Keep/keep-notes/tests/unit/embedding-validation.test.ts

import { test, expect } from '@playwright/test'
import { validateEmbedding, calculateL2Norm, normalizeEmbedding } from '../../lib/utils'

test.describe('Embedding Validation', () => {
  test.describe('validateEmbedding()', () => {
    test('should validate a normal embedding', () => {
      const embedding = [0.1, 0.2, 0.3, 0.4, 0.5]
      const result = validateEmbedding(embedding)

      expect(result.valid).toBe(true)
      expect(result.issues).toHaveLength(0)
    })

    test('should reject empty embedding', () => {
      const result = validateEmbedding([])

      expect(result.valid).toBe(false)
      expect(result.issues).toContain('Embedding is empty or has zero dimensionality')
    })

    test('should reject null embedding', () => {
      const result = validateEmbedding(null as any)

      expect(result.valid).toBe(false)
      expect(result.issues).toContain('Embedding is empty or has zero dimensionality')
    })

    test('should reject embedding with NaN values', () => {
      const embedding = [0.1, NaN, 0.3, 0.4, 0.5]
      const result = validateEmbedding(embedding)

      expect(result.valid).toBe(false)
      expect(result.issues).toContain('Embedding contains NaN values')
    })

    test('should reject embedding with Infinity values', () => {
      const embedding = [0.1, 0.2, Infinity, 0.4, 0.5]
      const result = validateEmbedding(embedding)

      expect(result.valid).toBe(false)
      expect(result.issues).toContain('Embedding contains Infinity values')
    })

    test('should reject zero vector', () => {
      const embedding = [0, 0, 0, 0, 0]
      const result = validateEmbedding(embedding)

      expect(result.valid).toBe(false)
      expect(result.issues).toContain('Embedding is a zero vector (all values are 0)')
    })

    test('should warn about L2 norm outside normal range', () => {
      // Very small norm
      const smallEmbedding = [0.01, 0.01, 0.01]
      const result1 = validateEmbedding(smallEmbedding)

      expect(result1.valid).toBe(false)
      expect(result1.issues.some(issue => issue.includes('L2 norm'))).toBe(true)

      // Very large norm
      const largeEmbedding = [2, 2, 2]
      const result2 = validateEmbedding(largeEmbedding)

      expect(result2.valid).toBe(false)
      expect(result2.issues.some(issue => issue.includes('L2 norm'))).toBe(true)
    })

    test('should detect multiple issues', () => {
      const embedding = [NaN, Infinity, 0]
      const result = validateEmbedding(embedding)

      expect(result.valid).toBe(false)
      expect(result.issues.length).toBeGreaterThan(1)
      expect(result.issues).toContain('Embedding contains NaN values')
      expect(result.issues).toContain('Embedding contains Infinity values')
      // Note: NaN and Infinity are not zero, so it won't detect zero vector
    })
  })

  test.describe('calculateL2Norm()', () => {
    test('should calculate correct L2 norm', () => {
      const vector = [3, 4]
      const norm = calculateL2Norm(vector)

      expect(norm).toBe(5) // sqrt(3^2 + 4^2) = 5
    })

    test('should return 0 for zero vector', () => {
      const vector = [0, 0, 0]
      const norm = calculateL2Norm(vector)

      expect(norm).toBe(0)
    })

    test('should handle negative values', () => {
      const vector = [-3, -4]
      const norm = calculateL2Norm(vector)

      expect(norm).toBe(5) // sqrt((-3)^2 + (-4)^2) = 5
    })
  })

  test.describe('normalizeEmbedding()', () => {
    test('should normalize a vector to unit L2 norm', () => {
      const embedding = [3, 4]
      const normalized = normalizeEmbedding(embedding)
      const norm = calculateL2Norm(normalized)

      expect(norm).toBeCloseTo(1.0, 5)
    })

    test('should preserve direction of vector', () => {
      const embedding = [1, 2, 3]
      const normalized = normalizeEmbedding(embedding)

      // Check that ratios are preserved
      expect(normalized[1] / normalized[0]).toBeCloseTo(embedding[1] / embedding[0], 5)
      expect(normalized[2] / normalized[1]).toBeCloseTo(embedding[2] / embedding[1], 5)
    })

    test('should return zero vector unchanged', () => {
      const embedding = [0, 0, 0]
      const normalized = normalizeEmbedding(embedding)

      expect(normalized).toEqual(embedding)
    })

    test('should handle already normalized vectors', () => {
      const embedding = [0.707, 0.707] // Already approximately unit norm
      const normalized = normalizeEmbedding(embedding)
      const norm = calculateL2Norm(normalized)

      expect(norm).toBeCloseTo(1.0, 5)
    })
  })
})