|
| 1 | +import { BedrockRuntime } from "@aws-sdk/client-bedrock-runtime"; |
| 2 | +import { S3Vectors } from "@aws-sdk/client-s3vectors"; |
| 3 | +import { beforeEach, describe, expect, test as it } from "vitest"; |
| 4 | + |
| 5 | +describe( |
| 6 | + S3Vectors.name, |
| 7 | + () => { |
| 8 | + const s3v = new S3Vectors({ |
| 9 | + region: "us-west-2", |
| 10 | + }); |
| 11 | + const rock = new BedrockRuntime({ |
| 12 | + region: "us-west-2", |
| 13 | + }); |
| 14 | + |
| 15 | + const VECTOR_BUCKET_NAME = "jsv3-e2e-vector"; |
| 16 | + |
| 17 | + beforeEach(async () => { |
| 18 | + async function checkForBucket() { |
| 19 | + const buckets = await s3v.listVectorBuckets({ |
| 20 | + prefix: VECTOR_BUCKET_NAME, |
| 21 | + }); |
| 22 | + |
| 23 | + return buckets.vectorBuckets?.some((bucket) => bucket.vectorBucketName === VECTOR_BUCKET_NAME); |
| 24 | + } |
| 25 | + |
| 26 | + const testBucket = await checkForBucket(); |
| 27 | + |
| 28 | + if (!testBucket) { |
| 29 | + await s3v.createVectorBucket({ |
| 30 | + vectorBucketName: VECTOR_BUCKET_NAME, |
| 31 | + }); |
| 32 | + await s3v.createIndex({ |
| 33 | + vectorBucketName: VECTOR_BUCKET_NAME, |
| 34 | + indexName: "characters", |
| 35 | + dataType: "float32", |
| 36 | + dimension: 1024, |
| 37 | + distanceMetric: "cosine", |
| 38 | + }); |
| 39 | + } |
| 40 | + |
| 41 | + while (true) { |
| 42 | + const testBucket = await checkForBucket(); |
| 43 | + if (!testBucket) { |
| 44 | + await new Promise((r) => setTimeout(r, 5000)); |
| 45 | + } else { |
| 46 | + break; |
| 47 | + } |
| 48 | + } |
| 49 | + }); |
| 50 | + |
| 51 | + it("can put vectors", async () => { |
| 52 | + const texts = [ |
| 53 | + "Santa Bear is a bear who wears a Santa outfit.", |
| 54 | + "Saucey is a tiger who is rotund.", |
| 55 | + `Tigey is a tiger who likes to say "woof".`, |
| 56 | + ]; |
| 57 | + |
| 58 | + const embeddings: number[][] = []; |
| 59 | + for (const text of texts) { |
| 60 | + const invoke = await rock.invokeModel({ |
| 61 | + modelId: "amazon.titan-embed-text-v2:0", |
| 62 | + body: JSON.stringify({ |
| 63 | + inputText: text, |
| 64 | + }), |
| 65 | + }); |
| 66 | + embeddings.push([...invoke.body].slice(0, 1024)); |
| 67 | + } |
| 68 | + |
| 69 | + await s3v.putVectors({ |
| 70 | + vectorBucketName: VECTOR_BUCKET_NAME, |
| 71 | + indexName: "characters", |
| 72 | + vectors: [ |
| 73 | + { |
| 74 | + key: "Santa Bear", |
| 75 | + data: { float32: embeddings[0] }, |
| 76 | + metadata: { source_text: texts[0], genre: "bear" }, |
| 77 | + }, |
| 78 | + { |
| 79 | + key: "Saucey", |
| 80 | + data: { float32: embeddings[1] }, |
| 81 | + metadata: { source_text: texts[1], genre: "tiger" }, |
| 82 | + }, |
| 83 | + { |
| 84 | + key: "Tigey", |
| 85 | + data: { float32: embeddings[2] }, |
| 86 | + metadata: { source_text: texts[2], genre: "tiger" }, |
| 87 | + }, |
| 88 | + ], |
| 89 | + }); |
| 90 | + }); |
| 91 | + |
| 92 | + it("can query vectors", async () => { |
| 93 | + const invoke = await rock.invokeModel({ |
| 94 | + modelId: "amazon.titan-embed-text-v2:0", |
| 95 | + body: JSON.stringify({ |
| 96 | + inputText: "characters who are tigers or bears", |
| 97 | + }), |
| 98 | + }); |
| 99 | + |
| 100 | + const embedding = invoke.body; |
| 101 | + |
| 102 | + const query = await s3v.queryVectors({ |
| 103 | + vectorBucketName: VECTOR_BUCKET_NAME, |
| 104 | + indexName: "characters", |
| 105 | + queryVector: { |
| 106 | + float32: [...embedding].slice(0, 1024), |
| 107 | + }, |
| 108 | + topK: 3, |
| 109 | + returnDistance: true, |
| 110 | + returnMetadata: true, |
| 111 | + }); |
| 112 | + |
| 113 | + expect(query.vectors?.some((v) => v.key === "Tigey")).toBeDefined(); |
| 114 | + expect(query.vectors?.some((v) => v.key === "Santa Bear")).toBeDefined(); |
| 115 | + expect(query.vectors?.some((v) => v.key === "Saucey")).toBeDefined(); |
| 116 | + }); |
| 117 | + }, |
| 118 | + 120_000 |
| 119 | +); |
0 commit comments