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