Merge branch 'worktree-agent-ad3aff0dffd17a6eb'

tests/unit/brainy/add.test.ts

@@ -335,15 +335,24 @@ describe('Brainy.add()', () => {
   })
   
   describe('edge cases', () => {
-    it('should reject empty string as data', async () => {
-      // Arrange
+    it('should accept an empty string as real (empty) data', async () => {
+      // Arrange — '' is legitimate content (e.g. an empty file's first
+      // write), not a missing field. Only null/undefined data (with no
+      // vector either) is "missing" — see the separate
+      // 'data and vector are both missing' test above.
       const params = createAddParams({
         data: '',
         type: 'thing'
       })
-      
-      // Act & Assert - Empty string is not valid data
-      await expect(brain.add(params)).rejects.toThrow('Invalid add() parameters: Missing required field \'data\'')
+
+      // Act
+      const id = await brain.add(params)
+
+      // Assert — stored and readable back as empty, not rejected
+      expect(id).toBeDefined()
+      const entity = await brain.get(id)
+      expect(entity).not.toBeNull()
+      expect(entity!.data).toBe('')
     })
     
     it('should handle very long text content', async () => {

tests/unit/brainy/open-path.test.ts

@@ -0,0 +1,199 @@
+/**
+ * OPEN-PATH tests: init() must never gate on the embedding model, the VFS
+ * root bootstrap must never touch the embedding engine, and a slow open
+ * must narrate its phases.
+ *
+ * Background: a production restart storm measured 90,017ms for a single
+ * brain init vs 1,117ms quiet — an ~80x contention multiplier — traced to
+ * every writer's init() eagerly awaiting the process-global WASM embedding
+ * engine before the VFS root even existed. See src/brainy.ts performInit()
+ * and src/vfs/VirtualFileSystem.ts doInitializeRoot().
+ */
+
+import { describe, it, expect, vi } from 'vitest'
+import { Brainy } from '../../../src/brainy'
+import { MemoryStorage } from '../../../src/storage/adapters/memoryStorage'
+import { embeddingManager } from '../../../src/embeddings/EmbeddingManager'
+import { createTestConfig } from '../../helpers/test-factory'
+
+/**
+ * The four signals `isDeterministicEmbedMode()` checks (see
+ * src/embeddings/deterministicEmbedMode.ts). The global unit-test setup
+ * (tests/setup-unit.ts) sets some of these for the whole file/run in some
+ * vitest configurations; other configurations leave them unset and run the
+ * real WASM engine instead. The background-warm tests below need the
+ * "not unit-test mode" branch of performInit() to actually execute, so they
+ * save/clear/restore all four explicitly — deterministic regardless of
+ * which config invoked this file, never relying on ambient state.
+ */
+function withRealEmbedderBranch<T>(fn: () => Promise<T>): Promise<T> {
+  const savedEnvDeterministic = process.env.BRAINY_DETERMINISTIC_EMBEDDINGS
+  const savedEnvUnitTest = process.env.BRAINY_UNIT_TEST
+  const g = globalThis as Record<string, unknown>
+  const savedGlobalDeterministic = g.__BRAINY_DETERMINISTIC_EMBED__
+  const savedGlobalUnitTest = g.__BRAINY_UNIT_TEST__
+
+  delete process.env.BRAINY_DETERMINISTIC_EMBEDDINGS
+  delete process.env.BRAINY_UNIT_TEST
+  delete g.__BRAINY_DETERMINISTIC_EMBED__
+  delete g.__BRAINY_UNIT_TEST__
+
+  const restore = () => {
+    if (savedEnvDeterministic !== undefined) process.env.BRAINY_DETERMINISTIC_EMBEDDINGS = savedEnvDeterministic
+    if (savedEnvUnitTest !== undefined) process.env.BRAINY_UNIT_TEST = savedEnvUnitTest
+    if (savedGlobalDeterministic !== undefined) g.__BRAINY_DETERMINISTIC_EMBED__ = savedGlobalDeterministic
+    if (savedGlobalUnitTest !== undefined) g.__BRAINY_UNIT_TEST__ = savedGlobalUnitTest
+  }
+
+  return fn().finally(restore)
+}
+
+/**
+ * A MemoryStorage whose init() takes an artificially long time — a
+ * controllable fake seam (not a wall-clock race) that reliably pushes
+ * performInit()'s "storage-init" phase (and therefore the total open time)
+ * past the 2000ms narration threshold, without touching the filesystem or
+ * relying on real contention.
+ */
+class SlowMemoryStorage extends MemoryStorage {
+  override async init(): Promise<void> {
+    await new Promise((resolve) => setTimeout(resolve, 2200))
+    await super.init()
+  }
+}
+
+describe('OPEN-PATH: init() never gates on the embedding model', () => {
+  it('bootstrapping a fresh store never calls the embedding engine (VFS root add is engine-untouched)', async () => {
+    const embedSpy = vi.spyOn(embeddingManager, 'embed')
+    const brain = new Brainy(createTestConfig())
+    try {
+      await brain.init()
+
+      // The VFS root's add() must never have reached the embedding engine —
+      // it carries an explicit placeholder vector instead (see
+      // VirtualFileSystem.doInitializeRoot()).
+      expect(embedSpy).not.toHaveBeenCalled()
+
+      // Sanity: the VFS is genuinely usable afterwards.
+      const files = await brain.vfs.readdir('/')
+      expect(files).toEqual([])
+    } finally {
+      await brain.close()
+      embedSpy.mockRestore()
+    }
+  })
+
+  it('starts the embedding-engine warm in the BACKGROUND — init() resolves before the warm does', async () => {
+    await withRealEmbedderBranch(async () => {
+      const events: string[] = []
+      let releaseWarm!: () => void
+      const warmGate = new Promise<void>((resolve) => {
+        releaseWarm = resolve
+      })
+
+      const initSpy = vi.spyOn(embeddingManager, 'init').mockImplementation(async () => {
+        events.push('warm-start')
+        await warmGate
+        events.push('warm-resolve')
+      })
+
+      const brain = new Brainy(createTestConfig())
+      try {
+        await brain.init()
+        events.push('init-resolved')
+
+        // init() started the warm but returned WITHOUT waiting for it.
+        expect(initSpy).toHaveBeenCalledTimes(1)
+        expect(events).toEqual(['warm-start', 'init-resolved'])
+
+        // Now let the fake warm finish and confirm it lands strictly after.
+        releaseWarm()
+        await new Promise((resolve) => setTimeout(resolve, 0))
+        expect(events).toEqual(['warm-start', 'init-resolved', 'warm-resolve'])
+      } finally {
+        await brain.close()
+        initSpy.mockRestore()
+      }
+    })
+  })
+
+  it('narrates a background warm FAILURE loudly instead of losing it silently', async () => {
+    await withRealEmbedderBranch(async () => {
+      const warnSpy = vi.spyOn(console, 'warn').mockImplementation(() => {})
+      const initSpy = vi
+        .spyOn(embeddingManager, 'init')
+        .mockRejectedValue(new Error('simulated cold-compile failure'))
+
+      const brain = new Brainy(createTestConfig())
+      try {
+        // init() itself must still resolve — a failed background warm is
+        // never fatal to open().
+        await expect(brain.init()).resolves.toBeUndefined()
+
+        // Give the background .catch() a microtask/macrotask to run.
+        await new Promise((resolve) => setTimeout(resolve, 0))
+
+        const failureLine = warnSpy.mock.calls
+          .map((args) => args.map(String).join(' '))
+          .find((line) => line.includes('background embedding-engine warm FAILED'))
+        expect(failureLine).toBeDefined()
+        expect(failureLine).toContain('simulated cold-compile failure')
+      } finally {
+        await brain.close()
+        initSpy.mockRestore()
+        warnSpy.mockRestore()
+      }
+    })
+  })
+
+  it('eagerEmbeddings: false starts no warm at all', async () => {
+    await withRealEmbedderBranch(async () => {
+      const initSpy = vi.spyOn(embeddingManager, 'init')
+      const brain = new Brainy({ ...createTestConfig(), eagerEmbeddings: false })
+      try {
+        await brain.init()
+        expect(initSpy).not.toHaveBeenCalled()
+      } finally {
+        await brain.close()
+        initSpy.mockRestore()
+      }
+    })
+  })
+
+  it('narrates a slow open with a per-phase ms breakdown once total time exceeds 2000ms', async () => {
+    const warnSpy = vi.spyOn(console, 'warn').mockImplementation(() => {})
+    const brain = new Brainy({ ...createTestConfig(), storage: new SlowMemoryStorage() })
+    try {
+      await brain.init()
+
+      const slowOpenLine = warnSpy.mock.calls
+        .map((args) => args.map(String).join(' '))
+        .find((line) => line.includes('[Brainy] slow open:'))
+
+      expect(slowOpenLine).toBeDefined()
+      expect(slowOpenLine).toContain('storage-init=')
+      expect(slowOpenLine).toContain('generation-store-open-fold=')
+      expect(slowOpenLine).toContain('index-init-gate=')
+      expect(slowOpenLine).toContain('vfs-bootstrap=')
+      expect(slowOpenLine).toContain('embedding-warm-started=')
+    } finally {
+      await brain.close()
+      warnSpy.mockRestore()
+    }
+  }, 20000)
+
+  it('stays silent about phase timing when open is fast (under 2000ms)', async () => {
+    const warnSpy = vi.spyOn(console, 'warn').mockImplementation(() => {})
+    const brain = new Brainy(createTestConfig())
+    try {
+      await brain.init()
+      const slowOpenLine = warnSpy.mock.calls
+        .map((args) => args.map(String).join(' '))
+        .find((line) => line.includes('[Brainy] slow open:'))
+      expect(slowOpenLine).toBeUndefined()
+    } finally {
+      await brain.close()
+      warnSpy.mockRestore()
+    }
+  })
+})

tests/unit/utils/paramValidation.test.ts

@@ -149,7 +149,33 @@ describe('Zero-Config Parameter Validation', () => {
         type: NounType.Document
       } as AddParams)).toThrow('Invalid add() parameters: Missing required field \'data\'')
     })
-    
+
+    it('should accept an empty string as real data — only null/undefined is "missing"', () => {
+      // A legitimate empty file's first write: '' is content, not absence.
+      expect(() => validateAddParams({
+        data: '',
+        type: NounType.Document
+      })).not.toThrow()
+
+      // null/undefined (with no vector) is still the genuine missing-field case.
+      expect(() => validateAddParams({
+        data: null as any,
+        type: NounType.Document
+      })).toThrow('Invalid add() parameters: Missing required field \'data\'')
+      expect(() => validateAddParams({
+        data: undefined,
+        type: NounType.Document
+      })).toThrow('Invalid add() parameters: Missing required field \'data\'')
+    })
+
+    it('deferEmbedding accepts empty-string data (real content, not absence)', () => {
+      expect(() => validateAddParams({
+        data: '',
+        type: NounType.Document,
+        deferEmbedding: true
+      } as AddParams)).not.toThrow()
+    })
+
     it('should validate NounType', () => {
       expect(() => validateAddParams({
         data: 'test',
@@ -190,7 +216,22 @@ describe('Zero-Config Parameter Validation', () => {
         id: 'test-id'
       })).toThrow('must specify at least one field to update')
     })
-    
+
+    it('empty-string data counts as a real field to update (truncating content)', () => {
+      expect(() => validateUpdateParams({
+        id: 'test-id',
+        data: ''
+      })).not.toThrow()
+    })
+
+    it('deferEmbedding accepts empty-string data on update', () => {
+      expect(() => validateUpdateParams({
+        id: 'test-id',
+        data: '',
+        deferEmbedding: true
+      } as UpdateParams)).not.toThrow()
+    })
+
     it('should validate NounType if changing', () => {
       expect(() => validateUpdateParams({
         id: 'test-id',

tests/unit/vfs-readdir-recursive.test.ts

@@ -0,0 +1,99 @@
+/**
+ * vfs.readdir()'s `recursive` option: typed since 7.30 but never read, so it
+ * silently behaved exactly like `recursive: false`. This pins the real,
+ * documented contract: a recursive listing returns every descendant (files
+ * AND directories, all depths) as paths RELATIVE TO THE QUERIED DIRECTORY —
+ * the same convention Node's `fs.readdir(dir, { recursive: true })` uses —
+ * for both the plain string-array form and the `withFileTypes` VFSDirent
+ * form (whose `name` carries that same relative path when recursive).
+ */
+
+import { describe, it, expect, beforeEach, afterEach } from 'vitest'
+import { Brainy } from '../../src/brainy.js'
+import type { VFSDirent } from '../../src/vfs/types.js'
+
+describe('vfs.readdir() recursive option', () => {
+  let brain: Brainy
+
+  beforeEach(async () => {
+    brain = new Brainy({
+      requireSubtype: false,
+      storage: { type: 'memory' },
+      silent: true
+    })
+    await brain.init()
+
+    // Build:
+    //   /a/b.txt
+    //   /a/sub/c.txt
+    //   /a/sub/deeper/d.txt
+    //   /a/sub2/            (empty directory)
+    await brain.vfs.writeFile('/a/b.txt', 'B')
+    await brain.vfs.writeFile('/a/sub/c.txt', 'C')
+    await brain.vfs.writeFile('/a/sub/deeper/d.txt', 'D')
+    await brain.vfs.mkdir('/a/sub2', { recursive: true })
+  })
+
+  afterEach(async () => {
+    await brain.close()
+  })
+
+  it('non-recursive (default) still returns only direct children, by basename', async () => {
+    const entries = await brain.vfs.readdir('/a') as string[]
+    expect([...entries].sort()).toEqual(['b.txt', 'sub', 'sub2'])
+  })
+
+  it('recursive: true returns every descendant as a path relative to the queried directory', async () => {
+    const entries = await brain.vfs.readdir('/a', { recursive: true }) as string[]
+    expect([...entries].sort()).toEqual([
+      'b.txt',
+      'sub',
+      'sub/c.txt',
+      'sub/deeper',
+      'sub/deeper/d.txt',
+      'sub2'
+    ])
+  })
+
+  it('recursive: true at the root has no leading slash on relative entries', async () => {
+    const entries = await brain.vfs.readdir('/', { recursive: true }) as string[]
+    expect(entries).toContain('a')
+    expect(entries).toContain('a/b.txt')
+    expect(entries).toContain('a/sub/deeper/d.txt')
+    for (const entry of entries) {
+      expect(entry.startsWith('/')).toBe(false)
+    }
+  })
+
+  it('recursive + withFileTypes: VFSDirent.name is the relative path, .path stays absolute', async () => {
+    const entries = await brain.vfs.readdir('/a', {
+      recursive: true,
+      withFileTypes: true
+    }) as VFSDirent[]
+
+    const byName = new Map(entries.map((e) => [e.name, e]))
+
+    const nested = byName.get('sub/deeper/d.txt')
+    expect(nested).toBeDefined()
+    expect(nested!.path).toBe('/a/sub/deeper/d.txt')
+    expect(nested!.type).toBe('file')
+
+    const nestedDir = byName.get('sub/deeper')
+    expect(nestedDir).toBeDefined()
+    expect(nestedDir!.path).toBe('/a/sub/deeper')
+    expect(nestedDir!.type).toBe('directory')
+
+    // Non-recursive VFSDirent behavior is unchanged: name is the basename.
+    const direct = await brain.vfs.readdir('/a', { withFileTypes: true }) as VFSDirent[]
+    const directEntry = direct.find((e) => e.path === '/a/b.txt')
+    expect(directEntry?.name).toBe('b.txt')
+  })
+
+  it('recursive + filter composes: only files survive a type filter', async () => {
+    const entries = await brain.vfs.readdir('/a', {
+      recursive: true,
+      filter: { type: 'file' }
+    }) as string[]
+    expect([...entries].sort()).toEqual(['b.txt', 'sub/c.txt', 'sub/deeper/d.txt'])
+  })
+})