brainy/tests/pagination.test.ts
David Snelling f898f0ce7b feat\!: migrate from TensorFlow.js to Transformers.js with ONNX Runtime
BREAKING CHANGE: Complete migration from TensorFlow.js to Transformers.js for embedding generation

This is a major architectural change that replaces TensorFlow.js (USE model) with Transformers.js (all-MiniLM-L6-v2) for significantly improved performance and reduced complexity.

Key Changes:
- Replace TensorFlow.js Universal Sentence Encoder with Transformers.js all-MiniLM-L6-v2
- Reduce model size from 525MB to 87MB (83% reduction)
- Reduce embedding dimensions from 512 to 384 (faster distance calculations)
- Remove TensorFlow.js Float32Array patching (caused ONNX conflicts)
- Implement smart bundled model detection for offline operation
- Add explicit model download script for Docker deployments
- Remove complex environment variables in favor of simple configuration
- Update all distance functions to use optimized pure JavaScript
- Remove TensorFlow-specific utilities and type definitions

Performance Improvements:
- Model loading: 5x faster (87MB vs 525MB)
- Memory usage: 75% reduction (~200-400MB vs ~1.5GB)
- Distance calculations: Faster pure JS vs GPU overhead for small vectors
- Cold start performance: Significantly improved

Files Changed:
- Updated package.json: New dependencies, simplified scripts
- Rewrote src/utils/embedding.ts: Complete Transformers.js implementation
- Updated src/utils/distance.ts: Optimized JavaScript distance functions
- Simplified src/setup.ts: Removed TensorFlow-specific patching
- Simplified src/utils/textEncoding.ts: Only Node.js TextEncoder/Decoder patches
- Deleted src/utils/robustModelLoader.ts: TensorFlow-specific loader
- Deleted src/types/tensorflowTypes.ts: TensorFlow type definitions
- Added scripts/download-models.cjs: Docker-compatible model downloader
- Added comprehensive documentation: README.md, OFFLINE_MODELS.md, analysis docs

Testing:
- All 19 tests passing
- Removed test mocking in favor of real implementation testing
- Updated test environment for Transformers.js compatibility
- Performance tests validate improved efficiency

This migration resolves production issues with Docker egress limitations and provides a more robust, performant foundation for vector operations.
2025-08-05 19:29:59 -07:00

255 lines
No EOL
7.8 KiB
TypeScript

/**
* Tests for offset-based pagination in search results
*/
import { describe, it, expect, beforeEach, afterEach } from 'vitest'
import { BrainyData } from '../src/brainyData.js'
import { cleanupWorkerPools } from '../src/utils/index.js'
describe('Pagination with Offset', () => {
let db: BrainyData
beforeEach(async () => {
// Initialize BrainyData with in-memory storage for testing
db = new BrainyData({
storage: {
forceMemoryStorage: true
},
logging: {
verbose: false
}
})
await db.init()
})
afterEach(async () => {
await db.clear()
await cleanupWorkerPools()
})
describe('Basic Offset Pagination', () => {
it('should return correct results with offset=0', async () => {
// Add test data
const testData = []
for (let i = 0; i < 20; i++) {
const item = {
id: `item-${i}`,
text: `test document ${i}`,
value: i
}
testData.push(item)
await db.add(item)
}
// Search without offset (default offset=0)
const results = await db.search('test document', 5)
expect(results.length).toBe(5)
// Results should be the top 5 most similar
const resultIds = results.map(r => r.metadata.id)
expect(resultIds.length).toBe(5)
})
it('should skip results with offset > 0', async () => {
// Add test data
const testData = []
for (let i = 0; i < 20; i++) {
const item = {
id: `item-${i}`,
text: `test document ${i}`,
value: i
}
testData.push(item)
await db.add(item)
}
// Get first page (no offset)
const firstPage = await db.search('test document', 5)
expect(firstPage.length).toBe(5)
const firstPageIds = firstPage.map(r => r.metadata.id)
// Get second page (offset=5)
const secondPage = await db.search('test document', 5, { offset: 5 })
expect(secondPage.length).toBe(5)
const secondPageIds = secondPage.map(r => r.metadata.id)
// Ensure no overlap between pages
const overlap = firstPageIds.filter(id => secondPageIds.includes(id))
expect(overlap.length).toBe(0)
})
it('should handle offset beyond available results', async () => {
// Add limited test data
for (let i = 0; i < 10; i++) {
await db.add({
id: `item-${i}`,
text: `test document ${i}`
})
}
// Search with offset beyond available results
const results = await db.search('test document', 5, { offset: 15 })
expect(results.length).toBe(0)
})
it('should return partial results when offset + k exceeds total', async () => {
// Add limited test data
for (let i = 0; i < 10; i++) {
await db.add({
id: `item-${i}`,
text: `test document ${i}`
})
}
// Search with offset that allows only partial results
const results = await db.search('test document', 5, { offset: 7 })
expect(results.length).toBe(3) // Only 3 results available after offset 7
})
})
describe('Pagination with Filters', () => {
it.skip('should paginate with noun type filters', async () => {
// TODO: This test requires proper noun type support in the add method
// Currently skipped as noun types are not directly supported in the add method
// Add test data with different noun types
for (let i = 0; i < 15; i++) {
await db.add({
id: `doc-${i}`,
text: `document ${i}`,
type: 'document'
}, undefined, 'document')
}
for (let i = 0; i < 15; i++) {
await db.add({
id: `note-${i}`,
text: `note ${i}`,
type: 'note'
}, undefined, 'note')
}
// Get first page of documents
const firstPage = await db.search('document', 5, {
nounTypes: ['document']
})
expect(firstPage.length).toBe(5)
expect(firstPage.every(r => r.metadata.type === 'document')).toBe(true)
// Get second page of documents
const secondPage = await db.search('document', 5, {
nounTypes: ['document'],
offset: 5
})
expect(secondPage.length).toBe(5)
expect(secondPage.every(r => r.metadata.type === 'document')).toBe(true)
// Ensure pages are different
const firstIds = firstPage.map(r => r.metadata.id)
const secondIds = secondPage.map(r => r.metadata.id)
const overlap = firstIds.filter(id => secondIds.includes(id))
expect(overlap.length).toBe(0)
})
it('should paginate with service filters', async () => {
// Add test data with different services
for (let i = 0; i < 20; i++) {
const metadata = {
id: `item-${i}`,
text: `test item ${i}`,
createdBy: {
augmentation: i < 10 ? 'service-a' : 'service-b'
}
}
await db.add(metadata)
}
// Get paginated results for service-a
const page1 = await db.search('test item', 3, {
service: 'service-a'
})
const page2 = await db.search('test item', 3, {
service: 'service-a',
offset: 3
})
// Check that results are from service-a
expect(page1.every(r => r.metadata.createdBy?.augmentation === 'service-a')).toBe(true)
expect(page2.every(r => r.metadata.createdBy?.augmentation === 'service-a')).toBe(true)
// Check no overlap
const page1Ids = page1.map(r => r.metadata.id)
const page2Ids = page2.map(r => r.metadata.id)
expect(page1Ids.filter(id => page2Ids.includes(id)).length).toBe(0)
})
})
describe('Pagination Consistency', () => {
it('should maintain consistent ordering across pages', async () => {
// Add test data
for (let i = 0; i < 30; i++) {
await db.add({
id: `item-${i.toString().padStart(2, '0')}`,
text: `consistent test ${i}`,
score: Math.random()
})
}
// Get all results in one query
const allResults = await db.search('consistent test', 30)
const allIds = allResults.map(r => r.metadata.id)
// Get results in pages
const page1 = await db.search('consistent test', 10, { offset: 0 })
const page2 = await db.search('consistent test', 10, { offset: 10 })
const page3 = await db.search('consistent test', 10, { offset: 20 })
const pagedIds = [
...page1.map(r => r.metadata.id),
...page2.map(r => r.metadata.id),
...page3.map(r => r.metadata.id)
]
// Check that paginated results match the full query
expect(pagedIds).toEqual(allIds)
})
it('should handle empty results gracefully', async () => {
// Search empty database with offset
const results = await db.search('nonexistent', 10, { offset: 5 })
expect(results).toEqual([])
})
})
describe('Vector Search with Offset', () => {
it('should paginate vector searches', async () => {
// Add test vectors
for (let i = 0; i < 20; i++) {
const vector = new Array(384).fill(0).map(() => Math.random())
await db.add({
id: `vec-${i}`,
vector: vector,
index: i
})
}
// Create a query vector
const queryVector = new Array(384).fill(0).map(() => Math.random())
// Get first page
const page1 = await db.search(queryVector, 5, { forceEmbed: false })
expect(page1.length).toBe(5)
// Get second page
const page2 = await db.search(queryVector, 5, {
forceEmbed: false,
offset: 5
})
expect(page2.length).toBe(5)
// Ensure different results
const page1Ids = page1.map(r => r.metadata.id)
const page2Ids = page2.map(r => r.metadata.id)
expect(page1Ids.filter(id => page2Ids.includes(id)).length).toBe(0)
})
})
})