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