memstead_base/engine/mutation/mod.rs
1//! Engine mutation entrypoints — split per mutation kind.
2//!
3//! Each sub-module implements one mutation of `Engine`: `create`,
4//! `update` (with batch), `delete`, `relate`, `rename`. The shared
5//! helpers (`today_iso`, `make_stub`, `gc_orphan_stubs`,
6//! `lookup_title_and_type`, `unknown_type_error`) plus the typed
7//! constants `PATCH_OLD_NOT_FOUND_CONTENT_CAP` and
8//! `RELATIONSHIP_CYCLE_PATH_CAP` live here.
9
10use std::collections::HashMap;
11
12use indexmap::IndexMap;
13
14use crate::entity::{Entity, EntityId};
15use crate::store::Store;
16
17use super::EngineError;
18
19pub mod create;
20pub mod delete;
21pub mod mem_sweep;
22pub mod parse_recovery;
23pub mod relate;
24pub mod rename;
25pub mod update;
26
27/// Look up an entity's `(title, entity_type)` pair in `store`. Both
28/// `None` for missing-from-store ids — matches full's `title_for` /
29/// `type_for` lossy-lookup contract. Used by [`Engine::changes_since`]
30/// to enrich id-only envelopes the backend returned with metadata
31/// from the in-memory store.
32pub(super) fn lookup_title_and_type(
33 store: &Store,
34 id: &EntityId,
35) -> (Option<String>, Option<String>) {
36 match store.get(id) {
37 Some(e) => (Some(e.title.clone()), Some(e.entity_type.clone())),
38 None => (None, None),
39 }
40}
41
42/// Maximum byte length of the truncated `current_content` snapshot
43/// that [`EngineError::PatchOldNotFound`] carries. Keeps the wire
44/// envelope bounded for sections with large bodies. Mirrors full's
45/// `memstead_git_branch::PATCH_OLD_NOT_FOUND_CONTENT_CAP`.
46pub const PATCH_OLD_NOT_FOUND_CONTENT_CAP: usize = 500;
47
48/// Maximum number of entity IDs retained in
49/// [`EngineError::RelationshipCycle::existing_path`]. Keeps the cycle
50/// envelope bounded for pathologically long chains. Mirrors full's
51/// `memstead_git_branch::RELATIONSHIP_CYCLE_PATH_CAP`.
52pub const RELATIONSHIP_CYCLE_PATH_CAP: usize = 20;
53
54/// Build an [`EngineError::UnknownType`] populated with the schema's
55/// declared type names (sorted) and a fuzzy suggestion. Mirrors full's
56/// `UnknownEntityType` recovery payload so MCP envelopes carry the
57/// same `name` / `schema_ref` / `declared` / `suggestion` keys
58/// regardless of which engine served the call.
59/// Render an entity for a write, refusing the one state the generator would
60/// silently make permanent.
61///
62/// **This is the mutation path's only way to bytes.** Calling
63/// `generate_markdown` directly from a mutation verb bypasses the guard, and
64/// the first version of this fix did exactly that: the check lived in
65/// `update_entity`, so `memstead relate` against the same entity walked
66/// straight past it and froze the absorption anyway (04/02, criterion 5,
67/// found by the plan's grade). A guard a new verb can miss by following the
68/// local idiom is not a guard; making the guarded call BE the idiom is.
69///
70/// The condition: a section whose stored body ends inside an unterminated
71/// fence has already absorbed every section after it, and the generator
72/// appends its closer AFTER those bytes. One write seals them inside a
73/// legitimately fenced block where nothing can tell them from prose the
74/// author meant to fence. It is refused rather than warned about because it
75/// is unrecoverable once it lands.
76///
77/// The way out needs no special case: a caller who replaces the absorbing
78/// section hands us an entity whose fence is closed, so the guard simply
79/// does not fire.
80pub(crate) fn render_for_write(
81 entity: &Entity,
82 type_def: &memstead_schema::TypeDefinition,
83) -> Result<String, EngineError> {
84 for (key, value) in &entity.sections {
85 if let Some(fence) = crate::markdown::closing_fence_if_unterminated(value.trim()) {
86 return Err(EngineError::UnterminatedFenceInStoredBody {
87 id: entity.id.to_string(),
88 section: key.clone(),
89 fence,
90 swallowed: crate::ops::integrity::swallowed_declared_sections(value, type_def),
91 });
92 }
93 }
94 Ok(crate::entity::generator::generate_markdown(
95 entity, type_def,
96 ))
97}
98
99pub(crate) fn unknown_type_error(schema: &memstead_schema::Schema, attempted: &str) -> EngineError {
100 let mut declared: Vec<String> = schema.types.keys().cloned().collect();
101 declared.sort();
102 let (sname, sver) = schema.id();
103 EngineError::UnknownType {
104 name: attempted.to_string(),
105 schema_ref: format!("{sname}@{sver}"),
106 declared,
107 suggestion: schema.suggest_type(attempted),
108 }
109}
110
111/// The lowercase wire string for a [`crate::pipeline::MediumType`] — the
112/// value the `INVALID_ANCHOR` recovery detail carries so an agent sees
113/// which medium's namespace rejected the grain. Matches the enum's
114/// `#[serde(rename_all = "lowercase")]` form.
115pub(crate) fn medium_type_wire(t: crate::pipeline::MediumType) -> &'static str {
116 use crate::pipeline::MediumType::*;
117 match t {
118 Codebase => "codebase",
119 Filesystem => "filesystem",
120 Graph => "graph",
121 Git => "git",
122 Web => "web",
123 }
124}
125
126impl super::Engine {
127 /// Resolve the single-source anchor-namespace context for `mem`, when
128 /// unambiguous. An `anchors[]` element carries no source name, so the
129 /// grain/namespace refusal ([`crate::anchor::AnchorValidationError::GrainNamespaceUnsupported`])
130 /// can only be fired deterministically when the mem's bindings declare
131 /// exactly one inline source; with zero or several the namespace check
132 /// is skipped (the vocabulary + hash-semantics rules still apply).
133 /// Returns the `(medium_type_wire, anchor_namespace)` pair the
134 /// validator consumes — the medium *half* of the lone source.
135 pub(crate) fn resolve_anchor_medium(&self, mem: &str) -> Option<(String, &'static str)> {
136 let mut sources = self
137 .pipeline_configs()
138 .bindings
139 .iter()
140 .filter(|r| r.mem == mem)
141 .flat_map(|r| r.config.sources.iter());
142 let first = sources.next()?;
143 if sources.next().is_some() {
144 // Ambiguous — the anchor does not name which source it targets;
145 // skip the namespace refinement rather than guess.
146 return None;
147 }
148 let caps = crate::binding::medium_capabilities(first.medium_type);
149 Some((
150 medium_type_wire(first.medium_type).to_string(),
151 caps.anchor_namespace,
152 ))
153 }
154
155 /// Validate the permissive `anchors[]` inputs for a mutation against
156 /// `mem`'s medium context into strict [`crate::anchor::Anchor`]s, or
157 /// refuse the whole mutation with a typed
158 /// [`EngineError::InvalidAnchor`]. Empty input yields an empty vec (no
159 /// sidecar write); a single malformed element aborts before any state
160 /// change so the entity is never written.
161 pub(crate) fn validate_anchor_inputs(
162 &self,
163 mem: &str,
164 inputs: &[crate::anchor::AnchorInput],
165 ) -> Result<Vec<crate::anchor::Anchor>, EngineError> {
166 if inputs.is_empty() {
167 return Ok(Vec::new());
168 }
169 let medium = self.resolve_anchor_medium(mem);
170 let medium_ref = medium.as_ref().map(|(t, ns)| (t.as_str(), *ns));
171 let mut anchors: Vec<crate::anchor::Anchor> = inputs
172 .iter()
173 .map(|i| i.validate(medium_ref).map_err(EngineError::from))
174 .collect::<Result<_, _>>()?;
175
176 // One payload, one row per triple (consistency-sweep 03/03,
177 // criterion 9). `(artifact, grain, class)` is the sidecar's merge
178 // identity, so a payload naming it twice used to collapse to the last
179 // occurrence and the caller was never told an anchor it wrote had
180 // vanished. The unit is THIS payload: the same triple arriving in a
181 // later call still replaces the stored row, which is what the
182 // carry-forward depends on.
183 {
184 let mut seen: std::collections::HashSet<(&str, &str, &str)> =
185 std::collections::HashSet::new();
186 for a in &anchors {
187 let key = (a.artifact.as_str(), a.grain.as_wire(), a.class.as_wire());
188 if !seen.insert(key) {
189 return Err(EngineError::from(
190 crate::anchor::AnchorValidationError::DuplicateAnchorTriple {
191 artifact: a.artifact.clone(),
192 grain: a.grain.as_wire(),
193 class: a.class.as_wire(),
194 },
195 ));
196 }
197 }
198 }
199
200 // Supplied `content` under the source's preparation: the context-free
201 // validator hashed the bytes under the default canonicalization; the
202 // seam knows the anchor's source and re-hashes through the registry's
203 // rule for that source (touchpoint A at write time), so the recorded
204 // hash is the one a later observation computes.
205 if inputs.iter().any(|i| i.content.is_some()) {
206 let joins = self.anchor_source_roots(mem);
207 for (input, anchor) in inputs.iter().zip(anchors.iter_mut()) {
208 let (Some(content), Some(source)) = (&input.content, &anchor.source) else {
209 continue;
210 };
211 let Some(join) = joins.get(source) else {
212 continue;
213 };
214 if join.preparation.is_none() {
215 continue;
216 }
217 match crate::preparation::path_prepared_hash(
218 join.preparation.as_deref(),
219 &anchor.artifact,
220 anchor.grain,
221 content.as_bytes(),
222 ) {
223 crate::preparation::PathPrepared::Hash(h) => anchor.hash = Some(h),
224 crate::preparation::PathPrepared::NoHash => {}
225 crate::preparation::PathPrepared::UnitAbsent => {
226 return Err(EngineError::from(
227 crate::anchor::AnchorValidationError::UnitAbsentFromContent {
228 artifact: anchor.artifact.clone(),
229 },
230 ));
231 }
232 }
233 }
234 }
235
236 // Source-vs-binding check: when an anchor names BOTH a producing
237 // binding and a source, and that binding hash still resolves in
238 // this workspace (reverse lookup — any later binding edit moves
239 // the hash and drops earlier anchors into the accept-any-name
240 // branch for good), the source must be one of the binding's
241 // declared names. The bindings load is skipped entirely unless
242 // some input needs it, and a missing workspace root or an
243 // unloadable store degrades to accept-any-name — validation
244 // must never require the binding to resolve.
245 if anchors
246 .iter()
247 .any(|a| a.binding.is_some() && a.source.is_some())
248 && let Some(root) = self.workspace_root()
249 && let Ok(configs) = crate::pipeline_store::load_pipeline_configs(root)
250 {
251 for a in &anchors {
252 let (Some(binding_hash), Some(source)) = (&a.binding, &a.source) else {
253 continue;
254 };
255 let Some(record) = configs
256 .bindings
257 .iter()
258 .find(|r| crate::binding::hash_binding(&r.config) == *binding_hash)
259 else {
260 continue; // unresolvable binding: accept any non-empty name
261 };
262 let declared: Vec<String> = record
263 .config
264 .sources
265 .iter()
266 .map(|s| s.name.clone())
267 .collect();
268 if !declared.iter().any(|n| n == source) {
269 return Err(EngineError::from(
270 crate::anchor::AnchorValidationError::SourceNotDeclared {
271 got: source.clone(),
272 declared,
273 },
274 ));
275 }
276 }
277 }
278
279 // Write time resolves or refuses (decision 26, plan 03a): a
280 // path-grain anchor whose artifact resolves under NO candidate join
281 // is a silently dead reference — refuse now, while the write moment
282 // still has the binding context to say what the right dialect is.
283 // Candidates in decision-29 priority: the source-join (the anchor's
284 // `source` name resolved to its declared pointer) first, then the
285 // workspace-relative form. An anchor naming a source the LOADED
286 // roster does not declare (a typo, a renamed binding) gets no join
287 // candidate — its workspace-relative form must resolve or the write
288 // refuses, because resolution will make the same roster lookup and
289 // orphan it at birth. The gate skips entirely only when it cannot
290 // know: no workspace root, or a pipeline store whose LOAD fails as a
291 // whole — validation never requires the binding store to be
292 // readable. (A corrupted individual binding file does not fail the
293 // load; it yields a reduced roster, and the gate then fail-closes on
294 // paths that resolve under no surviving candidate.)
295 if let Some(root) = self.workspace_root()
296 && crate::pipeline_store::load_pipeline_configs(root).is_ok()
297 {
298 let source_roots = self.anchor_source_roots(mem);
299 for a in &anchors {
300 match a.grain {
301 crate::anchor::AnchorGrain::Span
302 | crate::anchor::AnchorGrain::File
303 | crate::anchor::AnchorGrain::Tree => {}
304 crate::anchor::AnchorGrain::Url | crate::anchor::AnchorGrain::Entity => {
305 continue;
306 }
307 }
308 let base = crate::engine::query::anchor_base_path(&a.artifact);
309 // The shared decision-29 candidate rule — the same set
310 // resolution will later read, so the gate cannot accept a
311 // path resolution would read differently.
312 let candidates: Vec<String> = match a
313 .source
314 .as_ref()
315 .and_then(|source| source_roots.get(source))
316 {
317 Some(join) => crate::engine::query::artifact_candidates(&join.pointer, base),
318 None => vec![base.to_string()],
319 };
320 if !candidates.iter().any(|c| root.join(c).exists()) {
321 return Err(EngineError::from(
322 crate::anchor::AnchorValidationError::ArtifactUnresolvable {
323 artifact: a.artifact.clone(),
324 candidates,
325 },
326 ));
327 }
328 }
329 }
330
331 Ok(anchors)
332 }
333
334 /// Validate an update's `anchors_unset[]` payload up front — a
335 /// malformed selector (missing artifact, unknown grain/class wire
336 /// string) refuses the whole mutation with the same typed
337 /// `INVALID_ANCHOR` envelope as a malformed `anchors[]` element.
338 /// Empty payload → empty vec.
339 pub(crate) fn validate_anchor_unsets(
340 inputs: &[crate::anchor::AnchorUnsetInput],
341 ) -> Result<Vec<crate::anchor::AnchorUnset>, EngineError> {
342 inputs
343 .iter()
344 .map(|i| i.validate().map_err(EngineError::from))
345 .collect()
346 }
347
348 /// Record verify-observed prepared-content hashes onto **hash-less
349 /// hash-bearing** anchors in `mem_name`'s anchors sidecar — the
350 /// measurement-bookkeeping backfill the verify pass hands over via
351 /// [`crate::ingest::VerifyOutcome::hash_backfill`].
352 ///
353 /// This mutates **only** the engine-owned sidecar
354 /// ([`crate::anchor::ANCHOR_SIDECAR_PATH`]): no entity content, no
355 /// section, no `_hash` is touched — an anchor-only commit yields zero
356 /// entity deltas by construction. Guards enforced at this write seam,
357 /// not left to callers:
358 ///
359 /// - only a hash-bearing class (`anchored` / `derived`) may gain a hash —
360 /// an `authored` / `informed-by` anchor is never written, whatever the
361 /// caller observed;
362 /// - an anchor that already carries a hash is never overwritten — the
363 /// recorded hash is the drift baseline, so the backfill is idempotent
364 /// (a second identical call stages nothing and produces no commit).
365 ///
366 /// Returns how many anchors gained a hash. Zero writes ⇒ no commit.
367 pub fn record_anchor_observed_hashes(
368 &mut self,
369 mem_name: &str,
370 observed: &[crate::anchor::ObservedArtifactHash],
371 note: Option<&str>,
372 ) -> Result<usize, EngineError> {
373 if observed.is_empty() {
374 return Ok(0);
375 }
376 let mount_idx = self
377 .mounts
378 .iter()
379 .position(|m| m.mount.mem == mem_name)
380 .ok_or_else(|| self.unknown_mem_error(mem_name))?;
381 if self.mounts[mount_idx].mount.capability != crate::workspace::MountCapability::Write {
382 return Err(EngineError::ReadOnlyMount(mem_name.to_string()));
383 }
384 // Same posture as every other commit-producing write: probe for
385 // sibling-engine drift so the sidecar merge runs against current truth.
386 let _warnings = self.reload_if_stale(Some(mem_name));
387
388 let backend = self.mounts[mount_idx].backend.as_ref();
389 let mut sidecar = read_sidecar(backend)?;
390 let mut written = 0usize;
391 for obs in observed {
392 let Some(anchors) = sidecar.entities.get_mut(&obs.entity) else {
393 continue;
394 };
395 for a in anchors {
396 if a.class.is_hash_bearing() && a.hash.is_none() && a.artifact == obs.artifact {
397 a.hash = Some(obs.hash.clone());
398 // Stamp the origin (consistency-sweep 03/03, criterion 8):
399 // this baseline is the engine's inference from what it
400 // observed, not something an author pinned, and a reader
401 // comparing drift needs to know which.
402 a.hash_source = Some(crate::anchor::AnchorHashSource::Backfill);
403 written += 1;
404 }
405 }
406 }
407 if written == 0 {
408 return Ok(0);
409 }
410 backend.write_anchors_sidecar(&sidecar.to_bytes())?;
411 let ctx = crate::vcs::CommitContext {
412 actor: crate::vcs::Actor::Agent,
413 client: None,
414 tool: Some("record_anchor_observed_hashes"),
415 note: note.map(String::from),
416 role: self.current_role,
417 identity: self.current_identity.clone(),
418 logical_operation_id: None,
419 entity_ids: None,
420 };
421 let write_id = backend.commit(
422 &format!("memstead: anchor-hash backfill ({written} anchor(s))"),
423 &ctx,
424 )?;
425 self.record_self_write(mount_idx, &write_id);
426 // Anchor-hash backfill returns a count, not an agent-facing response,
427 // so an intervention report has nowhere to ride. Discarded knowingly:
428 // the merge itself still happened, so nothing was lost — only the
429 // notice that someone else had written (04/03, criterion 3).
430 let _intervention_has_no_channel_here = self.stamp_mutation_versions(mount_idx);
431 Ok(written)
432 }
433
434 /// Record on the mem which engine version and which resolved
435 /// schema performed the mutation that just committed
436 /// (`MemConfig.mutation_stamp`). Called by every mutation verb
437 /// after `record_self_write` — one shared implementation, per the
438 /// one-guard-on-all-write-paths principle — and deliberately NOT
439 /// by `apply_external_commit`: a replayed sibling commit was
440 /// stamped by the engine that performed it, and this engine must
441 /// not claim it.
442 ///
443 /// Write-cheap by construction: the config write happens only when
444 /// the stamp VALUE changes (a binary upgrade or a schema repin) —
445 /// steady-state mutations compare and return. The write is
446 /// best-effort: a failed stamp never fails the mutation that
447 /// preceded it. On git-branch backends the config rides the
448 /// `__MEMSTEAD` ref, so a stamp write never moves the mem branch
449 /// head — so a git-branch mem's change cursor (the `head` a prior
450 /// `changes_since` returned) stays valid across a stamp write.
451 pub(crate) fn stamp_mutation_versions(
452 &mut self,
453 mount_idx: usize,
454 ) -> Vec<crate::ops::WarningHint> {
455 let Some(state) = self.mounts.get(mount_idx) else {
456 return Vec::new();
457 };
458 let mem = state.mount.mem.clone();
459 let Some(schema) = self.schemas.get(&mem) else {
460 return Vec::new();
461 };
462 let (name, version) = schema.id();
463 // Full build version (semver + git build sha when present) so
464 // a rebuild between mutations is a recordable — and hence
465 // skew-detectable — event even between releases.
466 let stamp = memstead_schema::MutationStamp {
467 engine_version: crate::build_info::full_version().to_string(),
468 schema: format!("{name}@{version}"),
469 };
470 // A mem with no loaded config has nowhere to carry the stamp;
471 // skip silently (in-memory sketches, minimal fixtures).
472 let Some(config) = self
473 .mounts
474 .get(mount_idx)
475 .and_then(|s| s.mem_config.as_ref())
476 else {
477 return Vec::new();
478 };
479 // Skew at WRITE time, and before the restamp below erases the evidence
480 // (04/04, criterion 9). Boot-only detection meant a long-lived server
481 // that started under one binary and was written to by another never
482 // said so, and the very write that would have revealed it wrote the
483 // stamp that hid it. The write is never refused (criterion 10): a
484 // deliberate downgrade is the operator's business.
485 let mut warnings = Vec::new();
486 if let Some(prior) = config.mutation_stamp.as_ref()
487 && let Some(direction) = crate::build_info::skew_direction(
488 &prior.engine_version,
489 crate::build_info::full_version(),
490 )
491 {
492 warnings.push(crate::ops::WarningHint::EngineVersionSkew {
493 mem: mem.clone(),
494 stamped_engine: prior.engine_version.clone(),
495 running_engine: crate::build_info::full_version().to_string(),
496 stamped_schema: prior.schema.clone(),
497 direction,
498 });
499 }
500 if config.mutation_stamp.as_ref() == Some(&stamp) {
501 return warnings;
502 }
503 // Through the shared writer like the seven lifecycle setters
504 // (04/03, criterion 7). This one is why the damage looked
505 // spontaneous: it rides ordinary create/update/relate/rename/delete,
506 // so an operator saw a config field vanish during an innocuous entity
507 // write with no lifecycle call in sight. It stays exactly as dormant
508 // as before, because the equality guard above still decides whether
509 // to write at all; what changed is only what it writes over.
510 // The intervention rides the ENTITY mutation's own response: this
511 // writer has no response of its own, and the operator who sees a
512 // config field move during an innocuous entity write is owed the
513 // reason there (04/03, criterion 3, found by the plan's grade —
514 // an earlier draft discarded this with `let _`).
515 match self.write_mem_config_merged(
516 mount_idx,
517 &mem,
518 Some("engine version stamp"),
519 &move |c: &mut memstead_schema::config::MemConfig| {
520 c.mutation_stamp = Some(stamp.clone());
521 },
522 ) {
523 Ok((_, intervened)) if !intervened.is_empty() => {
524 warnings.push(crate::ops::WarningHint::ConfigWriteIntervened {
525 mem,
526 fields: intervened,
527 });
528 warnings
529 }
530 _ => warnings,
531 }
532 }
533}
534
535/// Stage a write of `entity_id`'s anchors into the mem's anchors sidecar
536/// through `backend`, merged over the existing sidecar at BOTH levels —
537/// other entities' rows survive (document level), and within the entity's
538/// own row `unsets` apply first, then each incoming anchor replaces the
539/// existing anchor with the same `(artifact, grain, class)` triple or
540/// appends ([`crate::anchor::AnchorSidecar::merge`]). Writing never
541/// removes an anchor the call did not name in `unsets`. The write is
542/// buffered into the SAME pending op set the entity write used — so the
543/// next [`crate::backend::MemBackend::commit`] carries entity + anchors
544/// as one atomic commit. Reads honour pending-buffer precedence, so
545/// successive stages within one transaction compose.
546/// Stage a mutation of the engine-owned derivations sidecar
547/// (agent-trust plan 12) so it rides the SAME commit as the edge
548/// write that produced it — the anchors-sidecar atomicity precedent.
549/// The sidecar travels through the backend's normal entity-path
550/// read/write under `.memstead/`, which every backend filters from
551/// entity listings and every archive/export path carries as-is.
552pub(crate) fn stage_derivation_sidecar(
553 backend: &dyn crate::backend::MemBackend,
554 mutate: impl FnOnce(&mut crate::derivation::DerivationSidecar),
555) -> Result<(), EngineError> {
556 let path = std::path::Path::new(crate::derivation::DERIVATION_SIDECAR_PATH);
557 let mut sidecar = match backend.read_entity(path)? {
558 Some(bytes) => crate::derivation::DerivationSidecar::from_bytes(&bytes).map_err(|e| {
559 EngineError::Backend(crate::backend::BackendError::Other(format!(
560 "derivations sidecar parse: {e}"
561 )))
562 })?,
563 None => crate::derivation::DerivationSidecar::default(),
564 };
565 mutate(&mut sidecar);
566 backend.write_entity(path, &sidecar.to_bytes())?;
567 Ok(())
568}
569
570/// True when `schema` declares `rel_type` as a derivation
571/// (`derivation: true` on the relationship definition) — the
572/// predicate every write path shares, so baseline recording cannot
573/// fork per verb.
574pub(crate) fn rel_type_declares_derivation(
575 schema: &memstead_schema::Schema,
576 rel_type: &str,
577) -> bool {
578 schema
579 .manifest
580 .relationships
581 .definitions
582 .iter()
583 .any(|d| d.name == rel_type && d.derivation)
584}
585
586pub(crate) fn stage_anchors_sidecar(
587 backend: &dyn crate::backend::MemBackend,
588 entity_id: &EntityId,
589 unsets: &[crate::anchor::AnchorUnset],
590 anchors: Vec<crate::anchor::Anchor>,
591) -> Result<(), EngineError> {
592 let mut sidecar = match backend.read_anchors_sidecar()? {
593 Some(bytes) => crate::anchor::AnchorSidecar::from_bytes(&bytes).map_err(|e| {
594 EngineError::Backend(crate::backend::BackendError::Other(format!(
595 "anchors sidecar parse: {e}"
596 )))
597 })?,
598 None => crate::anchor::AnchorSidecar::default(),
599 };
600 sidecar.merge(entity_id.as_ref(), unsets, anchors);
601 backend.write_anchors_sidecar(&sidecar.to_bytes())?;
602 Ok(())
603}
604
605/// Load the mem's anchors sidecar through `backend`, or the empty
606/// document when none exists yet. Shared by the delete / rename legs
607/// which must decide whether the entity actually has anchor rows before
608/// staging a sidecar write (so an entity with none stays byte-identical
609/// to a pre-anchor mutation).
610fn read_sidecar(
611 backend: &dyn crate::backend::MemBackend,
612) -> Result<crate::anchor::AnchorSidecar, EngineError> {
613 match backend.read_anchors_sidecar()? {
614 Some(bytes) => crate::anchor::AnchorSidecar::from_bytes(&bytes).map_err(|e| {
615 EngineError::Backend(crate::backend::BackendError::Other(format!(
616 "anchors sidecar parse: {e}"
617 )))
618 }),
619 None => Ok(crate::anchor::AnchorSidecar::default()),
620 }
621}
622
623/// Stage removal of `entity_id`'s anchor row into the same commit as an
624/// entity delete — a no-op (no sidecar write, so byte-identical to today)
625/// when the entity carries no anchors. Returns whether a write was staged.
626pub(crate) fn stage_anchors_removal(
627 backend: &dyn crate::backend::MemBackend,
628 entity_id: &EntityId,
629) -> Result<bool, EngineError> {
630 let mut sidecar = read_sidecar(backend)?;
631 if sidecar.get(entity_id.as_ref()).is_empty() {
632 return Ok(false);
633 }
634 sidecar.remove(entity_id.as_ref());
635 backend.write_anchors_sidecar(&sidecar.to_bytes())?;
636 Ok(true)
637}
638
639/// Stage a move of `from`'s anchor row to `to` into the same commit as an
640/// entity rename — leaving zero rows under the old id. A no-op (byte-
641/// identical to today) when the renamed entity carries no anchors. Returns
642/// whether a write was staged.
643pub(crate) fn stage_anchors_rename(
644 backend: &dyn crate::backend::MemBackend,
645 from: &EntityId,
646 to: &EntityId,
647) -> Result<bool, EngineError> {
648 let mut sidecar = read_sidecar(backend)?;
649 if sidecar.get(from.as_ref()).is_empty() {
650 return Ok(false);
651 }
652 sidecar.rename(from.as_ref(), to.as_ref());
653 backend.write_anchors_sidecar(&sidecar.to_bytes())?;
654 Ok(true)
655}
656
657// A `today_iso()` wall-clock convenience used to live here, for tests
658// comparing an auto-stamp against "roughly now". It is deliberately
659// gone: the stamp is second-resolution, so every such comparison races
660// the clock between the mutation and the assertion, and one of them
661// duly failed on a suite run that straddled midnight. Tests that need
662// a stamped value pin `Engine::set_mutation_clock` and derive the
663// expected string from the same instant via [`iso_from_system_time`].
664
665/// An instant as a full ISO-8601 datetime string `YYYY-MM-DDTHH:MM:SSZ`
666/// (UTC). Used by mutation paths that auto-stamp metadata fields
667/// (e.g. `last_modified` on update, `created_date` on create).
668///
669/// This is second-resolution (rather than
670/// date-only `YYYY-MM-DD`) so intra-day
671/// updates produce distinguishable timestamps and drift / staleness
672/// queries become per-update aware. The strict-mode date validator
673/// already accepts both forms (`^\d{4}-\d{2}-\d{2}(T\d{2}:\d{2}:\d{2}Z)?$`)
674/// so existing entities written with the date-only form continue to
675/// load; new writes carry the wider form.
676///
677/// Pure function: no allocation outside the `format!` invocation,
678/// no error path (the fallback to UNIX epoch on an instant before
679/// the epoch is acceptable for a best-effort timestamp).
680/// Howard-Hinnant civil-from-days for the date half; trivial modular
681/// arithmetic for the time half.
682pub(super) fn iso_from_system_time(t: std::time::SystemTime) -> String {
683 let now = t.duration_since(std::time::UNIX_EPOCH).unwrap_or_default();
684 let secs = now.as_secs();
685 let days = secs / 86400;
686 let secs_of_day = secs % 86400;
687 let hh = secs_of_day / 3600;
688 let mm = (secs_of_day % 3600) / 60;
689 let ss = secs_of_day % 60;
690 let z = days + 719468;
691 let era = z / 146097;
692 let doe = z - era * 146097;
693 let yoe = (doe - doe / 1460 + doe / 36524 - doe / 146096) / 365;
694 let y = yoe + era * 400;
695 let doy = doe - (365 * yoe + yoe / 4 - yoe / 100);
696 let mp = (5 * doy + 2) / 153;
697 let d = doy - (153 * mp + 2) / 5 + 1;
698 let m = if mp < 10 { mp + 3 } else { mp - 9 };
699 let y = if m <= 2 { y + 1 } else { y };
700 format!("{y:04}-{m:02}-{d:02}T{hh:02}:{mm:02}:{ss:02}Z")
701}
702
703/// Sweep stubs whose last incoming edge has just disappeared. Returns
704/// the dropped ids so callers can surface them to the agent (e.g. via
705/// [`DeleteEntityOutcome::orphan_stubs_removed`]).
706///
707/// Stubs are auto-created when a relate names an absent target — a
708/// "promise" that a real entity will land there later (see
709/// [`make_stub`]). When the last referrer drops its edge or is itself
710/// deleted, the promise has no holder and becomes pure bloat. Only
711/// stubs are eligible — real entities never count as orphans via this
712/// path.
713pub(super) fn gc_orphan_stubs(store: &mut Store) -> Vec<EntityId> {
714 let stub_ids: Vec<EntityId> = store
715 .all_entities()
716 .filter(|e| e.stub)
717 .map(|e| e.id.clone())
718 .collect();
719 gc_orphan_stubs_among(store, &stub_ids)
720}
721
722/// Scoped orphan-stub sweep: GC only the stubs *among `candidates`*
723/// whose last incoming edge has just disappeared, returning the dropped
724/// ids. This is the single home of the orphan-stub predicate (`stub &&
725/// no incoming`) — the three write paths that can sever a stub's last
726/// referrer all funnel through here so they cannot drift:
727/// [`gc_orphan_stubs`] (delete's full-store sweep) supplies every stub
728/// id; the `memstead_relate(remove)` path supplies the just-severed target;
729/// the `memstead_update` alias-resync path supplies the entity's
730/// pre-mutation body-link targets (the only edges that commit could
731/// have dropped). Scoping to a candidate set rather than walking the
732/// whole store keeps each path from GC'ing pre-existing orphans that
733/// aren't its responsibility. Candidates are de-duplicated; a candidate
734/// that is absent, not a stub, or still has a referrer is left
735/// untouched.
736pub(super) fn gc_orphan_stubs_among<'a>(
737 store: &mut Store,
738 candidates: impl IntoIterator<Item = &'a EntityId>,
739) -> Vec<EntityId> {
740 let mut removed: Vec<EntityId> = Vec::new();
741 let mut seen: std::collections::HashSet<&EntityId> = std::collections::HashSet::new();
742 for id in candidates {
743 if !seen.insert(id) {
744 continue;
745 }
746 if store.get(id).is_some_and(|e| e.stub) && store.incoming(id).is_empty() {
747 store.remove(id);
748 removed.push(id.clone());
749 }
750 }
751 removed
752}
753
754/// Shared target-id grammar validator. The wiki-link grammar gate
755/// runs on every relation-authoring path (`memstead_relate`,
756/// `memstead_create.relations[]`, future inline-relation surfaces) so a
757/// malformed target id (e.g. `bad@chars$here`) cannot land an
758/// auto-stub at the literal id — that stub would later fail every
759/// wiki-link parse that referenced it. Pre-Item-02 the gate lived
760/// only on `memstead_relate`; the create path admitted the same input
761/// silently.
762pub(super) fn validate_relation_target_grammar(target: &EntityId) -> Result<(), EngineError> {
763 if let Err(reason) = crate::entity::id::validate_mem_name_grammar(target.mem()) {
764 return Err(EngineError::InvalidEntityId {
765 id: target.to_string(),
766 reason,
767 });
768 }
769 if let Err(reason) = crate::entity::id::validate_id_path_grammar(target.path()) {
770 return Err(EngineError::InvalidEntityId {
771 id: target.to_string(),
772 reason,
773 });
774 }
775 Ok(())
776}
777
778/// Auto-stamp `auto_timestamp` metadata fields on an entity that's
779/// about to be re-written. Extracted from the update-path hot loop so
780/// the relate-path (add and remove) and the rename-path (the renaming
781/// entity plus every referrer the rewrite cascade touched) can
782/// invoke the same engine-driven stamp.
783///
784/// Walks the type's metadata-field declarations; any field flagged
785/// `auto_timestamp: true` (the default schema declares this on
786/// `last_modified`) is set to the supplied `today` ISO string. The
787/// helper is a no-op on schemas that declare no auto-timestamp
788/// fields. Callers pre-compute `today` via [`today_iso`] so a single
789/// mutation that touches multiple entities (rename's referrer rewrite
790/// cascade) stamps them all with the same value.
791pub(super) fn auto_stamp_timestamps(
792 entity: &mut Entity,
793 type_def: &memstead_schema::TypeDefinition,
794 today: &str,
795) {
796 for field_def in &type_def.metadata_fields {
797 if field_def.auto_timestamp {
798 entity.metadata.insert(
799 field_def.key.clone(),
800 crate::entity::MetadataValue::String(today.to_string()),
801 );
802 }
803 }
804}
805
806/// Build a stub [`Entity`] for an unresolved relate target. Callers
807/// declare the stub's origin via [`crate::entity::StubKind`] —
808/// `ForwardReference` for `memstead_relate` to an absent target,
809/// `Residual { since_commit, readonly_referrers }` for the
810/// delete/rename demote path. The kind persists for the engine
811/// instance's lifetime; a reload reduces every stub to `LoadTime`
812/// — the kind is annotation, not state.
813///
814/// The stub is in-store but unwritten to disk — `entity_type` empty,
815/// `file_path` empty, no metadata, no sections, `stub: true` and
816/// `stub_kind: Some(kind)` set together. A later
817/// [`Engine::create_entity`] at the same id promotes the stub to a
818/// real entity (loader / parse-result merge handles the upgrade
819/// path; `stub_kind` clears to `None`).
820pub(super) fn make_stub(id: &EntityId, kind: crate::entity::StubKind) -> Entity {
821 Entity {
822 id: id.clone(),
823 title: id.name().to_string(),
824 entity_type: String::new(),
825 mem: id.mem().to_string(),
826 file_path: String::new(),
827 metadata: IndexMap::new(),
828 sections: IndexMap::new(),
829 relationships: Vec::new(),
830 content_hash: String::new(),
831 stub: true,
832 stub_kind: Some(kind),
833 heading_spans: HashMap::new(),
834 raw_section_headings: Vec::new(),
835 }
836}
837
838/// Cross-mem add-path policy gate. Same-mem writes bypass; the
839/// `[cross_mem_links]` table only gates writes that cross the
840/// mem boundary. Cross-mem writes consult
841/// [`super::Engine::cross_mem_link_allowed`] in the edge's actual
842/// direction (`source_mem → target_mem`). Disallowed pairings
843/// surface [`EngineError::CrossMemLinkNotAllowed`] with the
844/// `(from_mem, to_mem)` payload an agent already sees on
845/// `memstead_relate`.
846///
847/// After the grant admits the pairing, a target absent from a
848/// `MountCapability::ReadOnly` mount refuses with
849/// [`EngineError::CrossMemTargetNotFound`]: the engine cannot
850/// persist a stub through the read-only boundary, and a read-only
851/// mem never gains the entity later — a missing target there is a
852/// wrong link, not a pending forward reference. Same-mem targets,
853/// cross-mem targets in Write mounts, and unmounted target mems all
854/// retain the auto-stub mechanic.
855///
856/// Funnel point for every add-shaped edge write — `memstead_relate`,
857/// `memstead_create.relations[]`, `memstead_update.declare_relations`,
858/// body-wiki-link alias synthesis, and any future add-path mutation
859/// surface route through one gate so the policy can't drift between
860/// sites. Remove-shaped writes (cleanup) remain permissive and call
861/// this helper not at all.
862pub(super) fn validate_cross_mem_add_policy(
863 engine: &super::Engine,
864 source_mem: &str,
865 target: &EntityId,
866) -> Result<(), EngineError> {
867 let target_mem = target.mem();
868 if source_mem == target_mem {
869 return Ok(());
870 }
871 if !engine.cross_mem_link_allowed(source_mem, target_mem) {
872 return Err(EngineError::CrossMemLinkNotAllowed {
873 from_mem: source_mem.to_string(),
874 to_mem: target_mem.to_string(),
875 });
876 }
877 if let Some(mount) = engine.mount(target_mem)
878 && mount.capability == crate::workspace::MountCapability::ReadOnly
879 && !engine.store.contains(target)
880 && !matches!(
881 probe_deferred_target(engine, target)?,
882 DeferredTargetProbe::Exists
883 )
884 {
885 return Err(EngineError::CrossMemTargetNotFound {
886 target_id: target.to_string(),
887 target_mem: target_mem.to_string(),
888 });
889 }
890 Ok(())
891}
892
893/// Storage verdict for a cross-mem target whose mem is mounted but
894/// DEFERRED (lazy, not yet loaded) — the write-time verification of
895/// flywheel W7/02. The check asks the mem's real storage through the
896/// cheap [`crate::backend::MemBackend::entity_exists`] probe
897/// (tree-lookup-class on git-branch, metadata-class on folder) and
898/// never triggers the mem's load: verification must not convert into
899/// a full-load side effect (plan 01's seam).
900pub(super) enum DeferredTargetProbe {
901 /// The target's mem is loaded (the store is the truth) or not
902 /// mounted at all (no storage handle to ask — the
903 /// forward-reference mechanic governs).
904 NotApplicable,
905 /// Storage holds the entity: the reference is verified against
906 /// real storage even though the mem is unloaded.
907 Exists,
908 /// Storage answers and the entity is not there.
909 Absent,
910}
911
912pub(super) fn probe_deferred_target(
913 engine: &super::Engine,
914 target: &EntityId,
915) -> Result<DeferredTargetProbe, EngineError> {
916 let rel_path = crate::entity::id::id_to_file_path(target);
917 if engine.mem_is_deferred(target.mem()) {
918 let Some(mounted) = engine.mounts.iter().find(|m| m.mount.mem == target.mem()) else {
919 return Ok(DeferredTargetProbe::NotApplicable);
920 };
921 return Ok(
922 if mounted
923 .backend
924 .entity_exists(std::path::Path::new(&rel_path))?
925 {
926 DeferredTargetProbe::Exists
927 } else {
928 DeferredTargetProbe::Absent
929 },
930 );
931 }
932 // UNMOUNTED mem: ask the workspace layer's discovery hook. No
933 // hook, or no discoverable storage → NotApplicable (the
934 // forward-reference mechanic governs, unchanged).
935 if engine.mount(target.mem()).is_none()
936 && let Some(prober) = &engine.unmounted_storage_prober
937 && let Some(storage) = prober(target.mem())
938 {
939 return Ok(
940 if storage
941 .backend
942 .entity_exists(std::path::Path::new(&rel_path))?
943 {
944 DeferredTargetProbe::Exists
945 } else {
946 DeferredTargetProbe::Absent
947 },
948 );
949 }
950 Ok(DeferredTargetProbe::NotApplicable)
951}
952
953/// The one-blob type read for a storage-verified deferred target: the
954/// shape check needs the target's real entity type, and a tree hit
955/// proves existence, not type. Reads exactly the resolved path's
956/// bytes and peeks the frontmatter `type:` — never the mem. Returns
957/// `None` when the entity is absent or declares no type (the shape
958/// gate then admits, the same posture the stub-bound case has
959/// always had — the check never guesses).
960/// The stub kind for a target being auto-stubbed on an add path: a
961/// target that storage VERIFIES inside a deferred mem gets the
962/// load-time kind — the stub is only plan 01's until-load
963/// representation of a link into an unloaded mem, not a forward
964/// reference to something that awaits creation. Everything else keeps
965/// `ForwardReference`. Kinds stay annotation-not-state either way.
966pub(super) fn deferred_verified_stub_kind(
967 engine: &super::Engine,
968 target: &EntityId,
969) -> Result<crate::entity::StubKind, EngineError> {
970 Ok(
971 if matches!(
972 probe_deferred_target(engine, target)?,
973 DeferredTargetProbe::Exists
974 ) {
975 crate::entity::StubKind::LoadTime
976 } else {
977 crate::entity::StubKind::ForwardReference
978 },
979 )
980}
981
982pub(super) fn peek_deferred_target_type(
983 engine: &super::Engine,
984 target: &EntityId,
985) -> Result<Option<String>, EngineError> {
986 let rel_path = crate::entity::id::id_to_file_path(target);
987 let bytes = if engine.mem_is_deferred(target.mem()) {
988 let Some(mounted) = engine.mounts.iter().find(|m| m.mount.mem == target.mem()) else {
989 return Ok(None);
990 };
991 mounted
992 .backend
993 .read_entity(std::path::Path::new(&rel_path))?
994 } else if engine.mount(target.mem()).is_none()
995 && let Some(prober) = &engine.unmounted_storage_prober
996 && let Some(storage) = prober(target.mem())
997 {
998 storage
999 .backend
1000 .read_entity(std::path::Path::new(&rel_path))?
1001 } else {
1002 return Ok(None);
1003 };
1004 let Some(bytes) = bytes else {
1005 return Ok(None);
1006 };
1007 Ok(String::from_utf8(bytes)
1008 .ok()
1009 .and_then(|c| crate::entity::parser::peek_type_from_frontmatter(&c)))
1010}
1011
1012/// The target-schema REF for cross-schema edge routing. Loaded mems
1013/// answer from the engine's schema catalogue; an UNMOUNTED mem with
1014/// discoverable storage answers from its stored config's pin via the
1015/// discovery hook (flywheel W7/02) — the routing check
1016/// (`validate_cross_mem_edge`) needs only the ref, never the full
1017/// schema, so the source schema's `cross_mem_relationships:` entry
1018/// keeps its authority without a mount. `None` falls back to the
1019/// intra-mem path, exactly the pre-existing posture.
1020pub(super) fn target_schema_ref_for_routing(
1021 engine: &super::Engine,
1022 target_mem: &str,
1023) -> Option<memstead_schema::SchemaRef> {
1024 if let Some(s) = engine.schemas.get(target_mem) {
1025 let (name, version) = s.id();
1026 return Some(memstead_schema::SchemaRef::new(name, version));
1027 }
1028 if engine.mount(target_mem).is_none()
1029 && let Some(prober) = &engine.unmounted_storage_prober
1030 && let Some(storage) = prober(target_mem)
1031 {
1032 return storage.schema;
1033 }
1034 None
1035}
1036
1037/// Outcome of the engine's edge-validation router for a single
1038/// inline / explicit relate. Carries the optional open-mode warning
1039/// from the intra-mem flow; the cross-mem flow has no
1040/// open-mode (cross-mem entries are declared vocabulary).
1041pub(super) enum EdgeRouteOutcome {
1042 Ok,
1043 OpenModeWarning(Box<crate::ops::WarningHint>),
1044}
1045
1046/// Run rel-type + shape validation for one edge, routing through
1047/// intra-mem vocabulary or the source schema's
1048/// `cross_mem_relationships:` section as appropriate.
1049///
1050/// The routing rule:
1051/// when `source_mem != target_mem` AND the target mem's
1052/// pinned schema differs from the source schema by name or by
1053/// version, the source schema's `cross_mem_relationships:` entry
1054/// for the target schema is the sole authority for both the
1055/// vocabulary check (`INVALID_REL_TYPE`) and the shape check
1056/// (`INVALID_REL_SHAPE`). If no matching entry exists, surface
1057/// [`EngineError::CrossMemEdgeNotDeclared`].
1058///
1059/// Otherwise (same-mem, same-schema cross-mem, or target mem
1060/// unmounted) the call falls through to the existing intra-mem
1061/// validators — the same behaviour the intra-mem path always had.
1062///
1063/// `check_shape` mirrors the relate path's add-only shape posture:
1064/// pass `false` to skip the shape check (currently only the
1065/// `memstead_relate --remove` path). The vocabulary check still fires
1066/// in that case, matching the intra-mem behaviour where
1067/// `validate_rel_type` runs on both add and remove.
1068// The nine parameters are one edge's full coordinates; a params struct
1069// would restate the same fields at every call site without grouping
1070// anything that travels together elsewhere.
1071#[allow(clippy::too_many_arguments)]
1072pub(super) fn route_edge_validation(
1073 engine: &super::Engine,
1074 rel_type: &str,
1075 from_type: &str,
1076 to_type: Option<&str>,
1077 source_mem: &str,
1078 target_mem: &str,
1079 from_id: &EntityId,
1080 to_id: &EntityId,
1081 check_shape: bool,
1082) -> Result<EdgeRouteOutcome, EngineError> {
1083 use crate::runtime_validator::{
1084 CrossMemRelCheck, RelationshipCheck, validate_cross_mem_edge, validate_rel_shape,
1085 validate_rel_type,
1086 };
1087 use memstead_schema::SchemaRef;
1088
1089 let source_schema = engine
1090 .schemas
1091 .get(source_mem)
1092 .expect("schema present for every registered mount");
1093
1094 let target_schema_ref: Option<SchemaRef> = if source_mem == target_mem {
1095 None
1096 } else {
1097 target_schema_ref_for_routing(engine, target_mem)
1098 };
1099 let cross_mem_different = match (&target_schema_ref, source_schema.id()) {
1100 (Some(target), (src_name, _)) => target.name != src_name,
1101 (None, _) => false,
1102 };
1103
1104 if cross_mem_different {
1105 let target_ref = target_schema_ref
1106 .as_ref()
1107 .expect("target_schema_ref is Some when cross_mem_different");
1108 if !check_shape {
1109 // Cleanup posture: cross-mem remove stays permissive so
1110 // pre-tightening edges remain droppable without first
1111 // re-declaring them. Mirrors the intra-mem shape gate's
1112 // add-only stance.
1113 return Ok(EdgeRouteOutcome::Ok);
1114 }
1115 match validate_cross_mem_edge(
1116 rel_type,
1117 from_type,
1118 to_type,
1119 source_schema.as_ref(),
1120 target_ref,
1121 ) {
1122 CrossMemRelCheck::Ok => Ok(EdgeRouteOutcome::Ok),
1123 CrossMemRelCheck::EdgeNotDeclared => {
1124 let (src_name, src_version) = source_schema.id();
1125 Err(EngineError::CrossMemEdgeNotDeclared {
1126 source_schema: SchemaRef::new(src_name, src_version).as_display(),
1127 target_schema: target_ref.as_display(),
1128 rel_type: rel_type.to_string(),
1129 from_id: from_id.to_string(),
1130 to_id: to_id.to_string(),
1131 })
1132 }
1133 CrossMemRelCheck::Invalid(v) => Err(EngineError::Validation(v)),
1134 }
1135 } else {
1136 let warning_hint = match validate_rel_type(rel_type, source_schema.as_ref())? {
1137 RelationshipCheck::Ok => None,
1138 RelationshipCheck::OpenWarning(message) => {
1139 Some(crate::ops::WarningHint::UndeclaredRelationshipOpen {
1140 rel_type: rel_type.to_string(),
1141 message,
1142 })
1143 }
1144 };
1145 if check_shape {
1146 validate_rel_shape(rel_type, from_type, to_type, source_schema.as_ref())?;
1147 }
1148 Ok(match warning_hint {
1149 Some(w) => EdgeRouteOutcome::OpenModeWarning(Box::new(w)),
1150 None => EdgeRouteOutcome::Ok,
1151 })
1152 }
1153}
1154
1155/// Cycle-family gate for one prospective edge — the single owner of
1156/// both refusals, shared by every edge-writing verb (`memstead_relate`,
1157/// `memstead_create.relations[]`, `memstead_update.declare_relations`, and the
1158/// batch paths, which stage prior items' edges into `store` so an
1159/// intra-batch cycle refuses like a stored one):
1160///
1161/// - **Self-loop on a listed no-self-loop rel-type.** `from == to` on
1162/// any rel-type the source type lists in `no_self_loop_relationships`
1163/// refuses, regardless of the `acyclic` flag — the declaration's one
1164/// effect (see `TypeDefinition::no_self_loop_relationships`).
1165/// - **Cycle on an acyclic rel-type.** An add closing a back-path
1166/// `to → … → from` (via [`crate::graph::query::would_cycle`]) refuses
1167/// with the existing path, capped at [`RELATIONSHIP_CYCLE_PATH_CAP`].
1168/// - **Cycle in a declared acyclicity set.** When the rel-type belongs
1169/// to a `relationships.acyclic_sets` set, an add closing a back-path
1170/// in the set's UNION subgraph (via
1171/// [`crate::graph::query::would_cycle_in_set`]) refuses; the payload
1172/// additionally echoes the set and the path's per-hop rel-types.
1173///
1174/// All refuse [`EngineError::RelationshipCycle`] (`RELATIONSHIP_CYCLE`)
1175/// with identical recovery detail on every path. Callers skip this on
1176/// remove paths — removal can only break cycles, never close one.
1177pub(super) fn validate_edge_acyclicity(
1178 store: &Store,
1179 schema: &memstead_schema::Schema,
1180 from: &EntityId,
1181 from_type: &str,
1182 to: &EntityId,
1183 rel_type: &str,
1184) -> Result<(), EngineError> {
1185 if from == to && schema.type_refuses_self_loop(from_type, rel_type) {
1186 return Err(EngineError::RelationshipCycle {
1187 rel_type: rel_type.to_string(),
1188 from: from.clone(),
1189 to: to.clone(),
1190 existing_path: vec![from.clone()],
1191 path_truncated: false,
1192 acyclic_set: None,
1193 existing_path_rel_types: None,
1194 });
1195 }
1196 if schema.relationship_acyclic(rel_type)
1197 && let Some(path) = crate::graph::query::would_cycle(store, from, to, rel_type)
1198 {
1199 let truncated = path.len() > RELATIONSHIP_CYCLE_PATH_CAP;
1200 let mut existing_path = path;
1201 if truncated {
1202 existing_path.truncate(RELATIONSHIP_CYCLE_PATH_CAP);
1203 }
1204 return Err(EngineError::RelationshipCycle {
1205 rel_type: rel_type.to_string(),
1206 from: from.clone(),
1207 to: to.clone(),
1208 existing_path,
1209 path_truncated: truncated,
1210 acyclic_set: None,
1211 existing_path_rel_types: None,
1212 });
1213 }
1214 // Cycle in a declared acyclicity SET: the union subgraph of the
1215 // set must stay acyclic, so the back-path may mix rel-types. The
1216 // refusal is additive — it echoes the declared set and one
1217 // rel-type per hop of the path.
1218 if let Some(set) = schema.acyclic_set_containing(rel_type)
1219 && let Some((path, path_rels)) =
1220 crate::graph::query::would_cycle_in_set(store, from, to, set)
1221 {
1222 let truncated = path.len() > RELATIONSHIP_CYCLE_PATH_CAP;
1223 let mut existing_path = path;
1224 let mut existing_path_rel_types = path_rels;
1225 if truncated {
1226 existing_path.truncate(RELATIONSHIP_CYCLE_PATH_CAP);
1227 existing_path_rel_types.truncate(existing_path.len().saturating_sub(1));
1228 }
1229 return Err(EngineError::RelationshipCycle {
1230 rel_type: rel_type.to_string(),
1231 from: from.clone(),
1232 to: to.clone(),
1233 existing_path,
1234 path_truncated: truncated,
1235 acyclic_set: Some(set.to_vec()),
1236 existing_path_rel_types: Some(existing_path_rel_types),
1237 });
1238 }
1239 Ok(())
1240}
1241
1242/// Validate the per-edge description posture declared on the rel-type
1243/// in the routing-appropriate definition (intra-mem when source and
1244/// target share the schema; cross-mem entry when they don't). Emits
1245/// `MissingRequiredDescription` / `DescriptionNotPermitted` on
1246/// violations; `optional` and unknown rel-types are no-ops (the
1247/// vocabulary / shape gates already catch undeclared names — posture
1248/// only fires for declared names).
1249///
1250/// `description` is the normalised value (empty / whitespace-only
1251/// collapses to `None` before reaching this gate). Called from every
1252/// add path: `memstead_relate`, `declare_relations` on `memstead_create` and
1253/// `memstead_update`.
1254pub(super) fn validate_description_posture(
1255 engine: &super::Engine,
1256 rel_type: &str,
1257 description: Option<&str>,
1258 source_mem: &str,
1259 target_mem: &str,
1260 from_id: &EntityId,
1261 to_id: &EntityId,
1262) -> Result<(), EngineError> {
1263 use memstead_schema::{PerEdgeDescription, SchemaRef};
1264
1265 let source_schema = engine
1266 .schemas
1267 .get(source_mem)
1268 .expect("schema present for every registered mount");
1269 let target_schema_ref: Option<SchemaRef> = if source_mem == target_mem {
1270 None
1271 } else {
1272 target_schema_ref_for_routing(engine, target_mem)
1273 };
1274 let cross_mem_different = match (&target_schema_ref, source_schema.id()) {
1275 (Some(target), (src_name, _)) => target.name != src_name,
1276 (None, _) => false,
1277 };
1278
1279 let posture = if cross_mem_different {
1280 // Look up the matching cross-mem entry's definition. If the
1281 // entry exists but the rel-type isn't enumerated under it, the
1282 // vocabulary gate (route_edge_validation) will surface
1283 // `CROSS_MEM_EDGE_NOT_DECLARED`; posture is a no-op there.
1284 let target_ref = target_schema_ref
1285 .as_ref()
1286 .expect("target_schema_ref is Some when cross_mem_different");
1287 source_schema
1288 .cross_mem_entries(&target_ref.name)
1289 .iter()
1290 .find_map(|entry| entry.definitions.iter().find(|d| d.name == rel_type))
1291 .map(|d| d.per_edge_description)
1292 } else {
1293 source_schema
1294 .relationship_def(rel_type)
1295 .map(|d| d.per_edge_description)
1296 };
1297
1298 match posture {
1299 Some(PerEdgeDescription::Required) if description.is_none() => {
1300 Err(EngineError::MissingRequiredDescription {
1301 rel_type: rel_type.to_string(),
1302 from_id: from_id.to_string(),
1303 to_id: to_id.to_string(),
1304 })
1305 }
1306 Some(PerEdgeDescription::Forbidden) if description.is_some() => {
1307 Err(EngineError::DescriptionNotPermitted {
1308 rel_type: rel_type.to_string(),
1309 from_id: from_id.to_string(),
1310 to_id: to_id.to_string(),
1311 })
1312 }
1313 _ => Ok(()),
1314 }
1315}
1316
1317/// Validate the manual-authoring posture declared on the rel-type.
1318/// Fires only on explicit-author paths (`memstead_relate`, inline
1319/// `relations:` on `memstead_create`, `declare_relations` on
1320/// `memstead_update`). The body-link → relation alias machinery
1321/// synthesises relations from wiki-links — that path bypasses this
1322/// gate by construction (it never calls this function), keeping the
1323/// alias path for `manual_authoring: forbidden` rel-types (e.g.
1324/// REFERENCES) intact.
1325pub(super) fn validate_manual_authoring_posture(
1326 engine: &super::Engine,
1327 rel_type: &str,
1328 source_mem: &str,
1329 from_id: &EntityId,
1330 to_id: &EntityId,
1331) -> Result<(), EngineError> {
1332 use memstead_schema::ManualAuthoring;
1333
1334 let source_schema = engine
1335 .schemas
1336 .get(source_mem)
1337 .expect("schema present for every registered mount");
1338 let posture = source_schema.relationship_manual_authoring(rel_type);
1339 if matches!(posture, ManualAuthoring::Forbidden) {
1340 let guidance = source_schema
1341 .relationship_when_to_use(rel_type)
1342 .unwrap_or_default();
1343 return Err(EngineError::RelationManualAuthoringForbidden {
1344 rel_type: rel_type.to_string(),
1345 from_id: from_id.to_string(),
1346 to_id: to_id.to_string(),
1347 guidance,
1348 });
1349 }
1350 Ok(())
1351}
1352
1353/// Alias-synthesis pass — populates `next.relationships` with engine-
1354/// emitted relations of the source schema's `alias_target_rel_type`
1355/// pointer for every body wiki-link not already backed by an
1356/// in-section-body explicit relation. Runs before the
1357/// `scan_wikilinks_without_relation` validator; after this pass the
1358/// validator finds zero missing wiki-links for the pointer rel-type.
1359///
1360/// Three cases:
1361/// 1. Schema has no pointer (`alias_target_rel_type` absent): no-op.
1362/// Caller's validator continues to refuse unbacked links exactly as
1363/// today.
1364/// 2. Schema has a pointer, body wiki-link target is in the same mem
1365/// OR cross-mem policy admits it: append `Relationship { rel_type:
1366/// pointer, target, description: None }` to `next.relationships` if
1367/// no relation of `(pointer, target)` is already present. Dedupe is
1368/// `(target, rel_type)` — a USES or DEPENDS_ON edge to the same
1369/// target does not suppress synthesis of the pointer rel-type.
1370/// 3. Schema has a pointer but a body wiki-link crosses a mem
1371/// boundary the workspace doesn't grant — or targets an entity
1372/// absent from a read-only mount: return the funnel's typed
1373/// refusal ([`EngineError::CrossMemLinkNotAllowed`] /
1374/// [`EngineError::CrossMemTargetNotFound`], via
1375/// [`validate_cross_mem_add_policy`]). The entire mutation
1376/// aborts — no partial state.
1377///
1378/// GC: when `prev` is `Some`, the pass also drops pointer-rel-type
1379/// relations whose target was a body wiki-link in `prev` but no longer
1380/// appears in `next.sections`. The loader forces `manual_authoring:
1381/// forbidden` on every schema's `alias_target_rel_type` pointer, so the
1382/// only path to a pointer-rel-type edge is the body-link channel; the
1383/// GC rule therefore reduces to "drop pointer-rel-type relations whose
1384/// target is not in the new body". Targeting prev's wiki-link set
1385/// specifically (rather than every pointer-rel-type relation) keeps the
1386/// pass correct even for an explicit-author relation that predates the
1387/// forbid posture.
1388///
1389/// Returns the list of relations the pass emitted (in body iteration
1390/// order) — `create.rs` / `update.rs` use it to surface
1391/// `relations_emitted` on the response envelope.
1392/// Returns the synthesised relations (in body iteration order) and a flag
1393/// signalling whether a body wiki-link to the entity's own id was dropped
1394/// (F11). The caller surfaces that as a `SELF_LINK_IGNORED` warning — the
1395/// pass has no warning channel of its own.
1396pub(super) fn synthesise_alias_relations(
1397 engine: &super::Engine,
1398 prev_body_targets: &std::collections::HashSet<EntityId>,
1399 next: &mut Entity,
1400) -> Result<AliasSynthesisOutcome, super::EngineError> {
1401 let schema = engine
1402 .schemas
1403 .get(next.mem.as_str())
1404 .expect("schema present for every registered mount");
1405 let Some(pointer) = schema.alias_target_rel_type().map(str::to_string) else {
1406 return Ok(AliasSynthesisOutcome::default());
1407 };
1408
1409 // 1. GC: drop pointer-rel-type relations whose target was a body
1410 // wiki-link in the prev entity state but isn't in next. Targets
1411 // not in prev's wiki-link set are explicit-author relations and
1412 // are never touched — the rule preserves explicit edges even
1413 // while the 5 built-ins still admit explicit REFERENCES.
1414 //
1415 // `extract_inline_links` is strict — non-slug-form targets refuse
1416 // here with the typed `InvalidWikiLinkTarget` envelope rather
1417 // than silently flowing into the GC's retain set as malformed
1418 // EntityIds. Section context comes from the iteration key.
1419 let mut next_targets: std::collections::HashSet<EntityId> = std::collections::HashSet::new();
1420 for (section_key, body) in next.sections.iter() {
1421 let ids = crate::entity::parser::extract_inline_links(body, &next.mem)
1422 .map_err(|errs| map_wiki_link_errors(section_key, errs))?;
1423 next_targets.extend(ids);
1424 }
1425 next.relationships.retain(|r| {
1426 !(r.rel_type == pointer
1427 && prev_body_targets.contains(&r.target)
1428 && !next_targets.contains(&r.target))
1429 });
1430
1431 // 2. Walk body wiki-links in section iteration order and append
1432 // one relation per `(target, pointer)` pair not already
1433 // present. Cross-mem gate fires on the first refusal.
1434 let existing: std::collections::HashSet<(String, EntityId)> = next
1435 .relationships
1436 .iter()
1437 .map(|r| (r.rel_type.clone(), r.target.clone()))
1438 .collect();
1439 let mut emitted: Vec<crate::entity::Relationship> = Vec::new();
1440 let mut already_synthesised: std::collections::HashSet<EntityId> =
1441 std::collections::HashSet::new();
1442 let mut self_link_ignored = false;
1443 let mut undeclared_dropped: Vec<UndeclaredCrossSchemaLink> = Vec::new();
1444 let mut undeclared_seen: std::collections::HashSet<EntityId> = std::collections::HashSet::new();
1445 for (section_key, body) in next.sections.iter() {
1446 let ids = crate::entity::parser::extract_inline_links(body, &next.mem)
1447 .map_err(|errs| map_wiki_link_errors(section_key, errs))?;
1448 for target in ids {
1449 // F11: a body wiki-link to the entity's own id is a vacuous
1450 // self-edge (renders as both Outgoing and Incoming, inflates
1451 // connectivity). Drop it — but don't refuse: the author may
1452 // have written their own slug. The caller surfaces
1453 // `SELF_LINK_IGNORED` so the dropped link stays observable.
1454 if target == next.id {
1455 self_link_ignored = true;
1456 continue;
1457 }
1458 let key = (pointer.clone(), target.clone());
1459 if existing.contains(&key) || already_synthesised.contains(&target) {
1460 continue;
1461 }
1462 validate_cross_mem_add_policy(engine, &next.mem, &target)?;
1463 // Schema-law gate: a link into a DIFFERENT schema is an edge
1464 // only where the source schema's cross_mem_relationships
1465 // declare the pointer rel-type for that destination (exact
1466 // entry or wildcard). Emitting anyway wrote an edge the
1467 // load-path filter then silently discarded on the next boot
1468 // — the write showed a relation the graph would not keep
1469 // (graph-plans 02 grading, 2026-08-28). Skip synthesis and
1470 // record the drop; the caller warns, typed, so the author
1471 // learns the citation is prose only. An unmounted or
1472 // schema-less target stays permissive (nothing to judge).
1473 if target.mem() != next.mem
1474 && let Some(target_ref) = target_schema_ref_for_routing(engine, target.mem())
1475 && target_ref.name != schema.id().0
1476 && !schema
1477 .cross_mem_entries(&target_ref.name)
1478 .iter()
1479 .any(|entry| entry.definitions.iter().any(|d| d.name == pointer))
1480 {
1481 if undeclared_seen.insert(target.clone()) {
1482 let (src_name, src_version) = schema.id();
1483 undeclared_dropped.push(UndeclaredCrossSchemaLink {
1484 target,
1485 source_schema: memstead_schema::SchemaRef::new(src_name, src_version)
1486 .as_display(),
1487 target_schema: target_ref.name,
1488 });
1489 }
1490 continue;
1491 }
1492 let rel = crate::entity::Relationship::new(pointer.clone(), target.clone());
1493 next.relationships.push(rel.clone());
1494 already_synthesised.insert(target);
1495 emitted.push(rel);
1496 }
1497 }
1498 Ok(AliasSynthesisOutcome {
1499 emitted,
1500 self_link_ignored,
1501 undeclared_dropped,
1502 })
1503}
1504
1505/// One body wiki-link the alias pass declined to turn into an edge
1506/// because the source schema declares no cross-mem entry carrying the
1507/// pointer rel-type for the target's schema. The caller surfaces each
1508/// as a `CROSS_SCHEMA_LINK_UNDECLARED` warning.
1509pub(super) struct UndeclaredCrossSchemaLink {
1510 pub target: EntityId,
1511 pub source_schema: String,
1512 pub target_schema: String,
1513}
1514
1515/// What [`synthesise_alias_relations`] did: the relations it emitted
1516/// (in body iteration order), whether a self-link was dropped (F11),
1517/// and the cross-schema links it declined for lack of a declaration.
1518#[derive(Default)]
1519pub(super) struct AliasSynthesisOutcome {
1520 pub emitted: Vec<crate::entity::Relationship>,
1521 pub self_link_ignored: bool,
1522 pub undeclared_dropped: Vec<UndeclaredCrossSchemaLink>,
1523}
1524
1525/// Map the first [`crate::entity::id::WikiLinkError`] from a body
1526/// wiki-link extraction into the typed [`EngineError`] envelope,
1527/// attaching the offending section's key. Errors after the first are
1528/// dropped — the agent reads the error, fixes the link, retries, and
1529/// surfaces the next one on the follow-up call. Keeps the envelope
1530/// shape stable (single typed payload rather than a list) so MCP /
1531/// CLI clients don't need a fan-out renderer.
1532pub(super) fn map_wiki_link_errors(
1533 section_key: &str,
1534 errors: Vec<crate::entity::id::WikiLinkError>,
1535) -> EngineError {
1536 use crate::entity::id::WikiLinkError;
1537 let first = errors
1538 .into_iter()
1539 .next()
1540 .expect("map_wiki_link_errors called with non-empty error list");
1541 match first {
1542 WikiLinkError::InvalidTarget {
1543 raw,
1544 suggested,
1545 reason,
1546 } => EngineError::InvalidWikiLinkTarget {
1547 raw,
1548 suggested,
1549 section: section_key.to_string(),
1550 link_source: "body_link".to_string(),
1551 reason,
1552 },
1553 WikiLinkError::InvalidMemName { raw, reason } => EngineError::InvalidWikiLinkMem {
1554 raw,
1555 section: section_key.to_string(),
1556 reason,
1557 },
1558 }
1559}
1560
1561/// Compute the set of body wiki-link targets in an entity. Used by
1562/// callers of `synthesise_alias_relations` to capture the pre-mutation
1563/// state once, before any borrow conflicts re-enter the engine's
1564/// schemas / store maps. Uses the lenient decoder — this snapshot
1565/// must tolerate on-disk drift on pre-strict entities whose bodies
1566/// may still contain non-conformant links; the strict gate fires
1567/// only on the post-mutation `next` state.
1568pub(super) fn collect_body_link_targets(entity: &Entity) -> std::collections::HashSet<EntityId> {
1569 entity
1570 .sections
1571 .iter()
1572 .flat_map(|(_, body)| {
1573 crate::entity::parser::extract_inline_links_lenient(body, &entity.mem)
1574 })
1575 .collect()
1576}
1577
1578/// Alias-existence invariant validator. Given the post-mutation entity
1579/// state, scan every section body for wiki-links whose target has no
1580/// corresponding explicit relation in `entity.relationships`. Returns
1581/// the list of `(section_key, target_id)` pairs that violate the
1582/// invariant — empty when the post-mutation state is clean.
1583///
1584/// Used by [`Engine::create_entity`] and [`Engine::update_entity`]
1585/// (and `batch_update`). The validator runs unconditionally — under
1586/// the alias model body wiki-links are foreign-key references on the
1587/// `## Relationships` table and every reference must be backed.
1588///
1589/// Sections from the auto-managed `## Relationships` heading are
1590/// not scanned (the engine generates them from the relations list
1591/// at write time; the parser keeps them out of
1592/// `entity.sections` so they never reach this function).
1593///
1594/// Reuses [`crate::entity::parser::extract_inline_links`] so the
1595/// lexical discipline (fenced-code masking, inline-code masking,
1596/// alias handling, cross-mem forms) matches every other validator
1597/// surface in the engine.
1598pub(super) fn scan_wikilinks_without_relation(
1599 next: &Entity,
1600 exempt: &std::collections::HashSet<EntityId>,
1601) -> Result<Vec<(String, EntityId)>, EngineError> {
1602 let explicit_targets: std::collections::HashSet<EntityId> = next
1603 .relationships
1604 .iter()
1605 .map(|r| r.target.clone())
1606 .collect();
1607 let mut missing: Vec<(String, EntityId)> = Vec::new();
1608 for (section_key, body) in next.sections.iter() {
1609 let ids = crate::entity::parser::extract_inline_links(body, &next.mem)
1610 .map_err(|errs| map_wiki_link_errors(section_key, errs))?;
1611 for target in ids {
1612 // A self-targeting body link is intentionally unbacked: the
1613 // alias pass drops its (vacuous) self-edge (F11), so it has no
1614 // backing relation by design and must not trip the
1615 // unbacked-link refusal here.
1616 if target == next.id {
1617 continue;
1618 }
1619 // A cross-schema link the alias pass declined for lack of a
1620 // declaration is likewise intentionally unbacked: the caller
1621 // warns (`CROSS_SCHEMA_LINK_UNDECLARED`) instead of the write
1622 // refusing over prose the schema deliberately keeps inert.
1623 if exempt.contains(&target) {
1624 continue;
1625 }
1626 if !explicit_targets.contains(&target)
1627 && !missing
1628 .iter()
1629 .any(|(k, t)| k == section_key && t == &target)
1630 {
1631 missing.push((section_key.clone(), target));
1632 }
1633 }
1634 }
1635 Ok(missing)
1636}
1637
1638#[cfg(test)]
1639mod tests {
1640
1641 use tempfile::TempDir;
1642
1643 use crate::backend::MemBackend;
1644 use crate::engine::test_helpers::*;
1645 use crate::engine::{CreateEntityArgs, Engine, UpdateEntityArgs};
1646
1647 use crate::storage::FilesystemMemWriter;
1648 use crate::vcs::CommitContext;
1649
1650 use indexmap::IndexMap;
1651
1652 #[test]
1653 fn with_ctx_wrappers_delegate_to_explicit_forms() {
1654 // Each *_with_ctx wrapper bundles a CommitContext and
1655 // routes through the corresponding 4-arg method. Verify
1656 // create → update → rename → delete via the wrappers
1657 // observably mutate the store the same way the explicit
1658 // forms would.
1659 let tmp = TempDir::new().unwrap();
1660 let mem_dir = tmp.path().to_path_buf();
1661 let writer = FilesystemMemWriter::new(mem_dir.clone());
1662 let mut engine = Engine::from_mounts(vec![(
1663 folder_mount("specs", mem_dir),
1664 Box::new(writer) as Box<dyn MemBackend>,
1665 )])
1666 .unwrap();
1667 let ctx = CommitContext::internal();
1668
1669 // create_entity_with_ctx
1670 let create_args = CreateEntityArgs {
1671 anchors: Vec::new(),
1672 mem: "specs".to_string(),
1673 title: "Seed".to_string(),
1674 entity_type: "spec".to_string(),
1675 sections: IndexMap::from_iter([
1676 ("identity".to_string(), "seed identity".to_string()),
1677 ("purpose".to_string(), "seed purpose".to_string()),
1678 ]),
1679 metadata: IndexMap::new(),
1680 relations: Vec::new(),
1681 dry_run: false,
1682 };
1683 let created = engine.create_entity_with_ctx(create_args, &ctx).unwrap();
1684 assert_eq!(created.title, "Seed");
1685 assert!(engine.store().get(&created.id).is_some());
1686
1687 // update_entity_with_ctx
1688 let update_args = UpdateEntityArgs {
1689 anchors: Vec::new(),
1690 id: created.id.clone(),
1691 expected_hash: Some(created.content_hash.clone()),
1692 sections: IndexMap::from_iter([("identity".to_string(), "updated".to_string())]),
1693 append_sections: IndexMap::new(),
1694 patch_sections: IndexMap::new(),
1695 metadata: IndexMap::new(),
1696 metadata_unset: Vec::new(),
1697 dry_run: false,
1698 declare_relations: Vec::new(),
1699 relations_unset: Vec::new(),
1700 anchors_unset: Vec::new(),
1701 };
1702 let updated = engine.update_entity_with_ctx(update_args, &ctx).unwrap();
1703 assert!(
1704 !updated.write_id.is_empty()
1705 || (updated.modified_sections.replaced.is_empty()
1706 && updated.modified_sections.appended.is_empty()
1707 && updated.modified_sections.patched.is_empty())
1708 );
1709
1710 // rename_entity_with_ctx
1711 let renamed = engine
1712 .rename_entity_with_ctx(&created.id, "Renamed", &updated.content_hash, &ctx)
1713 .unwrap();
1714 assert_ne!(renamed.old_id, renamed.new_id);
1715 assert!(engine.store().get(&renamed.new_id).is_some());
1716
1717 // delete_entity_with_ctx
1718 let deleted = engine
1719 .delete_entity_with_ctx(&renamed.new_id, &renamed.content_hash, &ctx)
1720 .unwrap();
1721 assert_eq!(deleted.id, renamed.new_id);
1722 assert!(engine.store().get(&renamed.new_id).is_none());
1723 }
1724
1725 /// Minimal on-disk MemConfig pinning `default@1.0.0`, with an
1726 /// optional pre-set mutation stamp — the carrier the stamp path
1727 /// and the boot skew check both read.
1728 fn write_config(dir: &std::path::Path, stamp: Option<memstead_schema::MutationStamp>) {
1729 let meta = dir.join(memstead_schema::MEM_META_DIR);
1730 std::fs::create_dir_all(&meta).unwrap();
1731 let mut config: memstead_schema::MemConfig =
1732 serde_json::from_str(r#"{"schema": "default@1.0.0"}"#).unwrap();
1733 config.mutation_stamp = stamp;
1734 std::fs::write(
1735 meta.join("config.json"),
1736 serde_json::to_vec_pretty(&config).unwrap(),
1737 )
1738 .unwrap();
1739 }
1740
1741 fn stamped_engine_fixture(mem_dir: std::path::PathBuf) -> Engine {
1742 Engine::from_mounts(vec![(
1743 folder_mount("specs", mem_dir.clone()),
1744 Box::new(FilesystemMemWriter::new(mem_dir)) as Box<dyn MemBackend>,
1745 )])
1746 .unwrap()
1747 }
1748
1749 fn disk_stamp(dir: &std::path::Path) -> Option<memstead_schema::MutationStamp> {
1750 let bytes =
1751 std::fs::read(dir.join(memstead_schema::MEM_META_DIR).join("config.json")).unwrap();
1752 let config: memstead_schema::MemConfig = serde_json::from_slice(&bytes).unwrap();
1753 config.mutation_stamp
1754 }
1755
1756 fn spec_create_args(title: &str) -> CreateEntityArgs {
1757 CreateEntityArgs {
1758 anchors: Vec::new(),
1759 mem: "specs".to_string(),
1760 title: title.to_string(),
1761 entity_type: "spec".to_string(),
1762 sections: IndexMap::from_iter([
1763 ("identity".to_string(), "seed identity".to_string()),
1764 ("purpose".to_string(), "seed purpose".to_string()),
1765 ]),
1766 metadata: IndexMap::new(),
1767 relations: Vec::new(),
1768 dry_run: false,
1769 }
1770 }
1771
1772 /// Criterion 3 (agent-trust plan 02): a mutation stamps the mem's
1773 /// engine-owned state with the running engine version and resolved
1774 /// schema; a read-only load writes nothing.
1775 #[test]
1776 fn mutation_writes_version_stamp_and_read_only_load_does_not() {
1777 let tmp = TempDir::new().unwrap();
1778 let mem_dir = tmp.path().to_path_buf();
1779 write_config(&mem_dir, None);
1780
1781 // Read-only session: boot and drop without mutating — the
1782 // stamp stays absent.
1783 drop(stamped_engine_fixture(mem_dir.clone()));
1784 assert!(
1785 disk_stamp(&mem_dir).is_none(),
1786 "a read-only load must not write a stamp"
1787 );
1788
1789 // A mutation stamps engine version + resolved schema.
1790 let mut engine = stamped_engine_fixture(mem_dir.clone());
1791 engine
1792 .create_entity_with_ctx(spec_create_args("Seed"), &CommitContext::internal())
1793 .unwrap();
1794 let stamp = disk_stamp(&mem_dir).expect("mutation must write the stamp");
1795 assert_eq!(stamp.engine_version, crate::build_info::full_version());
1796 assert_eq!(stamp.schema, "default@1.0.0");
1797
1798 // A second mutation under the same binary leaves the stamp at
1799 // the same value (the write path compares and no-ops).
1800 let mut engine = stamped_engine_fixture(mem_dir.clone());
1801 engine
1802 .create_entity_with_ctx(spec_create_args("Second"), &CommitContext::internal())
1803 .unwrap();
1804 let again = disk_stamp(&mem_dir).expect("stamp survives");
1805 assert_eq!(again, stamp);
1806 }
1807
1808 /// The reported damage, reproduced (04/03, criteria 7 and 8): a
1809 /// long-lived engine boots, a sibling writes the config out of band, and
1810 /// the engine's next ENTITY mutation stamps the version. Before the fix
1811 /// that stamp serialized the boot-time struct and the sibling's write was
1812 /// gone. No lifecycle call is involved anywhere in this test, which is why
1813 /// the loss looked spontaneous to the operator who reported it.
1814 ///
1815 /// The divergent stamp is written to disk rather than injected, because
1816 /// the running binary's version is a compile-time constant with no runtime
1817 /// seam; this is how the existing skew coverage reaches the condition too.
1818 #[test]
1819 fn a_sibling_config_write_survives_the_next_entity_mutation() {
1820 let tmp = TempDir::new().unwrap();
1821 let mem_dir = tmp.path().to_path_buf();
1822 // Seed a stamp that disagrees with this binary, so the stamp writer is
1823 // live rather than dormant: that is the two-binary topology the report
1824 // came from.
1825 write_config(
1826 &mem_dir,
1827 Some(memstead_schema::MutationStamp {
1828 engine_version: "0.0.1-other".to_string(),
1829 schema: "default@1.0.0".to_string(),
1830 }),
1831 );
1832
1833 // The long-lived engine boots and caches the config as it is now.
1834 let mut engine = stamped_engine_fixture(mem_dir.clone());
1835
1836 // A sibling process sets a description. The engine never learns:
1837 // a config-only write advances no entity head and appends no change
1838 // log line, so the staleness probe cannot see it.
1839 let path = mem_dir
1840 .join(memstead_schema::MEM_META_DIR)
1841 .join("config.json");
1842 let mut sibling: memstead_schema::MemConfig =
1843 serde_json::from_slice(&std::fs::read(&path).unwrap()).unwrap();
1844 sibling.description = Some("written by the sibling".to_string());
1845 std::fs::write(&path, serde_json::to_vec_pretty(&sibling).unwrap()).unwrap();
1846
1847 // An ordinary entity write. Nothing about it mentions config.
1848 engine
1849 .create_entity_with_ctx(spec_create_args("Seed"), &CommitContext::internal())
1850 .unwrap();
1851
1852 let after: memstead_schema::MemConfig =
1853 serde_json::from_slice(&std::fs::read(&path).unwrap()).unwrap();
1854 assert_eq!(
1855 after.description.as_deref(),
1856 Some("written by the sibling"),
1857 "the sibling's description must survive an entity mutation"
1858 );
1859 assert_eq!(
1860 after.mutation_stamp.map(|s| s.engine_version),
1861 Some(crate::build_info::full_version().to_string()),
1862 "and the stamp this engine came to write must still land"
1863 );
1864 }
1865
1866 /// Criterion 3 for the stamp writer: the intervention reaches the ENTITY
1867 /// mutation's own response. The stamp has no response of its own, and an
1868 /// earlier draft discarded the report with `let _`, so an operator whose
1869 /// config moved during an innocuous entity write was told nothing.
1870 #[test]
1871 fn the_stamps_intervention_rides_the_entity_mutations_response() {
1872 let tmp = TempDir::new().unwrap();
1873 let mem_dir = tmp.path().to_path_buf();
1874 write_config(
1875 &mem_dir,
1876 Some(memstead_schema::MutationStamp {
1877 engine_version: "0.0.1-other".to_string(),
1878 schema: "default@1.0.0".to_string(),
1879 }),
1880 );
1881 let mut engine = stamped_engine_fixture(mem_dir.clone());
1882
1883 let path = mem_dir
1884 .join(memstead_schema::MEM_META_DIR)
1885 .join("config.json");
1886 let mut sibling: memstead_schema::MemConfig =
1887 serde_json::from_slice(&std::fs::read(&path).unwrap()).unwrap();
1888 sibling.description = Some("theirs".to_string());
1889 std::fs::write(&path, serde_json::to_vec_pretty(&sibling).unwrap()).unwrap();
1890
1891 let outcome = engine
1892 .create_entity_with_ctx(spec_create_args("Seed"), &CommitContext::internal())
1893 .unwrap();
1894 assert!(
1895 outcome
1896 .warnings
1897 .iter()
1898 .any(|w| w.code() == "CONFIG_WRITE_INTERVENED"),
1899 "the entity mutation must report the config intervention: {:?}",
1900 outcome.warnings
1901 );
1902 }
1903
1904 /// Criterion 5: the folder backend's config write is a compare-and-set,
1905 /// not check-then-write. A write whose `expected` no longer matches the
1906 /// file must refuse rather than overwrite.
1907 #[test]
1908 fn the_folder_config_write_refuses_a_stale_expectation() {
1909 use crate::backend::MemBackend;
1910 let tmp = TempDir::new().unwrap();
1911 let mem_dir = tmp.path().to_path_buf();
1912 write_config(&mem_dir, None);
1913 let backend = FilesystemMemWriter::new(mem_dir.clone());
1914 let observed = backend.read_mem_config().unwrap().expect("config exists");
1915
1916 // Someone else writes.
1917 let path = mem_dir
1918 .join(memstead_schema::MEM_META_DIR)
1919 .join("config.json");
1920 std::fs::write(&path, br#"{"schema": "default@1.0.0", "title": "theirs"}"#).unwrap();
1921
1922 // A write against the stale expectation is refused, not applied.
1923 let wrote = backend
1924 .write_mem_config_cas(Some(&observed), b"{\"schema\": \"default@1.0.0\"}", None)
1925 .unwrap();
1926 assert!(!wrote, "a stale expectation must not overwrite");
1927 let on_disk = std::fs::read_to_string(&path).unwrap();
1928 assert!(
1929 on_disk.contains("theirs"),
1930 "their write survived: {on_disk}"
1931 );
1932
1933 // And against the current bytes it lands.
1934 let current = backend.read_mem_config().unwrap().unwrap();
1935 assert!(
1936 backend
1937 .write_mem_config_cas(Some(¤t), b"{\"schema\": \"default@1.0.0\"}", None)
1938 .unwrap(),
1939 "a current expectation writes"
1940 );
1941 }
1942
1943 /// Criterion 7's complement: the stamp does not become a busy writer. With
1944 /// a stamp that already agrees, an entity mutation must not touch the
1945 /// config at all, so a sibling's write is untouched for the boring reason
1946 /// rather than the interesting one.
1947 #[test]
1948 fn a_matching_stamp_still_writes_no_config_at_all() {
1949 let tmp = TempDir::new().unwrap();
1950 let mem_dir = tmp.path().to_path_buf();
1951 write_config(
1952 &mem_dir,
1953 Some(memstead_schema::MutationStamp {
1954 engine_version: crate::build_info::full_version().to_string(),
1955 schema: "default@1.0.0".to_string(),
1956 }),
1957 );
1958 let mut engine = stamped_engine_fixture(mem_dir.clone());
1959 let path = mem_dir
1960 .join(memstead_schema::MEM_META_DIR)
1961 .join("config.json");
1962 let before = std::fs::read(&path).unwrap();
1963 engine
1964 .create_entity_with_ctx(spec_create_args("Seed"), &CommitContext::internal())
1965 .unwrap();
1966 assert_eq!(
1967 std::fs::read(&path).unwrap(),
1968 before,
1969 "a mutation whose stamp already matches must write no config"
1970 );
1971 }
1972
1973 /// 04/04, criteria 9 and 10: skew reaches the write that meets it, before
1974 /// that write's own restamp erases the evidence, and the write still
1975 /// lands.
1976 ///
1977 /// Boot-only detection meant a long-lived server started under one binary
1978 /// and written to by another never said so, because the first mutation
1979 /// both revealed and hid the fact.
1980 #[test]
1981 fn skew_is_reported_at_the_write_and_the_write_still_lands() {
1982 let tmp = TempDir::new().unwrap();
1983 let mem_dir = tmp.path().to_path_buf();
1984 write_config(
1985 &mem_dir,
1986 Some(memstead_schema::MutationStamp {
1987 engine_version: "0.0.1".to_string(),
1988 schema: "default@1.0.0".to_string(),
1989 }),
1990 );
1991 let mut engine = stamped_engine_fixture(mem_dir.clone());
1992 let outcome = engine
1993 .create_entity_with_ctx(spec_create_args("Seed"), &CommitContext::internal())
1994 .unwrap();
1995
1996 let skew: Vec<_> = outcome
1997 .warnings
1998 .iter()
1999 .filter(|w| w.code() == "ENGINE_VERSION_SKEW")
2000 .collect();
2001 assert_eq!(
2002 skew.len(),
2003 1,
2004 "the write that meets the skew must report it: {:?}",
2005 outcome.warnings
2006 );
2007 assert!(
2008 matches!(
2009 skew[0],
2010 crate::ops::WarningHint::EngineVersionSkew {
2011 direction: crate::build_info::SkewDirection::StampedOlder,
2012 ..
2013 }
2014 ),
2015 "and say which way: {:?}",
2016 skew[0]
2017 );
2018 // Criterion 10: it landed. An older engine is not prevented from
2019 // writing; a deliberate downgrade is the operator's business.
2020 assert!(engine.get_entity(&outcome.id).is_some());
2021 assert_eq!(
2022 disk_stamp(&mem_dir).map(|s| s.engine_version),
2023 Some(crate::build_info::full_version().to_string()),
2024 "and the restamp still happened"
2025 );
2026
2027 // Second write, same binary: nothing left to report.
2028 let again = engine
2029 .create_entity_with_ctx(spec_create_args("Second"), &CommitContext::internal())
2030 .unwrap();
2031 assert!(
2032 !again
2033 .warnings
2034 .iter()
2035 .any(|w| w.code() == "ENGINE_VERSION_SKEW"),
2036 "the skew is resolved once restamped: {:?}",
2037 again.warnings
2038 );
2039 }
2040
2041 /// Criterion 8's complement at the write tier: a stamp from the same
2042 /// release with a different build hash is not skew, so a workspace whose
2043 /// binary is rebuilt from source is not told its engine disagrees on
2044 /// every mutation.
2045 #[test]
2046 fn a_rebuild_of_the_same_release_is_not_skew_at_the_write() {
2047 let tmp = TempDir::new().unwrap();
2048 let mem_dir = tmp.path().to_path_buf();
2049 write_config(
2050 &mem_dir,
2051 Some(memstead_schema::MutationStamp {
2052 engine_version: format!("{}+gdeadbee", crate::ENGINE_VERSION),
2053 schema: "default@1.0.0".to_string(),
2054 }),
2055 );
2056 let mut engine = stamped_engine_fixture(mem_dir.clone());
2057 let outcome = engine
2058 .create_entity_with_ctx(spec_create_args("Seed"), &CommitContext::internal())
2059 .unwrap();
2060 assert!(
2061 !outcome
2062 .warnings
2063 .iter()
2064 .any(|w| w.code() == "ENGINE_VERSION_SKEW"),
2065 "a differing build hash on the same version is not skew: {:?}",
2066 outcome.warnings
2067 );
2068 }
2069
2070 /// Criterion 3/4 (agent-trust plan 02): boot under a different
2071 /// binary version surfaces the warn-tier `ENGINE_VERSION_SKEW`
2072 /// naming both versions, on load warnings AND in `health()`;
2073 /// a stamp-less mem and a matching stamp are silent.
2074 #[test]
2075 fn boot_skew_warning_fires_only_on_disagreeing_stamp() {
2076 use crate::ops::WarningHint;
2077
2078 // Disagreeing stamp → warning on boot and in health.
2079 let tmp = TempDir::new().unwrap();
2080 let mem_dir = tmp.path().to_path_buf();
2081 write_config(
2082 &mem_dir,
2083 Some(memstead_schema::MutationStamp {
2084 engine_version: "0.0.1".to_string(),
2085 schema: "default@1.0.0".to_string(),
2086 }),
2087 );
2088 let engine = stamped_engine_fixture(mem_dir);
2089 let skew: Vec<_> = engine
2090 .load_warnings()
2091 .iter()
2092 .filter(|w| matches!(w, WarningHint::EngineVersionSkew { .. }))
2093 .collect();
2094 assert_eq!(skew.len(), 1, "one skewed mem, one warning: {skew:?}");
2095 if let WarningHint::EngineVersionSkew {
2096 mem,
2097 stamped_engine,
2098 running_engine,
2099 stamped_schema,
2100 direction,
2101 } = skew[0]
2102 {
2103 assert_eq!(mem, "specs");
2104 assert_eq!(stamped_engine, "0.0.1");
2105 assert_eq!(running_engine, crate::build_info::full_version());
2106 assert_eq!(stamped_schema, "default@1.0.0");
2107 // 0.0.1 against any shipped version: the mem is behind us.
2108 assert_eq!(*direction, crate::build_info::SkewDirection::StampedOlder);
2109 }
2110 let health = engine.health();
2111 assert!(
2112 health
2113 .warnings
2114 .iter()
2115 .any(|w| w.code() == "ENGINE_VERSION_SKEW"),
2116 "health() must surface the skew without an include gate: {:?}",
2117 health.warnings,
2118 );
2119
2120 // Matching stamp → silent.
2121 let tmp = TempDir::new().unwrap();
2122 let mem_dir = tmp.path().to_path_buf();
2123 write_config(
2124 &mem_dir,
2125 Some(memstead_schema::MutationStamp {
2126 engine_version: crate::build_info::full_version().to_string(),
2127 schema: "default@1.0.0".to_string(),
2128 }),
2129 );
2130 let engine = stamped_engine_fixture(mem_dir);
2131 assert!(
2132 !engine
2133 .load_warnings()
2134 .iter()
2135 .any(|w| matches!(w, WarningHint::EngineVersionSkew { .. })),
2136 "a matching stamp is not skew"
2137 );
2138
2139 // No stamp → silent (absence of a stamp is not skew).
2140 let tmp = TempDir::new().unwrap();
2141 let mem_dir = tmp.path().to_path_buf();
2142 write_config(&mem_dir, None);
2143 let engine = stamped_engine_fixture(mem_dir);
2144 assert!(
2145 !engine
2146 .load_warnings()
2147 .iter()
2148 .any(|w| matches!(w, WarningHint::EngineVersionSkew { .. })),
2149 "a stamp-less (pre-plan) mem boots without warning noise"
2150 );
2151 }
2152}