memstead_base/engine/mod.rs
1//! Unified engine.
2//!
3//! **One [`Engine`] type, three storage backends**: the engine sits
4//! above [`MemBackend`] and routes reads / writes to the backend
5//! named by each mount's mem. The MCP filesystem-mem server
6//! (`memstead_mcp::filesystem_server::FilesystemMcpServer`), every CLI
7//! lean subcommand, and the macOS UniFFI consumer all reach the
8//! engine through [`Engine::from_workspace_root`] (lean: folder +
9//! archive backends) or `memstead_git_branch::engine_from_workspace_root`
10//! (full: adds git-branch).
11//!
12//! ## Routing
13//!
14//! Each mount holds one mem. Lookup is by mem name: the first
15//! mount whose `mem` field equals the requested name wins. One mount
16//! per mem is enforced — duplicates are a configuration bug, not a
17//! feature, and the constructor rejects them.
18
19use std::cell::OnceCell;
20use std::collections::HashMap;
21use std::path::{Path, PathBuf};
22use std::sync::Arc;
23
24use memstead_schema::Schema;
25
26use crate::backend::{BackendError, MemBackend};
27use crate::graph::LouvainOutput;
28use crate::mem::MemRouterSnapshot;
29use crate::ops::WarningHint;
30#[cfg(not(target_arch = "wasm32"))]
31use crate::search_index::MemIndex;
32use crate::store::Store;
33use crate::workspace::{Mount, WorkspaceSettings};
34
35pub mod apply_commit;
36pub mod archive;
37pub mod boot;
38pub mod drift;
39pub mod error;
40pub mod events;
41#[cfg(feature = "file-watcher")]
42pub mod file_watcher;
43pub mod lifecycle;
44pub mod mutation;
45pub mod outcomes;
46pub mod query;
47
48pub use archive::FromArchiveBytesError;
49pub use error::{
50 BootError, EngineError, INLINE_LIST_CAP, MissingWikiLink, ReferrerInfo, SchemaSourceDiagnostic,
51 format_inline_list_overflow,
52};
53#[cfg(feature = "tokio")]
54pub use events::DEFAULT_BROADCAST_CAPACITY;
55pub use events::{EventCallback, MemChangedEvent, SubscriptionHandle};
56#[cfg(feature = "file-watcher")]
57pub use file_watcher::{FileWatcherError, MemRepoWatcher, watch_mem_repo};
58pub use mutation::delete::DeleteReferrers;
59pub use mutation::{PATCH_OLD_NOT_FOUND_CONTENT_CAP, RELATIONSHIP_CYCLE_PATH_CAP};
60pub use outcomes::{
61 CreateEntityArgs, CreateEntityOutcome, DeleteEntityArgs, DeleteEntityOutcome, RelateAction,
62 RelateEntityArgs, RelateEntityOutcome, RenameEntityArgs, RenameEntityOutcome, SetSchemaOutcome,
63 SetSchemaResult, UpdateEntityArgs, UpdateEntityOutcome,
64};
65
66pub use boot::{SchemaResolver, resolve_builtin_schema_pin_pub};
67
68/// One mem attachment, paired with the backend that serves it.
69/// Constructed by [`Engine::from_mounts`] and held internally.
70struct MountedBackend {
71 mount: Mount,
72 backend: Box<dyn MemBackend>,
73 /// Last cursor returned by `backend.current_head()`. Seeded in
74 /// [`Engine::from_mounts`]; refreshed by
75 /// [`Engine::reload_if_stale`] after a successful reload.
76 /// `None` means the backend doesn't track a head (folder /
77 /// archive) — drift detection is a no-op for this mount.
78 last_known_head: Option<String>,
79 /// Per-mem `.memstead/config.json` payload — surfaces via
80 /// [`Engine::mem_config_for`] for handlers that need
81 /// `write_guidance` / `extra` (`memstead_health
82 /// { include_config: true }`'s per-mem detail block).
83 ///
84 /// Loaded at construction for folder backends (read from
85 /// `<path>/.memstead/config.json`). Git-branch + archive backends
86 /// carry `None` for now — the read-from-storage-backend path
87 /// lifts in a follow-up session.
88 mem_config: Option<memstead_schema::config::MemConfig>,
89 /// Per-mem authoring-provenance payload read from the archive's
90 /// `.memstead/provenance.json` at construction (via
91 /// [`crate::backend::MemBackend::read_archive_provenance`]). `None`
92 /// when the backend carries no provenance member (a pre-provenance
93 /// archive, or a backend that does not surface one) — surfaced as
94 /// provenance-absent via [`Engine::archive_provenance_for`]. A
95 /// malformed payload is downgraded to `None` rather than failing the
96 /// mount: the member is additive.
97 archive_provenance: Option<memstead_schema::ArchiveProvenance>,
98}
99
100/// Unified engine. Holds a list of mounted backends and routes
101/// mem-named operations to the right one.
102///
103/// `Send` so the engine can sit behind a `Mutex` (today's pattern
104/// in the MCP server). The trait object's `Send + Sync` bound on
105/// `MemBackend` keeps the inner backends thread-safe; the engine
106/// itself is single-threaded by design (the lazy memos are
107/// `OnceCell`, which is `!Sync`).
108///
109/// `Debug` is hand-written to avoid requiring `Debug` on the
110/// `dyn MemBackend` trait object — backend impls are free to
111/// stay non-`Debug`.
112///
113/// ## Load-on-init
114///
115/// `Engine::from_mounts` walks each backend at construction time
116/// (`list_entities` + `read_entity` + parse) and populates a single
117/// shared [`Store`] with entities and edges from every mount. Each
118/// mount's schema resolves from its own pin (the backend config's
119/// schema, or the mount-record assertion as fallback) through the
120/// `SchemaResolver`, so `schemas` holds genuinely heterogeneous
121/// schemas in a multi-schema workspace. Per-file errors don't fail
122/// construction; they collect into [`Engine::load_errors`] for the
123/// operator to inspect.
124pub struct Engine {
125 mounts: Vec<MountedBackend>,
126 store: Store,
127 schemas: HashMap<String, Arc<Schema>>,
128 /// Workspace-authored schemas loaded from
129 /// `WorkspaceSettings.schemas_dir` at construction. Distinct from
130 /// `schemas` (per-mem, only schemas pinned by a mount): this
131 /// catalogue carries every workspace-loaded schema regardless of
132 /// whether a mem pins it. Surfaced via
133 /// [`Self::workspace_schemas`] for handlers that need to enumerate
134 /// schemas referenced by `mem_create_rules.schemas[]` but not
135 /// pinned by any mem — `memstead_overview` lists them in `## Schemas`
136 /// so an agent sees what could be pinned. Empty when no
137 /// `schemas_dir` was passed.
138 workspace_schemas: Vec<Arc<Schema>>,
139 /// Embedded built-in schemas loaded once at boot from
140 /// `memstead_schema::builtins::load_builtin_schemas()`. The boot path
141 /// uses this catalogue to resolve each mount's schema pin; storing
142 /// it on the engine lets read handlers (MCP's `memstead_schema`,
143 /// `memstead_overview`'s `## Schemas` rendering) surface every built-in
144 /// without re-walking the embedded directory. Schemas declared in
145 /// `workspace_schemas` shadow built-ins on `(name, version)`
146 /// collision — handlers walking both lists must check workspace
147 /// first.
148 builtin_schemas: Vec<Arc<Schema>>,
149 load_errors: Vec<(PathBuf, String)>,
150 /// Lazily-computed Louvain community detection across the
151 /// engine-wide store. Populated on first call to
152 /// [`Self::communities`]; invalidated by
153 /// [`Self::invalidate_communities`] which every mutation method
154 /// calls after a successful write. `OnceCell` is `!Sync`; the
155 /// engine is `Send` (it is moved into a `Mutex` by every consumer)
156 /// but not `Sync`.
157 community_memo: OnceCell<LouvainOutput>,
158 /// Lazily-computed per-mem search index map. Built on first call
159 /// to [`Self::search_indexes`] via [`build_all`]; invalidated by
160 /// [`Self::invalidate_search_indexes`] alongside the community
161 /// cache so every mutation triggers a fresh build on the next
162 /// search. Absent on `wasm32` targets — search lives behind the
163 /// bridge (see `EngineError::SearchUnavailable`).
164 #[cfg(not(target_arch = "wasm32"))]
165 search_indexes_memo: OnceCell<HashMap<String, MemIndex>>,
166 /// Workspace-level operator policy — mem create/delete rules,
167 /// cross-mem link permissions. Defaults to empty; populated via
168 /// [`Self::set_settings`] when [`Self::from_workspace_root`] (or
169 /// the full counterpart) reads `.memstead/workspace.toml`. Surfaced
170 /// read-only via [`Self::settings`] for MCP handlers and other
171 /// consumers.
172 settings: WorkspaceSettings,
173 /// Lazily-compiled [`crate::mem_management::CreateRuleSet`] over
174 /// `settings.mem_create_rules`. Built on first
175 /// [`Self::cross_mem_link_allowed`] call that needs synthesis;
176 /// invalidated by [`Self::set_settings`] (so a fresh policy
177 /// re-compiles on the next call). Compilation errors are logged
178 /// and the cache stays empty — synthesis is best-effort, the
179 /// resolver falls back to explicit-policy resolution. Operators
180 /// who want hard validation pre-compile via
181 /// [`crate::mem_management::CreateRuleSet::new`] before passing
182 /// settings.
183 create_rule_set_memo: OnceCell<crate::mem_management::CreateRuleSet>,
184 /// Workspace root path — set when the engine boots from a
185 /// workspace store ([`Self::from_workspace_root`] or the full
186 /// counterpart). `None` for tests + ad-hoc consumers that build
187 /// the engine directly from a mount list. Surfaced via
188 /// [`Self::workspace_root`] for handlers that need filesystem
189 /// context (e.g. [`Self::health`]'s outer-repo .gitignore
190 /// check).
191 workspace_root: Option<PathBuf>,
192 /// Typed warnings surfaced during mem load — drift findings
193 /// like [`WarningHint::SuspiciousNestedPrefix`] and
194 /// [`WarningHint::DuplicateSectionHeading`] that the loader
195 /// pipeline collects per entity. Empty for the V1 unified
196 /// engine; the field is in place so handlers and the health
197 /// surface can include them when the loader pipeline grows the
198 /// warning generators.
199 load_warnings: Vec<WarningHint>,
200 /// Pipeline configs (Medium / Facet / Projection / Ingest) loaded
201 /// from the workspace store at boot. Empty for engines built via
202 /// `from_mounts*` (tests, in-memory consumers) and for any workspace
203 /// that declares no pipelines; the workspace-root boot paths
204 /// (`from_workspace_root` and the full counterpart) populate it via
205 /// [`crate::pipeline_store::load_pipeline_configs`]. Read-only
206 /// runtime surface — exposed through [`Self::pipeline_configs`]; the
207 /// engine neither runs nor schedules pipelines (the ingest skill and
208 /// future consumers do).
209 pipeline_configs: crate::pipeline_store::PipelineConfigs,
210 /// Runtime snapshot of writable / visible mems. Derived from
211 /// the mount list at construction: writable mounts
212 /// (`MountCapability::Write`) register via `add_writable` with
213 /// the storage's directory path (folder → `path`, git-branch →
214 /// None, archive shouldn't be writable); read-only mounts
215 /// register via `add_writable` (folder/git-branch) or
216 /// `add_read_only` (archive). Used by MCP handlers that need the
217 /// writable/visible roster + per-mem origin (`memstead_health
218 /// include_config: true`, `memstead_overview`'s mem list,
219 /// `memstead_mem_create`'s collision check).
220 ///
221 /// Wrapped in `Arc` so the COW-snapshot discipline — clone the
222 /// snapshot, mutate the clone, swap the `Arc` — keeps writers
223 /// and concurrent readers from contending on the live mount
224 /// list.
225 mem_router: Arc<MemRouterSnapshot>,
226 /// Backend factory — function pointer used by
227 /// [`crate::mem_management::create_mem`] (and future runtime
228 /// mount-add paths) to materialise a [`MemBackend`] from a
229 /// [`Mount`] declaration. Defaults to
230 /// [`crate::workspace_store::instantiate_lean_backend`] so lean
231 /// (folder + archive only) consumers work out of the box. Full
232 /// consumers swap in `memstead_git_branch::storage::instantiate_full_backend`
233 /// via [`Self::set_backend_factory`] after constructing the engine —
234 /// `engine_from_workspace_root` does this once at boot. Function
235 /// pointer (not `Box<dyn Fn>`) because the backend factory is
236 /// stateless, `Send + Sync + Copy`, and one less allocation on the
237 /// hot path matters for the multi-mem pattern this engine is
238 /// designed around.
239 backend_factory: BackendFactory,
240 /// Git-branch ops bundle — function pointers for the per-mount
241 /// operations whose implementations live in `memstead-git-branch`
242 /// (and therefore can't sit on the `MemBackend` trait without
243 /// inverting the crate dependency). Full boot
244 /// (`memstead_git_branch::engine_from_workspace_root`) installs the
245 /// bundle via [`Self::set_git_branch_ops`]; lean consumers leave
246 /// it `None` and `Engine::changes_since` / `Engine::export_mem`
247 /// fall through to the folder/archive-only branches.
248 git_branch_ops: Option<GitBranchOps>,
249 /// Per-mem subscriber registry for [`MemChangedEvent`]s. Held
250 /// behind `Arc<Mutex<_>>` so [`SubscriptionHandle`]s — which own
251 /// the consumer's view of the subscription lifetime — can call
252 /// back into the registry on `Drop` without a self-reference cycle
253 /// to the engine. The emit path (in `record_self_write`) snapshots
254 /// the per-mem callback list under the lock, releases the lock,
255 /// and then invokes the callbacks — so a callback that re-enters
256 /// the engine for a read does not deadlock against the registry.
257 event_subscribers: Arc<std::sync::Mutex<events::SubscriberRegistry>>,
258 /// Reload-before-operation notices accumulated by
259 /// [`Self::reload_if_stale`] when an operation triggered a mem
260 /// reload. Built at reload time — when the backend's current head
261 /// equals the head we reloaded to, *before* any mutation in the
262 /// same operation commits — so the delta describes only the
263 /// sibling's change, never the engine's own follow-on write. The
264 /// response layer drains them via
265 /// [`Self::take_mem_changed_notices`] and attaches the structured
266 /// `mem_changed` notice to the operation's response. Every entity
267 /// op that can reload drains after; an undrained accumulation would
268 /// leak into the next operation's response, so callers that reload
269 /// must take.
270 pending_mem_changed: Vec<crate::ops::MemChangedNotice>,
271}
272
273/// Backend factory function pointer. Both flavours' existing
274/// `instantiate_*_backend` functions match this signature, so the
275/// type alias is what bridges the dependency direction (memstead-base
276/// can't depend on memstead-git-branch) without an extra trait.
277/// Stateless, `Send + Sync + Copy`.
278pub type BackendFactory =
279 fn(&Mount) -> Result<Box<dyn MemBackend>, crate::workspace_store::InstantiateError>;
280
281/// `Engine::changes_since` dispatch for git-branch mounts.
282///
283/// Signature matches `memstead_git_branch::ops::changes::changes_since` after
284/// adapting the `Store` parameter away (the engine performs enrichment
285/// downstream) and the `head_ref` parameter (`refs/heads/<branch>` is
286/// constructed inside the impl from `branch`).
287pub type GitBranchChangesSinceFn = fn(
288 gitdir: &Path,
289 branch: &str,
290 mem: &str,
291 since: &str,
292 rename_similarity: f32,
293) -> Result<crate::ops::BackendChanges, BackendError>;
294
295/// `Engine::export_mem` dispatch for git-branch mounts.
296///
297/// Signature mirrors `memstead_git_branch::ops::export::export_mem_from_branch`.
298pub type GitBranchExportFn = fn(
299 gitdir: &Path,
300 branch: &str,
301 mem: &str,
302 config: &memstead_schema::MemConfig,
303 output_path: &Path,
304 workspace_root: Option<&Path>,
305 workspace_schemas_dir: Option<&Path>,
306 // Engine-sourced authoring-provenance payload bytes (from the mount's
307 // `read_provenance` log) to embed at `.memstead/provenance.json`.
308 // `None` when the mem carried no noted mutations.
309 provenance_bytes: Option<&[u8]>,
310) -> Result<crate::ops::MemExportResult, BackendError>;
311
312/// `Engine::export_mem_to_bytes` dispatch for git-branch mounts.
313///
314/// Signature mirrors `memstead_git_branch::ops::export::export_mem_from_branch_to_bytes`.
315/// Symmetric to `GitBranchExportFn`: same inputs minus the output path,
316/// returns archive bytes plus metadata instead of writing to disk.
317pub type GitBranchExportToBytesFn = fn(
318 gitdir: &Path,
319 branch: &str,
320 mem: &str,
321 config: &memstead_schema::MemConfig,
322 workspace_root: Option<&Path>,
323 workspace_schemas_dir: Option<&Path>,
324 // Pre-built authoring-provenance payload bytes the engine sourced from
325 // the mount's `read_provenance` log, to embed at
326 // `.memstead/provenance.json`. `None` when the mem carried no noted
327 // mutations. The engine sources it (it holds the backend); the hook
328 // only embeds, keeping git history-walking out of the fn-pointer.
329 provenance_bytes: Option<&[u8]>,
330) -> Result<crate::ops::MemExportBytes, BackendError>;
331
332/// `Engine::diff` dispatch for git-branch mounts. Walks the two refs
333/// inside the workspace's mem-repo gitdir, produces a per-entity
334/// [`crate::ops::Diff`]. Refs are arbitrary `gix::rev_parse_single`
335/// inputs — branch names, commit SHAs, tag names. Resolves each
336/// independently so cross-branch (cross-mem) diffs work uniformly.
337pub type GitBranchDiffFn = fn(
338 gitdir: &Path,
339 mem: &str,
340 ref_a: &str,
341 ref_b: &str,
342 config: &crate::ops::DiffConfig,
343) -> Result<crate::ops::Diff, BackendError>;
344
345/// `Engine::fetch` dispatch for git-branch mounts.
346pub type GitBranchFetchFn = fn(
347 gitdir: &Path,
348 remote: &str,
349 refspecs: &[String],
350) -> Result<crate::ops::FetchOutcome, BackendError>;
351
352/// Read every `.md` blob at `ref_name` in `gitdir`, returning
353/// `(relative_path, utf8_content)` pairs. Skips `.memstead/` engine-internal
354/// entries and non-blob nodes. Used by the pre-merge schema-validation
355/// pass `Engine::pull` and `Engine::push` run before they advance the
356/// branch pointer / push to the remote.
357pub type GitBranchReadTreeFn =
358 fn(gitdir: &Path, ref_name: &str) -> Result<Vec<(String, String)>, BackendError>;
359
360/// `Engine::pull` dispatch for git-branch mounts.
361pub type GitBranchPullFn =
362 fn(gitdir: &Path, remote: &str, mem: &str) -> Result<crate::ops::PullOutcome, BackendError>;
363
364/// `Engine::push` dispatch for git-branch mounts.
365pub type GitBranchPushFn = fn(
366 gitdir: &Path,
367 remote: &str,
368 mem: &str,
369 force: bool,
370) -> Result<crate::ops::PushOutcome, BackendError>;
371
372/// `Engine::remote_add` dispatch — configures a named remote on the
373/// mem-repo gitdir (upsert: add, or set-url when it already exists).
374pub type GitBranchRemoteAddFn =
375 fn(gitdir: &Path, name: &str, url: &str) -> Result<crate::ops::RemoteAddOutcome, BackendError>;
376
377/// `Engine::branch_reset` dispatch for git-branch mounts. Returns the
378/// outcome on success; surfaces `BackendError::Other` carrying an
379/// in-band marker (`UNKNOWN_REF:<raw>` or
380/// `PUSHED_COMMITS_PROTECTED:<sha,sha,...>`) the engine layer
381/// un-marshals into typed `EngineError`s.
382pub type GitBranchBranchResetFn = fn(
383 gitdir: &Path,
384 branch: &str,
385 target_sha: &str,
386) -> Result<crate::ops::BranchResetOutcome, BackendError>;
387
388/// Residue-prune dispatch for git-branch mounts.
389/// The `create_mem` orchestrator calls this when
390/// `RecoveryAction::ForceOverwrite` is selected against pre-existing
391/// storage residue. Drops `refs/heads/<branch_full_path>` and the
392/// `__MEMSTEAD:mems/<branch_full_path>/config.json` blob in one
393/// ref-edit transaction (the same call the
394/// `MemBackend::delete_artifacts` impl wraps for delete-files
395/// flows). Surfaces as a function pointer so `memstead-engine` can
396/// drive a prune against an unmounted gitdir without depending on
397/// `memstead-git-branch`.
398pub type GitBranchPruneResidueFn =
399 fn(gitdir: &Path, branch_full_path: &str) -> Result<(), BackendError>;
400
401/// `Engine::install_schema` dispatch for the git-branch backend: write a
402/// schema package (`(relative-path, bytes)` pairs) onto the workspace's
403/// unified `__MEMSTEAD:schemas/<name>@<version>/` ref and return the
404/// resulting commit sha. Mirrors
405/// `memstead_git_branch::storage_memstead::write_schema_to_memstead_ref`.
406pub type GitBranchWriteSchemaFn = fn(
407 gitdir: &Path,
408 name: &str,
409 version: &str,
410 files: &[(String, Vec<u8>)],
411) -> Result<String, BackendError>;
412
413/// Bundle of git-branch-specific op dispatchers. Installed on the
414/// engine at full boot. Each field is one ops-method that previously
415/// lived on the `MemBackend` trait; moving them off the trait keeps
416/// the bytes-level primitive surface clean.
417#[derive(Clone, Copy)]
418pub struct GitBranchOps {
419 pub changes_since: GitBranchChangesSinceFn,
420 pub diff: GitBranchDiffFn,
421 pub branch_reset: GitBranchBranchResetFn,
422 pub fetch: GitBranchFetchFn,
423 pub pull: GitBranchPullFn,
424 pub push: GitBranchPushFn,
425 pub remote_add: GitBranchRemoteAddFn,
426 pub read_tree: GitBranchReadTreeFn,
427 pub export: GitBranchExportFn,
428 pub export_to_bytes: GitBranchExportToBytesFn,
429 pub prune_residue: GitBranchPruneResidueFn,
430 pub write_schema: GitBranchWriteSchemaFn,
431}
432
433impl std::fmt::Debug for Engine {
434 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
435 f.debug_struct("Engine")
436 .field(
437 "mems",
438 &self
439 .mounts
440 .iter()
441 .map(|m| m.mount.mem.as_str())
442 .collect::<Vec<_>>(),
443 )
444 .finish()
445 }
446}
447
448#[cfg(test)]
449mod in_memory_mem;
450
451#[cfg(test)]
452pub(super) mod test_helpers {
453 use std::io::Write as _;
454 use std::path::{Path, PathBuf};
455
456 use memstead_schema::SchemaRef;
457
458 use crate::backend::MemBackend;
459 use crate::storage::FilesystemMemWriter;
460 use crate::vcs::{Actor, ClientId};
461 use crate::workspace::{Mount, MountCapability, MountLifecycle, MountStorage};
462
463 use super::{CreateEntityArgs, CreateEntityOutcome, Engine, RelateEntityArgs};
464
465 use indexmap::IndexMap;
466 use tempfile::TempDir;
467
468 pub(crate) fn pin(name: &str) -> SchemaRef {
469 let version = match name {
470 "default" => semver::Version::new(1, 0, 0),
471 _ => semver::Version::new(0, 1, 0),
472 };
473 SchemaRef::new(name, version)
474 }
475
476 pub(crate) fn folder_mount(mem: &str, path: PathBuf) -> Mount {
477 Mount {
478 mem: mem.to_string(),
479 schema: Some(pin("default")),
480 storage: MountStorage::Folder { path },
481 capability: MountCapability::Write,
482 lifecycle: MountLifecycle::Eager,
483 cross_linkable: true,
484 migration_target: None,
485 }
486 }
487
488 pub(crate) fn in_memory_mount(mem: &str) -> Mount {
489 Mount {
490 mem: mem.to_string(),
491 schema: Some(pin("default")),
492 storage: MountStorage::InMemory,
493 capability: MountCapability::Write,
494 lifecycle: MountLifecycle::Eager,
495 cross_linkable: true,
496 migration_target: None,
497 }
498 }
499
500 pub(crate) fn archive_mount(mem: &str, path: PathBuf) -> Mount {
501 Mount {
502 mem: mem.to_string(),
503 schema: Some(pin("default")),
504 storage: MountStorage::Archive { path },
505 capability: MountCapability::ReadOnly,
506 lifecycle: MountLifecycle::Lazy,
507 cross_linkable: false,
508 migration_target: None,
509 }
510 }
511
512 /// Build a sealed archive at `tmp/<name>.mem` from
513 /// `(relative_path, bytes)` pairs and return the path.
514 pub(crate) fn build_archive(tmp: &Path, name: &str, entries: &[(&str, &[u8])]) -> PathBuf {
515 let path = tmp.join(format!("{name}.mem"));
516 let file = std::fs::File::create(&path).unwrap();
517 let mut writer = zip::ZipWriter::new(file);
518 let opts = zip::write::SimpleFileOptions::default();
519 for (rel, bytes) in entries {
520 writer.start_file(*rel, opts).unwrap();
521 writer.write_all(bytes).unwrap();
522 }
523 writer.finish().unwrap();
524 path
525 }
526
527 /// Write a schema manifest + minimal type bodies under
528 /// `<root>/<name>/`. Each type gets a body with a single
529 /// `body` section and `_default` hierarchy/propagation — enough
530 /// to load and parse markdown that uses that type. Used by tests
531 /// that need a custom schema with shape or vocabulary constraints.
532 pub(crate) fn write_schema_files_with_default_type(
533 root: &Path,
534 name: &str,
535 manifest: &str,
536 types: &[&str],
537 ) {
538 const TYPE_BODY: &str = r#"description: t
539when_to_use: Here
540sections:
541 - key: body
542 heading: Body
543 required: true
544 search_weight: 10.0
545 catch_all: true
546 write_rules: []
547metadata_fields: []
548title_weight: 100.0
549text_fields:
550 - body
551hierarchy_relationship: _default
552propagating_relationships: []
553updatable_fields:
554 - title
555 - body
556health_required_fields:
557 - body
558staleness_threshold_days: 90
559write_rules: []
560"#;
561 let dir = root.join(name);
562 std::fs::create_dir_all(dir.join("types")).unwrap();
563 std::fs::write(dir.join("schema.yaml"), manifest).unwrap();
564 for type_name in types {
565 let body = format!("name: {type_name}\n{TYPE_BODY}");
566 std::fs::write(dir.join("types").join(format!("{type_name}.yaml")), body).unwrap();
567 }
568 }
569
570 pub(crate) fn empty_create_args(mem: &str, title: &str) -> CreateEntityArgs {
571 // The
572 // create path refuses on missing required sections. The
573 // default `spec` type requires `identity` + `purpose`. Seed
574 // both with a single space so the test fixture remains a
575 // valid creation request — every test that uses this helper
576 // as a fixture builder continues to work, and tests that
577 // specifically exercise the refusal supply an explicit
578 // empty-sections payload (see the dedicated refusal tests).
579 let mut sections = IndexMap::new();
580 sections.insert("identity".to_string(), "fixture identity body".to_string());
581 sections.insert("purpose".to_string(), "fixture purpose body".to_string());
582 CreateEntityArgs {
583 mem: mem.to_string(),
584 title: title.to_string(),
585 entity_type: "spec".to_string(),
586 sections,
587 metadata: IndexMap::new(),
588 relations: Vec::new(),
589 dry_run: false,
590 }
591 }
592
593 pub(crate) fn cli_actor() -> (Actor, ClientId) {
594 (
595 Actor::Cli,
596 ClientId {
597 name: "claude-code".to_string(),
598 version: "2.1.0".to_string(),
599 },
600 )
601 }
602
603 pub(crate) fn engine_with_seed(tmp: &TempDir, title: &str) -> (Engine, CreateEntityOutcome) {
604 let mem_dir = tmp.path().to_path_buf();
605 let writer = FilesystemMemWriter::new(mem_dir.clone());
606 let mut engine = Engine::from_mounts(vec![(
607 folder_mount("specs", mem_dir),
608 Box::new(writer) as Box<dyn MemBackend>,
609 )])
610 .unwrap();
611 let (actor, client) = cli_actor();
612 let outcome = engine
613 .create_entity(
614 empty_create_args("specs", title),
615 actor,
616 Some(&client),
617 None,
618 )
619 .unwrap();
620 (engine, outcome)
621 }
622 pub(crate) fn build_demo_engine(tmp: &TempDir) -> Engine {
623 let mem_dir = tmp.path().to_path_buf();
624 let writer = FilesystemMemWriter::new(mem_dir.clone());
625 let mut engine = Engine::from_mounts(vec![(
626 folder_mount("specs", mem_dir),
627 Box::new(writer) as Box<dyn MemBackend>,
628 )])
629 .unwrap();
630 let (actor, client) = cli_actor();
631 let source = engine
632 .create_entity(
633 empty_create_args("specs", "Source One"),
634 actor,
635 Some(&client),
636 None,
637 )
638 .unwrap();
639 let target = engine
640 .create_entity(
641 empty_create_args("specs", "Target Two"),
642 actor,
643 Some(&client),
644 None,
645 )
646 .unwrap();
647 engine
648 .create_entity(
649 empty_create_args("specs", "Lonely Three"),
650 actor,
651 Some(&client),
652 None,
653 )
654 .unwrap();
655 engine
656 .relate_entity(
657 RelateEntityArgs {
658 source: source.id.clone(),
659 expected_hash: Some(source.content_hash.clone()),
660 rel_type: "USES".to_string(),
661 target: target.id.clone(),
662 remove: false,
663 description: None,
664 },
665 actor,
666 Some(&client),
667 None,
668 )
669 .unwrap();
670 engine
671 }
672}