scrapely 1.2.26 → 1.2.27

package/dist/actor.d.ts

@@ -11,7 +11,7 @@ export declare enum ACTOR_PERMISSION_LEVEL {
     /** User has full administrative access */
     ADMIN = "ADMIN"
 }
-import { ApifyClient } from 'scrapely-client';
+import type { ApifyClient } from 'scrapely-client';
 import type { ActorRun, Webhook } from 'scrapely-client';
 import type { ActorCallOptions, ActorStartOptions, RunAbortOptions, TaskCallOptions } from 'scrapely-client';
 import { type EventTypeName } from '@crawlee/core';
@@ -333,7 +333,8 @@ export declare class Actor {
      * succeeds.
      */
     private _lifecycleStarted;
-    private readonly _apifyClient;
+    /** Created on first use: scrapely-client is only loaded when something needs the API client. */
+    private _apifyClient;
     private readonly chargingManager;
     /**
      * Creates an Actor instance (Apify SDK-compatible alternative to static helpers).

package/dist/actor.js

@@ -31,6 +31,7 @@ const secrets_js_1 = require("./crypto/secrets.js");
 const charging_js_1 = require("./charging.js");
 const charging_js_2 = require("./charging.js");
 const charging_client_js_1 = require("./charging_client.js");
+const client_loader_js_1 = require("./client-loader.js");
 const proxy_configuration_js_1 = require("./proxy-configuration.js");
 const browser_configuration_js_1 = require("./browser-configuration.js");
 const utils_js_1 = require("./utils.js");
@@ -111,6 +112,7 @@ class Actor {
      * succeeds.
      */
     _lifecycleStarted = false;
+    /** Created on first use: scrapely-client is only loaded when something needs the API client. */
     _apifyClient;
     chargingManager;
     /**
@@ -119,12 +121,12 @@ class Actor {
     constructor(options = {}) {
         this.config =
             Object.keys(options).length === 0 ? configuration_js_1.Configuration.getGlobalConfig() : new configuration_js_1.Configuration(options);
-        this._apifyClient = this.newClient();
-        this.chargingManager = new charging_js_1.ChargingManager(this.config, this._apifyClient);
+        this.chargingManager = new charging_js_1.ChargingManager(this.config, () => this.apifyClient);
         Actor._instance ??= this;
     }
     /** Default ApifyClient instance for this Actor (Apify-compatible instance accessor). */
     get apifyClient() {
+        this._apifyClient ??= this.newClient();
         return this._apifyClient;
     }
     _ensureInit(methodName) {
@@ -278,8 +280,12 @@ class Actor {
                 };
                 this.on('migrating', gracefulShutdownHandlers.migrating);
             }
-            apifyClientInstance = this._apifyClient;
-            apifyClientAssigned = true;
+            // Same as before the client became lazy: this Actor's client (once
+            // created) is the process default returned by newClient().
+            if (this._apifyClient) {
+                apifyClientInstance = this._apifyClient;
+                apifyClientAssigned = true;
+            }
             await this.chargingManager.init();
             // Storage clients refuse default-dataset writes past this run's limit.
             (0, charging_js_1.setActiveChargingManager)(this.chargingManager);
@@ -337,7 +343,10 @@ class Actor {
                 }
                 (0, events_client_js_1.disconnectEvents)();
                 (0, patch_storage_js_1.disconnectStorage)();
-                if (apifyClientAssigned) {
+                // Also when the client was only created during this init (lazily,
+                // e.g. by the pricing lookup): never leave a failed init's client
+                // as the process default.
+                if (apifyClientAssigned || apifyClientInstance !== prevApifyClientInstance) {
                     apifyClientInstance = prevApifyClientInstance;
                 }
             }
@@ -424,7 +433,7 @@ class Actor {
         if (metamorphCount) {
             const count = parseInt(metamorphCount, 10);
             if (count > 0) {
-                const { KeyValueStore } = await import('crawlee');
+                const { KeyValueStore } = await import('@crawlee/core');
                 const store = await KeyValueStore.open();
                 const metamorphKey = `INPUT-METAMORPH-${count}`;
                 const metamorphInput = await store.getValue(metamorphKey);
@@ -455,7 +464,7 @@ class Actor {
             }
         }
         else {
-            const { KeyValueStore } = await import('crawlee');
+            const { KeyValueStore } = await import('@crawlee/core');
             const store = await KeyValueStore.open();
             const inputKey = Actor._actorConfig().get('inputKey', 'INPUT');
             input = await store.getValue(inputKey);
@@ -521,7 +530,7 @@ class Actor {
         Actor._ensureActorInit('pushData');
         if (eventNameOrOptions !== undefined && typeof eventNameOrOptions === 'object') {
             const opts = eventNameOrOptions;
-            const { Dataset } = await import('crawlee');
+            const { Dataset } = await import('@crawlee/core');
             const dataset = opts.dataset === undefined
                 ? await Dataset.open()
                 : typeof opts.dataset === 'string' ? await Dataset.open(opts.dataset) : opts.dataset;
@@ -564,7 +573,7 @@ class Actor {
             eventName === 'scrapely-default-dataset-item') {
             throw new Error(`Cannot charge for synthetic event '${eventName}' manually`);
         }
-        const { Dataset } = await import('crawlee');
+        const { Dataset } = await import('@crawlee/core');
         const dataset = await Dataset.open();
         if (eventName === undefined && !this.usesPushDataInterception(dataset)) {
             // Charge-limit safety net (also for local storage): this push is not
@@ -584,11 +593,16 @@ class Actor {
         return this.pushDataWithExplicitCharging(dataset, item, eventName);
     }
     usesPushDataInterception(dataset) {
-        return Boolean(dataset.client[charging_client_js_1.USES_PUSH_DATA_INTERCEPTION]);
+        // Only a scrapely-client dataset client carries the marker, so when the
+        // SDK's scrapely-client was never loaded no client can be intercepting.
+        const client = (0, client_loader_js_1.loadedScrapelyClient)();
+        if (!client)
+            return false;
+        return Boolean(dataset.client[client.USES_PUSH_DATA_INTERCEPTION]);
     }
     async pushDataViaInterceptedClient(dataset, item, eventName) {
         const context = { eventName };
-        await charging_client_js_1.pushDataChargingContext.run(context, async () => {
+        await (0, client_loader_js_1.loadedScrapelyClient)().pushDataChargingContext.run(context, async () => {
             await dataset.pushData(item);
         });
         return (context.chargeResult ?? {
@@ -613,7 +627,7 @@ class Actor {
     }
     static async getValue(key) {
         Actor._ensureActorInit('getValue');
-        const { KeyValueStore } = await import('crawlee');
+        const { KeyValueStore } = await import('@crawlee/core');
         const store = await KeyValueStore.open();
         return store.getValue(key);
     }
@@ -638,7 +652,7 @@ class Actor {
      */
     static async setValue(key, value, options) {
         Actor._ensureActorInit('setValue');
-        const { KeyValueStore } = await import('crawlee');
+        const { KeyValueStore } = await import('@crawlee/core');
         const store = await KeyValueStore.open();
         await store.setValue(key, value, options);
     }
@@ -711,19 +725,19 @@ class Actor {
         return storageOptions;
     }
     static async _openDataset(actorConfig, nameOrIdOrOptions, openOptions) {
-        const { Dataset } = await import('crawlee');
+        const { Dataset } = await import('@crawlee/core');
         const resolved = Actor._resolveStorageIdentifier(nameOrIdOrOptions, 'datasets', actorConfig);
         const storageOpts = Actor._buildStorageOpenOptions(actorConfig, openOptions);
         return Dataset.open(resolved, storageOpts);
     }
     static async _openKeyValueStore(actorConfig, nameOrIdOrOptions, openOptions) {
-        const { KeyValueStore } = await import('crawlee');
+        const { KeyValueStore } = await import('@crawlee/core');
         const resolved = Actor._resolveStorageIdentifier(nameOrIdOrOptions, 'keyValueStores', actorConfig);
         const storageOpts = Actor._buildStorageOpenOptions(actorConfig, openOptions);
         return KeyValueStore.open(resolved, storageOpts);
     }
     static async _openRequestQueue(actorConfig, nameOrIdOrOptions, openOptions) {
-        const { RequestQueue } = await import('crawlee');
+        const { RequestQueue } = await import('@crawlee/core');
         const resolved = Actor._resolveStorageIdentifier(nameOrIdOrOptions, 'requestQueues', actorConfig);
         const storageOpts = Actor._buildStorageOpenOptions(actorConfig, openOptions);
         return RequestQueue.open(resolved, storageOpts);
@@ -1361,7 +1375,7 @@ class Actor {
             stateKey = 'APIFY_GLOBAL_STATE';
             initialState = defaultValue;
         }
-        const { KeyValueStore } = await import('crawlee');
+        const { KeyValueStore } = await import('@crawlee/core');
         const kvStore = await KeyValueStore.open(options?.keyValueStoreName, {
             config: options?.config ?? Actor._actorConfig(),
         });

package/dist/charging.d.ts

@@ -70,7 +70,6 @@ export declare function mergeChargeResults(a: ChargeResult, b: ChargeResult): Ch
  */
 export declare class ChargingManager {
     private readonly configuration;
-    private readonly apifyClient;
     /** Set during actor shutdown; rejects new charge operations after the final flush. */
     private _closed;
     /** Sticky: the platform refused or capped a charge (HTTP 402 / fewer accepted) or the local gate trimmed one. */
@@ -177,7 +176,13 @@ export declare class ChargingManager {
     private readonly postAttempts;
     private readonly pendingMaxEntries;
     private readonly idempotencySafeAgeMs;
-    constructor(configuration: Configuration, apifyClient: ApifyClient);
+    private readonly apifyClientSource;
+    /**
+     * @param apifyClient - the API client, or a function returning it (called
+     *   on first use, so runs that never need the API never load the client).
+     */
+    constructor(configuration: Configuration, apifyClient: ApifyClient | (() => ApifyClient));
+    private get apifyClient();
     /** @internal Serialize allowance calculation, side effects, and local reservation. */
     _withOperationLock<T>(operation: () => Promise<T>): Promise<T>;
     private get isPayPerEvent();

package/dist/charging.js

@@ -149,7 +149,6 @@ function mergeChargeResults(a, b) {
  */
 class ChargingManager {
     configuration;
-    apifyClient;
     /** Set during actor shutdown; rejects new charge operations after the final flush. */
     _closed = false;
     /** Sticky: the platform refused or capped a charge (HTTP 402 / fewer accepted) or the local gate trimmed one. */
@@ -369,13 +368,22 @@ class ChargingManager {
     postAttempts = POST_ATTEMPTS;
     pendingMaxEntries = PENDING_MAX_ENTRIES;
     idempotencySafeAgeMs = exports.CHARGE_IDEMPOTENCY_SAFE_AGE_MS;
+    apifyClientSource;
+    /**
+     * @param apifyClient - the API client, or a function returning it (called
+     *   on first use, so runs that never need the API never load the client).
+     */
     constructor(configuration, apifyClient) {
         this.configuration = configuration;
-        this.apifyClient = apifyClient;
+        this.apifyClientSource = apifyClient;
         this.maxTotalChargeUsd = configuration.get('maxTotalChargeUsd') || Infinity;
         this.isAtHome = !!configuration.get('isAtHome');
         this.actorRunId = configuration.get('actorRunId') || '';
     }
+    get apifyClient() {
+        const source = this.apifyClientSource;
+        return typeof source === 'function' ? source() : source;
+    }
     /** @internal Serialize allowance calculation, side effects, and local reservation. */
     async _withOperationLock(operation) {
         const previous = this.operationTail;
@@ -402,9 +410,13 @@ class ChargingManager {
     async fetchPricingInfo() {
         const actorPricingInfo = this.configuration.get('actorPricingInfo');
         const chargedEventCounts = this.configuration.get('chargedEventCounts');
-        if (actorPricingInfo && chargedEventCounts) {
+        // The platform worker always injects the charged event counts, and the
+        // pricing info only when the actor has pricing (both from the same
+        // actor.pricingInfos[0] the run object reports). Counts without pricing
+        // therefore mean "no pricing": no API round trip at init.
+        if (chargedEventCounts) {
             return {
-                pricingInfo: JSON.parse(actorPricingInfo),
+                pricingInfo: actorPricingInfo ? JSON.parse(actorPricingInfo) : undefined,
                 chargedEventCounts: JSON.parse(chargedEventCounts),
                 maxTotalChargeUsd: this.configuration.get('maxTotalChargeUsd') || Infinity,
             };

package/dist/charging_client.d.ts

@@ -1,8 +1,6 @@
-import type { ApifyClientOptions, PpeAwarePushDataContext } from 'scrapely-client';
-import { ApifyClient, USES_PUSH_DATA_INTERCEPTION, pushDataChargingContext } from 'scrapely-client';
+import type { ApifyClient, ApifyClientOptions, PpeAwarePushDataContext } from 'scrapely-client';
 import { type ChargingManager } from './charging.js';
 import type { Configuration } from './configuration.js';
-export { USES_PUSH_DATA_INTERCEPTION, pushDataChargingContext };
 export type { PpeAwarePushDataContext };
 /**
  * Returns a `scrapely-client` `ApifyClient` configured with the SDK's charging

package/dist/charging_client.js

@@ -1,11 +1,8 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.pushDataChargingContext = exports.USES_PUSH_DATA_INTERCEPTION = void 0;
 exports.createChargingClient = createChargingClient;
-const scrapely_client_1 = require("scrapely-client");
-Object.defineProperty(exports, "USES_PUSH_DATA_INTERCEPTION", { enumerable: true, get: function () { return scrapely_client_1.USES_PUSH_DATA_INTERCEPTION; } });
-Object.defineProperty(exports, "pushDataChargingContext", { enumerable: true, get: function () { return scrapely_client_1.pushDataChargingContext; } });
 const charging_js_1 = require("./charging.js");
+const client_loader_js_1 = require("./client-loader.js");
 /**
  * Adapts the SDK's {@link ChargingManager} to `scrapely-client`'s
  * {@link ChargingHooks} protocol. The client only routes `pushItems()`
@@ -45,7 +42,8 @@ function createChargingHooks(actor) {
  * the pricing check happens lazily on every `dataset()` call.
  */
 function createChargingClient(options, actor) {
-    return new scrapely_client_1.ApifyClient({
+    const { ApifyClient } = (0, client_loader_js_1.scrapelyClient)();
+    return new ApifyClient({
         ...options,
         defaultDatasetId: actor.config.get('defaultDatasetId'),
         charging: createChargingHooks(actor),

package/dist/client-loader.d.ts

@@ -0,0 +1,20 @@
+/**
+ * scrapely-client, loaded on first use (axios, proxy-agent and friends cost
+ * ~0.7 s under gVisor, and most runs never need the API client).
+ *
+ * `USES_PUSH_DATA_INTERCEPTION` and `pushDataChargingContext` come from this
+ * module (not declared in the SDK) so the SDK and the client share a single
+ * symbol and a single async-local storage instance. Identity matters:
+ * `Actor.pushData()` checks dataset clients for this symbol, and the ambient
+ * charging context read inside the client's intercepted `pushItems()` must be
+ * the one `Actor.pushData()` entered.
+ *
+ * Loaded by the SDK entry: keep this file free of heavy imports.
+ */
+export type ScrapelyClientModule = typeof import('scrapely-client');
+/** The SDK's scrapely-client (loads it on first call). */
+export declare function scrapelyClient(): ScrapelyClientModule;
+/** The SDK's scrapely-client if it is already loaded, without loading it. */
+export declare function loadedScrapelyClient(): ScrapelyClientModule | undefined;
+/** Run `listener` once the SDK's scrapely-client is loaded (now, if it already is). */
+export declare function onScrapelyClientLoaded(listener: (mod: ScrapelyClientModule) => void): void;

package/dist/client-loader.js

@@ -0,0 +1,42 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.scrapelyClient = scrapelyClient;
+exports.loadedScrapelyClient = loadedScrapelyClient;
+exports.onScrapelyClientLoaded = onScrapelyClientLoaded;
+let scrapelyClientModule;
+const loadListeners = [];
+function loaded(mod) {
+    if (scrapelyClientModule)
+        return scrapelyClientModule;
+    scrapelyClientModule = mod;
+    for (const listener of loadListeners.splice(0))
+        listener(mod);
+    return mod;
+}
+/** The SDK's scrapely-client (loads it on first call). */
+function scrapelyClient() {
+    // eslint-disable-next-line @typescript-eslint/no-var-requires
+    return scrapelyClientModule ?? loaded(require('scrapely-client'));
+}
+/** The SDK's scrapely-client if it is already loaded, without loading it. */
+function loadedScrapelyClient() {
+    if (scrapelyClientModule)
+        return scrapelyClientModule;
+    try {
+        const cached = require.cache[require.resolve('scrapely-client')];
+        if (cached?.loaded)
+            return loaded(cached.exports);
+    }
+    catch {
+        // not resolvable — then nothing can be using it
+    }
+    return undefined;
+}
+/** Run `listener` once the SDK's scrapely-client is loaded (now, if it already is). */
+function onScrapelyClientLoaded(listener) {
+    const mod = loadedScrapelyClient();
+    if (mod)
+        listener(mod);
+    else
+        loadListeners.push(listener);
+}

package/dist/esm-loader-hooks.d.mts

@@ -0,0 +1,20 @@
+/**
+ * Off-thread ESM load hook (module.register) for Node versions without
+ * module.registerHooks (Node 20, 22.0-22.14). Registered by
+ * patch-browser-pool.ts only when puppeteer is installed for ESM imports.
+ *
+ * Appends one call to puppeteer-core's ESM PuppeteerNode.js that hands the
+ * class to patchPuppeteerNodeClass (published on a global by the preload in
+ * the main thread), before the package entry creates its default instance.
+ * Keep the pattern and suffix in sync with PUPPETEER_NODE_ESM_URL_RE /
+ * PUPPETEER_NODE_ESM_SUFFIX in patch-browser-pool.ts (a test checks this).
+ */
+export declare const PUPPETEER_NODE_ESM_URL_RE: RegExp;
+export declare const PUPPETEER_NODE_ESM_SUFFIX = "\n;globalThis[Symbol.for(\"scrapely.patchPuppeteerNodeClass\")]?.(PuppeteerNode);\n";
+interface LoadResult {
+    format?: string | null;
+    source?: string | ArrayBuffer | ArrayBufferView | null;
+    shortCircuit?: boolean;
+}
+export declare function load(url: string, context: unknown, nextLoad: (url: string, context: unknown) => Promise<LoadResult>): Promise<LoadResult>;
+export {};

package/dist/esm-loader-hooks.mjs

@@ -0,0 +1,21 @@
+/**
+ * Off-thread ESM load hook (module.register) for Node versions without
+ * module.registerHooks (Node 20, 22.0-22.14). Registered by
+ * patch-browser-pool.ts only when puppeteer is installed for ESM imports.
+ *
+ * Appends one call to puppeteer-core's ESM PuppeteerNode.js that hands the
+ * class to patchPuppeteerNodeClass (published on a global by the preload in
+ * the main thread), before the package entry creates its default instance.
+ * Keep the pattern and suffix in sync with PUPPETEER_NODE_ESM_URL_RE /
+ * PUPPETEER_NODE_ESM_SUFFIX in patch-browser-pool.ts (a test checks this).
+ */
+export const PUPPETEER_NODE_ESM_URL_RE = /\/puppeteer-core\/lib\/(esm\/)?puppeteer\/node\/PuppeteerNode\.js$/;
+export const PUPPETEER_NODE_ESM_SUFFIX = '\n;globalThis[Symbol.for("scrapely.patchPuppeteerNodeClass")]?.(PuppeteerNode);\n';
+export async function load(url, context, nextLoad) {
+    const result = await nextLoad(url, context);
+    if (result?.format === 'module' && result.source != null && PUPPETEER_NODE_ESM_URL_RE.test(url)) {
+        const source = typeof result.source === 'string' ? result.source : new TextDecoder().decode(result.source);
+        return { ...result, source: source + PUPPETEER_NODE_ESM_SUFFIX };
+    }
+    return result;
+}

package/dist/index.d.mts

@@ -0,0 +1,18 @@
+import type * as Sdk from './index.js';
+declare const sdk: typeof Sdk;
+export * from './index.js';
+/** `import scrapely from 'scrapely'`: the CommonJS exports object, as before this entry existed. */
+export default sdk;
+export declare const PlaywrightCrawler: typeof import("crawlee").PlaywrightCrawler;
+export declare const PuppeteerCrawler: typeof import("crawlee").PuppeteerCrawler;
+export declare const CheerioCrawler: typeof import("crawlee").CheerioCrawler;
+export declare const HttpCrawler: typeof import("crawlee").HttpCrawler;
+export declare const BasicCrawler: typeof import("crawlee").BasicCrawler;
+export declare const BrowserCrawler: typeof import("crawlee").BrowserCrawler;
+export declare const JSDOMCrawler: typeof import("crawlee").JSDOMCrawler;
+export declare const LinkeDOMCrawler: typeof import("crawlee").LinkeDOMCrawler;
+export declare const ApifyClient: typeof import("scrapely-client").ScrapelyClient;
+export declare const playwrightUtils: typeof Sdk.playwrightUtils;
+export declare const puppeteerUtils: typeof Sdk.puppeteerUtils;
+export declare const pushDataChargingContext: typeof Sdk.pushDataChargingContext;
+export declare let USES_PUSH_DATA_INTERCEPTION: typeof Sdk.USES_PUSH_DATA_INTERCEPTION | undefined;

package/dist/index.d.ts

@@ -1,13 +1,26 @@
 import './log-env.js';
+import type * as CrawleeBasic from '@crawlee/basic';
+import type * as CrawleeBrowser from '@crawlee/browser';
+import type * as CrawleeCheerio from '@crawlee/cheerio';
+import type * as CrawleeHttp from '@crawlee/http';
+import type * as CrawleeJsdom from '@crawlee/jsdom';
+import type * as CrawleeLinkedom from '@crawlee/linkedom';
+import type * as CrawleePlaywright from '@crawlee/playwright';
+import type * as CrawleePuppeteer from '@crawlee/puppeteer';
+import type * as ScrapelyClient from 'scrapely-client';
 export { setCdpUrl } from './patch-browser-pool.js';
 export { Actor, ACTOR_PERMISSION_LEVEL, type ActorInput, type InitOptions, type MainOptions, type ExitOptions, type RebootOptions, type AbortOptions, type PushDataChargeOptions, type MetamorphOptions, type RecordOptions, type OpenStorageOptions, type CallOptions, type CallTaskOptions, type StartOptions, type SetStatusMessageOptions, type WebhookOptions, type ApifyEnv, type UserFunc, type Token, type Timeout, EXIT_CODES, } from './actor';
 export type { StorageAlias, StorageId, StorageName, StorageIdentifier, StorageIdentifierWithoutAlias, } from './storage-identifiers.js';
-export { ApifyClient, type ApifyClientOptions, type ActorRun } from 'scrapely-client';
+export type { ApifyClientOptions, ActorRun } from 'scrapely-client';
+export declare const ApifyClient: typeof ScrapelyClient.ApifyClient;
+export type ApifyClient = ScrapelyClient.ApifyClient;
 export { Configuration, type ConfigurationOptions } from './configuration.js';
 export { DEGRADED_STATUS_MESSAGE, IpBanned, SolveAttemptError, SolverAuthError, SolverRateLimited, VendorUnavailable, degradedStatusMessage, detectVendor, isBlocked, reportSolveOutcome, solve, solverConfigured, solverStatus, stepAllowed, type Clearance, type SolveCookie, type SolveOptions, type SolverStepRequest, type SolverVendorStatus, } from './solver.js';
 export { ScrapelyEventManager as PlatformEventManager } from './events/scrapely-event-manager.js';
 export { ChargingManager, ChargeUncertainError, ChargeRejectedError, DEFAULT_DATASET_ITEM_EVENT, APIFY_DEFAULT_DATASET_ITEM_EVENT, SCRAPELY_DEFAULT_DATASET_ITEM_EVENT, hasDefaultDatasetItemEventInPricing, isDefaultDatasetItemEvent, resolveDefaultDatasetItemEventName, mergeChargeResults, pushDataAndCharge, pushDataWithCharges, validateRowCharges, type ChargeOptions, type ChargeResult, type ActorPricingInfo, } from './charging.js';
-export { createChargingClient, pushDataChargingContext, USES_PUSH_DATA_INTERCEPTION, type PpeAwarePushDataContext, } from './charging_client.js';
+export { createChargingClient, type PpeAwarePushDataContext, } from './charging_client.js';
+export declare const pushDataChargingContext: typeof ScrapelyClient.pushDataChargingContext;
+export declare const USES_PUSH_DATA_INTERCEPTION: typeof ScrapelyClient.USES_PUSH_DATA_INTERCEPTION;
 export { BillingTracker, type BillingClient } from './storage/billing-tracker.js';
 export { ApiStorageClientFactory, type ApiStorageConfig } from './storage/api-storage-client.js';
 export { ProxyConfiguration, isPlatformProxySentinel, validateProxySessionId } from './scrapely-proxy-configuration.js';
@@ -18,13 +31,21 @@ export type { LoggerOptions, KeyValueStoreIteratorOptions, KeyValueStoreOptions,
 export type { KeyConsumer } from './storage/key-value-store-client.js';
 export { utils, enqueueLinks, type EnqueueLinksOptions } from './utils.js';
 export { ScrapelyRequestList, type RequestListSource } from './request-list.js';
-export { PlaywrightCrawler } from '@crawlee/playwright';
-export { PuppeteerCrawler } from '@crawlee/puppeteer';
-export { CheerioCrawler } from '@crawlee/cheerio';
-export { HttpCrawler } from '@crawlee/http';
-export { BasicCrawler } from '@crawlee/basic';
-export { BrowserCrawler } from '@crawlee/browser';
-export { JSDOMCrawler } from '@crawlee/jsdom';
-export { LinkeDOMCrawler } from '@crawlee/linkedom';
-export { playwrightUtils } from '@crawlee/playwright';
-export { puppeteerUtils } from '@crawlee/puppeteer';
+export declare const PlaywrightCrawler: typeof CrawleePlaywright.PlaywrightCrawler;
+export type PlaywrightCrawler = CrawleePlaywright.PlaywrightCrawler;
+export declare const PuppeteerCrawler: typeof CrawleePuppeteer.PuppeteerCrawler;
+export type PuppeteerCrawler = CrawleePuppeteer.PuppeteerCrawler;
+export declare const CheerioCrawler: typeof CrawleeCheerio.CheerioCrawler;
+export type CheerioCrawler = CrawleeCheerio.CheerioCrawler;
+export declare const HttpCrawler: typeof CrawleeHttp.HttpCrawler;
+export type HttpCrawler<Context extends CrawleeHttp.InternalHttpCrawlingContext<any, any, CrawleeHttp.HttpCrawler<Context>>> = CrawleeHttp.HttpCrawler<Context>;
+export declare const BasicCrawler: typeof CrawleeBasic.BasicCrawler;
+export type BasicCrawler<Context extends CrawleeBasic.CrawlingContext = CrawleeBasic.BasicCrawlingContext> = CrawleeBasic.BasicCrawler<Context>;
+export declare const BrowserCrawler: typeof CrawleeBrowser.BrowserCrawler;
+export type BrowserCrawler<InternalBrowserPoolOptions extends import('@crawlee/browser-pool').BrowserPoolOptions = import('@crawlee/browser-pool').BrowserPoolOptions, LaunchOptions extends Record<string, any> | undefined = Record<string, any>, Context extends CrawleeBrowser.BrowserCrawlingContext = CrawleeBrowser.BrowserCrawlingContext, GoToOptions extends Record<string, any> = Record<string, any>> = CrawleeBrowser.BrowserCrawler<InternalBrowserPoolOptions, LaunchOptions, Context, GoToOptions>;
+export declare const JSDOMCrawler: typeof CrawleeJsdom.JSDOMCrawler;
+export type JSDOMCrawler = CrawleeJsdom.JSDOMCrawler;
+export declare const LinkeDOMCrawler: typeof CrawleeLinkedom.LinkeDOMCrawler;
+export type LinkeDOMCrawler = CrawleeLinkedom.LinkeDOMCrawler;
+export declare const playwrightUtils: typeof CrawleePlaywright.playwrightUtils;
+export declare const puppeteerUtils: typeof CrawleePuppeteer.puppeteerUtils;

package/dist/index.js

@@ -1,17 +1,15 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.log = exports.RequestQueue = exports.KeyValueStore = exports.Dataset = exports.createBrowserConfiguration = exports.BrowserConfiguration = exports.getPlatformProxyEnv = exports.createProxyConfiguration = exports.validateProxySessionId = exports.isPlatformProxySentinel = exports.ProxyConfiguration = exports.ApiStorageClientFactory = exports.BillingTracker = exports.USES_PUSH_DATA_INTERCEPTION = exports.pushDataChargingContext = exports.createChargingClient = exports.validateRowCharges = exports.pushDataWithCharges = exports.pushDataAndCharge = exports.mergeChargeResults = exports.resolveDefaultDatasetItemEventName = exports.isDefaultDatasetItemEvent = exports.hasDefaultDatasetItemEventInPricing = exports.SCRAPELY_DEFAULT_DATASET_ITEM_EVENT = exports.APIFY_DEFAULT_DATASET_ITEM_EVENT = exports.DEFAULT_DATASET_ITEM_EVENT = exports.ChargeRejectedError = exports.ChargeUncertainError = exports.ChargingManager = exports.PlatformEventManager = exports.stepAllowed = exports.solverStatus = exports.solverConfigured = exports.solve = exports.reportSolveOutcome = exports.isBlocked = exports.detectVendor = exports.degradedStatusMessage = exports.VendorUnavailable = exports.SolverRateLimited = exports.SolverAuthError = exports.SolveAttemptError = exports.IpBanned = exports.DEGRADED_STATUS_MESSAGE = exports.Configuration = exports.ApifyClient = exports.EXIT_CODES = exports.ACTOR_PERMISSION_LEVEL = exports.Actor = exports.setCdpUrl = void 0;
-exports.puppeteerUtils = exports.playwrightUtils = exports.LinkeDOMCrawler = exports.JSDOMCrawler = exports.BrowserCrawler = exports.BasicCrawler = exports.HttpCrawler = exports.CheerioCrawler = exports.PuppeteerCrawler = exports.PlaywrightCrawler = exports.ScrapelyRequestList = exports.enqueueLinks = exports.utils = exports.LoggerText = exports.LoggerJson = exports.Logger = exports.LogLevel = exports.Log = void 0;
+exports.Logger = exports.LogLevel = exports.Log = exports.log = exports.RequestQueue = exports.KeyValueStore = exports.Dataset = exports.createBrowserConfiguration = exports.BrowserConfiguration = exports.getPlatformProxyEnv = exports.createProxyConfiguration = exports.validateProxySessionId = exports.isPlatformProxySentinel = exports.ProxyConfiguration = exports.ApiStorageClientFactory = exports.BillingTracker = exports.createChargingClient = exports.validateRowCharges = exports.pushDataWithCharges = exports.pushDataAndCharge = exports.mergeChargeResults = exports.resolveDefaultDatasetItemEventName = exports.isDefaultDatasetItemEvent = exports.hasDefaultDatasetItemEventInPricing = exports.SCRAPELY_DEFAULT_DATASET_ITEM_EVENT = exports.APIFY_DEFAULT_DATASET_ITEM_EVENT = exports.DEFAULT_DATASET_ITEM_EVENT = exports.ChargeRejectedError = exports.ChargeUncertainError = exports.ChargingManager = exports.PlatformEventManager = exports.stepAllowed = exports.solverStatus = exports.solverConfigured = exports.solve = exports.reportSolveOutcome = exports.isBlocked = exports.detectVendor = exports.degradedStatusMessage = exports.VendorUnavailable = exports.SolverRateLimited = exports.SolverAuthError = exports.SolveAttemptError = exports.IpBanned = exports.DEGRADED_STATUS_MESSAGE = exports.Configuration = exports.EXIT_CODES = exports.ACTOR_PERMISSION_LEVEL = exports.Actor = exports.setCdpUrl = void 0;
+exports.ScrapelyRequestList = exports.enqueueLinks = exports.utils = exports.LoggerText = exports.LoggerJson = void 0;
 require("./log-env.js");
+const lazy_exports_js_1 = require("./lazy-exports.js");
 var patch_browser_pool_js_1 = require("./patch-browser-pool.js");
 Object.defineProperty(exports, "setCdpUrl", { enumerable: true, get: function () { return patch_browser_pool_js_1.setCdpUrl; } });
 var actor_1 = require("./actor");
 Object.defineProperty(exports, "Actor", { enumerable: true, get: function () { return actor_1.Actor; } });
 Object.defineProperty(exports, "ACTOR_PERMISSION_LEVEL", { enumerable: true, get: function () { return actor_1.ACTOR_PERMISSION_LEVEL; } });
 Object.defineProperty(exports, "EXIT_CODES", { enumerable: true, get: function () { return actor_1.EXIT_CODES; } });
-// Apify-compatible client exports (scrapely-client provides ApifyClient as an alias of ScrapelyClient)
-var scrapely_client_1 = require("scrapely-client");
-Object.defineProperty(exports, "ApifyClient", { enumerable: true, get: function () { return scrapely_client_1.ApifyClient; } });
 var configuration_js_1 = require("./configuration.js");
 Object.defineProperty(exports, "Configuration", { enumerable: true, get: function () { return configuration_js_1.Configuration; } });
 var solver_js_1 = require("./solver.js");
@@ -47,8 +45,6 @@ Object.defineProperty(exports, "pushDataWithCharges", { enumerable: true, get: f
 Object.defineProperty(exports, "validateRowCharges", { enumerable: true, get: function () { return charging_js_1.validateRowCharges; } });
 var charging_client_js_1 = require("./charging_client.js");
 Object.defineProperty(exports, "createChargingClient", { enumerable: true, get: function () { return charging_client_js_1.createChargingClient; } });
-Object.defineProperty(exports, "pushDataChargingContext", { enumerable: true, get: function () { return charging_client_js_1.pushDataChargingContext; } });
-Object.defineProperty(exports, "USES_PUSH_DATA_INTERCEPTION", { enumerable: true, get: function () { return charging_client_js_1.USES_PUSH_DATA_INTERCEPTION; } });
 var billing_tracker_js_1 = require("./storage/billing-tracker.js");
 Object.defineProperty(exports, "BillingTracker", { enumerable: true, get: function () { return billing_tracker_js_1.BillingTracker; } });
 // Storage exports - API-based storage 
@@ -85,28 +81,10 @@ Object.defineProperty(exports, "enqueueLinks", { enumerable: true, get: function
 // Request list (requestListSources input format; parity with the Python SDK's ScrapelyRequestList)
 var request_list_js_1 = require("./request-list.js");
 Object.defineProperty(exports, "ScrapelyRequestList", { enumerable: true, get: function () { return request_list_js_1.ScrapelyRequestList; } });
-// Crawler exports - re-export from crawlee packages for Apify SDK compatibility
-var playwright_1 = require("@crawlee/playwright");
-Object.defineProperty(exports, "PlaywrightCrawler", { enumerable: true, get: function () { return playwright_1.PlaywrightCrawler; } });
-var puppeteer_1 = require("@crawlee/puppeteer");
-Object.defineProperty(exports, "PuppeteerCrawler", { enumerable: true, get: function () { return puppeteer_1.PuppeteerCrawler; } });
-var cheerio_1 = require("@crawlee/cheerio");
-Object.defineProperty(exports, "CheerioCrawler", { enumerable: true, get: function () { return cheerio_1.CheerioCrawler; } });
-var http_1 = require("@crawlee/http");
-Object.defineProperty(exports, "HttpCrawler", { enumerable: true, get: function () { return http_1.HttpCrawler; } });
-var basic_1 = require("@crawlee/basic");
-Object.defineProperty(exports, "BasicCrawler", { enumerable: true, get: function () { return basic_1.BasicCrawler; } });
-var browser_1 = require("@crawlee/browser");
-Object.defineProperty(exports, "BrowserCrawler", { enumerable: true, get: function () { return browser_1.BrowserCrawler; } });
-var jsdom_1 = require("@crawlee/jsdom");
-Object.defineProperty(exports, "JSDOMCrawler", { enumerable: true, get: function () { return jsdom_1.JSDOMCrawler; } });
-var linkedom_1 = require("@crawlee/linkedom");
-Object.defineProperty(exports, "LinkeDOMCrawler", { enumerable: true, get: function () { return linkedom_1.LinkeDOMCrawler; } });
-// Utility exports for browser automation
-var playwright_2 = require("@crawlee/playwright");
-Object.defineProperty(exports, "playwrightUtils", { enumerable: true, get: function () { return playwright_2.playwrightUtils; } });
-var puppeteer_2 = require("@crawlee/puppeteer");
-Object.defineProperty(exports, "puppeteerUtils", { enumerable: true, get: function () { return puppeteer_2.puppeteerUtils; } });
 // Note: crawlee types are available through the real crawlee package which is installed
 // as a dependency. The build system creates a symlink from node_modules/crawlee to the
 // real crawlee package inside scrapely/node_modules/crawlee.
+// Must stay a plain call (not `export ... from`): cjs-module-lexer must not see
+// these names, or every ESM import of the SDK would load them (see lazy-exports.ts).
+// module.exports, not `exports`: tsx/esbuild replace module.exports in dev and tests.
+(0, lazy_exports_js_1.defineLazyExports)(module.exports);

package/dist/index.mjs

@@ -0,0 +1,39 @@
+/**
+ * ESM entry (`import ... from 'scrapely'`, and the `apify` shim's
+ * `export * from 'scrapely'`).
+ *
+ * Re-exports the CommonJS entry, plus ESM stand-ins for the names that load
+ * heavy packages (Crawlee crawlers, scrapely-client): Node reads every
+ * detected export of a CommonJS module when ESM imports it, so index.js hides
+ * those names from cjs-module-lexer and they are exported here instead. They
+ * load on first use (see lazy-exports.ts).
+ */
+import { createRequire } from 'node:module';
+const require = createRequire(import.meta.url);
+const sdk = require('./index.js');
+const { lazyExport } = require('./lazy-exports.js');
+const { onScrapelyClientLoaded } = require('./client-loader.js');
+export * from './index.js';
+/** `import scrapely from 'scrapely'`: the CommonJS exports object, as before this entry existed. */
+export default sdk;
+export const PlaywrightCrawler = lazyExport('PlaywrightCrawler', 'class');
+export const PuppeteerCrawler = lazyExport('PuppeteerCrawler', 'class');
+export const CheerioCrawler = lazyExport('CheerioCrawler', 'class');
+export const HttpCrawler = lazyExport('HttpCrawler', 'class');
+export const BasicCrawler = lazyExport('BasicCrawler', 'class');
+export const BrowserCrawler = lazyExport('BrowserCrawler', 'class');
+export const JSDOMCrawler = lazyExport('JSDOMCrawler', 'class');
+export const LinkeDOMCrawler = lazyExport('LinkeDOMCrawler', 'class');
+export const ApifyClient = lazyExport('ApifyClient', 'class');
+export const playwrightUtils = lazyExport('playwrightUtils', 'object');
+export const puppeteerUtils = lazyExport('puppeteerUtils', 'object');
+// The client's AsyncLocalStorage: methods (run, getStore, ...) act on the real instance.
+export const pushDataChargingContext = lazyExport('pushDataChargingContext', 'object');
+// A symbol cannot be stood in for: a live binding, set as soon as the SDK
+// loads scrapely-client (an Actor creating its API client, charging,
+// ApifyClient above, ...). Only clients from that module carry the marker,
+// so it is always set before any client could be checked for it.
+export let USES_PUSH_DATA_INTERCEPTION;
+onScrapelyClientLoaded((client) => {
+    USES_PUSH_DATA_INTERCEPTION = client.USES_PUSH_DATA_INTERCEPTION;
+});

package/dist/lazy-exports.d.ts

@@ -0,0 +1,14 @@
+/** name → [package, export]. */
+export declare const LAZY_EXPORTS: Readonly<Record<string, readonly [string, string]>>;
+/** Resolve one lazy export (loads its package on first call). */
+export declare function loadLazyExport(name: string): unknown;
+/** Define every lazy export as a self-replacing getter on a CommonJS exports object. */
+export declare function defineLazyExports(target: Record<string, unknown>): void;
+/**
+ * ESM stand-in for a lazy export: a Proxy that loads the real value on first
+ * use and forwards everything to it. For classes it is constructible and
+ * extendable (`new X()`, `class Y extends X`, `instanceof X`, statics); for
+ * objects (playwrightUtils, ...) property access works as usual. The one
+ * visible difference: the stand-in is not `===` the package's own export.
+ */
+export declare function lazyExport<T>(name: string, kind: 'class' | 'object'): T;