test(e2e): add dataset setup and teardown functions for use in CI (#561)

### Description

Added dataset setup and teardown functionality for E2E tests, with utilities for managing test datasets. This PR:

- Creates a dataset setup script that creates primary and secondary datasets if they don't exist
- Adds a dataset teardown script that cleans up test datasets in CI environments (users may want to have their datasets stick around for debugging purposes. we can enforce this differently if things get messy)
- Implements helper utilities like `sanityIdify` to create valid Sanity dataset names
- Adds a timer utility with visual feedback for long-running operations
- Fixes the CI environment variable type to be a boolean instead of string
- Normalizes some naming of setup / teardown functions

### What to review

- The dataset setup and teardown scripts in `packages/@repo/e2e/src/setup/datasets.setup.ts` and `packages/@repo/e2e/src/teardown/datasets.teardown.ts`
- The helper utilities for sanitizing dataset names and timing operations
- The type change for the CI environment variable from string to boolean
- The dependency changes in package.json files

### Testing
You can test this locally by pulling down the branch and updating the following CI variables (please look in 1pass if there's some you don't have):
```
SDK_E2E_DATASET_0=whatever-name
SDK_E2E_DATASET_1=whatever-name-2

CI=true
SDK_E2E_USER_ID=
SDK_E2E_USER_PASSWORD=
RECAPTCHA_E2E_STAGING_KEY=
```
and then running `pnpm test:e2e`

You should see console logging about creating and destroying datasets. When you go to https://www.sanity.work/organizations/oFvj4MZWQ/project/3j6vt2rg/datasets

You shouldn't see the datasets after the tests have run.

### Fun gif
![](https://media1.giphy.com/media/v1.Y2lkPTc5MGI3NjExZmFueXc3NDh6Y3dyZmwyY3ZtbjUwNDlodnB0NWd6YTNnamo0bXlvNyZlcD12MV9naWZzX3NlYXJjaCZjdD1n/dIfpG1Nst0dejPJgyF/giphy.webp)

Author: cngonzalez

package.json

@@ -25,6 +25,7 @@
     "clean": "run-s clean:build clean:deps",
     "clean:build": "turbo run clean",
     "clean:deps": "rimraf packages/*/node_modules apps/*/node_modules node_modules",
+    "cleanup:e2e": "turbo run cleanup --filter=@repo/e2e",
     "dev": "pnpm dev:kitchensink",
     "dev:kitchensink": "turbo run dev --filter=@sanity/kitchensink-react",
     "dev:e2e": "turbo run dev --filter=@sanity/kitchensink-react -- --mode=e2e",

packages/@repo/e2e/package.json

@@ -28,22 +28,27 @@
   ],
   "scripts": {
     "build": "pkg build --strict --clean --check",
+    "cleanup": "tsx src/scripts/cleanup-datasets.ts",
     "lint": "eslint .",
     "lint:fix": "eslint . --fix",
     "prepare": "pnpm build"
   },
   "dependencies": {
     "@sanity/client": "^6.12.5",
     "@sanity/uuid": "^3.0.0",
-    "dotenv": "^16.5.0"
+    "dotenv": "^16.5.0",
+    "lodash-es": "^4.17.21",
+    "ora": "^7.0.1"
   },
   "devDependencies": {
     "@playwright/test": "^1.52.0",
     "@repo/config-eslint": "workspace:*",
     "@repo/tsconfig": "workspace:*",
     "@sanity/pkg-utils": "^7.2.2",
+    "@types/lodash-es": "^4.17.12",
     "@types/node": "^22.10.5",
     "eslint": "^9.22.0",
+    "tsx": "^4.7.1",
     "typescript": "^5.0.0"
   },
   "peerDependencies": {

packages/@repo/e2e/src/helpers/getE2EEnv.ts

@@ -10,7 +10,7 @@ interface E2EEnv {
   /** The session token for authenticated tests */
   SDK_E2E_SESSION_TOKEN: string
   /** Whether we're running in CI */
-  CI?: string
+  CI?: boolean
   /** E2E test user ID */
   SDK_E2E_USER_ID?: string
   /** E2E test user password */
@@ -48,7 +48,7 @@ function findEnv(name: KnownEnvVar): string | undefined {
  * Get the E2E environment variables
  */
 export function getE2EEnv(): E2EEnv {
-  const CI = findEnv('CI')
+  const CI = findEnv('CI') === 'true'
   const SDK_E2E_PROJECT_ID = readEnv('SDK_E2E_PROJECT_ID')
   const SDK_E2E_DATASET_0 = readEnv('SDK_E2E_DATASET_0')
   const SDK_E2E_DATASET_1 = readEnv('SDK_E2E_DATASET_1')

packages/@repo/e2e/src/helpers/sanitizeDatasetName.ts

@@ -0,0 +1,9 @@
+import {deburr} from 'lodash-es'
+
+// we need to make a valid name for a sanity dataset from inputs
+// (people may have odd characters in branch names, etc)
+export function sanitizeDatasetName(input: string): string {
+  return deburr(input)
+    .replace(/[^a-zA-Z0-9_-]+/g, '_')
+    .replace(/^-/, '')
+}

packages/@repo/e2e/src/helpers/timer.ts

@@ -0,0 +1,13 @@
+import ora from 'ora'
+
+export function startTimer(label: string): {end: () => void} {
+  const spinner = ora(label).start()
+  const start = Date.now()
+  return {
+    end: () => spinner.succeed(`${label} (${formatMs(Date.now() - start)})`),
+  }
+}
+
+function formatMs(ms: number) {
+  return ms < 1000 ? `${ms}ms` : `${(ms / 1000).toFixed(2)}s`
+}

packages/@repo/e2e/src/scripts/cleanup-datasets.ts

@@ -0,0 +1,46 @@
+#!/usr/bin/env node
+
+/* eslint-disable no-console */
+import {getClient} from '../helpers/clients'
+import {getE2EEnv} from '../helpers/getE2EEnv'
+import {sanitizeDatasetName} from '../helpers/sanitizeDatasetName'
+import {startTimer} from '../helpers/timer'
+
+const env = getE2EEnv()
+
+// must be run as a separate script to avoid race conditions with the tests
+async function cleanupDatasets() {
+  const primaryDataset = sanitizeDatasetName(env.SDK_E2E_DATASET_0)
+  const secondaryDataset = sanitizeDatasetName(env.SDK_E2E_DATASET_1)
+
+  const client = getClient()
+  const timer = startTimer('Cleaning up test datasets')
+
+  try {
+    // Delete primary dataset
+    try {
+      await client.datasets.delete(primaryDataset)
+      console.log(`Successfully deleted primary dataset ${primaryDataset}`)
+    } catch (error) {
+      console.error(`Failed to delete primary dataset ${primaryDataset}:`, error)
+    }
+
+    // Delete secondary dataset
+    try {
+      await client.datasets.delete(secondaryDataset)
+      console.log(`Successfully deleted secondary dataset ${secondaryDataset}`)
+    } catch (error) {
+      console.error(`Failed to delete secondary dataset ${secondaryDataset}:`, error)
+    }
+
+    timer.end()
+  } catch (error) {
+    console.error('Failed to cleanup datasets:', error)
+    process.exit(1)
+  }
+}
+
+cleanupDatasets().catch((error) => {
+  console.error('Unhandled error during cleanup:', error)
+  process.exit(1)
+})

packages/@repo/e2e/src/setup/datasets.setup.ts

@@ -0,0 +1,35 @@
+import {test as setup} from '@playwright/test'
+
+import {getClient} from '../helpers/clients'
+import {getE2EEnv} from '../helpers/getE2EEnv'
+import {sanitizeDatasetName} from '../helpers/sanitizeDatasetName'
+import {startTimer} from '../helpers/timer'
+
+const env = getE2EEnv()
+
+setup('setup test datasets', async () => {
+  const primaryDataset = sanitizeDatasetName(env.SDK_E2E_DATASET_0)
+  const secondaryDataset = sanitizeDatasetName(env.SDK_E2E_DATASET_1)
+
+  const client = getClient()
+
+  const datasets = await client.datasets.list()
+
+  // Setup primary dataset
+  if (!datasets.find((ds) => ds.name === primaryDataset)) {
+    const timer = startTimer(`Creating primary dataset ${primaryDataset}`)
+    await client.datasets.create(primaryDataset, {
+      aclMode: 'public',
+    })
+    timer.end()
+  }
+
+  // Setup secondary dataset
+  if (!datasets.find((ds) => ds.name === secondaryDataset)) {
+    const timer = startTimer(`Creating secondary dataset ${secondaryDataset}`)
+    await client.datasets.create(secondaryDataset, {
+      aclMode: 'public',
+    })
+    timer.end()
+  }
+})

packages/@repo/e2e/src/teardown/global.teardown.ts renamed to packages/@repo/e2e/src/teardown/auth.teardown.ts

@@ -62,7 +62,6 @@
     "@sanity/message-protocol": "^0.12.0",
     "@sanity/mutate": "^0.12.4",
     "@sanity/types": "^3.83.0",
-    "@types/lodash-es": "^4.17.12",
     "groq": "3.88.1-typegen-experimental.0",
     "lodash-es": "^4.17.21",
     "reselect": "^5.1.1",
@@ -78,6 +77,7 @@
     "@sanity/browserslist-config": "^1.0.5",
     "@sanity/pkg-utils": "^7.2.2",
     "@sanity/prettier-config": "^1.0.3",
+    "@types/lodash-es": "^4.17.12",
     "@vitest/coverage-v8": "3.1.2",
     "eslint": "^9.22.0",
     "groq-js": "^1.16.1",