indexer-chunk-options.test.mjs
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | /** |
| 2 | * Hosted indexer chunk option tests. |
| 3 | * |
| 4 | * DeepInfra BGE embedding models enforce a 512-token context window. These |
| 5 | * tests keep the bridge defaults below that provider limit before any network |
| 6 | * embed request can fail with an over-context error. |
| 7 | */ |
| 8 | import { describe, it } from 'node:test'; |
| 9 | import assert from 'node:assert/strict'; |
| 10 | |
| 11 | import { |
| 12 | DEEPINFRA_SAFE_CHUNK_OVERLAP, |
| 13 | DEEPINFRA_SAFE_CHUNK_SIZE, |
| 14 | DEFAULT_CHUNK_OVERLAP, |
| 15 | DEFAULT_CHUNK_SIZE, |
| 16 | defaultBridgeEmbeddingModelForProvider, |
| 17 | resolveIndexerChunkOptions, |
| 18 | } from '../lib/indexer-chunk-options.mjs'; |
| 19 | import { splitByHeadingOrSize } from '../lib/chunk.mjs'; |
| 20 | |
| 21 | describe('indexer chunk options', () => { |
| 22 | it('unit: defaults DeepInfra bridge embeddings to a real DeepInfra model', () => { |
| 23 | assert.equal(defaultBridgeEmbeddingModelForProvider('deepinfra'), 'BAAI/bge-large-en-v1.5'); |
| 24 | assert.equal(defaultBridgeEmbeddingModelForProvider('openai'), 'text-embedding-3-small'); |
| 25 | assert.equal(defaultBridgeEmbeddingModelForProvider('voyage'), 'voyage-4-lite'); |
| 26 | assert.equal(defaultBridgeEmbeddingModelForProvider('ollama'), 'nomic-embed-text'); |
| 27 | }); |
| 28 | |
| 29 | it('integration: clamps hosted DeepInfra chunks below the 512-token model limit margin', () => { |
| 30 | const opts = resolveIndexerChunkOptions( |
| 31 | { INDEXER_CHUNK_SIZE: '2048', INDEXER_CHUNK_OVERLAP: '256' }, |
| 32 | { provider: 'deepinfra', model: 'BAAI/bge-large-en-v1.5' }, |
| 33 | ); |
| 34 | |
| 35 | assert.equal(opts.chunkSize, DEEPINFRA_SAFE_CHUNK_SIZE); |
| 36 | assert.equal(opts.chunkOverlap, DEEPINFRA_SAFE_CHUNK_OVERLAP); |
| 37 | }); |
| 38 | |
| 39 | it('end-to-end: DeepInfra-sized chunking splits a note that old defaults sent as one 2048-char chunk', () => { |
| 40 | const text = 'a '.repeat(1024).trim(); |
| 41 | const oldChunks = splitByHeadingOrSize(text, { |
| 42 | chunkSize: DEFAULT_CHUNK_SIZE, |
| 43 | chunkOverlap: DEFAULT_CHUNK_OVERLAP, |
| 44 | }); |
| 45 | const safeChunks = splitByHeadingOrSize(text, resolveIndexerChunkOptions({}, { |
| 46 | provider: 'deepinfra', |
| 47 | model: 'BAAI/bge-large-en-v1.5', |
| 48 | })); |
| 49 | |
| 50 | assert.equal(oldChunks.length, 1); |
| 51 | assert.equal(safeChunks.length > 1, true); |
| 52 | assert.equal(safeChunks.every((chunk) => chunk.length <= DEEPINFRA_SAFE_CHUNK_SIZE), true); |
| 53 | }); |
| 54 | |
| 55 | it('stress: invalid env values fall back to bounded provider defaults', () => { |
| 56 | const opts = resolveIndexerChunkOptions( |
| 57 | { INDEXER_CHUNK_SIZE: 'not-a-number', INDEXER_CHUNK_OVERLAP: '-5' }, |
| 58 | { provider: 'deepinfra', model: 'BAAI/bge-large-en-v1.5' }, |
| 59 | ); |
| 60 | |
| 61 | assert.deepEqual(opts, { |
| 62 | chunkSize: DEEPINFRA_SAFE_CHUNK_SIZE, |
| 63 | chunkOverlap: DEEPINFRA_SAFE_CHUNK_OVERLAP, |
| 64 | }); |
| 65 | }); |
| 66 | |
| 67 | it('data-integrity: non-DeepInfra providers keep the existing default chunk contract', () => { |
| 68 | assert.deepEqual(resolveIndexerChunkOptions({}, { provider: 'openai' }), { |
| 69 | chunkSize: DEFAULT_CHUNK_SIZE, |
| 70 | chunkOverlap: DEFAULT_CHUNK_OVERLAP, |
| 71 | }); |
| 72 | }); |
| 73 | |
| 74 | it('performance: explicit smaller DeepInfra chunk sizes are honored', () => { |
| 75 | const opts = resolveIndexerChunkOptions( |
| 76 | { INDEXER_CHUNK_SIZE: '768', INDEXER_CHUNK_OVERLAP: '80' }, |
| 77 | { provider: 'deepinfra', model: 'BAAI/bge-large-en-v1.5' }, |
| 78 | ); |
| 79 | |
| 80 | assert.deepEqual(opts, { chunkSize: 768, chunkOverlap: 80 }); |
| 81 | }); |
| 82 | |
| 83 | it('security: overlap is capped below chunk size so chunking always advances', () => { |
| 84 | const opts = resolveIndexerChunkOptions( |
| 85 | { INDEXER_CHUNK_SIZE: '32', INDEXER_CHUNK_OVERLAP: '9999' }, |
| 86 | { provider: 'openai', model: 'text-embedding-3-small' }, |
| 87 | ); |
| 88 | |
| 89 | assert.deepEqual(opts, { chunkSize: 32, chunkOverlap: 31 }); |
| 90 | }); |
| 91 | }); |