The best way to control Claude Code, Codex and more from any computer or phone
| 1 | import * as GCP from "@alchemy.run/gcp"; |
| 2 | import { expect } from "@alchemy.run/test"; |
| 3 | import * as Test from "@alchemy.run/test/Alchemy"; |
| 4 | import * as datalabeling from "@distilled.cloud/gcp/datalabeling_v1beta1"; |
| 5 | import * as Effect from "effect/Effect"; |
| 6 | import { logLevel, probe, currentProject, runLifecycle, waitUntilGone } from "./common.ts"; |
| 7 | |
| 8 | const { test } = Test.make({ providers: GCP.providers() }); |
| 9 | |
| 10 | test.provider( |
| 11 | "getProjectsEvaluationJobs on a missing job fails with a typed tag", |
| 12 | (stack) => |
| 13 | Effect.gen(function* () { |
| 14 | const project = yield* currentProject; |
| 15 | yield* stack.destroy(); |
| 16 | |
| 17 | const error = yield* Effect.flip( |
| 18 | probe( |
| 19 | datalabeling.getProjectsEvaluationJobs({ |
| 20 | name: `projects/${project}/evaluationJobs/alchemy-missing-job`, |
| 21 | }), |
| 22 | ), |
| 23 | ); |
| 24 | expect(error._tag).toEqual("NotFound"); |
| 25 | |
| 26 | yield* stack.destroy(); |
| 27 | }).pipe(logLevel), |
| 28 | { |
| 29 | tags: ["provider:gcp", "provider:gcp:datalabeling", "live"], |
| 30 | timeout: 90_000, |
| 31 | }, |
| 32 | ); |
| 33 | |
| 34 | test.provider.skipIf(!runLifecycle)( |
| 35 | "create, update, and delete an evaluation job", |
| 36 | (stack) => |
| 37 | Effect.gen(function* () { |
| 38 | const project = yield* currentProject; |
| 39 | yield* stack.destroy(); |
| 40 | |
| 41 | const created = yield* stack.deploy( |
| 42 | Effect.gen(function* () { |
| 43 | const specs = yield* GCP.DataLabeling.AnnotationSpecSet("Classes", { |
| 44 | displayName: "pets", |
| 45 | annotationSpecs: [{ displayName: "dog" }, { displayName: "cat" }], |
| 46 | }); |
| 47 | const job = yield* GCP.DataLabeling.EvaluationJob("DailyEval", { |
| 48 | annotationSpecSet: specs.name, |
| 49 | modelVersion: `projects/${project}/models/classifier/versions/v1`, |
| 50 | schedule: "every 24 hours", |
| 51 | description: "daily sample", |
| 52 | evaluationJobConfig: { |
| 53 | exampleCount: 50, |
| 54 | exampleSamplePercentage: 0.1, |
| 55 | evaluationConfig: {}, |
| 56 | inputConfig: { |
| 57 | dataType: "IMAGE", |
| 58 | annotationType: "IMAGE_CLASSIFICATION_ANNOTATION", |
| 59 | classificationMetadata: { isMultiLabel: false }, |
| 60 | bigquerySource: { |
| 61 | inputUri: `bq://${project}.eval.predictions`, |
| 62 | }, |
| 63 | }, |
| 64 | bigqueryImportKeys: { |
| 65 | data_json_key: "data", |
| 66 | label_json_key: "label", |
| 67 | label_score_json_key: "score", |
| 68 | }, |
| 69 | imageClassificationConfig: { |
| 70 | annotationSpecSet: specs.name, |
| 71 | }, |
| 72 | }, |
| 73 | }); |
| 74 | return { specs, job }; |
| 75 | }), |
| 76 | ); |
| 77 | |
| 78 | expect(created.job.name).toContain("/evaluationJobs/"); |
| 79 | expect(created.job.annotationSpecSet).toEqual(created.specs.name); |
| 80 | expect(created.job.description).toEqual("daily sample"); |
| 81 | expect(created.job.evaluationJobConfig?.exampleCount).toEqual(50); |
| 82 | |
| 83 | const fetched = yield* datalabeling.getProjectsEvaluationJobs({ |
| 84 | name: created.job.name, |
| 85 | }); |
| 86 | expect(fetched.name).toEqual(created.job.name); |
| 87 | expect(fetched.description).toContain("alchemy-id="); |
| 88 | |
| 89 | const updated = yield* stack.deploy( |
| 90 | Effect.gen(function* () { |
| 91 | const specs = yield* GCP.DataLabeling.AnnotationSpecSet("Classes", { |
| 92 | annotationSpecSetId: created.specs.annotationSpecSetId, |
| 93 | displayName: "pets", |
| 94 | annotationSpecs: [{ displayName: "dog" }, { displayName: "cat" }], |
| 95 | }); |
| 96 | const job = yield* GCP.DataLabeling.EvaluationJob("DailyEval", { |
| 97 | evaluationJobId: created.job.evaluationJobId, |
| 98 | annotationSpecSet: specs.name, |
| 99 | modelVersion: created.job.modelVersion ?? "", |
| 100 | schedule: created.job.schedule ?? "every 24 hours", |
| 101 | description: "daily sample", |
| 102 | evaluationJobConfig: { |
| 103 | ...created.job.evaluationJobConfig, |
| 104 | exampleCount: 80, |
| 105 | exampleSamplePercentage: 0.2, |
| 106 | imageClassificationConfig: { |
| 107 | annotationSpecSet: specs.name, |
| 108 | }, |
| 109 | }, |
| 110 | }); |
| 111 | return { specs, job }; |
| 112 | }), |
| 113 | ); |
| 114 | |
| 115 | expect(updated.job.name).toEqual(created.job.name); |
| 116 | expect(updated.job.evaluationJobConfig?.exampleCount).toEqual(80); |
| 117 | |
| 118 | yield* stack.destroy(); |
| 119 | const gone = yield* waitUntilGone( |
| 120 | datalabeling.getProjectsEvaluationJobs({ name: created.job.name }), |
| 121 | ); |
| 122 | expect(gone).toEqual("gone"); |
| 123 | }).pipe(logLevel), |
| 124 | { |
| 125 | tags: ["provider:gcp", "provider:gcp:datalabeling", "live"], |
| 126 | timeout: 90_000, |
| 127 | }, |
| 128 | ); |