3590 ..
3591 }
3592 | ModifyType {
3593 sig_id,
3594 from_stage_id,
3595 to_stage_id,
3596 }
3597 | ReplaceMatchArm {
3598 sig_id,
3599 from_stage_id,
3600 to_stage_id,
3601 ..
3602 }
3603 | RenameLocal {
3604 sig_id,
3605 from_stage_id,
3606 to_stage_id,
3607 ..
3608 }
3609 | InlineLet {
3610 sig_id,
3611 from_stage_id,
3612 to_stage_id,
3613 ..
3614 } => StageTransition::Replace {
3615 sig_id: sig_id.clone(),
3616 from: from_stage_id.clone(),
3617 to: to_stage_id.clone(),
3618 },
3619 RenameSymbol {
3620 from,
3621 to,
3622 body_stage_id,
3623 } => StageTransition::Rename {
3624 from: from.clone(),
3625 to: to.clone(),
3626 body_stage_id: body_stage_id.clone(),
3627 },
3628 AddImport { .. } | RemoveImport { .. } => StageTransition::ImportOnly,
3629 Merge { .. } => StageTransition::Merge {
3630 entries: Default::default(),
3631 },
3632 // #294: a Candidate proposes a stage without advancing
3633 // the branch. ImportOnly keeps the branch head untouched
3634 // — the stage IS published on disk (Store::propose_candidate
3635 // calls publish before apply), but no head delta lands.
3636 Candidate { .. } => StageTransition::ImportOnly,
3637 // A Promote advances the head exactly like ModifyBody
3638 // (or Create when the sig had no head). The winner
3639 // stage is the new branch state for that sig.
3640 Promote {
3641 sig_id,
3642 winner_stage_id,
3643 from_stage_id,
3644 ..
3645 } => match from_stage_id {
3646 Some(from) => StageTransition::Replace {
3647 sig_id: sig_id.clone(),
3648 from: from.clone(),
3649 to: winner_stage_id.clone(),
3650 },
3651 None => StageTransition::Create {
3652 sig_id: sig_id.clone(),
3653 stage_id: winner_stage_id.clone(),
3654 },
3655 },
3656 }
3657}
3658
3659/// Producer identity for TypeCheck attestations emitted by the
3660/// store-write gate. Pinned to this crate's name + version so an
3661/// attestation produced by a different `lex-store` revision is
3662/// distinguishable (content-hashed `produced_by`).
3663fn typecheck_producer() -> lex_vcs::ProducerDescriptor {
3664 lex_vcs::ProducerDescriptor {
3665 tool: "lex-store".into(),
3666 version: env!("CARGO_PKG_VERSION").into(),
3667 model: None,
3668 }
3669}
3670
3671/// Producer for attestations the hosted CI runner writes (#93). A
3672/// distinct tool name so a `require-attestation` gate — via the
3673/// producer-trust model — can weight "the hub verified this
3674/// server-side" above a client-attached `TypeCheck`.
3675fn hub_ci_producer() -> lex_vcs::ProducerDescriptor {
3676 lex_vcs::ProducerDescriptor {
3677 tool: "lex-hub-ci".into(),
3678 version: env!("CARGO_PKG_VERSION").into(),
3679 model: None,
3680 }
3681}
3682
3683/// Verdict of a hosted-CI run over a branch head (#93).
3684#[derive(Debug, Clone, serde::Serialize)]
3685pub struct HubCiVerdict {
3686 pub passed: bool,
3687 pub checked_stages: usize,
3688 pub attested_stages: usize,
3689 #[serde(skip_serializing_if = "Option::is_none")]
3690 pub detail: Option<String>,
3691}
3692
3693/// Producer for the replay-comparison attestation (#836 G3). Distinct
3694/// tool name so the comparison lex performed is attributable
3695/// separately from the (external) regeneration.
3696fn replay_producer() -> lex_vcs::ProducerDescriptor {
3697 lex_vcs::ProducerDescriptor {
3698 tool: "lex-store-replay".into(),
3699 version: env!("CARGO_PKG_VERSION").into(),
3700 model: None,
3701 }
3702}
3703
3704/// Human/audit label for a recorded model: `provider/name` (`@version`
3705/// when pinned).
3706fn model_label(m: &lex_vcs::ModelDescriptor) -> String {
3707 match &m.version {
3708 Some(v) => format!("{}/{}@{}", m.provider, m.name, v),
3709 None => format!("{}/{}", m.provider, m.name),
3710 }
3711}
3712
3713/// The `(sig_id, stage_id)` an op recorded producing, or `None` for a
3714/// transition that produces no stage (removal / import / merge) — those
3715/// have nothing to regenerate for a replay.
3716fn produced_sig_stage(t: &lex_vcs::StageTransition) -> Option<(String, String)> {
3717 use lex_vcs::StageTransition::*;
3718 match t {
3719 Create { sig_id, stage_id } => Some((sig_id.clone(), stage_id.clone())),
3720 Replace { sig_id, to, .. } => Some((sig_id.clone(), to.clone())),
3721 Rename { to, body_stage_id, .. } => Some((to.clone(), body_stage_id.clone())),
3722 Remove { .. } | ImportOnly | Merge { .. } => None,
3723 }
3724}
3725
3726/// Producer identity for `Examples::Passed` attestations emitted by
3727/// [`Store::record_examples_passed`] (#835). Distinct tool name so
3728/// the activity feed can tell an auto-emitted publish-time examples
3729/// verdict apart from an `lex agent-tool --examples` one.
3730fn examples_producer() -> lex_vcs::ProducerDescriptor {
3731 lex_vcs::ProducerDescriptor {
3732 tool: "lex-store::examples".into(),
3733 version: env!("CARGO_PKG_VERSION").into(),
3734 model: None,
3735 }
3736}
3737
3738/// Producer identity for `Review` attestations (#836). The reviewer's
3739/// own id lives in the kind; this records which tool minted the record.
3740fn review_producer(reviewer: &str) -> lex_vcs::ProducerDescriptor {
3741 lex_vcs::ProducerDescriptor {
3742 tool: format!("lex-store::review:{reviewer}"),
3743 version: env!("CARGO_PKG_VERSION").into(),
3744 model: None,
3745 }
3746}
3747
3748/// Producer identity for `RepairHint` attestations emitted by
3749/// `apply_operation_checked` on TypeError (#281). Distinct tool
3750/// name from `typecheck_producer` so consumers can filter the
3751/// activity feed for repair hints without scanning kinds.
3752fn repair_hint_producer() -> lex_vcs::ProducerDescriptor {
3753 lex_vcs::ProducerDescriptor {
3754 tool: "lex-store::repair_hint".into(),
3755 version: env!("CARGO_PKG_VERSION").into(),
3756 model: None,
3757 }
3758}
3759
3760/// Producer identity for `TrustWaived` attestations emitted by
3761/// the `required_attestations` gate on a trust-driven waiver
3762/// (#293). Distinct from `typecheck_producer` and `repair_hint`
3763/// so the audit trail clearly shows "the gate let this advance
3764/// through because trust > threshold."
3765fn trust_waived_producer() -> lex_vcs::ProducerDescriptor {
3766 lex_vcs::ProducerDescriptor {
3767 tool: "lex-store::trust_waived".into(),
3768 version: env!("CARGO_PKG_VERSION").into(),
3769 model: None,
3770 }
3771}
3772
3773/// Producer identity for `ProducerTrust` attestations emitted by
3774/// [`Store::recompute_producer_trust`]. The score-derivation
3775/// recompute is its own machine-emittable kind, distinct from
3776/// the gate-side `TrustWaived` emit (#293).
3777fn producer_trust_producer() -> lex_vcs::ProducerDescriptor {
3778 lex_vcs::ProducerDescriptor {
3779 tool: "lex-store::producer_trust".into(),
3780 version: env!("CARGO_PKG_VERSION").into(),
3781 model: None,
3782 }
3783}
3784
3785/// The set of stage_ids a transition introduces. These are the
3786/// stages a successful TypeCheck pass attests *about* — the new
3787/// head produced by Create/Replace, the renamed body, or the per-
3788/// sig resolution of a Merge. Removes and ImportOnly produce no
3789/// attestable stage; the program typechecks but no specific stage
3790/// is the subject of the claim.
3791/// One row of input to the producer-block / required-attestations
3792/// gates: `(op_id, stage_id, op_effects)`. The `stage_id` is
3793/// `None` for ops that don't touch a stage (imports, empty
3794/// merges) — the gate skips those.
3795type GateCandidate = (
3796 lex_vcs::OpId,
3797 Option<String>,
3798 std::collections::BTreeSet<String>,
3799);
3800
3801/// Effect set declared *by the operation itself* (#245). Used by
3802/// the `required_attestations` gate's `EffectsIntersect` clause.
3803///
3804/// Only `AddFunction` and `ChangeEffectSig` carry an effect set in
3805/// their op payload; for everything else this returns the empty
3806/// set, which means `EffectsIntersect` rules don't fire on those
3807/// ops. `Always` rules continue to fire regardless. A future
3808/// improvement is to extract effects from the candidate `Stage`
3809/// for `ModifyBody` ops, but the typed-effects-on-ops path (#247)
3810/// is the cleaner solution and lands separately.
3811fn op_declared_effects(kind: &lex_vcs::OperationKind) -> std::collections::BTreeSet<String> {
3812 use lex_vcs::OperationKind::*;
3813 match kind {
3814 AddFunction { effects, .. } => effects.clone(),
3815 ChangeEffectSig { to_effects, .. } => to_effects.clone(),
3816 _ => std::collections::BTreeSet::new(),
3817 }
3818}
3819
3820fn attestable_stage_ids(transition: &lex_vcs::StageTransition) -> Vec<String> {
3821 use lex_vcs::StageTransition::*;
3822 match transition {
3823 Create { stage_id, .. } => vec![stage_id.clone()],
3824 Replace { to, .. } => vec![to.clone()],
3825 Rename { body_stage_id, .. } => vec![body_stage_id.clone()],
3826 Merge { entries } => entries.values().filter_map(|opt| opt.clone()).collect(),
3827 Remove { .. } | ImportOnly => Vec::new(),
3828 }
3829}
3830
3831/// True when two `FnDecl`s are identical except for their body — the
3832/// precondition for a pure intra-body three-way merge (#838). The
3833/// signature fields are equal by construction when both share a
3834/// `sig_id`; this also guards the non-signature fields (`type_params`,
3835/// `examples`) so a side that changed those isn't silently dropped.
3836fn fndecl_same_except_body(a: &lex_ast::FnDecl, b: &lex_ast::FnDecl) -> bool {
3837 a.name == b.name
3838 && a.type_params == b.type_params
3839 && a.params == b.params
3840 && a.effects == b.effects
3841 && a.effect_row_var == b.effect_row_var
3842 && a.return_type == b.return_type
3843 && a.examples == b.examples
3844}
3845
3846fn write_canonical_json<T: Serialize>(path: &Path, value: &T) -> Result<(), StoreError> {
3847 let v = serde_json::to_value(value)?;
3848 let s = lex_ast::canon_json::to_canonical_string(&v);
3849 if let Some(parent) = path.parent() {
3850 fs::create_dir_all(parent)?;
3851 }
3852 fs::write(path, s)?;
3853 Ok(())
3854}
3855
3856/// Read the `name` of the `Let` expression at `let_node` inside
3857/// `stage`'s body. Used by [`Store::apply_rename_local`] to record
3858/// the rename source. Returns the same `TransformError` shapes as
3859/// the transformer itself so callers see a consistent error
3860/// vocabulary.
3861fn read_let_name(
3862 stage: &Stage,
3863 let_node: &lex_ast::NodeId,
3864) -> Result<String, lex_ast::TransformError> {
3865 // The transformer is itself a pure function; ask it to perform
3866 // a rename to a sentinel value and read the resulting let's
3867 // original name from the output. Cheaper than duplicating the
3868 // node-walk here, and stays correct as the transform evolves.
3869 //
3870 // We use a sentinel that's invalid as a Lex identifier so even
3871 // if the rename somehow lands, downstream parsing would
3872 // surface it loudly. (The transform path discards the renamed
3873 // value — we only need the *original* name.)
3874 let probed = lex_ast::rename_local(stage, let_node, "__lex_rename_probe__")?;
3875 let Stage::FnDecl(fd) = probed else {
3876 return Err(lex_ast::TransformError::NonFnTarget {
3877 stage_kind: "non-FnDecl",
3878 });
3879 };
3880 // Walk back to the probed let to read its old name from the
3881 // *original* stage — the probed stage's let has already been
3882 // renamed.
3883 let Stage::FnDecl(orig_fd) = stage else {
3884 return Err(lex_ast::TransformError::NonFnTarget {
3885 stage_kind: "non-FnDecl",
3886 });
3887 };
3888 // Path-based lookup matches the transformer's navigation.
3889 let path = parse_let_node_path(let_node.as_str())?;
3890 if path.is_empty() {
3891 return Err(lex_ast::TransformError::NotALet {
3892 at: let_node.as_str().into(),
3893 found_kind: "stage_root",
3894 });
3895 }
3896 if path[0] != orig_fd.params.len() + 1 {
3897 return Err(lex_ast::TransformError::UnknownNode {
3898 at: let_node.as_str().into(),
3899 });
3900 }
3901 let inner = &path[1..];
3902 let target = navigate_to_let(&orig_fd.body, inner, let_node.as_str())?;
3903 let _ = fd; // probed stage discarded
3904 Ok(target.to_string())
3905}
3906
3907fn parse_let_node_path(id: &str) -> Result<Vec<usize>, lex_ast::TransformError> {
3908 let s = id
3909 .strip_prefix("n_")
3910 .ok_or_else(|| lex_ast::TransformError::BadNodeId(id.into()))?;
3911 let mut parts = s.split('.');
3912 let head = parts
3913 .next()
3914 .ok_or_else(|| lex_ast::TransformError::BadNodeId(id.into()))?;
3915 if head != "0" {
3916 return Err(lex_ast::TransformError::BadNodeId(id.into()));
3917 }
3918 let mut out = Vec::new();
3919 for p in parts {
3920 out.push(
3921 p.parse::<usize>()
3922 .map_err(|_| lex_ast::TransformError::BadNodeId(id.into()))?,
3923 );
3924 }
3925 Ok(out)
3926}
3927
3928fn navigate_to_let<'a>(
3929 root: &'a lex_ast::CExpr,
3930 path: &[usize],
3931 at: &str,
3932) -> Result<&'a str, lex_ast::TransformError> {
3933 use lex_ast::CExpr::*;
3934 let mut current = root;
3935 for &idx in path {
3936 current = match current {
3937 Call { callee, args } => {
3938 if idx == 0 {
3939 callee
3940 } else {
3941 args.get(idx - 1)
3942 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?
3943 }
3944 }
3945 Let { value, body, .. } => match idx {
3946 0 => value,
3947 1 => body,
3948 _ => return Err(lex_ast::TransformError::UnknownNode { at: at.into() }),
3949 },
3950 Match { scrutinee, arms } => {
3951 if idx == 0 {
3952 scrutinee
3953 } else {
3954 let arm_off = idx - 1;
3955 if arm_off % 2 != 1 {
3956 return Err(lex_ast::TransformError::UnknownNode { at: at.into() });
3957 }
3958 let arm_index = arm_off / 2;
3959 &arms
3960 .get(arm_index)
3961 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?
3962 .body
3963 }
3964 }
3965 Block { statements, result } => {
3966 if idx < statements.len() {
3967 &statements[idx]
3968 } else if idx == statements.len() {
3969 result
3970 } else {
3971 return Err(lex_ast::TransformError::UnknownNode { at: at.into() });
3972 }
3973 }
3974 Constructor { args, .. }
3975 | TupleLit { items: args, .. }
3976 | ListLit { items: args, .. } => args
3977 .get(idx)
3978 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?,
3979 RecordLit { fields } => {
3980 &fields
3981 .get(idx)
3982 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?
3983 .value
3984 }
3985 FieldAccess { value, .. } if idx == 0 => value,
3986 Lambda { body, .. } if idx == 0 => body,
3987 BinOp { lhs, rhs, .. } => match idx {
3988 0 => lhs,
3989 1 => rhs,
3990 _ => return Err(lex_ast::TransformError::UnknownNode { at: at.into() }),
3991 },
3992 UnaryOp { expr, .. } if idx == 0 => expr,
3993 Return { value } if idx == 0 => value,
3994 _ => return Err(lex_ast::TransformError::UnknownNode { at: at.into() }),
3995 };
3996 }
3997 let Let { name, .. } = current else {
3998 return Err(lex_ast::TransformError::NotALet {
3999 at: at.into(),
4000 found_kind: lex_cexpr_kind(current),
4001 });
4002 };
4003 Ok(name)
4004}
4005
4006fn lex_cexpr_kind(e: &lex_ast::CExpr) -> &'static str {
4007 use lex_ast::CExpr::*;
4008 match e {
4009 Literal { .. } => "Literal",
4010 Var { .. } => "Var",
4011 Call { .. } => "Call",
4012 Let { .. } => "Let",
4013 Match { .. } => "Match",
4014 Block { .. } => "Block",
4015 Constructor { .. } => "Constructor",
4016 RecordLit { .. } => "RecordLit",
4017 TupleLit { .. } => "TupleLit",
4018 ListLit { .. } => "ListLit",
4019 FieldAccess { .. } => "FieldAccess",
4020 Lambda { .. } => "Lambda",
4021 BinOp { .. } => "BinOp",
4022 UnaryOp { .. } => "UnaryOp",
4023 Return { .. } => "Return",
4024 }
4025}
4026
4027/// Extract the declared `[budget(N)]` integer from a stage's
4028/// effect set, if any (#280 + #247). Returns `None` for stages
4029/// that aren't `FnDecl` or don't carry a budget effect — same
4030/// shape as `lex_vcs::budget_from_effects`.
4031fn budget_of_stage(stage: &Stage) -> Option<u64> {
4032 let fd = match stage {
4033 Stage::FnDecl(fd) => fd,
4034 _ => return None,
4035 };
4036 let mut min_cost: Option<u64> = None;
4037 for eff in &fd.effects {
4038 if eff.name != "budget" {
4039 continue;
4040 }
4041 if let Some(lex_ast::EffectArg::Int { value }) = &eff.arg {
4042 let n = *value as u64;
4043 min_cost = Some(min_cost.map(|c| c.min(n)).unwrap_or(n));
4044 }
4045 }
4046 min_cost
4047}
4048
4049/// Serialize a stage to its canonical-JSON byte form. Used by
4050/// `publish_signed` for delta encoding (#261 slice 3) — both the
4051/// "compute the diff" path and the "write a full snapshot"
4052/// fallback need exactly the same bytes.
4053fn canonical_bytes(stage: &Stage) -> Result<Vec<u8>, StoreError> {
4054 let v = serde_json::to_value(stage)?;
4055 Ok(lex_ast::canon_json::to_canonical_string(&v).into_bytes())
4056}
4057
4058#[allow(dead_code)]
4059fn read_json<T: DeserializeOwned>(path: &Path) -> Result<T, StoreError> {
4060 let bytes = fs::read(path)?;
4061 Ok(serde_json::from_slice(&bytes)?)
4062}