Files
Jarvis/packages/Memory/Tests/MemoryTests/MemoryStoreApplyOpTests.swift
james.maesandClaude Opus 4.7 cb584a2ac8 feat(d5d6): real memory database — vec0 statically linked, hard-fail boot, stats tool
Closes D-5 / D-6 (sqlite-vec dylib bundling) carry-forwards from the
2026-05-04 audit by adopting jkrukowski/SQLiteVec as a SPM dependency.
The package's CSQLiteVec target compiles sqlite3.c + sqlite-vec.c into
one statically-linked C target with `core_vec_init()` registering vec0
as an auto-extension. All sqlite3_* calls in Replay's SQLiteConnection
and Memory's MemoryStore resolve to this bundled SQLite (vec0 always
available) instead of Apple's stripped /usr/lib/libsqlite3.dylib.

The retired hand-rolled approach (dlsym pathway + JARVIS_VEC0_STUB_PATH
env-flag + Bundle.module placeholder text file) never worked on macOS
because Apple stripped sqlite3_load_extension from the system libsqlite3.
Memory init has been silently dormant since Phase 7 (2026-04 onward),
which is what made B-02 (history threading) appear permanently broken
to the user during dogfooding even after the 48ad8d6 tactical patch
landed.

Live verification 2026-05-11: fresh launch with empty jarvis.db creates
schema, persists turn rows across submits, B-02 multi-turn context
resolves correctly ("what breed is Toby?" → "Golden Retriever" after
prior turn established Toby is a Golden Retriever), and the new
get_memory_stats tool reports `turnsCount=6, vecVersion=v0.1.6,
sqliteVersion=3.51.0`.

Changes:

- packages/Replay/Package.swift
    + SQLiteVec SPM dependency (from: "0.0.14")
    + CSQLiteVec product wired into Replay target
- packages/Replay/Sources/Replay/SQLiteConnection.swift
    - import SQLite3
    + @_exported import CSQLiteVec
    All sqlite3_* call sites unchanged — linker now resolves to our
    bundled sqlite3 with vec0 statically linked in.
- packages/Memory/Package.swift
    + SQLiteVec SPM dependency (named product so MemoryStore can import
      CSQLiteVec directly for the core_vec_init() one-shot call)
    - resources: PLACEHOLDER.txt copy (defunct vec0.dylib stub slot)
- packages/Memory/Sources/Memory/MemoryStore.swift
    - import SQLite3 → import CSQLiteVec
    + _vecAutoExtensionRegistered static-let one-shot: calls
      CSQLiteVec.core_vec_init() which registers vec0 via
      sqlite3_auto_extension on the first MemoryStore construction.
      Idempotent (static-let initializer fires exactly once).
    - loadVecExtension(on:) dlsym/JARVIS_VEC0_STUB_PATH/Bundle.module
      pathway (retired; ~70 lines removed)
    Updated docstring to reflect the new hard-fail boot policy.
- packages/Memory/Sources/Memory/Resources/PLACEHOLDER.txt — deleted
- App/AppDelegate.swift installMemory()
    Reverses Track-D D-1 (75a10be, 2026-05-04): MemoryStore.init failures
    are now FATAL via Logger.critical + TCCAlertService.presentHardBlock
    + NSApp.terminate. The graceful-degradation path that kept the app
    booting with memoryStore=nil hid the entire B-02/B-04/B-05 silent-
    failure bug class. Aligns with the 2026-05-07 API-first pivot
    thesis: "no silent failures".
    `store` is now non-optional throughout installMemory; downstream
    callers no longer need `if let store = store`.
- packages/MCP/Sources/MCP/InProcess/GetMemoryStatsTool.swift (new)
    + MemoryStatsDispatching protocol
    + MemoryStats Codable struct (turnsCount, factsTotal, factsActive,
      sessionsCount, dbFileBytes, vecVersion, sqliteVersion,
      lastTurnAtUnixMs, lastFactAtUnixMs)
    + GetMemoryStatsTool conforming to InProcessTool
- App/MCP/InProcessMemoryAdapters.swift
    + MemoryStatsStoreAdapter — bridges the protocol to MemoryStore's
      public queryRowCount / querySingleString seams and FileManager
      for the on-disk file size. Read-only, no confirmation.
- App/AppDelegate.swift installSelfKnowledgeTools()
    + Registers GetMemoryStatsTool with MemoryStatsStoreAdapter.
- packages/Memory/Tests/MemoryTests/*
    Retired the JARVIS_VEC0_STUB_PATH XCTSkipUnless guards from 9 tests
    (vec0 is always available now — no need to env-gate). Rewrote
    MemoryStoreTests to drop the now-impossible "init throws when vec
    missing" case and assert facts_vec virtual table creation +
    vec_version() readback. Added seedTurnRow helper to
    MemoryStoreApplyOpTests so the now-enforced
    `facts.source_turn_id REFERENCES turns(id)` foreign key is honored
    in tests that previously passed when FK enforcement was dormant.

Test summary: Memory full suite 91 tests, 10 skipped (env-gated real-
Ollama tests), 0 failures. App target builds clean. Substrate verified
end-to-end via live launch (vec_version: v0.1.6 logged at boot;
get_memory_stats returns actual schema row counts).

Pending D-7 carry-forward: ollama pull nomic-embed-text (~270MB) for
embedder; user already has qwen2.5-coder:32b-instruct-q8_0 which is
a higher-quality variant of the project's hardcoded
qwen2.5-coder:32b — small ModelID config delta to follow.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-11 10:25:56 -05:00

230 lines
8.9 KiB
Swift

import XCTest
import Foundation
import AgentCore
import Replay
@testable import Memory
/// Plan 07-02 Task 1: applyOp + MemoryOp parsing + ReplayEvent.memoryMutation
/// round-trip. The DB-roundtrip tests are env-gated (JARVIS_VEC0_STUB_PATH);
/// the parse + encoded() tests run unconditionally.
final class MemoryStoreApplyOpTests: XCTestCase {
// MARK: - Mock sink
final class MockReplayLog: MemoryReplaySink, @unchecked Sendable {
let lock = NSLock()
struct Recorded {
let event: ReplayEvent
let triggerTurnId: Int64
}
private var _recorded: [Recorded] = []
var recorded: [Recorded] {
lock.lock(); defer { lock.unlock() }
return _recorded
}
func record(_ event: ReplayEvent, forTriggerTurnId triggerTurnId: Int64) {
lock.lock(); defer { lock.unlock() }
_recorded.append(Recorded(event: event, triggerTurnId: triggerTurnId))
}
}
// MARK: - Test 1: ADD inserts a fact
func testApplyOpADDInsertsFact() async throws {
let tmp = try Self.tempDB()
let store = try MemoryStore(databaseURL: tmp)
let mock = MockReplayLog()
await store.setReplayLog(mock)
// D-5/D-6 closure (2026-05-11): vec0 is now statically linked, schema
// migration applies cleanly, and `PRAGMA foreign_keys=ON` is honored.
// facts.source_turn_id REFERENCES turns(id) — must seed a turn row
// before applyOp passes sourceTurnId: 1.
try await Self.seedTurnRow(store, id: 1)
let fact = try await store.applyOp(
.add(subject: "Sarah", predicate: "works_at", object: "Acme", embedding: nil),
sourceTurnId: 1
)
XCTAssertNotNil(fact)
guard let fact else { return }
XCTAssertGreaterThan(fact.id, 0)
XCTAssertEqual(fact.subject, "Sarah")
XCTAssertNil(fact.validTo)
XCTAssertNil(fact.supersededBy)
let activeCount = try await store.queryRowCount(
"SELECT id FROM facts WHERE valid_to IS NULL AND subject='Sarah'"
)
XCTAssertEqual(activeCount, 1)
}
// MARK: - Test 2: UPDATE closes prior + inserts new
func testApplyOpUPDATEClosesPriorAndInsertsNew() async throws {
let tmp = try Self.tempDB()
let store = try MemoryStore(databaseURL: tmp)
// D-5/D-6 FK enforcement: two turn rows for sourceTurnId 1 and 2.
try await Self.seedTurnRow(store, id: 1)
try await Self.seedTurnRow(store, id: 2)
let prior = try await store.applyOp(
.add(subject: "Sarah", predicate: "works_at", object: "Acme", embedding: nil),
sourceTurnId: 1
)
XCTAssertNotNil(prior)
guard let prior else { return }
let new = try await store.applyOp(
.update(supersedes: prior.id, subject: "Sarah", predicate: "works_at", object: "Globex", embedding: nil),
sourceTurnId: 2
)
XCTAssertNotNil(new)
guard let new else { return }
XCTAssertGreaterThan(new.id, prior.id)
// Prior row's valid_to is set, superseded_by points to new.
let activeNow = try await store.queryRowCount(
"SELECT id FROM facts WHERE valid_to IS NULL AND subject='Sarah'"
)
XCTAssertEqual(activeNow, 1, "Exactly one active 'Sarah' fact after supersede.")
let priorClosed = try await store.queryRowCount(
"SELECT id FROM facts WHERE id=\(prior.id) AND valid_to IS NOT NULL AND superseded_by=\(new.id)"
)
XCTAssertEqual(priorClosed, 1, "Prior row must be closed and linked to the new row.")
}
// MARK: - Test 4: bad UPDATE rolls back
func testApplyOpRollsBackOnFailure() async throws {
let tmp = try Self.tempDB()
let store = try MemoryStore(databaseURL: tmp)
let beforeCount = try await store.queryRowCount("SELECT id FROM facts")
// supersedes id 9999 does not exist; UPDATE affects 0 rows; the
// applyOp transaction must roll back leaving facts unchanged.
do {
_ = try await store.applyOp(
.update(supersedes: 9999, subject: "Sarah", predicate: "works_at", object: "Globex", embedding: nil),
sourceTurnId: 7
)
XCTFail("Expected applyOpFailed when supersedes points at nonexistent id.")
} catch let MemoryError.applyOpFailed {
// expected
} catch {
XCTFail("Expected applyOpFailed, got \(error)")
}
let afterCount = try await store.queryRowCount("SELECT id FROM facts")
XCTAssertEqual(afterCount, beforeCount, "Rollback must leave row count unchanged.")
}
// MARK: - Test 5: NOOP records nothing
func testApplyOpEmitsNothingForNOOP() async throws {
// Construct a MemoryStore-less path: applyOp(.noop) returns nil
// before any SQL — but to keep the sink injection chain end-to-end
// we still need a store. Skip if vec0 unavailable.
let tmp = try Self.tempDB()
let store = try MemoryStore(databaseURL: tmp)
let mock = MockReplayLog()
await store.setReplayLog(mock)
let result = try await store.applyOp(.noop, sourceTurnId: 7)
XCTAssertNil(result)
XCTAssertTrue(mock.recorded.isEmpty,
"NOOPs are debug-only per RESEARCH §13; no replay event must be emitted.")
}
// MARK: - Test 6: ReplayEvent.memoryMutation kind round-trip
func testReplayEventMemoryMutationKindRoundtrip() {
let bytes = Data("{\"op\":\"ADD\",\"subject\":\"x\"}".utf8)
let event = ReplayEvent.memoryMutation(bytes)
let (kind, payload) = event.encoded()
XCTAssertEqual(kind, "memory_mutation")
XCTAssertEqual(payload, bytes)
}
// MARK: - Tests 7-8: parseApplyMemoryOps
func testParseApplyMemoryOpsHappyPath() throws {
let json: [String: Any] = ["ops": [
["op": "ADD", "subject": "Sarah", "predicate": "works_at", "object": "Acme"],
["op": "UPDATE", "supersedes_fact_id": 17, "subject": "Sarah", "predicate": "works_at", "object": "Globex"],
["op": "NOOP"],
]]
let data = try JSONSerialization.data(withJSONObject: json)
let ops = try MemoryOp.parseApplyMemoryOps(data)
XCTAssertEqual(ops.count, 3)
guard case let .add(s, p, o, _) = ops[0] else {
return XCTFail("expected ADD at [0], got \(ops[0])")
}
XCTAssertEqual(s, "Sarah")
XCTAssertEqual(p, "works_at")
XCTAssertEqual(o, "Acme")
guard case let .update(supersedes, s2, _, o2, _) = ops[1] else {
return XCTFail("expected UPDATE at [1], got \(ops[1])")
}
XCTAssertEqual(supersedes, 17)
XCTAssertEqual(s2, "Sarah")
XCTAssertEqual(o2, "Globex")
guard case .noop = ops[2] else {
return XCTFail("expected NOOP at [2], got \(ops[2])")
}
}
func testParseApplyMemoryOpsDropsInvalid() throws {
let json: [String: Any] = ["ops": [
// empty subject — dropped
["op": "ADD", "subject": "", "predicate": "p", "object": "o"],
// missing supersedes_fact_id — dropped
["op": "UPDATE", "subject": "x", "predicate": "y", "object": "z"],
// unknown op — dropped
["op": "FROBNICATE"],
]]
let data = try JSONSerialization.data(withJSONObject: json)
let ops = try MemoryOp.parseApplyMemoryOps(data)
XCTAssertTrue(ops.isEmpty, "Invalid entries must be silently dropped.")
}
func testParseApplyMemoryOpsThrowsOnMissingTopLevel() {
// missing top-level "ops" — throws extractorInvalidArgs
let bad = Data("{\"foo\":1}".utf8)
XCTAssertThrowsError(try MemoryOp.parseApplyMemoryOps(bad)) { err in
guard case MemoryError.extractorInvalidArgs = err else {
return XCTFail("Expected extractorInvalidArgs, got \(err)")
}
}
}
// MARK: - helpers
private static func tempDB() throws -> URL {
let tmp = FileManager.default.temporaryDirectory
.appendingPathComponent("memorytests-\(UUID().uuidString)", isDirectory: true)
try FileManager.default.createDirectory(at: tmp, withIntermediateDirectories: true)
return tmp.appendingPathComponent("jarvis.db")
}
/// D-5/D-6 helper: insert a placeholder turn row so applyOp's
/// `facts.source_turn_id REFERENCES turns(id)` foreign key resolves.
/// Returns when the row's auto-incremented id matches the requested
/// `id` — relies on inserts happening in order on a fresh DB.
fileprivate static func seedTurnRow(_ store: MemoryStore, id: Int64) async throws {
try await store.appendTurn(
sessionId: "applyop-test-seed",
role: "user",
content: "seed-\(id)",
source: "test",
createdAt: id
)
}
}