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 /// The resolved dependencies' exported **type declarations**, keyed by the
260 /// same import reference as [`Self::resolve_modules`] (#930 completeness).
261 /// The write-time gate registers these under each import's alias so a head
262 /// that references a dependency's *type* — not just its functions — checks.
263 ///
264 /// Defaulted to empty so existing resolvers keep compiling and simply
265 /// resolve function signatures only (their previous behavior); a resolver
266 /// that can supply type declarations overrides this.
267 fn resolve_module_types(
268 &self,
269 _stages: &[Stage],
270 _head_op: Option<&str>,
271 ) -> BTreeMap<String, Vec<lex_ast::TypeDecl>> {
272 BTreeMap::new()
273 }
274
275 /// Each resolved dependency import's **module mangle prefix**, keyed by the
276 /// same import reference (#963). When a dependency is resolved as a whole
277 /// package, its exported types are named by their path-derived prefix
278 /// (`error_<hash>.DbErr`); this lets the write-time gate map the import
279 /// alias to that prefix so an alias-qualified type reference (`e.DbErr`)
280 /// resolves to the same type the dependency's own signatures name.
281 ///
282 /// Defaulted to empty: a resolver that returns bare (alias-qualified) type
283 /// names supplies none, and the gate keeps the #930 alias-qualification
284 /// path.
285 fn resolve_module_prefixes(
286 &self,
287 _stages: &[Stage],
288 _head_op: Option<&str>,
289 ) -> BTreeMap<String, String> {
290 BTreeMap::new()
291 }
292}
293
294pub struct Store {
295 root: PathBuf,
296 /// Optional dependency resolver (#930). Injected by the client or the hub;
297 /// `None` in a bare store means the gate resolves an empty module map.
298 resolver: Option<std::sync::Arc<dyn DepResolver>>,
299}
300
301impl Store {
302 /// Open or create a store rooted at `root`.
303 pub fn open(root: impl AsRef<Path>) -> Result<Self, StoreError> {
304 let root = root.as_ref().to_path_buf();
305 fs::create_dir_all(root.join("stages"))?;
306 fs::create_dir_all(root.join("traces"))?;
307 let store = Self { root, resolver: None };
308 store.ensure_stage_index();
309 Ok(store)
310 }
311
312 /// Install the dependency resolver the write-time gate uses to type-check
313 /// heads that keep external `import` edges (#930). Builder-style so a
314 /// caller can write `Store::open(p)?.with_dep_resolver(r)`.
315 pub fn with_dep_resolver(mut self, resolver: std::sync::Arc<dyn DepResolver>) -> Self {
316 self.resolver = Some(resolver);
317 self
318 }
319
320 /// Set the dependency resolver in place (for a `Store` already owned, e.g.
321 /// behind a `Mutex` in the HTTP `State`).
322 pub fn set_dep_resolver(&mut self, resolver: std::sync::Arc<dyn DepResolver>) {
323 self.resolver = Some(resolver);
324 }
325
326 /// The resolved dependency module map for a head being gated — the
327 /// installed resolver's answer, or empty when none is installed (#930).
328 fn resolved_modules(
329 &self,
330 stages: &[Stage],
331 head_op: Option<&str>,
332 ) -> BTreeMap<String, lex_types::Ty> {
333 match &self.resolver {
334 Some(r) => r.resolve_modules(stages, head_op),
335 None => BTreeMap::new(),
336 }
337 }
338
339 /// The resolved dependencies' exported type declarations for a head being
340 /// gated (#930 completeness) — the installed resolver's answer, or empty
341 /// when none is installed.
342 fn resolved_module_types(
343 &self,
344 stages: &[Stage],
345 head_op: Option<&str>,
346 ) -> BTreeMap<String, Vec<lex_ast::TypeDecl>> {
347 match &self.resolver {
348 Some(r) => r.resolve_module_types(stages, head_op),
349 None => BTreeMap::new(),
350 }
351 }
352
353 /// The resolved dependency import → module-prefix map for a head being
354 /// gated (#963) — the installed resolver's answer, or empty when none is
355 /// installed (then the gate keeps the #930 alias-qualified path).
356 fn resolved_module_prefixes(
357 &self,
358 stages: &[Stage],
359 head_op: Option<&str>,
360 ) -> BTreeMap<String, String> {
361 match &self.resolver {
362 Some(r) => r.resolve_module_prefixes(stages, head_op),
363 None => BTreeMap::new(),
364 }
365 }
366
367 /// One-time migration for a store that predates the reverse
368 /// index (#822), or whose previous rebuild pass never finished
369 /// (e.g. the process was killed or its client disconnected
370 /// mid-request — server-side work keeps running either way, but
371 /// a *restart* genuinely stops it): build the index in a single
372 /// pass instead of leaving every subsequent `lookup_lifecycle`
373 /// call to discover its own entry via the slow per-call scan-
374 /// and-backfill fallback.
375 ///
376 /// That per-call fallback is fine for the rare individual miss
377 /// it was designed for, but pathological as a *bulk* cold-start
378 /// strategy: on a tenant with a few thousand functions it means
379 /// redoing an O(total sigs) scan from scratch for *each* of a
380 /// few thousand cold entries — O(total sigs²) — which measured
381 /// as a near-stall (page-cache thrashing) on a memory-
382 /// constrained host. A single pass over `list_sigs()` is
383 /// O(total sigs) total.
384 ///
385 /// Gated on a dedicated completion marker
386 /// (`stage_index.complete`), NOT on `stage_index.jsonl`'s mere
387 /// existence — a partially-built index file (left behind by an
388 /// interrupted rebuild, lazy or bulk) must still trigger a
389 /// re-run so the remaining entries get backfilled in one more
390 /// cheap O(total sigs) pass, not silently be mistaken for
391 /// "already done" and fall back to the slow per-call path for
392 /// whatever's left. `rebuild_stage_index` already skips entries
393 /// it finds present, so re-running it against a partial index
394 /// only does the work that remains. The marker is written only
395 /// after a full pass returns `Ok`, so a failed pass (e.g. an I/O
396 /// error partway through `list_sigs`) is retried on the next
397 /// open rather than being marked done.
398 ///
399 /// Runs once per `Store::open` call — which, in a long-lived
400 /// server (lex-hub caches one `Store` per tenant for the life of
401 /// the process), means once per tenant per process lifetime, not
402 /// once per request. Once the marker exists (the steady state
403 /// after the first successful run on any given host) this is a
404 /// single cheap file-existence check. Best-effort like the rest
405 /// of the index: any failure here just leaves the slower per-call
406 /// fallback as the only path, never breaks correctness.
407 fn ensure_stage_index(&self) {
408 if self.stage_index_complete_marker_path().exists() {
409 return;
410 }
411 if self.rebuild_stage_index().is_ok() {
412 let _ = fs::write(self.stage_index_complete_marker_path(), "");
413 }
414 }
415
416 fn stage_index_complete_marker_path(&self) -> PathBuf {
417 self.root.join("stage_index.complete")
418 }
419
420 /// Build (or top up) the reverse index in one pass over every
421 /// SigId in the store, rather than relying on `lookup_lifecycle`
422 /// to discover entries one at a time. Safe to call at any time,
423 /// including on a partially-built index (e.g. one left behind by
424 /// an interrupted request that was populating it lazily): already-
425 /// indexed stage_ids are skipped, so this only does the work that
426 /// remains. Returns the number of newly-added entries.
427 pub fn rebuild_stage_index(&self) -> Result<usize, StoreError> {
428 // A sig's lifecycle can list the same stage_id more than once
429 // (Draft, then later Active, then Deprecated all carry the
430 // same stage_id with a different status) -- track newly-seen
431 // keys locally too, not just what was already on disk at the
432 // start, so a repeated stage_id within one sig's transitions
433 // doesn't get appended to the index more than once.
434 let mut existing = self.load_stage_index();
435 let mut added = 0usize;
436 for sig in self.list_sigs()? {
437 let Ok(life) = self.read_lifecycle(&sig) else { continue };
438 for t in &life.transitions {
439 if !existing.contains_key(&t.stage_id) {
440 self.append_stage_index_entry(&t.stage_id, &sig);
441 existing.insert(t.stage_id.clone(), sig.clone());
442 added += 1;
443 }
444 }
445 }
446 Ok(added)
447 }
448
449 pub fn root(&self) -> &Path {
450 &self.root
451 }
452
453 // ── Generic content-addressed blobs (#5 / M6.1) ──────────────────────────
454 //
455 // The stage store holds typed Lex ASTs; loom-style artifacts (generated
456 // code, JSON, prose) are opaque text. These blob methods give the store a
457 // generic content-addressed object alongside stages, plus a lightweight
458 // ref namespace so callers can bind names (e.g. a sprint's node ids) to
459 // blob shas without touching the operation-log branch machinery.
460 //
461 // The sha is the lowercase hex SHA-256 of the content's UTF-8 bytes —
462 // identical to Lex's `crypto.sha256_str`, so a blob written here and an
463 // artifact content-addressed in loom's SQLite store share the same id and
464 // are interchangeable by reference. Store-scoped, so under lex-hub each
465 // tenant store gets its own blob space for free.
466
467 fn blobs_dir(&self) -> PathBuf {
468 self.root.join("blobs")
469 }
470
471 fn blob_refs_dir(&self) -> PathBuf {
472 self.root.join("blobrefs")
473 }
474
475 /// Content-address `content` and persist it under `<root>/blobs/<sha>`.
476 /// Returns the sha. Idempotent: re-putting identical content is a no-op.
477 /// Concurrency-safe — writes to a unique temp file then atomically renames
478 /// onto the content-addressed path, so parallel writers of the same content
479 /// can't corrupt it.
480 pub fn put_blob(&self, content: &str) -> Result<String, StoreError> {
481 use sha2::{Digest, Sha256};
482 let sha = hex::encode(Sha256::digest(content.as_bytes()));
483 let dir = self.blobs_dir();
484 let path = dir.join(&sha);
485 if path.exists() {
486 return Ok(sha);
487 }
488 fs::create_dir_all(&dir)?;
489 static SEQ: std::sync::atomic::AtomicU64 = std::sync::atomic::AtomicU64::new(0);
490 let n = SEQ.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
491 let tmp = dir.join(format!(".{sha}.{}.{n}.tmp", std::process::id()));
492 fs::write(&tmp, content.as_bytes())?;
493 // rename is atomic on the same filesystem; identical content makes a
494 // last-writer-wins race harmless.
495 fs::rename(&tmp, &path)?;
496 Ok(sha)
497 }
498
499 /// Read a blob by its sha. `UnknownBlob` if absent.
500 pub fn get_blob(&self, sha: &str) -> Result<String, StoreError> {
501 match fs::read_to_string(self.blobs_dir().join(sha)) {
502 Ok(s) => Ok(s),
503 Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
504 Err(StoreError::UnknownBlob(sha.to_string()))
505 }
506 Err(e) => Err(StoreError::Io(e)),
507 }
508 }
509
510 /// Whether a blob with this sha exists.
511 pub fn has_blob(&self, sha: &str) -> bool {
512 self.blobs_dir().join(sha).exists()
513 }
514
515 /// Bind `key` to a blob `sha` within `namespace` (e.g. namespace
516 /// `"loom/sprint-abc"`, key `"build-node"`). Overwrites an existing
517 /// binding. The namespace may contain `/`; neither namespace nor key may
518 /// contain a `..` path component.
519 pub fn set_blob_ref(&self, namespace: &str, key: &str, sha: &str) -> Result<(), StoreError> {
520 let dir = self.blob_ref_namespace_dir(namespace, key)?;
521 fs::create_dir_all(&dir)?;
522 fs::write(dir.join(key), sha.as_bytes())?;
523 Ok(())
524 }
525
526 /// Resolve `namespace`/`key` to a blob sha. `UnknownBlobRef` if unbound.
527 pub fn get_blob_ref(&self, namespace: &str, key: &str) -> Result<String, StoreError> {
528 let dir = self.blob_ref_namespace_dir(namespace, key)?;
529 match fs::read_to_string(dir.join(key)) {
530 Ok(s) => Ok(s.trim().to_string()),
531 Err(e) if e.kind() == std::io::ErrorKind::NotFound => Err(StoreError::UnknownBlobRef {
532 namespace: namespace.to_string(),
533 key: key.to_string(),
534 }),
535 Err(e) => Err(StoreError::Io(e)),
536 }
537 }
538
539 /// Blob-ref namespace for committed lockfiles (#930 phase 2b-1).
540 const LOCK_NS: &'static str = "lock";
541
542 /// Record the `lex.lock` committed with the package head `head_op` — the
543 /// exact dependency versions and op-log heads that head was built and
544 /// type-checks against (#930 phase 2b-1: "HEAD + its committed lex.lock
545 /// always type-checks"). Content-addressed via [`Self::put_blob`] and
546 /// bound under the `lock` namespace keyed by the head op, so it is
547 /// idempotent (a re-push converges) and travels with the package through
548 /// the same object-sync path as stages and intents. Keyed by head op
549 /// rather than by branch so re-verifying a *historical* head resolves it
550 /// against the lock that head actually committed, not whatever the branch
551 /// points at now.
552 pub fn set_committed_lock(&self, head_op: &str, lock_toml: &str) -> Result<(), StoreError> {
553 let sha = self.put_blob(lock_toml)?;
554 self.set_blob_ref(Self::LOCK_NS, head_op, &sha)
555 }
556
557 /// The `lex.lock` committed with `head_op`, or `None` when the head
558 /// carries no committed lock — a dependency-free package, or one
559 /// published before locks were committed (the write-time gate then has no
560 /// registry/git dependencies to resolve, exactly as today).
561 pub fn committed_lock(&self, head_op: &str) -> Result<Option<String>, StoreError> {
562 match self.get_blob_ref(Self::LOCK_NS, head_op) {
563 Ok(sha) => Ok(Some(self.get_blob(&sha)?)),
564 Err(StoreError::UnknownBlobRef { .. }) => Ok(None),
565 Err(e) => Err(e),
566 }
567 }
568
569 /// All `key → sha` bindings in a namespace (e.g. every artifact in a
570 /// sprint). Empty map if the namespace has no bindings yet.
571 pub fn list_blob_refs(
572 &self,
573 namespace: &str,
574 ) -> Result<std::collections::BTreeMap<String, String>, StoreError> {
575 let dir = self.blob_ref_namespace_dir(namespace, "x")?;
576 let mut out = std::collections::BTreeMap::new();
577 let entries = match fs::read_dir(&dir) {
578 Ok(e) => e,
579 Err(e) if e.kind() == std::io::ErrorKind::NotFound => return Ok(out),
580 Err(e) => return Err(StoreError::Io(e)),
581 };
582 for entry in entries {
583 let entry = entry?;
584 if entry.file_type()?.is_file() {
585 let key = entry.file_name().to_string_lossy().to_string();
586 let sha = fs::read_to_string(entry.path())?.trim().to_string();
587 out.insert(key, sha);
588 }
589 }
590 Ok(out)
591 }
592
593 // Resolve the on-disk dir for a (namespace, key), rejecting `..` traversal
594 // and `/` in the key. `key` is validated but not joined here (callers join
595 // it themselves so `list_blob_refs` can pass a dummy).
596 fn blob_ref_namespace_dir(&self, namespace: &str, key: &str) -> Result<PathBuf, StoreError> {
597 if key.contains('/') || key.contains('\\') || key.split('/').any(|c| c == "..") {
598 return Err(StoreError::UnknownBlobRef {
599 namespace: namespace.to_string(),
600 key: key.to_string(),
601 });
602 }
603 let mut dir = self.blob_refs_dir();
604 for comp in namespace.split('/') {
605 if comp == ".." || comp.contains('\\') {
606 return Err(StoreError::UnknownBlobRef {
607 namespace: namespace.to_string(),
608 key: key.to_string(),
609 });
610 }
611 if !comp.is_empty() {
612 dir.push(comp);
613 }
614 }
615 Ok(dir)
616 }
617
618 fn now() -> u64 {
619 SystemTime::now()
620 .duration_since(UNIX_EPOCH)
621 .map(|d| d.as_secs())
622 .unwrap_or(0)
623 }
624
625 fn sig_dir(&self, sig: &str) -> PathBuf {
626 self.root.join("stages").join(sig)
627 }
628 fn impl_dir(&self, sig: &str) -> PathBuf {
629 self.sig_dir(sig).join("implementations")
630 }
631 fn tests_dir(&self, sig: &str) -> PathBuf {
632 self.sig_dir(sig).join("tests")
633 }
634 fn specs_dir(&self, sig: &str) -> PathBuf {
635 self.sig_dir(sig).join("specs")
636 }
637 fn lifecycle_path(&self, sig: &str) -> PathBuf {
638 self.sig_dir(sig).join("lifecycle.json")
639 }
640
641 // ---- publish ----
642
643 /// Publish a stage as **Draft**. Returns the StageId.
644 /// Idempotent: republishing the same canonical AST returns the same
645 /// StageId without writing duplicates.
646 pub fn publish(&self, stage: &Stage) -> Result<String, StoreError> {
647 self.publish_signed(stage, None)
648 }
649
650 /// Like [`Self::publish`] but optionally attaches an Ed25519
651 /// signature over the StageId (#227). When `signer` is `Some`,
652 /// the persisted metadata gets a `signature` field that
653 /// downstream consumers can verify via
654 /// [`lex_vcs::verify_stage_id`].
655 ///
656 /// Idempotency: if a metadata file already exists the signature
657 /// is *not* re-written. This preserves "republishing is a no-op"
658 /// even across different signers — promoting a signed stage
659 /// requires a fresh stage hash anyway, so a metadata overwrite
660 /// would be the wrong primitive.
661 pub fn publish_signed(
662 &self,
663 stage: &Stage,
664 signer: Option<&lex_vcs::Keypair>,
665 ) -> Result<String, StoreError> {
666 let sig = sig_id(stage).ok_or(StoreError::CannotPublishImport)?;
667 let stage_id = stage_id(stage).ok_or(StoreError::CannotPublishImport)?;
668 let name = stage_name(stage).to_string();
669
670 fs::create_dir_all(self.impl_dir(&sig))?;
671 fs::create_dir_all(self.tests_dir(&sig))?;
672 fs::create_dir_all(self.specs_dir(&sig))?;
673
674 let ast_path = self.impl_dir(&sig).join(format!("{}.ast.json", stage_id));
675 let delta_path = self.impl_dir(&sig).join(format!("{}.delta.json", stage_id));
676 let meta_path = self
677 .impl_dir(&sig)
678 .join(format!("{}.metadata.json", stage_id));
679
680 // #261 slice 3: try delta encoding against the most recent
681 // prior stage in this sig's lifecycle. Falls back to a full
682 // snapshot when (a) no prior stage exists, (b) the diff
683 // ratio is over the threshold, or (c) the delta chain is
684 // already at its cap. The decision is internal — callers
685 // see the same `Stage` object on `get_ast` regardless.
686 if !ast_path.exists() && !delta_path.exists() {
687 self.persist_stage_bytes(&sig, &stage_id, stage, &ast_path, &delta_path)?;
688 }
689 if !meta_path.exists() {
690 let signature = signer.map(|kp| kp.sign_stage_id(&stage_id));
691 let metadata = Metadata {
692 stage_id: stage_id.clone(),
693 sig_id: sig.clone(),
694 name,
695 published_at: Self::now(),
696 note: None,
697 signature,
698 };
699 write_canonical_json(&meta_path, &metadata)?;
700 }
701
702 // Lifecycle: append a Draft transition for first publish.
703 let mut life = self.read_lifecycle(&sig).unwrap_or_else(|_| Lifecycle {
704 sig_id: sig.clone(),
705 ..Default::default()
706 });
707 if !life.transitions.iter().any(|t| t.stage_id == stage_id) {
708 life.transitions.push(Transition {
709 stage_id: stage_id.clone(),
710 from: StageStatus::Draft, // synthesized; "from" of first transition is itself
711 to: StageStatus::Draft,
712 at: Self::now(),
713 reason: None,
714 });
715 self.write_lifecycle(&sig, &life)?;
716 // Register the new stage_id's owning sig up front so a
717 // later `lookup_lifecycle` (e.g. `get_ast`) never needs
718 // to fall back to a full tenant-wide scan for it.
719 self.append_stage_index_entry(&stage_id, &sig);
720 }
721 Ok(stage_id)
722 }
723
724 // ---- lifecycle ----
725
726 pub fn activate(&self, stage_id: &str) -> Result<(), StoreError> {
727 let (sig, mut life) = self.lookup_lifecycle(stage_id)?;
728 // Demote any currently-Active impls for this SigId to Deprecated.
729 let active = life.current_active().map(|s| s.to_string());
730 if let Some(prev) = active {
731 if prev != stage_id {
732 life.transitions.push(Transition {
733 stage_id: prev,
734 from: StageStatus::Active,
735 to: StageStatus::Deprecated,
736 at: Self::now(),
737 reason: Some("superseded".into()),
738 });
739 }
740 }
741 let cur = life.status_of(stage_id);
742 if cur == Some(StageStatus::Tombstone) {
743 return Err(StoreError::InvalidTransition(
744 "tombstoned cannot be activated".into(),
745 ));
746 }
747 life.transitions.push(Transition {
748 stage_id: stage_id.into(),
749 from: cur.unwrap_or(StageStatus::Draft),
750 to: StageStatus::Active,
751 at: Self::now(),
752 reason: None,
753 });
754 self.write_lifecycle(&sig, &life)
755 }
756
757 pub fn deprecate(&self, stage_id: &str, reason: impl Into<String>) -> Result<(), StoreError> {
758 let (sig, mut life) = self.lookup_lifecycle(stage_id)?;
759 let cur = life
760 .status_of(stage_id)
761 .ok_or_else(|| StoreError::UnknownStage(stage_id.into()))?;
762 if cur != StageStatus::Active {
763 return Err(StoreError::InvalidTransition(format!(
764 "{cur:?} ⇒ Deprecated"
765 )));
766 }
767 life.transitions.push(Transition {
768 stage_id: stage_id.into(),
769 from: cur,
770 to: StageStatus::Deprecated,
771 at: Self::now(),
772 reason: Some(reason.into()),
773 });
774 self.write_lifecycle(&sig, &life)
775 }
776
777 pub fn tombstone(&self, stage_id: &str) -> Result<(), StoreError> {
778 let (sig, mut life) = self.lookup_lifecycle(stage_id)?;
779 let cur = life
780 .status_of(stage_id)
781 .ok_or_else(|| StoreError::UnknownStage(stage_id.into()))?;
782 if cur != StageStatus::Deprecated {
783 return Err(StoreError::InvalidTransition(format!(
784 "{cur:?} ⇒ Tombstone"
785 )));
786 }
787 life.transitions.push(Transition {
788 stage_id: stage_id.into(),
789 from: cur,
790 to: StageStatus::Tombstone,
791 at: Self::now(),
792 reason: None,
793 });
794 self.write_lifecycle(&sig, &life)
795 }
796
797 // ---- queries ----
798
799 /// The current Active StageId for a signature, or `None`.
800 pub fn resolve_sig(&self, sig: &str) -> Result<Option<String>, StoreError> {
801 let life = match self.read_lifecycle(sig) {
802 Ok(l) => l,
803 Err(_) => return Ok(None),
804 };
805 Ok(life.current_active().map(|s| s.to_string()))
806 }
807
808 /// Per-stage history for a SigId, ordered chronologically by
809 /// the *last* transition timestamp. Returns one entry per
810 /// distinct StageId that has ever been published under `sig`.
811 /// `Ok(vec![])` if the SigId doesn't exist in the store.
812 ///
813 /// Used by `lex blame` to render "where does this fn come from".
814 pub fn sig_history(&self, sig: &str) -> Result<Vec<StageHistoryEntry>, StoreError> {
815 let life = match self.read_lifecycle(sig) {
816 Ok(l) => l,
817 Err(_) => return Ok(Vec::new()),
818 };
819 // Collapse transitions: latest status + last_at per stage,
820 // plus the timestamp of the first Draft transition (≈ when
821 // the stage was published) when one exists.
822 let mut by_stage: indexmap::IndexMap<String, StageHistoryEntry> = indexmap::IndexMap::new();
823 for t in &life.transitions {
824 let entry = by_stage
825 .entry(t.stage_id.clone())
826 .or_insert(StageHistoryEntry {
827 stage_id: t.stage_id.clone(),
828 status: t.to,
829 last_at: t.at,
830 published_at: None,
831 });
832 entry.status = t.to;
833 entry.last_at = t.at;
834 if t.from == StageStatus::Draft && entry.published_at.is_none() {
835 entry.published_at = Some(t.at);
836 }
837 if t.to == StageStatus::Draft && entry.published_at.is_none() {
838 // Initial publication: Draft is the *destination*.
839 entry.published_at = Some(t.at);
840 }
841 }
842 let mut out: Vec<StageHistoryEntry> = by_stage.into_values().collect();
843 // Sort newest first so `lex blame` shows recent activity at top.
844 out.sort_by_key(|e| std::cmp::Reverse(e.last_at));
845 Ok(out)
846 }
847
848 pub fn get_ast(&self, stage_id: &str) -> Result<Stage, StoreError> {
849 let (sig, _) = self.lookup_lifecycle(stage_id)?;
850 let bytes = self.read_stage_canonical_bytes(&sig, stage_id)?;
851 Ok(serde_json::from_slice(&bytes)?)
852 }
853
854 /// Bulk AST fetch for callers that already know each stage's
855 /// **signature** — a branch head map, for instance, which is keyed
856 /// by SigId and whose values are the StageIds it points at.
857 ///
858 /// Prefer this over [`Self::get_asts_bulk`] whenever the SigId is in
859 /// hand, because resolving a StageId back to a SigId is not
860 /// reliable: a StageId hashes the structural signature plus the
861 /// implementation, deliberately *not* the name
862 /// (`docs/INVARIANTS.md`), so two functions that differ only in name
863 /// share one StageId while having two distinct SigIds — and two
864 /// separate ASTs, one under each sig directory. `stage_index` maps
865 /// each StageId to a single sig, so `get_ast`/`get_asts_bulk` return
866 /// whichever of those ASTs the index happens to name, i.e. the wrong
867 /// name half the time (#826). Reading straight from the sig the
868 /// caller already knows removes the ambiguity — and skips loading
869 /// the index at all.
870 ///
871 /// Returns results in the same order as `pairs`, `Err` for anything
872 /// that fails to resolve (mirroring `get_ast`'s error semantics).
873 pub fn get_asts_for_sigs_bulk(
874 &self,
875 pairs: &[(String, String)],
876 ) -> Vec<Result<Stage, StoreError>> {
877 pairs
878 .iter()
879 .map(|(sig_id, stage_id)| {
880 let bytes = self.read_stage_canonical_bytes(sig_id, stage_id)?;
881 Ok(serde_json::from_slice(&bytes)?)
882 })
883 .collect()
884 }
885
886 /// Bulk variant of [`Self::get_ast`] for callers resolving many
887 /// stage_ids at once (e.g. `pkg_publish_handler`'s `old_head`
888 /// scan over every live function in a tenant, once per publish
889 /// request). `get_ast` in a loop calls `lookup_lifecycle` once
890 /// per stage_id, and `lookup_lifecycle`'s index-hit path reads
891 /// and re-parses the *entire* `stage_index.jsonl` on every single
892 /// call — fine for one call, but O(index size × N) for N calls in
893 /// a row, which dominates once the index itself is large (#825's
894 /// follow-up: still correct and far better than the pre-index
895 /// full-tenant-scan-per-call behavior, but the per-call reparse
896 /// is itself a real, measured cost — 87.6s for 3,664 calls against
897 /// a ~14k-line index on the alpibrusl tenant).
898 ///
899 /// This loads the index once for the whole batch and keeps it in
900 /// memory across all `stage_ids`, only touching disk again to
901 /// append genuinely new entries (a positive backfill or a
902 /// negative "not found anywhere" cache, same as the single-call
903 /// path) — never to re-read what's already loaded.
904 ///
905 /// Returns results in the same order as `stage_ids`, `Err` for
906 /// anything that fails to resolve (mirroring `get_ast`'s error
907 /// semantics per call).
908 pub fn get_asts_bulk(&self, stage_ids: &[String]) -> Vec<Result<Stage, StoreError>> {
909 let mut index = self.load_stage_index();
910 let mut sigs_cache: BTreeMap<String, Option<Lifecycle>> = BTreeMap::new();
911 let mut all_sigs: Option<Vec<String>> = None;
912
913 stage_ids
914 .iter()
915 .map(|stage_id| {
916 self.lookup_lifecycle_bulk(stage_id, &mut index, &mut sigs_cache, &mut all_sigs)
917 .and_then(|sig| {
918 let bytes = self.read_stage_canonical_bytes(&sig, stage_id)?;
919 Ok(serde_json::from_slice(&bytes)?)
920 })
921 })
922 .collect()
923 }
924
925 /// Shared implementation behind [`Self::get_asts_bulk`]: identical
926 /// logic to [`Self::lookup_lifecycle`], but reads and writes the
927 /// caller-supplied `index` map instead of reloading it from disk
928 /// on every call, and memoizes `read_lifecycle` per sig and the
929 /// `list_sigs()` full-scan list across the whole batch. Disk
930 /// writes for newly-discovered entries (positive or negative)
931 /// still happen immediately, same as the single-call path — only
932 /// the repeated *reads* are batched away.
933 fn lookup_lifecycle_bulk(
934 &self,
935 stage_id: &str,
936 index: &mut BTreeMap<String, String>,
937 sigs_cache: &mut BTreeMap<String, Option<Lifecycle>>,
938 all_sigs: &mut Option<Vec<String>>,
939 ) -> Result<String, StoreError> {
940 if let Some(sig) = index.get(stage_id) {
941 if sig == MISSING_STAGE_MARKER {
942 return Err(StoreError::UnknownStage(stage_id.into()));
943 }
944 let life = sigs_cache
945 .entry(sig.clone())
946 .or_insert_with(|| self.read_lifecycle(sig).ok());
947 if let Some(life) = life {
948 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
949 return Ok(sig.clone());
950 }
951 }
952 }
953 if all_sigs.is_none() {
954 *all_sigs = Some(self.list_sigs()?);
955 }
956 for sig in all_sigs.as_ref().unwrap() {
957 let life = sigs_cache
958 .entry(sig.clone())
959 .or_insert_with(|| self.read_lifecycle(sig).ok());
960 if let Some(life) = life {
961 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
962 self.append_stage_index_entry(stage_id, sig);
963 index.insert(stage_id.to_string(), sig.clone());
964 return Ok(sig.clone());
965 }
966 }
967 }
968 self.append_stage_index_entry(stage_id, MISSING_STAGE_MARKER);
969 index.insert(stage_id.to_string(), MISSING_STAGE_MARKER.to_string());
970 Err(StoreError::UnknownStage(stage_id.into()))
971 }
972
973 /// Read the canonical bytes of a stage, walking back through
974 /// any delta chain (#261 slice 3). The recursion ends at a
975 /// `<stage_id>.ast.json` file (a full snapshot) or, in the
976 /// degenerate case of a missing chain, with `UnknownStage`.
977 fn read_stage_canonical_bytes(&self, sig: &str, stage_id: &str) -> Result<Vec<u8>, StoreError> {
978 let ast_path = self.impl_dir(sig).join(format!("{}.ast.json", stage_id));
979 if ast_path.exists() {
980 return Ok(fs::read(&ast_path)?);
981 }
982 let delta_path = self.impl_dir(sig).join(format!("{}.delta.json", stage_id));
983 if !delta_path.exists() {
984 return Err(StoreError::UnknownStage(stage_id.into()));
985 }
986 let delta_bytes = fs::read(&delta_path)?;
987 let delta: crate::delta::StageDelta = serde_json::from_slice(&delta_bytes)?;
988 let base_bytes = self.read_stage_canonical_bytes(sig, &delta.base_stage_id)?;
989 crate::delta::apply(&base_bytes, &delta).map_err(|e| {
990 StoreError::Io(std::io::Error::new(
991 std::io::ErrorKind::InvalidData,
992 format!("applying delta for {stage_id}: {e}"),
993 ))
994 })
995 }
996
997 /// Persist a freshly-published stage's canonical bytes (#261
998 /// slice 3). Tries delta encoding against the most recent
999 /// prior stage in the sig's lifecycle; falls back to a full
1000 /// snapshot when no base exists, the diff ratio is too high,
1001 /// or the delta chain is already at its cap.
1002 fn persist_stage_bytes(
1003 &self,
1004 sig: &str,
1005 stage_id: &str,
1006 stage: &Stage,
1007 ast_path: &Path,
1008 delta_path: &Path,
1009 ) -> Result<(), StoreError> {
1010 let new_bytes = canonical_bytes(stage)?;
1011 if let Some((base_stage_id, base_chain_length)) = self.pick_delta_base(sig, stage_id)? {
1012 let base_bytes = self.read_stage_canonical_bytes(sig, &base_stage_id)?;
1013 let (prefix, suffix, middle) = crate::delta::splice(&base_bytes, &new_bytes);
1014 let chain_length = base_chain_length + 1;
1015 if crate::delta::is_worth_encoding(middle.len(), new_bytes.len(), chain_length) {
1016 let delta = crate::delta::StageDelta {
1017 base_stage_id,
1018 chain_length,
1019 common_prefix: prefix,
1020 common_suffix: suffix,
1021 middle_hex: hex::encode(&middle),
1022 };
1023 write_canonical_json(delta_path, &delta)?;
1024 return Ok(());
1025 }
1026 }
1027 // Fall through: full snapshot.
1028 if let Some(parent) = ast_path.parent() {
1029 fs::create_dir_all(parent)?;
1030 }
1031 fs::write(ast_path, &new_bytes)?;
1032 Ok(())
1033 }
1034
1035 /// Pick a base stage for delta encoding from the given sig's
1036 /// lifecycle. Returns `(base_stage_id, base_chain_length)` for
1037 /// the most-recent non-tombstoned prior stage, or `None` when
1038 /// there is no candidate. The chain length is read off the
1039 /// base's `.delta.json` (if any) to enforce the cap.
1040 fn pick_delta_base(
1041 &self,
1042 sig: &str,
1043 new_stage_id: &str,
1044 ) -> Result<Option<(String, usize)>, StoreError> {
1045 let life = self.read_lifecycle(sig).ok();
1046 let Some(life) = life else {
1047 return Ok(None);
1048 };
1049 // Walk transitions newest-first; pick the first prior
1050 // stage that isn't this one and isn't tombstoned.
1051 let mut latest_per_stage: indexmap::IndexMap<&str, StageStatus> = indexmap::IndexMap::new();
1052 for t in &life.transitions {
1053 latest_per_stage.insert(&t.stage_id, t.to);
1054 }
1055 let mut candidates: Vec<&str> = latest_per_stage
1056 .iter()
1057 .filter(|(id, status)| **id != new_stage_id && **status != StageStatus::Tombstone)
1058 .map(|(id, _)| *id)
1059 .collect();
1060 // Reverse to get newest-first (transitions are append-only,
1061 // so latest_per_stage's iteration order matches insertion
1062 // order, oldest-first).
1063 candidates.reverse();
1064 let Some(&base) = candidates.first() else {
1065 return Ok(None);
1066 };
1067 let base_chain_length = self.delta_chain_length(sig, base)?;
1068 Ok(Some((base.to_string(), base_chain_length)))
1069 }
1070
1071 /// Length of the delta chain ending at `stage_id`. Zero when
1072 /// the stage is a full snapshot (`.ast.json` present); the
1073 /// stored `chain_length` from `.delta.json` otherwise.
1074 fn delta_chain_length(&self, sig: &str, stage_id: &str) -> Result<usize, StoreError> {
1075 let ast_path = self.impl_dir(sig).join(format!("{}.ast.json", stage_id));
1076 if ast_path.exists() {
1077 return Ok(0);
1078 }
1079 let delta_path = self.impl_dir(sig).join(format!("{}.delta.json", stage_id));
1080 if !delta_path.exists() {
1081 return Ok(0);
1082 }
1083 let bytes = fs::read(&delta_path)?;
1084 let delta: crate::delta::StageDelta = serde_json::from_slice(&bytes)?;
1085 Ok(delta.chain_length)
1086 }
1087
1088 pub fn get_metadata(&self, stage_id: &str) -> Result<Metadata, StoreError> {
1089 let (sig, _) = self.lookup_lifecycle(stage_id)?;
1090 let path = self
1091 .impl_dir(&sig)
1092 .join(format!("{}.metadata.json", stage_id));
1093 let bytes = fs::read(&path)?;
1094 Ok(serde_json::from_slice(&bytes)?)
1095 }
1096
1097 pub fn get_status(&self, stage_id: &str) -> Result<StageStatus, StoreError> {
1098 let (_sig, life) = self.lookup_lifecycle(stage_id)?;
1099 life.status_of(stage_id)
1100 .ok_or_else(|| StoreError::UnknownStage(stage_id.into()))
1101 }
1102
1103 pub fn list_stages_by_name(&self, name: &str) -> Result<Vec<String>, StoreError> {
1104 // Walk every SigId → check metadata of any implementation; if its
1105 // name matches, include the SigId.
1106 let mut out = Vec::new();
1107 let stages_dir = self.root.join("stages");
1108 if !stages_dir.exists() {
1109 return Ok(out);
1110 }
1111 for entry in fs::read_dir(&stages_dir)? {
1112 let entry = entry?;
1113 let sig_dir = entry.path();
1114 if !sig_dir.is_dir() {
1115 continue;
1116 }
1117 let sig = entry.file_name().to_string_lossy().to_string();
1118 // Look at any one metadata file under this SigId.
1119 let impls = self.impl_dir(&sig);
1120 if !impls.exists() {
1121 continue;
1122 }
1123 for f in fs::read_dir(impls)? {
1124 let f = f?;
1125 let p = f.path();
1126 if p.extension().is_some_and(|e| e == "json")
1127 && p.file_name()
1128 .is_some_and(|n| n.to_string_lossy().ends_with(".metadata.json"))
1129 {
1130 if let Ok(bytes) = fs::read(&p) {
1131 if let Ok(m) = serde_json::from_slice::<Metadata>(&bytes) {
1132 if m.name == name {
1133 if !out.contains(&sig) {
1134 out.push(sig.clone());
1135 }
1136 break;
1137 }
1138 }
1139 }
1140 }
1141 }
1142 }
1143 out.sort();
1144 Ok(out)
1145 }
1146
1147 pub fn list_sigs(&self) -> Result<Vec<String>, StoreError> {
1148 let stages_dir = self.root.join("stages");
1149 let mut out = Vec::new();
1150 if !stages_dir.exists() {
1151 return Ok(out);
1152 }
1153 for entry in fs::read_dir(stages_dir)? {
1154 let entry = entry?;
1155 if entry.file_type()?.is_dir() {
1156 out.push(entry.file_name().to_string_lossy().to_string());
1157 }
1158 }
1159 out.sort();
1160 Ok(out)
1161 }
1162
1163 // ---- tests/specs as metadata (§4.4) ----
1164
1165 pub fn attach_test(&self, sig: &str, test: &Test) -> Result<String, StoreError> {
1166 if !self.sig_dir(sig).exists() {
1167 return Err(StoreError::UnknownSig(sig.into()));
1168 }
1169 fs::create_dir_all(self.tests_dir(sig))?;
1170 let path = self.tests_dir(sig).join(format!("{}.json", test.id));
1171 write_canonical_json(&path, test)?;
1172 Ok(test.id.clone())
1173 }
1174
1175 pub fn list_tests(&self, sig: &str) -> Result<Vec<Test>, StoreError> {
1176 let dir = self.tests_dir(sig);
1177 if !dir.exists() {
1178 return Ok(Vec::new());
1179 }
1180 let mut out = Vec::new();
1181 for f in fs::read_dir(dir)? {
1182 let f = f?;
1183 if f.path().extension().is_some_and(|e| e == "json") {
1184 let bytes = fs::read(f.path())?;
1185 out.push(serde_json::from_slice(&bytes)?);
1186 }
1187 }
1188 Ok(out)
1189 }
1190
1191 pub fn attach_spec(&self, sig: &str, spec: &Spec) -> Result<String, StoreError> {
1192 if !self.sig_dir(sig).exists() {
1193 return Err(StoreError::UnknownSig(sig.into()));
1194 }
1195 fs::create_dir_all(self.specs_dir(sig))?;
1196 let path = self.specs_dir(sig).join(format!("{}.json", spec.id));
1197 write_canonical_json(&path, spec)?;
1198 Ok(spec.id.clone())
1199 }
1200
1201 pub fn list_specs(&self, sig: &str) -> Result<Vec<Spec>, StoreError> {
1202 let dir = self.specs_dir(sig);
1203 if !dir.exists() {
1204 return Ok(Vec::new());
1205 }
1206 let mut out = Vec::new();
1207 for f in fs::read_dir(dir)? {
1208 let f = f?;
1209 if f.path().extension().is_some_and(|e| e == "json") {
1210 let bytes = fs::read(f.path())?;
1211 out.push(serde_json::from_slice(&bytes)?);
1212 }
1213 }
1214 Ok(out)
1215 }
1216
1217 // ---- traces (§4.2 / M7) ----
1218
1219 // Native run-trace store — gated behind the `trace` feature (depends on
1220 // lex-trace). Off when a lower crate (lex-runtime) depends on lex-store to
1221 // avoid a dependency cycle; the blob/stage store below is unaffected.
1222 #[cfg(feature = "trace")]
1223 fn trace_path(&self, run_id: &str) -> PathBuf {
1224 self.root.join("traces").join(run_id).join("trace.json")
1225 }
1226
1227 #[cfg(feature = "trace")]
1228 pub fn save_trace(&self, tree: &lex_trace::TraceTree) -> Result<String, StoreError> {
1229 let path = self.trace_path(&tree.run_id);
1230 write_canonical_json(&path, tree)?;
1231 Ok(tree.run_id.clone())
1232 }
1233
1234 #[cfg(feature = "trace")]
1235 pub fn load_trace(&self, run_id: &str) -> Result<lex_trace::TraceTree, StoreError> {
1236 let bytes = fs::read(self.trace_path(run_id))?;
1237 Ok(serde_json::from_slice(&bytes)?)
1238 }
1239
1240 pub fn list_traces(&self) -> Result<Vec<String>, StoreError> {
1241 let dir = self.root.join("traces");
1242 if !dir.exists() {
1243 return Ok(Vec::new());
1244 }
1245 let mut out = Vec::new();
1246 for entry in fs::read_dir(dir)? {
1247 let entry = entry?;
1248 if entry.file_type()?.is_dir() {
1249 out.push(entry.file_name().to_string_lossy().to_string());
1250 }
1251 }
1252 out.sort();
1253 Ok(out)
1254 }
1255
1256 // ---- internals ----
1257
1258 /// `<root>/stage_index.jsonl` — an append-only, best-effort
1259 /// reverse index (`StageId` -> owning `SigId`), one JSON object
1260 /// per line. Backs `lookup_lifecycle`'s fast path; see its doc
1261 /// comment. Not a second source of truth: every entry is
1262 /// reconstructible from `stages/<sig>/lifecycle.json`, so a
1263 /// missing, truncated, or entirely absent index file only costs
1264 /// a slower lookup (the pre-existing full scan), never
1265 /// correctness — matching this module's "filesystem is the
1266 /// source of truth" stance (see the module doc comment) rather
1267 /// than introducing an actual second database.
1268 fn stage_index_path(&self) -> PathBuf {
1269 self.root.join("stage_index.jsonl")
1270 }
1271
1272 /// Best-effort load of the whole reverse index into memory.
1273 /// Tolerates a missing file (no index yet) and a corrupt or
1274 /// torn last line (a crash mid-append under the single-writer
1275 /// Tier-1 assumption) by skipping lines that don't parse,
1276 /// rather than failing the lookup that triggered the load.
1277 fn load_stage_index(&self) -> std::collections::BTreeMap<String, String> {
1278 let mut out = std::collections::BTreeMap::new();
1279 let Ok(raw) = fs::read_to_string(self.stage_index_path()) else {
1280 return out;
1281 };
1282 for line in raw.lines() {
1283 if let Ok(entry) = serde_json::from_str::<StageIndexEntry>(line) {
1284 out.insert(entry.stage_id, entry.sig_id);
1285 }
1286 }
1287 out
1288 }
1289
1290 /// Best-effort append of one new `(stage_id, sig_id)` pair.
1291 /// Failure (e.g. a read-only filesystem) only costs a future
1292 /// full scan for this stage_id, never correctness, so it's
1293 /// swallowed rather than propagated.
1294 fn append_stage_index_entry(&self, stage_id: &str, sig: &str) {
1295 use std::io::Write;
1296 let entry = StageIndexEntry { stage_id: stage_id.into(), sig_id: sig.into() };
1297 let Ok(line) = serde_json::to_string(&entry) else { return };
1298 if let Ok(mut f) = fs::OpenOptions::new()
1299 .create(true)
1300 .append(true)
1301 .open(self.stage_index_path())
1302 {
1303 let _ = writeln!(f, "{line}");
1304 }
1305 }
1306
1307 /// Find which SigId owns a StageId, and that sig's lifecycle.
1308 ///
1309 /// Before the reverse index (#822): a full scan over *every*
1310 /// SigId in the tenant (`list_sigs()`, not scoped to the
1311 /// package being looked at), reading and parsing each one's
1312 /// `lifecycle.json` until a match turned up. `get_ast` — called
1313 /// once per pre-existing function when building a publish
1314 /// request's `old_fns_by_name` (`lex-api/src/handlers.rs`) —
1315 /// calls this once per function, so a tenant with a few thousand
1316 /// published functions turned a single publish into millions of
1317 /// individual file reads; measured at roughly an hour on the
1318 /// `alpibrusl` tenant's ~2,400-function store.
1319 ///
1320 /// Now: check the persisted reverse index first (one sequential
1321 /// file read instead of up to N separate ones). A miss — the
1322 /// index doesn't exist yet, or this stage_id predates it — falls
1323 /// back to the full scan and backfills the index so the next
1324 /// lookup for the same stage_id is fast.
1325 fn lookup_lifecycle(&self, stage_id: &str) -> Result<(String, Lifecycle), StoreError> {
1326 let index = self.load_stage_index();
1327 if let Some(sig) = index.get(stage_id) {
1328 if sig == MISSING_STAGE_MARKER {
1329 // A previous full scan already established this
1330 // stage_id exists nowhere in the store. Re-scanning
1331 // would find nothing again -- see #825: a genuinely
1332 // orphaned reference (e.g. from data predating some
1333 // store migration) is looked up on *every* call that
1334 // needs it, forever, so without this negative cache
1335 // it silently costs a full O(total sigs) scan each
1336 // time, indistinguishable from the positive case at
1337 // the call site. Measured directly: on the alpibrusl
1338 // tenant, 988 of 3,664 branch-head entries are
1339 // orphaned this way, turning one `pkg publish`'s
1340 // old_fns_by_name build into ~16M wasted lifecycle
1341 // reads.
1342 return Err(StoreError::UnknownStage(stage_id.into()));
1343 }
1344 if let Ok(life) = self.read_lifecycle(sig) {
1345 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
1346 return Ok((sig.clone(), life));
1347 }
1348 }
1349 // Index entry is stale or wrong (shouldn't happen in
1350 // practice — sig ownership of a stage_id is permanent).
1351 // Fall through to the full scan below rather than trust it.
1352 }
1353 for sig in self.list_sigs()? {
1354 if let Ok(life) = self.read_lifecycle(&sig) {
1355 if life.transitions.iter().any(|t| t.stage_id == stage_id) {
1356 self.append_stage_index_entry(stage_id, &sig);
1357 return Ok((sig, life));
1358 }
1359 }
1360 }
1361 // Genuinely not found anywhere: cache that fact so the next
1362 // lookup for this exact stage_id is an index hit, not another
1363 // full scan. Safe even if this stage_id somehow gets a real
1364 // sig later (content-addressed publish is idempotent, so
1365 // "later" only means "a byte-identical stage republished
1366 // under a real sig") — `append_stage_index_entry`'s later,
1367 // real entry is a later line in the file, and `load_stage_index`
1368 // folds duplicate keys last-write-wins, so the real entry wins.
1369 self.append_stage_index_entry(stage_id, MISSING_STAGE_MARKER);
1370 Err(StoreError::UnknownStage(stage_id.into()))
1371 }
1372
1373 fn read_lifecycle(&self, sig: &str) -> Result<Lifecycle, StoreError> {
1374 let path = self.lifecycle_path(sig);
1375 if !path.exists() {
1376 return Ok(Lifecycle {
1377 sig_id: sig.into(),
1378 transitions: Vec::new(),
1379 });
1380 }
1381 let bytes = fs::read(&path)?;
1382 Ok(serde_json::from_slice(&bytes)?)
1383 }
1384
1385 fn write_lifecycle(&self, sig: &str, life: &Lifecycle) -> Result<(), StoreError> {
1386 write_canonical_json(&self.lifecycle_path(sig), life)
1387 }
1388
1389 /// Apply a published program to a branch as a sequence of typed
1390 /// operations. Returns the ordered list of op_ids + the new
1391 /// head_op. The caller (`lex publish` CLI, `lex serve`'s HTTP
1392 /// handler) is responsible for computing the `DiffReport` against
1393 /// the current branch head — the diff infrastructure lives in
1394 /// `lex-vcs::compute_diff` (previously `lex-cli`) to keep this
1395 /// layer from owning diffing logic.
1396 ///
1397 /// On success: every op in the returned list is durable in the
1398 /// op log and the branch's head_op points at the last one.
1399 /// On a no-op (no diff): returns empty `ops` and the existing
1400 /// `head_op` unchanged.
1401 pub fn publish_program(
1402 &self,
1403 branch: &str,
1404 stages: &[lex_ast::Stage],
1405 diff: &lex_vcs::DiffReport,
1406 new_imports: &lex_vcs::ImportMap,
1407 activate: bool,
1408 ) -> Result<PublishOutcome, StoreError> {
1409 self.publish_program_signed(branch, stages, diff, new_imports, activate, None)
1410 }
1411
1412 /// Signed variant of [`Self::publish_program`] (#227). Every
1413 /// stage written under this batch gets the same signer; per-stage
1414 /// keys aren't supported because the agent identity model treats
1415 /// a publish as a single authorial act.
1416 pub fn publish_program_signed(
1417 &self,
1418 branch: &str,
1419 stages: &[lex_ast::Stage],
1420 diff: &lex_vcs::DiffReport,
1421 new_imports: &lex_vcs::ImportMap,
1422 activate: bool,
1423 signer: Option<&lex_vcs::Keypair>,
1424 ) -> Result<PublishOutcome, StoreError> {
1425 // Single-file / test callers don't publish a mangled package, so
1426 // there are no module prefixes to record (`in_file` stays `None`).
1427 self.publish_program_with_intent(
1428 branch,
1429 stages,
1430 diff,
1431 new_imports,
1432 activate,
1433 signer,
1434 None,
1435 &std::collections::BTreeMap::new(),
1436 )
1437 }
1438
1439 /// [`Self::publish_program_signed`] plus an optional `intent_id`
1440 /// (#131 / #839): when given, every op this publish emits is stamped
1441 /// with it, so the op log records *why* the change happened — the
1442 /// prompt / model / session an agent was acting under — not only
1443 /// what it was. `lex recall --intent <id>` and `lex op replay` read
1444 /// it back. The caller records the [`lex_vcs::Intent`] in the
1445 /// [`lex_vcs::IntentLog`] beforehand; this only links ops to it.
1446 /// `None` is the existing (intent-less) behavior, so op ids for
1447 /// intent-less publishes are unchanged.
1448 // A batch publish legitimately takes the branch, program, diff,
1449 // imports, activate flag, signer, and now the intent — bundling
1450 // them into a struct for one optional field would obscure more
1451 // than it clarifies.
1452 #[allow(clippy::too_many_arguments)]
1453 pub fn publish_program_with_intent(
1454 &self,
1455 branch: &str,
1456 stages: &[lex_ast::Stage],
1457 diff: &lex_vcs::DiffReport,
1458 new_imports: &lex_vcs::ImportMap,
1459 activate: bool,
1460 signer: Option<&lex_vcs::Keypair>,
1461 intent_id: Option<lex_vcs::IntentId>,
1462 // Mangling prefix → package source file, for a multi-module
1463 // package publish; empty for a single file. Recorded as each
1464 // `AddFunction`/`AddType`'s `in_file` so `export-git` can
1465 // de-flatten the package (#894).
1466 module_prefixes: &std::collections::BTreeMap<String, String>,
1467 ) -> Result<PublishOutcome, StoreError> {
1468 use std::collections::{BTreeMap, BTreeSet};
1469
1470 // #130's write-time gate: verify the candidate program
1471 // typechecks (and effects are correctly declared) before
1472 // any disk side-effect. If anything fails, return the
1473 // structured envelope and leave the branch head unchanged
1474 // — the store's "always-valid HEAD" invariant only holds
1475 // because this is the only batch-publish path that
1476 // advances heads. Single-op writes via the lower-level
1477 // `apply_operation` are not gated yet (#130 follow-up).
1478 // #930: resolve any external dependency edges the head keeps
1479 // (empty when no resolver is installed or the head is inlined).
1480 let modules = self.resolved_modules(stages, None);
1481 let module_types = self.resolved_module_types(stages, None);
1482 let dep_prefixes = self.resolved_module_prefixes(stages, None);
1483 if let Err(errors) = lex_types::check_program_with_deps(stages, &modules, &module_types, &dep_prefixes) {
1484 return Err(StoreError::TypeError(errors));
1485 }
1486
1487 // Build old-side views from the current branch. There used to be
1488 // an `old_name_to_sig: BTreeMap<String, SigId>` built here too,
1489 // keyed by bare function name — but a bare name is not unique
1490 // across a package's files (#818: two files can legitimately
1491 // both declare a local `validate` helper with different
1492 // signatures), so a name-keyed map silently collapsed distinct
1493 // SigIds onto one. `diff` now carries each entry's own resolved
1494 // `old_sig_id` directly (see `diff_report`'s doc comments), so
1495 // `diff_to_ops` no longer needs this lookup at all.
1496 let old_head = self.branch_head(branch)?;
1497 // Read every live function's effects through the SigId the head
1498 // names, in one batch. Two reasons, both load-bearing:
1499 //
1500 // * Cost. This was a `get_ast` per live function, and
1501 // `get_ast`'s index-hit path re-reads and re-parses the whole
1502 // `stage_index.jsonl` on every call — O(index × live fns) per
1503 // publish, paid again for every `publish_program` call a
1504 // multi-file publish makes (#828; measured 34s for a no-op
1505 // republish of a real 21-file package against only 698 live
1506 // functions, nearly all of it here).
1507 // * Correctness. A StageId is name-independent, so two live
1508 // functions differing only in name share one and the index
1509 // maps it to a single sig — resolving by StageId therefore
1510 // attributed one function's effects to the *other* one's sig,
1511 // the same ambiguity #826 fixed in `pkg_publish_handler`.
1512 let head_pairs: Vec<(String, String)> = old_head
1513 .iter()
1514 .map(|(sig, stage)| (sig.clone(), stage.clone()))
1515 .collect();
1516 let old_effects: BTreeMap<String, BTreeSet<String>> = head_pairs
1517 .iter()
1518 .zip(self.get_asts_for_sigs_bulk(&head_pairs))
1519 .filter_map(|((sig, _), ast)| match ast.ok()? {
1520 lex_ast::Stage::FnDecl(fd) => {
1521 let s: BTreeSet<String> =
1522 fd.effects.iter().map(|e| e.name.clone()).collect();
1523 Some((sig.clone(), s))
1524 }
1525 _ => None,
1526 })
1527 .collect();
1528 let old_imports = self.derive_imports_from_oplog(branch)?;
1529
1530 let op_kinds = lex_vcs::diff_to_ops(lex_vcs::DiffInputs {
1531 old_head: &old_head,
1532 old_effects: &old_effects,
1533 old_imports: &old_imports,
1534 new_stages: stages,
1535 new_imports,
1536 diff,
1537 module_prefixes,
1538 })
1539 .map_err(|e| StoreError::InvalidTransition(format!("diff_to_ops: {e}")))?;
1540
1541 let mut ops_out: Vec<PublishOp> = Vec::new();
1542 let mut last_op_id: Option<lex_vcs::OpId> = None;
1543 for kind in op_kinds {
1544 // Persist the underlying stage AST/metadata if this op
1545 // produces or replaces one.
1546 if let Some(stg) = stage_for_kind(&kind, stages) {
1547 if !matches!(stg, lex_ast::Stage::Import(_)) {
1548 self.publish_signed(stg, signer)?;
1549 if activate {
1550 if let Some(stage_id_str) = stage_id(stg) {
1551 let _ = self.activate(&stage_id_str);
1552 }
1553 }
1554 }
1555 }
1556 let transition = transition_for_kind(&kind);
1557 let attestable = attestable_stage_ids(&transition);
1558 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
1559 let op =
1560 lex_vcs::Operation::new(kind.clone(), head_now.into_iter().collect::<Vec<_>>());
1561 // #131 / #839: stamp the caller's intent so the op log records
1562 // why this change happened, not just what it was. The CAS
1563 // retry path preserves `intent_id` when it rebuilds the op.
1564 let op = match &intent_id {
1565 Some(id) => op.with_intent(id.clone()),
1566 None => op,
1567 };
1568 let op_id = self.apply_operation(branch, op, transition)?;
1569 self.record_typecheck_passed(&attestable, &op_id)?;
1570 ops_out.push(PublishOp {
1571 op_id: op_id.clone(),
1572 kind: serde_json::to_value(&kind).map_err(StoreError::Serde)?,
1573 });
1574 last_op_id = Some(op_id);
1575 }
1576
1577 let head_op = match last_op_id {
1578 Some(id) => Some(id),
1579 // No ops applied; return whatever the head was already.
1580 None => self.get_branch(branch)?.and_then(|b| b.head_op),
1581 };
1582
1583 Ok(PublishOutcome {
1584 ops: ops_out,
1585 head_op,
1586 })
1587 }
1588
1589 pub fn derive_imports_from_oplog(
1590 &self,
1591 branch: &str,
1592 ) -> Result<lex_vcs::ImportMap, StoreError> {
1593 use lex_vcs::OperationKind::*;
1594 let log = lex_vcs::OpLog::open(self.root())?;
1595 let head = match self.get_branch(branch)?.and_then(|b| b.head_op) {
1596 Some(h) => h,
1597 None => return Ok(Default::default()),
1598 };
1599 let mut out: lex_vcs::ImportMap = Default::default();
1600 for r in log.walk_forward(&head, None)? {
1601 match r.op.kind {
1602 AddImport { in_file, module, alias } => {
1603 // The op omits the alias when it's the module's
1604 // default (last path segment) to keep its OpId
1605 // stable; rebuild it the same way on the way out.
1606 let alias =
1607 alias.unwrap_or_else(|| lex_vcs::default_import_alias(&module));
1608 out.entry(in_file)
1609 .or_default()
1610 .insert(lex_vcs::ImportRef { reference: module, alias });
1611 }
1612 RemoveImport { in_file, module } => {
1613 // Removal is keyed by reference (the op carries no
1614 // alias), so drop any binding of that module.
1615 if let Some(set) = out.get_mut(&in_file) {
1616 set.retain(|ir| ir.reference != module);
1617 }
1618 }
1619 _ => {}
1620 }
1621 }
1622 Ok(out)
1623 }
1624
1625 /// Apply an operation to a branch and advance its head_op.
1626 ///
1627 /// The single advance path. Validates parents via `lex_vcs::apply`,
1628 /// persists the operation via the op log, then atomically advances
1629 /// the branch file's head_op via `set_branch_head_op`.
1630 ///
1631 /// Errors:
1632 /// - `UnknownBranch`: branch does not exist (no op is persisted).
1633 /// - `Apply(ApplyError::StaleParent)`: the op's parents don't
1634 /// match the branch head — head is unchanged. Callers that
1635 /// want retry-on-stale (e.g. `lex publish` re-running against
1636 /// a moved head) match on this variant explicitly.
1637 /// - `Apply(ApplyError::UnknownMergeParent)`: a merge op's
1638 /// second parent isn't in the log.
1639 /// - `Io`: filesystem error during persist or branch advance.
1640 ///
1641 /// Crash recovery: between op persist and branch advance, a crash
1642 /// can leave an orphan op record in the log with no branch
1643 /// pointing at it. The op is content-addressed and cheap to
1644 /// re-derive from the same source. See
1645 /// Apply a single op against `branch`, gated on the candidate
1646 /// program typechecking. The per-op variant of #130's
1647 /// write-time gate — counterpart to [`Self::publish_program`]'s
1648 /// batch-mode check.
1649 ///
1650 /// `candidate` is the sequence of `Stage`s that *would* exist
1651 /// on this branch after the op is applied. Caller's
1652 /// responsibility: today neither `lex-store` nor `lex-vcs`
1653 /// reconstruct the candidate from the op + branch state on
1654 /// behalf of the caller. The natural callers (HTTP `POST
1655 /// /v1/publish` for a single op; agent harnesses driving
1656 /// merges via the future #134 API) already have the candidate
1657 /// in memory.
1658 ///
1659 /// On rejection: branch head unchanged, no op record persisted.
1660 /// Same atomicity guarantee as the publish path.
1661 ///
1662 /// # Why a separate method, not a flag on `apply_operation`
1663 ///
1664 /// `apply_operation` accepting `Option<&[Stage]>` and silently
1665 /// skipping the gate on `None` is exactly the kind of
1666 /// "secretly opt-out" path #130 is trying to remove. The honest
1667 /// split: `apply_operation` for the one caller that already
1668 /// typechecked its input up front (`publish_program`),
1669 /// `apply_operation_checked` for callers holding the candidate,
1670 /// [`Self::apply_operation_gated`] for single-parent callers
1671 /// that hold only the transition (`/v1/patch`), and
1672 /// [`Self::apply_merge_op_gated`] for merge commits (#833).
1673 pub fn apply_operation_checked(
1674 &self,
1675 branch: &str,
1676 op: lex_vcs::Operation,
1677 transition: lex_vcs::StageTransition,
1678 candidate: &[lex_ast::Stage],
1679 ) -> Result<lex_vcs::OpId, StoreError> {
1680 let modules = self.resolved_modules(candidate, None); // #930
1681 let module_types = self.resolved_module_types(candidate, None);
1682 let dep_prefixes = self.resolved_module_prefixes(candidate, None);
1683 if let Err(errors) = lex_types::check_program_with_deps(candidate, &modules, &module_types, &dep_prefixes) {
1684 // #281: emit a `RepairHint` attestation against each
1685 // candidate stage the transition was about to produce.
1686 // The op record itself isn't persisted (the gate is
1687 // pre-persistence), but the candidate stage IS — the
1688 // transform-flow methods publish before this call.
1689 // The attached hint lets `lex repair <op_id>` and
1690 // future LLM-assisted apply paths read the structured
1691 // errors without re-running the typecheck.
1692 let attestable = attestable_stage_ids(&transition);
1693 let failed_op_id = op.op_id();
1694 let _ = self.record_repair_hint(&attestable, &failed_op_id, &errors);
1695 return Err(StoreError::TypeError(errors));
1696 }
1697 // #292 slice 3: per-session budget gate. After typecheck
1698 // passes, refuse the op if it would push its session's
1699 // monotonic spend over the configured cap. Sessions
1700 // without an intent_id, or with an intent whose session
1701 // has no cap configured, sail through.
1702 self.check_session_budget(&op)?;
1703 let attestable = attestable_stage_ids(&transition);
1704 let op_effects = op_declared_effects(&op.kind);
1705 // #262: CAS retry loop. Single-parent ops can be safely
1706 // re-persisted under a new parent on contention (the kind
1707 // is invariant; only `parents` changes). Merge ops (already
1708 // 2-parent) come through the merge engine which has its own
1709 // coordination; we don't retry them here — we'll see the
1710 // first attempt's CAS fail and surface Contention.
1711 self.cas_retry_advance(branch, op, transition, |new_head| {
1712 self.record_typecheck_passed(&attestable, &new_head.op_id)?;
1713 self.run_required_attestations_gate(branch, &new_head.op_id, &attestable, &op_effects)
1714 })
1715 }
1716
1717 /// The program that would exist on `branch` after `transition`
1718 /// is applied: the branch head (snapshot-cached) with the
1719 /// transition replayed over it, every resulting `(sig, stage)`
1720 /// bulk-loaded. Exact for a **single-parent** transition — the
1721 /// candidate [`Self::apply_operation_gated`] wants. Not valid for
1722 /// a merge: a `StageTransition::Merge` records only the delta
1723 /// relative to dst, while the op-DAG replay that computes a
1724 /// merge's real head walks both parents (#833).
1725 pub fn candidate_program_for(
1726 &self,
1727 branch: &str,
1728 transition: &lex_vcs::StageTransition,
1729 ) -> Result<Vec<Stage>, StoreError> {
1730 let mut head = self.branch_head(branch)?;
1731 crate::branches::apply_transition(&mut head, transition);
1732 let pairs: Vec<(String, String)> = head.into_iter().collect();
1733 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect()
1734 }
1735
1736 /// [`Self::apply_operation_checked`] for a **single-parent** op
1737 /// where the caller holds only the transition: assembles the
1738 /// candidate via [`Self::candidate_program_for`] and runs the
1739 /// gate. Same rejection semantics — `TypeError`, a `RepairHint`
1740 /// attestation, head unchanged, nothing persisted. This is the
1741 /// write path for `/v1/patch` (#833). Merge ops must not use it
1742 /// (see `candidate_program_for`); they go through
1743 /// [`Self::apply_merge_op_gated`].
1744 pub fn apply_operation_gated(
1745 &self,
1746 branch: &str,
1747 op: lex_vcs::Operation,
1748 transition: lex_vcs::StageTransition,
1749 ) -> Result<lex_vcs::OpId, StoreError> {
1750 debug_assert!(
1751 op.parents.len() <= 1,
1752 "apply_operation_gated is single-parent only; merges use apply_merge_op_gated"
1753 );
1754 let candidate = self.candidate_program_for(branch, &transition)?;
1755 self.apply_operation_checked(branch, op, transition, &candidate)
1756 }
1757
1758 /// The gated write path for **merge** commits (`commit_merge`,
1759 /// `POST /v1/merge/<id>/commit`, `lex merge commit`).
1760 ///
1761 /// A `StageTransition::Merge` records only the delta relative to
1762 /// dst; the sig->stage map every consumer reads is recomputed by
1763 /// replaying the op DAG, which for a merge walks *both* parents
1764 /// and can surface sigs the delta never mentions. So the only way
1765 /// to know the true post-merge program is to replay it — land the
1766 /// op and read `branch_head`. This lands the merge op,
1767 /// type-checks the resulting head, and on a failure rolls the
1768 /// head back and returns `TypeError`.
1769 ///
1770 /// Before #833 the merge paths landed through the ungated
1771 /// `apply_operation`, so a merge whose result didn't compose
1772 /// (e.g. dst still calls `helper`, an agent-supplied resolution
1773 /// dropped it) advanced the head with nothing to catch it.
1774 ///
1775 /// Rollback leaves the rejected merge op as an unreachable record
1776 /// (reclaimed by `lex op gc`, the same orphan crash-recovery
1777 /// already tolerates). A stage the merge names that was never
1778 /// published surfaces as the underlying `StoreError` from the
1779 /// bulk read — the "never advance onto content that can't be
1780 /// loaded" invariant from the other side.
1781 pub fn apply_merge_op_gated(
1782 &self,
1783 branch: &str,
1784 op: lex_vcs::Operation,
1785 transition: lex_vcs::StageTransition,
1786 ) -> Result<lex_vcs::OpId, StoreError> {
1787 let head_before = self.get_branch(branch)?.and_then(|b| b.head_op);
1788 // Capture the stages this merge introduces before `transition`
1789 // is moved into `apply_operation`; used for the TypeCheck
1790 // attestation below.
1791 let attestable = attestable_stage_ids(&transition);
1792 let op_id = self.apply_operation(branch, op, transition)?;
1793
1794 let verdict = (|| -> Result<(), StoreError> {
1795 let head = self.branch_head(branch)?;
1796 let pairs: Vec<(String, String)> = head.into_iter().collect();
1797 let stages: Vec<Stage> =
1798 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
1799 // #930: per-head dep resolution for the merge path is a follow-up;
1800 // None lets a client resolver use its working-copy lock.
1801 let modules = self.resolved_modules(&stages, None);
1802 let module_types = self.resolved_module_types(&stages, None);
1803 let dep_prefixes = self.resolved_module_prefixes(&stages, None);
1804 if let Err(errors) = lex_types::check_program_with_deps(&stages, &modules, &module_types, &dep_prefixes) {
1805 return Err(StoreError::TypeError(errors));
1806 }
1807 Ok(())
1808 })();
1809
1810 if let Err(e) = verdict {
1811 // Roll the head back. The empty-dst case never reaches
1812 // here (it fast-forwards without a merge op), so
1813 // `head_before` is always `Some` on this arm.
1814 if let Some(prev) = head_before {
1815 self.set_branch_head_op(branch, prev)?;
1816 }
1817 return Err(e);
1818 }
1819 // #835: the merge's post-merge head type-checked, but until now
1820 // that verdict left no trace in the attestation log — so a
1821 // merged stage looked un-type-checked to `lex blame
1822 // --with-evidence` and the attestation queries, unlike a
1823 // published or patched stage. Emit `TypeCheck::Passed` for the
1824 // stages the merge introduced, mirroring the publish / patch
1825 // paths (`record_typecheck_passed`). Emitted only after the
1826 // check passes and the head is committed, so a rolled-back
1827 // merge records nothing.
1828 self.record_typecheck_passed(&attestable, &op_id)?;
1829 Ok(op_id)
1830 }
1831
1832 /// Type-check the program that would result from overlaying a merge
1833 /// `delta` onto `branch`'s current head — **without moving the
1834 /// head** (#834). `delta` maps `sig_id -> Some(stage)` to set that
1835 /// sig to `stage`, or `sig_id -> None` to remove it, exactly the
1836 /// `entries` a `StageTransition::Merge` records.
1837 ///
1838 /// This is the read-only, resolve-time counterpart of
1839 /// `apply_merge_op_gated`'s commit-time gate: it lets a merge
1840 /// session tell an agent *which resolution broke type-checking* the
1841 /// moment it is submitted, instead of only after a failed commit.
1842 /// `Ok(())` means the projected program composes; a type failure is
1843 /// `Err(StoreError::TypeError(..))`; a read failure is the
1844 /// corresponding `StoreError` I/O variant.
1845 pub fn typecheck_merge_projection(
1846 &self,
1847 branch: &str,
1848 delta: &std::collections::BTreeMap<String, Option<String>>,
1849 ) -> Result<(), StoreError> {
1850 let mut head = self.branch_head(branch)?;
1851 for (sig, stage) in delta {
1852 match stage {
1853 Some(s) => { head.insert(sig.clone(), s.clone()); }
1854 None => { head.remove(sig); }
1855 }
1856 }
1857 let pairs: Vec<(String, String)> = head.into_iter().collect();
1858 let stages: Vec<Stage> =
1859 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
1860 let modules = self.resolved_modules(&stages, None); // #930 (patch path)
1861 let module_types = self.resolved_module_types(&stages, None);
1862 let dep_prefixes = self.resolved_module_prefixes(&stages, None);
1863 if let Err(errors) = lex_types::check_program_with_deps(&stages, &modules, &module_types, &dep_prefixes) {
1864 return Err(StoreError::TypeError(errors));
1865 }
1866 Ok(())
1867 }
1868
1869 /// #838: attempt a typed three-way merge of a single sig's body for
1870 /// a `ModifyModify` conflict — the intra-function, better-than-git
1871 /// case where two agents edited *disjoint* subtrees of the same
1872 /// function (different match arms, different let bindings).
1873 ///
1874 /// `base` / `ours` (the dst side) / `theirs` (the src side) are the
1875 /// three stage ids the merge engine surfaced for `sig_id`. Loads
1876 /// the three `FnDecl`s, structurally merges the bodies
1877 /// ([`lex_vcs::merge_bodies`]), and accepts the result *only if* the
1878 /// merged function also type-checks against `dst_branch`'s head — a
1879 /// body that composes syntactically but not by type is still a
1880 /// conflict (#838). On success the merged stage is published
1881 /// (content-addressed, idempotent; orphaned and GC-reclaimable if
1882 /// the merge is never committed) and its id returned; `None` means
1883 /// "fall back to a whole-function conflict."
1884 ///
1885 /// Deliberately narrow for this slice: only pure body divergence is
1886 /// merged. If the two sides disagree on anything but the body
1887 /// (examples, type params — the signature is identical by
1888 /// construction, since all three share `sig_id`), or either stage
1889 /// isn't a function, it falls back to a conflict.
1890 pub fn try_semantic_body_merge(
1891 &self,
1892 dst_branch: &str,
1893 sig_id: &str,
1894 base: &str,
1895 ours: &str,
1896 theirs: &str,
1897 ) -> Result<Option<String>, StoreError> {
1898 use lex_ast::Stage::FnDecl;
1899 let (base_fd, ours_fd, theirs_fd) =
1900 match (self.get_ast(base), self.get_ast(ours), self.get_ast(theirs)) {
1901 (Ok(FnDecl(b)), Ok(FnDecl(o)), Ok(FnDecl(t))) => (b, o, t),
1902 // A non-function stage (type decl / import) or a stage
1903 // that can't be loaded isn't an intra-body merge.
1904 _ => return Ok(None),
1905 };
1906
1907 // Only the body may diverge between the two sides.
1908 if !fndecl_same_except_body(&ours_fd, &theirs_fd) {
1909 return Ok(None);
1910 }
1911
1912 let merged_body =
1913 match lex_vcs::merge_bodies(&base_fd.body, &ours_fd.body, &theirs_fd.body) {
1914 lex_vcs::BodyMerge::Merged(b) => b,
1915 lex_vcs::BodyMerge::Conflict => return Ok(None),
1916 };
1917
1918 let mut merged_fd = ours_fd.clone();
1919 merged_fd.body = merged_body;
1920 let merged_stage = lex_ast::Stage::FnDecl(merged_fd);
1921 let new_stage_id = match stage_id(&merged_stage) {
1922 Some(id) => id,
1923 None => return Ok(None),
1924 };
1925
1926 // Type-check the merged fn in context: dst's head with this sig
1927 // swapped to the merged stage. Requires the merged stage to be
1928 // loadable, so publish first (idempotent, content-addressed).
1929 self.publish(&merged_stage)?;
1930 let mut delta = std::collections::BTreeMap::new();
1931 delta.insert(sig_id.to_string(), Some(new_stage_id.clone()));
1932 match self.typecheck_merge_projection(dst_branch, &delta) {
1933 Ok(()) => Ok(Some(new_stage_id)),
1934 // Composes syntactically, not by type → still a conflict.
1935 Err(StoreError::TypeError(_)) => Ok(None),
1936 Err(e) => Err(e),
1937 }
1938 }
1939
1940 /// #836 G3: assemble everything a regenerator needs to *replay* an
1941 /// op — re-derive the change from its recorded cause. Returns the
1942 /// op's recorded intent (prompt / model / session), the target sig
1943 /// and the stage id it produced, and the program the change was
1944 /// made against (the parent state, rendered to source). An external
1945 /// harness feeds the prompt + parent program to the recorded model,
1946 /// then hands the regenerated stage back to [`Self::replay_compare`]
1947 /// (lex owns the deterministic comparison; the model call is the
1948 /// harness's, matching the rest of the architecture).
1949 ///
1950 /// Errors with `UnknownOp` if the op_id is unknown, or
1951 /// `InvalidTransition` if the op didn't produce a stage (a removal /
1952 /// import / merge has nothing to regenerate).
1953 pub fn replay_request(&self, op_id: &str) -> Result<ReplayRequest, StoreError> {
1954 let log = lex_vcs::OpLog::open(self.root())?;
1955 let record = log
1956 .get(&op_id.to_string())?
1957 .ok_or_else(|| StoreError::UnknownOp(op_id.to_string()))?;
1958 let (target_sig, expected_stage_id) = produced_sig_stage(&record.produces)
1959 .ok_or_else(|| StoreError::InvalidTransition(format!("op {op_id} produced no stage to replay")))?;
1960
1961 let (prompt, model, session_id) = match &record.op.intent_id {
1962 Some(id) => {
1963 let intents = lex_vcs::IntentLog::open(self.root())?;
1964 match intents.get(id)? {
1965 Some(i) => (Some(i.prompt), Some(model_label(&i.model)), Some(i.session_id)),
1966 None => (None, None, None),
1967 }
1968 }
1969 None => (None, None, None),
1970 };
1971
1972 // The program the op was applied against: the head state at its
1973 // (first) parent, rendered with the canonical printer. A root
1974 // op has no parent → empty program.
1975 let parent_program = match record.op.parents.first() {
1976 Some(parent) => self.program_source_at_op(parent)?,
1977 None => String::new(),
1978 };
1979
1980 // The target function's name + signature, from the recorded
1981 // stage — a regenerator needs the interface, not just the hash.
1982 let (target_name, target_signature) = match self.get_ast(&expected_stage_id) {
1983 Ok(lex_ast::Stage::FnDecl(fd)) => {
1984 (Some(fd.name.clone()), Some(lex_vcs::render_signature(&fd)))
1985 }
1986 _ => (None, None),
1987 };
1988
1989 Ok(ReplayRequest {
1990 op_id: op_id.to_string(),
1991 target_sig,
1992 target_name,
1993 target_signature,
1994 expected_stage_id,
1995 prompt,
1996 model,
1997 session_id,
1998 parent_program,
1999 })
2000 }
2001
2002 /// #836 G3: compare a regenerated `candidate` against what the op
2003 /// recorded producing, and emit the `Replay` attestation. The
2004 /// reproducibility claim made concrete — a faithful regeneration of
2005 /// the same function from the same cause yields the same
2006 /// content-addressed stage id.
2007 ///
2008 /// `reproduced` is true iff the candidate is the same sig *and* the
2009 /// same stage id the op recorded. A candidate for a different sig
2010 /// counts as "not reproduced" (`produced_stage_id: None`) rather
2011 /// than an error — it's a legitimate, if negative, replay result.
2012 /// The attestation is addressed to the op's recorded stage, so
2013 /// `list_for_stage` surfaces it alongside the TypeCheck/Examples
2014 /// evidence.
2015 pub fn replay_compare(
2016 &self,
2017 op_id: &str,
2018 candidate: &Stage,
2019 ) -> Result<ReplayOutcome, StoreError> {
2020 let (target_sig, expected_stage_id) = self.replay_target(op_id)?;
2021 let cand_sig = lex_ast::sig_id(candidate);
2022 let cand_stage = stage_id(candidate);
2023 let produced_stage_id = match (cand_sig.as_deref(), &cand_stage) {
2024 // Same function regenerated: the produced stage is
2025 // whatever it content-addresses to.
2026 (Some(s), Some(st)) if s == target_sig => Some(st.clone()),
2027 // A different sig (or an unhashable stage) isn't a
2028 // regeneration of this op's change.
2029 _ => None,
2030 };
2031 let reproduced = produced_stage_id.as_deref() == Some(expected_stage_id.as_str());
2032 let detail = if reproduced {
2033 None
2034 } else {
2035 Some("regeneration did not reproduce the recorded stage".to_string())
2036 };
2037 self.emit_replay(op_id, &expected_stage_id, produced_stage_id, reproduced, None, detail)
2038 }
2039
2040 /// Record a *negative* replay result for a regeneration that never
2041 /// yielded a comparable stage — the output didn't parse, or didn't
2042 /// define the target sig (#836 G3). Emits a `Replay { reproduced:
2043 /// false, produced_stage_id: None }` attestation with `reason` in
2044 /// its `Failed` detail, so an automated `lex op replay` run always
2045 /// records a verdict rather than aborting. `reason` is caller-supplied
2046 /// (e.g. "regenerated source did not parse").
2047 pub fn replay_record_miss(&self, op_id: &str, reason: &str) -> Result<ReplayOutcome, StoreError> {
2048 let (_target_sig, expected_stage_id) = self.replay_target(op_id)?;
2049 self.emit_replay(op_id, &expected_stage_id, None, false, None, Some(reason.to_string()))
2050 }
2051
2052 /// `(target_sig, expected_stage_id)` for a replayable op, or an
2053 /// error if the op is unknown or produced no stage.
2054 fn replay_target(&self, op_id: &str) -> Result<(String, String), StoreError> {
2055 let log = lex_vcs::OpLog::open(self.root())?;
2056 let record = log
2057 .get(&op_id.to_string())?
2058 .ok_or_else(|| StoreError::UnknownOp(op_id.to_string()))?;
2059 produced_sig_stage(&record.produces)
2060 .ok_or_else(|| StoreError::InvalidTransition(format!("op {op_id} produced no stage to replay")))
2061 }
2062
2063 /// Record a replay verdict the caller has already decided — used by
2064 /// the CLI's behavioral tier, which does the (VM-backed) equivalence
2065 /// check the store deliberately can't. `expected_stage_id` is looked
2066 /// up from the op. Set `behavioral_samples` to `Some(n)` when the
2067 /// candidate reproduced *behaviorally* over `n` sampled inputs rather
2068 /// than by exact stage-id match; the attestation then records that
2069 /// weaker-but-real claim distinctly.
2070 pub fn replay_record(
2071 &self,
2072 op_id: &str,
2073 produced_stage_id: Option<String>,
2074 reproduced: bool,
2075 behavioral_samples: Option<usize>,
2076 fail_detail: Option<String>,
2077 ) -> Result<ReplayOutcome, StoreError> {
2078 let (_target_sig, expected_stage_id) = self.replay_target(op_id)?;
2079 self.emit_replay(op_id, &expected_stage_id, produced_stage_id, reproduced, behavioral_samples, fail_detail)
2080 }
2081
2082 /// Compute the exact-match verdict for a candidate *without* emitting
2083 /// an attestation — `(expected_stage_id, produced_stage_id, exact)`.
2084 /// Lets a caller (the CLI) fall back to a behavioral check on a valid
2085 /// but non-identical candidate and emit a single verdict, instead of
2086 /// [`Self::replay_compare`]'s emit-immediately shape.
2087 pub fn replay_stage_of(
2088 &self,
2089 op_id: &str,
2090 candidate: &Stage,
2091 ) -> Result<(String, Option<String>, bool), StoreError> {
2092 let (target_sig, expected_stage_id) = self.replay_target(op_id)?;
2093 let cand_sig = lex_ast::sig_id(candidate);
2094 let cand_stage = stage_id(candidate);
2095 let produced_stage_id = match (cand_sig.as_deref(), &cand_stage) {
2096 (Some(s), Some(st)) if s == target_sig => Some(st.clone()),
2097 _ => None,
2098 };
2099 let exact = produced_stage_id.as_deref() == Some(expected_stage_id.as_str());
2100 Ok((expected_stage_id, produced_stage_id, exact))
2101 }
2102
2103 /// Emit the `Replay` attestation and build the outcome. Shared by
2104 /// [`Self::replay_compare`], [`Self::replay_record_miss`], and
2105 /// [`Self::replay_record`].
2106 fn emit_replay(
2107 &self,
2108 op_id: &str,
2109 expected_stage_id: &str,
2110 produced_stage_id: Option<String>,
2111 reproduced: bool,
2112 behavioral_samples: Option<usize>,
2113 fail_detail: Option<String>,
2114 ) -> Result<ReplayOutcome, StoreError> {
2115 let model = {
2116 let log = lex_vcs::OpLog::open(self.root())?;
2117 match log.get(&op_id.to_string())?.and_then(|r| r.op.intent_id) {
2118 Some(id) => lex_vcs::IntentLog::open(self.root())?
2119 .get(&id)?
2120 .map(|i| model_label(&i.model)),
2121 None => None,
2122 }
2123 };
2124 let result = if reproduced {
2125 lex_vcs::AttestationResult::Passed
2126 } else {
2127 lex_vcs::AttestationResult::Failed {
2128 detail: fail_detail.unwrap_or_else(|| "not reproduced".into()),
2129 }
2130 };
2131 let attestation = lex_vcs::Attestation::new(
2132 expected_stage_id.to_string(),
2133 Some(op_id.to_string()),
2134 None,
2135 lex_vcs::AttestationKind::Replay {
2136 expected_stage_id: expected_stage_id.to_string(),
2137 produced_stage_id: produced_stage_id.clone(),
2138 reproduced,
2139 behavioral_samples,
2140 model,
2141 },
2142 result,
2143 replay_producer(),
2144 None,
2145 );
2146 let attestation_id = attestation.attestation_id.clone();
2147 self.attestation_log()?.put(&attestation)?;
2148 Ok(ReplayOutcome {
2149 op_id: op_id.to_string(),
2150 expected_stage_id: expected_stage_id.to_string(),
2151 produced_stage_id,
2152 reproduced,
2153 behavioral_samples,
2154 attestation_id,
2155 })
2156 }
2157
2158 /// The program at an op (that op and all its ancestors applied), as
2159 /// canonical stages. The behavioral replay tier needs the whole
2160 /// program — a regenerated function may call helpers from its parent
2161 /// state, so it can only be run in context. Exposed for the CLI's
2162 /// equivalence check; `op_id` may be any op in the log.
2163 pub fn program_stages_at_op(&self, op_id: &str) -> Result<Vec<Stage>, StoreError> {
2164 let oid: lex_vcs::OpId = op_id.to_string();
2165 let log = lex_vcs::OpLog::open(self.root())?;
2166 let mut map: std::collections::BTreeMap<String, String> = std::collections::BTreeMap::new();
2167 for rec in log.walk_forward(&oid, None)? {
2168 crate::branches::apply_transition(&mut map, &rec.produces);
2169 }
2170 let pairs: Vec<(String, String)> = map.into_iter().collect();
2171 let stages: Vec<Stage> =
2172 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
2173 Ok(stages)
2174 }
2175
2176 /// The program at an op, rendered to source. Used to give a replay
2177 /// regenerator the context the change was made against.
2178 fn program_source_at_op(&self, op_id: &lex_vcs::OpId) -> Result<String, StoreError> {
2179 Ok(lex_ast::print_stages(&self.program_stages_at_op(op_id)?))
2180 }
2181
2182 /// Open the attestation log rooted at this store. The log lives
2183 /// under `<root>/attestations/`; opening is idempotent and cheap
2184 /// (`fs::create_dir_all`). Exposed publicly so consumers — `lex
2185 /// blame --with-evidence`, `GET /v1/stage/<id>/attestations` —
2186 /// can read what the store gate emitted without round-tripping
2187 /// through this crate's API surface.
2188 /// Recompute a producer's trust score from its recent
2189 /// attestation history and emit a fresh `ProducerTrust`
2190 /// attestation (#293). Score = `passed / (passed + failed
2191 /// + inconclusive)` over the last `window` attestations
2192 /// produced by `tool_id`, expressed in thousandths
2193 /// (`0..=1000`).
2194 ///
2195 /// Refuses to grant trust when the tool has an active
2196 /// `ProducerBlock` — the block wins as a hard veto. Returns
2197 /// `Ok(None)` for "no attestations to score" (a brand-new
2198 /// producer); the caller can choose how to handle it
2199 /// (typically: skip the publish until evidence accrues).
2200 ///
2201 /// `granted_by` is the identity of the actor running the
2202 /// recompute (typically the human admin, or "lex-ci-bot"
2203 /// for an automated nightly).
2204 pub fn recompute_producer_trust(
2205 &self,
2206 tool_id: &str,
2207 window: usize,
2208 granted_by: &str,
2209 ) -> Result<Option<lex_vcs::AttestationId>, StoreError> {
2210 let log = self.attestation_log()?;
2211 let all = log.list_all()?;
2212 // Hard veto: don't grant trust to a blocked tool.
2213 if lex_vcs::active_producer_block(&all, tool_id).is_some() {
2214 return Err(StoreError::InvalidTransition(format!(
2215 "cannot recompute trust for `{tool_id}` — \
2216 producer is currently blocked"
2217 )));
2218 }
2219 // Filter to attestations from this tool, newest-first by
2220 // timestamp, then take the window.
2221 let mut from_tool: Vec<&lex_vcs::Attestation> = all
2222 .iter()
2223 .filter(|a| a.produced_by.tool == tool_id)
2224 // Ignore self-referential trust attestations (we're
2225 // scoring evidence, not previous trust statements).
2226 .filter(|a| {
2227 !matches!(
2228 a.kind,
2229 lex_vcs::AttestationKind::ProducerTrust { .. }
2230 | lex_vcs::AttestationKind::TrustWaived { .. }
2231 )
2232 })
2233 .collect();
2234 from_tool.sort_by_key(|a| std::cmp::Reverse(a.timestamp));
2235 from_tool.truncate(window);
2236 if from_tool.is_empty() {
2237 return Ok(None);
2238 }
2239 let (mut passed, mut total) = (0u64, 0u64);
2240 for a in &from_tool {
2241 total += 1;
2242 if matches!(a.result, lex_vcs::AttestationResult::Passed) {
2243 passed += 1;
2244 }
2245 }
2246 let score = if total == 0 {
2247 0
2248 } else {
2249 let raw = (passed as f64) * 1000.0 / (total as f64);
2250 raw.round().clamp(0.0, 1000.0) as u32
2251 };
2252 let head_op = self
2253 .list_branches()?
2254 .into_iter()
2255 .find_map(|b| self.get_branch(&b).ok().flatten().and_then(|x| x.head_op))
2256 .unwrap_or_else(|| "fresh".into());
2257 let evidence = format!(
2258 "window={window}, sample={}, head_op={head_op:.16}",
2259 from_tool.len()
2260 );
2261 let attestation = lex_vcs::Attestation::new(
2262 tool_id.to_string(),
2263 None,
2264 None,
2265 lex_vcs::AttestationKind::ProducerTrust {
2266 tool_id: tool_id.into(),
2267 score_thousandths: score,
2268 evidence,
2269 granted_by: granted_by.into(),
2270 },
2271 lex_vcs::AttestationResult::Passed,
2272 producer_trust_producer(),
2273 None,
2274 );
2275 let id = attestation.attestation_id.clone();
2276 log.put(&attestation)?;
2277 Ok(Some(id))
2278 }
2279
2280 /// The latest live `ProducerTrust` score (thousandths, `0..=1000`) for
2281 /// every producer that currently has trust: the newest score per tool by
2282 /// timestamp, excluding any tool under an active `ProducerBlock` (a block
2283 /// is a hard veto over trust, matching `recompute_producer_trust`).
2284 ///
2285 /// Used to export a capsule trusted-keys keyring from *earned* trust — the
2286 /// producer id doubles as the publisher's signing key downstream, so this
2287 /// turns track record into the allowlist `capsule install` consumes.
2288 pub fn live_producer_trust_scores(
2289 &self,
2290 ) -> Result<std::collections::BTreeMap<String, u32>, StoreError> {
2291 let log = self.attestation_log()?;
2292 let all = log.list_all()?;
2293 // Newest score per tool.
2294 let mut latest: std::collections::BTreeMap<String, (u64, u32)> =
2295 std::collections::BTreeMap::new();
2296 for a in &all {
2297 if let lex_vcs::AttestationKind::ProducerTrust {
2298 tool_id,
2299 score_thousandths,
2300 ..
2301 } = &a.kind
2302 {
2303 let entry = latest.entry(tool_id.clone()).or_insert((0, 0));
2304 if a.timestamp >= entry.0 {
2305 *entry = (a.timestamp, *score_thousandths);
2306 }
2307 }
2308 }
2309 // Drop blocked producers; a block vetoes trust.
2310 let mut scores = std::collections::BTreeMap::new();
2311 for (tool, (_, score)) in latest {
2312 if lex_vcs::active_producer_block(&all, &tool).is_some() {
2313 continue;
2314 }
2315 scores.insert(tool, score);
2316 }
2317 Ok(scores)
2318 }
2319
2320 pub fn attestation_log(&self) -> Result<lex_vcs::AttestationLog, StoreError> {
2321 Ok(lex_vcs::AttestationLog::open(self.root())?)
2322 }
2323
2324 /// Emit one `TypeCheck::Passed` attestation per stage produced by
2325 /// a successful gated apply. Idempotent on `attestation_id` —
2326 /// re-running the same gate run dedups via content addressing.
2327 ///
2328 /// Failure modes: `io::Error` from the attestation log (disk
2329 /// full, perms). The op has already landed by the time this
2330 /// runs; an error here means the op is durable but the evidence
2331 /// is missing. We propagate so the caller sees the partial
2332 /// state rather than silently swallowing — re-attesting the
2333 /// same op against the same op_id is idempotent (content
2334 /// addressing) so a retry is safe once the underlying issue is
2335 /// fixed.
2336 fn record_typecheck_passed(
2337 &self,
2338 stage_ids: &[String],
2339 op_id: &lex_vcs::OpId,
2340 ) -> Result<(), StoreError> {
2341 if stage_ids.is_empty() {
2342 return Ok(());
2343 }
2344 let log = self.attestation_log()?;
2345 for stage_id in stage_ids {
2346 let attestation = lex_vcs::Attestation::new(
2347 stage_id.clone(),
2348 Some(op_id.clone()),
2349 None,
2350 lex_vcs::AttestationKind::TypeCheck,
2351 lex_vcs::AttestationResult::Passed,
2352 typecheck_producer(),
2353 None,
2354 );
2355 log.put(&attestation)?;
2356 }
2357 Ok(())
2358 }
2359
2360 /// The hosted CI runner (#93): independently re-run the write-time
2361 /// type-check gate on a branch head and record the verdict as a
2362 /// `lex-hub-ci`-produced `TypeCheck` attestation for the stages the
2363 /// advance introduced. Called after an `op push` fast-forwards the
2364 /// head, so `require-attestation type_check` gates are backed by a
2365 /// producer that actually verified the code server-side, not by
2366 /// whatever attestation a client chose to attach. Does NOT move or
2367 /// roll back the head — the client's own always-valid-HEAD gate is
2368 /// what refuses a bad publish; this produces the trusted verdict on
2369 /// top of an already-committed advance (so a client that bypassed
2370 /// its gate is caught by a `TypeCheck::Failed` from `lex-hub-ci`).
2371 ///
2372 /// `from_head` is the branch head *before* the advance; the ops
2373 /// between it and `to_head` are the ones whose stages get attested.
2374 /// Idempotent: attestations are content-addressed, so re-verifying
2375 /// the same head is a no-op.
2376 pub fn verify_head_and_attest(
2377 &self,
2378 branch: &str,
2379 from_head: Option<&str>,
2380 to_head: &str,
2381 ) -> Result<HubCiVerdict, StoreError> {
2382 // Reconstruct the program at the new head and re-check it.
2383 let head = self.branch_head(branch)?;
2384 let pairs: Vec<(String, String)> = head.into_iter().collect();
2385 let decls: Vec<Stage> =
2386 self.get_asts_for_sigs_bulk(&pairs).into_iter().collect::<Result<_, _>>()?;
2387 let checked_stages = decls.len();
2388 // #930: the SigId→stage map holds only fn/type declarations — the
2389 // head's `import` edges are AddImport ops, absent here. Reconstruct
2390 // them so a non-inlined head's `<alias>.name` references bind: the
2391 // resolver scans these imports to resolve each dependency, and the
2392 // checker's Pass 1 binds the alias to the resolved module. (Without
2393 // this the alias is unbound and the head fails as `unknown_identifier`,
2394 // even with the dependency correctly resolved.)
2395 let head_imports = crate::render::package_head_at_op(self, to_head)
2396 .map(|ph| {
2397 ph.flat_imports
2398 .into_iter()
2399 .map(|(reference, alias)| {
2400 Stage::Import(lex_ast::Import { reference, alias })
2401 })
2402 .collect::<Vec<_>>()
2403 })
2404 .unwrap_or_default();
2405 let mut stages = head_imports;
2406 stages.extend(decls);
2407 // #930: the hub gate resolves this head's external dependencies from
2408 // the lock committed with `to_head` (via the installed cross-store
2409 // resolver); empty when none is installed or the head is inlined.
2410 let modules = self.resolved_modules(&stages, Some(to_head));
2411 let module_types = self.resolved_module_types(&stages, Some(to_head));
2412 let dep_prefixes = self.resolved_module_prefixes(&stages, Some(to_head));
2413 let result = match lex_types::check_program_with_deps(&stages, &modules, &module_types, &dep_prefixes) {
2414 Ok(_) => lex_vcs::AttestationResult::Passed,
2415 Err(errors) => lex_vcs::AttestationResult::Failed {
2416 detail: serde_json::to_string(&errors).unwrap_or_else(|_| "type errors".into()),
2417 },
2418 };
2419 let passed = matches!(result, lex_vcs::AttestationResult::Passed);
2420
2421 // Stages introduced by THIS advance (from_head exclusive → to_head).
2422 let log = lex_vcs::OpLog::open(self.root())?;
2423 let to = to_head.to_string();
2424 let records = match from_head {
2425 Some(f) => log
2426 .walk_forward_since(&to, &f.to_string())?
2427 .unwrap_or_else(|| log.walk_forward(&to, None).unwrap_or_default()),
2428 None => log.walk_forward(&to, None)?,
2429 };
2430 let mut introduced: Vec<String> = Vec::new();
2431 for rec in &records {
2432 introduced.extend(attestable_stage_ids(&rec.produces));
2433 }
2434
2435 let alog = self.attestation_log()?;
2436 for sid in &introduced {
2437 let att = lex_vcs::Attestation::new(
2438 sid.clone(),
2439 Some(to_head.to_string()),
2440 None,
2441 lex_vcs::AttestationKind::TypeCheck,
2442 result.clone(),
2443 hub_ci_producer(),
2444 None,
2445 );
2446 alog.put(&att)?;
2447 }
2448
2449 let detail = match &result {
2450 lex_vcs::AttestationResult::Failed { detail } => Some(detail.clone()),
2451 _ => None,
2452 };
2453 Ok(HubCiVerdict { passed, checked_stages, attested_stages: introduced.len(), detail })
2454 }
2455
2456 /// Emit an `Examples::Passed` attestation for a published stage
2457 /// whose behavioral `examples {}` block was run and passed (#835,
2458 /// Tier 1). Mirrors [`Self::record_typecheck_passed`]. The
2459 /// behavioral run itself happens one layer up (lex-api / lex-cli)
2460 /// because it needs the bytecode compiler + VM, which this crate
2461 /// deliberately doesn't depend on; the store only records the
2462 /// verdict. `file_hash` uses the stage id — the stage fully
2463 /// determines its own examples.
2464 pub fn record_examples_passed(
2465 &self,
2466 stage_id: &str,
2467 op_id: &lex_vcs::OpId,
2468 count: usize,
2469 ) -> Result<(), StoreError> {
2470 let log = self.attestation_log()?;
2471 let attestation = lex_vcs::Attestation::new(
2472 stage_id.to_string(),
2473 Some(op_id.clone()),
2474 None,
2475 lex_vcs::AttestationKind::Examples { file_hash: stage_id.to_string(), count },
2476 lex_vcs::AttestationResult::Passed,
2477 examples_producer(),
2478 None,
2479 );
2480 log.put(&attestation)?;
2481 Ok(())
2482 }
2483
2484 /// Record a structured `Review` verdict on a stage (#836 G4).
2485 /// The verdict maps onto the attestation `result` so existing
2486 /// result-based tooling reads it: Approve->Passed,
2487 /// Reject->Failed, RequestChanges->Inconclusive.
2488 pub fn record_review(
2489 &self,
2490 stage_id: &str,
2491 op_id: Option<lex_vcs::OpId>,
2492 reviewer: &str,
2493 verdict: lex_vcs::ReviewVerdict,
2494 notes: Option<String>,
2495 ) -> Result<lex_vcs::AttestationId, StoreError> {
2496 let result = match verdict {
2497 lex_vcs::ReviewVerdict::Approve => lex_vcs::AttestationResult::Passed,
2498 lex_vcs::ReviewVerdict::Reject => lex_vcs::AttestationResult::Failed {
2499 detail: notes.clone().unwrap_or_else(|| "rejected".into()),
2500 },
2501 lex_vcs::ReviewVerdict::RequestChanges => lex_vcs::AttestationResult::Inconclusive {
2502 detail: notes.clone().unwrap_or_else(|| "changes requested".into()),
2503 },
2504 };
2505 let att = lex_vcs::Attestation::new(
2506 stage_id.to_string(),
2507 op_id,
2508 None,
2509 lex_vcs::AttestationKind::Review { reviewer: reviewer.to_string(), verdict, notes },
2510 result,
2511 review_producer(reviewer),
2512 None,
2513 );
2514 let id = att.attestation_id.clone();
2515 self.attestation_log()?.put(&att)?;
2516 Ok(id)
2517 }
2518
2519 /// The latest `Review` verdict recorded on a stage, if any
2520 /// (#836 G4). "Latest" is by attestation timestamp; ties keep the
2521 /// last one seen. Used by `promote_candidate` to honor a standing
2522 /// Reject.
2523 pub fn latest_review_verdict(
2524 &self,
2525 stage_id: &str,
2526 ) -> Result<Option<lex_vcs::ReviewVerdict>, StoreError> {
2527 let log = self.attestation_log()?;
2528 let mut latest: Option<(u64, lex_vcs::ReviewVerdict)> = None;
2529 for a in log.list_for_stage(&stage_id.to_string())? {
2530 if let lex_vcs::AttestationKind::Review { verdict, .. } = a.kind {
2531 if latest.as_ref().map(|(t, _)| a.timestamp >= *t).unwrap_or(true) {
2532 latest = Some((a.timestamp, verdict));
2533 }
2534 }
2535 }
2536 Ok(latest.map(|(_, v)| v))
2537 }
2538
2539 /// Consult `policy.session_budgets` for the op's session
2540 /// (resolved via `op.intent_id → Intent.session_id`) and
2541 /// refuse if applying would push the session's monotonic spend
2542 /// over the configured cap (#292 slice 3).
2543 ///
2544 /// Ops without an `intent_id`, or whose intent has no
2545 /// configured cap, return Ok without any disk read.
2546 fn check_session_budget(&self, op: &lex_vcs::Operation) -> Result<(), StoreError> {
2547 let Some(intent_id) = op.intent_id.as_deref() else {
2548 return Ok(());
2549 };
2550 let intent_log = lex_vcs::IntentLog::open(self.root())?;
2551 let Some(intent) = intent_log.get(&intent_id.to_string())? else {
2552 // Dangling intent — treat as "no session" and let it
2553 // sail through. Slice 1's ledger already documents
2554 // this as graceful-degradation semantics.
2555 return Ok(());
2556 };
2557 let policy = crate::policy::load(self.root())?.unwrap_or_default();
2558 let Some(cap) = policy.session_budgets.cap_for(&intent.session_id) else {
2559 return Ok(());
2560 };
2561 // Recompute the session's current spend + the contribution
2562 // from this op. Re-running the ledger walk on every gated
2563 // op is O(branch history); see #292 slice 1's note about
2564 // a future on-disk cache.
2565 let current = self.session_budget(&intent.session_id)?;
2566 let increment = crate::budget::monotonic_spend_of(&op.kind);
2567 let spent_after = current.spent.saturating_add(increment);
2568 if spent_after > cap {
2569 return Err(StoreError::BudgetExceeded {
2570 session_id: intent.session_id,
2571 cap,
2572 spent_after,
2573 });
2574 }
2575 Ok(())
2576 }
2577
2578 /// Emit `RepairHint` attestations for a TypeError-rejected op
2579 /// (#281). One per candidate stage in the transition. The hint
2580 /// records the *would-be* op_id (deterministic, content-
2581 /// addressed even though the op record was never persisted)
2582 /// and the structured errors.
2583 ///
2584 /// #306 slice 3: `suggested_transform` is populated from the
2585 /// static (rule_tag → likely_transform) table for the *first*
2586 /// error in the batch. The LLM-driven `lex repair --apply`
2587 /// flow can still overwrite this with a higher-quality
2588 /// suggestion; the static value is the floor, not the ceiling.
2589 ///
2590 /// Best-effort: a write failure here is swallowed by the
2591 /// caller (the original `TypeError` is the load-bearing
2592 /// signal; missing the hint is recoverable on a retry).
2593 fn record_repair_hint(
2594 &self,
2595 stage_ids: &[String],
2596 failed_op_id: &lex_vcs::OpId,
2597 errors: &[lex_types::TypeError],
2598 ) -> Result<(), StoreError> {
2599 if stage_ids.is_empty() {
2600 return Ok(());
2601 }
2602 let errors_json = serde_json::to_value(errors).map_err(StoreError::Serde)?;
2603 // #306 slice 3: look up the static suggested_transform for
2604 // the first error's rule_tag. Multiple errors per op are
2605 // possible — when they fire in lockstep (e.g. one bad let
2606 // binding propagates to several use sites), the first
2607 // error's rule_tag is usually the load-bearing one to fix.
2608 let suggested_transform = errors
2609 .first()
2610 .and_then(|e| lex_types::suggested_transform_for(e.rule_tag()));
2611 let log = self.attestation_log()?;
2612 for stage_id in stage_ids {
2613 let attestation = lex_vcs::Attestation::new(
2614 stage_id.clone(),
2615 None, // the failed op was never persisted; not the
2616 // attestation's op_id (which is for a
2617 // *successful* op).
2618 None,
2619 lex_vcs::AttestationKind::RepairHint {
2620 failed_op_id: failed_op_id.clone(),
2621 errors: errors_json.clone(),
2622 suggested_transform: suggested_transform.clone(),
2623 },
2624 lex_vcs::AttestationResult::Failed {
2625 detail: format!(
2626 "op {} rejected: {} type error(s)",
2627 failed_op_id,
2628 errors.len()
2629 ),
2630 },
2631 repair_hint_producer(),
2632 None,
2633 );
2634 log.put(&attestation)?;
2635 }
2636 Ok(())
2637 }
2638
2639 /// Emit `Trace` attestations linking an already-committed `op`
2640 /// to the run that produced it (#257). One attestation per
2641 /// produced stage (matching the `TypeCheck` emission contract
2642 /// — see [`Self::apply_operation_checked`]) with
2643 /// `op_id: Some(op_id)` set, so `lex trace --op <op_id>`
2644 /// surfaces the run.
2645 ///
2646 /// Returns the number of attestations emitted (zero for ops
2647 /// that produce no attestable stage, e.g. `Remove` /
2648 /// `ImportOnly`).
2649 ///
2650 /// Idempotent: re-emitting for the same
2651 /// `(run_id, root_target, op_id, stage_id, producer, result)`
2652 /// tuple dedups via content addressing.
2653 ///
2654 /// `op_id` must already exist in the op log — an unknown op
2655 /// surfaces as `StoreError::UnknownOp`.
2656 pub fn record_op_trace(
2657 &self,
2658 run_id: &str,
2659 root_target: &str,
2660 op_id: &lex_vcs::OpId,
2661 result: lex_vcs::AttestationResult,
2662 producer: lex_vcs::ProducerDescriptor,
2663 ) -> Result<usize, StoreError> {
2664 let log = lex_vcs::OpLog::open(self.root())?;
2665 let rec = log
2666 .get(op_id)?
2667 .ok_or_else(|| StoreError::UnknownOp(op_id.clone()))?;
2668 let stage_ids = attestable_stage_ids(&rec.produces);
2669 if stage_ids.is_empty() {
2670 return Ok(0);
2671 }
2672 let attlog = self.attestation_log()?;
2673 let mut emitted = 0;
2674 for stage_id in stage_ids {
2675 let attestation = lex_vcs::Attestation::new(
2676 stage_id,
2677 Some(op_id.clone()),
2678 None,
2679 lex_vcs::AttestationKind::Trace {
2680 run_id: run_id.into(),
2681 root_target: root_target.into(),
2682 },
2683 result.clone(),
2684 producer.clone(),
2685 None,
2686 );
2687 attlog.put(&attestation)?;
2688 emitted += 1;
2689 }
2690 Ok(emitted)
2691 }
2692
2693 /// Walk `ops_since(branch_head, base)` and emit per-stage
2694 /// `Trace` attestations for each new op, linking them to the
2695 /// run that produced them (#257). Used by `lex run --trace`
2696 /// after the VM exits: snapshot `base = branch_head` before
2697 /// the run, then call this with the post-run head.
2698 ///
2699 /// `base = None` means "every op currently reachable from the
2700 /// branch head" — generally not what you want for a single
2701 /// run; pass the pre-run head.
2702 ///
2703 /// Returns the total number of attestations emitted across
2704 /// every new op. Zero is the common case (the run committed no
2705 /// ops).
2706 ///
2707 /// Idempotent on the per-op level via [`Self::record_op_trace`].
2708 pub fn record_run_committed_ops_since(
2709 &self,
2710 run_id: &str,
2711 root_target: &str,
2712 branch: &str,
2713 base: Option<&lex_vcs::OpId>,
2714 result: lex_vcs::AttestationResult,
2715 producer: lex_vcs::ProducerDescriptor,
2716 ) -> Result<usize, StoreError> {
2717 let head = match self.get_branch(branch)?.and_then(|b| b.head_op) {
2718 Some(h) => h,
2719 None => return Ok(0),
2720 };
2721 let log = lex_vcs::OpLog::open(self.root())?;
2722 let new_ops = log.ops_since(&head, base)?;
2723 let mut total = 0;
2724 for rec in new_ops {
2725 total += self.record_op_trace(
2726 run_id,
2727 root_target,
2728 &rec.op_id,
2729 result.clone(),
2730 producer.clone(),
2731 )?;
2732 }
2733 Ok(total)
2734 }
2735
2736 /// Apply a typed `ReplaceMatchArm` transform (#280) and emit a
2737 /// `OperationKind::ReplaceMatchArm` op that records the
2738 /// semantic shape of the edit, not just the byte effect.
2739 ///
2740 /// Steps:
2741 /// 1. Load the source stage's canonical bytes (delta-aware).
2742 /// 2. Run [`lex_ast::replace_match_arm`] to produce the new
2743 /// `Stage`. Pure function, no I/O.
2744 /// 3. Publish the new stage. Idempotent on the
2745 /// content-addressed `to_stage_id`.
2746 /// 4. Assemble the candidate program (every active stage on
2747 /// the branch, with the rewritten one swapped in) and call
2748 /// [`Self::apply_operation_checked`] — re-typechecks and
2749 /// runs every existing gate (TypeCheck attestation,
2750 /// required_attestations, producer-block walk-back).
2751 ///
2752 /// Failure modes:
2753 /// * [`StoreError::TransformError`] — transform didn't apply.
2754 /// The branch is unchanged; no stage published.
2755 /// * [`StoreError::TypeError`] — transform produced an
2756 /// ill-typed program. The new stage is on disk (idempotent
2757 /// on its content hash) but the branch is unchanged. Same
2758 /// "publish without advance" semantics as #245.
2759 /// * Everything else from `apply_operation_checked`.
2760 pub fn apply_replace_match_arm(
2761 &self,
2762 branch: &str,
2763 from_stage_id: &str,
2764 match_node: &lex_ast::NodeId,
2765 arm_index: usize,
2766 new_body: lex_ast::CExpr,
2767 ) -> Result<lex_vcs::OpId, StoreError> {
2768 let from_stage = self.get_ast(from_stage_id)?;
2769 let new_stage = lex_ast::replace_match_arm(&from_stage, match_node, arm_index, new_body)
2770 .map_err(StoreError::TransformError)?;
2771 let sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2772 let to_stage_id = self.publish(&new_stage)?;
2773 if to_stage_id == from_stage_id {
2774 // No-op transform — the new body was structurally
2775 // identical to the old. Refuse rather than advancing
2776 // the branch with an empty edit.
2777 return Err(StoreError::InvalidTransition(format!(
2778 "replace_match_arm produced the same stage_id `{from_stage_id}`"
2779 )));
2780 }
2781
2782 // Assemble the candidate program: every active stage on
2783 // the branch, with `from_stage_id` swapped for `new_stage`.
2784 let head = self.branch_head(branch)?;
2785 let mut candidate: Vec<lex_ast::Stage> = Vec::with_capacity(head.len());
2786 for (other_sig, other_stage_id) in &head {
2787 if other_sig == &sig {
2788 candidate.push(new_stage.clone());
2789 } else {
2790 candidate.push(self.get_ast(other_stage_id)?);
2791 }
2792 }
2793 // If the source sig isn't on the current branch head, the
2794 // transform is operating on a stage that hasn't been added
2795 // yet — refuse rather than risking a candidate program
2796 // that doesn't reflect the branch's actual state.
2797 if !head.contains_key(&sig) {
2798 return Err(StoreError::InvalidTransition(format!(
2799 "sig `{sig}` not on branch `{branch}`'s head"
2800 )));
2801 }
2802
2803 // #247: budget delta captured for `lex op log --budget-drift`.
2804 let from_budget = budget_of_stage(&from_stage);
2805 let to_budget = budget_of_stage(&new_stage);
2806
2807 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2808 let kind = lex_vcs::OperationKind::ReplaceMatchArm {
2809 sig_id: sig.clone(),
2810 from_stage_id: from_stage_id.to_string(),
2811 to_stage_id: to_stage_id.clone(),
2812 match_node: match_node.as_str().to_string(),
2813 arm_index,
2814 from_budget,
2815 to_budget,
2816 };
2817 let transition = lex_vcs::StageTransition::Replace {
2818 sig_id: sig.clone(),
2819 from: from_stage_id.to_string(),
2820 to: to_stage_id.clone(),
2821 };
2822 let op = lex_vcs::Operation::new(kind, head_now.into_iter().collect::<Vec<_>>());
2823 self.apply_operation_checked(branch, op, transition, &candidate)
2824 }
2825
2826 /// Apply a typed `RenameLocal` transform (#280) — rename a
2827 /// `let`-bound local within a fn body and emit a matching
2828 /// `OperationKind::RenameLocal`. Same end-to-end shape as
2829 /// [`Self::apply_replace_match_arm`]; see that method for the
2830 /// failure-mode taxonomy.
2831 pub fn apply_rename_local(
2832 &self,
2833 branch: &str,
2834 from_stage_id: &str,
2835 let_node: &lex_ast::NodeId,
2836 new_name: &str,
2837 ) -> Result<lex_vcs::OpId, StoreError> {
2838 let from_stage = self.get_ast(from_stage_id)?;
2839 // Read the old name before running the transform, so the
2840 // op log records the rename target rather than just the
2841 // new value.
2842 let old_name = read_let_name(&from_stage, let_node).map_err(StoreError::TransformError)?;
2843 let new_stage = lex_ast::rename_local(&from_stage, let_node, new_name)
2844 .map_err(StoreError::TransformError)?;
2845 let sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2846 let to_stage_id = self.publish(&new_stage)?;
2847 if to_stage_id == from_stage_id {
2848 return Err(StoreError::InvalidTransition(format!(
2849 "rename_local produced the same stage_id `{from_stage_id}`"
2850 )));
2851 }
2852 let head = self.branch_head(branch)?;
2853 let mut candidate: Vec<lex_ast::Stage> = Vec::with_capacity(head.len());
2854 for (other_sig, other_stage_id) in &head {
2855 if other_sig == &sig {
2856 candidate.push(new_stage.clone());
2857 } else {
2858 candidate.push(self.get_ast(other_stage_id)?);
2859 }
2860 }
2861 if !head.contains_key(&sig) {
2862 return Err(StoreError::InvalidTransition(format!(
2863 "sig `{sig}` not on branch `{branch}`'s head"
2864 )));
2865 }
2866 let from_budget = budget_of_stage(&from_stage);
2867 let to_budget = budget_of_stage(&new_stage);
2868 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2869 let kind = lex_vcs::OperationKind::RenameLocal {
2870 sig_id: sig.clone(),
2871 from_stage_id: from_stage_id.to_string(),
2872 to_stage_id: to_stage_id.clone(),
2873 let_node: let_node.as_str().to_string(),
2874 old_name,
2875 new_name: new_name.to_string(),
2876 from_budget,
2877 to_budget,
2878 };
2879 let transition = lex_vcs::StageTransition::Replace {
2880 sig_id: sig.clone(),
2881 from: from_stage_id.to_string(),
2882 to: to_stage_id.clone(),
2883 };
2884 let op = lex_vcs::Operation::new(kind, head_now.into_iter().collect::<Vec<_>>());
2885 self.apply_operation_checked(branch, op, transition, &candidate)
2886 }
2887
2888 /// Apply a typed `InlineLet` transform (#280) — eliminate a
2889 /// `let x := v; body` by substituting `v` for every unshadowed
2890 /// `x` in `body`, then replacing the `Let` node with the
2891 /// substituted body. Same end-to-end shape as
2892 /// [`Self::apply_replace_match_arm`].
2893 pub fn apply_inline_let(
2894 &self,
2895 branch: &str,
2896 from_stage_id: &str,
2897 let_node: &lex_ast::NodeId,
2898 ) -> Result<lex_vcs::OpId, StoreError> {
2899 let from_stage = self.get_ast(from_stage_id)?;
2900 let binding_name =
2901 read_let_name(&from_stage, let_node).map_err(StoreError::TransformError)?;
2902 let new_stage =
2903 lex_ast::inline_let(&from_stage, let_node).map_err(StoreError::TransformError)?;
2904 let sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2905 let to_stage_id = self.publish(&new_stage)?;
2906 if to_stage_id == from_stage_id {
2907 return Err(StoreError::InvalidTransition(format!(
2908 "inline_let produced the same stage_id `{from_stage_id}`"
2909 )));
2910 }
2911 let head = self.branch_head(branch)?;
2912 let mut candidate: Vec<lex_ast::Stage> = Vec::with_capacity(head.len());
2913 for (other_sig, other_stage_id) in &head {
2914 if other_sig == &sig {
2915 candidate.push(new_stage.clone());
2916 } else {
2917 candidate.push(self.get_ast(other_stage_id)?);
2918 }
2919 }
2920 if !head.contains_key(&sig) {
2921 return Err(StoreError::InvalidTransition(format!(
2922 "sig `{sig}` not on branch `{branch}`'s head"
2923 )));
2924 }
2925 let from_budget = budget_of_stage(&from_stage);
2926 let to_budget = budget_of_stage(&new_stage);
2927 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
2928 let kind = lex_vcs::OperationKind::InlineLet {
2929 sig_id: sig.clone(),
2930 from_stage_id: from_stage_id.to_string(),
2931 to_stage_id: to_stage_id.clone(),
2932 let_node: let_node.as_str().to_string(),
2933 binding_name,
2934 from_budget,
2935 to_budget,
2936 };
2937 let transition = lex_vcs::StageTransition::Replace {
2938 sig_id: sig.clone(),
2939 from: from_stage_id.to_string(),
2940 to: to_stage_id.clone(),
2941 };
2942 let op = lex_vcs::Operation::new(kind, head_now.into_iter().collect::<Vec<_>>());
2943 self.apply_operation_checked(branch, op, transition, &candidate)
2944 }
2945
2946 /// Apply a typed `ExtractFunction` transform (#280 slice 4) —
2947 /// extract a sub-expression of `from_stage_id`'s body into a
2948 /// new top-level fn defined by `spec`, and emit two ops tied
2949 /// together by a shared synthetic Intent so `lex op log
2950 /// --intent <id>` groups them.
2951 ///
2952 /// The two ops:
2953 /// 1. `AddFunction { sig_id: <new_fn_sig>, stage_id: <new_fn_stage> }`
2954 /// 2. `ModifyBody { sig_id: <source_sig>, from_stage_id, to_stage_id: <modified> }`
2955 ///
2956 /// The shared Intent's prompt is structured (`extract_function:
2957 /// <new_fn_name>` plus the source identity) so downstream
2958 /// tooling can recover the typed-transform shape from the
2959 /// op-log + intent-log join.
2960 ///
2961 /// Returns `(add_fn_op_id, modify_body_op_id)`.
2962 pub fn apply_extract_function(
2963 &self,
2964 branch: &str,
2965 from_stage_id: &str,
2966 expr_node: &lex_ast::NodeId,
2967 spec: lex_ast::ExtractFnSpec,
2968 ) -> Result<(lex_vcs::OpId, lex_vcs::OpId), StoreError> {
2969 let from_stage = self.get_ast(from_stage_id)?;
2970 let new_fn_name = spec.name.clone();
2971 let (modified_stage, new_fn_stage) =
2972 lex_ast::extract_function(&from_stage, expr_node, spec)
2973 .map_err(StoreError::TransformError)?;
2974
2975 let source_sig = lex_ast::sig_id(&from_stage).ok_or(StoreError::CannotPublishImport)?;
2976 let new_fn_sig = lex_ast::sig_id(&new_fn_stage).ok_or(StoreError::CannotPublishImport)?;
2977 if source_sig == new_fn_sig {
2978 return Err(StoreError::InvalidTransition(format!(
2979 "extract_function produced a sig matching the source `{source_sig}`"
2980 )));
2981 }
2982 let new_fn_stage_id = self.publish(&new_fn_stage)?;
2983 let modified_stage_id = self.publish(&modified_stage)?;
2984 if modified_stage_id == from_stage_id {
2985 return Err(StoreError::InvalidTransition(format!(
2986 "extract_function produced the same stage_id `{from_stage_id}` for the source"
2987 )));
2988 }
2989
2990 let head = self.branch_head(branch)?;
2991 if !head.contains_key(&source_sig) {
2992 return Err(StoreError::InvalidTransition(format!(
2993 "sig `{source_sig}` not on branch `{branch}`'s head"
2994 )));
2995 }
2996
2997 // Synthesize an Intent linking the two ops. The session_id
2998 // / model fields here are not load-bearing — they exist to
2999 // make the IntentId content-addressed; downstream tooling
3000 // reads `prompt` to reconstruct the typed-transform shape.
3001 let intent = lex_vcs::Intent::new(
3002 format!(
3003 "[lex.transform.extract_function]\nnew_fn={new_fn_name}\nsource_sig={source_sig}\nfrom_stage={from_stage_id}\nexpr_node={node}",
3004 node = expr_node.as_str(),
3005 ),
3006 "lex-store::apply_extract_function",
3007 lex_vcs::ModelDescriptor {
3008 provider: "lex-store".into(),
3009 name: env!("CARGO_PKG_VERSION").into(),
3010 version: None,
3011 },
3012 None,
3013 );
3014 let intent_id = intent.intent_id.clone();
3015 lex_vcs::IntentLog::open(self.root())?.put(&intent)?;
3016
3017 // Step 1 — emit the AddFunction op for the new fn. Build
3018 // the candidate program by appending the new fn to every
3019 // stage on the current branch head.
3020 let new_fn_effects: std::collections::BTreeSet<String> = match &new_fn_stage {
3021 lex_ast::Stage::FnDecl(fd) => fd.effects.iter().map(|e| e.name.clone()).collect(),
3022 _ => Default::default(),
3023 };
3024 let new_fn_budget = budget_of_stage(&new_fn_stage);
3025 let mut candidate_with_new_fn: Vec<lex_ast::Stage> = Vec::with_capacity(head.len() + 1);
3026 for stage_id in head.values() {
3027 candidate_with_new_fn.push(self.get_ast(stage_id)?);
3028 }
3029 candidate_with_new_fn.push(new_fn_stage.clone());
3030 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
3031 let add_op = lex_vcs::Operation::new(
3032 lex_vcs::OperationKind::AddFunction {
3033 sig_id: new_fn_sig.clone(),
3034 stage_id: new_fn_stage_id.clone(),
3035 effects: new_fn_effects,
3036 budget_cost: new_fn_budget,
3037 // Single-op apply path — no package context here.
3038 in_file: None,
3039 },
3040 head_now.into_iter().collect::<Vec<_>>(),
3041 )
3042 .with_intent(intent_id.clone());
3043 let add_transition = lex_vcs::StageTransition::Create {
3044 sig_id: new_fn_sig.clone(),
3045 stage_id: new_fn_stage_id.clone(),
3046 };
3047 let add_op_id =
3048 self.apply_operation_checked(branch, add_op, add_transition, &candidate_with_new_fn)?;
3049
3050 // Step 2 — emit the ModifyBody op for the source. Build
3051 // the candidate program by replacing the source's stage
3052 // with `modified_stage` and keeping the new fn alongside.
3053 let from_budget = budget_of_stage(&from_stage);
3054 let to_budget = budget_of_stage(&modified_stage);
3055 let mut candidate_with_modified: Vec<lex_ast::Stage> = Vec::with_capacity(head.len() + 1);
3056 for (other_sig, other_stage_id) in &head {
3057 if other_sig == &source_sig {
3058 candidate_with_modified.push(modified_stage.clone());
3059 } else {
3060 candidate_with_modified.push(self.get_ast(other_stage_id)?);
3061 }
3062 }
3063 candidate_with_modified.push(new_fn_stage.clone());
3064 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
3065 let modify_op = lex_vcs::Operation::new(
3066 lex_vcs::OperationKind::ModifyBody {
3067 sig_id: source_sig.clone(),
3068 from_stage_id: from_stage_id.to_string(),
3069 to_stage_id: modified_stage_id.clone(),
3070 from_budget,
3071 to_budget,
3072 },
3073 head_now.into_iter().collect::<Vec<_>>(),
3074 )
3075 .with_intent(intent_id);
3076 let modify_transition = lex_vcs::StageTransition::Replace {
3077 sig_id: source_sig,
3078 from: from_stage_id.to_string(),
3079 to: modified_stage_id,
3080 };
3081 let modify_op_id = self.apply_operation_checked(
3082 branch,
3083 modify_op,
3084 modify_transition,
3085 &candidate_with_modified,
3086 )?;
3087
3088 Ok((add_op_id, modify_op_id))
3089 }
3090
3091 /// Propose a stage for `sig_id` without advancing the branch
3092 /// head (#294). Multiple agents can call this concurrently
3093 /// for the same sig — every call lands a fresh `Candidate`
3094 /// op chained off the current head_op. The branch head stays
3095 /// where it was; a later [`Self::promote_candidate`] picks
3096 /// the winner.
3097 ///
3098 /// The caller is responsible for typechecking `new_stage`
3099 /// against whatever program context they consider valid —
3100 /// `propose_candidate` doesn't run the gate. Type errors
3101 /// surface at promotion time, where the candidate is
3102 /// composed back into a candidate program via the standard
3103 /// `apply_operation_checked` path.
3104 ///
3105 /// The stage is published (idempotent on content hash). The
3106 /// `intent_id` is required so downstream consumers can
3107 /// distinguish proposals by author.
3108 pub fn propose_candidate(
3109 &self,
3110 branch: &str,
3111 new_stage: &lex_ast::Stage,
3112 intent_id: &lex_vcs::IntentId,
3113 ) -> Result<lex_vcs::OpId, StoreError> {
3114 let sig = lex_ast::sig_id(new_stage).ok_or(StoreError::CannotPublishImport)?;
3115 let stage_id = self.publish(new_stage)?;
3116 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
3117 let op = lex_vcs::Operation::new(
3118 lex_vcs::OperationKind::Candidate {
3119 sig_id: sig,
3120 stage_id,
3121 },
3122 head_now.into_iter().collect::<Vec<_>>(),
3123 )
3124 .with_intent(intent_id.clone());
3125 let transition = lex_vcs::StageTransition::ImportOnly;
3126 self.apply_operation(branch, op, transition)
3127 }
3128
3129 /// List every live `Candidate` op for `sig_id` — i.e. those
3130 /// not yet referenced by any `Promote` op (either as the
3131 /// winner or in the `supersedes` set). Used by `lex stage
3132 /// candidates`. Results are sorted by op_id for
3133 /// reproducibility.
3134 pub fn list_candidates(&self, sig_id: &str) -> Result<Vec<CandidateInfo>, StoreError> {
3135 let log = lex_vcs::OpLog::open(self.root())?;
3136 let all = log.list_all()?;
3137 // Collect the set of candidate op_ids referenced by any
3138 // Promote for this sig. Those candidates are no longer
3139 // live.
3140 let mut referenced: std::collections::BTreeSet<lex_vcs::OpId> = Default::default();
3141 for rec in &all {
3142 if let lex_vcs::OperationKind::Promote {
3143 sig_id: s,
3144 winner_candidate,
3145 supersedes,
3146 ..
3147 } = &rec.op.kind
3148 {
3149 if s != sig_id {
3150 continue;
3151 }
3152 referenced.insert(winner_candidate.clone());
3153 for sup in supersedes {
3154 referenced.insert(sup.clone());
3155 }
3156 }
3157 }
3158 let mut out: Vec<CandidateInfo> = Vec::new();
3159 for rec in all {
3160 let lex_vcs::OperationKind::Candidate {
3161 sig_id: s,
3162 stage_id,
3163 } = &rec.op.kind
3164 else {
3165 continue;
3166 };
3167 if s != sig_id {
3168 continue;
3169 }
3170 if referenced.contains(&rec.op_id) {
3171 continue;
3172 }
3173 out.push(CandidateInfo {
3174 op_id: rec.op_id.clone(),
3175 stage_id: stage_id.clone(),
3176 intent_id: rec.op.intent_id.clone(),
3177 });
3178 }
3179 out.sort_by(|a, b| a.op_id.cmp(&b.op_id));
3180 Ok(out)
3181 }
3182
3183 /// Promote a previously-landed `Candidate` op as the new
3184 /// branch head for its sig (#294). Emits a `Promote` op
3185 /// listing every other live `Candidate` for the same sig
3186 /// in its `supersedes` field. After this lands,
3187 /// [`Self::list_candidates`] returns an empty set for the
3188 /// sig.
3189 ///
3190 /// Re-typechecks the candidate program (winner stage + the
3191 /// rest of the branch) through `apply_operation_checked`, so
3192 /// a candidate that doesn't compose with the current branch
3193 /// state surfaces as `StoreError::TypeError`.
3194 pub fn promote_candidate(
3195 &self,
3196 branch: &str,
3197 candidate_op_id: &lex_vcs::OpId,
3198 ) -> Result<lex_vcs::OpId, StoreError> {
3199 let log = lex_vcs::OpLog::open(self.root())?;
3200 let candidate_rec = log
3201 .get(candidate_op_id)?
3202 .ok_or_else(|| StoreError::UnknownOp(candidate_op_id.clone()))?;
3203 let (sig, winner_stage_id) = match &candidate_rec.op.kind {
3204 lex_vcs::OperationKind::Candidate { sig_id, stage_id } => {
3205 (sig_id.clone(), stage_id.clone())
3206 }
3207 other => {
3208 return Err(StoreError::InvalidTransition(format!(
3209 "op `{candidate_op_id}` is a `{:?}`, not a Candidate",
3210 other
3211 )))
3212 }
3213 };
3214
3215 // #836 G4: a candidate carrying a standing `Reject` review must
3216 // not be promoted. "Standing" = the latest `Review` on the
3217 // winner's stage is a Reject; a later `Approve` (or
3218 // `RequestChanges`, which is advisory, not a veto) lifts it.
3219 // Safe by default: a candidate with no review, or an approved
3220 // one, promotes exactly as before.
3221 if let Some(lex_vcs::ReviewVerdict::Reject) = self.latest_review_verdict(&winner_stage_id)? {
3222 return Err(StoreError::InvalidTransition(format!(
3223 "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"
3224 )));
3225 }
3226
3227 // Gather every OTHER live candidate for this sig — the
3228 // ones this Promote will supersede.
3229 let live = self.list_candidates(&sig)?;
3230 let mut supersedes: Vec<lex_vcs::OpId> = live
3231 .iter()
3232 .filter(|c| &c.op_id != candidate_op_id)
3233 .map(|c| c.op_id.clone())
3234 .collect();
3235 supersedes.sort();
3236
3237 // Assemble candidate program: winner stage in place of
3238 // the sig's current head (if any), plus every other sig
3239 // unchanged.
3240 let head = self.branch_head(branch)?;
3241 let winner_stage = self.get_ast(&winner_stage_id)?;
3242 let mut candidate_program: Vec<lex_ast::Stage> = Vec::with_capacity(head.len() + 1);
3243 let mut found = false;
3244 for (other_sig, other_stage_id) in &head {
3245 if other_sig == &sig {
3246 candidate_program.push(winner_stage.clone());
3247 found = true;
3248 } else {
3249 candidate_program.push(self.get_ast(other_stage_id)?);
3250 }
3251 }
3252 if !found {
3253 // Sig doesn't have a head yet — append the winner
3254 // stage to make it a Create.
3255 candidate_program.push(winner_stage.clone());
3256 }
3257 let from_stage_id = head.get(&sig).cloned();
3258 // Budget delta from old head to winner — same shape as
3259 // ModifyBody.
3260 let from_budget = from_stage_id
3261 .as_deref()
3262 .and_then(|s| self.get_ast(s).ok())
3263 .and_then(|s| budget_of_stage(&s));
3264 let to_budget = budget_of_stage(&winner_stage);
3265
3266 let head_now = self.get_branch(branch)?.and_then(|b| b.head_op);
3267 let op = lex_vcs::Operation::new(
3268 lex_vcs::OperationKind::Promote {
3269 sig_id: sig.clone(),
3270 winner_candidate: candidate_op_id.clone(),
3271 winner_stage_id: winner_stage_id.clone(),
3272 supersedes,
3273 from_stage_id: from_stage_id.clone(),
3274 from_budget,
3275 to_budget,
3276 },
3277 head_now.into_iter().collect::<Vec<_>>(),
3278 );
3279 let transition = match &from_stage_id {
3280 Some(from) => lex_vcs::StageTransition::Replace {
3281 sig_id: sig,
3282 from: from.clone(),
3283 to: winner_stage_id,
3284 },
3285 None => lex_vcs::StageTransition::Create {
3286 sig_id: sig,
3287 stage_id: winner_stage_id,
3288 },
3289 };
3290 self.apply_operation_checked(branch, op, transition, &candidate_program)
3291 }
3292
3293 /// `set_branch_head_op` for the durability story on the branch
3294 /// file itself.
3295 pub fn apply_operation(
3296 &self,
3297 branch: &str,
3298 op: lex_vcs::Operation,
3299 transition: lex_vcs::StageTransition,
3300 ) -> Result<lex_vcs::OpId, StoreError> {
3301 let attestable = attestable_stage_ids(&transition);
3302 let op_effects = op_declared_effects(&op.kind);
3303 self.cas_retry_advance(branch, op, transition, |new_head| {
3304 self.run_required_attestations_gate(branch, &new_head.op_id, &attestable, &op_effects)
3305 })
3306 }
3307
3308 /// CAS retry loop for #262. Single-parent ops are rebuilt on
3309 /// each iteration with the current branch head as parent;
3310 /// the per-iteration callback runs the gate (and TypeCheck
3311 /// emission, for the checked path) between persist and CAS.
3312 /// Merge ops (with 2 parents already set) skip the rebuild —
3313 /// their parents are caller-supplied and meaningful — and get
3314 /// a single attempt; on CAS failure they surface `Contention`.
3315 fn cas_retry_advance<F>(
3316 &self,
3317 branch: &str,
3318 op: lex_vcs::Operation,
3319 transition: lex_vcs::StageTransition,
3320 mut between_persist_and_cas: F,
3321 ) -> Result<lex_vcs::OpId, StoreError>
3322 where
3323 F: FnMut(&lex_vcs::NewHead) -> Result<(), StoreError>,
3324 {
3325 // 32 retries handles up to ~32 concurrent writers racing on
3326 // the same branch tip. Beyond that, surfacing `Contention`
3327 // is the right signal — clients should back off or batch.
3328 const MAX_ATTEMPTS: u32 = 32;
3329 // Single-parent ops can be rebuilt on retry; merge ops
3330 // can't (their two parents are meaningful, supplied by the
3331 // merge engine). For merges, single attempt: if CAS
3332 // fails, surface Contention.
3333 let is_rebuildable = op.parents.len() <= 1;
3334 let kind = op.kind.clone();
3335 let intent_id = op.intent_id.clone();
3336
3337 let mut last_io_err: Option<StoreError> = None;
3338 let mut current_op = op;
3339 let current_transition = transition;
3340 // Only rebuild on retries — attempt 1 honors the caller's
3341 // exact op so a user-supplied bogus parent (parents =
3342 // ["someone-else"]) surfaces as `StaleParent` instead of
3343 // being silently corrected.
3344 //
3345 // Exception (#262 follow-up): an op with `parents = []`
3346 // means "I don't care; chain off whatever the current
3347 // head is." Under concurrent apply, attempt 1 can read
3348 // `head_op = Some(opA)` after a sibling writer landed,
3349 // and the persist's parent check fails StaleParent
3350 // unprompted. Rebuild attempt 1 for the empty-parents
3351 // case so the legitimate-race path retries cleanly.
3352 let mut rebuilt_already = false;
3353 for attempt in 1..=MAX_ATTEMPTS {
3354 // Read the current head BEFORE we persist — this is
3355 // the value we'll compare against in the CAS.
3356 let parent = self.get_branch(branch)?.and_then(|b| b.head_op);
3357
3358 // Rebuild the op against the current head, but only
3359 // on retries (not the caller's first attempt) and
3360 // only for single-parent operations. Multi-parent
3361 // (merge) ops are passed through unchanged.
3362 //
3363 // Empty-parents ops also rebuild on attempt 1 (see
3364 // the exception note above) so concurrent apply
3365 // doesn't false-positive on StaleParent.
3366 let should_rebuild = is_rebuildable
3367 && (rebuilt_already || (current_op.parents.is_empty() && parent.is_some()));
3368 if should_rebuild {
3369 current_op = lex_vcs::Operation {
3370 kind: kind.clone(),
3371 parents: parent.iter().cloned().collect(),
3372 intent_id: intent_id.clone(),
3373 };
3374 }
3375
3376 // Persist (idempotent). On `StaleParent` from a retry
3377 // attempt (where we already rebuilt), the head changed
3378 // between our `get_branch` and this `lex_vcs::apply`
3379 // — race; rebuild and continue. On `StaleParent` from
3380 // attempt 1 (caller's input), propagate.
3381 let new_head = match self.persist_op_only_with_parent(
3382 branch,
3383 parent.as_ref(),
3384 current_op.clone(),
3385 current_transition.clone(),
3386 ) {
3387 Ok(nh) => nh,
3388 Err(StoreError::Apply(lex_vcs::ApplyError::StaleParent { .. }))
3389 if is_rebuildable && rebuilt_already =>
3390 {
3391 rebuilt_already = true;
3392 continue;
3393 }
3394 Err(e) => return Err(e),
3395 };
3396
3397 // Run the caller's between-persist-and-cas hook
3398 // (TypeCheck emission + gate). If this fails, the op
3399 // record is durable but orphaned — same semantics as
3400 // pre-#262.
3401 between_persist_and_cas(&new_head)?;
3402
3403 // CAS the branch head. On success: done. On mismatch:
3404 // someone advanced in parallel; retry.
3405 match self.set_branch_head_op_cas(branch, parent, new_head.op_id.clone()) {
3406 Ok(()) => return Ok(new_head.op_id),
3407 Err(crate::branches::CasFailed::Mismatch { .. }) if is_rebuildable => {
3408 // Try again with the new head as parent.
3409 rebuilt_already = true;
3410 continue;
3411 }
3412 Err(crate::branches::CasFailed::Mismatch { .. }) => {
3413 // Merge op: surface immediately — we can't
3414 // rebuild without rerunning the merge engine.
3415 let _ = attempt;
3416 return Err(StoreError::Contention {
3417 branch: branch.into(),
3418 attempts: 1,
3419 });
3420 }
3421 Err(crate::branches::CasFailed::UnknownBranch(b)) => {
3422 return Err(StoreError::UnknownBranch(b));
3423 }
3424 Err(crate::branches::CasFailed::Io(e)) => {
3425 last_io_err = Some(StoreError::Io(std::io::Error::other(e)));
3426 continue;
3427 }
3428 }
3429 }
3430 // Retries exhausted. Prefer surfacing the most recent IO
3431 // error if we hit one; otherwise it's pure CAS contention.
3432 match last_io_err {
3433 Some(e) => Err(e),
3434 None => Err(StoreError::Contention {
3435 branch: branch.into(),
3436 attempts: MAX_ATTEMPTS,
3437 }),
3438 }
3439 }
3440
3441 /// Persist an op against an explicitly-supplied parent. Used
3442 /// by the CAS retry loop in `cas_retry_advance` so the
3443 /// `lex_vcs::apply` parent check matches what we read at the
3444 /// top of the loop iteration (avoids a TOCTOU race against
3445 /// `persist_op_only`'s second read).
3446 fn persist_op_only_with_parent(
3447 &self,
3448 branch: &str,
3449 parent: Option<&lex_vcs::OpId>,
3450 op: lex_vcs::Operation,
3451 transition: lex_vcs::StageTransition,
3452 ) -> Result<lex_vcs::NewHead, StoreError> {
3453 if branch != DEFAULT_BRANCH && self.get_branch(branch)?.is_none() {
3454 return Err(StoreError::UnknownBranch(branch.into()));
3455 }
3456 let log = lex_vcs::OpLog::open(self.root())?;
3457 lex_vcs::apply(&log, parent, op, transition).map_err(|e| match e {
3458 lex_vcs::ApplyError::Persist(io) => StoreError::Io(io),
3459 other => StoreError::Apply(other),
3460 })
3461 }
3462
3463 /// Run the `required_attestations` gate (#245) and the
3464 /// retroactive producer-block gate (#248) over a single op
3465 /// against the store's `policy.json` and attestation log.
3466 ///
3467 /// Failure modes (in order):
3468 ///
3469 /// 1. Producer-block first: if any attestation on the op's
3470 /// stage is from a quarantined tool, refuse with
3471 /// `ProducerBlocked` (#248). Surfaces *before* the
3472 /// required-attestations gate so a clearly-malicious record
3473 /// isn't masked by a missing-Spec error.
3474 /// 2. Required-attestations next: if any required attestation
3475 /// kind is missing, refuse with `BranchAdvanceBlocked`
3476 /// (#245).
3477 ///
3478 /// Loads the policy / attestation log lazily; with no policy
3479 /// file and no `ProducerBlock` attestations the gate is a no-op
3480 /// (default-permissive — matches pre-#245 stores).
3481 fn run_required_attestations_gate(
3482 &self,
3483 branch: &str,
3484 op_id: &lex_vcs::OpId,
3485 stage_ids: &[String],
3486 op_effects: &std::collections::BTreeSet<String>,
3487 ) -> Result<(), StoreError> {
3488 // Build the candidate slice for the new op. Ops with no
3489 // attestable stage (imports, empty merges) get a single
3490 // `None`-stage tuple; both gates skip those.
3491 let new_op_candidate: Vec<(
3492 lex_vcs::OpId,
3493 Option<String>,
3494 std::collections::BTreeSet<String>,
3495 )> = if stage_ids.is_empty() {
3496 vec![(op_id.clone(), None, op_effects.clone())]
3497 } else {
3498 stage_ids
3499 .iter()
3500 .map(|sid| (op_id.clone(), Some(sid.clone()), op_effects.clone()))
3501 .collect()
3502 };
3503 let attest_log = self.attestation_log()?;
3504
3505 // #248 + #256: producer-block gate, walk-back style.
3506 //
3507 // The naive #248 gate only checked the new op's stage. That
3508 // missed contamination on ancestors — once `lex attest
3509 // retro-block` lands, every previously-gated op stays in
3510 // the chain even though its attestations are now from a
3511 // quarantined producer.
3512 //
3513 // #256 fixes this by walking the chain from `head_op` back
3514 // to `last_gate_checkpoint` (or genesis when the checkpoint
3515 // is invalidated), collecting each ancestor's attestable
3516 // stages, and running `check_producer_block` on the
3517 // combined set. After a successful advance,
3518 // `set_branch_head_op` moves the checkpoint to the new
3519 // head (steady-state O(new ops) per advance).
3520 let walk_back_candidate = self.collect_ancestor_candidates(branch)?;
3521 let mut producer_block_candidate = walk_back_candidate;
3522 producer_block_candidate.extend(new_op_candidate.iter().cloned());
3523 crate::policy::check_producer_block(&attest_log, &producer_block_candidate)
3524 .map_err(StoreError::ProducerBlocked)?;
3525
3526 // #245: required-attestations gate. Forward-going only —
3527 // only the new op is checked. Walking back makes no sense
3528 // here: the policy is "this advance must carry these
3529 // attestations," not "every prior op must have."
3530 let policy = match crate::policy::load(self.root())? {
3531 Some(p) if !p.required_attestations.is_empty() => p,
3532 _ => return Ok(()),
3533 };
3534 let waivers =
3535 crate::policy::check_required_attestations(&attest_log, &new_op_candidate, &policy)
3536 .map_err(StoreError::BranchAdvanceBlocked)?;
3537 // #293: emit one `TrustWaived` attestation per waiver so
3538 // the audit trail records every skip. Idempotent on
3539 // attestation_id (content-addressed dedup) — re-running
3540 // the gate with the same state writes the same files.
3541 for w in waivers {
3542 let att = lex_vcs::Attestation::new(
3543 w.stage_id,
3544 Some(op_id.clone()),
3545 None,
3546 lex_vcs::AttestationKind::TrustWaived {
3547 producer: w.producer,
3548 score_thousandths: w.score_thousandths,
3549 threshold_thousandths: w.threshold_thousandths,
3550 kind_tag: w.kind_tag,
3551 },
3552 lex_vcs::AttestationResult::Passed,
3553 trust_waived_producer(),
3554 None,
3555 );
3556 attest_log.put(&att)?;
3557 }
3558 Ok(())
3559 }
3560
3561 /// Walk the branch from `head_op` back to `last_gate_checkpoint`
3562 /// (exclusive) and return the `(op_id, stage_id, op_effects)`
3563 /// tuples for every attestable stage touched by an ancestor
3564 /// (#256). Empty when the branch is fresh, when the checkpoint
3565 /// equals the head, or when the head is None.
3566 fn collect_ancestor_candidates(&self, branch: &str) -> Result<Vec<GateCandidate>, StoreError> {
3567 let b = match self.get_branch(branch)? {
3568 Some(b) => b,
3569 None => return Ok(Vec::new()),
3570 };
3571 let Some(head) = b.head_op else {
3572 return Ok(Vec::new());
3573 };
3574 if Some(&head) == b.last_gate_checkpoint.as_ref() {
3575 // Steady-state common case: previous advance left the
3576 // checkpoint at head. Nothing to re-walk.
3577 return Ok(Vec::new());
3578 }
3579
3580 let log = lex_vcs::OpLog::open(self.root())?;
3581 let walk = log.walk_back(&head, None)?;
3582 let stop_at = b.last_gate_checkpoint.clone();
3583 let mut out = Vec::new();
3584 for rec in walk {
3585 if Some(&rec.op_id) == stop_at.as_ref() {
3586 break;
3587 }
3588 let stages = attestable_stage_ids(&rec.produces);
3589 let effects = op_declared_effects(&rec.op.kind);
3590 if stages.is_empty() {
3591 out.push((rec.op_id.clone(), None, effects));
3592 } else {
3593 for sid in stages {
3594 out.push((rec.op_id.clone(), Some(sid), effects.clone()));
3595 }
3596 }
3597 }
3598 Ok(out)
3599 }
3600}
3601
3602fn stage_name(stage: &Stage) -> &str {
3603 match stage {
3604 Stage::FnDecl(fd) => &fd.name,
3605 Stage::TypeDecl(td) => &td.name,
3606 Stage::Import(i) => &i.alias,
3607 }
3608}
3609
3610fn stage_for_kind<'a>(
3611 kind: &lex_vcs::OperationKind,
3612 stages: &'a [lex_ast::Stage],
3613) -> Option<&'a lex_ast::Stage> {
3614 use lex_vcs::OperationKind::*;
3615 let target_sig = match kind {
3616 AddFunction { sig_id, .. }
3617 | ModifyBody { sig_id, .. }
3618 | ChangeEffectSig { sig_id, .. }
3619 | AddType { sig_id, .. }
3620 | ModifyType { sig_id, .. } => Some(sig_id.clone()),
3621 RenameSymbol { to, .. } => Some(to.clone()),
3622 _ => None,
3623 };
3624 let target_sig = target_sig?;
3625 stages
3626 .iter()
3627 .find(|s| sig_id(s).as_deref() == Some(target_sig.as_str()))
3628}
3629
3630fn transition_for_kind(kind: &lex_vcs::OperationKind) -> lex_vcs::StageTransition {
3631 use lex_vcs::OperationKind::*;
3632 use lex_vcs::StageTransition;
3633 match kind {
3634 AddFunction {
3635 sig_id, stage_id, ..
3636 }
3637 | AddType { sig_id, stage_id, .. } => StageTransition::Create {
3638 sig_id: sig_id.clone(),
3639 stage_id: stage_id.clone(),
3640 },
3641 RemoveFunction {
3642 sig_id,
3643 last_stage_id,
3644 }
3645 | RemoveType {
3646 sig_id,
3647 last_stage_id,
3648 } => StageTransition::Remove {
3649 sig_id: sig_id.clone(),
3650 last: last_stage_id.clone(),
3651 },
3652 ModifyBody {
3653 sig_id,
3654 from_stage_id,
3655 to_stage_id,
3656 ..
3657 }
3658 | ChangeEffectSig {
3659 sig_id,
3660 from_stage_id,
3661 to_stage_id,
3662 ..
3663 }
3664 | ModifyType {
3665 sig_id,
3666 from_stage_id,
3667 to_stage_id,
3668 }
3669 | ReplaceMatchArm {
3670 sig_id,
3671 from_stage_id,
3672 to_stage_id,
3673 ..
3674 }
3675 | RenameLocal {
3676 sig_id,
3677 from_stage_id,
3678 to_stage_id,
3679 ..
3680 }
3681 | InlineLet {
3682 sig_id,
3683 from_stage_id,
3684 to_stage_id,
3685 ..
3686 } => StageTransition::Replace {
3687 sig_id: sig_id.clone(),
3688 from: from_stage_id.clone(),
3689 to: to_stage_id.clone(),
3690 },
3691 RenameSymbol {
3692 from,
3693 to,
3694 body_stage_id,
3695 } => StageTransition::Rename {
3696 from: from.clone(),
3697 to: to.clone(),
3698 body_stage_id: body_stage_id.clone(),
3699 },
3700 AddImport { .. } | RemoveImport { .. } => StageTransition::ImportOnly,
3701 Merge { .. } => StageTransition::Merge {
3702 entries: Default::default(),
3703 },
3704 // #294: a Candidate proposes a stage without advancing
3705 // the branch. ImportOnly keeps the branch head untouched
3706 // — the stage IS published on disk (Store::propose_candidate
3707 // calls publish before apply), but no head delta lands.
3708 Candidate { .. } => StageTransition::ImportOnly,
3709 // A Promote advances the head exactly like ModifyBody
3710 // (or Create when the sig had no head). The winner
3711 // stage is the new branch state for that sig.
3712 Promote {
3713 sig_id,
3714 winner_stage_id,
3715 from_stage_id,
3716 ..
3717 } => match from_stage_id {
3718 Some(from) => StageTransition::Replace {
3719 sig_id: sig_id.clone(),
3720 from: from.clone(),
3721 to: winner_stage_id.clone(),
3722 },
3723 None => StageTransition::Create {
3724 sig_id: sig_id.clone(),
3725 stage_id: winner_stage_id.clone(),
3726 },
3727 },
3728 }
3729}
3730
3731/// Producer identity for TypeCheck attestations emitted by the
3732/// store-write gate. Pinned to this crate's name + version so an
3733/// attestation produced by a different `lex-store` revision is
3734/// distinguishable (content-hashed `produced_by`).
3735fn typecheck_producer() -> lex_vcs::ProducerDescriptor {
3736 lex_vcs::ProducerDescriptor {
3737 tool: "lex-store".into(),
3738 version: env!("CARGO_PKG_VERSION").into(),
3739 model: None,
3740 }
3741}
3742
3743/// Producer for attestations the hosted CI runner writes (#93). A
3744/// distinct tool name so a `require-attestation` gate — via the
3745/// producer-trust model — can weight "the hub verified this
3746/// server-side" above a client-attached `TypeCheck`.
3747fn hub_ci_producer() -> lex_vcs::ProducerDescriptor {
3748 lex_vcs::ProducerDescriptor {
3749 tool: "lex-hub-ci".into(),
3750 version: env!("CARGO_PKG_VERSION").into(),
3751 model: None,
3752 }
3753}
3754
3755/// Verdict of a hosted-CI run over a branch head (#93).
3756#[derive(Debug, Clone, serde::Serialize)]
3757pub struct HubCiVerdict {
3758 pub passed: bool,
3759 pub checked_stages: usize,
3760 pub attested_stages: usize,
3761 #[serde(skip_serializing_if = "Option::is_none")]
3762 pub detail: Option<String>,
3763}
3764
3765/// Producer for the replay-comparison attestation (#836 G3). Distinct
3766/// tool name so the comparison lex performed is attributable
3767/// separately from the (external) regeneration.
3768fn replay_producer() -> lex_vcs::ProducerDescriptor {
3769 lex_vcs::ProducerDescriptor {
3770 tool: "lex-store-replay".into(),
3771 version: env!("CARGO_PKG_VERSION").into(),
3772 model: None,
3773 }
3774}
3775
3776/// Human/audit label for a recorded model: `provider/name` (`@version`
3777/// when pinned).
3778fn model_label(m: &lex_vcs::ModelDescriptor) -> String {
3779 match &m.version {
3780 Some(v) => format!("{}/{}@{}", m.provider, m.name, v),
3781 None => format!("{}/{}", m.provider, m.name),
3782 }
3783}
3784
3785/// The `(sig_id, stage_id)` an op recorded producing, or `None` for a
3786/// transition that produces no stage (removal / import / merge) — those
3787/// have nothing to regenerate for a replay.
3788fn produced_sig_stage(t: &lex_vcs::StageTransition) -> Option<(String, String)> {
3789 use lex_vcs::StageTransition::*;
3790 match t {
3791 Create { sig_id, stage_id } => Some((sig_id.clone(), stage_id.clone())),
3792 Replace { sig_id, to, .. } => Some((sig_id.clone(), to.clone())),
3793 Rename { to, body_stage_id, .. } => Some((to.clone(), body_stage_id.clone())),
3794 Remove { .. } | ImportOnly | Merge { .. } => None,
3795 }
3796}
3797
3798/// Producer identity for `Examples::Passed` attestations emitted by
3799/// [`Store::record_examples_passed`] (#835). Distinct tool name so
3800/// the activity feed can tell an auto-emitted publish-time examples
3801/// verdict apart from an `lex agent-tool --examples` one.
3802fn examples_producer() -> lex_vcs::ProducerDescriptor {
3803 lex_vcs::ProducerDescriptor {
3804 tool: "lex-store::examples".into(),
3805 version: env!("CARGO_PKG_VERSION").into(),
3806 model: None,
3807 }
3808}
3809
3810/// Producer identity for `Review` attestations (#836). The reviewer's
3811/// own id lives in the kind; this records which tool minted the record.
3812fn review_producer(reviewer: &str) -> lex_vcs::ProducerDescriptor {
3813 lex_vcs::ProducerDescriptor {
3814 tool: format!("lex-store::review:{reviewer}"),
3815 version: env!("CARGO_PKG_VERSION").into(),
3816 model: None,
3817 }
3818}
3819
3820/// Producer identity for `RepairHint` attestations emitted by
3821/// `apply_operation_checked` on TypeError (#281). Distinct tool
3822/// name from `typecheck_producer` so consumers can filter the
3823/// activity feed for repair hints without scanning kinds.
3824fn repair_hint_producer() -> lex_vcs::ProducerDescriptor {
3825 lex_vcs::ProducerDescriptor {
3826 tool: "lex-store::repair_hint".into(),
3827 version: env!("CARGO_PKG_VERSION").into(),
3828 model: None,
3829 }
3830}
3831
3832/// Producer identity for `TrustWaived` attestations emitted by
3833/// the `required_attestations` gate on a trust-driven waiver
3834/// (#293). Distinct from `typecheck_producer` and `repair_hint`
3835/// so the audit trail clearly shows "the gate let this advance
3836/// through because trust > threshold."
3837fn trust_waived_producer() -> lex_vcs::ProducerDescriptor {
3838 lex_vcs::ProducerDescriptor {
3839 tool: "lex-store::trust_waived".into(),
3840 version: env!("CARGO_PKG_VERSION").into(),
3841 model: None,
3842 }
3843}
3844
3845/// Producer identity for `ProducerTrust` attestations emitted by
3846/// [`Store::recompute_producer_trust`]. The score-derivation
3847/// recompute is its own machine-emittable kind, distinct from
3848/// the gate-side `TrustWaived` emit (#293).
3849fn producer_trust_producer() -> lex_vcs::ProducerDescriptor {
3850 lex_vcs::ProducerDescriptor {
3851 tool: "lex-store::producer_trust".into(),
3852 version: env!("CARGO_PKG_VERSION").into(),
3853 model: None,
3854 }
3855}
3856
3857/// The set of stage_ids a transition introduces. These are the
3858/// stages a successful TypeCheck pass attests *about* — the new
3859/// head produced by Create/Replace, the renamed body, or the per-
3860/// sig resolution of a Merge. Removes and ImportOnly produce no
3861/// attestable stage; the program typechecks but no specific stage
3862/// is the subject of the claim.
3863/// One row of input to the producer-block / required-attestations
3864/// gates: `(op_id, stage_id, op_effects)`. The `stage_id` is
3865/// `None` for ops that don't touch a stage (imports, empty
3866/// merges) — the gate skips those.
3867type GateCandidate = (
3868 lex_vcs::OpId,
3869 Option<String>,
3870 std::collections::BTreeSet<String>,
3871);
3872
3873/// Effect set declared *by the operation itself* (#245). Used by
3874/// the `required_attestations` gate's `EffectsIntersect` clause.
3875///
3876/// Only `AddFunction` and `ChangeEffectSig` carry an effect set in
3877/// their op payload; for everything else this returns the empty
3878/// set, which means `EffectsIntersect` rules don't fire on those
3879/// ops. `Always` rules continue to fire regardless. A future
3880/// improvement is to extract effects from the candidate `Stage`
3881/// for `ModifyBody` ops, but the typed-effects-on-ops path (#247)
3882/// is the cleaner solution and lands separately.
3883fn op_declared_effects(kind: &lex_vcs::OperationKind) -> std::collections::BTreeSet<String> {
3884 use lex_vcs::OperationKind::*;
3885 match kind {
3886 AddFunction { effects, .. } => effects.clone(),
3887 ChangeEffectSig { to_effects, .. } => to_effects.clone(),
3888 _ => std::collections::BTreeSet::new(),
3889 }
3890}
3891
3892fn attestable_stage_ids(transition: &lex_vcs::StageTransition) -> Vec<String> {
3893 use lex_vcs::StageTransition::*;
3894 match transition {
3895 Create { stage_id, .. } => vec![stage_id.clone()],
3896 Replace { to, .. } => vec![to.clone()],
3897 Rename { body_stage_id, .. } => vec![body_stage_id.clone()],
3898 Merge { entries } => entries.values().filter_map(|opt| opt.clone()).collect(),
3899 Remove { .. } | ImportOnly => Vec::new(),
3900 }
3901}
3902
3903/// True when two `FnDecl`s are identical except for their body — the
3904/// precondition for a pure intra-body three-way merge (#838). The
3905/// signature fields are equal by construction when both share a
3906/// `sig_id`; this also guards the non-signature fields (`type_params`,
3907/// `examples`) so a side that changed those isn't silently dropped.
3908fn fndecl_same_except_body(a: &lex_ast::FnDecl, b: &lex_ast::FnDecl) -> bool {
3909 a.name == b.name
3910 && a.type_params == b.type_params
3911 && a.params == b.params
3912 && a.effects == b.effects
3913 && a.effect_row_var == b.effect_row_var
3914 && a.return_type == b.return_type
3915 && a.examples == b.examples
3916}
3917
3918fn write_canonical_json<T: Serialize>(path: &Path, value: &T) -> Result<(), StoreError> {
3919 let v = serde_json::to_value(value)?;
3920 let s = lex_ast::canon_json::to_canonical_string(&v);
3921 if let Some(parent) = path.parent() {
3922 fs::create_dir_all(parent)?;
3923 }
3924 fs::write(path, s)?;
3925 Ok(())
3926}
3927
3928/// Read the `name` of the `Let` expression at `let_node` inside
3929/// `stage`'s body. Used by [`Store::apply_rename_local`] to record
3930/// the rename source. Returns the same `TransformError` shapes as
3931/// the transformer itself so callers see a consistent error
3932/// vocabulary.
3933fn read_let_name(
3934 stage: &Stage,
3935 let_node: &lex_ast::NodeId,
3936) -> Result<String, lex_ast::TransformError> {
3937 // The transformer is itself a pure function; ask it to perform
3938 // a rename to a sentinel value and read the resulting let's
3939 // original name from the output. Cheaper than duplicating the
3940 // node-walk here, and stays correct as the transform evolves.
3941 //
3942 // We use a sentinel that's invalid as a Lex identifier so even
3943 // if the rename somehow lands, downstream parsing would
3944 // surface it loudly. (The transform path discards the renamed
3945 // value — we only need the *original* name.)
3946 let probed = lex_ast::rename_local(stage, let_node, "__lex_rename_probe__")?;
3947 let Stage::FnDecl(fd) = probed else {
3948 return Err(lex_ast::TransformError::NonFnTarget {
3949 stage_kind: "non-FnDecl",
3950 });
3951 };
3952 // Walk back to the probed let to read its old name from the
3953 // *original* stage — the probed stage's let has already been
3954 // renamed.
3955 let Stage::FnDecl(orig_fd) = stage else {
3956 return Err(lex_ast::TransformError::NonFnTarget {
3957 stage_kind: "non-FnDecl",
3958 });
3959 };
3960 // Path-based lookup matches the transformer's navigation.
3961 let path = parse_let_node_path(let_node.as_str())?;
3962 if path.is_empty() {
3963 return Err(lex_ast::TransformError::NotALet {
3964 at: let_node.as_str().into(),
3965 found_kind: "stage_root",
3966 });
3967 }
3968 if path[0] != orig_fd.params.len() + 1 {
3969 return Err(lex_ast::TransformError::UnknownNode {
3970 at: let_node.as_str().into(),
3971 });
3972 }
3973 let inner = &path[1..];
3974 let target = navigate_to_let(&orig_fd.body, inner, let_node.as_str())?;
3975 let _ = fd; // probed stage discarded
3976 Ok(target.to_string())
3977}
3978
3979fn parse_let_node_path(id: &str) -> Result<Vec<usize>, lex_ast::TransformError> {
3980 let s = id
3981 .strip_prefix("n_")
3982 .ok_or_else(|| lex_ast::TransformError::BadNodeId(id.into()))?;
3983 let mut parts = s.split('.');
3984 let head = parts
3985 .next()
3986 .ok_or_else(|| lex_ast::TransformError::BadNodeId(id.into()))?;
3987 if head != "0" {
3988 return Err(lex_ast::TransformError::BadNodeId(id.into()));
3989 }
3990 let mut out = Vec::new();
3991 for p in parts {
3992 out.push(
3993 p.parse::<usize>()
3994 .map_err(|_| lex_ast::TransformError::BadNodeId(id.into()))?,
3995 );
3996 }
3997 Ok(out)
3998}
3999
4000fn navigate_to_let<'a>(
4001 root: &'a lex_ast::CExpr,
4002 path: &[usize],
4003 at: &str,
4004) -> Result<&'a str, lex_ast::TransformError> {
4005 use lex_ast::CExpr::*;
4006 let mut current = root;
4007 for &idx in path {
4008 current = match current {
4009 Call { callee, args } => {
4010 if idx == 0 {
4011 callee
4012 } else {
4013 args.get(idx - 1)
4014 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?
4015 }
4016 }
4017 Let { value, body, .. } => match idx {
4018 0 => value,
4019 1 => body,
4020 _ => return Err(lex_ast::TransformError::UnknownNode { at: at.into() }),
4021 },
4022 Match { scrutinee, arms } => {
4023 if idx == 0 {
4024 scrutinee
4025 } else {
4026 let arm_off = idx - 1;
4027 if arm_off % 2 != 1 {
4028 return Err(lex_ast::TransformError::UnknownNode { at: at.into() });
4029 }
4030 let arm_index = arm_off / 2;
4031 &arms
4032 .get(arm_index)
4033 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?
4034 .body
4035 }
4036 }
4037 Block { statements, result } => {
4038 if idx < statements.len() {
4039 &statements[idx]
4040 } else if idx == statements.len() {
4041 result
4042 } else {
4043 return Err(lex_ast::TransformError::UnknownNode { at: at.into() });
4044 }
4045 }
4046 Constructor { args, .. }
4047 | TupleLit { items: args, .. }
4048 | ListLit { items: args, .. } => args
4049 .get(idx)
4050 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?,
4051 RecordLit { fields } => {
4052 &fields
4053 .get(idx)
4054 .ok_or_else(|| lex_ast::TransformError::UnknownNode { at: at.into() })?
4055 .value
4056 }
4057 FieldAccess { value, .. } if idx == 0 => value,
4058 Lambda { body, .. } if idx == 0 => body,
4059 BinOp { lhs, rhs, .. } => match idx {
4060 0 => lhs,
4061 1 => rhs,
4062 _ => return Err(lex_ast::TransformError::UnknownNode { at: at.into() }),
4063 },
4064 UnaryOp { expr, .. } if idx == 0 => expr,
4065 Return { value } if idx == 0 => value,
4066 _ => return Err(lex_ast::TransformError::UnknownNode { at: at.into() }),
4067 };
4068 }
4069 let Let { name, .. } = current else {
4070 return Err(lex_ast::TransformError::NotALet {
4071 at: at.into(),
4072 found_kind: lex_cexpr_kind(current),
4073 });
4074 };
4075 Ok(name)
4076}
4077
4078fn lex_cexpr_kind(e: &lex_ast::CExpr) -> &'static str {
4079 use lex_ast::CExpr::*;
4080 match e {
4081 Literal { .. } => "Literal",
4082 Var { .. } => "Var",
4083 Call { .. } => "Call",
4084 Let { .. } => "Let",
4085 Match { .. } => "Match",
4086 Block { .. } => "Block",
4087 Constructor { .. } => "Constructor",
4088 RecordLit { .. } => "RecordLit",
4089 TupleLit { .. } => "TupleLit",
4090 ListLit { .. } => "ListLit",
4091 FieldAccess { .. } => "FieldAccess",
4092 Lambda { .. } => "Lambda",
4093 BinOp { .. } => "BinOp",
4094 UnaryOp { .. } => "UnaryOp",
4095 Return { .. } => "Return",
4096 }
4097}
4098
4099/// Extract the declared `[budget(N)]` integer from a stage's
4100/// effect set, if any (#280 + #247). Returns `None` for stages
4101/// that aren't `FnDecl` or don't carry a budget effect — same
4102/// shape as `lex_vcs::budget_from_effects`.
4103fn budget_of_stage(stage: &Stage) -> Option<u64> {
4104 let fd = match stage {
4105 Stage::FnDecl(fd) => fd,
4106 _ => return None,
4107 };
4108 let mut min_cost: Option<u64> = None;
4109 for eff in &fd.effects {
4110 if eff.name != "budget" {
4111 continue;
4112 }
4113 if let Some(lex_ast::EffectArg::Int { value }) = &eff.arg {
4114 let n = *value as u64;
4115 min_cost = Some(min_cost.map(|c| c.min(n)).unwrap_or(n));
4116 }
4117 }
4118 min_cost
4119}
4120
4121/// Serialize a stage to its canonical-JSON byte form. Used by
4122/// `publish_signed` for delta encoding (#261 slice 3) — both the
4123/// "compute the diff" path and the "write a full snapshot"
4124/// fallback need exactly the same bytes.
4125fn canonical_bytes(stage: &Stage) -> Result<Vec<u8>, StoreError> {
4126 let v = serde_json::to_value(stage)?;
4127 Ok(lex_ast::canon_json::to_canonical_string(&v).into_bytes())
4128}
4129
4130#[allow(dead_code)]
4131fn read_json<T: DeserializeOwned>(path: &Path) -> Result<T, StoreError> {
4132 let bytes = fs::read(path)?;
4133 Ok(serde_json::from_slice(&bytes)?)
4134}