@crawlee/puppeteer 4.0.0-beta.121 → 4.0.0-beta.122

package/internals/enqueue-links/click-elements.js

@@ -1,9 +1,25 @@
 import { URL } from 'node:url';
-import { applyRequestTransform, constructUrlPatternObjects, createRequestOptions, filterRequestOptionsByPatterns, Request, serviceLocator, } from '@crawlee/browser';
-import ow from 'ow';
+import { applyRequestTransform, constructUrlPatternObjects, createRequestOptions, filterRequestOptionsByPatterns, parseArgument, urlPatternSchema, Request, schemas, serviceLocator, } from '@crawlee/browser';
+import { z } from 'zod';
 import { addInterceptRequestHandler, removeInterceptRequestHandler } from '../utils/puppeteer_request_interception.js';
 const STARTING_Z_INDEX = 2147400000;
 const getLog = () => serviceLocator.getChildLog('Puppeteer Click Elements');
+const enqueueLinksByClickingElementsOptionsSchema = z.strictObject({
+    page: schemas.objectWithKeys(['goto', 'evaluate']),
+    requestManager: schemas.objectWithKeys(['fetchNextRequest', 'addRequestsBatched']),
+    selector: z.string(),
+    userData: schemas.anyObject.optional(),
+    clickOptions: schemas.anyObject.optional(),
+    include: schemas.arrayOf(urlPatternSchema, 'URL patterns').optional(),
+    exclude: schemas.arrayOf(urlPatternSchema, 'URL patterns').optional(),
+    transformRequestFunction: schemas.anyFunction.optional(),
+    waitForPageIdleSecs: schemas.anyNumber.default(1),
+    maxWaitForPageIdleSecs: schemas.anyNumber.default(5),
+    label: z.string().optional(),
+    forefront: z.boolean().optional(),
+    skipNavigation: z.boolean().optional(),
+    onSkippedRequest: schemas.anyFunction.optional(),
+});
 /**
  * The function finds elements matching a specific CSS selector in a Puppeteer page,
  * clicks all those elements using a mouse move and a left mouse button click and intercepts
@@ -45,24 +61,8 @@ const getLog = () => serviceLocator.getChildLog('Puppeteer Click Elements');
  * @returns Promise that resolves to {@link BatchAddRequestsResult} object.
  */
 export async function enqueueLinksByClickingElements(options) {
-    const urlPatternValidator = ow.any(ow.string, ow.regExp, ow.object.hasKeys('glob'), ow.object.hasKeys('regexp'));
-    ow(options, ow.object.exactShape({
-        page: ow.object.hasKeys('goto', 'evaluate'),
-        requestManager: ow.object.hasKeys('fetchNextRequest', 'addRequestsBatched'),
-        selector: ow.string,
-        userData: ow.optional.object,
-        clickOptions: ow.optional.object,
-        include: ow.optional.array.ofType(urlPatternValidator),
-        exclude: ow.optional.array.ofType(urlPatternValidator),
-        transformRequestFunction: ow.optional.function,
-        waitForPageIdleSecs: ow.optional.number,
-        maxWaitForPageIdleSecs: ow.optional.number,
-        label: ow.optional.string,
-        forefront: ow.optional.boolean,
-        skipNavigation: ow.optional.boolean,
-        onSkippedRequest: ow.optional.function,
-    }));
-    const { page, requestManager, selector, clickOptions, include, exclude, transformRequestFunction, waitForPageIdleSecs = 1, maxWaitForPageIdleSecs = 5, forefront, onSkippedRequest, } = options;
+    const parsedOptions = parseArgument(options, enqueueLinksByClickingElementsOptionsSchema, 'EnqueueLinksByClickingElementsOptions');
+    const { page, requestManager, selector, clickOptions, include, exclude, transformRequestFunction, waitForPageIdleSecs, maxWaitForPageIdleSecs, forefront, onSkippedRequest, } = parsedOptions;
     const waitForPageIdleMillis = waitForPageIdleSecs * 1000;
     const maxWaitForPageIdleMillis = maxWaitForPageIdleSecs * 1000;
     const urlExcludePatternObjects = exclude?.length ? constructUrlPatternObjects(exclude) : [];
@@ -74,7 +74,7 @@ export async function enqueueLinksByClickingElements(options) {
         maxWaitForPageIdleMillis,
         clickOptions,
     });
-    const requestOptions = createRequestOptions(interceptedRequests, options);
+    const requestOptions = createRequestOptions(interceptedRequests, parsedOptions);
     const skippedByFilters = [];
     let filteredOptions = filterRequestOptionsByPatterns(requestOptions, urlPatternObjects.length > 0 ? urlPatternObjects : undefined, urlExcludePatternObjects, undefined, (url) => skippedByFilters.push(url));
     if (onSkippedRequest && skippedByFilters.length > 0) {

package/internals/puppeteer-crawler.d.ts

@@ -4,6 +4,7 @@ import type { PuppeteerPlugin } from '@crawlee/browser-pool';
 import type { Dictionary } from '@crawlee/types';
 // @ts-ignore optional peer dependency or compatibility with es2022
 import type { HTTPResponse, LaunchOptions, Page } from 'puppeteer';
+import { z } from 'zod';
 import type { EnqueueLinksByClickingElementsOptions } from './enqueue-links/click-elements.js';
 import type { PuppeteerLaunchContext } from './puppeteer-launcher.js';
 import type { InterceptHandler } from './utils/puppeteer_request_interception.js';
@@ -123,99 +124,121 @@ export declare class PuppeteerCrawler<ContextExtension = Dictionary<never>, Exte
     browserPlugins: [PuppeteerPlugin];
 }, LaunchOptions, PuppeteerCrawlingContext, ContextExtension, ExtendedContext, Routes> {
     protected static optionsShape: {
-// @ts-ignore optional peer dependency or compatibility with es2022
-        browserPoolOptions: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        navigationTimeoutSecs: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        preNavigationHooks: import("ow").ArrayPredicate<unknown> & import("ow").BasePredicate<unknown[] | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        postNavigationHooks: import("ow").ArrayPredicate<unknown> & import("ow").BasePredicate<unknown[] | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        launchContext: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        headless: import("ow").AnyPredicate<string | boolean>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        browserPool: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        remoteBrowser: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        saveResponseCookies: import("ow").BooleanPredicate & import("ow").BasePredicate<boolean | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        proxyConfiguration: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        contextPipelineBuilder: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        extendContext: import("ow").Predicate<Function> & import("ow").BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestList: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestQueue: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestManager: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestHandler: import("ow").Predicate<Function> & import("ow").BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        requestHandlerTimeoutSecs: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        errorHandler: import("ow").Predicate<Function> & import("ow").BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        failedRequestHandler: import("ow").Predicate<Function> & import("ow").BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxRequestRetries: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        sameDomainDelaySecs: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxRequestsPerCrawl: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxCrawlDepth: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        taskLoopOptions: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        concurrencySystem: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        sessionPool: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        statusMessageLoggingInterval: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        statusMessageCallback: import("ow").Predicate<Function> & import("ow").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 & import("ow").BasePredicate<boolean | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        respectRobotsTxtFile: import("ow").AnyPredicate<boolean | object>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        transactionalStorage: import("ow").BasePredicate<boolean | Partial<import("@crawlee/browser").StorageWritePolicy> | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        onSkippedRequest: import("ow").Predicate<Function> & import("ow").BasePredicate<Function | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        httpClient: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        configuration: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        storageBackend: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        eventManager: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        logger: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        minConcurrency: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxConcurrency: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        maxRequestsPerMinute: import("ow").NumberPredicate & import("ow").BasePredicate<number | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        keepAlive: import("ow").BooleanPredicate & import("ow").BasePredicate<boolean | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        statistics: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        id: import("ow").StringPredicate & import("ow").BasePredicate<string | undefined>;
+        browserPoolOptions: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        navigationTimeoutSecs: z.ZodDefault<z.ZodCustom<number, number>>;
+        preNavigationHooks: z.ZodDefault<z.ZodCustom<unknown[], unknown[]>>;
+        postNavigationHooks: z.ZodDefault<z.ZodCustom<unknown[], unknown[]>>;
+        launchContext: z.ZodDefault<z.ZodCustom<Dictionary, Dictionary>>;
+        headless: z.ZodOptional<z.ZodUnion<readonly [z.ZodBoolean, z.ZodString]>>;
+        browserPool: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        remoteBrowser: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        saveResponseCookies: z.ZodDefault<z.ZodBoolean>;
+        proxyConfiguration: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        ignoreIframes: z.ZodDefault<z.ZodBoolean>;
+        ignoreShadowRoots: z.ZodDefault<z.ZodBoolean>;
+        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>>>;
+        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.$strict>]>>;
+        onSkippedRequest: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+// @ts-ignore optional peer dependency or compatibility with es2022
+        httpClient: z.ZodOptional<z.ZodCustom<import("@crawlee/http-client").BaseHttpClient, import("@crawlee/http-client").BaseHttpClient>>;
+// @ts-ignore optional peer dependency or compatibility with es2022
+        configuration: z.ZodOptional<z.ZodCustom<import("@crawlee/browser").Configuration, import("@crawlee/browser").Configuration>>;
+        storageBackend: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+// @ts-ignore optional peer dependency or compatibility with es2022
+        eventManager: z.ZodOptional<z.ZodCustom<import("@crawlee/browser").EventManager, import("@crawlee/browser").EventManager>>;
+        logger: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        minConcurrency: z.ZodOptional<z.ZodCustom<number, number>>;
+        maxConcurrency: z.ZodOptional<z.ZodCustom<number, number>>;
+        maxRequestsPerMinute: z.ZodOptional<z.ZodCustom<number, number>>;
+        keepAlive: z.ZodOptional<z.ZodBoolean>;
+        statistics: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        id: z.ZodOptional<z.ZodString>;
     };
+    protected static optionsSchema: z.ZodObject<{
+        browserPoolOptions: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        navigationTimeoutSecs: z.ZodDefault<z.ZodCustom<number, number>>;
+        preNavigationHooks: z.ZodDefault<z.ZodCustom<unknown[], unknown[]>>;
+        postNavigationHooks: z.ZodDefault<z.ZodCustom<unknown[], unknown[]>>;
+        launchContext: z.ZodDefault<z.ZodCustom<Dictionary, Dictionary>>;
+        headless: z.ZodOptional<z.ZodUnion<readonly [z.ZodBoolean, z.ZodString]>>;
+        browserPool: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        remoteBrowser: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        saveResponseCookies: z.ZodDefault<z.ZodBoolean>;
+        proxyConfiguration: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        ignoreIframes: z.ZodDefault<z.ZodBoolean>;
+        ignoreShadowRoots: z.ZodDefault<z.ZodBoolean>;
+        contextPipelineBuilder: z.ZodOptional<z.ZodCustom<Dictionary, Dictionary>>;
+        extendContext: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        requestList: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$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.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+        statusMessageLoggingInterval: z.ZodDefault<z.ZodCustom<number, number>>;
+        statusMessageCallback: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+        additionalHttpErrorStatusCodes: z.ZodDefault<z.ZodArray<z.ZodCustom<number, number>>>;
+        ignoreHttpErrorStatusCodes: z.ZodDefault<z.ZodArray<z.ZodCustom<number, number>>>;
+        blockedStatusCodes: z.ZodOptional<z.ZodArray<z.ZodCustom<number, number>>>;
+        retryOnBlocked: z.ZodDefault<z.ZodBoolean>;
+        respectRobotsTxtFile: z.ZodDefault<z.ZodUnion<readonly [z.ZodBoolean, z.ZodCustom<Dictionary, Dictionary>]>>;
+        transactionalStorage: z.ZodOptional<z.ZodUnion<readonly [z.ZodBoolean, z.ZodObject<{
+            requestQueue: z.ZodOptional<z.ZodEnum<{
+                deferred: "deferred";
+                writeThrough: "writeThrough";
+            }>>;
+        }, z.core.$strict>]>>;
+        onSkippedRequest: z.ZodOptional<z.ZodCustom<(...args: any[]) => unknown, (...args: any[]) => unknown>>;
+// @ts-ignore optional peer dependency or compatibility with es2022
+        httpClient: z.ZodOptional<z.ZodCustom<import("@crawlee/http-client").BaseHttpClient, import("@crawlee/http-client").BaseHttpClient>>;
+// @ts-ignore optional peer dependency or compatibility with es2022
+        configuration: z.ZodOptional<z.ZodCustom<import("@crawlee/browser").Configuration, import("@crawlee/browser").Configuration>>;
+        storageBackend: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$ZodTypeInternals<Dictionary<any>, unknown>>>;
+// @ts-ignore optional peer dependency or compatibility with es2022
+        eventManager: z.ZodOptional<z.ZodCustom<import("@crawlee/browser").EventManager, import("@crawlee/browser").EventManager>>;
+        logger: z.ZodOptional<z.ZodType<Dictionary<any>, unknown, z.core.$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 `PuppeteerCrawler` parameters are passed via an options object.
      */

package/internals/puppeteer-crawler.js

@@ -1,6 +1,6 @@
 import { BrowserCrawler, RequestState, Router } from '@crawlee/browser';
-import { serviceLocator } from '@crawlee/core';
-import ow from 'ow';
+import { parseArgument, schemas, serviceLocator } from '@crawlee/core';
+import { z } from 'zod';
 import { PuppeteerLauncher } from './puppeteer-launcher.js';
 import { gotoExtended, puppeteerUtils } from './utils/puppeteer_utils.js';
 /**
@@ -71,16 +71,17 @@ import { gotoExtended, puppeteerUtils } from './utils/puppeteer_utils.js';
 export class PuppeteerCrawler extends BrowserCrawler {
     static optionsShape = {
         ...BrowserCrawler.optionsShape,
-        browserPoolOptions: ow.optional.object,
+        browserPoolOptions: schemas.anyObject.optional(),
     };
+    static optionsSchema = z.strictObject(PuppeteerCrawler.optionsShape);
     /**
      * All `PuppeteerCrawler` parameters are passed via an options object.
      */
     constructor(options = {}) {
-        ow(options, 'PuppeteerCrawlerOptions', ow.object.exactShape(PuppeteerCrawler.optionsShape));
-        const { launchContext = {}, headless, proxyConfiguration, contextPipelineBuilder, ...browserCrawlerOptions } = options;
+        const parsedOptions = parseArgument(options, PuppeteerCrawler.optionsSchema, 'PuppeteerCrawlerOptions');
+        const { launchContext, headless, proxyConfiguration, contextPipelineBuilder, ...browserCrawlerOptions } = parsedOptions;
         const browserPoolOptions = {
-            ...options.browserPoolOptions,
+            ...parsedOptions.browserPoolOptions,
         };
         if (launchContext.proxyUrl) {
             throw new Error('PuppeteerCrawlerOptions.launchContext.proxyUrl is not allowed in PuppeteerCrawler.' +
@@ -95,7 +96,7 @@ export class PuppeteerCrawler extends BrowserCrawler {
             launchContext.launchOptions ??= {};
             launchContext.launchOptions.headless = headless;
         }
-        const puppeteerLauncher = new PuppeteerLauncher(launchContext, options.configuration);
+        const puppeteerLauncher = new PuppeteerLauncher(launchContext, parsedOptions.configuration);
         browserPoolOptions.browserPlugins = [puppeteerLauncher.createBrowserPlugin()];
         super({
             ...browserCrawlerOptions,

package/internals/puppeteer-launcher.d.ts

@@ -3,6 +3,7 @@ import { BrowserLauncher, Configuration } from '@crawlee/browser';
 import { PuppeteerPlugin } from '@crawlee/browser-pool';
 // @ts-ignore optional peer dependency or compatibility with es2022
 import type { Browser } from 'puppeteer';
+import { z } from 'zod';
 /**
  * Apify extends the launch options of Puppeteer.
  * You can use any of the Puppeteer compatible
@@ -69,24 +70,30 @@ export declare class PuppeteerLauncher extends BrowserLauncher<PuppeteerPlugin,
     readonly configuration: Configuration;
     protected static optionsShape: {
 // @ts-ignore optional peer dependency or compatibility with es2022
-        launcher: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
+        launcher: z.ZodOptional<z.ZodCustom<import("@crawlee/types").Dictionary, import("@crawlee/types").Dictionary>>;
+        proxyUrl: z.ZodOptional<z.ZodURL>;
+        useChrome: z.ZodOptional<z.ZodBoolean>;
+        useIncognitoPages: z.ZodOptional<z.ZodBoolean>;
+        browserPerProxy: z.ZodOptional<z.ZodBoolean>;
+        ignoreProxyCertificate: z.ZodOptional<z.ZodBoolean>;
+        userDataDir: z.ZodOptional<z.ZodString>;
 // @ts-ignore optional peer dependency or compatibility with es2022
-        proxyUrl: import("ow").StringPredicate & import("ow").BasePredicate<string | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        useChrome: import("ow").BooleanPredicate & import("ow").BasePredicate<boolean | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        useIncognitoPages: import("ow").BooleanPredicate & import("ow").BasePredicate<boolean | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        browserPerProxy: import("ow").BooleanPredicate & import("ow").BasePredicate<boolean | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        ignoreProxyCertificate: import("ow").BooleanPredicate & import("ow").BasePredicate<boolean | undefined>;
-// @ts-ignore optional peer dependency or compatibility with es2022
-        userDataDir: import("ow").StringPredicate & import("ow").BasePredicate<string | undefined>;
+        launchOptions: z.ZodOptional<z.ZodCustom<import("@crawlee/types").Dictionary, import("@crawlee/types").Dictionary>>;
+        userAgent: z.ZodOptional<z.ZodString>;
+    };
+    protected static optionsSchema: z.ZodObject<{
 // @ts-ignore optional peer dependency or compatibility with es2022
-        launchOptions: import("ow").ObjectPredicate<object> & import("ow").BasePredicate<object | undefined>;
+        launcher: z.ZodOptional<z.ZodCustom<import("@crawlee/types").Dictionary, import("@crawlee/types").Dictionary>>;
+        proxyUrl: z.ZodOptional<z.ZodURL>;
+        useChrome: z.ZodOptional<z.ZodBoolean>;
+        useIncognitoPages: z.ZodOptional<z.ZodBoolean>;
+        browserPerProxy: z.ZodOptional<z.ZodBoolean>;
+        ignoreProxyCertificate: z.ZodOptional<z.ZodBoolean>;
+        userDataDir: z.ZodOptional<z.ZodString>;
 // @ts-ignore optional peer dependency or compatibility with es2022
-        userAgent: import("ow").StringPredicate & import("ow").BasePredicate<string | undefined>;
-    };
+        launchOptions: z.ZodOptional<z.ZodCustom<import("@crawlee/types").Dictionary, import("@crawlee/types").Dictionary>>;
+        userAgent: z.ZodOptional<z.ZodString>;
+    }, z.core.$strict>;
     /**
      * All `PuppeteerLauncher` parameters are passed via an launchContext object.
      */

package/internals/puppeteer-launcher.js

@@ -1,6 +1,6 @@
-import { BrowserLauncher, Configuration } from '@crawlee/browser';
+import { BrowserLauncher, Configuration, parseArgument, schemas } from '@crawlee/browser';
 import { PuppeteerPlugin } from '@crawlee/browser-pool';
-import ow from 'ow';
+import { z } from 'zod';
 /**
  * `PuppeteerLauncher` is based on the `BrowserLauncher`. It launches `puppeteer` browser instance.
  * @ignore
@@ -9,14 +9,14 @@ export class PuppeteerLauncher extends BrowserLauncher {
     configuration;
     static optionsShape = {
         ...BrowserLauncher.optionsShape,
-        launcher: ow.optional.object,
+        launcher: schemas.anyObject.optional(),
     };
+    static optionsSchema = z.strictObject(PuppeteerLauncher.optionsShape);
     /**
      * All `PuppeteerLauncher` parameters are passed via an launchContext object.
      */
     constructor(launchContext = {}, configuration = Configuration.getGlobalConfiguration()) {
-        ow(launchContext, 'PuppeteerLauncher', ow.object.exactShape(PuppeteerLauncher.optionsShape));
-        const { launcher = BrowserLauncher.requireLauncherOrThrow('puppeteer', 'apify/actor-node-puppeteer-chrome'), ...browserLauncherOptions } = launchContext;
+        const { launcher = BrowserLauncher.requireLauncherOrThrow('puppeteer', 'apify/actor-node-puppeteer-chrome'), ...browserLauncherOptions } = parseArgument(launchContext, PuppeteerLauncher.optionsSchema, 'PuppeteerLaunchContext');
         super({
             ...browserLauncherOptions,
             launcher,

package/internals/utils/puppeteer_request_interception.js

@@ -1,6 +1,6 @@
 import { EventEmitter } from 'node:events';
-import { serviceLocator } from '@crawlee/browser';
-import ow from 'ow';
+import { parseArgument, schemas, serviceLocator } from '@crawlee/browser';
+const pageSchema = schemas.objectWithKeys(['goto', 'evaluate']);
 // We use weak maps here so that the content gets discarded after page gets closed.
 const pageInterceptRequestHandlersMap = new WeakMap(); // Maps page to an array of request interception handlers.
 const pageInterceptRequestMasterHandlerMap = new WeakMap(); // Maps page to master request interception handler.
@@ -137,8 +137,8 @@ async function handleRequest(request, interceptRequestHandlers) {
  * @param handler Request interception handler.
  */
 export async function addInterceptRequestHandler(page, handler) {
-    ow(page, ow.object.hasKeys('goto', 'evaluate'));
-    ow(handler, ow.function);
+    parseArgument(page, pageSchema);
+    parseArgument(handler, schemas.anyFunction);
     if (!pageInterceptRequestHandlersMap.has(page)) {
         pageInterceptRequestHandlersMap.set(page, []);
     }
@@ -174,8 +174,8 @@ export async function addInterceptRequestHandler(page, handler) {
  * @param handler Request interception handler.
  */
 export async function removeInterceptRequestHandler(page, handler) {
-    ow(page, ow.object.hasKeys('goto', 'evaluate'));
-    ow(handler, ow.function);
+    parseArgument(page, pageSchema);
+    parseArgument(handler, schemas.anyFunction);
     const handlersArray = pageInterceptRequestHandlersMap.get(page).filter((item) => item !== handler);
     pageInterceptRequestHandlersMap.set(page, handlersArray);
     if (handlersArray.length === 0) {

package/internals/utils/puppeteer_utils.js

@@ -20,9 +20,9 @@
 import { readFile } from 'node:fs/promises';
 import { createRequire } from 'node:module';
 import vm from 'node:vm';
-import { Configuration, KeyValueStore, serviceLocator, validators } from '@crawlee/browser';
+import { Configuration, KeyValueStore, parseArgument, schemas, serviceLocator, validators } from '@crawlee/browser';
 import { expandShadowRoots, sleep } from '@crawlee/utils';
-import ow from 'ow';
+import { z } from 'zod';
 import { LruCache } from '@apify/datastructures';
 import { enqueueLinksByClickingElements } from '../enqueue-links/click-elements.js';
 import { addInterceptRequestHandler, removeInterceptRequestHandler } from './puppeteer_request_interception.js';
@@ -30,6 +30,37 @@ const require = createRequire(import.meta.url);
 const jqueryPath = require.resolve('jquery');
 const MAX_INJECT_FILE_CACHE_SIZE = 10;
 const DEFAULT_BLOCK_REQUEST_URL_PATTERNS = ['.css', '.jpg', '.jpeg', '.png', '.svg', '.gif', '.woff', '.pdf', '.zip'];
+const filePathSchema = z.string();
+const injectFileOptionsSchema = z.strictObject({
+    surviveNavigations: z.boolean().optional(),
+});
+const responseUrlRulesSchema = schemas.arrayOf(z.union([z.string(), z.instanceof(RegExp)]), 'strings or RegExps');
+const gotoExtendedRequestSchema = z.looseObject({
+    url: z.url(),
+    method: z.string().optional(),
+    headers: schemas.anyObject.optional(),
+    payload: z.union([z.string(), z.instanceof(Uint8Array)]).optional(),
+});
+const blockRequestsOptionsSchema = z.strictObject({
+    urlPatterns: schemas.arrayOf(z.string(), 'strings').default(DEFAULT_BLOCK_REQUEST_URL_PATTERNS),
+    extraUrlPatterns: schemas.arrayOf(z.string(), 'strings').default(() => []),
+});
+const infiniteScrollOptionsSchema = z.strictObject({
+    timeoutSecs: schemas.anyNumber.default(0),
+    maxScrollHeight: schemas.anyNumber.default(0),
+    waitForSecs: schemas.anyNumber.default(4),
+    scrollDownAndUp: z.boolean().default(false),
+    buttonSelector: z.string().optional(),
+    stopScrollCallback: schemas.anyFunction.optional(),
+});
+const saveSnapshotOptionsSchema = z.strictObject({
+    key: z.string().min(1).default('SNAPSHOT'),
+    screenshotQuality: schemas.anyNumber.default(50),
+    saveScreenshot: z.boolean().default(true),
+    saveHtml: z.boolean().default(true),
+    keyValueStoreName: z.string().optional(),
+    configuration: schemas.anyObject.optional(),
+});
 const getLog = () => serviceLocator.getChildLog('Puppeteer Utils');
 /**
  * Cache contents of previously injected files to limit file system access.
@@ -47,18 +78,16 @@ const injectedFilesCache = new LruCache({ maxLength: MAX_INJECT_FILE_CACHE_SIZE
  * @param [options]
  */
 export async function injectFile(page, filePath, options = {}) {
-    ow(page, ow.object.validate(validators.browserPage));
-    ow(filePath, ow.string);
-    ow(options, ow.object.exactShape({
-        surviveNavigations: ow.optional.boolean,
-    }));
+    parseArgument(page, validators.browserPage);
+    parseArgument(filePath, filePathSchema);
+    const { surviveNavigations } = parseArgument(options, injectFileOptionsSchema);
     let contents = injectedFilesCache.get(filePath);
     if (!contents) {
         contents = await readFile(filePath, 'utf8');
         injectedFilesCache.add(filePath, contents);
     }
     const evalP = page.evaluate(contents);
-    if (options.surviveNavigations) {
+    if (surviveNavigations) {
         page.on('framenavigated', async () => page
             .evaluate(contents)
             .catch((error) => getLog().warning('An error occurred during the script injection!', { error })));
@@ -92,7 +121,7 @@ export async function injectFile(page, filePath, options = {}) {
  * @param [options.surviveNavigations] Opt-out option to disable the JQuery reinjection after navigation.
  */
 export async function injectJQuery(page, options) {
-    ow(page, ow.object.validate(validators.browserPage));
+    parseArgument(page, validators.browserPage);
     return injectFile(page, jqueryPath, { surviveNavigations: options?.surviveNavigations ?? true });
 }
 /**
@@ -108,7 +137,7 @@ export async function injectJQuery(page, options) {
  * @param ignoreShadowRoots
  */
 export async function parseWithCheerio(page, ignoreShadowRoots = false, ignoreIframes = false) {
-    ow(page, ow.object.validate(validators.browserPage));
+    parseArgument(page, validators.browserPage);
     if (page.frames().length > 1 && !ignoreIframes) {
         const frames = await page.$$('iframe');
         await Promise.all(frames.map(async (frame) => {
@@ -187,12 +216,8 @@ export async function parseWithCheerio(page, ignoreShadowRoots = false, ignoreIf
  * @param [options]
  */
 export async function blockRequests(page, options = {}) {
-    ow(page, ow.object.validate(validators.browserPage));
-    ow(options, ow.object.exactShape({
-        urlPatterns: ow.optional.array.ofType(ow.string),
-        extraUrlPatterns: ow.optional.array.ofType(ow.string),
-    }));
-    const { urlPatterns = DEFAULT_BLOCK_REQUEST_URL_PATTERNS, extraUrlPatterns = [] } = options;
+    parseArgument(page, validators.browserPage);
+    const { urlPatterns, extraUrlPatterns } = parseArgument(options, blockRequestsOptionsSchema);
     const patternsToBlock = [...urlPatterns, ...extraUrlPatterns];
     // We use CDP commands instead of request interception as the latter disables caching, which is not ideal
     await sendCDPCommand(page, 'Network.setBlockedURLs', { urls: patternsToBlock });
@@ -249,9 +274,9 @@ export const blockResources = async (page, resourceTypes = ['stylesheet', 'font'
  * @deprecated
  */
 export async function cacheResponses(page, cache, responseUrlRules) {
-    ow(page, ow.object.validate(validators.browserPage));
-    ow(cache, ow.object);
-    ow(responseUrlRules, ow.array.ofType(ow.any(ow.string, ow.regExp)));
+    parseArgument(page, validators.browserPage);
+    parseArgument(cache, schemas.anyObject);
+    parseArgument(responseUrlRules, responseUrlRulesSchema);
     serviceLocator
         .getLogger()
         .deprecated('utils.puppeteer.cacheResponses() has a high impact on performance ' +
@@ -344,14 +369,9 @@ export function compileScript(scriptString, context = Object.create(null)) {
  * @param [gotoOptions] Custom options for `page.goto()`.
  */
 export async function gotoExtended(page, request, gotoOptions = {}) {
-    ow(page, ow.object.validate(validators.browserPage));
-    ow(request, ow.object.partialShape({
-        url: ow.string.url,
-        method: ow.optional.string,
-        headers: ow.optional.object,
-        payload: ow.optional.any(ow.string, ow.uint8Array),
-    }));
-    ow(gotoOptions, ow.object);
+    parseArgument(page, validators.browserPage);
+    parseArgument(request, gotoExtendedRequestSchema);
+    parseArgument(gotoOptions, schemas.anyObject);
     gotoOptions = { ...gotoOptions };
     if (gotoOptions.waitUntil === 'networkidle') {
         gotoOptions.waitUntil = 'networkidle0';
@@ -407,16 +427,8 @@ export async function gotoExtended(page, request, gotoOptions = {}) {
  * @param [options]
  */
 export async function infiniteScroll(page, options = {}) {
-    ow(page, ow.object.validate(validators.browserPage));
-    ow(options, ow.object.exactShape({
-        timeoutSecs: ow.optional.number,
-        maxScrollHeight: ow.optional.number,
-        waitForSecs: ow.optional.number,
-        scrollDownAndUp: ow.optional.boolean,
-        buttonSelector: ow.optional.string,
-        stopScrollCallback: ow.optional.function,
-    }));
-    const { timeoutSecs = 0, maxScrollHeight = 0, waitForSecs = 4, scrollDownAndUp = false, buttonSelector, stopScrollCallback, } = options;
+    parseArgument(page, validators.browserPage);
+    const { timeoutSecs, maxScrollHeight, waitForSecs, scrollDownAndUp, buttonSelector, stopScrollCallback } = parseArgument(options, infiniteScrollOptionsSchema);
     let finished;
     const startTime = Date.now();
     const CHECK_INTERVAL_MILLIS = 1000;
@@ -508,16 +520,8 @@ export async function infiniteScroll(page, options = {}) {
  * @param [options]
  */
 export async function saveSnapshot(page, options = {}) {
-    ow(page, ow.object.validate(validators.browserPage));
-    ow(options, ow.object.exactShape({
-        key: ow.optional.string.nonEmpty,
-        screenshotQuality: ow.optional.number,
-        saveScreenshot: ow.optional.boolean,
-        saveHtml: ow.optional.boolean,
-        keyValueStoreName: ow.optional.string,
-        configuration: ow.optional.object,
-    }));
-    const { key = 'SNAPSHOT', screenshotQuality = 50, saveScreenshot = true, saveHtml = true, keyValueStoreName, configuration, } = options;
+    parseArgument(page, validators.browserPage);
+    const { key, screenshotQuality, saveScreenshot, saveHtml, keyValueStoreName, configuration } = parseArgument(options, saveSnapshotOptionsSchema);
     try {
         const store = await KeyValueStore.open(keyValueStoreName ? { name: keyValueStoreName } : null, {
             configuration: configuration ?? Configuration.getGlobalConfiguration(),
@@ -564,7 +568,7 @@ ${error.message}
     return idcacPlaywright;
 }
 export async function closeCookieModals(page) {
-    ow(page, ow.object.validate(validators.browserPage));
+    parseArgument(page, validators.browserPage);
     const idcac = await getIdcacPlaywright();
     if (idcac?.getInjectableScript()) {
         await page.evaluate(idcac.getInjectableScript());

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@crawlee/puppeteer",
-    "version": "4.0.0-beta.121",
+    "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"
@@ -48,16 +48,16 @@
     },
     "dependencies": {
         "@apify/datastructures": "^2.0.3",
-        "@crawlee/browser": "4.0.0-beta.121",
-        "@crawlee/browser-pool": "4.0.0-beta.121",
-        "@crawlee/core": "4.0.0-beta.121",
-        "@crawlee/types": "4.0.0-beta.121",
-        "@crawlee/utils": "4.0.0-beta.121",
+        "@crawlee/browser": "4.0.0-beta.122",
+        "@crawlee/browser-pool": "4.0.0-beta.122",
+        "@crawlee/core": "4.0.0-beta.122",
+        "@crawlee/types": "4.0.0-beta.122",
+        "@crawlee/utils": "4.0.0-beta.122",
         "cheerio": "^1.0.0",
         "devtools-protocol": "*",
         "jquery": "^3.7.1",
-        "ow": "^2.0.0",
-        "tslib": "^2.8.1"
+        "tslib": "^2.8.1",
+        "zod": "^4.4.3"
     },
     "peerDependencies": {
         "idcac-playwright": "^0.2.0",
@@ -78,5 +78,5 @@
             }
         }
     },
-    "gitHead": "5027317de626f5ba6de5047ae9341a898258cc5a"
+    "gitHead": "2c3e87fefdb9e1fca4c144f03167d8524cfdc2e5"
 }