@crawlee/basic 4.0.0-beta.120 → 4.0.0-beta.122

package/internals/basic-crawler.d.ts

@@ -1,9 +1,10 @@
-import type { AddRequestsBatchedOptions, AddRequestsBatchedResult, ConcurrencySystemOptions, Configuration, CrawleeLogger, CrawlingContext, DatasetExportOptions, EnqueueLinksOptions, EventManager, FinalStatistics, GetUserDataFromRequest, IConcurrencySystem, IProxyConfiguration, IRequestLoader, IRequestManager, IStatistics, Request, RequestsLike, RouterHandler, RouterRoutes, SkippedRequestCallback, Source, StatisticState, StorageIdentifier, StorageWritePolicy, TaskLoopPredicates, TypedRequestsLike } from '@crawlee/core';
-import { ConcurrencySystem, ContextPipeline, Dataset, RequestQueue } from '@crawlee/core';
-import type { Awaitable, BaseHttpClient, BatchAddRequestsResult, Dictionary, ISession, ISessionPool, ProxyInfo, SetStatusMessageOptions, StorageBackend } from '@crawlee/types';
+import type { AddRequestsBatchedOptions, AddRequestsBatchedResult, ConcurrencySystemOptions, CrawleeLogger, CrawlingContext, DatasetExportOptions, EnqueueLinksOptions, FinalStatistics, GetUserDataFromRequest, IConcurrencySystem, IProxyConfiguration, IRequestLoader, IRequestManager, IStatistics, Request, RequestsLike, RouterHandler, RouterRoutes, SkippedRequestCallback, Source, StatisticState, StorageIdentifier, StorageWritePolicy, TaskLoopPredicates, TypedRequestsLike } from '@crawlee/core';
+import { ConcurrencySystem, Configuration, ContextPipeline, Dataset, EventManager, RequestQueue } from '@crawlee/core';
+import { BaseHttpClient } from '@crawlee/http-client';
+import type { Awaitable, BatchAddRequestsResult, Dictionary, ISession, ISessionPool, ProxyInfo, SetStatusMessageOptions, StorageBackend } from '@crawlee/types';
 import { RobotsTxtFile } from '@crawlee/utils';
-import { type BasePredicate } from 'ow';
 import type { ReadonlyDeep, SetRequired } from 'type-fest';
+import { z } from 'zod';
 import { TimeoutError } from '@apify/timeout';
 export interface BasicCrawlingContext<UserData extends Dictionary = Dictionary> extends CrawlingContext<UserData> {
 }
@@ -514,80 +515,93 @@ export declare class BasicCrawler<Context extends CrawlingContext = CrawlingCont
     protected readonly onSkippedRequest?: SkippedRequestCallback;
     protected readonly identity: CrawlerIdentity;
     protected static optionsShape: {
-// @ts-ignore optional peer dependency or compatibility with es2022
-        contextPipelineBuilder: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        extendContext: import("ow").Predicate<Function> & BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestList: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestQueue: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestManager: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestHandler: import("ow").Predicate<Function> & BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestHandlerTimeoutSecs: import("ow").NumberPredicate & BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        errorHandler: import("ow").Predicate<Function> & BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        failedRequestHandler: import("ow").Predicate<Function> & BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxRequestRetries: import("ow").NumberPredicate & BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        sameDomainDelaySecs: import("ow").NumberPredicate & BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxRequestsPerCrawl: import("ow").NumberPredicate & BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxCrawlDepth: import("ow").NumberPredicate & BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        taskLoopOptions: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        concurrencySystem: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        sessionPool: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        proxyConfiguration: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        statusMessageLoggingInterval: import("ow").NumberPredicate & BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        statusMessageCallback: import("ow").Predicate<Function> & BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        additionalHttpErrorStatusCodes: import("ow").ArrayPredicate<number>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        ignoreHttpErrorStatusCodes: import("ow").ArrayPredicate<number>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        blockedStatusCodes: import("ow").ArrayPredicate<number>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        retryOnBlocked: import("ow").BooleanPredicate & BasePredicate<boolean | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        respectRobotsTxtFile: import("ow").AnyPredicate<boolean | object>;
-        transactionalStorage: BasePredicate<boolean | Partial<StorageWritePolicy> | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        onSkippedRequest: import("ow").Predicate<Function> & BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        httpClient: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        configuration: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        storageBackend: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        eventManager: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        logger: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        minConcurrency: import("ow").NumberPredicate & BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxConcurrency: import("ow").NumberPredicate & BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxRequestsPerMinute: import("ow").NumberPredicate & BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        keepAlive: import("ow").BooleanPredicate & BasePredicate<boolean | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        statistics: import("ow").ObjectPredicate<object> & BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        id: import("ow").StringPredicate & BasePredicate<string | undefined>;
+        contextPipelineBuilder: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        extendContext: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        requestList: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        requestQueue: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        requestManager: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        requestHandler: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        requestHandlerTimeoutSecs: z.ZodOptional<z.ZodCustom<number, number>>;
+        errorHandler: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        failedRequestHandler: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        maxRequestRetries: z.ZodDefault<z.ZodCustom<number, number>>;
+        sameDomainDelaySecs: z.ZodDefault<z.ZodCustom<number, number>>;
+        maxRequestsPerCrawl: z.ZodOptional<z.ZodCustom<number, number>>;
+        maxCrawlDepth: z.ZodOptional<z.ZodCustom<number, number>>;
+        taskLoopOptions: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        concurrencySystem: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        sessionPool: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        proxyConfiguration: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        statusMessageLoggingInterval: z.ZodDefault<z.ZodCustom<number, number>>;
+        statusMessageCallback: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        additionalHttpErrorStatusCodes: z.ZodDefault<z.ZodArray<z.ZodCustom<number, number>>>;
+        ignoreHttpErrorStatusCodes: z.ZodDefault<z.ZodArray<z.ZodCustom<number, number>>>;
+        blockedStatusCodes: z.ZodOptional<z.ZodArray<z.ZodCustom<number, number>>>;
+        retryOnBlocked: z.ZodDefault<z.ZodBoolean>;
+        respectRobotsTxtFile: z.ZodDefault<z.ZodUnion<readonly [z.ZodBoolean, z.ZodCustom<Dictionary, Dictionary>]>>;
+        transactionalStorage: z.ZodOptional<z.ZodUnion<readonly [z.ZodBoolean, z.ZodObject<{
+            requestQueue: z.ZodOptional<z.ZodEnum<{
+                deferred: "deferred";
+                writeThrough: "writeThrough";
+            }>>;
+        }, z.core.$strict>]>>;
+        onSkippedRequest: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        httpClient: z.ZodOptional<z.ZodCustom<BaseHttpClient, BaseHttpClient>>;
+        configuration: z.ZodOptional<z.ZodCustom<Configuration, Configuration>>;
+        storageBackend: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        eventManager: z.ZodOptional<z.ZodCustom<EventManager, EventManager>>;
+        logger: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        minConcurrency: z.ZodOptional<z.ZodCustom<number, number>>;
+        maxConcurrency: z.ZodOptional<z.ZodCustom<number, number>>;
+        maxRequestsPerMinute: z.ZodOptional<z.ZodCustom<number, number>>;
+        keepAlive: z.ZodOptional<z.ZodBoolean>;
+        statistics: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        id: z.ZodOptional<z.ZodString>;
     };
+    protected static optionsSchema: z.ZodObject<{
+        contextPipelineBuilder: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        extendContext: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        requestList: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        requestQueue: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        requestManager: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        requestHandler: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        requestHandlerTimeoutSecs: z.ZodOptional<z.ZodCustom<number, number>>;
+        errorHandler: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        failedRequestHandler: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        maxRequestRetries: z.ZodDefault<z.ZodCustom<number, number>>;
+        sameDomainDelaySecs: z.ZodDefault<z.ZodCustom<number, number>>;
+        maxRequestsPerCrawl: z.ZodOptional<z.ZodCustom<number, number>>;
+        maxCrawlDepth: z.ZodOptional<z.ZodCustom<number, number>>;
+        taskLoopOptions: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        concurrencySystem: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        sessionPool: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        proxyConfiguration: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        statusMessageLoggingInterval: z.ZodDefault<z.ZodCustom<number, number>>;
+        statusMessageCallback: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        additionalHttpErrorStatusCodes: z.ZodDefault<z.ZodArray<z.ZodCustom<number, number>>>;
+        ignoreHttpErrorStatusCodes: z.ZodDefault<z.ZodArray<z.ZodCustom<number, number>>>;
+        blockedStatusCodes: z.ZodOptional<z.ZodArray<z.ZodCustom<number, number>>>;
+        retryOnBlocked: z.ZodDefault<z.ZodBoolean>;
+        respectRobotsTxtFile: z.ZodDefault<z.ZodUnion<readonly [z.ZodBoolean, z.ZodCustom<Dictionary, Dictionary>]>>;
+        transactionalStorage: z.ZodOptional<z.ZodUnion<readonly [z.ZodBoolean, z.ZodObject<{
+            requestQueue: z.ZodOptional<z.ZodEnum<{
+                deferred: "deferred";
+                writeThrough: "writeThrough";
+            }>>;
+        }, z.core.$strict>]>>;
+        onSkippedRequest: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        httpClient: z.ZodOptional<z.ZodCustom<BaseHttpClient, BaseHttpClient>>;
+        configuration: z.ZodOptional<z.ZodCustom<Configuration, Configuration>>;
+        storageBackend: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        eventManager: z.ZodOptional<z.ZodCustom<EventManager, EventManager>>;
+        logger: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        minConcurrency: z.ZodOptional<z.ZodCustom<number, number>>;
+        maxConcurrency: z.ZodOptional<z.ZodCustom<number, number>>;
+        maxRequestsPerMinute: z.ZodOptional<z.ZodCustom<number, number>>;
+        keepAlive: z.ZodOptional<z.ZodBoolean>;
+        statistics: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        id: z.ZodOptional<z.ZodString>;
+    }, z.core.$strict>;
     /**
      * All `BasicCrawler` parameters are passed via an options object.
      */

package/internals/basic-crawler.js

@@ -1,19 +1,20 @@
 import { mkdir, writeFile } from 'node:fs/promises';
 import { dirname } from 'node:path';
-import { AutoscaledPool, bindMethodsToServiceLocator, BLOCKED_STATUS_CODES, ConcurrencySystem, ContextPipeline, ContextPipelineCleanupError, ContextPipelineInitializationError, ContextPipelineInterruptedError, createStorageTransaction, CriticalError, currentStorageTransaction, Dataset, enqueueLinks, EnqueueStrategy, EventType, getObjectType, KeyValueStore, log, LogLevel, mergeCookies, MissingSessionError, NavigationSkippedError, NonRetryableError, OwnedOrInjected, purgeDefaultStorages, RequestHandlerError, parseRetryAfterHeader, RequestThrottledError, RequestManagerTandem, RequestQueue, RequestState, RetryRequestError, supportsDomainThrottling, Router, ServiceLocator, serviceLocator, Session, SessionError, SessionPool, Statistics, validateUserData, validators, withDirectStorageAccess, } from '@crawlee/core';
-import { FetchHttpClient } from '@crawlee/http-client';
+import { AutoscaledPool, bindMethodsToServiceLocator, BLOCKED_STATUS_CODES, ConcurrencySystem, Configuration, ContextPipeline, ContextPipelineCleanupError, ContextPipelineInitializationError, ContextPipelineInterruptedError, createStorageTransaction, CriticalError, currentStorageTransaction, Dataset, enqueueLinks, EnqueueStrategy, EventManager, EventType, getObjectType, KeyValueStore, log, LogLevel, mergeCookies, MissingSessionError, NavigationSkippedError, NonRetryableError, OwnedOrInjected, parseArgument, purgeDefaultStorages, RequestHandlerError, parseRetryAfterHeader, RequestThrottledError, RequestManagerTandem, RequestQueue, RequestState, RetryRequestError, supportsDomainThrottling, Router, schemas, ServiceLocator, serviceLocator, Session, SessionError, SessionPool, Statistics, validateUserData, validators, withDirectStorageAccess, } from '@crawlee/core';
+import { BaseHttpClient, FetchHttpClient } from '@crawlee/http-client';
 import { isAsyncIterable, isIterable, ROTATE_PROXY_ERRORS } from '@crawlee/utils/internal';
 import { RobotsTxtFile } from '@crawlee/utils';
-import ow, { ArgumentError } from 'ow';
 import { getDomain } from 'tldts';
+import { z } from 'zod';
 import { LruCache } from '@apify/datastructures';
 import { addTimeoutToPromise, extendTimeout, TimeoutError } from '@apify/timeout';
 import { cryptoRandomObjectId } from '@apify/utilities';
 import { extendTimeoutKey, navigationDeadlineKey, raceWithTimeout, timeoutExpiredKey, } from './request-timeout.js';
 import { createSendRequest } from './send-request.js';
-class LazyDefaultHttpClient {
+class LazyDefaultHttpClient extends BaseHttpClient {
     #delegatePromise;
     constructor(options) {
+        super(options);
         this.#delegatePromise = import('@crawlee/impit-client')
             .then(({ ImpitHttpClient }) => new ImpitHttpClient(options))
             .catch(() => {
@@ -22,6 +23,9 @@ class LazyDefaultHttpClient {
             return new FetchHttpClient(options);
         });
     }
+    fetch() {
+        throw new Error('LazyDefaultHttpClient delegates `sendRequest` entirely; `fetch` is never called.');
+    }
     async sendRequest(...args) {
         return (await this.#delegatePromise).sendRequest(...args);
     }
@@ -200,72 +204,77 @@ export class BasicCrawler {
     identity;
     #contextPipelineOptions;
     static optionsShape = {
-        contextPipelineBuilder: ow.optional.object,
-        extendContext: ow.optional.function,
-        requestList: ow.optional.object.validate(validators.requestList),
-        requestQueue: ow.optional.object.validate(validators.requestQueue),
-        requestManager: ow.optional.object,
+        contextPipelineBuilder: schemas.anyObject.optional(),
+        extendContext: schemas.anyFunction.optional(),
+        requestList: validators.requestList.optional(),
+        requestQueue: validators.requestQueue.optional(),
+        requestManager: validators.requestManager.optional(),
         // Subclasses override this function instead of passing it
         // in constructor, so this validation needs to apply only
         // if the user creates an instance of BasicCrawler directly.
-        requestHandler: ow.optional.function,
-        requestHandlerTimeoutSecs: ow.optional.number,
-        errorHandler: ow.optional.function,
-        failedRequestHandler: ow.optional.function,
-        maxRequestRetries: ow.optional.number,
-        sameDomainDelaySecs: ow.optional.number,
-        maxRequestsPerCrawl: ow.optional.number,
-        maxCrawlDepth: ow.optional.number,
-        taskLoopOptions: ow.optional.object,
-        concurrencySystem: ow.optional.object,
-        sessionPool: ow.optional.object.validate(validators.sessionPool),
-        proxyConfiguration: ow.optional.object.validate(validators.proxyConfiguration),
-        statusMessageLoggingInterval: ow.optional.number,
-        statusMessageCallback: ow.optional.function,
-        additionalHttpErrorStatusCodes: ow.optional.array.ofType(ow.number),
-        ignoreHttpErrorStatusCodes: ow.optional.array.ofType(ow.number),
-        blockedStatusCodes: ow.optional.array.ofType(ow.number),
-        retryOnBlocked: ow.optional.boolean,
-        respectRobotsTxtFile: ow.optional.any(ow.boolean, ow.object),
-        transactionalStorage: ow.optional.any(ow.boolean, ow.object.exactShape({
-            requestQueue: ow.optional.string.oneOf(['deferred', 'writeThrough']),
-        })),
-        onSkippedRequest: ow.optional.function,
-        httpClient: ow.optional.object,
-        configuration: ow.optional.object,
-        storageBackend: ow.optional.object,
-        eventManager: ow.optional.object,
-        logger: ow.optional.object,
+        requestHandler: schemas.anyFunction.optional(),
+        requestHandlerTimeoutSecs: schemas.anyNumber.optional(),
+        errorHandler: schemas.anyFunction.optional(),
+        failedRequestHandler: schemas.anyFunction.optional(),
+        maxRequestRetries: schemas.anyNumber.default(3),
+        sameDomainDelaySecs: schemas.anyNumber.default(0),
+        maxRequestsPerCrawl: schemas.anyNumber.optional(),
+        maxCrawlDepth: schemas.anyNumber.optional(),
+        // No zod default — subclasses provide their own fallback (e.g. HTTP-optimized pool options).
+        taskLoopOptions: schemas.anyObject.optional(),
+        concurrencySystem: schemas.anyObject.optional(),
+        sessionPool: validators.sessionPool.optional(),
+        proxyConfiguration: validators.proxyConfiguration.optional(),
+        statusMessageLoggingInterval: schemas.anyNumber.default(10),
+        statusMessageCallback: schemas.anyFunction.optional(),
+        additionalHttpErrorStatusCodes: schemas.arrayOf(schemas.anyNumber, 'numbers').default(() => []),
+        ignoreHttpErrorStatusCodes: schemas.arrayOf(schemas.anyNumber, 'numbers').default(() => []),
+        blockedStatusCodes: schemas.arrayOf(schemas.anyNumber, 'numbers').optional(),
+        retryOnBlocked: z.boolean().default(false),
+        respectRobotsTxtFile: z.union([z.boolean(), schemas.anyObject]).default(false),
+        transactionalStorage: z
+            .union([z.boolean(), z.strictObject({ requestQueue: z.enum(['deferred', 'writeThrough']).optional() })])
+            .optional(),
+        onSkippedRequest: schemas.anyFunction.optional(),
+        httpClient: schemas.httpClient.optional(),
+        configuration: z.instanceof(Configuration).optional(),
+        storageBackend: validators.storageBackend.optional(),
+        eventManager: z.instanceof(EventManager).optional(),
+        logger: validators.logger.optional(),
         // AutoscaledPool shorthands
-        minConcurrency: ow.optional.number,
-        maxConcurrency: ow.optional.number,
-        maxRequestsPerMinute: ow.optional.number.integerOrInfinite.positive.greaterThanOrEqual(1),
-        keepAlive: ow.optional.boolean,
-        statistics: ow.optional.object,
-        id: ow.optional.string,
+        minConcurrency: schemas.anyNumber.optional(),
+        maxConcurrency: schemas.anyNumber.optional(),
+        maxRequestsPerMinute: schemas.anyNumber
+            .refine((value) => Number.isInteger(value) || value === Infinity, 'Expected an integer or infinite number')
+            .refine((value) => value >= 1, 'Expected a number greater than or equal to 1')
+            .optional(),
+        keepAlive: z.boolean().optional(),
+        statistics: schemas.anyObject.optional(),
+        id: z.string().optional(),
     };
+    static optionsSchema = z.strictObject(BasicCrawler.optionsShape);
     /**
      * All `BasicCrawler` parameters are passed via an options object.
      */
     constructor(options = {}) {
-        ow(options, 'BasicCrawlerOptions', ow.object.exactShape(BasicCrawler.optionsShape));
+        const parsedOptions = parseArgument(options, BasicCrawler.optionsSchema, 'BasicCrawlerOptions');
         const { 
         // oxlint-disable-next-line typescript/no-deprecated -- still accepted and folded into `requestManager` for back-compat
         requestList, 
         // oxlint-disable-next-line typescript/no-deprecated -- still accepted and folded into `requestManager` for back-compat
-        requestQueue, requestManager, maxRequestRetries = 3, sameDomainDelaySecs = 0, maxRequestsPerCrawl, maxCrawlDepth, taskLoopOptions = {}, concurrencySystem, keepAlive, sessionPool, proxyConfiguration, additionalHttpErrorStatusCodes = [], ignoreHttpErrorStatusCodes = [], 
+        requestQueue, requestManager, maxRequestRetries, sameDomainDelaySecs, maxRequestsPerCrawl, maxCrawlDepth, taskLoopOptions = {}, concurrencySystem, keepAlive, sessionPool, proxyConfiguration, additionalHttpErrorStatusCodes, ignoreHttpErrorStatusCodes, 
         // Service locator options
         configuration, storageBackend, eventManager, logger, 
         // AutoscaledPool shorthands
-        minConcurrency, maxConcurrency, maxRequestsPerMinute, blockedStatusCodes: blockedStatusCodesInput, retryOnBlocked = false, respectRobotsTxtFile = false, transactionalStorage, onSkippedRequest, requestHandler, requestHandlerTimeoutSecs, errorHandler, failedRequestHandler, statusMessageLoggingInterval = 10, statusMessageCallback, statistics, httpClient, id, } = options;
+        minConcurrency, maxConcurrency, maxRequestsPerMinute, blockedStatusCodes: blockedStatusCodesInput, retryOnBlocked, respectRobotsTxtFile, transactionalStorage, onSkippedRequest, requestHandler, requestHandlerTimeoutSecs, errorHandler, failedRequestHandler, statusMessageLoggingInterval, statusMessageCallback, statistics, httpClient, id, } = parsedOptions;
         // All concurrency configuration lives on the `ConcurrencySystem`, so the shortcuts have nowhere to go once
         // one is supplied - and silently dropping a `maxConcurrency` the user asked for is how crawls end up
         // hammering a site.
         if (concurrencySystem !== undefined &&
             (minConcurrency !== undefined || maxConcurrency !== undefined || maxRequestsPerMinute !== undefined)) {
-            throw new ArgumentError('The `minConcurrency`/`maxConcurrency`/`maxRequestsPerMinute` shortcuts cannot be combined with ' +
+            throw new Error('The `minConcurrency`/`maxConcurrency`/`maxRequestsPerMinute` shortcuts cannot be combined with ' +
                 '`concurrencySystem` - they configure the default `ConcurrencySystem` that a supplied one ' +
-                'replaces. Pass them to the `ConcurrencySystem` constructor instead.', this.constructor);
+                'replaces. Pass them to the `ConcurrencySystem` constructor instead.');
         }
         // Create per-crawler service locator if custom services were provided.
         // This wraps every method on the crawler instance so that calls to the global `serviceLocator`
@@ -277,14 +286,18 @@ export class BasicCrawler {
             eventManager ||
             logger ||
             (configuration !== undefined && configuration !== serviceLocator.getConfiguration())) {
-            const scopedServiceLocator = new ServiceLocator(configuration, eventManager, storageBackend, logger);
+            // Inherit the ambient locator's already-set services for anything not explicitly
+            // provided - e.g. only passing a `logger` must not detach the crawler from a globally
+            // configured storage backend.
+            const ambientServices = serviceLocator.getServicesIfSet();
+            const scopedServiceLocator = new ServiceLocator(configuration ?? ambientServices.configuration, eventManager ?? ambientServices.eventManager, storageBackend ?? ambientServices.storageBackend, logger ?? ambientServices.logger);
             serviceLocatorScope = bindMethodsToServiceLocator(scopedServiceLocator, this);
         }
         try {
             serviceLocatorScope.enterScope();
             this.#contextPipelineOptions = {
-                contextPipelineBuilder: options.contextPipelineBuilder,
-                extendContext: options.extendContext,
+                contextPipelineBuilder: parsedOptions.contextPipelineBuilder,
+                extendContext: parsedOptions.extendContext,
             };
             this.#log = serviceLocator.getLogger().child({ prefix: this.constructor.name });
             // Initialize the Configuration instance to avoid lazy loading in the components
@@ -1005,9 +1018,9 @@ export class BasicCrawler {
         const isAllowedBasedOnRobotsTxtFile = this.isAllowedBasedOnRobotsTxtFile.bind(this);
         const maxCrawlDepth = this.maxCrawlDepth;
         const validateRequestUserData = this.validateRequestUserData.bind(this);
-        ow(requests, ow.object
-            .is((value) => isIterable(value) || isAsyncIterable(value))
-            .message((value) => `Expected an iterable or async iterable, got ${getObjectType(value)}`));
+        if (!isIterable(requests) && !isAsyncIterable(requests)) {
+            throw new Error(`Expected an iterable or async iterable, got ${getObjectType(requests)}`);
+        }
         async function* filteredRequests() {
             for await (const request of requests) {
                 const url = typeof request === 'string' ? request : request.url;

package/internals/send-request.d.ts

@@ -1,5 +1,6 @@
 import type { Request as CrawleeRequest } from '@crawlee/core';
-import type { BaseHttpClient, HttpRequestOptions, ISession, SendRequestOptions } from '@crawlee/types';
+import type { BaseHttpClient } from '@crawlee/http-client';
+import type { HttpRequestOptions, ISession, SendRequestOptions } from '@crawlee/types';
 /**
  * Prepares a function to be used as the `sendRequest` context helper.
  *

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@crawlee/basic",
-    "version": "4.0.0-beta.120",
+    "version": "4.0.0-beta.122",
     "description": "The scalable web crawling and scraping library for JavaScript/Node.js. Enables development of data extraction and web automation jobs (not only) with headless Chrome and Puppeteer.",
     "engines": {
         "node": ">=22.0.0"
@@ -42,18 +42,18 @@
         "@apify/datastructures": "^2.0.0",
         "@apify/timeout": "^0.4.4",
         "@apify/utilities": "^2.15.5",
-        "@crawlee/core": "4.0.0-beta.120",
-        "@crawlee/http-client": "4.0.0-beta.120",
-        "@crawlee/types": "4.0.0-beta.120",
-        "@crawlee/utils": "4.0.0-beta.120",
+        "@crawlee/core": "4.0.0-beta.122",
+        "@crawlee/http-client": "4.0.0-beta.122",
+        "@crawlee/types": "4.0.0-beta.122",
+        "@crawlee/utils": "4.0.0-beta.122",
         "csv-stringify": "^6.5.2",
-        "ow": "^2.0.0",
         "tldts": "^7.0.6",
         "tslib": "^2.8.1",
-        "type-fest": "^4.41.0"
+        "type-fest": "^4.41.0",
+        "zod": "^4.4.3"
     },
     "optionalDependencies": {
-        "@crawlee/impit-client": "^4.0.0-beta.120"
+        "@crawlee/impit-client": "^4.0.0-beta.122"
     },
     "lerna": {
         "command": {
@@ -62,5 +62,5 @@
             }
         }
     },
-    "gitHead": "da5d427c4c1c9dcaea95b151e9c4c7310100885d"
+    "gitHead": "2c3e87fefdb9e1fca4c144f03167d8524cfdc2e5"
 }