memstead_base/engine/query.rs
1//! Engine read paths — accessors and queries.
2//!
3//! Read-only methods on `Engine`: store / schema / mount accessors,
4//! per-mem path helpers (`gitdir_for` / `worktree_for`), aggregated
5//! views (`communities`, `orphans`, `stubs`, `most_connected`,
6//! `missing_required_outgoing`), per-mem summaries (`health`,
7//! `status`, `context`), search (`list`, `search`,
8//! `search_indexes`), and the bytes-level read wrappers
9//! (`list_entities`, `read_entity`, `read_provenance`). Capability and
10//! cross-mem link gating live here too — they're consulted by
11//! handlers before any mutation reaches the backend.
12
13use std::cell::OnceCell;
14use std::collections::HashMap;
15use std::path::{Path, PathBuf};
16use std::sync::Arc;
17
18use memstead_schema::Schema;
19
20use crate::engine_fallback_type;
21use crate::entity::{Entity, EntityId};
22use crate::graph::{LouvainOutput, community::detect_communities};
23use crate::mem::MemRouterSnapshot;
24use crate::ops::{ContextResult, Direction, NeighborInfo, SearchResult, SearchScope, WarningHint};
25use crate::provenance::Provenance;
26#[cfg(not(target_arch = "wasm32"))]
27use crate::search_index::{MemIndex, build_all};
28use crate::store::Store;
29use crate::workspace::{MountCapability, MountStorage, WorkspaceSettings};
30
31use super::{BackendFactory, Engine, EngineError, MountedBackend};
32
33impl Engine {
34 /// In-memory store populated at construction time from every
35 /// mount's backend. Read-only at this point in the rebuild —
36 /// mutation paths land in a later session.
37 pub fn store(&self) -> &Store {
38 &self.store
39 }
40
41 /// Per-mem schema, keyed by mount's mem name. Each entry is the
42 /// schema resolved from that mount's pin at boot, so the map holds
43 /// genuinely heterogeneous schemas in a multi-schema workspace.
44 pub fn schemas(&self) -> &HashMap<String, Arc<Schema>> {
45 &self.schemas
46 }
47
48 /// Workspace-authored schemas loaded from
49 /// `WorkspaceSettings.schemas_dir` at construction. Distinct from
50 /// [`Self::schemas`] (per-mem, only schemas pinned by a mount):
51 /// this slice carries every workspace-loaded schema regardless of
52 /// whether a mem pins it. Used by `memstead_overview` to enumerate
53 /// schemas referenced by `mem_create_rules.schemas[]` but not
54 /// pinned by any mem — agents see what could be pinned without
55 /// looking up the workspace.toml directly.
56 pub fn workspace_schemas(&self) -> &[Arc<Schema>] {
57 &self.workspace_schemas
58 }
59
60 /// Embedded built-in schemas loaded once at boot. Handlers
61 /// resolving a schema pin by `<name>@<version>` (MCP's `memstead_schema`,
62 /// `memstead_overview` rendering) walk mem-pinned, workspace, and
63 /// built-in catalogues in order — built-ins are the catch-all when
64 /// no mem or workspace dir pins the schema. Workspace schemas
65 /// shadow built-ins on `(name, version)` collision; resolve from
66 /// `workspace_schemas()` first.
67 pub fn builtin_schemas(&self) -> &[Arc<Schema>] {
68 &self.builtin_schemas
69 }
70
71 /// Classify a schema's trust origin — the single authority every read
72 /// surface consults before serving a schema's instruction-prose.
73 ///
74 /// A schema is [`OriginClass::FirstParty`] iff it is an engine built-in
75 /// **or** pinned by a writable mount in this workspace. Built-ins are
76 /// compiled into the binary — unforgeable. A non-built-in schema earns
77 /// first-party status only once the operator *adopts* it by writably
78 /// mounting a mem that pins it: writing into a mem is the act that
79 /// legitimately needs a schema's authoring prose (`system_message`,
80 /// `write_rules`, …), and the mount's writable posture is set by the
81 /// consumer's own config — a publisher cannot forge it.
82 ///
83 /// Everything else is [`OriginClass::ThirdParty`]: a schema present in
84 /// the catalogue but pinned only by read-only mounts (a registry-
85 /// installed read-mem or an adopted foreign folder/clone), or one the
86 /// engine cannot vouch for at all. Its prose is served structural-only
87 /// so a stranger's free-text never reaches a consuming agent as
88 /// instructions. This classifies by the mount graph — never by scanning
89 /// the schema's content, which a publisher controls — and `ThirdParty`
90 /// is the safe default for any ambiguous origin.
91 ///
92 /// Note a read-only mount pinning a *built-in* schema (e.g. a registry
93 /// mem on `default@1.0.0`) resolves to the consumer's own clean copy
94 /// and stays first-party — the de-framing targets only foreign,
95 /// non-built-in schemas that no writable mem has adopted.
96 pub fn schema_origin(&self, schema: &Arc<Schema>) -> crate::render::OriginClass {
97 use crate::render::OriginClass;
98 let (name, version) = schema.id();
99 // Built-in schemas are compiled in — first-party, unforgeable.
100 let is_builtin = self.builtin_schemas.iter().any(|s| {
101 let id = s.id();
102 id.0 == name && id.1 == version
103 });
104 if is_builtin {
105 return OriginClass::FirstParty;
106 }
107 // Adoption signal: some writable mount pins this exact schema, so
108 // the operator authors against it here.
109 let canon = format!("{name}@{version}");
110 let pinned_by_writable = self.mounts().iter().any(|m| {
111 m.schema.as_ref().map(|s| s.to_string()).as_deref() == Some(canon.as_str())
112 && self.mem_router().is_writable(&m.mem)
113 });
114 if pinned_by_writable {
115 OriginClass::FirstParty
116 } else {
117 OriginClass::ThirdParty
118 }
119 }
120
121 /// Classify a mem's *data* trust origin — the authority every read
122 /// surface consults before serving an entity's content (bodies,
123 /// snippets, titles). A writable mount is [`OriginClass::FirstParty`]:
124 /// its content is authored in this workspace. Anything else — a
125 /// read-only mount (a registry-installed read-mem or an adopted
126 /// foreign folder/clone) or an unknown mem — is
127 /// [`OriginClass::ThirdParty`], so the consuming agent/host treats the
128 /// content as quoted, untrusted data.
129 ///
130 /// This reads the deployment's declaration when one exists (see
131 /// [`Self::declare_mem_origin`]), else the mount's already-decided
132 /// writable/read-only posture (fixed at adopt/mount time) — it never
133 /// scans content, and both levers are consumer-side config, so a
134 /// publisher cannot forge first-party. Distinct from
135 /// [`Self::schema_origin`], which governs a schema's
136 /// instruction-prose: the data channel and the instruction channel
137 /// are separate vectors with separate authorities.
138 pub fn mem_origin_class(&self, mem: &str) -> crate::render::OriginClass {
139 if let Some(declared) = self.declared_origins.get(mem) {
140 return *declared;
141 }
142 if self.mem_router().is_writable(mem) {
143 crate::render::OriginClass::FirstParty
144 } else {
145 crate::render::OriginClass::ThirdParty
146 }
147 }
148
149 /// Declare a mem's data-trust origin as a deployment fact — the
150 /// embedding process (a curated hosted read tier, an app that vouches
151 /// for a bundled mem) overrides the writability inference for one mem.
152 /// Composition-layer-only by design: not persisted, not reachable over
153 /// MCP, never derived from mem content — the operator running the
154 /// process is the only authority that can set it, so a served mem the
155 /// deployment does *not* vouch for keeps reporting third-party on
156 /// every surface. (Deliberately absent from the CLI: that surface
157 /// operates a workspace, not a deployment; the CLI counterpart would be
158 /// a workspace-config knob no use case demands yet.)
159 pub fn declare_mem_origin(
160 &mut self,
161 mem: impl Into<String>,
162 origin: crate::render::OriginClass,
163 ) {
164 self.declared_origins.insert(mem.into(), origin);
165 }
166
167 /// Per-file errors collected during load. Non-fatal: the engine
168 /// continues with whatever did parse. Empty when every backend's
169 /// content parses cleanly.
170 pub fn load_errors(&self) -> &[(PathBuf, String)] {
171 &self.load_errors
172 }
173
174 /// Resolve a VISIBLE mem to its folder-storage root on disk.
175 ///
176 /// Unknown or quarantined names refuse `UNKNOWN_MEM` (the same
177 /// visibility gate `search` and the conflicts door apply); a
178 /// visible mount whose storage is not folder-backed returns
179 /// `Ok(None)` so callers branch on backend applicability without
180 /// inventing a typed error. Read-only mounts resolve too — this is
181 /// a read accessor, not a write gate. Generic by design: any
182 /// consumer that needs a folder mem's disk root (per-mem changelog
183 /// readers, export tooling, future doors) gets the same answer.
184 pub fn folder_mem_root(&self, mem: &str) -> Result<Option<PathBuf>, EngineError> {
185 let mount = self
186 .mounts
187 .iter()
188 .find(|m| m.mount.mem == mem)
189 .ok_or_else(|| self.unknown_mem_error(mem))?;
190 if self.quarantine_reason(mem).is_some() {
191 return Err(self.unknown_mem_error(mem));
192 }
193 match &mount.mount.storage {
194 crate::workspace::MountStorage::Folder { path } => Ok(Some(path.clone())),
195 _ => Ok(None),
196 }
197 }
198
199 /// Workspace-level operator policy (mem create/delete rules,
200 /// cross-mem links). Defaults to empty; populated via
201 /// [`Engine::set_settings`] after construction. Surfaced for MCP
202 /// handlers (`memstead_health { include_config: true }`,
203 /// `memstead_overview`'s lifecycle-namespaces section) and other
204 /// consumers that need to read workspace policy.
205 pub fn settings(&self) -> &WorkspaceSettings {
206 &self.settings
207 }
208
209 /// The pipeline configs — the v2 single-record binding store — loaded
210 /// from the workspace at boot: the read-only queryable surface the
211 /// loader exposes. Empty for engines not booted from a workspace root,
212 /// or for a workspace that declares no pipelines. The ingest skill
213 /// and future MCP tools consume this structured form
214 /// rather than re-reading the JSON folders.
215 pub fn pipeline_configs(&self) -> &crate::pipeline_store::BindingConfigs {
216 &self.pipeline_configs
217 }
218
219 /// The pipeline configs serialized as a JSON string — the read
220 /// counterpart of the `add_projection_json` edit entry point.
221 /// Serialization-boundary callers (where serde does not live)
222 /// get the store in one call and deserialize on their side.
223 ///
224 /// Shape: `{ "bindings": [{ mem, name, config }] }` — the v2
225 /// single-record store (`config` carries the whole binding: inline
226 /// `sources`, `operations`, everything). The `mediums` / `facets` /
227 /// `ingests` keys are **gone** with their record kinds. This reads the
228 /// live binding store fresh (like the brief path) rather than the
229 /// in-memory snapshot, so an edit shows back immediately. A missing
230 /// root or a legacy/unreadable store yields the fallback empty object.
231 pub fn pipeline_configs_json(&self) -> String {
232 let empty = || "{\"bindings\":[]}".to_string();
233 let Some(root) = self.workspace_root() else {
234 return empty();
235 };
236 match crate::pipeline_store::load_pipeline_configs(root) {
237 Ok(configs) => serde_json::to_string(&configs).unwrap_or_else(|_| empty()),
238 Err(_) => empty(),
239 }
240 }
241
242 /// Overwrite the in-memory pipeline configs. The workspace-root boot
243 /// paths call this after [`crate::pipeline_store::load_pipeline_configs`];
244 /// exposed so the full boot helper (a separate crate) can populate the
245 /// same surface.
246 pub fn set_pipeline_configs(&mut self, configs: crate::pipeline_store::BindingConfigs) {
247 self.pipeline_configs = configs;
248 }
249
250 /// Build a [`WarningHint::NoteMissing`] when the workspace has
251 /// `[mutations].require_notes = true` and the caller omitted (or
252 /// passed a blank/whitespace-only) `note`; `None` otherwise.
253 ///
254 /// This is the single enforcement point for the `require_notes`
255 /// provenance nudge. Every mutation that accepts a `note` calls it
256 /// on its commit-landing path and pushes the result onto the
257 /// outcome's `warnings`, so both the CLI and the MCP transports
258 /// inherit identical behaviour from the engine response rather than
259 /// each re-deriving the policy at its own boundary (the drift that
260 /// left the policy decorative on the CLI). `tool` becomes the
261 /// warning's `details.tool` — callers pass the engine-level verb
262 /// (`create_entity`, `update_entity`, `relate_entity`,
263 /// `delete_entity`, `rename_entity`, `create_mem`,
264 /// `delete_mem`), matching the commit `Tool:` provenance trailer.
265 /// The mutation still commits — the policy nudges, it never blocks.
266 pub fn note_missing_warning(&self, tool: &str, note: Option<&str>) -> Option<WarningHint> {
267 if !self.settings.mutations.require_notes.unwrap_or(false) {
268 return None;
269 }
270 let has_note = note.map(|n| !n.trim().is_empty()).unwrap_or(false);
271 if has_note {
272 return None;
273 }
274 Some(WarningHint::NoteMissing {
275 tool: tool.to_string(),
276 })
277 }
278
279 /// Backend factory currently installed on this engine. Returned by
280 /// value because [`BackendFactory`] is a function pointer (`Copy`).
281 /// Used by [`crate::mem_management::create_mem`] to materialise
282 /// the backend for a freshly-registered mount; consumers that need
283 /// to instantiate a backend ad-hoc can call this directly.
284 pub fn backend_factory(&self) -> BackendFactory {
285 self.backend_factory
286 }
287
288 /// Git-branch ops bundle currently installed on this engine.
289 /// `None` on lean-flavor engines that don't see mem-repo
290 /// mounts. Returned by value because [`super::GitBranchOps`] is
291 /// `Copy`. `create_mem` reaches for
292 /// the bundle to drive `prune_residue` against an unmounted
293 /// gitdir when the `ForceOverwrite` recovery action is selected.
294 pub fn git_branch_ops(&self) -> Option<super::GitBranchOps> {
295 self.git_branch_ops
296 }
297
298 /// Convenience: look up a parsed entity by id. Returns `None` for
299 /// unknown ids, including stub entries created for unresolved
300 /// inline-link targets — callers that want to distinguish real
301 /// from stub branch on `Entity::stub`.
302 pub fn get_entity(&self, id: &EntityId) -> Option<&Entity> {
303 self.store.get(id)
304 }
305
306 /// The stored provenance anchors for `id`, read from its mem's
307 /// anchors sidecar. Empty for an entity with none, an unknown mem, or
308 /// a backend that does not persist anchors (a pre-anchor archive / any
309 /// sealed read-only mount). Additive read surface (E3a): the
310 /// resolution *model* lives in [`crate::anchor`]
311 /// ([`crate::anchor::resolve_anchor`] / [`crate::anchor::compose_entity_anchors`]);
312 /// the live per-anchor *state* (which requires observing the source
313 /// artifacts through the medium/preparation pipeline) is E3b's concern.
314 /// The anchors sidecar's parse error for `mem`, if it has one.
315 ///
316 /// The anchor readers below degrade a malformed sidecar to "no anchors",
317 /// which keeps a read path alive but makes a corrupt file
318 /// indistinguishable from an empty one. For a *reader* that is the right
319 /// trade; for anything that draws a conclusion from the absence of
320 /// anchors it is not — a fidelity pass would report every artifact
321 /// uncovered and call it a finding, when the truth is that it could not
322 /// read the file. Callers that need that distinction ask here first and
323 /// refuse. `None` means the sidecar is absent (legitimately no anchors
324 /// yet) or parses cleanly; the binding store draws the same distinction
325 /// with its quarantine path.
326 pub fn anchors_sidecar_error(&self, mem: &str) -> Option<String> {
327 let mount = self.mounts.iter().find(|m| m.mount.mem == mem)?;
328 // Three distinct ways to be unreadable, and only one of them is a
329 // parse error. `.ok().flatten()` would collapse the first into
330 // "absent", which is the very confusion this exists to prevent.
331 let bytes = match mount.backend.read_anchors_sidecar() {
332 // A backend error — permission denied, an IO fault. The file may
333 // be perfectly well-formed; we simply could not look at it.
334 Err(e) => return Some(format!("could not read the sidecar: {e}")),
335 // Genuinely absent: a mem with no anchors yet. Not an error.
336 Ok(None) => return None,
337 Ok(Some(b)) => b,
338 };
339 // An empty or whitespace-only file parses as "no anchors" by a
340 // deliberate tolerance in `from_bytes`. That tolerance is right for a
341 // reader and wrong here: a sidecar truncated to zero by an
342 // interrupted write is not a mem that never had anchors.
343 if bytes.iter().all(|b| b.is_ascii_whitespace()) {
344 return Some(
345 "the sidecar file is empty — an interrupted write leaves this state, and it is not the same as having no anchors; remove the file if the mem genuinely has none"
346 .to_string(),
347 );
348 }
349 match crate::anchor::AnchorSidecar::from_bytes(&bytes) {
350 Ok(_) => None,
351 Err(e) => Some(e.to_string()),
352 }
353 }
354
355 pub fn entity_anchors(&self, id: &EntityId) -> Vec<crate::anchor::Anchor> {
356 let Some(mount) = self.mounts.iter().find(|m| m.mount.mem == id.mem()) else {
357 return Vec::new();
358 };
359 let Ok(Some(bytes)) = mount.backend.read_anchors_sidecar() else {
360 return Vec::new();
361 };
362 match crate::anchor::AnchorSidecar::from_bytes(&bytes) {
363 Ok(sc) => sc.get(id.as_ref()).to_vec(),
364 // Deliberate degrade-to-empty for the read path; a caller that
365 // must not confuse "unreadable" with "none" checks
366 // [`Self::anchors_sidecar_error`] first.
367 Err(_) => Vec::new(),
368 }
369 }
370
371 /// The stored anchors for `id`, each paired with its **live** resolution
372 /// state when the engine could observe the source artifact this pass.
373 ///
374 /// Additive over [`Self::entity_anchors`]: the durable data is unchanged;
375 /// `state` is the [`crate::anchor::resolve_anchor`] outcome against an
376 /// observation the engine produces here. A `path`-namespace anchor
377 /// (codebase / filesystem / git) is observed against the working tree at
378 /// the current HEAD; an `entity`-namespace anchor is observed against the
379 /// live graph by [`Self::observe_entity_anchor`]:
380 ///
381 /// - artifact absent ⇒ [`AnchorState::Orphaned`](crate::anchor::AnchorState::Orphaned);
382 /// - artifact present, non-hash class (`authored` / `informed-by`) ⇒
383 /// [`Resolves`](crate::anchor::AnchorState::Resolves);
384 /// - artifact present, hash-bearing class (`anchored` / `derived`) ⇒ the
385 /// prepared-content hash comparison decides:
386 /// [`Resolves`](crate::anchor::AnchorState::Resolves) on a match,
387 /// [`Drifted`](crate::anchor::AnchorState::Drifted) on a stable-medium
388 /// mismatch, [`Recheck`](crate::anchor::AnchorState::Recheck) on an
389 /// unstable medium or when a hash is unavailable on either side (a
390 /// hash-less anchor, a `tree` grain, an unreadable artifact).
391 ///
392 /// For an `entity` grain the same table applies, read off the store
393 /// rather than the filesystem: the entity missing (or present only as a
394 /// stub) is `Orphaned`, and a hash-bearing class compares the canonical
395 /// rendered markdown.
396 ///
397 /// `state` is `None` (unobserved — never a fabricated state) when there is
398 /// no workspace root, when the grain is `url` (the engine never fetches;
399 /// a url anchor's hash is the registry's prepared form of the content
400 /// its observer supplied at write time), when an `entity` anchor's
401 /// source declares a preparation the registry does not know (the form
402 /// cannot be computed), or when an `entity` anchor's mem is **not
403 /// mounted** — an unmounted mem is not a mem of deleted entities, and
404 /// saying so would route a deletion proposal to prune.
405 pub fn entity_anchors_resolved(&self, id: &EntityId) -> Vec<ResolvedAnchor> {
406 let anchors = self.entity_anchors(id);
407 let source_roots = self.anchor_source_roots(id.mem());
408 let none = SuppliedObservations::new();
409 anchors
410 .into_iter()
411 .map(|anchor| self.resolve_one(anchor, &source_roots, &none))
412 .collect()
413 }
414
415 /// Observe one stored anchor and pair it with its resolution — the
416 /// shared step behind every anchor read.
417 fn resolve_one(
418 &self,
419 anchor: crate::anchor::Anchor,
420 source_roots: &std::collections::BTreeMap<String, AnchorSourceJoin>,
421 supplied: &SuppliedObservations,
422 ) -> ResolvedAnchor {
423 let observed = self.observe_anchor(&anchor, source_roots, supplied);
424 let (state, observed_hash, observed_at) = match observed {
425 Some(o) => (Some(o.state), o.hash, o.at),
426 None => (None, None, None),
427 };
428 ResolvedAnchor {
429 anchor,
430 state,
431 observed_hash,
432 observed_at,
433 }
434 }
435
436 /// Per-anchor observation — THE one resolution mechanism, shared by
437 /// binding-backed verify (`mem_anchors_resolved`, which the ingest
438 /// render/report/prune/findings paths consume), the per-entity read
439 /// (`entity_anchors_resolved`), and the standalone
440 /// `verify_mem_anchors` operation. Path-shaped grains
441 /// (`span`/`file`/`tree`) observe under the **decision-29 candidate
442 /// priority** (backlog-sweep plan 03a): the anchor's artifact path is
443 /// SOURCE-relative first — when its `source` name resolves through
444 /// `source_roots` to a declared pointer, the pointer-joined path is
445 /// authoritative — and workspace-relative only as the fallback, tried
446 /// when the source-join does not resolve. A path resolving under both
447 /// joins is decided by that priority, deterministically. An anchor
448 /// without a `source` (a hand-authored mem, a binding-less write)
449 /// observes workspace-relative exactly as before. A `url`
450 /// grain has no engine-side observation and returns `None` (the report
451 /// vocabulary's `unresolvable`) — the engine never fetches; its hash is
452 /// recorded from observation-supplied content at write time — as does a
453 /// workspace-root-less engine. An `entity`
454 /// grain is not a filesystem path but is not unobservable either — it
455 /// resolves against the live graph via
456 /// [`Self::observe_entity_anchor`], under the preparation its source
457 /// declares (touchpoint A of [`crate::preparation`]: the registry
458 /// decides the prepared form the artifact hashes as). This replaces the retired `single_path_medium_root` gate,
459 /// whose single-source assumption nulled every anchor of a mem with
460 /// zero or several bindings — the honest per-anchor answer supersedes
461 /// the all-or-nothing mem-level one.
462 ///
463 /// A `url` grain is the one grain the engine cannot observe itself: it
464 /// resolves from a SUPPLIED observation when the caller passed one for
465 /// its artifact (`memstead verify-anchors --observations`), else from
466 /// the row's recorded `last_observed` (sidecar version 2) — an
467 /// observation that was made, carrying its own date so the row ages
468 /// visibly — else it is unobserved (`None`), never fabricated. A
469 /// supplied `absent` resolves `recheck`, not `orphaned`: an observer
470 /// failing to retrieve a web resource is not the medium saying the
471 /// artifact is gone, and prune must never act on it.
472 fn observe_anchor(
473 &self,
474 anchor: &crate::anchor::Anchor,
475 source_roots: &std::collections::BTreeMap<String, AnchorSourceJoin>,
476 supplied: &SuppliedObservations,
477 ) -> Option<Observed> {
478 if anchor.grain == crate::anchor::AnchorGrain::Url {
479 if let Some(obs) = supplied.get(&anchor.artifact) {
480 return Some(match &obs.outcome {
481 crate::anchor::SuppliedOutcome::Absent => Observed {
482 state: crate::anchor::AnchorState::Recheck,
483 hash: None,
484 at: Some(obs.at.clone()),
485 },
486 crate::anchor::SuppliedOutcome::Present { hash } => Observed {
487 state: crate::anchor::resolve_anchor(
488 anchor,
489 &crate::anchor::ArtifactObservation::Present {
490 current_hash: Some(hash.clone()),
491 },
492 ),
493 hash: Some(hash.clone()),
494 at: Some(obs.at.clone()),
495 },
496 });
497 }
498 return anchor.last_observed.as_ref().map(|rec| Observed {
499 state: rec.state,
500 hash: rec.hash.clone(),
501 at: Some(rec.at.clone()),
502 });
503 }
504 let join = anchor
505 .source
506 .as_deref()
507 .and_then(|name| source_roots.get(name));
508 // An `entity`-grain anchor points into a mem's graph, not a file
509 // tree. It has always returned `None` here — "unobserved this pass" —
510 // which meant it could never be drifted, never be orphaned, and always
511 // blocked prune. That is the bail the S1b pilot demonstrated: a
512 // deliberately stale anchor over a changed source entity went unflagged
513 // while the capability matrix claimed full parity.
514 if anchor.grain == crate::anchor::AnchorGrain::Entity {
515 return self
516 .observe_entity_anchor(anchor, join.and_then(|j| j.preparation.as_deref()))
517 .map(Observed::live);
518 }
519 let root = self.workspace_root.as_deref()?;
520 observe_path_anchor(root, anchor, join).map(Observed::live)
521 }
522
523 /// Observe an `entity`-grain anchor against the live graph — the entity-
524 /// namespace counterpart of [`observe_path_anchor`], and the mechanism
525 /// that makes a graph-medium binding's drift real.
526 ///
527 /// The artifact is an entity id. Present/absent comes from the store, so
528 /// this works uniformly across backends — a git-branch mem has no
529 /// working-tree file to stat, which is exactly why observation cannot go
530 /// through the filesystem here. The compared form is the preparation
531 /// registry's **prepared form** for `preparation` (the anchor's source's
532 /// declared preparation, [`crate::preparation::entity_prepared_hash`]):
533 /// the **canonical rendered markdown** when the source declares none —
534 /// byte-for-byte today's form — or the load-bearing serialization under
535 /// `entity-load-bearing`; hashed with the same `prepared_content_hash`
536 /// the path arm uses, so an anchor's recorded hash means the same thing
537 /// in both namespaces. An identifier the registry does not know cannot
538 /// be prepared: the anchor is reported unobserved (`None`), never hashed
539 /// under a fabricated form.
540 ///
541 /// A stub is treated as absent: a stub is the engine's placeholder for an
542 /// unresolved reference, not the entity the anchor claims to pin. Scoring
543 /// it as present would let a dangling anchor resolve clean.
544 fn observe_entity_anchor(
545 &self,
546 anchor: &crate::anchor::Anchor,
547 preparation: Option<&str>,
548 ) -> Option<(crate::anchor::AnchorState, Option<String>)> {
549 let id = EntityId::canonical(&anchor.artifact);
550
551 // The mem the anchor points into must be MOUNTED before a store miss
552 // can mean anything. If it is not, every entity in it is missing from
553 // the store, and reporting `Orphaned` would say "the source deleted
554 // these" about entities sitting untouched on disk. `None` — genuinely
555 // unobserved — is the honest answer.
556 //
557 // This guard lives here, at the one observation site, and not at the
558 // callers. An earlier fix put it in `run_verify` alone; `prune` reaches
559 // anchor resolution by its own path (`mem_anchors_resolved`), so the
560 // sync brief went on proposing the deletion of every destination
561 // entity — a data-loss suggestion routed to the graph's only
562 // maintenance writer, from a mem merely being unmounted. A guard that
563 // protects one caller is not a guard on the behaviour.
564 if !self.mounts.iter().any(|m| m.mount.mem == id.mem()) {
565 return None;
566 }
567
568 let entity = self.store.get(&id).filter(|e| !e.stub);
569 let Some(entity) = entity else {
570 return Some((
571 crate::anchor::resolve_anchor(anchor, &crate::anchor::ArtifactObservation::Absent),
572 None,
573 ));
574 };
575 let current_hash = if anchor.class.is_hash_bearing() {
576 let type_def = self
577 .schema_for(id.mem())
578 .and_then(|schema| schema.get_type(&entity.entity_type));
579 Some(crate::preparation::entity_prepared_hash(
580 entity,
581 type_def.as_deref(),
582 preparation,
583 )?)
584 } else {
585 None
586 };
587 let observation = crate::anchor::ArtifactObservation::Present {
588 current_hash: current_hash.clone(),
589 };
590 Some((
591 crate::anchor::resolve_anchor(anchor, &observation),
592 current_hash,
593 ))
594 }
595
596 /// The `source name → join` map for `mem`'s bindings: the filesystem
597 /// roots that anchors written in the source dialect join onto (decision
598 /// 26: anchor artifact paths are source-relative first) and the
599 /// preparation each source declares (touchpoint A: what the registry
600 /// prepares the artifact as before hashing). Empty when the workspace
601 /// has no root, the pipeline store does not load, or `mem` has no
602 /// bindings — resolution then degrades to the workspace-relative dialect
603 /// alone with no preparation, which is exactly the hand-authored-mem
604 /// posture.
605 pub(crate) fn anchor_source_roots(
606 &self,
607 mem: &str,
608 ) -> std::collections::BTreeMap<String, AnchorSourceJoin> {
609 let mut roots = std::collections::BTreeMap::new();
610 let Some(root) = self.workspace_root.as_deref() else {
611 return roots;
612 };
613 let Ok(configs) = crate::pipeline_store::load_pipeline_configs(root) else {
614 return roots;
615 };
616 for record in configs.bindings.iter().filter(|r| r.mem == mem) {
617 for source in &record.config.sources {
618 roots
619 .entry(source.name.clone())
620 .or_insert_with(|| AnchorSourceJoin {
621 pointer: source.pointer.clone(),
622 preparation: source.preparation.clone(),
623 source: source.clone(),
624 deny_paths: record.config.deny_paths.clone(),
625 });
626 }
627 }
628 roots
629 }
630
631 /// Reverse anchor lookup: every `(entity_id, anchor)` across all mems
632 /// whose anchor references `artifact_path`. This is the query the
633 /// rebuilt check-realization hook consumes — given the file an agent
634 /// just edited, which entities anchored to it. A `span`/`file`/`tree`
635 /// anchor references the path when its base path (locator suffix
636 /// `@commit` / `#span` stripped) equals the path, or — for a `tree`
637 /// grain — when the path lies under the tree. Path-shaped grains only;
638 /// `url` / `entity` anchors are matched by exact base equality.
639 pub fn anchors_referencing_artifact(
640 &self,
641 artifact_path: &str,
642 ) -> Vec<(EntityId, crate::anchor::Anchor)> {
643 let mut out = Vec::new();
644 for mount in &self.mounts {
645 let Ok(Some(bytes)) = mount.backend.read_anchors_sidecar() else {
646 continue;
647 };
648 let Ok(sc) = crate::anchor::AnchorSidecar::from_bytes(&bytes) else {
649 continue;
650 };
651 // Source-dialect anchors (decision 26) reference the same
652 // artifact under its pointer-joined workspace form — match both.
653 let source_roots = self.anchor_source_roots(&mount.mount.mem);
654 for (eid, anchors) in &sc.entities {
655 for a in anchors {
656 // The shared decision-29 candidate rule: the join
657 // candidate exists only where the rule produces one (a
658 // climbing `../…` artifact never joins); the
659 // workspace-relative form is the `anchor_references_path`
660 // arm below.
661 let joined = a
662 .source
663 .as_deref()
664 .and_then(|name| source_roots.get(name))
665 .map(|join| {
666 artifact_candidates(&join.pointer, anchor_base_path(&a.artifact))
667 })
668 .and_then(|mut c| (c.len() > 1).then(|| c.remove(0)));
669 if anchor_references_path(a, artifact_path)
670 || joined.is_some_and(|j| {
671 path_references(
672 &j,
673 a.grain == crate::anchor::AnchorGrain::Tree,
674 artifact_path,
675 )
676 })
677 {
678 out.push((EntityId(eid.clone()), a.clone()));
679 }
680 }
681 }
682 }
683 out
684 }
685
686 /// Every `(entity_id, resolved anchor)` in `mem`, read from its anchors
687 /// sidecar once and each paired with its **live** resolution state (the
688 /// same observation [`Self::entity_anchors_resolved`] produces per entity,
689 /// computed here mem-wide in a single sidecar read). Empty for an unknown
690 /// mem, a backend that persists no anchors, or a mem with none.
691 ///
692 /// Additive read surface: the durable data is unchanged; `state` is the
693 /// [`crate::anchor::resolve_anchor`] outcome against an observation the
694 /// engine produces — the working tree for a `path`-namespace anchor, the
695 /// live graph for an `entity` one — or `None` when unobserved (never
696 /// fabricated). The verify pipeline consumes it to adjudicate a mem's
697 /// anchors against the source; audit/health can reuse it.
698 /// Whether `mem`'s anchor sidecar records ANY anchor row — the cheap
699 /// existence check [`Self::mem_anchors_resolved`] cannot serve: that
700 /// walk OBSERVES every anchor (hashing live source artifacts, and for
701 /// path anchors enumerating the facet's file scope) when a caller only
702 /// needs to know the sidecar is non-empty. Parses the sidecar and stops
703 /// there; false for an unknown mem, a backend without anchors, or an
704 /// empty sidecar — the same population `mem_anchors_resolved` would
705 /// report empty for, minus the observation cost.
706 pub fn mem_has_anchors(&self, mem: &str) -> bool {
707 let Some(mount) = self.mounts.iter().find(|m| m.mount.mem == mem) else {
708 return false;
709 };
710 let Ok(Some(bytes)) = mount.backend.read_anchors_sidecar() else {
711 return false;
712 };
713 let Ok(sc) = crate::anchor::AnchorSidecar::from_bytes(&bytes) else {
714 return false;
715 };
716 sc.entities.values().any(|anchors| !anchors.is_empty())
717 }
718
719 pub fn mem_anchors_resolved(&self, mem: &str) -> Vec<(EntityId, ResolvedAnchor)> {
720 self.mem_anchors_resolved_with(mem, &SuppliedObservations::new())
721 }
722
723 /// [`Self::mem_anchors_resolved`] with observer-supplied observations
724 /// for the grains the engine cannot observe itself (`url`), keyed by
725 /// artifact. Rows without a supplied observation resolve as they would
726 /// without the map.
727 pub fn mem_anchors_resolved_with(
728 &self,
729 mem: &str,
730 supplied: &SuppliedObservations,
731 ) -> Vec<(EntityId, ResolvedAnchor)> {
732 let Some(mount) = self.mounts.iter().find(|m| m.mount.mem == mem) else {
733 return Vec::new();
734 };
735 let Ok(Some(bytes)) = mount.backend.read_anchors_sidecar() else {
736 return Vec::new();
737 };
738 let Ok(sc) = crate::anchor::AnchorSidecar::from_bytes(&bytes) else {
739 return Vec::new();
740 };
741 let mut out = Vec::new();
742 let source_roots = self.anchor_source_roots(mem);
743 for (eid, anchors) in &sc.entities {
744 for anchor in anchors {
745 out.push((
746 EntityId(eid.clone()),
747 self.resolve_one(anchor.clone(), &source_roots, supplied),
748 ));
749 }
750 }
751 out
752 }
753
754 /// Standalone anchor verification — "do my sources still say what I
755 /// recorded?" for one mem, regardless of how it was built. Walks the
756 /// mem's anchor sidecar through the shared per-anchor mechanism
757 /// ([`Self::observe_anchor`] via [`Self::mem_anchors_resolved`]) and
758 /// classifies every anchor into the report vocabulary: `resolved`
759 /// (source present, hash matches or non-hash class), `drifted`
760 /// (present, hash differs, stability `stable`), `recheck` (hash
761 /// differs under `unstable`, or a hash is missing on either side),
762 /// `unresolvable` (source absent, or a grain/medium the mechanism
763 /// does not reach — never fabricated into drift). Read-only on mem
764 /// content: pure sidecar read + filesystem observation, no commit on
765 /// any backend. A mem with no anchors returns an empty report.
766 pub fn verify_mem_anchors(&self, mem: &str) -> Result<MemAnchorVerification, EngineError> {
767 self.verify_mem_anchors_with(mem, &SuppliedObservations::new())
768 }
769
770 /// The stale axis defers to anchor state: an entity carrying at least
771 /// one adjudicated hash-bearing anchor (`resolves`, `drifted` or
772 /// `recheck`) reads by its anchors, not by the day threshold. `drifted`
773 /// and `recheck` list the entity as their own condition whatever its
774 /// age; `resolves` keeps it off the list, and when the threshold would
775 /// have listed it the row moves to `anchor_fresh` so the reading names
776 /// the clock that overruled the threshold. Entities with no adjudicated
777 /// anchor (none at all, or only `unresolvable` / `unobserved` rows)
778 /// keep the day-threshold reading, byte for byte. Derived at read time
779 /// from the same verification the anchors axis runs; nothing stored.
780 fn apply_anchor_clock(&self, summary: &mut crate::ops::HealthSummary, mem: Option<&str>) {
781 use std::collections::HashMap;
782 // Dominant adjudicated state per entity: drifted > recheck > resolves.
783 fn rank(state: &str) -> Option<u8> {
784 match state {
785 "drifted" => Some(3),
786 "recheck" => Some(2),
787 "resolves" => Some(1),
788 _ => None,
789 }
790 }
791 let mut by_entity: HashMap<String, (u8, String)> = HashMap::new();
792 let mut mems: Vec<String> = self.mem_names().iter().map(|s| s.to_string()).collect();
793 mems.retain(|m| mem.is_none_or(|scope| scope == m));
794 for m in mems {
795 let Ok(report) = self.verify_mem_anchors(&m) else {
796 continue;
797 };
798 for row in &report.anchors {
799 let Some(r) = rank(&row.state) else { continue };
800 let entry = by_entity
801 .entry(row.entity_id.clone())
802 .or_insert((0, String::new()));
803 if r > entry.0 {
804 *entry = (r, row.state.clone());
805 }
806 }
807 }
808 if by_entity.is_empty() {
809 return;
810 }
811 let today = crate::ops::health::days_since_epoch();
812 let mut stale = std::mem::take(&mut summary.stale_entities);
813 // Day-threshold rows on anchored entities leave the list: a
814 // resolving anchor makes them fresh, a drifted or recheck one is
815 // re-listed below under its own condition.
816 let mut fresh = Vec::new();
817 stale.retain(|row| match by_entity.get(row.id.0.as_str()) {
818 None => true,
819 Some((_, state)) => {
820 if state == "resolves" {
821 fresh.push(crate::ops::StaleEntity {
822 id: row.id.clone(),
823 title: row.title.clone(),
824 days_since_modified: row.days_since_modified,
825 anchor_state: Some(state.clone()),
826 });
827 }
828 false
829 }
830 });
831 let mut by_anchor: Vec<crate::ops::StaleEntity> = by_entity
832 .iter()
833 .filter(|(_, (_, state))| state != "resolves")
834 .filter_map(|(id, (_, state))| {
835 let entity = self.store.get(&crate::EntityId(id.clone()))?;
836 if entity.stub || mem.is_some_and(|scope| entity.mem != scope) {
837 return None;
838 }
839 let days = entity
840 .metadata
841 .get("last_modified")
842 .and_then(|v| crate::ops::health::parse_iso_to_days(&v.to_frontmatter_string()))
843 .map(|d| today.saturating_sub(d))
844 .unwrap_or(0);
845 Some(crate::ops::StaleEntity {
846 id: entity.id.clone(),
847 title: entity.title.clone(),
848 days_since_modified: days,
849 anchor_state: Some(state.clone()),
850 })
851 })
852 .collect();
853 stale.append(&mut by_anchor);
854 stale.sort_by(|a, b| a.id.0.cmp(&b.id.0));
855 fresh.sort_by(|a, b| a.id.0.cmp(&b.id.0));
856 summary.stale_entities = stale;
857 summary.anchor_fresh = fresh;
858 }
859
860 /// [`Self::verify_mem_anchors`] with observer-supplied observations
861 /// (the `--observations` file of `memstead verify-anchors`). A `url`
862 /// row whose artifact has a supplied observation adjudicates through
863 /// the shared funnel exactly like a file row; a supplied row matching
864 /// no `url` anchor of the mem is reported in `unmatched_observations`
865 /// and changes nothing. The report's `recordable_observations` are the
866 /// `last_observed` records the caller commits with
867 /// [`Self::record_anchor_observations`] — the verification itself
868 /// writes nothing.
869 pub fn verify_mem_anchors_with(
870 &self,
871 mem: &str,
872 supplied: &SuppliedObservations,
873 ) -> Result<MemAnchorVerification, EngineError> {
874 if !self.mem_router.is_visible(mem) {
875 return Err(self.unknown_mem_error(mem));
876 }
877 let now = self.now_iso();
878 // An unreadable sidecar is a condition, never zero rows: the readers
879 // below would degrade it to "no anchors" and the report would then
880 // describe a clean mem it never measured.
881 if let Some(why) = self.anchors_sidecar_error(mem) {
882 return Ok(MemAnchorVerification {
883 mem: mem.to_string(),
884 sidecar_error: Some(why),
885 ..Default::default()
886 });
887 }
888 let mut report = MemAnchorVerification {
889 mem: mem.to_string(),
890 unreconciled: self
891 .entity_set_is_reconcilable(mem)
892 .err()
893 .map(str::to_string),
894 ..Default::default()
895 };
896 let reconciled = report.unreconciled.is_none();
897 let mut matched: std::collections::BTreeSet<String> = std::collections::BTreeSet::new();
898 for (eid, resolved) in self.mem_anchors_resolved_with(mem, supplied) {
899 let is_url = resolved.anchor.grain == crate::anchor::AnchorGrain::Url;
900 let supplied_here = is_url && supplied.contains_key(&resolved.anchor.artifact);
901 if supplied_here {
902 matched.insert(resolved.anchor.artifact.clone());
903 }
904 let observed_at = resolved.observed_at.clone();
905 let unobserved_for_days = observed_at
906 .as_deref()
907 .and_then(|at| crate::anchor::days_between(at, &now));
908 if let (true, Some(state), Some(at)) = (supplied_here, resolved.state, &observed_at) {
909 report.recordable_observations.push(RecordedObservation {
910 entity: eid.to_string(),
911 artifact: resolved.anchor.artifact.clone(),
912 observation: crate::anchor::AnchorObservation {
913 at: at.clone(),
914 hash: resolved.observed_hash.clone(),
915 state,
916 },
917 });
918 }
919 // The entity end first: a row whose holder is gone is adjudicated
920 // against its artifact alone otherwise, and a matching hash then
921 // reports it as `resolved` for an entity that does not exist.
922 if reconciled && self.entity_is_absent(&eid) {
923 report.dangling += 1;
924 report.anchors.push(VerifiedAnchor {
925 entity_id: eid.to_string(),
926 artifact: resolved.anchor.artifact.clone(),
927 grain: resolved.anchor.grain.as_wire().to_string(),
928 class: resolved.anchor.class.as_wire().to_string(),
929 state: "dangling".to_string(),
930 observed_hash: resolved.observed_hash,
931 observed_at,
932 unobserved_for_days,
933 observation_supplied: supplied_here,
934 });
935 continue;
936 }
937 let state = match resolved.state {
938 Some(crate::anchor::AnchorState::Resolves) => {
939 report.resolves += 1;
940 crate::anchor::AnchorState::Resolves.as_wire()
941 }
942 Some(crate::anchor::AnchorState::Drifted) => {
943 report.drifted += 1;
944 "drifted"
945 }
946 Some(crate::anchor::AnchorState::Recheck) => {
947 report.recheck += 1;
948 "recheck"
949 }
950 Some(crate::anchor::AnchorState::Orphaned) => {
951 report.unresolvable += 1;
952 "unresolvable"
953 }
954 // Split from `unresolvable` (03/05, criterion 2): the artifact
955 // being GONE is a measurement; the pass not reaching the
956 // artifact at all is the absence of one, and the repairs
957 // differ.
958 None => {
959 report.unobserved += 1;
960 "unobserved"
961 }
962 };
963 report.anchors.push(VerifiedAnchor {
964 entity_id: eid.to_string(),
965 artifact: resolved.anchor.artifact.clone(),
966 grain: resolved.anchor.grain.as_wire().to_string(),
967 class: resolved.anchor.class.as_wire().to_string(),
968 state: state.to_string(),
969 observed_hash: resolved.observed_hash,
970 observed_at,
971 unobserved_for_days,
972 observation_supplied: supplied_here,
973 });
974 }
975 report.unmatched_observations = supplied
976 .keys()
977 .filter(|artifact| !matched.contains(*artifact))
978 .cloned()
979 .collect();
980 Ok(report)
981 }
982
983 /// Mem names the engine knows about, in declaration order.
984 /// Cheap; useful for callers that need to enumerate before
985 /// dispatching by mem.
986 pub fn mem_names(&self) -> Vec<&str> {
987 self.mounts.iter().map(|m| m.mount.mem.as_str()).collect()
988 }
989
990 /// Derivation-staleness report for one mem (agent-trust plan 12):
991 /// every EXPLICIT edge whose rel-type the mem's schema declares
992 /// `derivation: true`, compared against its recorded baseline.
993 /// Baseline differs from the target's current hash → `stale`;
994 /// no baseline recorded (edge predates the declaration, or was
995 /// load-derived) → `unbaselined`, distinctly — never fabricated
996 /// as fresh or stale. Fresh edges are not reported. A mem whose
997 /// schema declares no derivation rel-types returns the empty
998 /// report; an unreadable sidecar reads as empty (every edge
999 /// unbaselined) rather than an error.
1000 pub fn derivation_report(
1001 &self,
1002 mem: &str,
1003 ) -> Result<Vec<crate::ops::health::DerivationFinding>, EngineError> {
1004 if !self.mem_router.is_visible(mem) {
1005 return Err(self.unknown_mem_error(mem));
1006 }
1007 let Some(schema) = self.schemas.get(mem) else {
1008 return Ok(Vec::new());
1009 };
1010 let declared: std::collections::HashSet<&str> = schema
1011 .manifest
1012 .relationships
1013 .definitions
1014 .iter()
1015 .filter(|d| d.derivation)
1016 .map(|d| d.name.as_str())
1017 .collect();
1018 if declared.is_empty() {
1019 return Ok(Vec::new());
1020 }
1021 let sidecar = self
1022 .mounts
1023 .iter()
1024 .find(|m| m.mount.mem == mem)
1025 .and_then(|m| {
1026 m.backend
1027 .read_entity(Path::new(crate::derivation::DERIVATION_SIDECAR_PATH))
1028 .ok()
1029 .flatten()
1030 })
1031 .and_then(|bytes| crate::derivation::DerivationSidecar::from_bytes(&bytes).ok())
1032 .unwrap_or_default();
1033
1034 let mut out = Vec::new();
1035 let mut sources: Vec<&crate::entity::Entity> = self
1036 .store
1037 .all_entities()
1038 .filter(|e| !e.stub && e.id.mem() == mem)
1039 .collect();
1040 sources.sort_by(|a, b| a.id.as_ref().cmp(b.id.as_ref()));
1041 for entity in sources {
1042 for edge in self.store.outgoing(&entity.id) {
1043 if !declared.contains(edge.rel_type.as_str())
1044 || edge.source != crate::store::EdgeSource::Explicit
1045 {
1046 continue;
1047 }
1048 let current = self
1049 .store
1050 .get(&edge.target)
1051 .map(|t| t.content_hash.clone())
1052 .unwrap_or_default();
1053 match sidecar.get(entity.id.as_ref(), &edge.rel_type, edge.target.as_ref()) {
1054 None => out.push(crate::ops::health::DerivationFinding {
1055 source: entity.id.clone(),
1056 rel_type: edge.rel_type.clone(),
1057 target: edge.target.clone(),
1058 state: "unbaselined".to_string(),
1059 baseline: None,
1060 current,
1061 }),
1062 Some(baseline) if baseline != current => {
1063 out.push(crate::ops::health::DerivationFinding {
1064 source: entity.id.clone(),
1065 rel_type: edge.rel_type.clone(),
1066 target: edge.target.clone(),
1067 state: "stale".to_string(),
1068 baseline: Some(baseline.to_string()),
1069 current,
1070 })
1071 }
1072 Some(_) => {}
1073 }
1074 }
1075 }
1076 Ok(out)
1077 }
1078
1079 /// Public-shape mount record for `mem`, or `None` for an unknown
1080 /// mem.
1081 ///
1082 /// Surfaces the operator-facing
1083 /// [`crate::workspace::Mount`] (mem name, schema pin, storage
1084 /// reference, capability, lifecycle, cross_linkable) so MCP / CLI
1085 /// handlers can branch on backend-specific shapes via
1086 /// [`crate::workspace::MountStorage`] when they need accessors
1087 /// that don't make sense on every backend (e.g. gitdir / branch
1088 /// for `memstead_health { include_config: true }`'s git-class
1089 /// payload). Backends that want the equivalent of full's
1090 /// `engine.gitdir_for(mem)` match
1091 /// `engine.mount(mem).map(|m| &m.storage)` against
1092 /// `MountStorage::GitBranch { gitdir, branch }` and walk
1093 /// directly — keeps the engine surface backend-neutral.
1094 ///
1095 /// Counterpart to [`Self::mem_names`] which lists every mount.
1096 pub fn mount(&self, mem: &str) -> Option<&crate::workspace::Mount> {
1097 self.mounts
1098 .iter()
1099 .find(|m| m.mount.mem == mem)
1100 .map(|m| &m.mount)
1101 }
1102
1103 /// Orphan count attributed to each mem's pinned schema, over the
1104 /// given `orphan_ids` (the caller pre-filters them by any mem scope).
1105 /// Lets a health surface show that ingest-mem isolates (orphans by
1106 /// design) and code-mem debt land in different schema buckets rather
1107 /// than one blended, misleading total. Mems with no settled pin
1108 /// bucket under the empty string.
1109 pub fn orphans_by_schema(
1110 &self,
1111 orphan_ids: &[EntityId],
1112 ) -> std::collections::BTreeMap<String, usize> {
1113 let mut by_schema = std::collections::BTreeMap::new();
1114 for id in orphan_ids {
1115 let schema = self
1116 .store()
1117 .get(id)
1118 .and_then(|e| self.mount(&e.mem))
1119 .and_then(|m| m.schema.as_ref().map(|s| s.as_display()))
1120 .unwrap_or_default();
1121 *by_schema.entry(schema).or_insert(0) += 1;
1122 }
1123 by_schema
1124 }
1125
1126 /// Community count attributed to each schema across `mems`: a cluster
1127 /// counts toward every schema whose mems it touches, so these figures
1128 /// can sum above the global community count — the same "touches"
1129 /// semantic as the mem-scoped count. Per-schema dedup keeps a cluster
1130 /// touching two mems of one schema from being counted twice.
1131 pub fn communities_by_schema(
1132 &self,
1133 mems: &[String],
1134 ) -> std::collections::BTreeMap<String, usize> {
1135 let louvain = self.communities();
1136 let mut buckets: std::collections::BTreeMap<String, std::collections::BTreeSet<String>> =
1137 std::collections::BTreeMap::new();
1138 for name in mems {
1139 let schema = self
1140 .mount(name)
1141 .and_then(|m| m.schema.as_ref().map(|s| s.as_display()))
1142 .unwrap_or_default();
1143 let clusters = crate::graph::community::clusters_in_mem(self.store(), louvain, name);
1144 buckets.entry(schema).or_default().extend(clusters);
1145 }
1146 buckets
1147 .into_iter()
1148 .map(|(schema, set)| (schema, set.len()))
1149 .collect()
1150 }
1151
1152 /// All mounts the engine knows about, in declaration order.
1153 /// Counterpart to [`Self::mem_names`] when the caller needs
1154 /// the full mount shape (e.g. to enumerate by storage variant).
1155 pub fn mounts(&self) -> Vec<&crate::workspace::Mount> {
1156 self.mounts.iter().map(|m| &m.mount).collect()
1157 }
1158
1159 /// Names of mems whose mount declares
1160 /// [`crate::workspace::MountCapability::Write`], in declaration
1161 /// order. Convenience over `mounts().iter().filter(...).map(...)`
1162 /// for handlers that gate by writable status (`memstead_health`,
1163 /// `memstead_overview`'s mem roster, the lifecycle tools'
1164 /// candidate list). Read-only mounts (archive backends) are
1165 /// excluded.
1166 pub fn writable_mem_names(&self) -> Vec<&str> {
1167 self.mounts
1168 .iter()
1169 .filter(|m| m.mount.capability == MountCapability::Write)
1170 .map(|m| m.mount.mem.as_str())
1171 .collect()
1172 }
1173
1174 /// The default writable mem — the target a mutation lands in when
1175 /// it omits `mem`. `None` when no writable mem is mounted.
1176 ///
1177 /// Defined as the **first writable mount in declaration order**, i.e.
1178 /// the seed / earliest-created writable mem. This is a *stable*
1179 /// designation, not a function of the current name set: new mems
1180 /// register via `register_writable_mem`, which pushes onto the end
1181 /// of the mount list (and `mounts.json` preserves that order across
1182 /// reboots), so creating an additional mem never moves the default
1183 /// — even one whose name sorts ahead alphabetically. Deleting the
1184 /// current default promotes the next-earliest writable mem; that is
1185 /// the only thing that shifts it. Both the MCP `resolve_mem` and the
1186 /// CLI's omitted-`--mem` path resolve through here so the two
1187 /// surfaces always agree (the
1188 /// pre-fix MCP path read `writable_mems().iter().next()` off an
1189 /// unordered `HashSet`, which silently retargeted writes when a second
1190 /// mem appeared).
1191 pub fn default_writable_mem(&self) -> Option<&str> {
1192 self.mounts
1193 .iter()
1194 .find(|m| m.mount.capability == MountCapability::Write)
1195 .map(|m| m.mount.mem.as_str())
1196 }
1197
1198 /// On-disk folder path for a folder-backed mount, or `None` for
1199 /// any other backend (git-branch, archive) or unknown mem.
1200 /// Convenience over `engine.mount(mem).map(|m| &m.storage)` +
1201 /// matching on `MountStorage::Folder { path }`. Used by
1202 /// handlers that need a filesystem path for a folder mem
1203 /// (e.g. `memstead_health { include_config: true }`'s
1204 /// `mems[].vcs.worktree` field for folder mounts).
1205 pub fn folder_path_for_mem(&self, mem: &str) -> Option<&Path> {
1206 match self.mount(mem).map(|m| &m.storage) {
1207 Some(crate::workspace::MountStorage::Folder { path }) => Some(path.as_path()),
1208 _ => None,
1209 }
1210 }
1211
1212 /// Runtime snapshot of writable / visible mems. Handlers that
1213 /// need the writable roster (`memstead_health`'s `writable_mems` /
1214 /// `read_mems`), per-mem origin tag (`include_config:
1215 /// true`'s `mems[].origin`), or visibility check
1216 /// (`memstead_overview`'s mem list, the lifecycle tools' collision
1217 /// guard) consume the router here. Returned by reference — the
1218 /// `Arc` is held on the engine; callers that need a clonable
1219 /// handle can `Arc::clone` the engine's field directly when that
1220 /// surface arrives.
1221 pub fn mem_router(&self) -> &MemRouterSnapshot {
1222 &self.mem_router
1223 }
1224
1225 /// Resolve the gitdir for a writable mem. Used by `memstead_health
1226 /// { include_config: true }` to surface per-mem `vcs.gitdir`
1227 /// so outer-repo bookkeeping clients can `git -C <gitdir>` per
1228 /// mem without hardcoding the layout.
1229 ///
1230 /// - `EngineError::UnknownMem` when the name does not resolve.
1231 /// - `EngineError::Mem` when the mount's storage is not
1232 /// git-branch-backed (folder, archive — they have no gitdir).
1233 pub fn gitdir_for(&self, mem_name: &str) -> Result<PathBuf, EngineError> {
1234 let m = self
1235 .mount(mem_name)
1236 .ok_or_else(|| self.unknown_mem_error(mem_name))?;
1237 match &m.storage {
1238 MountStorage::GitBranch { gitdir, .. } => Ok(gitdir.clone()),
1239 MountStorage::Folder { .. } | MountStorage::Archive { .. } | MountStorage::InMemory => {
1240 Err(EngineError::Mem(format!(
1241 "mem '{mem_name}' has no resolved gitdir"
1242 )))
1243 }
1244 }
1245 }
1246
1247 /// Resolve the worktree for a writable mem. Used by
1248 /// `memstead_health { include_config: true }` to surface per-mem
1249 /// `vcs.worktree`.
1250 ///
1251 /// - `EngineError::UnknownMem` when the name does not resolve.
1252 /// - `EngineError::Mem` when the mount's backend has no
1253 /// worktree concept (git-branch with no working tree, archive).
1254 ///
1255 /// Folder mounts surface their on-disk path. Git-branch mounts
1256 /// follow the `dir: Some(...)` composition pattern: when the
1257 /// workspace root contains a folder named after the mem with a
1258 /// `.memstead/config.json` marker, that folder is the worktree
1259 /// (disk-shape composition). Otherwise — pure mem-repo-backed
1260 /// — return Err.
1261 pub fn worktree_for(&self, mem_name: &str) -> Result<PathBuf, EngineError> {
1262 let m = self
1263 .mount(mem_name)
1264 .ok_or_else(|| self.unknown_mem_error(mem_name))?;
1265 match &m.storage {
1266 MountStorage::Folder { path } => Ok(path.clone()),
1267 MountStorage::GitBranch { .. } => {
1268 if let Some(root) = self.workspace_root.as_deref() {
1269 let candidate = root.join(mem_name);
1270 if candidate
1271 .join(crate::mem::MEM_META_DIR)
1272 .join("config.json")
1273 .is_file()
1274 {
1275 return Ok(candidate.canonicalize().unwrap_or(candidate));
1276 }
1277 }
1278 Err(EngineError::Mem(format!(
1279 "mem '{mem_name}' has no working tree (mem-repo-backed)"
1280 )))
1281 }
1282 MountStorage::Archive { .. } => Err(EngineError::Mem(format!(
1283 "mem '{mem_name}' is archive-backed and has no worktree"
1284 ))),
1285 MountStorage::InMemory => Err(EngineError::Mem(format!(
1286 "mem '{mem_name}' is in-memory and has no worktree"
1287 ))),
1288 }
1289 }
1290
1291 /// Per-mem `.memstead/config.json` payload, when available. Used
1292 /// by `memstead_health { include_config: true }` to surface the
1293 /// opaque `write_guidance` map and the catch-all `extra` fields
1294 /// per mem.
1295 ///
1296 /// Folder-backed mounts return `Some(&MemConfig)` when
1297 /// `<path>/.memstead/config.json` parsed cleanly at construction.
1298 /// Git-branch and archive backends return `None` until the
1299 /// read-from-storage-backend path lifts (the V1 unified engine
1300 /// loads configs only from folder layouts; the file lives
1301 /// inside the gitdir / archive for the other backends and
1302 /// needs a backend-level read primitive).
1303 ///
1304 /// Unknown mem names return `None` (no error variant — the
1305 /// accessor is intentionally lenient because memstead_health emits
1306 /// an empty detail block per missing config rather than
1307 /// aborting the call).
1308 pub fn mem_config_for(&self, mem: &str) -> Option<&memstead_schema::config::MemConfig> {
1309 self.mounts
1310 .iter()
1311 .find(|m| m.mount.mem == mem)
1312 .and_then(|m| m.mem_config.as_ref())
1313 }
1314
1315 /// The authoring-provenance payload an installed mem carries, read
1316 /// from the archive's `.memstead/provenance.json` at construction.
1317 /// `None` when the mem carries none (a pre-provenance archive, a
1318 /// runtime-created mem, or a backend that does not surface one) —
1319 /// the read path reports provenance as absent. Unknown mem names
1320 /// return `None`.
1321 pub fn archive_provenance_for(&self, mem: &str) -> Option<&memstead_schema::ArchiveProvenance> {
1322 self.mounts
1323 .iter()
1324 .find(|m| m.mount.mem == mem)
1325 .and_then(|m| m.archive_provenance.as_ref())
1326 }
1327
1328 /// Iterate `(mem_name, &MemConfig)` for every mount whose
1329 /// mem-config payload loaded at construction. Used by callers
1330 /// that walk every writable mount's config (`memstead health`'s
1331 /// per-mem dump, the workspace-dump CLI). The yielded `&str` is
1332 /// the authoritative mem leaf from the mount record.
1333 ///
1334 /// Folder-backed mounts yield when their `.memstead/config.json`
1335 /// parsed cleanly. Git-branch and archive backends are silent in
1336 /// V1 (the same deferred-read-from-storage gap that
1337 /// [`Self::mem_config_for`] documents).
1338 pub fn mem_configs_named(
1339 &self,
1340 ) -> impl Iterator<Item = (&str, &memstead_schema::config::MemConfig)> {
1341 self.mounts
1342 .iter()
1343 .filter_map(|m| m.mem_config.as_ref().map(|c| (m.mount.mem.as_str(), c)))
1344 }
1345
1346 /// Every configured mount, with its config when one was readable.
1347 ///
1348 /// The counterpart to [`Self::mem_configs_named`], whose contract is "mems
1349 /// WITH config" and which is therefore right to omit the rest. The trouble
1350 /// was that callers wanting to enumerate mounts reached for it anyway: a
1351 /// folder mount whose directory is gone returns no config rather than an
1352 /// error, boot stores none, and the mount became invisible to every one of
1353 /// them. Nine call sites shared that blind spot, and the omission was a
1354 /// side effect of config readability rather than anything about the mount
1355 /// (04/05, criteria 1 and 8).
1356 ///
1357 /// Callers that genuinely want only configured mems keep using the other
1358 /// one; drivers that want every mount ask for every mount.
1359 pub fn mounts_with_optional_config(
1360 &self,
1361 ) -> impl Iterator<Item = (&str, Option<&memstead_schema::config::MemConfig>)> {
1362 self.mounts
1363 .iter()
1364 .map(|m| (m.mount.mem.as_str(), m.mem_config.as_ref()))
1365 }
1366
1367 /// Resolved `Arc<Schema>` for a writable mem by name. `None`
1368 /// when the name is not a registered mount.
1369 ///
1370 /// Cheap — `Arc::clone` over the per-mem schema map. Resolved
1371 /// schemas are stored in `HashMap<String, Arc<Schema>>` so the
1372 /// lookup is a single hash hit + clone.
1373 pub fn schema_for(&self, mem: &str) -> Option<std::sync::Arc<memstead_schema::Schema>> {
1374 self.schemas.get(mem).cloned()
1375 }
1376
1377 /// Cached current branch-tip cursor (typically a 40-char hex
1378 /// SHA for git-branch backends; `None` for fresh mems or
1379 /// backends that don't track a head — folder / archive).
1380 ///
1381 /// The value is the per-mount `last_known_head`, seeded at
1382 /// construction by `backend.current_head()` and refreshed by
1383 /// [`Self::reload_if_stale`] / mutation paths after a
1384 /// successful commit.
1385 ///
1386 /// - `EngineError::UnknownMem` when the name does not resolve.
1387 pub fn mem_head_sha(&self, mem_name: &str) -> Result<Option<String>, EngineError> {
1388 let m = self
1389 .mounts
1390 .iter()
1391 .find(|m| m.mount.mem == mem_name)
1392 .ok_or_else(|| self.unknown_mem_error(mem_name))?;
1393 Ok(m.last_known_head.clone())
1394 }
1395
1396 /// Whether a sibling writer has advanced this mem's backend past
1397 /// the engine's cached `last_known_head` — a read-only drift probe
1398 /// that does **not** reload (unlike [`Self::reload_if_stale`]). One
1399 /// `backend.current_head()` read compared against the cached cursor;
1400 /// the comparison clears once the engine re-reads (a `reload` /
1401 /// `reload_if_stale` refreshes `last_known_head` to the live tip).
1402 ///
1403 /// Only git-branch backends track a head, so folder / archive /
1404 /// in-memory mounts always report `false`. A backend that errors on
1405 /// the probe (transient refdb hiccup) reports `false` rather than
1406 /// surfacing the error — drift is advisory, and the next real
1407 /// operation's reload path is the authoritative sync.
1408 ///
1409 /// - `EngineError::UnknownMem` when the name does not resolve.
1410 pub fn mem_drifted(&self, mem_name: &str) -> Result<bool, EngineError> {
1411 let m = self
1412 .mounts
1413 .iter()
1414 .find(|m| m.mount.mem == mem_name)
1415 .ok_or_else(|| self.unknown_mem_error(mem_name))?;
1416 let live = m.backend.current_head().ok().flatten();
1417 Ok(live != m.last_known_head)
1418 }
1419
1420 /// Workspace root the engine booted from, when one is known.
1421 /// `None` for engines built directly from a mount list (tests,
1422 /// ad-hoc consumers). Set by [`Self::from_workspace_root`] and
1423 /// the full counterpart.
1424 pub fn workspace_root(&self) -> Option<&Path> {
1425 self.workspace_root.as_deref()
1426 }
1427
1428 /// Typed warnings surfaced during mem load — drift findings
1429 /// the loader pipeline collects per entity. Empty for V1; the
1430 /// accessor surfaces them so handlers can merge into health
1431 /// summaries uniformly.
1432 pub fn load_warnings(&self) -> &[WarningHint] {
1433 &self.load_warnings
1434 }
1435
1436 /// The quarantine roster: mems that failed their mem-level boot
1437 /// step and serve nothing until repaired + reloaded. Empty on a
1438 /// fully healthy workspace. Surfaced on overview and health.
1439 pub fn quarantined_mems(&self) -> &[crate::engine::QuarantinedMem] {
1440 &self.quarantined
1441 }
1442
1443 /// The quarantine entry for `mem`, when it is quarantined.
1444 pub fn quarantine_reason(&self, mem: &str) -> Option<&crate::engine::QuarantinedMem> {
1445 self.quarantined.iter().find(|q| q.mount.mem == mem)
1446 }
1447
1448 /// Whether the store's entity set for `mem` can be trusted to answer the
1449 /// question "does this entity still exist?".
1450 ///
1451 /// WHY this is a question and not an assumption: an anchor sidecar is
1452 /// keyed by entity id, and a key with no entity behind it is a dangling
1453 /// row (consistency-sweep 03/02). Detecting one means reading a NEGATIVE
1454 /// from the store, and a negative is only evidence when the store is known
1455 /// to hold everything the mem has. Four states break that, and each of
1456 /// them would otherwise turn every anchor in the mem into a false dangling
1457 /// report: the mem is not mounted at all, it is quarantined (serving
1458 /// nothing), its lazy load has not run yet (mounted, entities absent), or
1459 /// a file in it failed to parse, in which case an id missing from the
1460 /// store may be a load failure rather than a deleted entity.
1461 ///
1462 /// The last case is deliberately COARSE for non-folder mounts: load-error
1463 /// paths are normalized to absolute only for folder mounts, so a
1464 /// git-branch mem's errors carry mem-relative paths that two mems can
1465 /// spell identically. Attributing them by name would be a guess, and a
1466 /// wrong guess here fabricates dangling rows. Any load error at all
1467 /// therefore blocks reconciliation for a non-folder mem. The caller states
1468 /// the block rather than skipping silently, which is the honest direction.
1469 pub fn entity_set_is_reconcilable(&self, mem: &str) -> Result<(), &'static str> {
1470 if self.quarantine_reason(mem).is_some() {
1471 return Err("the mem is quarantined and serves no entities");
1472 }
1473 let Some(mount) = self.mounts.iter().find(|m| m.mount.mem == mem) else {
1474 return Err("the mem is not mounted here");
1475 };
1476 if mount.deferred {
1477 return Err("the mem's lazy entity load has not run this session");
1478 }
1479 if self.load_errors.is_empty() {
1480 return Ok(());
1481 }
1482 match &mount.mount.storage {
1483 crate::workspace::MountStorage::Folder { path } => {
1484 if self.load_errors.iter().any(|(p, _)| p.starts_with(path)) {
1485 Err(
1486 "a file in this mem failed to parse, so an id missing from the store may be a load failure rather than a deleted entity",
1487 )
1488 } else {
1489 Ok(())
1490 }
1491 }
1492 _ => Err(
1493 "this workspace has files that failed to parse, and their paths cannot be attributed to one mem",
1494 ),
1495 }
1496 }
1497
1498 /// Whether `id` names an entity this mem no longer holds. A STUB counts
1499 /// as missing: a stub is the placeholder an unresolved wiki-link target
1500 /// leaves behind, not an entity anyone wrote, so an anchor keyed to one
1501 /// is dangling exactly as if nothing were there.
1502 ///
1503 /// Only meaningful once [`Self::entity_set_is_reconcilable`] says yes.
1504 pub fn entity_is_absent(&self, id: &EntityId) -> bool {
1505 self.store.get(id).is_none_or(|e| e.stub)
1506 }
1507
1508 /// Mems whose lazy entity load is still DEFERRED — on the mount
1509 /// roster with a resolved schema pin, but with no entities in the
1510 /// store yet. Read surfaces that render per-mem counts or
1511 /// distributions consult this: a count over a deferred mem's slice
1512 /// of the store is a count over nothing, and rendering it as a
1513 /// bare zero is the silent absence the lazy-mount contract forbids.
1514 /// Either trigger the load ([`Self::ensure_mems_loaded`]) or render
1515 /// the load state explicitly.
1516 pub fn deferred_mems(&self) -> Vec<&str> {
1517 self.mounts
1518 .iter()
1519 .filter(|m| m.deferred)
1520 .map(|m| m.mount.mem.as_str())
1521 .collect()
1522 }
1523
1524 /// Whether `mem` is a lazy mount whose entity load has not run yet.
1525 pub fn mem_is_deferred(&self, mem: &str) -> bool {
1526 self.mounts.iter().any(|m| m.deferred && m.mount.mem == mem)
1527 }
1528
1529 /// The typed error for a mem name that did not resolve to a
1530 /// serving mount: `MEM_QUARANTINED` (carrying the underlying boot
1531 /// failure and its repair command) when the mem is on the
1532 /// quarantine roster, `UNKNOWN_MEM` otherwise. Every lookup site
1533 /// that fails to find a mem routes here so a quarantined mem is
1534 /// never misreported as unknown — honest absence, with the reason.
1535 /// Resolve an entity id that may lack its mem prefix — the one
1536 /// rule every id-taking mutation verb applies before it reads the
1537 /// id's mem. A full id (`mem--slug`) returns unchanged with no
1538 /// hint, byte for byte the path it always took. A bare slug is
1539 /// looked up across every mounted mem (loading deferred mems
1540 /// first, so a lazily mounted carrier is not invisible): exactly
1541 /// one carrier resolves the id and returns the
1542 /// [`crate::ops::WarningHint::ShortIdResolved`] announcement the
1543 /// verb rides on its outcome; zero or several carriers refuse
1544 /// [`EngineError::EntityIdMissingMem`] naming every candidate.
1545 /// Before this rule a bare slug reached the verbs as an id whose
1546 /// mem was the empty string, and the caller was told a mem called
1547 /// "" did not exist (B7 grader finding, 2026-09-02).
1548 pub fn resolve_entity_id(
1549 &mut self,
1550 id: &EntityId,
1551 ) -> Result<(EntityId, Option<crate::ops::WarningHint>), EngineError> {
1552 if !id.mem().is_empty() {
1553 return Ok((id.clone(), None));
1554 }
1555 // Grammar before resolution: a malformed bare string is an
1556 // invalid id, not a slug nobody carries.
1557 let slug = id.0.as_str();
1558 if let Err(reason) = crate::entity::id::validate_id_path_grammar(slug) {
1559 return Err(EngineError::InvalidEntityId {
1560 id: id.0.clone(),
1561 reason,
1562 });
1563 }
1564 self.ensure_mems_loaded(None);
1565 let mut candidates: Vec<EntityId> = self
1566 .store
1567 .all_entities()
1568 .filter(|e| !e.stub && e.id.path() == slug)
1569 .map(|e| e.id.clone())
1570 .collect();
1571 candidates.sort_by(|a, b| a.0.cmp(&b.0));
1572 candidates.dedup();
1573 if candidates.len() == 1 {
1574 let resolved = candidates.remove(0);
1575 let hint = crate::ops::WarningHint::ShortIdResolved {
1576 given: id.0.clone(),
1577 resolved: resolved.clone(),
1578 };
1579 return Ok((resolved, Some(hint)));
1580 }
1581 Err(EngineError::EntityIdMissingMem {
1582 id: id.0.clone(),
1583 candidates: candidates.into_iter().map(|c| c.0).collect(),
1584 })
1585 }
1586
1587 pub fn unknown_mem_error(&self, mem: &str) -> EngineError {
1588 match self.quarantine_reason(mem) {
1589 Some(q) => EngineError::MemQuarantined {
1590 mem: mem.to_string(),
1591 reason_code: q.reason_code.clone(),
1592 reason_message: q.reason_message.clone(),
1593 },
1594 None if self.recently_unmounted(mem) => EngineError::MemUnmounted {
1595 mem: mem.to_string(),
1596 },
1597 None => EngineError::UnknownMem(mem.to_string()),
1598 }
1599 }
1600
1601 /// The workspace-level boot diagnosis a diagnostic-shell engine
1602 /// carries (`None` on ordinarily booted engines).
1603 pub fn boot_diagnosis(&self) -> Option<(&str, &str)> {
1604 self.boot_diagnosis
1605 .as_ref()
1606 .map(|(c, m)| (c.as_str(), m.as_str()))
1607 }
1608
1609 /// Build a mem-less diagnostic-shell engine for a workspace whose
1610 /// boot failed at the WORKSPACE level (nothing loadable — e.g. an
1611 /// unparseable store). It serves no mems and no entities; its one
1612 /// job is answering overview/health with the typed boot diagnosis
1613 /// so a session can always ask WHY the graph is gone — the MCP
1614 /// server serves this instead of exiting into `-32000 Connection
1615 /// closed` (degrade, never disappear).
1616 pub fn diagnostic_shell(reason_code: String, reason_message: String) -> Engine {
1617 let mut engine =
1618 Engine::from_mounts(Vec::new()).expect("an empty mount list always constructs");
1619 engine.boot_diagnosis = Some((reason_code, reason_message));
1620 engine
1621 }
1622
1623 /// Append boot-path quarantine entries recorded outside
1624 /// `from_mounts_inner` (backend-instantiation failures happen
1625 /// before the mount list reaches the engine constructor). Boot
1626 /// paths only — quarantine is a boot judgment, never a runtime
1627 /// mutation.
1628 pub fn extend_quarantine(&mut self, entries: Vec<crate::engine::QuarantinedMem>) {
1629 self.quarantined.extend(entries);
1630 }
1631
1632 // ---------------------------------------------------------------
1633 // Read-side delegates onto the kernel ops/graph functions.
1634 //
1635 // The mem-router engine exposed each of these directly so the
1636 // MCP layer could call them without reaching into the store. The
1637 // unified engine mirrors that surface so the MCP migration is a
1638 // straight rename rather than a re-architecture.
1639 //
1640 // Multi-mem cache strategy: per-mem community detection and
1641 // per-mem search indexes are unnecessary at this layer — the
1642 // engine-wide store already carries every mount's edges; Louvain
1643 // and tantivy run once across the union. `mem_schemas` for
1644 // health/search is the engine's existing `schemas` field as-is.
1645 // ---------------------------------------------------------------
1646
1647 /// Lazy community-detection cache. First call runs Louvain
1648 /// against the current store using one pinned schema for
1649 /// `community.{resolution, seed}` and the per-rel weights.
1650 /// Subsequent calls return the cached result. Mutations invalidate
1651 /// the cache via [`Self::invalidate_communities`].
1652 ///
1653 /// One detection run per engine. The partition is workspace-global,
1654 /// so it needs a single source for the Louvain parameters; that
1655 /// source is the schema of the lexicographically-first mem name —
1656 /// a stable key, so the partition is deterministic across processes
1657 /// even when mounts pin heterogeneous schemas. For a single-schema
1658 /// workspace every mem's schema is identical, so the choice of
1659 /// key is immaterial there.
1660 pub fn communities(&self) -> &LouvainOutput {
1661 // Generation sanity: a stored memo must match the live store
1662 // generation — the mutation hooks clear stale memos before any
1663 // read can arrive here. A mismatch would mean a mutation path
1664 // skipped its invalidation call (the exact silent-staleness
1665 // bug the generation exists to catch).
1666 if let Some((memo_key, _)) = self.community_memo.get() {
1667 debug_assert_eq!(
1668 *memo_key,
1669 self.derived_key(),
1670 "community memo key lags the engine — a mutation path missed invalidate_communities"
1671 );
1672 }
1673 &self
1674 .community_memo
1675 .get_or_init(|| (self.derived_key(), self.compute_communities()))
1676 .1
1677 }
1678
1679 /// The current validity key for derived-structure memos: store
1680 /// generation plus schemas epoch (see [`super::DerivedKey`]).
1681 pub fn derived_key(&self) -> super::DerivedKey {
1682 super::DerivedKey {
1683 store_generation: self.store.generation(),
1684 schemas_epoch: self.schemas_epoch,
1685 }
1686 }
1687
1688 fn compute_communities(&self) -> LouvainOutput {
1689 {
1690 // Select the parameter schema by a stable key (smallest
1691 // mem name) rather than unordered-map iteration, so the
1692 // partition does not vary between processes. Fall back to
1693 // the builtin default for the empty-mounts case (caller
1694 // still gets a valid empty Louvain result against an empty
1695 // store).
1696 let schema = self
1697 .schemas
1698 .iter()
1699 .min_by(|a, b| a.0.cmp(b.0))
1700 .map(|(_, s)| s.clone())
1701 .unwrap_or_else(Schema::builtin_default);
1702 let manifest = &schema.manifest;
1703 let resolution = manifest.community.resolution;
1704 let seed = manifest.community.seed;
1705 let schema_for_weights = schema.clone();
1706 detect_communities(&self.store, resolution, seed, move |rel_type| {
1707 schema_for_weights
1708 .manifest
1709 .relationships
1710 .definitions
1711 .iter()
1712 .find(|d| d.name == rel_type)
1713 .map(|d| d.default_weight as f64)
1714 .unwrap_or(1.0)
1715 })
1716 }
1717 }
1718
1719 /// Drop the cached community detection result and the grounded
1720 /// labelling memo — unless the store still sits at the generation
1721 /// each memo was computed from (flywheel W8/01). The keep case is
1722 /// exactly the batch-rollback path: the restored snapshot restored
1723 /// the generation with it, so the memo describes the live state
1724 /// and recomputing it would be pure waste. Every real mutation
1725 /// bumps the generation first, so those clears behave as before.
1726 /// Coupling the labelling reset here means every site that already
1727 /// invalidates communities (all mutation paths, drift reload,
1728 /// quarantine attach/detach, apply-commit) invalidates the
1729 /// labelling too, so a stale label can never outlive the state
1730 /// change that moved it.
1731 pub fn invalidate_communities(&mut self) {
1732 let key = self.derived_key();
1733 if !matches!(self.community_memo.get(), Some((k, _)) if *k == key) {
1734 self.community_memo = OnceCell::new();
1735 }
1736 if !matches!(self.labelling_memo.get(), Some((k, _)) if *k == key) {
1737 self.labelling_memo = OnceCell::new();
1738 }
1739 }
1740
1741 /// The grounded labelling of one mem — `None` when its pinned
1742 /// schema declares no `relationships.labelling`. Computed on
1743 /// first access for every declaring mem, memoised until the next
1744 /// invalidation; generation-keyed like `community_memo`.
1745 pub fn mem_labelling(&self, mem: &str) -> Option<&crate::ops::labelling::MemLabelling> {
1746 // Generation sanity, same contract as `communities()`: a
1747 // stored memo must match the live derived key — a mismatch
1748 // means a mutation path missed invalidate_communities.
1749 if let Some((memo_key, _)) = self.labelling_memo.get() {
1750 debug_assert_eq!(
1751 *memo_key,
1752 self.derived_key(),
1753 "labelling memo key lags the engine — a mutation path missed invalidate_communities"
1754 );
1755 }
1756 let (_, map) = self.labelling_memo.get_or_init(|| {
1757 let mut out = std::collections::HashMap::new();
1758 for (mem_name, schema) in &self.schemas {
1759 if let Some(lab) = crate::ops::labelling::labelling_of(schema) {
1760 out.insert(
1761 mem_name.clone(),
1762 crate::ops::labelling::compute_mem_labelling(
1763 &self.store,
1764 mem_name,
1765 &lab.attack,
1766 ),
1767 );
1768 }
1769 }
1770 (self.derived_key(), out)
1771 });
1772 map.get(mem)
1773 }
1774
1775 /// One entity's served labelling view — `None` when the entity is
1776 /// a stub or its mem's schema declares no labelling; serving
1777 /// surfaces then keep their byte-identical payloads. The shape
1778 /// block is present exactly when the declaration carries a
1779 /// `support` walk.
1780 pub fn computed_labelling(
1781 &self,
1782 entity: &Entity,
1783 ) -> Option<crate::ops::labelling::LabellingView> {
1784 use crate::ops::labelling::{Label, compute_shape, labelling_of};
1785 if entity.stub {
1786 return None;
1787 }
1788 let schema = self.schemas.get(entity.mem.as_str())?;
1789 let lab = labelling_of(schema)?;
1790 let mem_lab = self.mem_labelling(entity.mem.as_str())?;
1791 let label = *mem_lab.labels.get(entity.id.0.as_str())?;
1792 let defeated_by = if label == Label::Defeated {
1793 mem_lab.accepted_attackers_of(entity.id.0.as_str())
1794 } else {
1795 Vec::new()
1796 };
1797 let undecided_by = if label == Label::Undecided {
1798 mem_lab.undecided_attackers_of(entity.id.0.as_str())
1799 } else {
1800 Vec::new()
1801 };
1802 let shape = lab.support.as_ref().map(|walk| {
1803 let label_of = |id: &EntityId| -> Option<Label> {
1804 self.mem_labelling(id.mem())
1805 .and_then(|ml| ml.labels.get(id.0.as_str()).copied())
1806 };
1807 compute_shape(&self.store, &entity.id, walk, &label_of)
1808 });
1809 Some(crate::ops::labelling::LabellingView {
1810 label,
1811 defeated_by,
1812 undecided_by,
1813 shape,
1814 })
1815 }
1816
1817 /// The `labelling` health axis payload — per declaring mem:
1818 /// counts per label, the defeated list with its accepted
1819 /// attackers, the undecided list with its open attacker set, and
1820 /// the excluded cross-mem attack-edge count. One composer shared
1821 /// by the CLI health command and both MCP flavours.
1822 pub fn health_labelling_axis(&self, mem_filter: Option<&str>) -> serde_json::Value {
1823 use crate::ops::labelling::Label;
1824 let mut mems = serde_json::Map::new();
1825 let mut mem_names: Vec<&String> = self.schemas.keys().collect();
1826 mem_names.sort();
1827 for mem in mem_names {
1828 if let Some(v) = mem_filter
1829 && mem != v
1830 {
1831 continue;
1832 }
1833 let Some(ml) = self.mem_labelling(mem) else {
1834 continue;
1835 };
1836 let mut accepted = 0usize;
1837 let mut defeated: Vec<serde_json::Value> = Vec::new();
1838 let mut undecided: Vec<serde_json::Value> = Vec::new();
1839 for (id, label) in &ml.labels {
1840 match label {
1841 Label::Accepted => accepted += 1,
1842 Label::Defeated => defeated.push(serde_json::json!({
1843 "id": id,
1844 "defeated_by": ml.accepted_attackers_of(id),
1845 })),
1846 Label::Undecided => undecided.push(serde_json::json!({
1847 "id": id,
1848 "undecided_by": ml.undecided_attackers_of(id),
1849 })),
1850 }
1851 }
1852 mems.insert(
1853 mem.clone(),
1854 serde_json::json!({
1855 "counts": {
1856 "accepted": accepted,
1857 "defeated": defeated.len(),
1858 "undecided": undecided.len(),
1859 },
1860 "defeated": defeated,
1861 "undecided": undecided,
1862 "cross_mem_edges_excluded": ml.cross_mem_edges_excluded,
1863 }),
1864 );
1865 }
1866 serde_json::Value::Object(mems)
1867 }
1868
1869 /// Real entities with no incoming or outgoing edges — leaf-declared
1870 /// types exempt (their edge-less entities are terminal by
1871 /// construction; see [`Self::leaf_population`]).
1872 pub fn orphans(&self) -> Vec<EntityId> {
1873 crate::graph::query::find_orphans_with_schemas(&self.store, &self.schemas)
1874 }
1875
1876 /// Count of real entities per leaf-declared type, keyed
1877 /// `<schema_ref>:<type>` — the visible population the orphan
1878 /// exemption covers.
1879 pub fn leaf_population(&self) -> std::collections::BTreeMap<String, usize> {
1880 crate::graph::query::leaf_population(&self.store, &self.schemas)
1881 }
1882
1883 /// Stub entities with their referencer ids.
1884 pub fn stubs(&self) -> Vec<(EntityId, Vec<EntityId>)> {
1885 crate::graph::query::find_stubs(&self.store)
1886 }
1887
1888 /// Top `limit` entities by total degree.
1889 pub fn most_connected(&self, limit: usize) -> Vec<crate::graph::query::Connectivity> {
1890 crate::graph::query::most_connected(&self.store, limit)
1891 }
1892
1893 /// Entities whose type's `required_outgoing` blocks are not yet
1894 /// satisfied. `mem_filter = None` scans every mem; `Some(v)`
1895 /// scans only that mem.
1896 pub fn missing_required_outgoing(
1897 &self,
1898 mem_filter: Option<&str>,
1899 ) -> Vec<crate::ops::health::MissingRequiredOutgoingReport> {
1900 crate::ops::health::collect_missing_required_outgoing(
1901 &self.store,
1902 mem_filter,
1903 &self.schemas,
1904 )
1905 }
1906
1907 /// Standing violations of declared `constraints` (the health
1908 /// `constraints` include) — every non-stub entity whose type
1909 /// declares constraints its current state violates, in
1910 /// deterministic `(mem, id)` order.
1911 pub fn constraint_findings(
1912 &self,
1913 mem_filter: Option<&str>,
1914 ) -> Vec<crate::ops::health::ConstraintFindingReport> {
1915 let check_provider = self.check_standing_provider();
1916 crate::ops::health::collect_constraint_findings(
1917 &self.store,
1918 mem_filter,
1919 &self.schemas,
1920 Some(&check_provider),
1921 )
1922 }
1923
1924 /// The evaluated aggregate signals for one entity — `None` when
1925 /// the mem has no schema, the type is unknown or a stub, or the
1926 /// type declares no signals; serving surfaces then keep their
1927 /// byte-identical payloads.
1928 pub fn computed_signals(
1929 &self,
1930 entity: &Entity,
1931 ) -> Option<Vec<crate::ops::signals::ComputedSignal>> {
1932 if entity.stub {
1933 return None;
1934 }
1935 let schema = self.schemas.get(entity.mem.as_str())?;
1936 let td = schema.types.get(entity.entity_type.as_str())?;
1937 if td.signals.is_empty() {
1938 return None;
1939 }
1940 Some(crate::ops::signals::compute_signals(
1941 &self.store,
1942 td,
1943 &entity.id,
1944 ))
1945 }
1946
1947 /// Every entity carrying at least one signal above `none` — the
1948 /// include-gated `signals` health axis.
1949 pub fn signal_reports(
1950 &self,
1951 mem_filter: Option<&str>,
1952 ) -> Vec<crate::ops::health::SignalReport> {
1953 crate::ops::health::collect_signal_reports(&self.store, mem_filter, &self.schemas)
1954 }
1955
1956 /// The `signals` health axis payload — the entity roster plus
1957 /// per-level counts. One composer shared by the CLI health
1958 /// command and both MCP flavours so the axis cannot drift
1959 /// between surfaces.
1960 pub fn health_signals_axis(&self, mem_filter: Option<&str>) -> serde_json::Value {
1961 use memstead_schema::SignalLevel;
1962 let reports = self.signal_reports(mem_filter);
1963 let mut notice = 0usize;
1964 let mut warn = 0usize;
1965 for r in &reports {
1966 for s in &r.signals {
1967 match s.level {
1968 Some(SignalLevel::Notice) => notice += 1,
1969 Some(SignalLevel::Warn) => warn += 1,
1970 None => {}
1971 }
1972 }
1973 }
1974 serde_json::json!({
1975 "entities": reports,
1976 "counts": { "notice": notice, "warn": warn },
1977 })
1978 }
1979
1980 /// Defective section-format declarations the loaded schemas carry
1981 /// (lenient boot recorded them; install would have refused).
1982 pub fn schema_format_defects(&self) -> Vec<crate::ops::health::SchemaFormatDefect> {
1983 crate::ops::health::collect_schema_format_defects(&self.schemas)
1984 }
1985
1986 /// Conformance-axis integrity findings for one mem — which
1987 /// entities a write would refuse under the effective schema, and
1988 /// why. `target_schema = None` lints against the mem's current
1989 /// pin; `Some(ref)` lints against that schema instead (resolved
1990 /// among mem-pinned, workspace, and built-in schemas).
1991 pub fn conformance_findings(
1992 &self,
1993 mem: &str,
1994 target_schema: Option<&memstead_schema::SchemaRef>,
1995 ) -> Result<Vec<crate::ops::integrity::IntegrityFinding>, EngineError> {
1996 let pinned = self
1997 .schemas
1998 .get(mem)
1999 .ok_or_else(|| self.unknown_mem_error(mem))?;
2000 let effective: Arc<Schema> = match target_schema {
2001 None => pinned.clone(),
2002 Some(target) => self.resolve_schema_by_ref(target).ok_or_else(|| {
2003 let consulted: Vec<_> = self
2004 .workspace_schemas
2005 .iter()
2006 .chain(self.builtin_schemas.iter())
2007 .cloned()
2008 .collect();
2009 EngineError::SchemaNotFound {
2010 mem: mem.to_string(),
2011 pin: target.as_display(),
2012 sources: crate::engine::error::SchemaSourceDiagnostic::for_failed_pin(
2013 &target.name,
2014 &target.version,
2015 &consulted,
2016 ),
2017 install_hint: None,
2018 }
2019 .with_schema_install_probe(self.workspace_root())
2020 })?,
2021 };
2022 Ok(crate::ops::integrity::conformance_findings(
2023 &self.store,
2024 mem,
2025 &effective,
2026 &self.schemas,
2027 ))
2028 }
2029
2030 /// What `mem`'s entity BODIES carry that their types do not declare
2031 /// (consistency-sweep 04/01): headings absorbed into the catch-all,
2032 /// headings repeated so that later bodies were not kept, and frontmatter
2033 /// keys the next write will drop.
2034 ///
2035 /// A folder mem's ledger set against its file set, per mem.
2036 ///
2037 /// **Folder mems only, and that is the point** (04/04, criterion 4). On a
2038 /// git-branch mem the change set is a real two-tree diff against the
2039 /// committed tree, so ledger-versus-files divergence is structurally
2040 /// impossible; emitting an always-clean version of this check there would
2041 /// be a surface asserting something it never had to establish, which is
2042 /// the failure class this bundle exists to remove. Such a mem is absent
2043 /// from the map rather than present and empty.
2044 pub fn ledger_reconciliation(
2045 &self,
2046 ) -> std::collections::BTreeMap<String, crate::filesystem::changelog::LedgerReconciliation>
2047 {
2048 let mut out = std::collections::BTreeMap::new();
2049 for m in &self.mounts {
2050 let crate::workspace::MountStorage::Folder { path } = &m.mount.storage else {
2051 continue;
2052 };
2053 if let Ok(r) = crate::filesystem::changelog::reconcile_ledger(path) {
2054 out.insert(m.mount.mem.clone(), r);
2055 }
2056 }
2057 out
2058 }
2059
2060 /// Separate from [`Self::conformance_findings`] on purpose. These are
2061 /// observations, not violations: absorbing an undeclared heading is the
2062 /// catch-all working as designed, and reporting it as a finding would fail
2063 /// every mem that uses the feature. What a reader needs here is whether the
2064 /// content SURVIVES, which each observation states.
2065 pub fn body_observations(
2066 &self,
2067 mem: &str,
2068 target_schema: Option<&memstead_schema::SchemaRef>,
2069 ) -> Result<Vec<crate::ops::integrity::BodyObservation>, EngineError> {
2070 let pinned = self
2071 .schemas
2072 .get(mem)
2073 .ok_or_else(|| self.unknown_mem_error(mem))?;
2074 let effective: Arc<Schema> = match target_schema {
2075 None => pinned.clone(),
2076 Some(target) => {
2077 self.resolve_schema_by_ref(target)
2078 .ok_or_else(|| EngineError::SchemaNotFound {
2079 mem: mem.to_string(),
2080 pin: target.as_display(),
2081 sources: Vec::new(),
2082 install_hint: None,
2083 })?
2084 }
2085 };
2086 Ok(crate::ops::integrity::body_observations(
2087 &self.store,
2088 mem,
2089 &effective,
2090 ))
2091 }
2092
2093 /// Resolve an exact `name@version` ref against every schema this
2094 /// engine can see: mem-pinned, workspace-authored, built-in.
2095 /// `None` when no loaded schema matches.
2096 pub(crate) fn resolve_schema_by_ref(
2097 &self,
2098 target: &memstead_schema::SchemaRef,
2099 ) -> Option<Arc<Schema>> {
2100 self.schemas
2101 .values()
2102 .chain(self.workspace_schemas.iter())
2103 .chain(self.builtin_schemas.iter())
2104 .find(|s| {
2105 let (name, version) = s.id();
2106 name == target.name && version == target.version
2107 })
2108 .cloned()
2109 }
2110
2111 /// The mem's `Mount.schema` expectation assertion, when set.
2112 /// `None` for unknown mems *and* for mems whose mount carries no
2113 /// assertion (the authoritative pin then lives in the backend
2114 /// config; the resolved active schema, not this, is the effective pin).
2115 pub fn schema_pin(&self, mem: &str) -> Option<memstead_schema::SchemaRef> {
2116 self.mounts
2117 .iter()
2118 .find(|m| m.mount.mem == mem)
2119 .and_then(|m| m.mount.schema.clone())
2120 }
2121
2122 /// The mem's in-flight migration target, when dual-pin state is
2123 /// active. `None` for settled or unknown mems.
2124 pub fn migration_target(&self, mem: &str) -> Option<memstead_schema::SchemaRef> {
2125 self.mounts
2126 .iter()
2127 .find(|m| m.mount.mem == mem)
2128 .and_then(|m| m.mount.migration_target.clone())
2129 }
2130
2131 /// Consistency-axis integrity findings for one mem — the
2132 /// pre-existing graph-coherence categories (dangling links, stubs)
2133 /// plus cross-mem edges the workspace grant table no longer permits,
2134 /// projected into the `{ id, axis, code, detail }` finding shape.
2135 pub fn consistency_findings(
2136 &self,
2137 mem: &str,
2138 ) -> Result<Vec<crate::ops::integrity::IntegrityFinding>, EngineError> {
2139 if !self.schemas.contains_key(mem) {
2140 return Err(self.unknown_mem_error(mem));
2141 }
2142 let mut findings = crate::ops::integrity::consistency_findings(
2143 &self.store,
2144 mem,
2145 // The one grant resolver, the same one the write gate calls. Every
2146 // consumer of the axis reaches it through this funnel, so there is
2147 // no site where a second answer could be written (04/07).
2148 &|from, to| self.cross_mem_link_allowed(from, to),
2149 // The mount set, so an edge into a mem that is not mounted is
2150 // reported as the dangling finding it is and never as a grant
2151 // the table still names (backlog-decisions plan B8).
2152 &|target| self.mount(target).is_some(),
2153 );
2154 // The provenance layer's own consistency: a sidecar the engine cannot
2155 // read is a finding on the mem, beside the entity-level ones. Without
2156 // it every anchor surface reads zero rows over this mem and calls
2157 // that clean (backlog, found by the evidence-engine grader 2026-09-02).
2158 if let Some(why) = self.anchors_sidecar_error(mem) {
2159 findings.push(crate::ops::integrity::IntegrityFinding {
2160 id: mem.to_string(),
2161 axis: crate::ops::integrity::IntegrityAxis::Consistency,
2162 code: "ANCHORS_SIDECAR_UNREADABLE".to_string(),
2163 detail: serde_json::json!({
2164 "mem": mem,
2165 "reason": why,
2166 "repair": "the anchors sidecar is unreadable, so every anchor surface for this mem reports a condition instead of rows; fix or remove the sidecar file, then re-record anchors",
2167 }),
2168 });
2169 }
2170 Ok(findings)
2171 }
2172
2173 /// Every cross-mem edge the workspace's current grant resolution does
2174 /// not permit, across every visible mem.
2175 ///
2176 /// A projection of [`Self::consistency_findings`] filtered to the one
2177 /// code, not a second scan: the revoke path and the health axis must
2178 /// never be able to answer differently about the same edge, and the
2179 /// surest way to guarantee that is for one of them to BE the other
2180 /// (04/07, criterion 8).
2181 ///
2182 /// Call it after the grant edit has landed and the settings have been
2183 /// reloaded — the answer is "what does the CURRENT policy leave
2184 /// unbacked", which is what an operator revoking a grant wants to know.
2185 ///
2186 /// Takes `&mut self` because it must load lazily-deferred mems first. A
2187 /// deferred mem's entities are not in the store, so scanning without the
2188 /// load would report zero ungranted edges for it and call that clean —
2189 /// which is precisely the silent all-clear this whole axis exists to
2190 /// prevent. The long-lived server engines carry lazy mounts; the CLI's
2191 /// fresh boot does not, so this only bites on the surface where it is
2192 /// hardest to notice.
2193 pub fn ungranted_cross_mem_edges(&mut self) -> Vec<crate::ops::integrity::IntegrityFinding> {
2194 self.ensure_mems_loaded(None);
2195 let mut mems: Vec<&String> = self.schemas.keys().collect();
2196 mems.sort();
2197 mems.into_iter()
2198 .filter_map(|mem| self.consistency_findings(mem).ok())
2199 .flatten()
2200 .filter(|f| f.code == "CROSS_MEM_EDGE_UNGRANTED")
2201 .collect()
2202 }
2203
2204 /// The edges that went from permitted to unpermitted between two
2205 /// readings of [`Self::ungranted_cross_mem_edges`] — what a policy edit
2206 /// just orphaned, as opposed to what was already orphaned before it.
2207 ///
2208 /// A revocation that reported the whole standing set would blame this
2209 /// edit for every edge some earlier unrelated revocation left behind,
2210 /// which is a different and less useful claim (04/07, criterion 5).
2211 pub fn newly_ungranted(
2212 before: &[crate::ops::integrity::IntegrityFinding],
2213 after: Vec<crate::ops::integrity::IntegrityFinding>,
2214 ) -> Vec<crate::ops::integrity::IntegrityFinding> {
2215 let seen: std::collections::BTreeSet<(String, String)> = before
2216 .iter()
2217 .map(|f| (f.id.clone(), f.detail["target_id"].to_string()))
2218 .collect();
2219 after
2220 .into_iter()
2221 .filter(|f| !seen.contains(&(f.id.clone(), f.detail["target_id"].to_string())))
2222 .collect()
2223 }
2224
2225 /// Engine-wide health summary across every mount.
2226 pub fn health(&self) -> crate::ops::HealthSummary {
2227 self.health_inner(None)
2228 }
2229
2230 /// Health summary scoped to one visible mem. The scans and
2231 /// structural counts narrow to that mem; workspace-level facts
2232 /// (quarantine roster, boot diagnosis, workspace-scoped warnings)
2233 /// stay global — an agent scoping to one mem must still see them.
2234 /// A name that is quarantined or not on the visible roster refuses
2235 /// `UNKNOWN_MEM` — the same gate `search` applies to its `mem`
2236 /// filter, so "no such mem" and "healthy mem, nothing to report"
2237 /// can never be confused. `None` is the engine-wide sweep.
2238 pub fn health_scoped(
2239 &self,
2240 mem: Option<&str>,
2241 ) -> Result<crate::ops::HealthSummary, crate::EngineError> {
2242 if let Some(name) = mem
2243 && (self.quarantine_reason(name).is_some() || !self.mem_router.is_visible(name))
2244 {
2245 return Err(self.unknown_mem_error(name));
2246 }
2247 Ok(self.health_inner(mem))
2248 }
2249
2250 fn health_inner(&self, mem: Option<&str>) -> crate::ops::HealthSummary {
2251 let fallback = engine_fallback_type();
2252 let mut summary =
2253 crate::ops::health::compute_health(&self.store, fallback.as_ref(), &self.schemas, mem);
2254 self.apply_anchor_clock(&mut summary, mem);
2255 // Merge in load-time drift warnings so every caller of
2256 // Engine::health — MCP handler, Swift FFI, direct CLI —
2257 // sees the SuspiciousNestedPrefix / DuplicateSectionHeading
2258 // findings without reaching into private engine state. The
2259 // MCP handler further appends request-scoped warnings on
2260 // top. Mirrors full's merge.
2261 if !self.load_warnings.is_empty() {
2262 let mut merged = self.load_warnings.clone();
2263 merged.append(&mut summary.warnings);
2264 summary.warnings = merged;
2265 }
2266 // A standing property, reported here rather than on every boot: a
2267 // folder mem's drift cursor is its own ledger, which only the engine
2268 // writes, so an edit made to its files by anything else is invisible
2269 // and reads keep serving the pre-edit content. Silence about that is
2270 // the one outcome 04/04's criterion 3 forbids. Git-branch mems are
2271 // absent: their change set is a real two-tree diff, so the condition
2272 // cannot arise (criterion 4).
2273 for m in &self.mounts {
2274 if matches!(
2275 m.mount.storage,
2276 crate::workspace::MountStorage::Folder { .. }
2277 ) && mem.is_none_or(|scope| scope == m.mount.mem)
2278 {
2279 summary
2280 .warnings
2281 .push(crate::ops::WarningHint::OutOfBandEditsUndetected {
2282 mem: m.mount.mem.clone(),
2283 });
2284 }
2285 }
2286 // Quarantine roster — a boot-honesty fact, present whenever
2287 // non-empty, never behind an include gate. Empty (and omitted
2288 // from the wire) on a healthy workspace.
2289 summary.quarantined = self
2290 .quarantined
2291 .iter()
2292 .map(|q| crate::ops::QuarantinedMemReport {
2293 mem: q.mount.mem.clone(),
2294 reason_code: q.reason_code.clone(),
2295 reason_message: q.reason_message.clone(),
2296 })
2297 .collect();
2298 // Per-file load failures ride the report unconditionally, like
2299 // the quarantine roster — each entry's message names the remedy
2300 // (the merge-conflict refusal names `memstead conflicts
2301 // resolve`), and a remedy only a library accessor carries is a
2302 // capability nobody finds at the moment it is needed.
2303 summary.load_errors = self
2304 .load_errors
2305 .iter()
2306 .map(|(path, msg)| crate::ops::LoadErrorReport {
2307 file: path.display().to_string(),
2308 error: msg.clone(),
2309 })
2310 .collect();
2311 summary.boot_diagnosis = self
2312 .boot_diagnosis
2313 .as_ref()
2314 .map(|(code, message)| serde_json::json!({ "code": code, "message": message }));
2315 // Surface OUTER_REPO_NOT_IGNORING_MEM_REPO when the
2316 // workspace is embedded inside a git repository whose
2317 // .gitignore does not list `mem-repo/`. Skipped when
2318 // workspace_root is unset (engine built ad-hoc from a mount
2319 // list).
2320 if let Some(root) = self.workspace_root.as_deref()
2321 && let Some(outer) = crate::workspace_root::find_enclosing_git_repo(root)
2322 && !crate::workspace_root::outer_repo_ignores_mem_repo(&outer, root)
2323 {
2324 summary
2325 .warnings
2326 .push(WarningHint::OuterRepoNotIgnoringMemRepo {
2327 outer_repo_root: outer.display().to_string(),
2328 workspace_root: root.display().to_string(),
2329 });
2330 }
2331 // Authoring-drift axis: for every pinned schema whose sealed
2332 // copy carries an install-provenance stamp, report a MISSING
2333 // authoring package (stamped path gone) or a DIVERGED one
2334 // (present but no longer parsed-equivalent to the seal).
2335 // Unstamped schemas — sealed pre-stamp, built-ins, archive
2336 // installs — produce no finding. Read-only on both copies.
2337 summary.warnings.extend(self.authoring_drift_findings());
2338 // Rot axis for the pins the drift axis skips: an UNSTAMPED
2339 // sealed package whose content no longer passes current-
2340 // language authoring validation gets its own low-tier hint —
2341 // the holding runs fine on the tolerant seal, but the package
2342 // (and the unlocatable authoring source it came from) is no
2343 // longer installable, and nothing else would say so before the
2344 // next install attempt. A parsing unstamped package stays
2345 // silent; stamped pins are the drift axis's business.
2346 summary.warnings.extend(self.unstamped_rot_findings());
2347 // Under a mem scope, mem-attributable warnings narrow to the
2348 // scoped mem; workspace- and request-scoped warnings return
2349 // `None` from `source_mem()` and stay visible regardless.
2350 // Mirrors the full flavour's compose filter.
2351 if let Some(v) = mem {
2352 summary
2353 .warnings
2354 .retain(|w| w.source_mem().is_none_or(|wv| wv == v));
2355 }
2356 summary
2357 }
2358
2359 /// Compute the authoring-drift findings for every stamped pinned
2360 /// schema. See the call site in [`Self::health`] for the axis
2361 /// contract; returns an empty list when no workspace root is set
2362 /// (ad-hoc mount-list engines have no authoring tree to check).
2363 fn authoring_drift_findings(&self) -> Vec<WarningHint> {
2364 let Some(root) = self.workspace_root.as_deref() else {
2365 return Vec::new();
2366 };
2367 // Group pinning mems by (name, version) — BTreeMap for a
2368 // deterministic finding order.
2369 let mut pins: std::collections::BTreeMap<(String, String), Vec<String>> =
2370 std::collections::BTreeMap::new();
2371 for (mem, schema) in &self.schemas {
2372 let (name, version) = schema.id();
2373 pins.entry((name.to_string(), version.to_string()))
2374 .or_default()
2375 .push(mem.clone());
2376 }
2377 let mut out = Vec::new();
2378 for ((name, version), mut mems) in pins {
2379 mems.sort();
2380 let Some(stamped_path) = self.read_install_provenance(root, &name, &version) else {
2381 continue;
2382 };
2383 let schema_ref = format!("{name}@{version}");
2384 // A workspace-relative stamp (the portable form the installer
2385 // writes for in-workspace authoring dirs) resolves against
2386 // THIS workspace root, so the axis checks real drift on every
2387 // clone instead of reporting another machine's absolute path
2388 // as missing. Absolute stamps stay machine-pinned as-is.
2389 let stamped = std::path::Path::new(&stamped_path);
2390 let resolved: std::path::PathBuf = if stamped.is_absolute() {
2391 stamped.to_path_buf()
2392 } else {
2393 root.join(stamped)
2394 };
2395 let authoring = resolved.as_path();
2396 if !authoring.is_dir() {
2397 out.push(WarningHint::SchemaAuthoringSourceMissing {
2398 schema_ref,
2399 stamped_path,
2400 mems,
2401 });
2402 continue;
2403 }
2404 let sealed = self
2405 .schemas
2406 .get(&mems[0])
2407 .expect("mems collected from self.schemas keys")
2408 .clone();
2409 match memstead_schema::load_schema_from_dir(authoring) {
2410 Err(e) => out.push(WarningHint::SchemaAuthoringSourceDiverged {
2411 schema_ref,
2412 stamped_path,
2413 mems,
2414 detail: format!("the authoring package no longer loads: {e}"),
2415 }),
2416 Ok(authored) => {
2417 if schema_parsed_fingerprint(&authored) != schema_parsed_fingerprint(&sealed) {
2418 out.push(WarningHint::SchemaAuthoringSourceDiverged {
2419 schema_ref,
2420 stamped_path,
2421 mems,
2422 detail: "the parsed authoring package differs from the sealed copy \
2423 the engine runs on"
2424 .to_string(),
2425 });
2426 }
2427 }
2428 }
2429 }
2430 out
2431 }
2432
2433 /// Compute the rot findings for every UNSTAMPED pinned schema: read
2434 /// the sealed package's content back (folder seal directory, or the
2435 /// `__MEMSTEAD:schemas/` ref via the ops bundle) and run the
2436 /// authoring-tier check over it. A pin with a stamp is skipped (the
2437 /// divergence axis owns it); a pin with no readable sealed package
2438 /// — built-ins resolving from the embedded catalogue — is skipped
2439 /// too (nothing on disk can rot). See the call site in
2440 /// [`Self::health`] for the axis contract.
2441 fn unstamped_rot_findings(&self) -> Vec<WarningHint> {
2442 let Some(root) = self.workspace_root.as_deref() else {
2443 return Vec::new();
2444 };
2445 let mut pins: std::collections::BTreeMap<(String, String), Vec<String>> =
2446 std::collections::BTreeMap::new();
2447 for (mem, schema) in &self.schemas {
2448 let (name, version) = schema.id();
2449 pins.entry((name.to_string(), version.to_string()))
2450 .or_default()
2451 .push(mem.clone());
2452 }
2453 let mut out = Vec::new();
2454 for ((name, version), mut mems) in pins {
2455 mems.sort();
2456 if self
2457 .read_install_provenance(root, &name, &version)
2458 .is_some()
2459 {
2460 continue; // stamped — the divergence axis checks it
2461 }
2462 let schema_ref = format!("{name}@{version}");
2463 // Folder seal: the sealed package is a real directory the
2464 // authoring loader can probe directly.
2465 let sealed_dir = root.join(".memstead").join("schemas").join(&schema_ref);
2466 let detail: Option<String> = if sealed_dir.join("schema.yaml").is_file() {
2467 memstead_schema::load_schema_from_dir(&sealed_dir)
2468 .err()
2469 .map(|e| e.to_string())
2470 } else if let Some((manifest, types)) = self.read_sealed_package_yamls(&name, &version)
2471 {
2472 memstead_schema::loader::check_package_reauthorable(&manifest, &types)
2473 .err()
2474 .map(|e| e.to_string())
2475 } else {
2476 None // no sealed copy anywhere — embedded builtin
2477 };
2478 if let Some(detail) = detail {
2479 out.push(WarningHint::SchemaUnstampedSourceRot {
2480 schema_ref,
2481 mems,
2482 detail,
2483 });
2484 }
2485 }
2486 out
2487 }
2488
2489 /// Read a sealed package's `schema.yaml` + `types/*.yaml` back from
2490 /// the `__MEMSTEAD:schemas/` ref, reconstructing the type-file names
2491 /// from the pinned parsed schema's type roster (seal-time authoring
2492 /// enforces stem == declared type name). `None` when the ops bundle
2493 /// or the package is absent — the embedded-builtin state.
2494 fn read_sealed_package_yamls(
2495 &self,
2496 name: &str,
2497 version: &str,
2498 ) -> Option<(String, Vec<(String, String)>)> {
2499 let ops = self.git_branch_ops()?;
2500 let root = self.workspace_root.as_deref()?;
2501 let gitdir = self
2502 .mounts
2503 .iter()
2504 .find_map(|m| match &m.mount.storage {
2505 crate::workspace::MountStorage::GitBranch { gitdir, .. } => Some(gitdir.clone()),
2506 _ => None,
2507 })
2508 .or_else(|| {
2509 let g = root.join("mem-repo").join(".git");
2510 g.is_dir().then_some(g)
2511 })?;
2512 let read = |rel: &str| -> Option<String> {
2513 (ops.read_schema_file)(&gitdir, name, version, rel)
2514 .ok()
2515 .flatten()
2516 .and_then(|bytes| String::from_utf8(bytes).ok())
2517 };
2518 let manifest = read("schema.yaml")?;
2519 let schema = self
2520 .schemas
2521 .values()
2522 .find(|s| {
2523 let (n, v) = s.id();
2524 n == name && v.to_string() == version
2525 })?
2526 .clone();
2527 let mut type_names: Vec<String> = schema.types.keys().cloned().collect();
2528 type_names.sort();
2529 let mut types = Vec::new();
2530 for t in type_names {
2531 if let Some(body) = read(&format!("types/{t}.yaml")) {
2532 types.push((t, body));
2533 }
2534 }
2535 Some((manifest, types))
2536 }
2537
2538 /// Read the install-provenance stamp for a sealed schema package,
2539 /// checking the folder location first
2540 /// (`.memstead/schemas/<name>@<version>/`) and falling back to the
2541 /// `__MEMSTEAD:schemas/` ref via the git-branch ops bundle when
2542 /// wired. `None` when no stamp exists anywhere — the normal state
2543 /// for pre-stamp seals, built-ins, and archive installs.
2544 fn read_install_provenance(&self, root: &Path, name: &str, version: &str) -> Option<String> {
2545 let folder_stamp = root
2546 .join(".memstead")
2547 .join("schemas")
2548 .join(format!("{name}@{version}"))
2549 .join(memstead_schema::INSTALL_PROVENANCE_FILE);
2550 let bytes = if folder_stamp.is_file() {
2551 std::fs::read(&folder_stamp).ok()
2552 } else {
2553 let ops = self.git_branch_ops()?;
2554 let gitdir = self
2555 .mounts
2556 .iter()
2557 .find_map(|m| match &m.mount.storage {
2558 crate::workspace::MountStorage::GitBranch { gitdir, .. } => {
2559 Some(gitdir.clone())
2560 }
2561 _ => None,
2562 })
2563 .or_else(|| {
2564 let g = root.join("mem-repo").join(".git");
2565 g.is_dir().then_some(g)
2566 })?;
2567 (ops.read_schema_file)(
2568 &gitdir,
2569 name,
2570 version,
2571 memstead_schema::INSTALL_PROVENANCE_FILE,
2572 )
2573 .ok()
2574 .flatten()
2575 }?;
2576 let v: serde_json::Value = serde_json::from_slice(&bytes).ok()?;
2577 v.get("authoring_path")?.as_str().map(String::from)
2578 }
2579
2580 /// Engine-wide [`crate::ops::Status`] across every mount — the graph
2581 /// counts behind `memstead status` (renamed from `stats` with the
2582 /// command, D11; fields unchanged).
2583 pub fn status(&self) -> crate::ops::Status {
2584 let mut types_in_use: Vec<String> = self
2585 .store
2586 .all_entities()
2587 .filter(|e| !e.stub && !e.entity_type.is_empty())
2588 .map(|e| e.entity_type.clone())
2589 .collect();
2590 types_in_use.sort();
2591 types_in_use.dedup();
2592
2593 let mut edge_types: std::collections::BTreeMap<String, usize> =
2594 std::collections::BTreeMap::new();
2595 for id in self.store.all_ids() {
2596 for edge in self.store.outgoing(id) {
2597 *edge_types.entry(edge.rel_type.clone()).or_insert(0) += 1;
2598 }
2599 }
2600
2601 crate::ops::Status {
2602 entity_count: self.store.all_entities().filter(|e| !e.stub).count(),
2603 edge_count: self.store.edge_count(),
2604 edge_types,
2605 community_count: self.communities().count,
2606 mem_count: self.mounts.len(),
2607 types_in_use,
2608 }
2609 }
2610
2611 /// Build a [`ContextResult`] for `id`: the community cluster id
2612 /// (or `None` when the entity is a stub or not present), plus the
2613 /// outgoing + incoming neighbour lists.
2614 pub fn context(&self, id: &EntityId) -> Option<ContextResult> {
2615 let entity = self.store.get(id)?;
2616 let community = self
2617 .communities()
2618 .entity_cluster_map
2619 .get(id.as_ref())
2620 .cloned();
2621 let mut neighbors = Vec::new();
2622 for edge in self.store.outgoing(id) {
2623 if let Some(target) = self.store.get(&edge.target) {
2624 neighbors.push(NeighborInfo {
2625 id: target.id.clone(),
2626 title: target.title.clone(),
2627 relationship: edge.rel_type.clone(),
2628 direction: Direction::Outgoing,
2629 });
2630 }
2631 }
2632 for edge in self.store.incoming(id) {
2633 if let Some(source) = self.store.get(&edge.from) {
2634 neighbors.push(NeighborInfo {
2635 id: source.id.clone(),
2636 title: source.title.clone(),
2637 relationship: edge.rel_type.clone(),
2638 direction: Direction::Incoming,
2639 });
2640 }
2641 }
2642 Some(ContextResult {
2643 entity_id: entity.id.clone(),
2644 community,
2645 neighbors,
2646 })
2647 }
2648
2649 /// Lazily-built per-mem search index map. The map carries one
2650 /// entry per writable mem. Build cost scales with entity count;
2651 /// expect hundreds-of-ms for thousand-entity workspaces. Not
2652 /// available on `wasm32` targets — search lives behind the bridge
2653 /// (see [`Self::search`] for the typed refuse).
2654 #[cfg(not(target_arch = "wasm32"))]
2655 pub fn search_indexes(&self) -> &HashMap<String, MemIndex> {
2656 if let Some((memo_key, _)) = self.search_indexes_memo.get() {
2657 debug_assert_eq!(
2658 *memo_key,
2659 self.derived_key(),
2660 "search memo key lags the engine — a mutation path missed invalidate_search_indexes"
2661 );
2662 }
2663 &self
2664 .search_indexes_memo
2665 .get_or_init(|| (self.derived_key(), build_all(&self.store, &self.schemas)))
2666 .1
2667 }
2668
2669 /// Drop the cached per-mem search index map. No-op on `wasm32`
2670 /// where no index exists; the method stays present so mutation
2671 /// hooks can call it unconditionally.
2672 /// Incrementally maintain the search-index memo for a known
2673 /// touched-id set (flywheel W8/01, criterion 1): replace or remove
2674 /// exactly the touched documents in place and advance the memo's
2675 /// key, instead of dropping the whole map. Semantics:
2676 ///
2677 /// - Memo empty → nothing to maintain; the next read builds fresh.
2678 /// - Memo current (key matches) → no-op (rollback already landed).
2679 /// - Schemas epoch moved → the index FIELD SET may have changed:
2680 /// the named, scoped fallback — drop the memo for a full
2681 /// rebuild. Never a silent widening: this is the one case the
2682 /// plan names (schema-shape change).
2683 /// - Otherwise: per touched id, a real non-stub entity in the
2684 /// store is re-indexed (delete-then-add on the id term), and an
2685 /// absent or stub entry is removed — stubs stay excluded exactly
2686 /// as the bulk build excludes them. Touched mems' writers
2687 /// commit; any tantivy error falls back to dropping the memo
2688 /// (warn-logged), never to serving a stale index.
2689 #[cfg(not(target_arch = "wasm32"))]
2690 pub(crate) fn maintain_search_indexes(&mut self, touched: &[crate::EntityId]) {
2691 let current = super::DerivedKey {
2692 store_generation: self.store.generation(),
2693 schemas_epoch: self.schemas_epoch,
2694 };
2695 let Some((memo_key, indexes)) = self.search_indexes_memo.get_mut() else {
2696 return;
2697 };
2698 if *memo_key == current {
2699 return;
2700 }
2701 if memo_key.schemas_epoch != current.schemas_epoch {
2702 self.search_indexes_memo = OnceCell::new();
2703 return;
2704 }
2705 let mut touched_mems: std::collections::HashSet<&str> = std::collections::HashSet::new();
2706 for id in touched {
2707 let Some(idx) = indexes.get_mut(id.mem()) else {
2708 continue;
2709 };
2710 let result = match self.store.get(id) {
2711 Some(entity) if !entity.stub => idx.index_entity(entity),
2712 _ => idx.remove_entity(id),
2713 };
2714 if let Err(e) = result {
2715 tracing::warn!(
2716 id = id.as_ref(),
2717 error = %e,
2718 "incremental index maintenance failed; dropping the memo for a full rebuild"
2719 );
2720 self.search_indexes_memo = OnceCell::new();
2721 return;
2722 }
2723 touched_mems.insert(id.mem());
2724 }
2725 for (mem, idx) in indexes.iter_mut() {
2726 if !touched_mems.contains(mem.as_str()) {
2727 continue;
2728 }
2729 if let Err(e) = idx.commit() {
2730 tracing::warn!(
2731 mem = mem.as_str(),
2732 error = %e,
2733 "incremental index commit failed; dropping the memo for a full rebuild"
2734 );
2735 self.search_indexes_memo = OnceCell::new();
2736 return;
2737 }
2738 }
2739 *memo_key = current;
2740 }
2741
2742 /// No-op shim on `wasm32`, mirroring `invalidate_search_indexes`
2743 /// so mutation paths call it unconditionally.
2744 #[cfg(target_arch = "wasm32")]
2745 pub(crate) fn maintain_search_indexes(&mut self, _touched: &[crate::EntityId]) {}
2746
2747 /// Unconditionally drop the search-index memo, regardless of its
2748 /// generation. The forced variant exists for embedders that want
2749 /// to release the index's memory in a long-lived process (or force
2750 /// a from-scratch rebuild for verification); the generation-checked
2751 /// [`Self::invalidate_search_indexes`] stays the mutation-path
2752 /// hook.
2753 pub fn drop_search_indexes(&mut self) {
2754 #[cfg(not(target_arch = "wasm32"))]
2755 {
2756 self.search_indexes_memo = OnceCell::new();
2757 }
2758 }
2759
2760 pub fn invalidate_search_indexes(&mut self) {
2761 #[cfg(not(target_arch = "wasm32"))]
2762 {
2763 // Same generation check as `invalidate_communities`: keep
2764 // the memo when the store still sits at its generation
2765 // (the batch-rollback case), clear otherwise.
2766 if let Some((memo_key, _)) = self.search_indexes_memo.get()
2767 && *memo_key == self.derived_key()
2768 {
2769 return;
2770 }
2771 self.search_indexes_memo = OnceCell::new();
2772 }
2773 }
2774
2775 /// Filter the in-memory store by metadata only (no text match).
2776 #[cfg(not(target_arch = "wasm32"))]
2777 pub fn list(&self, scope: &SearchScope) -> crate::ops::ListResult {
2778 let fallback = engine_fallback_type();
2779 crate::ops::search::list(&self.store, scope, fallback.as_ref(), &self.schemas)
2780 }
2781
2782 /// Run a search against the lazily-built index map. Returns
2783 /// [`EngineError::SearchUnavailable`] on `wasm32` targets — browser
2784 /// consumers route search to the bridge; the local
2785 /// engine never builds a tantivy index in WASM. Native targets get
2786 /// the same shape as before, wrapped in `Ok`.
2787 pub fn search(&self, scope: &SearchScope) -> Result<SearchResult, EngineError> {
2788 // A mem filter naming a quarantined OR nonexistent mem refuses
2789 // typed `UNKNOWN_MEM`, matching every other mem-naming surface. A
2790 // success with 0 hits (the old nonexistent-mem behaviour, with a
2791 // missing-index warning) is indistinguishable from a true empty
2792 // result — the one thing a typed surface must never be.
2793 if let Some(mem) = scope.mem.as_deref()
2794 && (self.quarantine_reason(mem).is_some() || !self.mem_router.is_visible(mem))
2795 {
2796 return Err(self.unknown_mem_error(mem));
2797 }
2798 #[cfg(target_arch = "wasm32")]
2799 {
2800 let _ = scope;
2801 return Err(EngineError::SearchUnavailable);
2802 }
2803 #[cfg(not(target_arch = "wasm32"))]
2804 {
2805 let fallback = engine_fallback_type();
2806 Ok(crate::ops::search::search(
2807 &self.store,
2808 scope,
2809 fallback.as_ref(),
2810 self.search_indexes(),
2811 &self.schemas,
2812 ))
2813 }
2814 }
2815
2816 /// All mem-relative entity paths under `mem`. Delegates to
2817 /// the backend's `list_entities`. Order is backend-defined.
2818 pub fn list_entities(&self, mem: &str) -> Result<Vec<PathBuf>, EngineError> {
2819 let m = self.find_mount(mem)?;
2820 m.backend.list_entities().map_err(EngineError::Backend)
2821 }
2822
2823 /// Raw bytes for a single entity (`Ok(None)` if absent).
2824 pub fn read_entity(&self, mem: &str, rel_path: &Path) -> Result<Option<Vec<u8>>, EngineError> {
2825 let m = self.find_mount(mem)?;
2826 m.backend
2827 .read_entity(rel_path)
2828 .map_err(EngineError::Backend)
2829 }
2830
2831 /// Provenance entries for `mem` since `cursor`. Cursor shape is
2832 /// backend-specific (RFC-3339 timestamp for folder, commit SHA for
2833 /// git-branch); `None` means "from the beginning".
2834 pub fn read_provenance(
2835 &self,
2836 mem: &str,
2837 cursor: Option<&str>,
2838 ) -> Result<Vec<Provenance>, EngineError> {
2839 let m = self.find_mount(mem)?;
2840 m.backend
2841 .read_provenance(cursor)
2842 .map_err(EngineError::Backend)
2843 }
2844
2845 /// Capability declared on the mount for `mem`. Surfaced for
2846 /// callers that need to gate before dispatching a write — the
2847 /// engine itself does not yet enforce capability (mutation paths
2848 /// land in a later session).
2849 pub fn capability(&self, mem: &str) -> Result<crate::workspace::MountCapability, EngineError> {
2850 let m = self.find_mount(mem)?;
2851 Ok(m.mount.capability)
2852 }
2853
2854 /// Returns `true` when `from`'s source mem is mounted with
2855 /// [`crate::workspace::MountCapability::ReadOnly`]. Returns
2856 /// `false` for Write-Mems and for mems whose mount is absent
2857 /// from the router (no mount → no ReadOnly assertion can be
2858 /// made; the absence is treated as not-ReadOnly so consumers
2859 /// don't trip on transient lookup misses).
2860 ///
2861 /// Plan body §"Single edge source in the store" specifies this
2862 /// helper as the derived-on-demand alternative to adding a new
2863 /// field on [`crate::store::Edge`]. Strict-invariant validators
2864 /// and surfaces that want to highlight cross-mount references
2865 /// call this rather than pattern-matching on a per-edge marker.
2866 /// The information is fully derivable from the current mount
2867 /// roster, so no new state needs to live on the edge itself.
2868 pub fn edge_is_from_readonly(&self, from: &EntityId) -> bool {
2869 match self.capability(from.mem()) {
2870 Ok(crate::workspace::MountCapability::ReadOnly) => true,
2871 Ok(crate::workspace::MountCapability::Write) | Err(_) => false,
2872 }
2873 }
2874
2875 /// Whether a cross-mem edge from `from_mem` to `to_mem` is
2876 /// permitted under the current [`crate::WorkspaceSettings`]
2877 /// cross-mem link policy.
2878 ///
2879 /// Resolution rules (matches full's `mem_router` semantics):
2880 /// 1. Same-mem edge (`from_mem == to_mem`) → always
2881 /// allowed; the policy gates *cross*-mem edges only.
2882 /// 2. Explicit `cross_mem_links[from_mem]`:
2883 /// - `"*"` (wildcard) → allowed regardless of target.
2884 /// - `["a", ...]` (allowlist) → allowed iff `to_mem` is in
2885 /// the list.
2886 /// 3. Per-create-rule `default_cross_links` synthesis — if
2887 /// rule (1) didn't grant permission and `from_mem` matches
2888 /// a `[[mem_management.create]]` rule whose
2889 /// `default_cross_links` is set, the synthesised value
2890 /// contributes:
2891 /// - `"*"` → allowed regardless of target.
2892 /// - `["a", ...]` → allowed iff `to_mem` is in the list.
2893 /// 4. Otherwise → denied (default-deny posture).
2894 ///
2895 /// The synthesis layer compiles a [`crate::mem_management::CreateRuleSet`]
2896 /// lazily on first call and caches it; [`Self::set_settings`]
2897 /// invalidates the cache. Compilation failure (malformed glob
2898 /// in a rule) logs a warning and the synthesis layer is silently
2899 /// skipped — the resolver still returns `true` from explicit
2900 /// policy alone, so a half-broken config doesn't lock out edges
2901 /// the operator did intend to allow. Operators who want hard
2902 /// validation pre-compile via
2903 /// [`crate::mem_management::CreateRuleSet::new`] before
2904 /// calling [`Self::set_settings`].
2905 ///
2906 /// The MCP `memstead_relate` handler's cross-mem gate consumes
2907 /// this method directly.
2908 pub fn cross_mem_link_allowed(&self, from_mem: &str, to_mem: &str) -> bool {
2909 use memstead_schema::workspace_config::CrossLinkValue;
2910 if from_mem == to_mem {
2911 return true;
2912 }
2913
2914 // Step 1: explicit cross_mem_links policy.
2915 if let Some(value) = self.settings.cross_mem_links.get(from_mem) {
2916 match value {
2917 CrossLinkValue::Wildcard => return true,
2918 CrossLinkValue::List(targets) => {
2919 if targets.iter().any(|t| t == to_mem) {
2920 return true;
2921 }
2922 // Fall through to synthesis check — a List that
2923 // doesn't include the target may still allow it
2924 // via per-rule default_cross_links union.
2925 }
2926 }
2927 }
2928
2929 // Step 2: per-create-rule default_cross_links synthesis.
2930 let rule_set = self.create_rule_set_memo.get_or_init(|| {
2931 crate::mem_management::CreateRuleSet::new(
2932 self.settings.mem_create_rules.clone(),
2933 )
2934 .unwrap_or_else(|err| {
2935 tracing::warn!(
2936 error = %err,
2937 "cross_mem_link_allowed: failed to compile mem_create_rules — synthesis disabled (resolver falls back to explicit-policy-only)"
2938 );
2939 crate::mem_management::CreateRuleSet::default()
2940 })
2941 });
2942
2943 // Compose the same `<mem_path>/<name>` candidate the create-rule
2944 // composer matched against. The rule globs are keyed on the composed
2945 // lifecycle path (e.g. `memstead/project`, compiled with
2946 // `literal_separator`), not the bare leaf name — matching
2947 // `from_mem` alone silently misses, so synthesis denied a link
2948 // that `memstead_overview` rendered as rule-granted (the
2949 // leaf-vs-composed-path divergence). Flat-layout mems (no
2950 // hierarchical path) keep the bare leaf, matching their bare rule.
2951 let candidate = match self.mount(from_mem).and_then(|m| m.mem_path()) {
2952 Some(path) => format!("{path}/{from_mem}"),
2953 None => from_mem.to_string(),
2954 };
2955 if let Some(matched) = rule_set.first_match(std::path::Path::new(&candidate))
2956 && let Some(synth) = matched.default_cross_links.as_ref()
2957 {
2958 return match synth {
2959 CrossLinkValue::Wildcard => true,
2960 CrossLinkValue::List(targets) => targets.iter().any(|t| t == to_mem),
2961 };
2962 }
2963
2964 false
2965 }
2966
2967 pub(super) fn find_mount(&self, mem: &str) -> Result<&MountedBackend, EngineError> {
2968 self.mounts
2969 .iter()
2970 .find(|m| m.mount.mem == mem)
2971 .ok_or_else(|| self.unknown_mem_error(mem))
2972 }
2973}
2974
2975/// The base path of an anchor artifact ref — the locator suffixes a
2976/// medium may append (`@<commit>`, `#<span>`) stripped so the reverse
2977/// lookup compares paths, not versioned/located refs.
2978pub(crate) fn anchor_base_path(artifact: &str) -> &str {
2979 let cut = artifact.find(['@', '#']).unwrap_or(artifact.len());
2980 &artifact[..cut]
2981}
2982
2983/// One mem's standalone anchor-verification report — the counts plus
2984/// the per-anchor rows, in sidecar order.
2985#[derive(Debug, Clone, Default, serde::Serialize)]
2986pub struct MemAnchorVerification {
2987 pub mem: String,
2988 /// The anchors sidecar could not be read: the parse or IO reason. When
2989 /// set, no row was counted — every count below is zero because nothing
2990 /// was measured, not because the mem is clean — `fully_adjudicated`
2991 /// reads false and the population reads unknown. One condition, one
2992 /// code (`ANCHORS_SIDECAR_UNREADABLE`), rendered by every surface from
2993 /// this field; no surface parses the sidecar on its own.
2994 #[serde(skip_serializing_if = "Option::is_none")]
2995 pub sidecar_error: Option<String>,
2996 /// Source present, hash matches (or a non-hash class whose source
2997 /// exists). The wire name is the `AnchorState` wire form, `resolves`;
2998 /// `resolved` was retired on 2026-09-02 (one name per state).
2999 pub resolves: usize,
3000 /// Source present, hash differs, stability `stable` — real drift.
3001 pub drifted: usize,
3002 /// Hash differs under `unstable` stability, or a hash is missing on
3003 /// either side — flagged for re-examination, never called drift.
3004 pub recheck: usize,
3005 /// Source absent: a MEASURED failure. The artifact the anchor names is
3006 /// not there.
3007 pub unresolvable: usize,
3008 /// The anchor could not be observed at all this pass, so nothing about it
3009 /// was measured (consistency-sweep 03/05, criterion 2). Its own count,
3010 /// because `unresolvable` used to swallow it: a reader on the surface you
3011 /// reach WITHOUT a binding could not tell a measured failure from an
3012 /// absent measurement, which is the one distinction that surface exists to
3013 /// make.
3014 pub unobserved: usize,
3015 /// Rows whose ENTITY is gone (consistency-sweep 03/02). Its own class,
3016 /// counted apart from the states above: those describe the artifact end,
3017 /// and a vanished entity says nothing about the source. Folding it into
3018 /// `unresolvable` would name the wrong repair.
3019 pub dangling: usize,
3020 /// Why the entity end could not be reconciled this pass, when it could
3021 /// not. `dangling: 0` means "none found" only when this is `None`.
3022 pub unreconciled: Option<String>,
3023 pub anchors: Vec<VerifiedAnchor>,
3024 /// Supplied observations whose artifact matched no `url` anchor of the
3025 /// mem — reported, never silently dropped, never applied to anything.
3026 #[serde(default, skip_serializing_if = "Vec::is_empty")]
3027 pub unmatched_observations: Vec<String>,
3028 /// The `last_observed` records this verification produced from supplied
3029 /// observations, for the caller to commit through
3030 /// [`Engine::record_anchor_observations`]. Engine-internal.
3031 #[serde(skip)]
3032 pub recordable_observations: Vec<RecordedObservation>,
3033}
3034
3035/// One observation to record onto a sidecar row (the `last_observed`
3036/// field), addressed by the `(entity, artifact)` pair.
3037#[derive(Debug, Clone, PartialEq, Eq)]
3038pub struct RecordedObservation {
3039 pub entity: String,
3040 pub artifact: String,
3041 pub observation: crate::anchor::AnchorObservation,
3042}
3043
3044/// Observer-supplied observations keyed by artifact — the input of
3045/// [`Engine::verify_mem_anchors_with`].
3046pub type SuppliedObservations =
3047 std::collections::BTreeMap<String, crate::anchor::SuppliedObservation>;
3048
3049/// What one observation of an anchor yielded: the resolved state, the
3050/// hash it saw (when any), and — for an observation that was not made
3051/// live this pass — when it was made.
3052#[derive(Debug, Clone, PartialEq, Eq)]
3053struct Observed {
3054 state: crate::anchor::AnchorState,
3055 hash: Option<String>,
3056 at: Option<String>,
3057}
3058
3059impl Observed {
3060 /// A live observation made this pass (path and entity grains).
3061 fn live((state, hash): (crate::anchor::AnchorState, Option<String>)) -> Self {
3062 Observed {
3063 state,
3064 hash,
3065 at: None,
3066 }
3067 }
3068}
3069
3070impl MemAnchorVerification {
3071 /// The population statement that must accompany this report's figures
3072 /// (consistency-sweep 03/05, criteria 1 and 3): what the figures were
3073 /// computed over, and how much of it the pass could not adjudicate.
3074 ///
3075 /// A resolution figure alone is read as health. Every W3 finding made that
3076 /// figure mean less than a reader assumes, and none of them made it wrong
3077 /// in a way anyone could see. Rendering the figure and its population as
3078 /// ONE unit is what stops the next such finding being invisible: a surface
3079 /// cannot show the number and omit the caveat, because it gets both from
3080 /// here or neither.
3081 pub fn population_statement(&self) -> String {
3082 // `recheck` belongs on ONE side of this sentence. A first version put
3083 // it in both: counted as adjudicated and then reported as not, so the
3084 // same rows appeared twice and the two numbers could not be reconciled
3085 // by a reader. A recheck row is a row whose drift could NOT be
3086 // asserted, which is the definition of unadjudicated.
3087 if let Some(why) = &self.sidecar_error {
3088 return format!(
3089 "population unknown: the anchors sidecar could not be read ({why}); no row was \
3090 counted, and zero counts here are not a clean mem"
3091 );
3092 }
3093 let adjudicated = self.resolves + self.drifted + self.unresolvable;
3094 let unadjudicated = self.recheck + self.unobserved;
3095 let mut s = format!(
3096 "over {} counted row(s): {adjudicated} adjudicated, {unadjudicated} not (recheck {}, unobserved {})",
3097 adjudicated + unadjudicated,
3098 self.recheck,
3099 self.unobserved
3100 );
3101 if self.dangling > 0 {
3102 s.push_str(&format!(
3103 "; {} row(s) excluded, naming an entity the mem no longer holds",
3104 self.dangling
3105 ));
3106 }
3107 if let Some(why) = &self.unreconciled {
3108 s.push_str(&format!(
3109 "; the entity end was NOT reconciled ({why}), so dangling rows would not have been detected"
3110 ));
3111 }
3112 s
3113 }
3114
3115 /// Whether this axis adjudicated everything it counted. False means the
3116 /// figures above rest on an incomplete measurement, which is not the same
3117 /// as a failed one.
3118 pub fn fully_adjudicated(&self) -> bool {
3119 self.sidecar_error.is_none()
3120 && self.recheck == 0
3121 && self.unobserved == 0
3122 && self.unreconciled.is_none()
3123 }
3124}
3125
3126/// One anchor's verification row.
3127#[derive(Debug, Clone, serde::Serialize)]
3128pub struct VerifiedAnchor {
3129 pub entity_id: String,
3130 pub artifact: String,
3131 pub grain: String,
3132 pub class: String,
3133 /// `resolves` | `drifted` | `recheck` | `unresolvable` (artifact gone) |
3134 /// `unobserved` (not measured this pass) | `dangling` (the entity is
3135 /// gone). The wire vocabulary of this field, which is NOT the engine's
3136 /// `AnchorState` enum: that has four variants describing the artifact
3137 /// end, and the last two here are conditions beside them.
3138 pub state: String,
3139 #[serde(skip_serializing_if = "Option::is_none")]
3140 pub observed_hash: Option<String>,
3141 /// When the observation this row's state rests on was made — present
3142 /// only for a row resolved from a supplied or recorded observation (a
3143 /// `url` row); live observations carry none.
3144 #[serde(skip_serializing_if = "Option::is_none")]
3145 pub observed_at: Option<String>,
3146 /// Whole days between `observed_at` and this run — how long the row has
3147 /// gone unobserved. `Some(0)` for an observation made today.
3148 #[serde(skip_serializing_if = "Option::is_none")]
3149 pub unobserved_for_days: Option<u64>,
3150 /// Whether this row's state came from an observation supplied to this
3151 /// run (as opposed to a live or a previously recorded one).
3152 #[serde(default, skip_serializing_if = "std::ops::Not::not")]
3153 pub observation_supplied: bool,
3154}
3155
3156/// Whether `anchor` references `path`. `tree`-grain anchors match `path`
3157/// itself and anything beneath the tree; every other grain matches by
3158/// exact base-path equality.
3159/// A stored anchor paired with its live resolution state, when observable.
3160/// See [`Engine::entity_anchors_resolved`] for how `state` is produced and
3161/// when it is `None` (unobserved, never fabricated).
3162#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize)]
3163pub struct ResolvedAnchor {
3164 /// The durable anchor record (flattened on the wire so the resolved shape
3165 /// is the stored anchor plus a `state` field).
3166 #[serde(flatten)]
3167 pub anchor: crate::anchor::Anchor,
3168 /// The live resolution state, or `None` when the engine could not observe
3169 /// the source artifact this pass: a `url` grain, no workspace root, an
3170 /// ambiguous or absent path medium, or an `entity` grain whose mem is not
3171 /// mounted. That last case is load-bearing — an unmounted mem is not a mem
3172 /// of deleted entities, so it must read as unobserved rather than
3173 /// `Orphaned`, which prune would act on.
3174 #[serde(skip_serializing_if = "Option::is_none")]
3175 pub state: Option<crate::anchor::AnchorState>,
3176 /// The prepared-content hash the observation computed this pass —
3177 /// present only for a hash-bearing (`anchored` / `derived`) anchor whose
3178 /// artifact could be read: a `file` / `span` anchor resolving to a
3179 /// readable file, or an `entity` anchor whose mem is mounted (hashed over
3180 /// the canonical rendered markdown, so the value means the same thing in
3181 /// both namespaces). The verify pass's backfill leg records it onto a
3182 /// hash-less anchor. Engine-internal
3183 /// observation detail, deliberately not serialized: the wire shape stays
3184 /// the stored anchor plus `state`.
3185 #[serde(skip)]
3186 pub observed_hash: Option<String>,
3187 /// When the observation `state` rests on was made, for a row whose
3188 /// observation is supplied or recorded rather than live (a `url` row).
3189 /// Additive: absent on every live-observed row.
3190 #[serde(skip_serializing_if = "Option::is_none")]
3191 pub observed_at: Option<String>,
3192}
3193
3194/// What an anchor's `source` name resolves to in its mem's bindings: the
3195/// declared pointer (the filesystem root a source-dialect artifact path
3196/// joins onto, decision 26) and the declared preparation (what the
3197/// preparation registry prepares the artifact as before hashing —
3198/// touchpoint A of [`crate::preparation`]).
3199#[derive(Debug, Clone, PartialEq, Eq)]
3200pub(crate) struct AnchorSourceJoin {
3201 /// The source's declared `pointer`.
3202 pub(crate) pointer: String,
3203 /// The source's declared `preparation`, if any.
3204 pub(crate) preparation: Option<String>,
3205 /// The declaring source itself — its scope is what a `tree` anchor's
3206 /// prepared form enumerates under a code-map preparation.
3207 pub(crate) source: crate::pipeline::Source,
3208 /// The binding's `deny_paths`, applied on top of the source scope.
3209 pub(crate) deny_paths: Vec<String>,
3210}
3211
3212/// Observe a single path-namespace anchor against `root` (its medium's
3213/// filesystem root) and resolve its live state plus — for a present
3214/// hash-bearing (`anchored` / `derived`) `file` / `span` anchor — the
3215/// artifact's **prepared-content hash**
3216/// ([`crate::anchor::prepared_content_hash`]). `None` when the anchor's
3217/// grain does not reference a filesystem path.
3218///
3219/// The computed hash is what lets [`crate::anchor::resolve_anchor`]
3220/// adjudicate `drifted` vs `resolves` deterministically against the recorded
3221/// hash. The prepared form is the registry's rule for the anchor's
3222/// source's preparation ([`crate::preparation::path_prepared_hash`]): a
3223/// `span` anchor hashes its whole containing file (the span locator selects
3224/// within it; the file is the hashed unit), except under a **delivery
3225/// preparation**, where a `<path>#<key>` span names one delivery unit (the
3226/// unit's own text is the hashed unit, and a key the file no longer yields
3227/// is an absent artifact); under a **code-map** preparation a file or span
3228/// hashes the interface digest, and a `tree` hashes the code map of every
3229/// scoped file under it. A `tree` under any other preparation hashes the
3230/// plain per-file prepared-content map of its scoped files, so tree anchors
3231/// adjudicate deterministically like file anchors. A `tree` with no
3232/// resolvable source-join (the enumeration scope is undefined without one),
3233/// a partial enumeration, and a read failure observe no hash — those
3234/// resolve `recheck`, never a fabricated `drifted`. Non-hash classes
3235/// (`authored` / `informed-by`) skip the read entirely, so an anchor-less or
3236/// hash-free mem pays no observation cost.
3237fn observe_path_anchor(
3238 root: &Path,
3239 anchor: &crate::anchor::Anchor,
3240 join: Option<&AnchorSourceJoin>,
3241) -> Option<(crate::anchor::AnchorState, Option<String>)> {
3242 use crate::anchor::AnchorGrain;
3243 match anchor.grain {
3244 AnchorGrain::Span | AnchorGrain::File | AnchorGrain::Tree => {}
3245 AnchorGrain::Url | AnchorGrain::Entity => return None,
3246 }
3247 let source_pointer = join.map(|j| j.pointer.as_str());
3248 let preparation = join.and_then(|j| j.preparation.as_deref());
3249 let base = anchor_base_path(&anchor.artifact);
3250 // Decision 29: the source-join is authoritative — an artifact path is
3251 // source-relative first (joined onto the declaring source's pointer,
3252 // which may deliberately leave the workspace root for out-of-root
3253 // pointers); the workspace-relative form is tried only when the
3254 // source-join does not resolve. The candidate set is the shared
3255 // [`artifact_candidates`] rule, so resolution, the write gate, and the
3256 // population matcher read one artifact the same way.
3257 let path = {
3258 let candidates = artifact_candidates(source_pointer.unwrap_or(""), base);
3259 candidates
3260 .iter()
3261 .map(|c| root.join(c))
3262 .find(|p| p.exists())
3263 .unwrap_or_else(|| root.join(base))
3264 };
3265 if !path.exists() {
3266 return Some((
3267 crate::anchor::resolve_anchor(anchor, &crate::anchor::ArtifactObservation::Absent),
3268 None,
3269 ));
3270 }
3271 let current_hash = if !anchor.class.is_hash_bearing() {
3272 None
3273 } else if matches!(anchor.grain, AnchorGrain::File | AnchorGrain::Span) && path.is_file() {
3274 match std::fs::read(&path).ok().map(|bytes| {
3275 crate::preparation::path_prepared_hash(
3276 preparation,
3277 &anchor.artifact,
3278 anchor.grain,
3279 &bytes,
3280 )
3281 }) {
3282 Some(crate::preparation::PathPrepared::Hash(h)) => Some(h),
3283 Some(crate::preparation::PathPrepared::UnitAbsent) => {
3284 return Some((
3285 crate::anchor::resolve_anchor(
3286 anchor,
3287 &crate::anchor::ArtifactObservation::Absent,
3288 ),
3289 None,
3290 ));
3291 }
3292 Some(crate::preparation::PathPrepared::NoHash) | None => None,
3293 }
3294 } else if anchor.grain == AnchorGrain::Tree
3295 && path.is_dir()
3296 && let Some(join) = join
3297 {
3298 // The tree's prepared form: a digest over every scoped file under
3299 // the tree, by the declaring source's own scope and the binding's
3300 // deny paths — the code map under a code-map preparation, the plain
3301 // per-file prepared-content map otherwise, so a tree anchor
3302 // adjudicates deterministically instead of resting in `recheck`
3303 // forever. The path the anchor names is workspace-relative or
3304 // source-relative; the enumeration is workspace-relative, so compare
3305 // the resolved absolute paths. A PARTIAL enumeration (malformed or
3306 // retired-dialect scope pattern) observes no hash — a digest over a
3307 // set that is not the population would silently change a stored
3308 // tree-anchor hash; no-hash resolves `recheck`, the same posture as
3309 // a failed read.
3310 let enumeration = crate::ingest::cursor::enumerate_facet_files_reported(
3311 &join.source,
3312 &join.deny_paths,
3313 root,
3314 );
3315 if enumeration.is_partial() {
3316 None
3317 } else if preparation == Some(crate::preparation::CODE_MAP) {
3318 let files: Vec<(String, String)> = enumeration
3319 .files
3320 .into_iter()
3321 .filter(|f| root.join(f).starts_with(&path))
3322 .filter_map(|f| {
3323 std::fs::read(root.join(&f))
3324 .ok()
3325 .map(|bytes| (f, String::from_utf8_lossy(&bytes).into_owned()))
3326 })
3327 .collect();
3328 Some(crate::anchor::prepared_content_hash(
3329 crate::preparation::code_map_tree_digest(&files).as_bytes(),
3330 ))
3331 } else {
3332 let files: Vec<(String, Vec<u8>)> = enumeration
3333 .files
3334 .into_iter()
3335 .filter(|f| root.join(f).starts_with(&path))
3336 .filter_map(|f| std::fs::read(root.join(&f)).ok().map(|bytes| (f, bytes)))
3337 .collect();
3338 Some(crate::anchor::prepared_content_hash(
3339 crate::preparation::plain_tree_digest(&files).as_bytes(),
3340 ))
3341 }
3342 } else {
3343 None
3344 };
3345 let observation = crate::anchor::ArtifactObservation::Present {
3346 current_hash: current_hash.clone(),
3347 };
3348 Some((
3349 crate::anchor::resolve_anchor(anchor, &observation),
3350 current_hash,
3351 ))
3352}
3353
3354/// Deterministic fingerprint of a PARSED schema for the
3355/// authoring-drift equivalence check. Compares semantic content, never
3356/// raw bytes: YAML comments (the CLI-injected editor-header lines) and
3357/// whitespace vanish at parse time, and `Schema.types` — a `HashMap`
3358/// with nondeterministic iteration order — is rendered sorted by type
3359/// name so two loads of equivalent packages always fingerprint alike.
3360fn schema_parsed_fingerprint(schema: &memstead_schema::Schema) -> String {
3361 let mut keys: Vec<&String> = schema.types.keys().collect();
3362 keys.sort();
3363 let types: Vec<String> = keys
3364 .iter()
3365 .map(|k| format!("{k}={:?}", schema.types[k.as_str()]))
3366 .collect();
3367 format!(
3368 "{:?}|{}|{}",
3369 schema.manifest,
3370 schema.version,
3371 types.join(";")
3372 )
3373}
3374
3375fn anchor_references_path(anchor: &crate::anchor::Anchor, path: &str) -> bool {
3376 let base = anchor_base_path(&anchor.artifact);
3377 path_references(base, anchor.grain == crate::anchor::AnchorGrain::Tree, path)
3378}
3379
3380/// Whether `base` (a file path, or a tree root when `is_tree`) references
3381/// `path` — exact match, or containment for a tree.
3382fn path_references(base: &str, is_tree: bool, path: &str) -> bool {
3383 if base == path {
3384 return true;
3385 }
3386 if is_tree {
3387 let prefix = base.strip_suffix('/').unwrap_or(base);
3388 return path.starts_with(&format!("{prefix}/"));
3389 }
3390 false
3391}
3392
3393/// Join a source pointer and a source-relative artifact path into the
3394/// pointer-joined (workspace-relative) form — the decision-26 dialect
3395/// bridge. Plain string concatenation with a separator: the pointer is
3396/// workspace-relative (and may climb out via `..`), the artifact is
3397/// source-relative; no canonicalization here, the filesystem resolves it.
3398pub(crate) fn join_pointer(pointer: &str, base: &str) -> String {
3399 let pointer = pointer.trim_end_matches('/');
3400 if pointer.is_empty() || pointer == "." {
3401 base.to_string()
3402 } else {
3403 format!("{pointer}/{base}")
3404 }
3405}
3406
3407/// How a source-relative artifact id resolved ACROSS a binding's primary
3408/// sources — the cross-source counterpart of [`artifact_candidates`], which
3409/// answers only within ONE source.
3410///
3411/// The two questions are genuinely different and only this one can be
3412/// ambiguous. `artifact_candidates` settles which reading wins under a single
3413/// pointer (decision 29: source-join first, workspace-relative fallback). It
3414/// takes one pointer and has no opinion on two sources both carrying
3415/// `docs/a.md`. That case has exactly one home, the exclude fold, and used to
3416/// take the first source that matched — so the source listed first in the
3417/// binding silently won, which is the wrong-target write this type exists to
3418/// prevent.
3419#[derive(Debug, Clone, PartialEq, Eq)]
3420pub(crate) enum CrossSourceArtifact {
3421 /// No source's join lands on a member of the enumerated set.
3422 Unresolved,
3423 /// Exactly one does; this is the canonical id to record.
3424 Unique(String),
3425 /// Several do. Sorted and deduplicated, so the refusal names them in a
3426 /// stable order and the caller never has to pick.
3427 Ambiguous(Vec<String>),
3428}
3429
3430/// Resolve a requested artifact id against a binding's primary sources.
3431///
3432/// `canonical_for` maps one source's base and the requested id to the
3433/// workspace-relative form, returning `None` when that form is not a member
3434/// of the enumerated source set. The caller owns that predicate because the
3435/// membership test differs by surface; what is shared, and what lives here,
3436/// is the rule that several matches are an ambiguity to refuse rather than a
3437/// choice to make silently.
3438pub(crate) fn resolve_across_sources<'a, I, F>(
3439 bases: I,
3440 requested: &str,
3441 canonical_for: F,
3442) -> CrossSourceArtifact
3443where
3444 I: IntoIterator<Item = &'a std::path::PathBuf>,
3445 F: Fn(&std::path::PathBuf, &str) -> Option<String>,
3446{
3447 let mut hits: Vec<String> = bases
3448 .into_iter()
3449 .filter_map(|base| canonical_for(base, requested))
3450 .collect();
3451 hits.sort();
3452 hits.dedup();
3453 match hits.len() {
3454 0 => CrossSourceArtifact::Unresolved,
3455 1 => CrossSourceArtifact::Unique(hits.remove(0)),
3456 _ => CrossSourceArtifact::Ambiguous(hits),
3457 }
3458}
3459
3460/// The candidate workspace-relative forms an anchor artifact could denote
3461/// under its declaring source's pointer, in the ratified priority (bundle
3462/// decision 29): the source-join first, the workspace-relative form as the
3463/// fallback. **The single implementation of that rule** — resolution, the
3464/// write-time gate, and the population scope matcher all construct their
3465/// candidate set here, so one artifact cannot read differently across the
3466/// three (each site then applies its own predicate: existence for
3467/// resolution and the gate, glob membership for the matcher).
3468///
3469/// One lexical clarification rides with the rule: an artifact that climbs
3470/// (`../…`) never joins. An artifact of a source never climbs OUT of that
3471/// source, so such a path is already the workspace-relative form; joining
3472/// anyway fabricates `<ptr>/../…`, which the filesystem then resolves into a
3473/// sibling tree — a false resolution for an existence test and a false
3474/// in-scope for a `**` glob. An artifact already carrying the pointer prefix
3475/// is NOT suppressed: on a self-nested layout both readings exist and the
3476/// decision's priority — source-join wins — settles it deterministically.
3477pub(crate) fn artifact_candidates(pointer: &str, base: &str) -> Vec<String> {
3478 let pointer = pointer.trim_end_matches('/');
3479 if pointer.is_empty() || pointer == "." {
3480 return vec![base.to_string()];
3481 }
3482 if base.starts_with("../") || base == ".." {
3483 return vec![base.to_string()];
3484 }
3485 vec![format!("{pointer}/{base}"), base.to_string()]
3486}
3487
3488#[cfg(test)]
3489mod tests {
3490 use std::path::Path;
3491
3492 use tempfile::TempDir;
3493
3494 use crate::backend::{BackendError, MemBackend};
3495 use crate::engine::test_helpers::*;
3496 use crate::engine::{Engine, EngineError, RelateEntityArgs};
3497 use crate::entity::EntityId;
3498 use crate::ops::{Direction, SearchScope, WarningHint};
3499 use crate::provenance::Provenance;
3500 use crate::storage::{ArchiveBackend, FilesystemMemWriter, MemWriter};
3501
3502 use crate::vcs::CommitContext;
3503 use crate::workspace::{Mount, MountCapability, MountLifecycle, MountStorage};
3504
3505 /// The shared decision-29 candidate rule: source-join first,
3506 /// workspace-relative fallback; a pointer-less (or `.`) source has one
3507 /// reading; a climbing `../…` artifact never joins (the fabricated
3508 /// `<ptr>/../…` would resolve into a sibling tree).
3509 #[test]
3510 fn artifact_candidates_follow_decision_29_priority() {
3511 use super::artifact_candidates;
3512 assert_eq!(artifact_candidates("", "a/b.rs"), vec!["a/b.rs"]);
3513 assert_eq!(artifact_candidates(".", "a/b.rs"), vec!["a/b.rs"]);
3514 assert_eq!(artifact_candidates("./", "a/b.rs"), vec!["a/b.rs"]);
3515 assert_eq!(
3516 artifact_candidates("sub", "a/b.rs"),
3517 vec!["sub/a/b.rs", "a/b.rs"]
3518 );
3519 assert_eq!(
3520 artifact_candidates("sub/", "a/b.rs"),
3521 vec!["sub/a/b.rs", "a/b.rs"]
3522 );
3523 // Self-nesting is settled by priority, not suppression: the artifact
3524 // already carrying the pointer prefix still offers the join first.
3525 assert_eq!(
3526 artifact_candidates("sub", "sub/x.rs"),
3527 vec!["sub/sub/x.rs", "sub/x.rs"]
3528 );
3529 // A climbing artifact is already the workspace-relative form.
3530 assert_eq!(
3531 artifact_candidates("sub", "../dev/x.md"),
3532 vec!["../dev/x.md"]
3533 );
3534 assert_eq!(
3535 artifact_candidates("../dev", "../dev/x.md"),
3536 vec!["../dev/x.md"]
3537 );
3538 assert_eq!(artifact_candidates("sub", ".."), vec![".."]);
3539 }
3540
3541 /// `schema_origin` is the trust-classification authority: a built-in
3542 /// (or workspace-authored) schema is first-party; a schema whose
3543 /// `(name, version)` is in neither catalogue is third-party — the safe
3544 /// default for an origin the engine cannot vouch for.
3545 #[test]
3546 fn schema_origin_classifies_builtin_first_party_and_unknown_third_party() {
3547 use std::sync::Arc;
3548
3549 use crate::render::OriginClass;
3550
3551 let tmp = TempDir::new().unwrap();
3552 let engine = Engine::from_mounts(vec![(
3553 folder_mount("specs", tmp.path().to_path_buf()),
3554 Box::new(FilesystemMemWriter::new(tmp.path().to_path_buf())) as Box<dyn MemBackend>,
3555 )])
3556 .unwrap();
3557
3558 // A built-in schema (the catalogue the engine resolved against).
3559 let builtin = engine.builtin_schemas()[0].clone();
3560 assert_eq!(
3561 engine.schema_origin(&builtin),
3562 OriginClass::FirstParty,
3563 "a built-in schema is first-party"
3564 );
3565
3566 // A schema whose version is in no catalogue — a stand-in for a
3567 // schema that entered from outside the workspace. Same name, a
3568 // version the engine never loaded.
3569 let foreign = Arc::new(memstead_schema::Schema {
3570 manifest: builtin.manifest.clone(),
3571 version: semver::Version::new(99, 0, 0),
3572 types: builtin.types.clone(),
3573 });
3574 assert_eq!(
3575 engine.schema_origin(&foreign),
3576 OriginClass::ThirdParty,
3577 "a schema in neither catalogue classifies third-party (safe default)"
3578 );
3579 }
3580
3581 /// `mem_origin_class` classifies a writable mount first-party (its
3582 /// content is authored in this workspace) and a read-only mount
3583 /// third-party (registry-installed read-mem or adopted foreign
3584 /// folder/clone — quoted, untrusted data). An unknown mem is
3585 /// third-party (the safe default).
3586 #[test]
3587 fn mem_origin_class_writable_first_party_readonly_third_party() {
3588 use crate::render::OriginClass;
3589
3590 let tmp = TempDir::new().unwrap();
3591 // Writable folder mem.
3592 let writable_dir = tmp.path().join("writable");
3593 std::fs::create_dir_all(&writable_dir).unwrap();
3594 let writer = FilesystemMemWriter::new(writable_dir.clone());
3595
3596 // Read-only archive mem.
3597 let body = "---\ntype: spec\n---\n# Ext\n\n## Identity\n\nFrom an archive.\n";
3598 let archive_path = build_archive(tmp.path(), "ext", &[("ext.md", body.as_bytes())]);
3599
3600 let engine = Engine::from_mounts(vec![
3601 (
3602 folder_mount("local", writable_dir),
3603 Box::new(writer) as Box<dyn MemBackend>,
3604 ),
3605 (
3606 archive_mount("external", archive_path.clone()),
3607 Box::new(ArchiveBackend::new(archive_path)) as Box<dyn MemBackend>,
3608 ),
3609 ])
3610 .unwrap();
3611
3612 assert_eq!(
3613 engine.mem_origin_class("local"),
3614 OriginClass::FirstParty,
3615 "a writable mount is first-party"
3616 );
3617 assert_eq!(
3618 engine.mem_origin_class("external"),
3619 OriginClass::ThirdParty,
3620 "a read-only mount is third-party"
3621 );
3622 assert_eq!(
3623 engine.mem_origin_class("no-such-mem"),
3624 OriginClass::ThirdParty,
3625 "an unknown mem is third-party (safe default)"
3626 );
3627 }
3628
3629 /// `declare_mem_origin` lets the embedding deployment vouch for one
3630 /// read-only mount as first-party (the curated hosted read tier),
3631 /// overriding the writability inference for that mem only — sibling
3632 /// read-only mounts keep the safe third-party default.
3633 #[test]
3634 fn declared_origin_overrides_inference_per_mem() {
3635 use crate::render::OriginClass;
3636
3637 let tmp = TempDir::new().unwrap();
3638 let body = "---\ntype: spec\n---\n# Ext\n\n## Identity\n\nFrom an archive.\n";
3639 let vouched_path = build_archive(tmp.path(), "vouched", &[("v.md", body.as_bytes())]);
3640 let other_path = build_archive(tmp.path(), "other", &[("o.md", body.as_bytes())]);
3641
3642 let mut engine = Engine::from_mounts(vec![
3643 (
3644 archive_mount("vouched", vouched_path.clone()),
3645 Box::new(ArchiveBackend::new(vouched_path)) as Box<dyn MemBackend>,
3646 ),
3647 (
3648 archive_mount("other", other_path.clone()),
3649 Box::new(ArchiveBackend::new(other_path)) as Box<dyn MemBackend>,
3650 ),
3651 ])
3652 .unwrap();
3653
3654 engine.declare_mem_origin("vouched", OriginClass::FirstParty);
3655
3656 assert_eq!(
3657 engine.mem_origin_class("vouched"),
3658 OriginClass::FirstParty,
3659 "the deployment's declaration wins over the read-only inference"
3660 );
3661 assert_eq!(
3662 engine.mem_origin_class("other"),
3663 OriginClass::ThirdParty,
3664 "an undeclared sibling mount keeps the safe default"
3665 );
3666 }
3667
3668 /// The adopt-gate: a non-built-in schema is first-party only once a
3669 /// writable mount pins it (the operator authors against it here).
3670 /// Pinned only by a read-only mount — a registry read-mem or an
3671 /// adopted foreign folder/clone — it stays third-party, so
3672 /// `memstead_schema` serves it structural-only.
3673 #[test]
3674 fn schema_origin_third_party_until_pinned_by_a_writable_mount() {
3675 use memstead_schema::SchemaRef;
3676
3677 use crate::render::OriginClass;
3678
3679 let manifest = r#"name: trust-test
3680version: 0.1.0
3681description: adopt-gate test schema
3682when_to_use: tests
3683types:
3684 - doc
3685relationships:
3686 mode: strict
3687 definitions:
3688 - name: _default
3689 description: fallback
3690 default_weight: 1.0
3691community:
3692 resolution: 1.0
3693 seed: 42
3694"#;
3695 let pin = SchemaRef::new("trust-test", semver::Version::new(0, 1, 0));
3696
3697 let mk_engine = |cap: MountCapability| -> Engine {
3698 let tmp = TempDir::new().unwrap();
3699 let schemas_dir = tmp.path().join("schemas");
3700 std::fs::create_dir_all(&schemas_dir).unwrap();
3701 write_schema_files_with_default_type(&schemas_dir, "trust-test", manifest, &["doc"]);
3702 let mem_dir = tmp.path().join("mem");
3703 std::fs::create_dir_all(&mem_dir).unwrap();
3704 let mount = Mount {
3705 mem: "v".to_string(),
3706 schema: Some(pin.clone()),
3707 storage: MountStorage::Folder {
3708 path: mem_dir.clone(),
3709 },
3710 capability: cap,
3711 lifecycle: MountLifecycle::Eager,
3712 cross_linkable: true,
3713 migration_target: None,
3714 };
3715 let backend = Box::new(FilesystemMemWriter::new(mem_dir)) as Box<dyn MemBackend>;
3716 // Keep `tmp` alive for the engine's lifetime by leaking it —
3717 // the test process is short-lived and the folder must outlast
3718 // the closure.
3719 std::mem::forget(tmp);
3720 Engine::from_mounts_with_schemas_dir(vec![(mount, backend)], Some(&schemas_dir))
3721 .unwrap()
3722 };
3723
3724 // Read-only mount: the foreign schema is never adopted → third-party.
3725 let ro = mk_engine(MountCapability::ReadOnly);
3726 let schema = ro.schemas().get("v").expect("schema resolved").clone();
3727 assert_eq!(
3728 ro.schema_origin(&schema),
3729 OriginClass::ThirdParty,
3730 "a non-built-in schema pinned only by a read-only mount is third-party"
3731 );
3732
3733 // Writable mount pinning the same schema: adopted → first-party.
3734 let rw = mk_engine(MountCapability::Write);
3735 let schema = rw.schemas().get("v").expect("schema resolved").clone();
3736 assert_eq!(
3737 rw.schema_origin(&schema),
3738 OriginClass::FirstParty,
3739 "a writable mount pinning the schema adopts it → first-party"
3740 );
3741 }
3742
3743 /// Consumer read path: an installed (archive-backed) mem that ships
3744 /// a `.memstead/provenance.json` payload surfaces per-entity authoring
3745 /// provenance through `archive_provenance_for`. A noted entity carries
3746 /// its rationale; an entity authored without a note is absent from the
3747 /// payload and reads as provenance-absent (no fabricated value); the
3748 /// `history` disposition records that full history is not shipped.
3749 #[test]
3750 fn archive_provenance_surfaces_per_entity_and_reports_absence() {
3751 use memstead_schema::History;
3752
3753 let tmp = TempDir::new().unwrap();
3754 let config = br#"{"format":3,"name":"seed","version":"0.1.0","schema":"default@1.0.0"}"#;
3755 let alpha = b"---\ntype: spec\n---\n# Alpha\n\n## Identity\n\na\n\n## Purpose\n\np\n";
3756 let beta = b"---\ntype: spec\n---\n# Beta\n\n## Identity\n\nb\n\n## Purpose\n\np\n";
3757 // alpha noted; beta deliberately absent from the payload.
3758 let provenance = br#"{"format":1,"history":"summarised","entities":{"alpha":{"rationale":"why alpha exists","kind":"create","timestamp":"2026-06-24T00:00:00Z","actor":"agent"}}}"#;
3759 let archive = build_archive(
3760 tmp.path(),
3761 "seed",
3762 &[
3763 (".memstead/config.json", config),
3764 ("alpha.md", alpha),
3765 ("beta.md", beta),
3766 (".memstead/provenance.json", provenance),
3767 ],
3768 );
3769 let engine = Engine::from_mounts(vec![(
3770 archive_mount("seed", archive.clone()),
3771 Box::new(ArchiveBackend::new(archive)) as Box<dyn MemBackend>,
3772 )])
3773 .unwrap();
3774
3775 let prov = engine
3776 .archive_provenance_for("seed")
3777 .expect("provenance payload read from the archive");
3778 assert_eq!(
3779 prov.history,
3780 History::Summarised,
3781 "history-not-shipped is observable"
3782 );
3783 assert_eq!(
3784 prov.entity("alpha").and_then(|r| r.rationale.as_deref()),
3785 Some("why alpha exists"),
3786 "noted entity surfaces its rationale"
3787 );
3788 assert!(
3789 prov.entity("beta").is_none(),
3790 "unnoted entity is absent (reported absent, not fabricated)"
3791 );
3792 }
3793
3794 /// A pre-provenance archive (no `.memstead/provenance.json`) reads as
3795 /// provenance uniformly absent — the additive contract: a newer engine
3796 /// installing an old archive reports no provenance, never an error.
3797 #[test]
3798 fn archive_without_provenance_reports_absent() {
3799 let tmp = TempDir::new().unwrap();
3800 let config = br#"{"format":3,"name":"seed","version":"0.1.0","schema":"default@1.0.0"}"#;
3801 let alpha = b"---\ntype: spec\n---\n# Alpha\n\n## Identity\n\na\n\n## Purpose\n\np\n";
3802 let archive = build_archive(
3803 tmp.path(),
3804 "seed",
3805 &[(".memstead/config.json", config), ("alpha.md", alpha)],
3806 );
3807 let engine = Engine::from_mounts(vec![(
3808 archive_mount("seed", archive.clone()),
3809 Box::new(ArchiveBackend::new(archive)) as Box<dyn MemBackend>,
3810 )])
3811 .unwrap();
3812 assert!(
3813 engine.archive_provenance_for("seed").is_none(),
3814 "an archive without a provenance payload reports provenance absent"
3815 );
3816 }
3817
3818 #[test]
3819 fn folder_mount_routes_reads_to_filesystem_backend() {
3820 let tmp = TempDir::new().unwrap();
3821 let mem_dir = tmp.path().to_path_buf();
3822 let writer = FilesystemMemWriter::new(mem_dir.clone());
3823 // MemWriter and MemBackend share method names; the
3824 // module-top `use` brings both into scope. Seed via fully-
3825 // qualified MemWriter calls so dot-syntax stays unambiguous.
3826 <FilesystemMemWriter as MemWriter>::write_entity(&writer, Path::new("a.md"), b"alpha")
3827 .unwrap();
3828 <FilesystemMemWriter as MemWriter>::commit(&writer, "seed", &CommitContext::internal())
3829 .unwrap();
3830
3831 let engine = Engine::from_mounts(vec![(
3832 folder_mount("specs", mem_dir),
3833 Box::new(writer) as Box<dyn MemBackend>,
3834 )])
3835 .unwrap();
3836
3837 let mut paths: Vec<String> = engine
3838 .list_entities("specs")
3839 .unwrap()
3840 .into_iter()
3841 .map(|p| p.to_string_lossy().into_owned())
3842 .collect();
3843 paths.sort();
3844 assert_eq!(paths, vec!["a.md".to_string()]);
3845
3846 assert_eq!(
3847 engine.read_entity("specs", Path::new("a.md")).unwrap(),
3848 Some(b"alpha".to_vec())
3849 );
3850 }
3851
3852 #[test]
3853 fn heterogeneous_mounts_route_to_correct_backend() {
3854 let tmp = TempDir::new().unwrap();
3855
3856 // Folder mem.
3857 let folder_dir = tmp.path().join("folder-mem");
3858 std::fs::create_dir_all(&folder_dir).unwrap();
3859 let folder_writer = FilesystemMemWriter::new(folder_dir.clone());
3860 <FilesystemMemWriter as MemWriter>::write_entity(
3861 &folder_writer,
3862 Path::new("local.md"),
3863 b"local",
3864 )
3865 .unwrap();
3866 <FilesystemMemWriter as MemWriter>::commit(
3867 &folder_writer,
3868 "seed",
3869 &CommitContext::internal(),
3870 )
3871 .unwrap();
3872
3873 // Archive mem.
3874 let archive_path = build_archive(
3875 tmp.path(),
3876 "external",
3877 &[("ext.md", b"external"), ("dir/nested.md", b"nested")],
3878 );
3879
3880 let engine = Engine::from_mounts(vec![
3881 (
3882 folder_mount("local", folder_dir),
3883 Box::new(folder_writer) as Box<dyn MemBackend>,
3884 ),
3885 (
3886 archive_mount("external", archive_path.clone()),
3887 Box::new(ArchiveBackend::new(archive_path)),
3888 ),
3889 ])
3890 .unwrap();
3891
3892 // Routes correctly by mem name.
3893 assert_eq!(engine.mem_names(), vec!["local", "external"]);
3894 assert_eq!(
3895 engine.read_entity("local", Path::new("local.md")).unwrap(),
3896 Some(b"local".to_vec())
3897 );
3898 assert_eq!(
3899 engine.read_entity("external", Path::new("ext.md")).unwrap(),
3900 Some(b"external".to_vec())
3901 );
3902 assert_eq!(
3903 engine
3904 .read_entity("external", Path::new("dir/nested.md"))
3905 .unwrap(),
3906 Some(b"nested".to_vec())
3907 );
3908 // Cross-routing: reading a path from the wrong mem → None
3909 // (the backend doesn't have it), not an error.
3910 assert_eq!(
3911 engine.read_entity("local", Path::new("ext.md")).unwrap(),
3912 None
3913 );
3914 assert_eq!(
3915 engine
3916 .read_entity("external", Path::new("local.md"))
3917 .unwrap(),
3918 None
3919 );
3920 }
3921
3922 #[test]
3923 fn edge_is_from_readonly_classifies_every_edge_by_source_mount_capability() {
3924 // `engine.edge_is_from_readonly` is the derived-on-demand
3925 // alternative to adding a per-edge marker: construct a mixed
3926 // workspace (one Write-Mem + one ReadOnly archive with
3927 // cross-mem wiki-links) and walk every edge in the store,
3928 // asserting each edge's source-mount capability.
3929 let tmp = TempDir::new().unwrap();
3930
3931 // Write folder mem `local` with a spec-shaped entity that
3932 // declares an explicit cross-mem relation into the archive
3933 // (under the alias model edges originate from `## Relationships`).
3934 let folder_dir = tmp.path().join("local-mem");
3935 std::fs::create_dir_all(&folder_dir).unwrap();
3936 let folder_writer = FilesystemMemWriter::new(folder_dir.clone());
3937 let local_md = b"---\ntype: spec\n---\n# Note\n\n## Identity\n\nsee [[external:archived]] for prior context.\n\n## Relationships\n\n- **REFERENCES**: [[external:archived]]\n";
3938 <FilesystemMemWriter as MemWriter>::write_entity(
3939 &folder_writer,
3940 Path::new("note.md"),
3941 local_md,
3942 )
3943 .unwrap();
3944 <FilesystemMemWriter as MemWriter>::commit(
3945 &folder_writer,
3946 "seed",
3947 &CommitContext::internal(),
3948 )
3949 .unwrap();
3950
3951 // ReadOnly archive mem `external` with a spec-shaped entity
3952 // declaring an explicit cross-mem relation back to the local
3953 // note.
3954 let archive_md = b"---\ntype: spec\n---\n# Archived\n\n## Identity\n\nrefers back to [[local:note]] for the current revision.\n\n## Relationships\n\n- **REFERENCES**: [[local:note]]\n";
3955 let archive_path = build_archive(tmp.path(), "external", &[("archived.md", archive_md)]);
3956
3957 let engine = Engine::from_mounts(vec![
3958 (
3959 folder_mount("local", folder_dir),
3960 Box::new(folder_writer) as Box<dyn MemBackend>,
3961 ),
3962 (
3963 archive_mount("external", archive_path.clone()),
3964 Box::new(ArchiveBackend::new(archive_path)),
3965 ),
3966 ])
3967 .unwrap();
3968
3969 // Sanity: both entities are real, both mems are mounted.
3970 let local_id = EntityId::new("local", "note");
3971 let archived_id = EntityId::new("external", "archived");
3972 assert!(engine.get_entity(&local_id).is_some());
3973 assert!(engine.get_entity(&archived_id).is_some());
3974 assert!(matches!(
3975 engine.capability("local").unwrap(),
3976 MountCapability::Write
3977 ));
3978 assert!(matches!(
3979 engine.capability("external").unwrap(),
3980 MountCapability::ReadOnly
3981 ));
3982
3983 // Walk every edge in the store. For each (from, edge) pair,
3984 // `edge_is_from_readonly(from)` must return true iff the
3985 // source mount's capability is ReadOnly. The fixture's two
3986 // wiki-links produce one edge from each mem — both halves
3987 // exercise both branches of the helper.
3988 let mut seen_write_edge = false;
3989 let mut seen_readonly_edge = false;
3990 for from in engine.store().all_ids().cloned().collect::<Vec<_>>() {
3991 for _edge in engine.store().outgoing(&from) {
3992 let is_ro = engine.edge_is_from_readonly(&from);
3993 match engine.capability(from.mem()).unwrap() {
3994 MountCapability::Write => {
3995 assert!(
3996 !is_ro,
3997 "edge from write mem {} reported as ReadOnly",
3998 from.mem()
3999 );
4000 seen_write_edge = true;
4001 }
4002 MountCapability::ReadOnly => {
4003 assert!(
4004 is_ro,
4005 "edge from readonly mem {} reported as Write",
4006 from.mem()
4007 );
4008 seen_readonly_edge = true;
4009 }
4010 }
4011 }
4012 }
4013 assert!(
4014 seen_write_edge,
4015 "fixture must produce at least one edge from a write mem"
4016 );
4017 assert!(
4018 seen_readonly_edge,
4019 "fixture must produce at least one edge from a readonly mem"
4020 );
4021
4022 // Helper also reports `false` for mems absent from the
4023 // router — no mount → no ReadOnly assertion can be made.
4024 let phantom = EntityId::new("missing-mem", "phantom");
4025 assert!(
4026 !engine.edge_is_from_readonly(&phantom),
4027 "absent mount must not be reported as ReadOnly"
4028 );
4029 }
4030
4031 // ---- Engine::changes_since wrapper ------------------------------
4032
4033 #[test]
4034 fn cross_mem_link_allowed_same_mem_always_true() {
4035 // Self-edges (from == to) bypass the cross-mem policy
4036 // entirely — the policy gates *cross*-mem edges only.
4037 let tmp = TempDir::new().unwrap();
4038 let engine = build_demo_engine(&tmp);
4039 assert!(engine.cross_mem_link_allowed("specs", "specs"));
4040 // Even when the mem doesn't exist (not enrolled in
4041 // settings.cross_mem_links), same-mem returns true —
4042 // the engine doesn't validate mem existence here, just the
4043 // policy.
4044 assert!(engine.cross_mem_link_allowed("anywhere", "anywhere"));
4045 }
4046
4047 #[test]
4048 fn cross_mem_link_allowed_absent_denies_by_default() {
4049 // No entry in cross_mem_links for `from_mem` → denied.
4050 // Default-deny is the V1 posture; operators opt in.
4051 let tmp = TempDir::new().unwrap();
4052 let engine = build_demo_engine(&tmp);
4053 assert!(!engine.cross_mem_link_allowed("specs", "engine"));
4054 assert!(!engine.cross_mem_link_allowed("missing", "anywhere"));
4055 }
4056
4057 #[test]
4058 fn cross_mem_link_allowed_wildcard_admits_any_target() {
4059 use memstead_schema::workspace_config::CrossLinkValue;
4060 let tmp = TempDir::new().unwrap();
4061 let mut engine = build_demo_engine(&tmp);
4062 let mut settings = crate::workspace::WorkspaceSettings::default();
4063 settings
4064 .cross_mem_links
4065 .insert("specs".to_string(), CrossLinkValue::Wildcard);
4066 engine.set_settings(settings);
4067 assert!(engine.cross_mem_link_allowed("specs", "engine"));
4068 assert!(engine.cross_mem_link_allowed("specs", "macos"));
4069 assert!(engine.cross_mem_link_allowed("specs", "any-other"));
4070 // Reverse direction is independent — no policy entry for
4071 // engine→specs means denied.
4072 assert!(!engine.cross_mem_link_allowed("engine", "specs"));
4073 }
4074
4075 #[test]
4076 fn cross_mem_link_allowed_allowlist_enforces_membership() {
4077 use memstead_schema::workspace_config::CrossLinkValue;
4078 let tmp = TempDir::new().unwrap();
4079 let mut engine = build_demo_engine(&tmp);
4080 let mut settings = crate::workspace::WorkspaceSettings::default();
4081 settings.cross_mem_links.insert(
4082 "specs".to_string(),
4083 CrossLinkValue::List(vec!["engine".to_string(), "macos".to_string()]),
4084 );
4085 engine.set_settings(settings);
4086 assert!(engine.cross_mem_link_allowed("specs", "engine"));
4087 assert!(engine.cross_mem_link_allowed("specs", "macos"));
4088 assert!(!engine.cross_mem_link_allowed("specs", "external"));
4089 }
4090
4091 #[test]
4092 fn cross_mem_link_allowed_synthesises_from_matching_create_rule_wildcard() {
4093 // No explicit cross_mem_links entry, but a create rule
4094 // matches `from_mem` and carries default_cross_links = "*".
4095 // Synthesis grants permission to any target.
4096 use memstead_schema::workspace_config::CrossLinkValue;
4097 let tmp = TempDir::new().unwrap();
4098 let mut engine = build_demo_engine(&tmp);
4099 let mut settings = crate::workspace::WorkspaceSettings::default();
4100 settings
4101 .mem_create_rules
4102 .push(crate::workspace::CreateRuleSetting {
4103 pattern: "exec-*".to_string(),
4104 schemas: vec!["default".to_string()],
4105 default_cross_links: Some(CrossLinkValue::Wildcard),
4106 });
4107 engine.set_settings(settings);
4108 // No explicit policy; synthesis grants permission for any
4109 // target because the rule's value is Wildcard.
4110 assert!(engine.cross_mem_link_allowed("exec-foo", "specs"));
4111 assert!(engine.cross_mem_link_allowed("exec-foo", "engine"));
4112 // Mem that doesn't match any rule → still denied.
4113 assert!(!engine.cross_mem_link_allowed("orphan", "specs"));
4114 }
4115
4116 /// #42: synthesis matches a hierarchical mem by composing the same
4117 /// `<mem_path>/<name>` candidate the create-rule glob is keyed on,
4118 /// not the bare leaf. Before the fix, `from_mem = "project"` could
4119 /// never match a `memstead/*` rule (the leaf-vs-composed-path
4120 /// divergence), so enforcement denied a link `memstead_overview`
4121 /// rendered as rule-granted.
4122 #[test]
4123 fn cross_mem_link_allowed_synthesises_for_hierarchical_mem() {
4124 use memstead_schema::workspace_config::CrossLinkValue;
4125 let tmp = TempDir::new().unwrap();
4126 let mem_dir = tmp.path().to_path_buf();
4127 // Mount `project` with a hierarchical branch so its `mem_path()`
4128 // is "memstead" and the composed candidate is "memstead/project".
4129 // The Folder backend handles loading; only the Mount's storage
4130 // feeds `mem_path()`.
4131 let mount = Mount {
4132 mem: "project".into(),
4133 schema: Some(pin("default")),
4134 storage: MountStorage::GitBranch {
4135 gitdir: mem_dir.join(".git"),
4136 branch: "memstead/project".into(),
4137 },
4138 capability: MountCapability::Write,
4139 lifecycle: MountLifecycle::Eager,
4140 cross_linkable: true,
4141 migration_target: None,
4142 };
4143 let writer = FilesystemMemWriter::new(mem_dir.clone());
4144 let mut engine =
4145 Engine::from_mounts(vec![(mount, Box::new(writer) as Box<dyn MemBackend>)]).unwrap();
4146 let mut settings = crate::workspace::WorkspaceSettings::default();
4147 settings
4148 .mem_create_rules
4149 .push(crate::workspace::CreateRuleSetting {
4150 pattern: "memstead/*".to_string(),
4151 schemas: vec!["default".to_string()],
4152 default_cross_links: Some(CrossLinkValue::List(vec!["engine".to_string()])),
4153 });
4154 engine.set_settings(settings);
4155 assert!(
4156 engine.cross_mem_link_allowed("project", "engine"),
4157 "synthesis must match via the composed `memstead/project` candidate"
4158 );
4159 assert!(
4160 !engine.cross_mem_link_allowed("project", "macos"),
4161 "a target outside the rule's default_cross_links is still denied"
4162 );
4163 }
4164
4165 #[test]
4166 fn cross_mem_link_allowed_synthesises_from_matching_create_rule_list() {
4167 // Create rule's default_cross_links is a list — synthesis
4168 // grants permission to listed targets only.
4169 use memstead_schema::workspace_config::CrossLinkValue;
4170 let tmp = TempDir::new().unwrap();
4171 let mut engine = build_demo_engine(&tmp);
4172 let mut settings = crate::workspace::WorkspaceSettings::default();
4173 settings
4174 .mem_create_rules
4175 .push(crate::workspace::CreateRuleSetting {
4176 pattern: "exec-*".to_string(),
4177 schemas: vec!["default".to_string()],
4178 default_cross_links: Some(CrossLinkValue::List(vec!["specs".to_string()])),
4179 });
4180 engine.set_settings(settings);
4181 assert!(engine.cross_mem_link_allowed("exec-foo", "specs"));
4182 // Target not in the synthesised list → denied.
4183 assert!(!engine.cross_mem_link_allowed("exec-foo", "engine"));
4184 }
4185
4186 #[test]
4187 fn cross_mem_link_allowed_explicit_policy_wins_over_synthesis() {
4188 // Explicit cross_mem_links wildcard fires first; the
4189 // synthesis layer is never consulted (and would deny).
4190 use memstead_schema::workspace_config::CrossLinkValue;
4191 let tmp = TempDir::new().unwrap();
4192 let mut engine = build_demo_engine(&tmp);
4193 let mut settings = crate::workspace::WorkspaceSettings::default();
4194 settings
4195 .cross_mem_links
4196 .insert("exec-foo".to_string(), CrossLinkValue::Wildcard);
4197 // The synthesis layer would deny `exec-foo → engine` (no
4198 // matching rule), but explicit policy returns true first.
4199 engine.set_settings(settings);
4200 assert!(engine.cross_mem_link_allowed("exec-foo", "engine"));
4201 }
4202
4203 #[test]
4204 fn cross_mem_link_allowed_synthesis_unions_into_explicit_list() {
4205 // Explicit list = ["specs"]; create rule synthesises = ["macos"].
4206 // Effective allowed targets: union ({specs, macos}).
4207 use memstead_schema::workspace_config::CrossLinkValue;
4208 let tmp = TempDir::new().unwrap();
4209 let mut engine = build_demo_engine(&tmp);
4210 let mut settings = crate::workspace::WorkspaceSettings::default();
4211 settings.cross_mem_links.insert(
4212 "exec-foo".to_string(),
4213 CrossLinkValue::List(vec!["specs".to_string()]),
4214 );
4215 settings
4216 .mem_create_rules
4217 .push(crate::workspace::CreateRuleSetting {
4218 pattern: "exec-*".to_string(),
4219 schemas: vec!["default".to_string()],
4220 default_cross_links: Some(CrossLinkValue::List(vec!["macos".to_string()])),
4221 });
4222 engine.set_settings(settings);
4223 // Explicit allowlist contains specs → allowed.
4224 assert!(engine.cross_mem_link_allowed("exec-foo", "specs"));
4225 // Synthesis layer adds macos → allowed.
4226 assert!(engine.cross_mem_link_allowed("exec-foo", "macos"));
4227 // Neither layer allows engine → denied.
4228 assert!(!engine.cross_mem_link_allowed("exec-foo", "engine"));
4229 }
4230
4231 #[test]
4232 fn cross_mem_link_allowed_set_settings_invalidates_compiled_rule_cache() {
4233 // After set_settings, a fresh policy must be reflected on the
4234 // next call — the lazy memo can't return stale rules.
4235 use memstead_schema::workspace_config::CrossLinkValue;
4236 let tmp = TempDir::new().unwrap();
4237 let mut engine = build_demo_engine(&tmp);
4238
4239 // First settings: a rule allows exec-* → specs via synthesis.
4240 let mut s1 = crate::workspace::WorkspaceSettings::default();
4241 s1.mem_create_rules
4242 .push(crate::workspace::CreateRuleSetting {
4243 pattern: "exec-*".to_string(),
4244 schemas: vec!["default".to_string()],
4245 default_cross_links: Some(CrossLinkValue::List(vec!["specs".to_string()])),
4246 });
4247 engine.set_settings(s1);
4248 assert!(engine.cross_mem_link_allowed("exec-foo", "specs"));
4249
4250 // Replace settings: the rule no longer carries
4251 // default_cross_links. Cache must invalidate so the next
4252 // call sees the new policy.
4253 let mut s2 = crate::workspace::WorkspaceSettings::default();
4254 s2.mem_create_rules
4255 .push(crate::workspace::CreateRuleSetting {
4256 pattern: "exec-*".to_string(),
4257 schemas: vec!["default".to_string()],
4258 default_cross_links: None,
4259 });
4260 engine.set_settings(s2);
4261 assert!(!engine.cross_mem_link_allowed("exec-foo", "specs"));
4262 }
4263
4264 #[test]
4265 fn cross_mem_link_allowed_malformed_glob_falls_back_to_explicit_policy() {
4266 // Malformed pattern in a create rule causes CreateRuleSet
4267 // compilation to fail; the resolver logs and disables
4268 // synthesis, but explicit cross_mem_links still works.
4269 use memstead_schema::workspace_config::CrossLinkValue;
4270 let tmp = TempDir::new().unwrap();
4271 let mut engine = build_demo_engine(&tmp);
4272 let mut settings = crate::workspace::WorkspaceSettings::default();
4273 settings
4274 .mem_create_rules
4275 .push(crate::workspace::CreateRuleSetting {
4276 pattern: "[unclosed".to_string(),
4277 schemas: vec!["default".to_string()],
4278 default_cross_links: Some(CrossLinkValue::Wildcard),
4279 });
4280 // Explicit policy still works.
4281 settings
4282 .cross_mem_links
4283 .insert("specs".to_string(), CrossLinkValue::Wildcard);
4284 engine.set_settings(settings);
4285 // Explicit policy: specs → engine allowed.
4286 assert!(engine.cross_mem_link_allowed("specs", "engine"));
4287 // Synthesis disabled (compilation failed); rule's would-be
4288 // wildcard doesn't apply.
4289 assert!(!engine.cross_mem_link_allowed("orphan", "anything"));
4290 }
4291
4292 #[test]
4293 fn cross_mem_link_allowed_empty_list_denies_all_cross_mem_targets() {
4294 // [cross_mem_links] specs = [] is the explicit
4295 // "intentionally locked down" shape — same effect as
4296 // default-deny but operator-acknowledged.
4297 use memstead_schema::workspace_config::CrossLinkValue;
4298 let tmp = TempDir::new().unwrap();
4299 let mut engine = build_demo_engine(&tmp);
4300 let mut settings = crate::workspace::WorkspaceSettings::default();
4301 settings
4302 .cross_mem_links
4303 .insert("specs".to_string(), CrossLinkValue::List(Vec::new()));
4304 engine.set_settings(settings);
4305 // Same-mem still passes — policy only gates cross-mem.
4306 assert!(engine.cross_mem_link_allowed("specs", "specs"));
4307 // Cross-mem denied to every target.
4308 assert!(!engine.cross_mem_link_allowed("specs", "engine"));
4309 assert!(!engine.cross_mem_link_allowed("specs", "anything"));
4310 }
4311
4312 #[test]
4313 fn from_mounts_load_warnings_merge_into_health_summary() {
4314 let tmp = TempDir::new().unwrap();
4315 let mem_dir = tmp.path().to_path_buf();
4316 let body = "---\ntype: spec\n---\n# Dup2\n\n## Identity\n\na.\n\n## Identity\n\nb.\n";
4317 std::fs::write(mem_dir.join("dup2.md"), body).unwrap();
4318
4319 let writer = FilesystemMemWriter::new(mem_dir.clone());
4320 let engine = Engine::from_mounts(vec![(
4321 folder_mount("specs", mem_dir),
4322 Box::new(writer) as Box<dyn MemBackend>,
4323 )])
4324 .unwrap();
4325
4326 let summary = engine.health();
4327 assert!(
4328 summary
4329 .warnings
4330 .iter()
4331 .any(|w| matches!(w, WarningHint::DuplicateSectionHeading { .. })),
4332 "health() must merge load_warnings into summary.warnings: {:?}",
4333 summary.warnings,
4334 );
4335 }
4336
4337 #[test]
4338 fn workspace_root_accessor_is_none_for_engine_built_from_mounts() {
4339 let tmp = TempDir::new().unwrap();
4340 let mem_dir = tmp.path().to_path_buf();
4341 let writer = FilesystemMemWriter::new(mem_dir.clone());
4342 // Newest default generation so the clean-boot assertion below
4343 // isn't tripped by the SCHEMA_GENERATIONS_BEHIND hint.
4344 let mut mount = folder_mount("specs", mem_dir);
4345 mount.schema = Some("default@1.3.0".parse().unwrap());
4346 let engine =
4347 Engine::from_mounts(vec![(mount, Box::new(writer) as Box<dyn MemBackend>)]).unwrap();
4348 assert!(
4349 engine.workspace_root().is_none(),
4350 "from_mounts has no workspace path",
4351 );
4352 // An entity-less mount reports itself empty; nothing else.
4353 assert!(
4354 engine
4355 .load_warnings()
4356 .iter()
4357 .all(|w| w.code() == "MOUNT_UNBACKED"),
4358 "{:?}",
4359 engine.load_warnings()
4360 );
4361 }
4362
4363 #[test]
4364 fn health_omits_outer_repo_warning_when_workspace_root_unset() {
4365 let tmp = TempDir::new().unwrap();
4366 let mem_dir = tmp.path().to_path_buf();
4367 let writer = FilesystemMemWriter::new(mem_dir.clone());
4368 let engine = Engine::from_mounts(vec![(
4369 folder_mount("specs", mem_dir),
4370 Box::new(writer) as Box<dyn MemBackend>,
4371 )])
4372 .unwrap();
4373 let health = engine.health();
4374 assert!(
4375 !health
4376 .warnings
4377 .iter()
4378 .any(|w| matches!(w, WarningHint::OuterRepoNotIgnoringMemRepo { .. })),
4379 "outer-repo check must skip when workspace_root is None",
4380 );
4381 }
4382
4383 #[test]
4384 fn writable_mem_names_filters_by_capability() {
4385 let tmp = TempDir::new().unwrap();
4386 let mem_dir = tmp.path().to_path_buf();
4387 let writer = FilesystemMemWriter::new(mem_dir.clone());
4388 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4389
4390 let engine = Engine::from_mounts(vec![
4391 (
4392 folder_mount("writable", mem_dir),
4393 Box::new(writer) as Box<dyn MemBackend>,
4394 ),
4395 (
4396 archive_mount("sealed", archive_path.clone()),
4397 Box::new(ArchiveBackend::new(archive_path)),
4398 ),
4399 ])
4400 .unwrap();
4401
4402 // Only the writable mount surfaces; the archive (read-only)
4403 // is filtered out.
4404 let names = engine.writable_mem_names();
4405 assert_eq!(names, vec!["writable"]);
4406 }
4407
4408 /// The default writable mem is
4409 /// the FIRST writable mount in declaration order — the stable seed,
4410 /// not the alphabetically-first name. `test` is declared first;
4411 /// `other` sorts ahead alphabetically but is declared second, so it
4412 /// is NOT the default. This is the invariant that stops a second
4413 /// mem from silently retargeting omitted-`mem` writes.
4414 #[test]
4415 fn default_writable_mem_is_declaration_first_not_alphabetical() {
4416 let tmp = TempDir::new().unwrap();
4417 let test_dir = tmp.path().join("test");
4418 let other_dir = tmp.path().join("other");
4419 std::fs::create_dir_all(&test_dir).unwrap();
4420 std::fs::create_dir_all(&other_dir).unwrap();
4421
4422 let engine = Engine::from_mounts(vec![
4423 (
4424 folder_mount("test", test_dir.clone()),
4425 Box::new(FilesystemMemWriter::new(test_dir)) as Box<dyn MemBackend>,
4426 ),
4427 (
4428 folder_mount("other", other_dir.clone()),
4429 Box::new(FilesystemMemWriter::new(other_dir)) as Box<dyn MemBackend>,
4430 ),
4431 ])
4432 .unwrap();
4433
4434 assert_eq!(
4435 engine.default_writable_mem(),
4436 Some("test"),
4437 "default must be the declaration-first writable mem, not the alphabetically-first",
4438 );
4439 }
4440
4441 /// Reverse declaration order to prove the default tracks declaration
4442 /// order rather than a fixed name: with `other` declared first it
4443 /// becomes the default. Together with the test above this pins the
4444 /// lean as mount order, not name sort.
4445 #[test]
4446 fn default_writable_mem_follows_declaration_order() {
4447 let tmp = TempDir::new().unwrap();
4448 let other_dir = tmp.path().join("other");
4449 let test_dir = tmp.path().join("test");
4450 std::fs::create_dir_all(&other_dir).unwrap();
4451 std::fs::create_dir_all(&test_dir).unwrap();
4452
4453 let engine = Engine::from_mounts(vec![
4454 (
4455 folder_mount("other", other_dir.clone()),
4456 Box::new(FilesystemMemWriter::new(other_dir)) as Box<dyn MemBackend>,
4457 ),
4458 (
4459 folder_mount("test", test_dir.clone()),
4460 Box::new(FilesystemMemWriter::new(test_dir)) as Box<dyn MemBackend>,
4461 ),
4462 ])
4463 .unwrap();
4464
4465 assert_eq!(engine.default_writable_mem(), Some("other"));
4466 }
4467
4468 /// A read-only-only workspace has no default writable mem.
4469 #[test]
4470 fn default_writable_mem_none_without_writable_mount() {
4471 let tmp = TempDir::new().unwrap();
4472 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4473 let engine = Engine::from_mounts(vec![(
4474 archive_mount("sealed", archive_path.clone()),
4475 Box::new(ArchiveBackend::new(archive_path)) as Box<dyn MemBackend>,
4476 )])
4477 .unwrap();
4478 assert_eq!(engine.default_writable_mem(), None);
4479 }
4480
4481 #[test]
4482 fn folder_path_for_mem_returns_path_for_folder_mounts_only() {
4483 let tmp = TempDir::new().unwrap();
4484 let mem_dir = tmp.path().join("specs");
4485 std::fs::create_dir_all(&mem_dir).unwrap();
4486 let writer = FilesystemMemWriter::new(mem_dir.clone());
4487 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4488
4489 let engine = Engine::from_mounts(vec![
4490 (
4491 folder_mount("specs", mem_dir.clone()),
4492 Box::new(writer) as Box<dyn MemBackend>,
4493 ),
4494 (
4495 archive_mount("sealed", archive_path.clone()),
4496 Box::new(ArchiveBackend::new(archive_path)),
4497 ),
4498 ])
4499 .unwrap();
4500
4501 // Folder mount returns its path.
4502 assert_eq!(engine.folder_path_for_mem("specs"), Some(mem_dir.as_path()),);
4503 // Archive mount returns None — caller branches on storage type.
4504 assert_eq!(engine.folder_path_for_mem("sealed"), None);
4505 // Unknown mem returns None — same as Engine::mount.
4506 assert_eq!(engine.folder_path_for_mem("missing"), None);
4507 }
4508
4509 #[test]
4510 fn mount_accessor_returns_public_mount_shape() {
4511 // Build a heterogeneous engine and verify Engine::mount /
4512 // Engine::mounts surface the operator-facing Mount records.
4513 // Handlers branch on MountStorage variants through this
4514 // accessor (replacing full's gitdir_for / worktree_for /
4515 // mem_head_sha / mem_config_for direct-engine
4516 // accessors).
4517 let tmp = TempDir::new().unwrap();
4518 let mem_dir = tmp.path().to_path_buf();
4519 let writer = FilesystemMemWriter::new(mem_dir.clone());
4520 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4521
4522 let engine = Engine::from_mounts(vec![
4523 (
4524 folder_mount("writable", mem_dir.clone()),
4525 Box::new(writer) as Box<dyn MemBackend>,
4526 ),
4527 (
4528 archive_mount("sealed", archive_path.clone()),
4529 Box::new(ArchiveBackend::new(archive_path.clone())),
4530 ),
4531 ])
4532 .unwrap();
4533
4534 // Known mems: each returns a Mount whose storage variant
4535 // matches what the caller passed at construction.
4536 let folder = engine.mount("writable").expect("known mem");
4537 assert!(matches!(folder.storage, MountStorage::Folder { .. }));
4538 assert_eq!(folder.capability, MountCapability::Write);
4539
4540 let archive = engine.mount("sealed").expect("known mem");
4541 match &archive.storage {
4542 MountStorage::Archive { path } => assert_eq!(path, &archive_path),
4543 other => panic!("expected Archive storage, got {other:?}"),
4544 }
4545 assert_eq!(archive.capability, MountCapability::ReadOnly);
4546
4547 // Unknown mem — None, no panic, no error.
4548 assert!(engine.mount("missing").is_none());
4549
4550 // Engine::mounts enumerates every mount in declaration order.
4551 let mounts = engine.mounts();
4552 assert_eq!(mounts.len(), 2);
4553 assert_eq!(mounts[0].mem, "writable");
4554 assert_eq!(mounts[1].mem, "sealed");
4555 }
4556
4557 #[test]
4558 fn mem_router_writable_set_matches_writable_mount_capability() {
4559 // Build an engine with one writable folder mount and one
4560 // read-only archive mount; the router's writable set must
4561 // equal the writable mount's name only.
4562 let tmp = TempDir::new().unwrap();
4563 let mem_dir = tmp.path().join("specs");
4564 std::fs::create_dir_all(&mem_dir).unwrap();
4565 let writer = FilesystemMemWriter::new(mem_dir.clone());
4566 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4567
4568 let engine = Engine::from_mounts(vec![
4569 (
4570 folder_mount("specs", mem_dir.clone()),
4571 Box::new(writer) as Box<dyn MemBackend>,
4572 ),
4573 (
4574 archive_mount("ext", archive_path.clone()),
4575 Box::new(ArchiveBackend::new(archive_path)),
4576 ),
4577 ])
4578 .unwrap();
4579
4580 let router = engine.mem_router();
4581 assert!(router.is_writable("specs"));
4582 assert!(!router.is_writable("ext"));
4583 assert!(router.is_visible("specs"));
4584 assert!(router.is_visible("ext"));
4585 let writable: std::collections::HashSet<&String> = router.writable_mems().iter().collect();
4586 assert_eq!(writable.len(), 1);
4587 assert!(writable.contains(&"specs".to_string()));
4588 }
4589
4590 #[test]
4591 fn mem_router_origin_is_explicit_toml_for_workspace_mounts() {
4592 // Every mount built via `from_mounts` lands as
4593 // `MemOrigin::ExplicitToml` — the file-adapter origin.
4594 // `RuntimeCreated` is reserved for `memstead_mem_create`
4595 // runtime registrations once that handler migrates onto
4596 // the unified engine.
4597 let tmp = TempDir::new().unwrap();
4598 let mem_dir = tmp.path().join("specs");
4599 std::fs::create_dir_all(&mem_dir).unwrap();
4600 let writer = FilesystemMemWriter::new(mem_dir.clone());
4601
4602 let engine = Engine::from_mounts(vec![(
4603 folder_mount("specs", mem_dir),
4604 Box::new(writer) as Box<dyn MemBackend>,
4605 )])
4606 .unwrap();
4607
4608 let origin = engine
4609 .mem_router()
4610 .origin_for_mem("specs")
4611 .expect("known mem");
4612 assert_eq!(origin.kind(), "explicit");
4613 }
4614
4615 #[test]
4616 fn mem_router_dir_for_writable_folder_mount_matches_storage_path() {
4617 // Folder-backed writable mounts surface the storage path
4618 // via `dir_for_mem`. Handlers consuming the router for
4619 // per-mem path resolution rely on this.
4620 let tmp = TempDir::new().unwrap();
4621 let mem_dir = tmp.path().join("specs");
4622 std::fs::create_dir_all(&mem_dir).unwrap();
4623 let writer = FilesystemMemWriter::new(mem_dir.clone());
4624
4625 let engine = Engine::from_mounts(vec![(
4626 folder_mount("specs", mem_dir.clone()),
4627 Box::new(writer) as Box<dyn MemBackend>,
4628 )])
4629 .unwrap();
4630
4631 assert_eq!(
4632 engine.mem_router().dir_for_mem("specs"),
4633 Some(mem_dir.as_path()),
4634 );
4635 assert_eq!(engine.mem_router().dir_for_mem("unknown"), None);
4636 }
4637
4638 #[test]
4639 fn mem_router_archive_path_for_read_only_archive_mount() {
4640 // Read-only archive mounts register via `add_read_only` so
4641 // `archive_path_for_mem` resolves the archive's on-disk
4642 // location.
4643 let tmp = TempDir::new().unwrap();
4644 let mem_dir = tmp.path().join("specs");
4645 std::fs::create_dir_all(&mem_dir).unwrap();
4646 let writer = FilesystemMemWriter::new(mem_dir.clone());
4647 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4648
4649 let engine = Engine::from_mounts(vec![
4650 (
4651 folder_mount("specs", mem_dir),
4652 Box::new(writer) as Box<dyn MemBackend>,
4653 ),
4654 (
4655 archive_mount("ext", archive_path.clone()),
4656 Box::new(ArchiveBackend::new(archive_path.clone())),
4657 ),
4658 ])
4659 .unwrap();
4660
4661 let router = engine.mem_router();
4662 assert_eq!(
4663 router.archive_path_for_mem("ext"),
4664 Some(archive_path.as_path()),
4665 );
4666 // Writable folder mount has no archive path.
4667 assert_eq!(router.archive_path_for_mem("specs"), None);
4668 }
4669
4670 #[test]
4671 fn read_mem_config_via_backend_trait_folder_reads_bytes() {
4672 // Direct trait call against FilesystemMemWriter. Verifies
4673 // the backend-side primitive returns the raw bytes the
4674 // engine then parses.
4675 let tmp = TempDir::new().unwrap();
4676 let mem_dir = tmp.path().to_path_buf();
4677 std::fs::create_dir_all(mem_dir.join(".memstead")).unwrap();
4678 let body = br#"{
4679 "format": 1,
4680 "schema": "default@1.0.0",
4681 "writeGuidance": { "tone": "neutral" }
4682 }"#;
4683 std::fs::write(mem_dir.join(".memstead").join("config.json"), body).unwrap();
4684
4685 let writer = FilesystemMemWriter::new(mem_dir);
4686 let result = MemBackend::read_mem_config(&writer).unwrap();
4687 let bytes = result.expect("config bytes must surface");
4688 let parsed: serde_json::Value = serde_json::from_slice(&bytes).unwrap();
4689 assert_eq!(parsed["schema"], "default@1.0.0");
4690 }
4691
4692 #[test]
4693 fn read_mem_config_via_backend_trait_folder_missing_returns_none() {
4694 let tmp = TempDir::new().unwrap();
4695 let mem_dir = tmp.path().to_path_buf();
4696 let writer = FilesystemMemWriter::new(mem_dir);
4697 let result = MemBackend::read_mem_config(&writer).unwrap();
4698 assert!(result.is_none());
4699 }
4700
4701 #[test]
4702 fn read_mem_config_via_backend_trait_archive_reads_bytes() {
4703 // Build an archive containing .memstead/config.json and verify
4704 // the ArchiveBackend impl returns its bytes.
4705 let tmp = TempDir::new().unwrap();
4706 let archive_path = tmp.path().join("seed.mem");
4707 let body = br#"{
4708 "format": 1,
4709 "schema": "default@1.0.0",
4710 "writeGuidance": { "tone": "archive" }
4711 }"#;
4712 {
4713 let file = std::fs::File::create(&archive_path).unwrap();
4714 let mut writer = zip::ZipWriter::new(file);
4715 writer
4716 .start_file(
4717 ".memstead/config.json",
4718 zip::write::SimpleFileOptions::default(),
4719 )
4720 .unwrap();
4721 use std::io::Write;
4722 writer.write_all(body).unwrap();
4723 writer.finish().unwrap();
4724 }
4725
4726 let backend = ArchiveBackend::new(archive_path);
4727 let result = MemBackend::read_mem_config(&backend).unwrap();
4728 let bytes = result.expect("config bytes must surface");
4729 let parsed: serde_json::Value = serde_json::from_slice(&bytes).unwrap();
4730 assert_eq!(parsed["writeGuidance"]["tone"], "archive");
4731 }
4732
4733 #[test]
4734 fn mem_config_for_returns_none_when_no_config_file_present() {
4735 // Folder backend without a `.memstead/config.json` file. The
4736 // accessor must lenient — return None, not error.
4737 let tmp = TempDir::new().unwrap();
4738 let mem_dir = tmp.path().to_path_buf();
4739 let writer = FilesystemMemWriter::new(mem_dir.clone());
4740 let engine = Engine::from_mounts(vec![(
4741 folder_mount("specs", mem_dir),
4742 Box::new(writer) as Box<dyn MemBackend>,
4743 )])
4744 .unwrap();
4745 assert!(engine.mem_config_for("specs").is_none());
4746 }
4747
4748 #[test]
4749 fn mem_config_for_returns_some_when_config_file_present() {
4750 // Drop a valid `.memstead/config.json` into the mem dir,
4751 // build the engine, and assert the accessor surfaces a
4752 // MemConfig with the right shape (write_guidance entries
4753 // round-trip).
4754 let tmp = TempDir::new().unwrap();
4755 let mem_dir = tmp.path().to_path_buf();
4756 std::fs::create_dir_all(mem_dir.join(".memstead")).unwrap();
4757 let config_body = r#"{
4758 "format": 1,
4759 "schema": "default@1.0.0",
4760 "writeGuidance": {
4761 "tone": "neutral",
4762 "voice": "active"
4763 }
4764 }"#;
4765 std::fs::write(mem_dir.join(".memstead").join("config.json"), config_body).unwrap();
4766
4767 let writer = FilesystemMemWriter::new(mem_dir.clone());
4768 let engine = Engine::from_mounts(vec![(
4769 folder_mount("specs", mem_dir),
4770 Box::new(writer) as Box<dyn MemBackend>,
4771 )])
4772 .unwrap();
4773
4774 let cfg = engine
4775 .mem_config_for("specs")
4776 .expect("mem_config should load");
4777 assert_eq!(cfg.write_guidance.len(), 2);
4778 assert_eq!(
4779 cfg.write_guidance.get("tone").and_then(|v| v.as_str()),
4780 Some("neutral"),
4781 );
4782 assert_eq!(
4783 cfg.write_guidance.get("voice").and_then(|v| v.as_str()),
4784 Some("active"),
4785 );
4786 }
4787
4788 #[test]
4789 fn mem_config_for_unknown_mem_returns_none() {
4790 // Lenient accessor — unknown names get None, not Err.
4791 let tmp = TempDir::new().unwrap();
4792 let mem_dir = tmp.path().to_path_buf();
4793 let writer = FilesystemMemWriter::new(mem_dir.clone());
4794 let engine = Engine::from_mounts(vec![(
4795 folder_mount("specs", mem_dir),
4796 Box::new(writer) as Box<dyn MemBackend>,
4797 )])
4798 .unwrap();
4799 assert!(engine.mem_config_for("missing").is_none());
4800 }
4801
4802 #[test]
4803 fn mem_config_for_archive_mount_returns_none() {
4804 // Archive backends carry mem_config = None in V1 (the
4805 // read-from-storage path is deferred to a follow-up).
4806 let tmp = TempDir::new().unwrap();
4807 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4808 let engine = Engine::from_mounts(vec![(
4809 archive_mount("ext", archive_path.clone()),
4810 Box::new(ArchiveBackend::new(archive_path)) as Box<dyn MemBackend>,
4811 )])
4812 .unwrap();
4813 assert!(engine.mem_config_for("ext").is_none());
4814 }
4815
4816 #[test]
4817 fn mem_configs_named_iterates_only_mounts_with_config() {
4818 // Two folder mounts; one has a config file, one doesn't.
4819 // The iterator yields exactly the configured one — verifies
4820 // the filter_map shape and that the name comes from the
4821 // mount record (authoritative), not the config body.
4822 let tmp = TempDir::new().unwrap();
4823 let with_config = tmp.path().join("specs");
4824 let without_config = tmp.path().join("memos");
4825 std::fs::create_dir_all(with_config.join(".memstead")).unwrap();
4826 std::fs::create_dir_all(&without_config).unwrap();
4827 let config_body = r#"{
4828 "format": 1,
4829 "schema": "default@1.0.0",
4830 "writeGuidance": { "tone": "neutral" }
4831 }"#;
4832 std::fs::write(
4833 with_config.join(".memstead").join("config.json"),
4834 config_body,
4835 )
4836 .unwrap();
4837
4838 let engine = Engine::from_mounts(vec![
4839 (
4840 folder_mount("specs", with_config.clone()),
4841 Box::new(FilesystemMemWriter::new(with_config)) as Box<dyn MemBackend>,
4842 ),
4843 (
4844 folder_mount("memos", without_config.clone()),
4845 Box::new(FilesystemMemWriter::new(without_config)) as Box<dyn MemBackend>,
4846 ),
4847 ])
4848 .unwrap();
4849
4850 let yielded: Vec<(&str, usize)> = engine
4851 .mem_configs_named()
4852 .map(|(name, cfg)| (name, cfg.write_guidance.len()))
4853 .collect();
4854 assert_eq!(yielded, vec![("specs", 1)]);
4855 }
4856
4857 #[test]
4858 fn schema_for_returns_some_for_known_mem_and_none_for_unknown() {
4859 // Every mount registers a schema (resolved from its pin at
4860 // boot). Lookup by mem name surfaces the same Arc that
4861 // mutations resolve internally; unknown names return None.
4862 let tmp = TempDir::new().unwrap();
4863 let mem_dir = tmp.path().to_path_buf();
4864 let writer = FilesystemMemWriter::new(mem_dir.clone());
4865 let engine = Engine::from_mounts(vec![(
4866 folder_mount("specs", mem_dir),
4867 Box::new(writer) as Box<dyn MemBackend>,
4868 )])
4869 .unwrap();
4870 assert!(engine.schema_for("specs").is_some());
4871 assert!(engine.schema_for("missing").is_none());
4872 }
4873
4874 #[test]
4875 fn gitdir_for_unknown_mem_returns_unknown_mem() {
4876 let tmp = TempDir::new().unwrap();
4877 let mem_dir = tmp.path().to_path_buf();
4878 let writer = FilesystemMemWriter::new(mem_dir.clone());
4879 let engine = Engine::from_mounts(vec![(
4880 folder_mount("specs", mem_dir),
4881 Box::new(writer) as Box<dyn MemBackend>,
4882 )])
4883 .unwrap();
4884 let err = engine.gitdir_for("missing").unwrap_err();
4885 assert!(matches!(err, EngineError::UnknownMem(v) if v == "missing"));
4886 }
4887
4888 #[test]
4889 fn gitdir_for_folder_mount_returns_no_gitdir_error() {
4890 // Folder mounts do not have a gitdir — full's contract surfaces
4891 // a mem-level error, not UnknownMem. Mirror that here.
4892 let tmp = TempDir::new().unwrap();
4893 let mem_dir = tmp.path().to_path_buf();
4894 let writer = FilesystemMemWriter::new(mem_dir.clone());
4895 let engine = Engine::from_mounts(vec![(
4896 folder_mount("specs", mem_dir),
4897 Box::new(writer) as Box<dyn MemBackend>,
4898 )])
4899 .unwrap();
4900 let err = engine.gitdir_for("specs").unwrap_err();
4901 match err {
4902 EngineError::Mem(msg) => assert!(msg.contains("no resolved gitdir")),
4903 other => panic!("expected EngineError::Mem, got {other:?}"),
4904 }
4905 }
4906
4907 #[test]
4908 fn worktree_for_folder_mount_returns_storage_path() {
4909 let tmp = TempDir::new().unwrap();
4910 let mem_dir = tmp.path().to_path_buf();
4911 let writer = FilesystemMemWriter::new(mem_dir.clone());
4912 let engine = Engine::from_mounts(vec![(
4913 folder_mount("specs", mem_dir.clone()),
4914 Box::new(writer) as Box<dyn MemBackend>,
4915 )])
4916 .unwrap();
4917 let worktree = engine.worktree_for("specs").unwrap();
4918 assert_eq!(worktree, mem_dir);
4919 }
4920
4921 #[test]
4922 fn worktree_for_unknown_mem_returns_unknown_mem() {
4923 let tmp = TempDir::new().unwrap();
4924 let mem_dir = tmp.path().to_path_buf();
4925 let writer = FilesystemMemWriter::new(mem_dir.clone());
4926 let engine = Engine::from_mounts(vec![(
4927 folder_mount("specs", mem_dir),
4928 Box::new(writer) as Box<dyn MemBackend>,
4929 )])
4930 .unwrap();
4931 let err = engine.worktree_for("missing").unwrap_err();
4932 assert!(matches!(err, EngineError::UnknownMem(v) if v == "missing"));
4933 }
4934
4935 #[test]
4936 fn worktree_for_archive_mount_returns_archive_backed_error() {
4937 let tmp = TempDir::new().unwrap();
4938 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4939 let engine = Engine::from_mounts(vec![(
4940 archive_mount("ext", archive_path.clone()),
4941 Box::new(ArchiveBackend::new(archive_path)) as Box<dyn MemBackend>,
4942 )])
4943 .unwrap();
4944 let err = engine.worktree_for("ext").unwrap_err();
4945 match err {
4946 EngineError::Mem(msg) => assert!(msg.contains("archive-backed")),
4947 other => panic!("expected EngineError::Mem, got {other:?}"),
4948 }
4949 }
4950
4951 #[test]
4952 fn mem_head_sha_for_folder_mount_is_none() {
4953 // Folder backend doesn't track a head; current_head() returns
4954 // Ok(None) at construction; mem_head_sha returns Ok(None).
4955 let tmp = TempDir::new().unwrap();
4956 let mem_dir = tmp.path().to_path_buf();
4957 let writer = FilesystemMemWriter::new(mem_dir.clone());
4958 let engine = Engine::from_mounts(vec![(
4959 folder_mount("specs", mem_dir),
4960 Box::new(writer) as Box<dyn MemBackend>,
4961 )])
4962 .unwrap();
4963 let head = engine.mem_head_sha("specs").unwrap();
4964 assert_eq!(head, None);
4965 }
4966
4967 #[test]
4968 fn mem_head_sha_unknown_mem_returns_unknown_mem() {
4969 let tmp = TempDir::new().unwrap();
4970 let mem_dir = tmp.path().to_path_buf();
4971 let writer = FilesystemMemWriter::new(mem_dir.clone());
4972 let engine = Engine::from_mounts(vec![(
4973 folder_mount("specs", mem_dir),
4974 Box::new(writer) as Box<dyn MemBackend>,
4975 )])
4976 .unwrap();
4977 let err = engine.mem_head_sha("missing").unwrap_err();
4978 assert!(matches!(err, EngineError::UnknownMem(v) if v == "missing"));
4979 }
4980
4981 #[test]
4982 fn capability_surfaces_per_mount() {
4983 let tmp = TempDir::new().unwrap();
4984 let mem_dir = tmp.path().to_path_buf();
4985 let writer = FilesystemMemWriter::new(mem_dir.clone());
4986 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4987
4988 let engine = Engine::from_mounts(vec![
4989 (
4990 folder_mount("writable", mem_dir),
4991 Box::new(writer) as Box<dyn MemBackend>,
4992 ),
4993 (
4994 archive_mount("read-only", archive_path.clone()),
4995 Box::new(ArchiveBackend::new(archive_path)),
4996 ),
4997 ])
4998 .unwrap();
4999
5000 assert_eq!(
5001 engine.capability("writable").unwrap(),
5002 MountCapability::Write
5003 );
5004 assert_eq!(
5005 engine.capability("read-only").unwrap(),
5006 MountCapability::ReadOnly
5007 );
5008 assert!(matches!(
5009 engine.capability("missing"),
5010 Err(EngineError::UnknownMem(_))
5011 ));
5012 }
5013
5014 #[test]
5015 fn read_provenance_routes_through_backend() {
5016 let tmp = TempDir::new().unwrap();
5017 let mem_dir = tmp.path().to_path_buf();
5018 let writer = FilesystemMemWriter::new(mem_dir.clone());
5019
5020 // Append a provenance record via the backend trait directly,
5021 // then read it back through the engine.
5022 let backend_handle: &dyn MemBackend = &writer;
5023 backend_handle
5024 .append_provenance(&Provenance::new(
5025 std::time::UNIX_EPOCH + std::time::Duration::from_secs(1_700_000_000),
5026 crate::ProvenanceKind::Create,
5027 Some("v:e".into()),
5028 crate::vcs::Actor::Cli,
5029 None,
5030 Some("first".into()),
5031 ))
5032 .unwrap();
5033
5034 let engine = Engine::from_mounts(vec![(
5035 folder_mount("specs", mem_dir),
5036 Box::new(writer) as Box<dyn MemBackend>,
5037 )])
5038 .unwrap();
5039
5040 let records = engine.read_provenance("specs", None).unwrap();
5041 assert_eq!(records.len(), 1);
5042 assert_eq!(records[0].kind, crate::ProvenanceKind::Create);
5043 assert_eq!(records[0].entity.as_deref(), Some("v:e"));
5044 assert_eq!(records[0].note.as_deref(), Some("first"));
5045 }
5046
5047 #[test]
5048 fn archive_mount_returns_sealed_indirectly_through_backend_layer() {
5049 // The engine doesn't yet expose mutation methods, but an
5050 // archive backend held on a Mount with ReadOnly capability is
5051 // still a `&dyn MemBackend` whose write methods return
5052 // Sealed. This test locks the trait routing — when the engine
5053 // gains write methods in a later session, capability gating +
5054 // backend Sealed errors must agree.
5055 let tmp = TempDir::new().unwrap();
5056 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
5057 let backend = ArchiveBackend::new(archive_path);
5058 match MemBackend::write_entity(&backend, Path::new("x.md"), b"x") {
5059 Err(BackendError::Sealed) => {}
5060 other => panic!("expected Sealed, got {other:?}"),
5061 }
5062 }
5063
5064 // ---- Read-side delegates ----------------------------------------
5065 //
5066 // These tests pin the surface that the MCP migration consumes
5067 // (stats, health, context, communities, search, list, orphans,
5068 // stubs, most_connected, missing_required_outgoing). They run
5069 // against a folder-mount engine with a small fixture of created
5070 // entities and one relate edge — enough to exercise both the
5071 // graph-query path and the cache-invalidation hooks.
5072
5073 /// Generation-keyed memos (flywheel W8/01). Four claims: repeated
5074 /// reads serve the memo; a REFUSED engine batch leaves the memo
5075 /// untouched (the refusal path calls no hook — pinned so it stays
5076 /// cheap); a memo computed from an INTERIM (mid-batch) state never
5077 /// survives a rollback as fresh — the store-carried generation is
5078 /// what lets the invalidation hook adjudicate that correctly; and
5079 /// a real mutation invalidates, with the recomputation identical
5080 /// to a from-scratch detection (the criterion-3 identity oracle
5081 /// for the mechanism).
5082 #[test]
5083 fn generation_keyed_memos_survive_rollback_and_track_mutations() {
5084 use indexmap::IndexMap;
5085
5086 let tmp = TempDir::new().unwrap();
5087 let mut engine = build_demo_engine(&tmp);
5088
5089 // 1. Repeated reads: cell filled once, generation stable.
5090 let _ = engine.communities();
5091 let memo_gen_before = engine.community_memo.get().expect("memo filled").0;
5092 let _ = engine.communities();
5093 assert_eq!(
5094 engine.community_memo.get().expect("still filled").0,
5095 memo_gen_before,
5096 "repeated reads must serve the memo"
5097 );
5098
5099 // 2. A REFUSED engine batch rolls the store back and leaves
5100 // the memo untouched — refusal stays recompute-free.
5101 let bare = |id: crate::EntityId| crate::engine::UpdateEntityArgs {
5102 anchors: Vec::new(),
5103 anchors_unset: Vec::new(),
5104 id,
5105 expected_hash: None,
5106 sections: IndexMap::new(),
5107 append_sections: IndexMap::new(),
5108 patch_sections: IndexMap::new(),
5109 sections_unset: Vec::new(),
5110 metadata: IndexMap::new(),
5111 metadata_unset: Vec::new(),
5112 declare_relations: Vec::new(),
5113 dry_run: false,
5114 relations_unset: Vec::new(),
5115 };
5116 let mut real = bare(crate::EntityId::new("specs", "source-one"));
5117 real.append_sections
5118 .insert("identity".to_string(), "appended line".to_string());
5119 let missing = bare(crate::EntityId::new("specs", "does-not-exist"));
5120 let (actor, client) = cli_actor();
5121 let result = engine
5122 .batch_update(
5123 vec![(real, None), (missing, None)],
5124 actor,
5125 Some(&client),
5126 false,
5127 )
5128 .expect("refused batch returns a report-all envelope");
5129 assert!(
5130 !result.applied,
5131 "the missing target refuses the whole batch"
5132 );
5133 assert_eq!(
5134 engine.community_memo.get().map(|m| m.0),
5135 Some(memo_gen_before),
5136 "a refused batch must leave the pre-batch memo standing"
5137 );
5138 assert_eq!(
5139 engine.store().generation(),
5140 memo_gen_before.store_generation,
5141 "rollback restored the store to the memo's generation"
5142 );
5143
5144 // 3. The dangerous direction: a memo computed from an INTERIM
5145 // state (simulated batch staging) must not survive the
5146 // rollback as fresh. The store-carried generation is what the
5147 // invalidation hook adjudicates with.
5148 let snapshot = engine.store.clone();
5149 let interim_id = crate::EntityId::new("specs", "interim-only");
5150 let mut interim = engine
5151 .store
5152 .get(&crate::EntityId::new("specs", "source-one"))
5153 .expect("demo entity present")
5154 .clone();
5155 interim.id = interim_id.clone();
5156 interim.title = "Interim Only".to_string();
5157 engine.store.upsert(interim_id, interim);
5158 engine.invalidate_communities();
5159 let _ = engine.communities();
5160 let interim_gen = engine.community_memo.get().expect("interim memo").0;
5161 assert_ne!(interim_gen, memo_gen_before);
5162 engine.store = snapshot; // rollback, generation restored with it
5163 engine.invalidate_communities();
5164 engine.invalidate_search_indexes();
5165 if let Some((g, _)) = engine.community_memo.get() {
5166 assert_ne!(
5167 *g, interim_gen,
5168 "a rolled-back interim state must never be served as fresh"
5169 );
5170 }
5171 assert!(
5172 !engine
5173 .communities()
5174 .entity_cluster_map
5175 .keys()
5176 .any(|id| id.contains("interim-only")),
5177 "the partition served after rollback reflects the restored store, not the interim one"
5178 );
5179
5180 // 4. A real mutation invalidates; the recomputation equals a
5181 // from-scratch detection and sees the new entity.
5182 let (actor, client) = cli_actor();
5183 engine
5184 .create_entity(
5185 empty_create_args("specs", "Fourth Entity"),
5186 actor,
5187 Some(&client),
5188 None,
5189 )
5190 .unwrap();
5191 assert!(
5192 engine
5193 .communities()
5194 .entity_cluster_map
5195 .keys()
5196 .any(|id| id.contains("fourth-entity")),
5197 "recomputed partition sees the new entity"
5198 );
5199 let fresh = {
5200 let schema = engine
5201 .schemas
5202 .iter()
5203 .min_by(|a, b| a.0.cmp(b.0))
5204 .map(|(_, s)| s.clone())
5205 .expect("demo engine has a schema");
5206 let schema_for_weights = schema.clone();
5207 crate::graph::community::detect_communities(
5208 engine.store(),
5209 schema.manifest.community.resolution,
5210 schema.manifest.community.seed,
5211 move |rel_type| {
5212 schema_for_weights
5213 .manifest
5214 .relationships
5215 .definitions
5216 .iter()
5217 .find(|d| d.name == rel_type)
5218 .map(|d| d.default_weight as f64)
5219 .unwrap_or(1.0)
5220 },
5221 )
5222 };
5223 assert_eq!(
5224 engine.communities().entity_cluster_map,
5225 fresh.entity_cluster_map,
5226 "memo must equal a from-scratch detection over the current store"
5227 );
5228 }
5229
5230 fn build_demo_engine(tmp: &TempDir) -> Engine {
5231 let mem_dir = tmp.path().to_path_buf();
5232 let writer = FilesystemMemWriter::new(mem_dir.clone());
5233 let mut engine = Engine::from_mounts(vec![(
5234 folder_mount("specs", mem_dir),
5235 Box::new(writer) as Box<dyn MemBackend>,
5236 )])
5237 .unwrap();
5238 let (actor, client) = cli_actor();
5239 let source = engine
5240 .create_entity(
5241 empty_create_args("specs", "Source One"),
5242 actor,
5243 Some(&client),
5244 None,
5245 )
5246 .unwrap();
5247 let target = engine
5248 .create_entity(
5249 empty_create_args("specs", "Target Two"),
5250 actor,
5251 Some(&client),
5252 None,
5253 )
5254 .unwrap();
5255 engine
5256 .create_entity(
5257 empty_create_args("specs", "Lonely Three"),
5258 actor,
5259 Some(&client),
5260 None,
5261 )
5262 .unwrap();
5263 engine
5264 .relate_entity(
5265 RelateEntityArgs {
5266 source: source.id.clone(),
5267 expected_hash: Some(source.content_hash.clone()),
5268 rel_type: "USES".to_string(),
5269 target: target.id.clone(),
5270 remove: false,
5271 description: None,
5272 dry_run: false,
5273 },
5274 actor,
5275 Some(&client),
5276 None,
5277 )
5278 .unwrap();
5279 engine
5280 }
5281
5282 #[test]
5283 fn status_reports_per_engine_counts() {
5284 let tmp = TempDir::new().unwrap();
5285 let engine = build_demo_engine(&tmp);
5286 let stats = engine.status();
5287 assert_eq!(stats.entity_count, 3);
5288 assert_eq!(stats.edge_count, 1);
5289 assert_eq!(stats.mem_count, 1);
5290 assert_eq!(stats.types_in_use, vec!["spec".to_string()]);
5291 assert_eq!(stats.edge_types.get("USES"), Some(&1));
5292 }
5293
5294 #[test]
5295 fn orphans_lists_unconnected_real_entities() {
5296 let tmp = TempDir::new().unwrap();
5297 let engine = build_demo_engine(&tmp);
5298 let orphans = engine.orphans();
5299 assert_eq!(orphans.len(), 1);
5300 assert_eq!(orphans[0].as_ref(), "specs--lonely-three");
5301 }
5302
5303 /// #49: the orphan/community headlines can be attributed per pinned
5304 /// schema. Single-mem here, so one bucket — but it proves the
5305 /// attribution keys by `schema_of(mem)` and that the per-schema
5306 /// counts sum to the raw total (which a health surface keeps verbatim).
5307 #[test]
5308 fn schema_breakdowns_attribute_to_mem_pin() {
5309 let tmp = TempDir::new().unwrap();
5310 let engine = build_demo_engine(&tmp);
5311
5312 let orphans = engine.orphans();
5313 let orphans_by_schema = engine.orphans_by_schema(&orphans);
5314 assert_eq!(
5315 orphans_by_schema.values().sum::<usize>(),
5316 orphans.len(),
5317 "per-schema orphan counts must sum to the raw total"
5318 );
5319 assert_eq!(orphans_by_schema.len(), 1, "one mem ⇒ one schema bucket");
5320 let (schema, count) = orphans_by_schema.iter().next().unwrap();
5321 assert!(!schema.is_empty(), "specs mem is pinned: {schema:?}");
5322 assert_eq!(*count, 1);
5323
5324 // communities_by_schema buckets the demo mem's clusters under the
5325 // same pin; with one schema, its values sum to the global count.
5326 let mems: Vec<String> = engine.mounts().iter().map(|m| m.mem.clone()).collect();
5327 let communities_by_schema = engine.communities_by_schema(&mems);
5328 assert_eq!(communities_by_schema.len(), 1);
5329 assert_eq!(
5330 communities_by_schema.values().sum::<usize>(),
5331 engine.communities().count,
5332 );
5333 }
5334
5335 #[test]
5336 fn stubs_lists_unresolved_link_targets() {
5337 let tmp = TempDir::new().unwrap();
5338 let mem_dir = tmp.path().to_path_buf();
5339 let writer = FilesystemMemWriter::new(mem_dir.clone());
5340 let mut engine = Engine::from_mounts(vec![(
5341 folder_mount("specs", mem_dir),
5342 Box::new(writer) as Box<dyn MemBackend>,
5343 )])
5344 .unwrap();
5345 let (actor, client) = cli_actor();
5346 let source = engine
5347 .create_entity(
5348 empty_create_args("specs", "Holder"),
5349 actor,
5350 Some(&client),
5351 None,
5352 )
5353 .unwrap();
5354 // Relate to a non-existent target — relate_entity creates a
5355 // stub for the target so the edge can land.
5356 engine
5357 .relate_entity(
5358 RelateEntityArgs {
5359 source: source.id.clone(),
5360 expected_hash: Some(source.content_hash.clone()),
5361 rel_type: "USES".to_string(),
5362 target: EntityId::new("specs", "ghost"),
5363 remove: false,
5364 description: None,
5365 dry_run: false,
5366 },
5367 actor,
5368 Some(&client),
5369 None,
5370 )
5371 .unwrap();
5372 let stubs = engine.stubs();
5373 assert!(
5374 stubs.iter().any(|(id, _)| id.as_ref() == "specs--ghost"),
5375 "expected ghost stub: {stubs:?}"
5376 );
5377 }
5378
5379 #[test]
5380 fn most_connected_orders_by_degree() {
5381 let tmp = TempDir::new().unwrap();
5382 let engine = build_demo_engine(&tmp);
5383 let top = engine.most_connected(5);
5384 assert_eq!(top.len(), 3);
5385 // Source and Target each have one edge; Lonely has zero.
5386 let zero_degree: Vec<_> = top
5387 .iter()
5388 .filter(|c| c.total == 0)
5389 .map(|c| c.id.as_ref().to_string())
5390 .collect();
5391 assert_eq!(zero_degree, vec!["specs--lonely-three".to_string()]);
5392 }
5393
5394 #[test]
5395 fn health_returns_per_engine_summary() {
5396 let tmp = TempDir::new().unwrap();
5397 let engine = build_demo_engine(&tmp);
5398 let health = engine.health();
5399 // `memstead_create` refuses on missing required sections, so
5400 // entities built through `empty_create_args` carry the
5401 // helper-seeded `identity` + `purpose` bodies and no longer
5402 // surface as missing-fields. Health remains the read-side
5403 // tolerance surface for legacy on-disk drift — covered by
5404 // the loader-tolerance tests that hand-craft pre-strict
5405 // markdown files.
5406 assert!(
5407 health
5408 .missing_fields
5409 .iter()
5410 .all(|r| r.id.as_ref() != "specs--source-one"),
5411 "post-strict-create fixture must not surface as missing-fields; got {:?}",
5412 health.missing_fields,
5413 );
5414 }
5415
5416 #[test]
5417 fn context_carries_neighbors_and_community() {
5418 let tmp = TempDir::new().unwrap();
5419 let engine = build_demo_engine(&tmp);
5420 let source_id = EntityId::new("specs", "source-one");
5421 let ctx = engine.context(&source_id).unwrap();
5422 assert_eq!(ctx.entity_id, source_id);
5423 assert_eq!(ctx.neighbors.len(), 1);
5424 assert_eq!(ctx.neighbors[0].relationship, "USES");
5425 assert!(matches!(ctx.neighbors[0].direction, Direction::Outgoing));
5426 }
5427
5428 #[test]
5429 fn communities_caches_louvain_until_invalidated() {
5430 let tmp = TempDir::new().unwrap();
5431 let mut engine = build_demo_engine(&tmp);
5432 // Population reflects the current store at first call.
5433 let entities_before = engine.communities().entity_cluster_map.len();
5434 // Cache hit — repeat call returns same data.
5435 assert_eq!(
5436 engine.communities().entity_cluster_map.len(),
5437 entities_before
5438 );
5439 // Mutation invalidates the cache; next call re-runs against
5440 // the post-mutation store and includes the new entity.
5441 let (actor, client) = cli_actor();
5442 engine
5443 .create_entity(
5444 empty_create_args("specs", "Disturber"),
5445 actor,
5446 Some(&client),
5447 None,
5448 )
5449 .unwrap();
5450 let entities_after = engine.communities().entity_cluster_map.len();
5451 assert_eq!(
5452 entities_after,
5453 entities_before + 1,
5454 "create_entity should have invalidated community cache and added the new entity"
5455 );
5456 }
5457
5458 #[test]
5459 fn list_filters_by_metadata_only() {
5460 let tmp = TempDir::new().unwrap();
5461 let engine = build_demo_engine(&tmp);
5462 let scope = SearchScope {
5463 entity_type: Some("spec".to_string()),
5464 ..Default::default()
5465 };
5466 let result = engine.list(&scope);
5467 // Three real spec entities created; stubs / non-spec types absent.
5468 assert_eq!(result.hits.len(), 3);
5469 }
5470
5471 #[test]
5472 fn list_applies_schema_declared_filter_on_non_default_schema_mem() {
5473 // A mem pinned to `planning` (non-default schema). The
5474 // `decision` type declares `status` with `filterable: equality`.
5475 // Pre-fix, filter dispatch consulted only the built-in default
5476 // schema via `type_by_name`, missed `status`, silently bypassed
5477 // the filter, and emitted the misleading "unknown filter key"
5478 // warning. Post-fix, the filter is honored and no warning fires.
5479 let tmp = TempDir::new().unwrap();
5480 let mem_dir = tmp.path().to_path_buf();
5481 let writer = FilesystemMemWriter::new(mem_dir.clone());
5482 let mount = Mount {
5483 mem: "planning".to_string(),
5484 schema: Some(memstead_schema::SchemaRef::new(
5485 "planning",
5486 semver::Version::new(0, 1, 0),
5487 )),
5488 storage: MountStorage::Folder { path: mem_dir },
5489 capability: MountCapability::Write,
5490 lifecycle: MountLifecycle::Eager,
5491 cross_linkable: true,
5492 migration_target: None,
5493 };
5494 let mut engine =
5495 Engine::from_mounts(vec![(mount, Box::new(writer) as Box<dyn MemBackend>)]).unwrap();
5496 let (actor, client) = cli_actor();
5497
5498 // Two decisions with different status values; required fields
5499 // (decision/context/consequences sections, decided_on, deciders)
5500 // get placeholder defaults — the test only cares about the
5501 // status field's filterability.
5502 for (title, status) in &[("Skip Postgres", "accepted"), ("Use SQLite", "proposed")] {
5503 let mut metadata = indexmap::IndexMap::new();
5504 metadata.insert("status".to_string(), status.to_string());
5505 metadata.insert("deciders".to_string(), "alice".to_string());
5506 metadata.insert("decided_on".to_string(), "2026-05-19".to_string());
5507 let args = crate::engine::CreateEntityArgs {
5508 anchors: Vec::new(),
5509 mem: "planning".to_string(),
5510 title: title.to_string(),
5511 entity_type: "decision".to_string(),
5512 sections: indexmap::IndexMap::from_iter([
5513 ("decision".to_string(), "We chose this.".to_string()),
5514 ("context".to_string(), "Single-user dev.".to_string()),
5515 ("consequences".to_string(), "Lose multi-writer.".to_string()),
5516 ]),
5517 metadata,
5518 relations: Vec::new(),
5519 dry_run: false,
5520 };
5521 engine
5522 .create_entity(args, actor, Some(&client), None)
5523 .unwrap();
5524 }
5525
5526 // Filter on the schema-declared filterable field.
5527 let scope = SearchScope {
5528 entity_type: Some("decision".to_string()),
5529 filters: std::collections::HashMap::from([(
5530 "status".to_string(),
5531 "accepted".to_string(),
5532 )]),
5533 ..Default::default()
5534 };
5535 let result = engine.list(&scope);
5536 assert_eq!(
5537 result.hits.len(),
5538 1,
5539 "filter on schema-declared field must select only matching entities"
5540 );
5541 assert_eq!(result.hits[0].title, "Skip Postgres");
5542 assert!(
5543 result.warnings.is_empty(),
5544 "no warning should fire when the filter is declared by the mem's pinned schema: {:?}",
5545 result.warnings
5546 );
5547 }
5548
5549 #[test]
5550 fn search_returns_results_against_built_index() {
5551 let tmp = TempDir::new().unwrap();
5552 let engine = build_demo_engine(&tmp);
5553 let scope = SearchScope {
5554 query: Some(crate::ops::Query {
5555 any: vec!["source".to_string()],
5556 ..Default::default()
5557 }),
5558 ..Default::default()
5559 };
5560 let result = engine.search(&scope).expect("native search returns Ok");
5561 assert!(result.total >= 1, "expected ≥1 hit for source: {result:?}");
5562 assert!(
5563 result
5564 .hits
5565 .iter()
5566 .any(|h| h.id.as_ref() == "specs--source-one"),
5567 "expected source-one in hits: {result:?}"
5568 );
5569 }
5570
5571 // ---- Engine::from_workspace_root (lean boot path) --------------
5572}