chore: import upstream snapshot with attribution

apps/sim/tools/apify/get_dataset_items.ts

@@ -0,0 +1,98 @@
+import type { GetDatasetItemsParams, GetDatasetItemsResult } from '@/tools/apify/types'
+import type { ToolConfig } from '@/tools/types'
+
+export const apifyGetDatasetItemsTool: ToolConfig<GetDatasetItemsParams, GetDatasetItemsResult> = {
+  id: 'apify_get_dataset_items',
+  name: 'APIFY Get Dataset Items',
+  description: 'Retrieve items stored in an APIFY dataset',
+  version: '1.0.0',
+
+  params: {
+    apiKey: {
+      type: 'string',
+      required: true,
+      visibility: 'user-only',
+      description: 'APIFY API token from console.apify.com/account#/integrations',
+    },
+    datasetId: {
+      type: 'string',
+      required: true,
+      visibility: 'user-or-llm',
+      description: 'Dataset ID to read items from. Example: "9RnD3Pql2vGZkc5H5"',
+    },
+    itemLimit: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Max items to return (1-250000). Default: all items. Example: 500',
+    },
+    offset: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Number of items to skip at the start. Default: 0',
+    },
+    fields: {
+      type: 'string',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Comma-separated list of fields to include. Example: "title,url,price"',
+    },
+  },
+
+  request: {
+    url: (params) => {
+      const encodedDatasetId = encodeURIComponent(params.datasetId.trim())
+      const baseUrl = `https://api.apify.com/v2/datasets/${encodedDatasetId}/items`
+      const queryParams = new URLSearchParams()
+
+      queryParams.set('clean', 'true')
+      if (params.itemLimit) {
+        const limit = Math.max(1, Math.min(params.itemLimit, 250000))
+        queryParams.set('limit', limit.toString())
+      }
+      if (params.offset != null) {
+        queryParams.set('offset', params.offset.toString())
+      }
+      if (params.fields) {
+        queryParams.set('fields', params.fields)
+      }
+
+      return `${baseUrl}?${queryParams.toString()}`
+    },
+    method: 'GET',
+    headers: (params) => ({
+      Authorization: `Bearer ${params.apiKey}`,
+    }),
+  },
+
+  transformResponse: async (response, params) => {
+    if (!response.ok) {
+      const errorText = await response.text()
+      return {
+        success: false,
+        output: { success: false, datasetId: params?.datasetId ?? '', items: [], count: 0 },
+        error: `APIFY API error: ${errorText}`,
+      }
+    }
+
+    const items = await response.json()
+    const list = Array.isArray(items) ? items : []
+    return {
+      success: true,
+      output: {
+        success: true,
+        datasetId: params?.datasetId ?? '',
+        items: list,
+        count: list.length,
+      },
+    }
+  },
+
+  outputs: {
+    success: { type: 'boolean', description: 'Whether the items were retrieved' },
+    datasetId: { type: 'string', description: 'Dataset ID the items were read from' },
+    items: { type: 'array', description: 'Items stored in the dataset' },
+    count: { type: 'number', description: 'Number of items returned' },
+  },
+}

apps/sim/tools/apify/get_run.ts

@@ -0,0 +1,92 @@
+import type { ApifyRun, GetRunParams, GetRunResult } from '@/tools/apify/types'
+import type { ToolConfig } from '@/tools/types'
+
+export const apifyGetRunTool: ToolConfig<GetRunParams, GetRunResult> = {
+  id: 'apify_get_run',
+  name: 'APIFY Get Run',
+  description: 'Get the status and details of an APIFY actor run',
+  version: '1.0.0',
+
+  params: {
+    apiKey: {
+      type: 'string',
+      required: true,
+      visibility: 'user-only',
+      description: 'APIFY API token from console.apify.com/account#/integrations',
+    },
+    runId: {
+      type: 'string',
+      required: true,
+      visibility: 'user-or-llm',
+      description: 'Actor run ID to fetch. Example: "HG7ML7M8z78YcAPEB"',
+    },
+  },
+
+  request: {
+    url: (params) =>
+      `https://api.apify.com/v2/actor-runs/${encodeURIComponent(params.runId.trim())}`,
+    method: 'GET',
+    headers: (params) => ({
+      Authorization: `Bearer ${params.apiKey}`,
+    }),
+  },
+
+  transformResponse: async (response, params) => {
+    if (!response.ok) {
+      const errorText = await response.text()
+      return {
+        success: false,
+        output: {
+          success: false,
+          runId: params?.runId ?? '',
+          status: 'ERROR',
+          startedAt: null,
+          finishedAt: null,
+          datasetId: null,
+          keyValueStoreId: null,
+          stats: null,
+        },
+        error: `APIFY API error: ${errorText}`,
+      }
+    }
+
+    const data = await response.json()
+    const run = data.data as ApifyRun
+    return {
+      success: true,
+      output: {
+        success: true,
+        runId: run.id,
+        status: run.status,
+        startedAt: run.startedAt ?? null,
+        finishedAt: run.finishedAt ?? null,
+        datasetId: run.defaultDatasetId ?? null,
+        keyValueStoreId: run.defaultKeyValueStoreId ?? null,
+        stats: run.stats ?? null,
+      },
+    }
+  },
+
+  outputs: {
+    success: { type: 'boolean', description: 'Whether the run was found' },
+    runId: { type: 'string', description: 'APIFY run ID' },
+    status: { type: 'string', description: 'Run status (READY, RUNNING, SUCCEEDED, FAILED, etc.)' },
+    startedAt: {
+      type: 'string',
+      description: 'When the run started (ISO timestamp)',
+      optional: true,
+    },
+    finishedAt: {
+      type: 'string',
+      description: 'When the run finished (ISO timestamp)',
+      optional: true,
+    },
+    datasetId: { type: 'string', description: 'Default dataset ID for the run', optional: true },
+    keyValueStoreId: {
+      type: 'string',
+      description: 'Default key-value store ID for the run',
+      optional: true,
+    },
+    stats: { type: 'json', description: 'Run statistics (memory, CPU, duration)', optional: true },
+  },
+}

apps/sim/tools/apify/index.ts

@@ -0,0 +1,6 @@
+export { apifyGetDatasetItemsTool } from './get_dataset_items'
+export { apifyGetRunTool } from './get_run'
+export { apifyRunActorAsyncTool } from './run_actor_async'
+export { apifyRunActorSyncTool } from './run_actor_sync'
+export { apifyRunTaskTool } from './run_task'
+export * from './types'

apps/sim/tools/apify/run_actor_async.ts

@@ -0,0 +1,228 @@
+import { sleep } from '@sim/utils/helpers'
+import { DEFAULT_EXECUTION_TIMEOUT_MS } from '@/lib/core/execution-limits'
+import type { ApifyRun, RunActorParams, RunActorResult } from '@/tools/apify/types'
+import type { ToolConfig } from '@/tools/types'
+
+const POLL_INTERVAL_MS = 5000
+const MAX_POLL_TIME_MS = DEFAULT_EXECUTION_TIMEOUT_MS
+
+export const apifyRunActorAsyncTool: ToolConfig<RunActorParams, RunActorResult> = {
+  id: 'apify_run_actor_async',
+  name: 'APIFY Run Actor (Async)',
+  description: 'Run an APIFY actor asynchronously with polling for long-running tasks',
+  version: '1.0.0',
+
+  params: {
+    apiKey: {
+      type: 'string',
+      required: true,
+      visibility: 'user-only',
+      description: 'APIFY API token from console.apify.com/account#/integrations',
+    },
+    actorId: {
+      type: 'string',
+      required: true,
+      visibility: 'user-or-llm',
+      description:
+        'Actor ID or username/actor-name. Examples: "apify/web-scraper", "janedoe/my-actor", "moJRLRc85AitArpNN"',
+    },
+    input: {
+      type: 'string',
+      required: false,
+      visibility: 'user-or-llm',
+      description:
+        'Actor input as JSON string. Example: {"startUrls": [{"url": "https://example.com"}], "maxPages": 10}',
+    },
+    waitForFinish: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Initial wait time in seconds (0-60) before polling starts. Example: 30',
+    },
+    itemLimit: {
+      type: 'number',
+      required: false,
+      default: 100,
+      visibility: 'user-or-llm',
+      description: 'Max dataset items to fetch (1-250000). Default: 100. Example: 500',
+    },
+    memory: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description:
+        'Memory in megabytes allocated for the actor run (128-32768). Example: 1024 for 1GB, 2048 for 2GB',
+    },
+    timeout: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description:
+        'Timeout in seconds for the actor run. Example: 300 for 5 minutes, 3600 for 1 hour',
+    },
+    build: {
+      type: 'string',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Actor build to run. Examples: "latest", "beta", "1.2.3", "build-tag-name"',
+    },
+  },
+
+  request: {
+    url: (params) => {
+      const encodedActorId = encodeURIComponent(params.actorId.trim())
+      const baseUrl = `https://api.apify.com/v2/acts/${encodedActorId}/runs`
+      const queryParams = new URLSearchParams()
+
+      if (params.waitForFinish !== undefined) {
+        const waitTime = Math.max(0, Math.min(params.waitForFinish, 60))
+        queryParams.set('waitForFinish', waitTime.toString())
+      }
+      if (params.memory) {
+        queryParams.set('memory', params.memory.toString())
+      }
+      if (params.timeout) {
+        queryParams.set('timeout', params.timeout.toString())
+      }
+      if (params.build) {
+        queryParams.set('build', params.build)
+      }
+
+      const query = queryParams.toString()
+      return query ? `${baseUrl}?${query}` : baseUrl
+    },
+    method: 'POST',
+    headers: (params) => ({
+      Authorization: `Bearer ${params.apiKey}`,
+      'Content-Type': 'application/json',
+    }),
+    body: (params) => {
+      let inputData = {}
+      if (params.input) {
+        try {
+          inputData = JSON.parse(params.input)
+        } catch (e) {
+          throw new Error('Invalid JSON in input parameter')
+        }
+      }
+      return inputData
+    },
+  },
+
+  transformResponse: async (response) => {
+    if (!response.ok) {
+      const errorText = await response.text()
+      return {
+        success: false,
+        output: { success: false, runId: '', status: 'ERROR' },
+        error: `APIFY API error: ${errorText}`,
+      }
+    }
+
+    const data = await response.json()
+    const run = data.data as ApifyRun
+    return {
+      success: true,
+      output: {
+        success: true,
+        runId: run.id,
+        status: run.status,
+        datasetId: run.defaultDatasetId,
+      },
+    }
+  },
+
+  postProcess: async (result, params) => {
+    if (!result.success) {
+      return result
+    }
+
+    const runId = result.output.runId
+
+    let elapsedTime = 0
+
+    while (elapsedTime < MAX_POLL_TIME_MS) {
+      await sleep(POLL_INTERVAL_MS)
+      elapsedTime += POLL_INTERVAL_MS
+
+      const statusResponse = await fetch(`https://api.apify.com/v2/actor-runs/${runId}`, {
+        headers: {
+          Authorization: `Bearer ${params.apiKey}`,
+        },
+      })
+
+      if (!statusResponse.ok) {
+        return {
+          success: false,
+          output: { success: false, runId, status: 'UNKNOWN' },
+          error: 'Failed to fetch run status',
+        }
+      }
+
+      const statusData = await statusResponse.json()
+      const run = statusData.data as ApifyRun
+
+      if (
+        run.status === 'SUCCEEDED' ||
+        run.status === 'FAILED' ||
+        run.status === 'ABORTED' ||
+        run.status === 'TIMED-OUT'
+      ) {
+        if (run.status === 'SUCCEEDED' && run.defaultDatasetId) {
+          const limit = Math.max(1, Math.min(params.itemLimit || 100, 250000))
+          const itemsResponse = await fetch(
+            `https://api.apify.com/v2/datasets/${run.defaultDatasetId}/items?limit=${limit}`,
+            {
+              headers: {
+                Authorization: `Bearer ${params.apiKey}`,
+              },
+            }
+          )
+
+          if (itemsResponse.ok) {
+            const items = await itemsResponse.json()
+            return {
+              success: true,
+              output: {
+                success: true,
+                runId,
+                status: run.status,
+                datasetId: run.defaultDatasetId,
+                items,
+              },
+            }
+          }
+        }
+
+        return {
+          success: run.status === 'SUCCEEDED',
+          output: {
+            success: run.status === 'SUCCEEDED',
+            runId,
+            status: run.status,
+            datasetId: run.defaultDatasetId,
+          },
+          error: run.status !== 'SUCCEEDED' ? `Actor run ${run.status}` : undefined,
+        }
+      }
+    }
+
+    return {
+      success: false,
+      output: {
+        success: false,
+        runId,
+        status: 'TIMEOUT',
+      },
+      error: 'Actor run timed out after 5 minutes of polling',
+    }
+  },
+
+  outputs: {
+    success: { type: 'boolean', description: 'Whether the actor run succeeded' },
+    runId: { type: 'string', description: 'APIFY run ID' },
+    status: { type: 'string', description: 'Run status (SUCCEEDED, FAILED, etc.)' },
+    datasetId: { type: 'string', description: 'Dataset ID containing results' },
+    items: { type: 'array', description: 'Dataset items (if completed)' },
+  },
+}

apps/sim/tools/apify/run_actor_sync.ts

@@ -0,0 +1,118 @@
+import type { RunActorParams, RunActorResult } from '@/tools/apify/types'
+import type { ToolConfig } from '@/tools/types'
+
+export const apifyRunActorSyncTool: ToolConfig<RunActorParams, RunActorResult> = {
+  id: 'apify_run_actor_sync',
+  name: 'APIFY Run Actor (Sync)',
+  description: 'Run an APIFY actor synchronously and get results (max 5 minutes)',
+  version: '1.0.0',
+
+  params: {
+    apiKey: {
+      type: 'string',
+      required: true,
+      visibility: 'user-only',
+      description: 'APIFY API token from console.apify.com/account#/integrations',
+    },
+    actorId: {
+      type: 'string',
+      required: true,
+      visibility: 'user-or-llm',
+      description:
+        'Actor ID or username/actor-name. Examples: "apify/web-scraper", "janedoe/my-actor", "moJRLRc85AitArpNN"',
+    },
+    input: {
+      type: 'string',
+      required: false,
+      visibility: 'user-or-llm',
+      description:
+        'Actor input as JSON string. Example: {"startUrls": [{"url": "https://example.com"}], "maxPages": 10}',
+    },
+    memory: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description:
+        'Memory in megabytes allocated for the actor run (128-32768). Example: 1024 for 1GB, 2048 for 2GB',
+    },
+    timeout: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description:
+        'Timeout in seconds for the actor run. Example: 300 for 5 minutes, 3600 for 1 hour',
+    },
+    build: {
+      type: 'string',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Actor build to run. Examples: "latest", "beta", "1.2.3", "build-tag-name"',
+    },
+  },
+
+  request: {
+    url: (params) => {
+      const encodedActorId = encodeURIComponent(params.actorId.trim())
+      const baseUrl = `https://api.apify.com/v2/acts/${encodedActorId}/run-sync-get-dataset-items`
+      const queryParams = new URLSearchParams()
+
+      if (params.memory) {
+        queryParams.set('memory', params.memory.toString())
+      }
+      if (params.timeout) {
+        queryParams.set('timeout', params.timeout.toString())
+      }
+      if (params.build) {
+        queryParams.set('build', params.build)
+      }
+
+      const query = queryParams.toString()
+      return query ? `${baseUrl}?${query}` : baseUrl
+    },
+    method: 'POST',
+    headers: (params) => ({
+      Authorization: `Bearer ${params.apiKey}`,
+      'Content-Type': 'application/json',
+    }),
+    body: (params) => {
+      let inputData = {}
+      if (params.input) {
+        try {
+          inputData = JSON.parse(params.input)
+        } catch (e) {
+          throw new Error('Invalid JSON in input parameter')
+        }
+      }
+      return inputData
+    },
+  },
+
+  transformResponse: async (response) => {
+    if (!response.ok) {
+      const errorText = await response.text()
+      return {
+        success: false,
+        output: { success: false, runId: '', status: 'ERROR', items: [] },
+        error: `APIFY API error: ${errorText}`,
+      }
+    }
+
+    const items = await response.json()
+    return {
+      success: true,
+      output: {
+        success: true,
+        runId: 'sync-execution',
+        status: 'SUCCEEDED',
+        items,
+      },
+    }
+  },
+
+  outputs: {
+    success: { type: 'boolean', description: 'Whether the actor run succeeded' },
+    runId: { type: 'string', description: 'APIFY run ID' },
+    status: { type: 'string', description: 'Run status (SUCCEEDED, FAILED, etc.)' },
+    items: { type: 'array', description: 'Dataset items (if completed)' },
+  },
+}

apps/sim/tools/apify/run_task.ts

@@ -0,0 +1,123 @@
+import type { RunTaskParams, RunTaskResult } from '@/tools/apify/types'
+import type { ToolConfig } from '@/tools/types'
+
+export const apifyRunTaskTool: ToolConfig<RunTaskParams, RunTaskResult> = {
+  id: 'apify_run_task',
+  name: 'APIFY Run Task',
+  description: 'Run a saved APIFY actor task synchronously and get dataset items (max 5 minutes)',
+  version: '1.0.0',
+
+  params: {
+    apiKey: {
+      type: 'string',
+      required: true,
+      visibility: 'user-only',
+      description: 'APIFY API token from console.apify.com/account#/integrations',
+    },
+    taskId: {
+      type: 'string',
+      required: true,
+      visibility: 'user-or-llm',
+      description:
+        'Task ID or username/task-name. Examples: "janedoe/my-task", "moJRLRc85AitArpNN"',
+    },
+    input: {
+      type: 'string',
+      required: false,
+      visibility: 'user-or-llm',
+      description:
+        'JSON string that overrides the task\'s saved input. Example: {"startUrls": [{"url": "https://example.com"}]}',
+    },
+    itemLimit: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Max dataset items to return (1-250000). Example: 500',
+    },
+    memory: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Memory in megabytes allocated for the run (128-32768). Example: 1024 for 1GB',
+    },
+    timeout: {
+      type: 'number',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Timeout in seconds for the run. Example: 300 for 5 minutes',
+    },
+    build: {
+      type: 'string',
+      required: false,
+      visibility: 'user-or-llm',
+      description: 'Actor build to run. Examples: "latest", "beta", "1.2.3"',
+    },
+  },
+
+  request: {
+    url: (params) => {
+      const encodedTaskId = encodeURIComponent(params.taskId.trim())
+      const baseUrl = `https://api.apify.com/v2/actor-tasks/${encodedTaskId}/run-sync-get-dataset-items`
+      const queryParams = new URLSearchParams()
+
+      if (params.itemLimit) {
+        const limit = Math.max(1, Math.min(params.itemLimit, 250000))
+        queryParams.set('limit', limit.toString())
+      }
+      if (params.memory) {
+        queryParams.set('memory', params.memory.toString())
+      }
+      if (params.timeout) {
+        queryParams.set('timeout', params.timeout.toString())
+      }
+      if (params.build) {
+        queryParams.set('build', params.build)
+      }
+
+      const query = queryParams.toString()
+      return query ? `${baseUrl}?${query}` : baseUrl
+    },
+    method: 'POST',
+    headers: (params) => ({
+      Authorization: `Bearer ${params.apiKey}`,
+      'Content-Type': 'application/json',
+    }),
+    body: (params) => {
+      if (params.input) {
+        try {
+          return JSON.parse(params.input)
+        } catch {
+          throw new Error('Invalid JSON in input parameter')
+        }
+      }
+      return {}
+    },
+  },
+
+  transformResponse: async (response) => {
+    if (!response.ok) {
+      const errorText = await response.text()
+      return {
+        success: false,
+        output: { success: false, status: 'ERROR', items: [] },
+        error: `APIFY API error: ${errorText}`,
+      }
+    }
+
+    const items = await response.json()
+    return {
+      success: true,
+      output: {
+        success: true,
+        status: 'SUCCEEDED',
+        items: Array.isArray(items) ? items : [],
+      },
+    }
+  },
+
+  outputs: {
+    success: { type: 'boolean', description: 'Whether the task run succeeded' },
+    status: { type: 'string', description: 'Run status (SUCCEEDED, FAILED, etc.)' },
+    items: { type: 'array', description: 'Dataset items produced by the run' },
+  },
+}

apps/sim/tools/apify/types.ts

@@ -0,0 +1,95 @@
+import type { ToolResponse } from '@/tools/types'
+
+/** Apify actor run object returned by the run/status endpoints. */
+export interface ApifyRun {
+  id: string
+  actId: string
+  status:
+    | 'READY'
+    | 'RUNNING'
+    | 'SUCCEEDED'
+    | 'FAILED'
+    | 'ABORTED'
+    | 'TIMED-OUT'
+    | 'ABORTING'
+    | 'TIMING-OUT'
+  startedAt?: string
+  finishedAt?: string
+  defaultDatasetId?: string
+  defaultKeyValueStoreId?: string
+  stats?: Record<string, unknown>
+}
+
+export interface RunActorParams {
+  apiKey: string
+  actorId: string
+  input?: string
+  waitForFinish?: number // For async tool: 0-60 seconds initial wait
+  itemLimit?: number // For async tool: 1-250000 items, default 100
+  memory?: number // Memory in MB (128-32768)
+  timeout?: number // Timeout in seconds
+  build?: string // Actor build to run (e.g., "latest", "beta", build tag/number)
+}
+
+export interface RunActorResult extends ToolResponse {
+  output: {
+    success: boolean
+    runId: string
+    status: string
+    datasetId?: string
+    items?: unknown[]
+  }
+}
+
+export interface RunTaskParams {
+  apiKey: string
+  taskId: string
+  input?: string
+  memory?: number
+  timeout?: number
+  build?: string
+  itemLimit?: number
+}
+
+export interface RunTaskResult extends ToolResponse {
+  output: {
+    success: boolean
+    status: string
+    items: unknown[]
+  }
+}
+
+export interface GetDatasetItemsParams {
+  apiKey: string
+  datasetId: string
+  itemLimit?: number
+  offset?: number
+  fields?: string
+}
+
+export interface GetDatasetItemsResult extends ToolResponse {
+  output: {
+    success: boolean
+    datasetId: string
+    items: unknown[]
+    count: number
+  }
+}
+
+export interface GetRunParams {
+  apiKey: string
+  runId: string
+}
+
+export interface GetRunResult extends ToolResponse {
+  output: {
+    success: boolean
+    runId: string
+    status: string
+    startedAt: string | null
+    finishedAt: string | null
+    datasetId: string | null
+    keyValueStoreId: string | null
+    stats: Record<string, unknown> | null
+  }
+}