apify 3.1.2-beta.44 → 3.1.2-beta.69

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/src/actor.ts DELETED
@@ -1,1614 +0,0 @@
1
- import ow from 'ow';
2
- import { createPrivateKey } from 'node:crypto';
3
- import { decryptInputSecrets } from '@apify/input_secrets';
4
- import { ENV_VARS, INTEGER_ENV_VARS } from '@apify/consts';
5
- import { addTimeoutToPromise } from '@apify/timeout';
6
- import log from '@apify/log';
7
- import type {
8
- ActorStartOptions,
9
- ApifyClientOptions,
10
- RunAbortOptions,
11
- TaskStartOptions,
12
- Webhook,
13
- WebhookEventType,
14
- } from 'apify-client';
15
- import {
16
- ActorRun as ClientActorRun,
17
- ApifyClient,
18
- } from 'apify-client';
19
- import type {
20
- ConfigurationOptions,
21
- EventManager,
22
- EventTypeName,
23
- IStorage,
24
- RecordOptions,
25
- UseStateOptions,
26
- } from '@crawlee/core';
27
- import {
28
- Configuration as CoreConfiguration,
29
- Dataset,
30
- EventType,
31
- RequestQueue,
32
- StorageManager,
33
- purgeDefaultStorages,
34
- } from '@crawlee/core';
35
- import type { Awaitable, Constructor, Dictionary, StorageClient } from '@crawlee/types';
36
- import { sleep, snakeCaseToCamelCase } from '@crawlee/utils';
37
- import { logSystemInfo, printOutdatedSdkWarning } from './utils';
38
- import { PlatformEventManager } from './platform_event_manager';
39
- import type { ProxyConfigurationOptions } from './proxy_configuration';
40
- import { ProxyConfiguration } from './proxy_configuration';
41
- import { KeyValueStore } from './key_value_store';
42
- import { Configuration } from './configuration';
43
-
44
- /**
45
- * `Actor` class serves as an alternative approach to the static helpers exported from the package. It allows to pass configuration
46
- * that will be used on the instance methods. Environment variables will have precedence over this configuration.
47
- * See {@apilink Configuration} for details about what can be configured and what are the default values.
48
- */
49
- export class Actor<Data extends Dictionary = Dictionary> {
50
- /** @internal */
51
- static _instance: Actor;
52
-
53
- /**
54
- * Configuration of this SDK instance (provided to its constructor). See {@apilink Configuration} for details.
55
- * @internal
56
- */
57
- readonly config: Configuration;
58
-
59
- /**
60
- * Default {@apilink ApifyClient} instance.
61
- * @internal
62
- */
63
- readonly apifyClient: ApifyClient;
64
-
65
- /**
66
- * Default {@apilink EventManager} instance.
67
- * @internal
68
- */
69
- readonly eventManager: EventManager;
70
-
71
- /**
72
- * Whether the actor instance was initialized. This is set by calling {@apilink Actor.init}.
73
- */
74
- initialized = false;
75
-
76
- /**
77
- * Set if the actor called a method that requires the instance to be initialized, but did not do so.
78
- * A call to `init` after this warning is emitted is considered an invalid state and will throw an error.
79
- */
80
- private warnedAboutMissingInitCall = false;
81
-
82
- constructor(options: ConfigurationOptions = {}) {
83
- // use default configuration object if nothing overridden (it fallbacks to env vars)
84
- this.config = Object.keys(options).length === 0 ? Configuration.getGlobalConfig() : new Configuration(options);
85
- this.apifyClient = this.newClient();
86
- this.eventManager = new PlatformEventManager(this.config);
87
- }
88
-
89
- /**
90
- * Runs the main user function that performs the job of the actor
91
- * and terminates the process when the user function finishes.
92
- *
93
- * **The `Actor.main()` function is optional** and is provided merely for your convenience.
94
- * It is mainly useful when you're running your code as an actor on the [Apify platform](https://apify.com/actors).
95
- * However, if you want to use Apify SDK tools directly inside your existing projects, e.g.
96
- * running in an [Express](https://expressjs.com/) server, on
97
- * [Google Cloud functions](https://cloud.google.com/functions)
98
- * or [AWS Lambda](https://aws.amazon.com/lambda/), it's better to avoid
99
- * it since the function terminates the main process when it finishes!
100
- *
101
- * The `Actor.main()` function performs the following actions:
102
- *
103
- * - When running on the Apify platform (i.e. `APIFY_IS_AT_HOME` environment variable is set),
104
- * it sets up a connection to listen for platform events.
105
- * For example, to get a notification about an imminent migration to another server.
106
- * See {@apilink Actor.events} for details.
107
- * - It checks that either `APIFY_TOKEN` or `APIFY_LOCAL_STORAGE_DIR` environment variable
108
- * is defined. If not, the functions sets `APIFY_LOCAL_STORAGE_DIR` to `./apify_storage`
109
- * inside the current working directory. This is to simplify running code examples.
110
- * - It invokes the user function passed as the `userFunc` parameter.
111
- * - If the user function returned a promise, waits for it to resolve.
112
- * - If the user function throws an exception or some other error is encountered,
113
- * prints error details to console so that they are stored to the log.
114
- * - Exits the Node.js process, with zero exit code on success and non-zero on errors.
115
- *
116
- * The user function can be synchronous:
117
- *
118
- * ```javascript
119
- * await Actor.main(() => {
120
- * // My synchronous function that returns immediately
121
- * console.log('Hello world from actor!');
122
- * });
123
- * ```
124
- *
125
- * If the user function returns a promise, it is considered asynchronous:
126
- * ```javascript
127
- * import { gotScraping } from 'got-scraping';
128
- *
129
- * await Actor.main(() => {
130
- * // My asynchronous function that returns a promise
131
- * return gotScraping('http://www.example.com').then((html) => {
132
- * console.log(html);
133
- * });
134
- * });
135
- * ```
136
- *
137
- * To simplify your code, you can take advantage of the `async`/`await` keywords:
138
- *
139
- * ```javascript
140
- * import { gotScraping } from 'got-scraping';
141
- *
142
- * await Actor.main(async () => {
143
- * // My asynchronous function
144
- * const html = await request('http://www.example.com');
145
- * console.log(html);
146
- * });
147
- * ```
148
- *
149
- * @param userFunc User function to be executed. If it returns a promise,
150
- * the promise will be awaited. The user function is called with no arguments.
151
- * @param options
152
- * @ignore
153
- */
154
- main<T>(userFunc: UserFunc, options?: MainOptions): Promise<T> {
155
- if (!userFunc || typeof userFunc !== 'function') {
156
- throw new Error(`First parameter for Actor.main() must be a function (was '${userFunc === null ? 'null' : typeof userFunc}').`);
157
- }
158
-
159
- return (async () => {
160
- await this.init(options);
161
- let ret: T;
162
-
163
- try {
164
- ret = await userFunc() as T;
165
- await this.exit(options);
166
- } catch (err: any) {
167
- log.exception(err, err.message);
168
- await this.exit({ exitCode: EXIT_CODES.ERROR_USER_FUNCTION_THREW });
169
- }
170
-
171
- return ret!;
172
- })();
173
- }
174
-
175
- /**
176
- * @ignore
177
- */
178
- async init(options: InitOptions = {}): Promise<void> {
179
- if (this.initialized) {
180
- return;
181
- }
182
-
183
- // If the warning about forgotten init call was emitted, we will not continue the init procedure.
184
- if (this.warnedAboutMissingInitCall) {
185
- throw new Error([
186
- 'Actor.init() was called after a method that would access a storage client was used.',
187
- 'This in an invalid state. Please make sure to call Actor.init() before such methods are called.',
188
- ].join('\n'));
189
- }
190
-
191
- this.initialized = true;
192
-
193
- logSystemInfo();
194
- printOutdatedSdkWarning();
195
-
196
- // reset global config instance to respect APIFY_ prefixed env vars
197
- CoreConfiguration.globalConfig = Configuration.getGlobalConfig();
198
-
199
- if (this.isAtHome()) {
200
- this.config.set('availableMemoryRatio', 1);
201
- this.config.set('disableBrowserSandbox', true); // for browser launcher, adds `--no-sandbox` to args
202
- this.config.useStorageClient(this.apifyClient);
203
- this.config.useEventManager(this.eventManager);
204
- } else if (options.storage) {
205
- this.config.useStorageClient(options.storage);
206
- }
207
-
208
- // Init the event manager the config uses
209
- await this.config.getEventManager().init();
210
-
211
- await purgeDefaultStorages(this.config);
212
- Configuration.storage.enterWith(this.config);
213
- }
214
-
215
- /**
216
- * @ignore
217
- */
218
- async exit(messageOrOptions?: string | ExitOptions, options: ExitOptions = {}): Promise<void> {
219
- options = typeof messageOrOptions === 'string' ? { ...options, statusMessage: messageOrOptions } : { ...messageOrOptions, ...options };
220
- options.exit ??= true;
221
- options.exitCode ??= EXIT_CODES.SUCCESS;
222
- options.timeoutSecs ??= 30;
223
-
224
- // Close the event manager and emit the final PERSIST_STATE event
225
- await this.config.getEventManager().close();
226
-
227
- // Emit the exit event
228
- this.config.getEventManager().emit(EventType.EXIT, options);
229
-
230
- // Wait for all event listeners to be processed
231
- log.debug(`Waiting for all event listeners to complete their execution (with ${options.timeoutSecs} seconds timeout)`);
232
- await addTimeoutToPromise(
233
- () => this.config.getEventManager().waitForAllListenersToComplete(),
234
- options.timeoutSecs * 1000,
235
- `Waiting for all event listeners to complete their execution timed out after ${options.timeoutSecs} seconds`,
236
- );
237
-
238
- const client = this.config.getStorageClient();
239
-
240
- if (client.teardown) {
241
- let finished = false;
242
- setTimeout(() => {
243
- if (!finished) {
244
- log.info('Waiting for the storage to write its state to file system.');
245
- }
246
- }, 1000);
247
- await client.teardown();
248
- finished = true;
249
- }
250
-
251
- if (options.exitCode > 0) {
252
- options.statusMessage ??= `Actor finished with an error (exit code ${options.exitCode})`;
253
- log.error(options.statusMessage);
254
- } else {
255
- options.statusMessage ??= `Actor finished successfully (exit code ${options.exitCode})`;
256
- log.info(options.statusMessage);
257
- }
258
-
259
- if (!options.exit) {
260
- return;
261
- }
262
-
263
- process.exit(options.exitCode);
264
- }
265
-
266
- /**
267
- * @ignore
268
- */
269
- async fail(messageOrOptions?: string | ExitOptions, options: ExitOptions = {}): Promise<void> {
270
- return this.exit(messageOrOptions, { exitCode: 1, ...options });
271
- }
272
-
273
- /**
274
- * @ignore
275
- */
276
- on(event: EventTypeName, listener: (...args: any[]) => any): void {
277
- this.config.getEventManager().on(event, listener);
278
- }
279
-
280
- /**
281
- * @ignore
282
- */
283
- off(event: EventTypeName, listener?: (...args: any[]) => any): void {
284
- this.config.getEventManager().off(event, listener);
285
- }
286
-
287
- /**
288
- * Runs an actor on the Apify platform using the current user account (determined by the `APIFY_TOKEN` environment variable).
289
- *
290
- * The result of the function is an {@apilink ActorRun} object that contains details about the actor run.
291
- *
292
- * If you want to run an actor task rather than an actor, please use the {@apilink Actor.callTask} function instead.
293
- *
294
- * For more information about actors, read the [documentation](https://docs.apify.com/actor).
295
- *
296
- * **Example usage:**
297
- *
298
- * ```javascript
299
- * const run = await Actor.call('apify/hello-world', { myInput: 123 });
300
- * ```
301
- *
302
- * @param actorId
303
- * Allowed formats are `username/actor-name`, `userId/actor-name` or actor ID.
304
- * @param [input]
305
- * Input for the actor. If it is an object, it will be stringified to
306
- * JSON and its content type set to `application/json; charset=utf-8`.
307
- * Otherwise the `options.contentType` parameter must be provided.
308
- * @param [options]
309
- * @ignore
310
- */
311
- async call(actorId: string, input?: unknown, options: CallOptions = {}): Promise<ClientActorRun> {
312
- const { token, ...rest } = options;
313
- const client = token ? this.newClient({ token }) : this.apifyClient;
314
-
315
- return client.actor(actorId).call(input, rest);
316
- }
317
-
318
- /**
319
- * Runs an actor on the Apify platform using the current user account (determined by the `APIFY_TOKEN` environment variable),
320
- * unlike `Actor.call`, this method just starts the run without waiting for finish.
321
- *
322
- * The result of the function is an {@apilink ActorRun} object that contains details about the actor run.
323
- *
324
- * For more information about actors, read the
325
- * [documentation](https://docs.apify.com/actor).
326
- *
327
- * **Example usage:**
328
- *
329
- * ```javascript
330
- * const run = await Actor.start('apify/hello-world', { myInput: 123 });
331
- * ```
332
- *
333
- * @param actorId
334
- * Allowed formats are `username/actor-name`, `userId/actor-name` or actor ID.
335
- * @param [input]
336
- * Input for the actor. If it is an object, it will be stringified to
337
- * JSON and its content type set to `application/json; charset=utf-8`.
338
- * Otherwise the `options.contentType` parameter must be provided.
339
- * @param [options]
340
- * @ignore
341
- */
342
- async start(actorId: string, input?: unknown, options: CallOptions = {}): Promise<ClientActorRun> {
343
- const { token, ...rest } = options;
344
- const client = token ? this.newClient({ token }) : this.apifyClient;
345
-
346
- return client.actor(actorId).start(input, rest);
347
- }
348
-
349
- /**
350
- * Aborts given actor run on the Apify platform using the current user account (determined by the `APIFY_TOKEN` environment variable).
351
- *
352
- * The result of the function is an {@apilink ActorRun} object that contains details about the actor run.
353
- *
354
- * For more information about actors, read the
355
- * [documentation](https://docs.apify.com/actor).
356
- *
357
- * **Example usage:**
358
- *
359
- * ```javascript
360
- * const run = await Actor.abort(runId);
361
- * ```
362
- * @ignore
363
- */
364
- async abort(runId: string, options: AbortOptions = {}): Promise<ClientActorRun> {
365
- const { token, statusMessage, ...rest } = options;
366
- const client = token ? this.newClient({ token }) : this.apifyClient;
367
-
368
- if (statusMessage) {
369
- await this.setStatusMessage(statusMessage);
370
- }
371
-
372
- return client.run(runId).abort(rest);
373
- }
374
-
375
- /**
376
- * Runs an actor task on the Apify platform using the current user account (determined by the `APIFY_TOKEN` environment variable).
377
- *
378
- * The result of the function is an {@apilink ActorRun} object that contains details about the actor run.
379
- *
380
- * Note that an actor task is a saved input configuration and options for an actor.
381
- * If you want to run an actor directly rather than an actor task, please use the
382
- * {@apilink Actor.call} function instead.
383
- *
384
- * For more information about actor tasks, read the [documentation](https://docs.apify.com/tasks).
385
- *
386
- * **Example usage:**
387
- *
388
- * ```javascript
389
- * const run = await Actor.callTask('bob/some-task');
390
- * ```
391
- *
392
- * @param taskId
393
- * Allowed formats are `username/task-name`, `userId/task-name` or task ID.
394
- * @param [input]
395
- * Input overrides for the actor task. If it is an object, it will be stringified to
396
- * JSON and its content type set to `application/json; charset=utf-8`.
397
- * Provided input will be merged with actor task input.
398
- * @param [options]
399
- * @ignore
400
- */
401
- async callTask(taskId: string, input?: Dictionary, options: CallTaskOptions = {}): Promise<ClientActorRun> {
402
- const { token, ...rest } = options;
403
- const client = token ? this.newClient({ token }) : this.apifyClient;
404
-
405
- return client.task(taskId).call(input, rest);
406
- }
407
-
408
- /**
409
- * Transforms this actor run to an actor run of a given actor. The system stops the current container and starts
410
- * the new container instead. All the default storages are preserved and the new input is stored under the `INPUT-METAMORPH-1` key
411
- * in the same default key-value store.
412
- *
413
- * @param targetActorId
414
- * Either `username/actor-name` or actor ID of an actor to which we want to metamorph.
415
- * @param [input]
416
- * Input for the actor. If it is an object, it will be stringified to
417
- * JSON and its content type set to `application/json; charset=utf-8`.
418
- * Otherwise, the `options.contentType` parameter must be provided.
419
- * @param [options]
420
- * @ignore
421
- */
422
- async metamorph(targetActorId: string, input?: unknown, options: MetamorphOptions = {}): Promise<void> {
423
- if (!this.isAtHome()) {
424
- log.warning('Actor.metamorph() is only supported when running on the Apify platform.');
425
- return;
426
- }
427
-
428
- const {
429
- customAfterSleepMillis = this.config.get('metamorphAfterSleepMillis'),
430
- ...metamorphOpts
431
- } = options;
432
- const runId = this.config.get('actorRunId')!;
433
- await this.apifyClient.run(runId).metamorph(targetActorId, input, metamorphOpts);
434
-
435
- // Wait some time for container to be stopped.
436
- await sleep(customAfterSleepMillis);
437
- }
438
-
439
- /**
440
- * Internally reboots this actor. The system stops the current container and starts
441
- * a new container with the same run ID.
442
- *
443
- * @ignore
444
- */
445
- async reboot(): Promise<void> {
446
- if (!this.isAtHome()) {
447
- log.warning('Actor.reboot() is only supported when running on the Apify platform.');
448
- return;
449
- }
450
-
451
- // Waiting for all the listeners to finish, as `.metamorph()` kills the container.
452
- await Promise.all([
453
- // `persistState` for individual RequestLists, RequestQueue... instances to be persisted
454
- ...this.config.getEventManager().listeners(EventType.PERSIST_STATE).map((x) => x()),
455
- // `migrating` to pause Apify crawlers
456
- ...this.config.getEventManager().listeners(EventType.MIGRATING).map((x) => x()),
457
- ]);
458
-
459
- const actorId = this.config.get('actorId')!;
460
- await this.metamorph(actorId);
461
- }
462
-
463
- /**
464
- * Creates an ad-hoc webhook for the current actor run, which lets you receive a notification when the actor run finished or failed.
465
- * For more information about Apify actor webhooks, please see the [documentation](https://docs.apify.com/webhooks).
466
- *
467
- * Note that webhooks are only supported for actors running on the Apify platform.
468
- * In local environment, the function will print a warning and have no effect.
469
- *
470
- * @param options
471
- * @returns The return value is the Webhook object.
472
- * For more information, see the [Get webhook](https://apify.com/docs/api/v2#/reference/webhooks/webhook-object/get-webhook) API endpoint.
473
- * @ignore
474
- */
475
- async addWebhook(options: WebhookOptions): Promise<Webhook | undefined> {
476
- ow(options, ow.object.exactShape({
477
- eventTypes: ow.array.ofType<WebhookEventType>(ow.string),
478
- requestUrl: ow.string,
479
- payloadTemplate: ow.optional.string,
480
- idempotencyKey: ow.optional.string,
481
- }));
482
-
483
- const { eventTypes, requestUrl, payloadTemplate, idempotencyKey } = options;
484
-
485
- if (!this.isAtHome()) {
486
- log.warning('Actor.addWebhook() is only supported when running on the Apify platform. The webhook will not be invoked.');
487
- return undefined;
488
- }
489
-
490
- const runId = this.config.get('actorRunId')!;
491
- if (!runId) {
492
- throw new Error(`Environment variable ${ENV_VARS.ACTOR_RUN_ID} is not set!`);
493
- }
494
-
495
- return this.apifyClient.webhooks().create({
496
- isAdHoc: true,
497
- eventTypes,
498
- condition: {
499
- actorRunId: runId,
500
- },
501
- requestUrl,
502
- payloadTemplate,
503
- idempotencyKey,
504
- });
505
- }
506
-
507
- /**
508
- * Sets the status message for the current actor run.
509
- *
510
- * @param options
511
- * @returns The return value is the Run object.
512
- * For more information, see the [Actor Runs](https://docs.apify.com/api/v2#/reference/actor-runs/) API endpoints.
513
- * @ignore
514
- */
515
- async setStatusMessage(statusMessage: string): Promise<ClientActorRun> {
516
- ow(statusMessage, ow.string);
517
-
518
- const runId = this.config.get('actorRunId')!;
519
- if (!runId) {
520
- throw new Error(`Environment variable ${ENV_VARS.ACTOR_RUN_ID} is not set!`);
521
- }
522
-
523
- return this.apifyClient.run(runId).update({ statusMessage });
524
- }
525
-
526
- /**
527
- * Stores an object or an array of objects to the default {@apilink Dataset} of the current actor run.
528
- *
529
- * This is just a convenient shortcut for {@apilink Dataset.pushData}.
530
- * For example, calling the following code:
531
- * ```javascript
532
- * await Actor.pushData({ myValue: 123 });
533
- * ```
534
- *
535
- * is equivalent to:
536
- * ```javascript
537
- * const dataset = await Actor.openDataset();
538
- * await dataset.pushData({ myValue: 123 });
539
- * ```
540
- *
541
- * For more information, see {@apilink Actor.openDataset} and {@apilink Dataset.pushData}
542
- *
543
- * **IMPORTANT**: Make sure to use the `await` keyword when calling `pushData()`,
544
- * otherwise the actor process might finish before the data are stored!
545
- *
546
- * @param item Object or array of objects containing data to be stored in the default dataset.
547
- * The objects must be serializable to JSON and the JSON representation of each object must be smaller than 9MB.
548
- * @ignore
549
- */
550
- async pushData(item: Data | Data[]): Promise<void> {
551
- this._ensureActorInit('pushData');
552
-
553
- const dataset = await this.openDataset();
554
- return dataset.pushData(item);
555
- }
556
-
557
- /**
558
- * Opens a dataset and returns a promise resolving to an instance of the {@apilink Dataset} class.
559
- *
560
- * Datasets are used to store structured data where each object stored has the same attributes,
561
- * such as online store products or real estate offers.
562
- * The actual data is stored either on the local filesystem or in the cloud.
563
- *
564
- * For more details and code examples, see the {@apilink Dataset} class.
565
- *
566
- * @param [datasetIdOrName]
567
- * ID or name of the dataset to be opened. If `null` or `undefined`,
568
- * the function returns the default dataset associated with the actor run.
569
- * @param [options]
570
- * @ignore
571
- */
572
- async openDataset(
573
- datasetIdOrName?: string | null,
574
- options: OpenStorageOptions = {},
575
- ): Promise<Dataset<Data>> {
576
- ow(datasetIdOrName, ow.optional.string);
577
- ow(options, ow.object.exactShape({
578
- forceCloud: ow.optional.boolean,
579
- }));
580
-
581
- this._ensureActorInit('openDataset');
582
-
583
- return this._openStorage<Dataset<Data>>(Dataset, datasetIdOrName, options);
584
- }
585
-
586
- /**
587
- * Gets a value from the default {@apilink KeyValueStore} associated with the current actor run.
588
- *
589
- * This is just a convenient shortcut for {@apilink KeyValueStore.getValue}.
590
- * For example, calling the following code:
591
- * ```javascript
592
- * const value = await Actor.getValue('my-key');
593
- * ```
594
- *
595
- * is equivalent to:
596
- * ```javascript
597
- * const store = await Actor.openKeyValueStore();
598
- * const value = await store.getValue('my-key');
599
- * ```
600
- *
601
- * To store the value to the default key-value store, you can use the {@apilink Actor.setValue} function.
602
- *
603
- * For more information, see {@apilink Actor.openKeyValueStore}
604
- * and {@apilink KeyValueStore.getValue}.
605
- *
606
- * @param key Unique record key.
607
- * @returns
608
- * Returns a promise that resolves to an object, string
609
- * or [`Buffer`](https://nodejs.org/api/buffer.html), depending
610
- * on the MIME content type of the record, or `null`
611
- * if the record is missing.
612
- * @ignore
613
- */
614
- async getValue<T = unknown>(key: string): Promise<T | null> {
615
- this._ensureActorInit('getValue');
616
-
617
- const store = await this.openKeyValueStore();
618
- return store.getValue<T>(key);
619
- }
620
-
621
- /**
622
- * Stores or deletes a value in the default {@apilink KeyValueStore} associated with the current actor run.
623
- *
624
- * This is just a convenient shortcut for {@apilink KeyValueStore.setValue}.
625
- * For example, calling the following code:
626
- * ```javascript
627
- * await Actor.setValue('OUTPUT', { foo: "bar" });
628
- * ```
629
- *
630
- * is equivalent to:
631
- * ```javascript
632
- * const store = await Actor.openKeyValueStore();
633
- * await store.setValue('OUTPUT', { foo: "bar" });
634
- * ```
635
- *
636
- * To get a value from the default key-value store, you can use the {@apilink Actor.getValue} function.
637
- *
638
- * For more information, see {@apilink Actor.openKeyValueStore}
639
- * and {@apilink KeyValueStore.getValue}.
640
- *
641
- * @param key
642
- * Unique record key.
643
- * @param value
644
- * Record data, which can be one of the following values:
645
- * - If `null`, the record in the key-value store is deleted.
646
- * - If no `options.contentType` is specified, `value` can be any JavaScript object, and it will be stringified to JSON.
647
- * - If `options.contentType` is set, `value` is taken as is, and it must be a `String` or [`Buffer`](https://nodejs.org/api/buffer.html).
648
- * For any other value an error will be thrown.
649
- * @param [options]
650
- * @ignore
651
- */
652
- async setValue<T>(key: string, value: T | null, options: RecordOptions = {}): Promise<void> {
653
- this._ensureActorInit('setValue');
654
-
655
- const store = await this.openKeyValueStore();
656
- return store.setValue(key, value, options);
657
- }
658
-
659
- /**
660
- * Gets the actor input value from the default {@apilink KeyValueStore} associated with the current actor run.
661
- *
662
- * This is just a convenient shortcut for [`keyValueStore.getValue('INPUT')`](core/class/KeyValueStore#getValue).
663
- * For example, calling the following code:
664
- * ```javascript
665
- * const input = await Actor.getInput();
666
- * ```
667
- *
668
- * is equivalent to:
669
- * ```javascript
670
- * const store = await Actor.openKeyValueStore();
671
- * await store.getValue('INPUT');
672
- * ```
673
- *
674
- * Note that the `getInput()` function does not cache the value read from the key-value store.
675
- * If you need to use the input multiple times in your actor,
676
- * it is far more efficient to read it once and store it locally.
677
- *
678
- * For more information, see {@apilink Actor.openKeyValueStore}
679
- * and {@apilink KeyValueStore.getValue}.
680
- *
681
- * @returns
682
- * Returns a promise that resolves to an object, string
683
- * or [`Buffer`](https://nodejs.org/api/buffer.html), depending
684
- * on the MIME content type of the record, or `null`
685
- * if the record is missing.
686
- * @ignore
687
- */
688
- async getInput<T = Dictionary | string | Buffer>(): Promise<T | null> {
689
- this._ensureActorInit('getInput');
690
-
691
- const inputSecretsPrivateKeyFile = this.config.get('inputSecretsPrivateKeyFile');
692
- const inputSecretsPrivateKeyPassphrase = this.config.get('inputSecretsPrivateKeyPassphrase');
693
- const input = await this.getValue<T>(this.config.get('inputKey'));
694
- if (ow.isValid(input, ow.object.nonEmpty) && inputSecretsPrivateKeyFile && inputSecretsPrivateKeyPassphrase) {
695
- const privateKey = createPrivateKey({
696
- key: Buffer.from(inputSecretsPrivateKeyFile, 'base64'),
697
- passphrase: inputSecretsPrivateKeyPassphrase,
698
- });
699
- return decryptInputSecrets<T>({ input, privateKey });
700
- }
701
- return input;
702
- }
703
-
704
- /**
705
- * Opens a key-value store and returns a promise resolving to an instance of the {@apilink KeyValueStore} class.
706
- *
707
- * Key-value stores are used to store records or files, along with their MIME content type.
708
- * The records are stored and retrieved using a unique key.
709
- * The actual data is stored either on a local filesystem or in the Apify cloud.
710
- *
711
- * For more details and code examples, see the {@apilink KeyValueStore} class.
712
- *
713
- * @param [storeIdOrName]
714
- * ID or name of the key-value store to be opened. If `null` or `undefined`,
715
- * the function returns the default key-value store associated with the actor run.
716
- * @param [options]
717
- * @ignore
718
- */
719
- async openKeyValueStore(storeIdOrName?: string | null, options: OpenStorageOptions = {}): Promise<KeyValueStore> {
720
- ow(storeIdOrName, ow.optional.string);
721
- ow(options, ow.object.exactShape({
722
- forceCloud: ow.optional.boolean,
723
- }));
724
-
725
- this._ensureActorInit('openKeyValueStore');
726
-
727
- return this._openStorage(KeyValueStore, storeIdOrName, options);
728
- }
729
-
730
- /**
731
- * Opens a request queue and returns a promise resolving to an instance
732
- * of the {@apilink RequestQueue} class.
733
- *
734
- * {@apilink RequestQueue} represents a queue of URLs to crawl, which is stored either on local filesystem or in the cloud.
735
- * The queue is used for deep crawling of websites, where you start with several URLs and then
736
- * recursively follow links to other pages. The data structure supports both breadth-first
737
- * and depth-first crawling orders.
738
- *
739
- * For more details and code examples, see the {@apilink RequestQueue} class.
740
- *
741
- * @param [queueIdOrName]
742
- * ID or name of the request queue to be opened. If `null` or `undefined`,
743
- * the function returns the default request queue associated with the actor run.
744
- * @param [options]
745
- * @ignore
746
- */
747
- async openRequestQueue(queueIdOrName?: string | null, options: OpenStorageOptions = {}): Promise<RequestQueue> {
748
- ow(queueIdOrName, ow.optional.string);
749
- ow(options, ow.object.exactShape({
750
- forceCloud: ow.optional.boolean,
751
- }));
752
-
753
- this._ensureActorInit('openRequestQueue');
754
-
755
- return this._openStorage(RequestQueue, queueIdOrName, options);
756
- }
757
-
758
- /**
759
- * Creates a proxy configuration and returns a promise resolving to an instance
760
- * of the {@apilink ProxyConfiguration} class that is already initialized.
761
- *
762
- * Configures connection to a proxy server with the provided options. Proxy servers are used to prevent target websites from blocking
763
- * your crawlers based on IP address rate limits or blacklists. Setting proxy configuration in your crawlers automatically configures
764
- * them to use the selected proxies for all connections.
765
- *
766
- * For more details and code examples, see the {@apilink ProxyConfiguration} class.
767
- *
768
- * ```javascript
769
- *
770
- * // Returns initialized proxy configuration class
771
- * const proxyConfiguration = await Actor.createProxyConfiguration({
772
- * groups: ['GROUP1', 'GROUP2'] // List of Apify proxy groups
773
- * countryCode: 'US'
774
- * });
775
- *
776
- * const crawler = new CheerioCrawler({
777
- * // ...
778
- * proxyConfiguration,
779
- * requestHandler({ proxyInfo }) {
780
- * const usedProxyUrl = proxyInfo.url; // Getting the proxy URL
781
- * }
782
- * })
783
- *
784
- * ```
785
- *
786
- * For compatibility with existing Actor Input UI (Input Schema), this function
787
- * returns `undefined` when the following object is passed as `proxyConfigurationOptions`.
788
- *
789
- * ```
790
- * { useApifyProxy: false }
791
- * ```
792
- * @ignore
793
- */
794
- async createProxyConfiguration(
795
- proxyConfigurationOptions: ProxyConfigurationOptions & { useApifyProxy?: boolean } = {},
796
- ): Promise<ProxyConfiguration | undefined> {
797
- // Compatibility fix for Input UI where proxy: None returns { useApifyProxy: false }
798
- // Without this, it would cause proxy to use the zero config / auto mode.
799
- const { useApifyProxy, ...options } = proxyConfigurationOptions;
800
- const dontUseApifyProxy = useApifyProxy === false;
801
- const dontUseCustomProxies = !proxyConfigurationOptions.proxyUrls;
802
-
803
- if (dontUseApifyProxy && dontUseCustomProxies) {
804
- return undefined;
805
- }
806
-
807
- const proxyConfiguration = new ProxyConfiguration(options, this.config);
808
- await proxyConfiguration.initialize();
809
-
810
- return proxyConfiguration;
811
- }
812
-
813
- /**
814
- * Returns a new {@apilink ApifyEnv} object which contains information parsed from all the `APIFY_XXX` environment variables.
815
- *
816
- * For the list of the `APIFY_XXX` environment variables, see
817
- * [Actor documentation](https://docs.apify.com/actor/run#environment-variables).
818
- * If some variables are not defined or are invalid, the corresponding value in the resulting object will be null.
819
- * @ignore
820
- */
821
- getEnv(): ApifyEnv {
822
- // NOTE: Don't throw if env vars are invalid to simplify local development and debugging of actors
823
- const env = process.env || {};
824
- const envVars = {} as ApifyEnv;
825
-
826
- for (const [shortName, fullName] of Object.entries(ENV_VARS)) {
827
- const camelCaseName = snakeCaseToCamelCase(shortName) as keyof ApifyEnv;
828
- let value: string | number | Date | undefined = env[fullName];
829
-
830
- // Parse dates and integers.
831
- if (value && fullName.endsWith('_AT')) {
832
- const unix = Date.parse(value);
833
- value = unix > 0 ? new Date(unix) : undefined;
834
- } else if ((INTEGER_ENV_VARS as readonly string[]).includes(fullName)) {
835
- value = parseInt(value!, 10);
836
- }
837
-
838
- Reflect.set(envVars, camelCaseName, value || value === 0 ? value : null);
839
- }
840
-
841
- return envVars;
842
- }
843
-
844
- /**
845
- * Returns a new instance of the Apify API client. The `ApifyClient` class is provided
846
- * by the [apify-client](https://www.npmjs.com/package/apify-client)
847
- * NPM package, and it is automatically configured using the `APIFY_API_BASE_URL`, and `APIFY_TOKEN`
848
- * environment variables. You can override the token via the available options. That's useful
849
- * if you want to use the client as a different Apify user than the SDK internals are using.
850
- * @ignore
851
- */
852
- newClient(options: ApifyClientOptions = {}): ApifyClient {
853
- const { storageDir, ...storageClientOptions } = this.config.get('storageClientOptions') as Dictionary;
854
- return new ApifyClient({
855
- baseUrl: this.config.get('apiBaseUrl'),
856
- token: this.config.get('token'),
857
- ...storageClientOptions,
858
- ...options, // allow overriding the instance configuration
859
- });
860
- }
861
-
862
- /**
863
- * Returns `true` when code is running on Apify platform and `false` otherwise (for example locally).
864
- * @ignore
865
- */
866
- isAtHome(): boolean {
867
- return !!process.env[ENV_VARS.IS_AT_HOME];
868
- }
869
-
870
- /**
871
- * Easily create and manage state values. All state values are automatically persisted.
872
- *
873
- * Values can be modified by simply using the assignment operator.
874
- *
875
- * @param name The name of the store to use.
876
- * @param defaultValue If the store does not yet have a value in it, the value will be initialized with the `defaultValue` you provide.
877
- * @param options An optional object parameter where a custom `keyValueStoreName` and `config` can be passed in.
878
- */
879
- async useState<State extends Dictionary = Dictionary>(
880
- name?: string,
881
- defaultValue = {} as State,
882
- options?: UseStateOptions,
883
- ) {
884
- const kvStore = await KeyValueStore.open(options?.keyValueStoreName, { config: options?.config || Configuration.getGlobalConfig() });
885
- return kvStore.getAutoSavedValue<State>(name || 'APIFY_GLOBAL_STATE', defaultValue);
886
- }
887
-
888
- /**
889
- * Easily create and manage state values. All state values are automatically persisted.
890
- *
891
- * Values can be modified by simply using the assignment operator.
892
- *
893
- * @param name The name of the store to use.
894
- * @param defaultValue If the store does not yet have a value in it, the value will be initialized with the `defaultValue` you provide.
895
- * @param options An optional object parameter where a custom `keyValueStoreName` and `config` can be passed in.
896
- */
897
- static async useState<State extends Dictionary = Dictionary>(
898
- name?: string,
899
- defaultValue = {} as State,
900
- options?: UseStateOptions,
901
- ) {
902
- return Actor.getDefaultInstance().useState<State>(name, defaultValue, options);
903
- }
904
-
905
- /**
906
- * Runs the main user function that performs the job of the actor
907
- * and terminates the process when the user function finishes.
908
- *
909
- * **The `Actor.main()` function is optional** and is provided merely for your convenience.
910
- * It is mainly useful when you're running your code as an actor on the [Apify platform](https://apify.com/actors).
911
- * However, if you want to use Apify SDK tools directly inside your existing projects, e.g.
912
- * running in an [Express](https://expressjs.com/) server, on
913
- * [Google Cloud functions](https://cloud.google.com/functions)
914
- * or [AWS Lambda](https://aws.amazon.com/lambda/), it's better to avoid
915
- * it since the function terminates the main process when it finishes!
916
- *
917
- * The `Actor.main()` function performs the following actions:
918
- *
919
- * - When running on the Apify platform (i.e. `APIFY_IS_AT_HOME` environment variable is set),
920
- * it sets up a connection to listen for platform events.
921
- * For example, to get a notification about an imminent migration to another server.
922
- * See {@apilink Actor.events} for details.
923
- * - It checks that either `APIFY_TOKEN` or `APIFY_LOCAL_STORAGE_DIR` environment variable
924
- * is defined. If not, the functions sets `APIFY_LOCAL_STORAGE_DIR` to `./apify_storage`
925
- * inside the current working directory. This is to simplify running code examples.
926
- * - It invokes the user function passed as the `userFunc` parameter.
927
- * - If the user function returned a promise, waits for it to resolve.
928
- * - If the user function throws an exception or some other error is encountered,
929
- * prints error details to console so that they are stored to the log.
930
- * - Exits the Node.js process, with zero exit code on success and non-zero on errors.
931
- *
932
- * The user function can be synchronous:
933
- *
934
- * ```javascript
935
- * await Actor.main(() => {
936
- * // My synchronous function that returns immediately
937
- * console.log('Hello world from actor!');
938
- * });
939
- * ```
940
- *
941
- * If the user function returns a promise, it is considered asynchronous:
942
- * ```javascript
943
- * import { gotScraping } from 'got-scraping';
944
- *
945
- * await Actor.main(() => {
946
- * // My asynchronous function that returns a promise
947
- * return gotScraping('http://www.example.com').then((html) => {
948
- * console.log(html);
949
- * });
950
- * });
951
- * ```
952
- *
953
- * To simplify your code, you can take advantage of the `async`/`await` keywords:
954
- *
955
- * ```javascript
956
- * import { gotScraping } from 'got-scraping';
957
- *
958
- * await Actor.main(async () => {
959
- * // My asynchronous function
960
- * const html = await gotScraping('http://www.example.com');
961
- * console.log(html);
962
- * });
963
- * ```
964
- *
965
- * @param userFunc User function to be executed. If it returns a promise,
966
- * the promise will be awaited. The user function is called with no arguments.
967
- * @param options
968
- */
969
- static main<T>(userFunc: UserFunc<T>, options?: MainOptions): Promise<T> {
970
- return Actor.getDefaultInstance().main<T>(userFunc, options);
971
- }
972
-
973
- static async init(options: InitOptions = {}): Promise<void> {
974
- return Actor.getDefaultInstance().init(options);
975
- }
976
-
977
- static async exit(messageOrOptions?: string | ExitOptions, options: ExitOptions = {}): Promise<void> {
978
- return Actor.getDefaultInstance().exit(messageOrOptions, options);
979
- }
980
-
981
- static async fail(messageOrOptions?: string | ExitOptions, options: ExitOptions = {}): Promise<void> {
982
- return Actor.getDefaultInstance().fail(messageOrOptions, options);
983
- }
984
-
985
- static on(event: EventTypeName, listener: (...args: any[]) => any): void {
986
- Actor.getDefaultInstance().on(event, listener);
987
- }
988
-
989
- static off(event: EventTypeName, listener?: (...args: any[]) => any): void {
990
- Actor.getDefaultInstance().off(event, listener);
991
- }
992
-
993
- /**
994
- * Runs an actor on the Apify platform using the current user account (determined by the `APIFY_TOKEN` environment variable).
995
- *
996
- * The result of the function is an {@apilink ActorRun} object that contains details about the actor run.
997
- *
998
- * If you want to run an actor task rather than an actor, please use the {@apilink Actor.callTask} function instead.
999
- *
1000
- * For more information about actors, read the [documentation](https://docs.apify.com/actor).
1001
- *
1002
- * **Example usage:**
1003
- *
1004
- * ```javascript
1005
- * const run = await Actor.call('apify/hello-world', { myInput: 123 });
1006
- * ```
1007
- *
1008
- * @param actorId
1009
- * Allowed formats are `username/actor-name`, `userId/actor-name` or actor ID.
1010
- * @param [input]
1011
- * Input for the actor. If it is an object, it will be stringified to
1012
- * JSON and its content type set to `application/json; charset=utf-8`.
1013
- * Otherwise the `options.contentType` parameter must be provided.
1014
- * @param [options]
1015
- */
1016
- static async call(actorId: string, input?: unknown, options: CallOptions = {}): Promise<ClientActorRun> {
1017
- return Actor.getDefaultInstance().call(actorId, input, options);
1018
- }
1019
-
1020
- /**
1021
- * Runs an actor task on the Apify platform using the current user account (determined by the `APIFY_TOKEN` environment variable).
1022
- *
1023
- * The result of the function is an {@apilink ActorRun} object that contains details about the actor run.
1024
- *
1025
- * Note that an actor task is a saved input configuration and options for an actor.
1026
- * If you want to run an actor directly rather than an actor task, please use the
1027
- * {@apilink Actor.call} function instead.
1028
- *
1029
- * For more information about actor tasks, read the [documentation](https://docs.apify.com/tasks).
1030
- *
1031
- * **Example usage:**
1032
- *
1033
- * ```javascript
1034
- * const run = await Actor.callTask('bob/some-task');
1035
- * ```
1036
- *
1037
- * @param taskId
1038
- * Allowed formats are `username/task-name`, `userId/task-name` or task ID.
1039
- * @param [input]
1040
- * Input overrides for the actor task. If it is an object, it will be stringified to
1041
- * JSON and its content type set to `application/json; charset=utf-8`.
1042
- * Provided input will be merged with actor task input.
1043
- * @param [options]
1044
- */
1045
- static async callTask(taskId: string, input?: Dictionary, options: CallTaskOptions = {}): Promise<ClientActorRun> {
1046
- return Actor.getDefaultInstance().callTask(taskId, input, options);
1047
- }
1048
-
1049
- /**
1050
- * Runs an actor on the Apify platform using the current user account (determined by the `APIFY_TOKEN` environment variable),
1051
- * unlike `Actor.call`, this method just starts the run without waiting for finish.
1052
- *
1053
- * The result of the function is an {@apilink ActorRun} object that contains details about the actor run.
1054
- *
1055
- * For more information about actors, read the
1056
- * [documentation](https://docs.apify.com/actor).
1057
- *
1058
- * **Example usage:**
1059
- *
1060
- * ```javascript
1061
- * const run = await Actor.start('apify/hello-world', { myInput: 123 });
1062
- * ```
1063
- *
1064
- * @param actorId
1065
- * Allowed formats are `username/actor-name`, `userId/actor-name` or actor ID.
1066
- * @param [input]
1067
- * Input for the actor. If it is an object, it will be stringified to
1068
- * JSON and its content type set to `application/json; charset=utf-8`.
1069
- * Otherwise the `options.contentType` parameter must be provided.
1070
- * @param [options]
1071
- */
1072
- static async start(actorId: string, input?: Dictionary, options: CallOptions = {}): Promise<ClientActorRun> {
1073
- return Actor.getDefaultInstance().start(actorId, input, options);
1074
- }
1075
-
1076
- /**
1077
- * Aborts given actor run on the Apify platform using the current user account (determined by the `APIFY_TOKEN` environment variable).
1078
- *
1079
- * The result of the function is an {@apilink ActorRun} object that contains details about the actor run.
1080
- *
1081
- * For more information about actors, read the
1082
- * [documentation](https://docs.apify.com/actor).
1083
- *
1084
- * **Example usage:**
1085
- *
1086
- * ```javascript
1087
- * const run = await Actor.abort(runId);
1088
- * ```
1089
- */
1090
- static async abort(runId: string, options: AbortOptions = {}): Promise<ClientActorRun> {
1091
- return Actor.getDefaultInstance().abort(runId, options);
1092
- }
1093
-
1094
- /**
1095
- * Transforms this actor run to an actor run of a given actor. The system stops the current container and starts
1096
- * the new container instead. All the default storages are preserved and the new input is stored under the `INPUT-METAMORPH-1` key
1097
- * in the same default key-value store.
1098
- *
1099
- * @param targetActorId
1100
- * Either `username/actor-name` or actor ID of an actor to which we want to metamorph.
1101
- * @param [input]
1102
- * Input for the actor. If it is an object, it will be stringified to
1103
- * JSON and its content type set to `application/json; charset=utf-8`.
1104
- * Otherwise, the `options.contentType` parameter must be provided.
1105
- * @param [options]
1106
- */
1107
- static async metamorph(targetActorId: string, input?: unknown, options: MetamorphOptions = {}): Promise<void> {
1108
- return Actor.getDefaultInstance().metamorph(targetActorId, input, options);
1109
- }
1110
-
1111
- /**
1112
- * Internally reboots this actor run. The system stops the current container and starts
1113
- * a new container with the same run id.
1114
- */
1115
- static async reboot(): Promise<void> {
1116
- return Actor.getDefaultInstance().reboot();
1117
- }
1118
-
1119
- /**
1120
- * Creates an ad-hoc webhook for the current actor run, which lets you receive a notification when the actor run finished or failed.
1121
- * For more information about Apify actor webhooks, please see the [documentation](https://docs.apify.com/webhooks).
1122
- *
1123
- * Note that webhooks are only supported for actors running on the Apify platform.
1124
- * In local environment, the function will print a warning and have no effect.
1125
- *
1126
- * @param options
1127
- * @returns The return value is the Webhook object.
1128
- * For more information, see the [Get webhook](https://apify.com/docs/api/v2#/reference/webhooks/webhook-object/get-webhook) API endpoint.
1129
- */
1130
- static async addWebhook(options: WebhookOptions): Promise<Webhook | undefined> {
1131
- return Actor.getDefaultInstance().addWebhook(options);
1132
- }
1133
-
1134
- /**
1135
- * Sets the status message for the current actor run.
1136
- *
1137
- * @param options
1138
- * @returns The return value is the Run object.
1139
- * For more information, see the [Actor Runs](https://docs.apify.com/api/v2#/reference/actor-runs/) API endpoints.
1140
- */
1141
- static async setStatusMessage(statusMessage: string): Promise<ClientActorRun> {
1142
- return Actor.getDefaultInstance().setStatusMessage(statusMessage);
1143
- }
1144
-
1145
- /**
1146
- * Stores an object or an array of objects to the default {@apilink Dataset} of the current actor run.
1147
- *
1148
- * This is just a convenient shortcut for {@apilink Dataset.pushData}.
1149
- * For example, calling the following code:
1150
- * ```javascript
1151
- * await Actor.pushData({ myValue: 123 });
1152
- * ```
1153
- *
1154
- * is equivalent to:
1155
- * ```javascript
1156
- * const dataset = await Actor.openDataset();
1157
- * await dataset.pushData({ myValue: 123 });
1158
- * ```
1159
- *
1160
- * For more information, see {@apilink Actor.openDataset} and {@apilink Dataset.pushData}
1161
- *
1162
- * **IMPORTANT**: Make sure to use the `await` keyword when calling `pushData()`,
1163
- * otherwise the actor process might finish before the data are stored!
1164
- *
1165
- * @param item Object or array of objects containing data to be stored in the default dataset.
1166
- * The objects must be serializable to JSON and the JSON representation of each object must be smaller than 9MB.
1167
- */
1168
- static async pushData<Data extends Dictionary = Dictionary>(item: Data | Data[]): Promise<void> {
1169
- return Actor.getDefaultInstance().pushData(item);
1170
- }
1171
-
1172
- /**
1173
- * Opens a dataset and returns a promise resolving to an instance of the {@apilink Dataset} class.
1174
- *
1175
- * Datasets are used to store structured data where each object stored has the same attributes,
1176
- * such as online store products or real estate offers.
1177
- * The actual data is stored either on the local filesystem or in the cloud.
1178
- *
1179
- * For more details and code examples, see the {@apilink Dataset} class.
1180
- *
1181
- * @param [datasetIdOrName]
1182
- * ID or name of the dataset to be opened. If `null` or `undefined`,
1183
- * the function returns the default dataset associated with the actor run.
1184
- * @param [options]
1185
- */
1186
- static async openDataset<Data extends Dictionary = Dictionary>(
1187
- datasetIdOrName?: string | null, options: OpenStorageOptions = {},
1188
- ): Promise<Dataset<Data>> {
1189
- return Actor.getDefaultInstance().openDataset(datasetIdOrName, options);
1190
- }
1191
-
1192
- /**
1193
- * Gets a value from the default {@apilink KeyValueStore} associated with the current actor run.
1194
- *
1195
- * This is just a convenient shortcut for {@apilink KeyValueStore.getValue}.
1196
- * For example, calling the following code:
1197
- * ```javascript
1198
- * const value = await Actor.getValue('my-key');
1199
- * ```
1200
- *
1201
- * is equivalent to:
1202
- * ```javascript
1203
- * const store = await Actor.openKeyValueStore();
1204
- * const value = await store.getValue('my-key');
1205
- * ```
1206
- *
1207
- * To store the value to the default key-value store, you can use the {@apilink Actor.setValue} function.
1208
- *
1209
- * For more information, see {@apilink Actor.openKeyValueStore}
1210
- * and {@apilink KeyValueStore.getValue}.
1211
- *
1212
- * @param key Unique record key.
1213
- * @returns
1214
- * Returns a promise that resolves to an object, string
1215
- * or [`Buffer`](https://nodejs.org/api/buffer.html), depending
1216
- * on the MIME content type of the record, or `null`
1217
- * if the record is missing.
1218
- */
1219
- static async getValue<T = unknown>(key: string): Promise<T | null> {
1220
- return Actor.getDefaultInstance().getValue(key);
1221
- }
1222
-
1223
- /**
1224
- * Stores or deletes a value in the default {@apilink KeyValueStore} associated with the current actor run.
1225
- *
1226
- * This is just a convenient shortcut for {@apilink KeyValueStore.setValue}.
1227
- * For example, calling the following code:
1228
- * ```javascript
1229
- * await Actor.setValue('OUTPUT', { foo: "bar" });
1230
- * ```
1231
- *
1232
- * is equivalent to:
1233
- * ```javascript
1234
- * const store = await Actor.openKeyValueStore();
1235
- * await store.setValue('OUTPUT', { foo: "bar" });
1236
- * ```
1237
- *
1238
- * To get a value from the default key-value store, you can use the {@apilink Actor.getValue} function.
1239
- *
1240
- * For more information, see {@apilink Actor.openKeyValueStore}
1241
- * and {@apilink KeyValueStore.getValue}.
1242
- *
1243
- * @param key
1244
- * Unique record key.
1245
- * @param value
1246
- * Record data, which can be one of the following values:
1247
- * - If `null`, the record in the key-value store is deleted.
1248
- * - If no `options.contentType` is specified, `value` can be any JavaScript object, and it will be stringified to JSON.
1249
- * - If `options.contentType` is set, `value` is taken as is, and it must be a `String` or [`Buffer`](https://nodejs.org/api/buffer.html).
1250
- * For any other value an error will be thrown.
1251
- * @param [options]
1252
- */
1253
- static async setValue<T>(key: string, value: T | null, options: RecordOptions = {}): Promise<void> {
1254
- return Actor.getDefaultInstance().setValue(key, value, options);
1255
- }
1256
-
1257
- /**
1258
- * Gets the actor input value from the default {@apilink KeyValueStore} associated with the current actor run.
1259
- *
1260
- * This is just a convenient shortcut for {@apilink KeyValueStore.getValue | `keyValueStore.getValue('INPUT')`}.
1261
- * For example, calling the following code:
1262
- * ```javascript
1263
- * const input = await Actor.getInput();
1264
- * ```
1265
- *
1266
- * is equivalent to:
1267
- * ```javascript
1268
- * const store = await Actor.openKeyValueStore();
1269
- * await store.getValue('INPUT');
1270
- * ```
1271
- *
1272
- * Note that the `getInput()` function does not cache the value read from the key-value store.
1273
- * If you need to use the input multiple times in your actor,
1274
- * it is far more efficient to read it once and store it locally.
1275
- *
1276
- * For more information, see {@apilink Actor.openKeyValueStore} and {@apilink KeyValueStore.getValue}.
1277
- *
1278
- * @returns
1279
- * Returns a promise that resolves to an object, string
1280
- * or [`Buffer`](https://nodejs.org/api/buffer.html), depending
1281
- * on the MIME content type of the record, or `null`
1282
- * if the record is missing.
1283
- */
1284
- static async getInput<T = Dictionary | string | Buffer>(): Promise<T | null> {
1285
- return Actor.getDefaultInstance().getInput();
1286
- }
1287
-
1288
- /**
1289
- * Opens a key-value store and returns a promise resolving to an instance of the {@apilink KeyValueStore} class.
1290
- *
1291
- * Key-value stores are used to store records or files, along with their MIME content type.
1292
- * The records are stored and retrieved using a unique key.
1293
- * The actual data is stored either on a local filesystem or in the Apify cloud.
1294
- *
1295
- * For more details and code examples, see the {@apilink KeyValueStore} class.
1296
- *
1297
- * @param [storeIdOrName]
1298
- * ID or name of the key-value store to be opened. If `null` or `undefined`,
1299
- * the function returns the default key-value store associated with the actor run.
1300
- * @param [options]
1301
- */
1302
- static async openKeyValueStore(storeIdOrName?: string | null, options: OpenStorageOptions = {}): Promise<KeyValueStore> {
1303
- return Actor.getDefaultInstance().openKeyValueStore(storeIdOrName, options);
1304
- }
1305
-
1306
- /**
1307
- * Opens a request queue and returns a promise resolving to an instance
1308
- * of the {@apilink RequestQueue} class.
1309
- *
1310
- * {@apilink RequestQueue} represents a queue of URLs to crawl, which is stored either on local filesystem or in the cloud.
1311
- * The queue is used for deep crawling of websites, where you start with several URLs and then
1312
- * recursively follow links to other pages. The data structure supports both breadth-first
1313
- * and depth-first crawling orders.
1314
- *
1315
- * For more details and code examples, see the {@apilink RequestQueue} class.
1316
- *
1317
- * @param [queueIdOrName]
1318
- * ID or name of the request queue to be opened. If `null` or `undefined`,
1319
- * the function returns the default request queue associated with the actor run.
1320
- * @param [options]
1321
- */
1322
- static async openRequestQueue(queueIdOrName?: string | null, options: OpenStorageOptions = {}): Promise<RequestQueue> {
1323
- return Actor.getDefaultInstance().openRequestQueue(queueIdOrName, options);
1324
- }
1325
-
1326
- /**
1327
- * Creates a proxy configuration and returns a promise resolving to an instance
1328
- * of the {@apilink ProxyConfiguration} class that is already initialized.
1329
- *
1330
- * Configures connection to a proxy server with the provided options. Proxy servers are used to prevent target websites from blocking
1331
- * your crawlers based on IP address rate limits or blacklists. Setting proxy configuration in your crawlers automatically configures
1332
- * them to use the selected proxies for all connections.
1333
- *
1334
- * For more details and code examples, see the {@apilink ProxyConfiguration} class.
1335
- *
1336
- * ```javascript
1337
- *
1338
- * // Returns initialized proxy configuration class
1339
- * const proxyConfiguration = await Actor.createProxyConfiguration({
1340
- * groups: ['GROUP1', 'GROUP2'] // List of Apify proxy groups
1341
- * countryCode: 'US'
1342
- * });
1343
- *
1344
- * const crawler = new CheerioCrawler({
1345
- * // ...
1346
- * proxyConfiguration,
1347
- * requestHandler({ proxyInfo }) {
1348
- * const usedProxyUrl = proxyInfo.url; // Getting the proxy URL
1349
- * }
1350
- * })
1351
- *
1352
- * ```
1353
- *
1354
- * For compatibility with existing Actor Input UI (Input Schema), this function
1355
- * returns `undefined` when the following object is passed as `proxyConfigurationOptions`.
1356
- *
1357
- * ```
1358
- * { useApifyProxy: false }
1359
- * ```
1360
- */
1361
- static async createProxyConfiguration(
1362
- proxyConfigurationOptions: ProxyConfigurationOptions & { useApifyProxy?: boolean } = {},
1363
- ): Promise<ProxyConfiguration | undefined> {
1364
- return Actor.getDefaultInstance().createProxyConfiguration(proxyConfigurationOptions);
1365
- }
1366
-
1367
- /**
1368
- * Returns a new {@apilink ApifyEnv} object which contains information parsed from all the `APIFY_XXX` environment variables.
1369
- *
1370
- * For the list of the `APIFY_XXX` environment variables, see
1371
- * [Actor documentation](https://docs.apify.com/actor/run#environment-variables).
1372
- * If some of the variables are not defined or are invalid, the corresponding value in the resulting object will be null.
1373
- */
1374
- static getEnv(): ApifyEnv {
1375
- return Actor.getDefaultInstance().getEnv();
1376
- }
1377
-
1378
- /**
1379
- * Returns a new instance of the Apify API client. The `ApifyClient` class is provided
1380
- * by the [apify-client](https://www.npmjs.com/package/apify-client)
1381
- * NPM package, and it is automatically configured using the `APIFY_API_BASE_URL`, and `APIFY_TOKEN`
1382
- * environment variables. You can override the token via the available options. That's useful
1383
- * if you want to use the client as a different Apify user than the SDK internals are using.
1384
- */
1385
- static newClient(options: ApifyClientOptions = {}): ApifyClient {
1386
- return Actor.getDefaultInstance().newClient(options);
1387
- }
1388
-
1389
- /**
1390
- * Returns `true` when code is running on Apify platform and `false` otherwise (for example locally).
1391
- */
1392
- static isAtHome(): boolean {
1393
- return Actor.getDefaultInstance().isAtHome();
1394
- }
1395
-
1396
- /** Default {@apilink ApifyClient} instance. */
1397
- static get apifyClient(): ApifyClient {
1398
- return Actor.getDefaultInstance().apifyClient;
1399
- }
1400
-
1401
- /** Default {@apilink Configuration} instance. */
1402
- static get config(): Configuration {
1403
- return Actor.getDefaultInstance().config;
1404
- }
1405
-
1406
- /** @internal */
1407
- static getDefaultInstance(): Actor {
1408
- this._instance ??= new Actor();
1409
- return this._instance;
1410
- }
1411
-
1412
- private _openStorage<T extends IStorage>(storageClass: Constructor<T>, id?: string, options: OpenStorageOptions = {}) {
1413
- const client = options.forceCloud ? this.apifyClient : undefined;
1414
- return StorageManager.openStorage<T>(storageClass, id, client, this.config);
1415
- }
1416
-
1417
- private _ensureActorInit(methodCalled: string) {
1418
- // If we already warned the user once, don't do it again to prevent spam
1419
- if (this.warnedAboutMissingInitCall) {
1420
- return;
1421
- }
1422
-
1423
- if (this.initialized) {
1424
- return;
1425
- }
1426
-
1427
- this.warnedAboutMissingInitCall = true;
1428
-
1429
- log.warning([
1430
- `Actor.${methodCalled}() was called but the actor instance was not initialized.`,
1431
- 'Did you forget to call Actor.init()?',
1432
- ].join('\n'));
1433
- }
1434
- }
1435
-
1436
- export interface InitOptions {
1437
- storage?: StorageClient;
1438
- }
1439
-
1440
- export interface MainOptions extends ExitOptions, InitOptions {}
1441
-
1442
- /**
1443
- * Parsed representation of the `APIFY_XXX` environmental variables.
1444
- * This object is returned by the {@apilink Actor.getEnv} function.
1445
- */
1446
- export interface ApifyEnv {
1447
- /**
1448
- * ID of the actor (APIFY_ACTOR_ID)
1449
- */
1450
- actorId: string | null;
1451
-
1452
- /**
1453
- * ID of the actor run (APIFY_ACTOR_RUN_ID)
1454
- */
1455
- actorRunId: string | null;
1456
-
1457
- /**
1458
- * ID of the actor task (APIFY_ACTOR_TASK_ID)
1459
- */
1460
- actorTaskId: string | null;
1461
-
1462
- /**
1463
- * ID of the user who started the actor - note that it might be
1464
- * different than the owner ofthe actor (APIFY_USER_ID)
1465
- */
1466
- userId: string | null;
1467
-
1468
- /**
1469
- * Authentication token representing privileges given to the actor run,
1470
- * it can be passed to various Apify APIs (APIFY_TOKEN)
1471
- */
1472
- token: string | null;
1473
-
1474
- /**
1475
- * Date when the actor was started (APIFY_STARTED_AT)
1476
- */
1477
- startedAt: Date | null;
1478
-
1479
- /**
1480
- * Date when the actor will time out (APIFY_TIMEOUT_AT)
1481
- */
1482
- timeoutAt: Date | null;
1483
-
1484
- /**
1485
- * ID of the key-value store where input and output data of this
1486
- * actor is stored (APIFY_DEFAULT_KEY_VALUE_STORE_ID)
1487
- */
1488
- defaultKeyValueStoreId: string | null;
1489
-
1490
- /**
1491
- * ID of the dataset where input and output data of this
1492
- * actor is stored (APIFY_DEFAULT_DATASET_ID)
1493
- */
1494
- defaultDatasetId: string | null;
1495
-
1496
- /**
1497
- * Amount of memory allocated for the actor,
1498
- * in megabytes (APIFY_MEMORY_MBYTES)
1499
- */
1500
- memoryMbytes: number | null;
1501
- }
1502
-
1503
- export type UserFunc<T = unknown> = () => Awaitable<T>;
1504
-
1505
- export interface CallOptions extends ActorStartOptions {
1506
- /**
1507
- * User API token that is used to run the actor. By default, it is taken from the `APIFY_TOKEN` environment variable.
1508
- */
1509
- token?: string;
1510
- }
1511
-
1512
- export interface CallTaskOptions extends TaskStartOptions {
1513
- /**
1514
- * User API token that is used to run the actor. By default, it is taken from the `APIFY_TOKEN` environment variable.
1515
- */
1516
- token?: string;
1517
- }
1518
-
1519
- export interface AbortOptions extends RunAbortOptions {
1520
- /**
1521
- * User API token that is used to run the actor. By default, it is taken from the `APIFY_TOKEN` environment variable.
1522
- */
1523
- token?: string;
1524
-
1525
- /** Exit with given status message */
1526
- statusMessage?: string;
1527
- }
1528
-
1529
- export interface WebhookOptions {
1530
- /**
1531
- * Array of event types, which you can set for actor run, see
1532
- * the [actor run events](https://docs.apify.com/webhooks/events#actor-run) in the Apify doc.
1533
- */
1534
- eventTypes: readonly WebhookEventType[];
1535
-
1536
- /**
1537
- * URL which will be requested using HTTP POST request, when actor run will reach the set event type.
1538
- */
1539
- requestUrl: string;
1540
-
1541
- /**
1542
- * Payload template is a JSON-like string that describes the structure of the webhook POST request payload.
1543
- * It uses JSON syntax, extended with a double curly braces syntax for injecting variables `{{variable}}`.
1544
- * Those variables are resolved at the time of the webhook's dispatch, and a list of available variables with their descriptions
1545
- * is available in the [Apify webhook documentation](https://docs.apify.com/webhooks).
1546
- * If `payloadTemplate` is omitted, the default payload template is used
1547
- * ([view docs](https://docs.apify.com/webhooks/actions#payload-template)).
1548
- */
1549
- payloadTemplate?: string;
1550
-
1551
- /**
1552
- * Idempotency key enables you to ensure that a webhook will not be added multiple times in case of
1553
- * an actor restart or other situation that would cause the `addWebhook()` function to be called again.
1554
- * We suggest using the actor run ID as the idempotency key. You can get the run ID by calling
1555
- * {@apilink Actor.getEnv} function.
1556
- */
1557
- idempotencyKey?: string;
1558
- }
1559
-
1560
- export interface MetamorphOptions {
1561
- /**
1562
- * Content type for the `input`. If not specified,
1563
- * `input` is expected to be an object that will be stringified to JSON and content type set to
1564
- * `application/json; charset=utf-8`. If `options.contentType` is specified, then `input` must be a
1565
- * `String` or `Buffer`.
1566
- */
1567
- contentType?: string;
1568
-
1569
- /**
1570
- * Tag or number of the target actor build to metamorph into (e.g. `beta` or `1.2.345`).
1571
- * If not provided, the run uses build tag or number from the default actor run configuration (typically `latest`).
1572
- */
1573
- build?: string;
1574
-
1575
- /** @internal */
1576
- customAfterSleepMillis?: number;
1577
- }
1578
-
1579
- export interface ExitOptions {
1580
- /** Exit with given status message */
1581
- statusMessage?: string;
1582
- /**
1583
- * Amount of time, in seconds, to wait for all event handlers to finish before exiting the process.
1584
- * @default 30
1585
- */
1586
- timeoutSecs?: number;
1587
- /** Exit code, defaults to 0 */
1588
- exitCode?: number;
1589
- /** Call `process.exit()`? Defaults to true */
1590
- exit?: boolean;
1591
- }
1592
-
1593
- export interface OpenStorageOptions {
1594
- /**
1595
- * If set to `true` then the cloud storage is used even if the `APIFY_LOCAL_STORAGE_DIR`
1596
- * environment variable is set. This way it is possible to combine local and cloud storage.
1597
- * @default false
1598
- */
1599
- forceCloud?: boolean;
1600
- }
1601
-
1602
- export { ClientActorRun as ActorRun };
1603
-
1604
- /**
1605
- * Exit codes for the actor process.
1606
- * The error codes must be in the range 1-128, to avoid collision with signal exits
1607
- * and to ensure Docker will handle them correctly!
1608
- * @internal should be removed if we decide to remove `Actor.main()`
1609
- */
1610
- export const EXIT_CODES = {
1611
- SUCCESS: 0,
1612
- ERROR_USER_FUNCTION_THREW: 91,
1613
- ERROR_UNKNOWN: 92,
1614
- };