@crawlee/core 3.17.1-beta.5 → 3.17.1-beta.50

package/autoscaling/autoscaled_pool.js

@@ -415,14 +415,17 @@ class AutoscaledPool {
         this.isStopped = true;
         await new Promise((resolve, reject) => {
             let timeout;
+            let interval;
             if (timeoutSecs) {
                 timeout = setTimeout(() => {
+                    // Clean up the polling interval to prevent it from leaking on timeout.
+                    clearInterval(interval);
                     const err = new Error("The pool's running tasks did not finish" +
                         `in ${timeoutSecs} secs after pool.pause() invocation.`);
                     reject(err);
                 }, timeoutSecs);
             }
-            const interval = setInterval(() => {
+            interval = setInterval(() => {
                 if (this._currentConcurrency <= 0) {
                     // Clean up timeout and interval to prevent process hanging.
                     if (timeout)

package/autoscaling/snapshotter.d.ts

@@ -35,13 +35,13 @@ export interface SnapshotterOptions {
     /**
      * Defines the maximum number of new rate limit errors within
      * the given interval.
-     * @default 1
+     * @default 3
      */
     maxClientErrors?: number;
     /**
      * Sets the interval in seconds for which a history of resource snapshots
      * will be kept. Increasing this to very high numbers will affect performance.
-     * @default 60
+     * @default 30
      */
     snapshotHistorySecs?: number;
     /** @internal */

package/cookie_utils.js

@@ -107,6 +107,8 @@ function mergeCookies(url, sourceCookies) {
             if (!cookieString)
                 continue;
             const cookie = tough_cookie_1.Cookie.parse(cookieString);
+            if (!cookie)
+                throw new errors_1.CookieParseError(cookieString);
             const similarKeyCookie = jar.getCookiesSync(url).find((c) => {
                 return cookie.key !== c.key && cookie.key.toLowerCase() === c.key.toLowerCase();
             });

package/http_clients/got-scraping-http-client.js

@@ -22,6 +22,14 @@ class GotScrapingHttpClient {
         });
         return {
             ...gotResult,
+            complete: gotResult.complete,
+            headers: gotResult.headers,
+            ip: gotResult.ip,
+            redirectUrls: gotResult.redirectUrls,
+            statusCode: gotResult.statusCode,
+            statusMessage: gotResult.statusMessage,
+            trailers: gotResult.trailers,
+            url: gotResult.url,
             body: gotResult.body,
             request: { url: request.url, ...gotResult.request },
         };

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@crawlee/core",
-    "version": "3.17.1-beta.5",
+    "version": "3.17.1-beta.50",
     "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": ">=16.0.0"
@@ -59,9 +59,9 @@
         "@apify/pseudo_url": "^2.0.30",
         "@apify/timeout": "^0.3.0",
         "@apify/utilities": "^2.7.10",
-        "@crawlee/memory-storage": "3.17.1-beta.5",
-        "@crawlee/types": "3.17.1-beta.5",
-        "@crawlee/utils": "3.17.1-beta.5",
+        "@crawlee/memory-storage": "3.17.1-beta.50",
+        "@crawlee/types": "3.17.1-beta.50",
+        "@crawlee/utils": "3.17.1-beta.50",
         "@sapphire/async-queue": "^1.5.1",
         "@vladfrangu/async_event_emitter": "^2.2.2",
         "csv-stringify": "^6.2.0",
@@ -83,5 +83,5 @@
             }
         }
     },
-    "gitHead": "9479ce41601143fc01d56047f69d72ff54a33d1b"
+    "gitHead": "adfd432d80ce08c0f0f04bbfce7dc53efc28aeb0"
 }

package/router.d.ts

@@ -2,13 +2,26 @@ import type { Dictionary } from '@crawlee/types';
 import type { CrawlingContext, LoadedRequest, RestrictedCrawlingContext } from './crawlers/crawler_commons';
 import type { Request } from './request';
 import type { Awaitable } from './typedefs';
-export interface RouterHandler<Context extends Omit<RestrictedCrawlingContext, 'enqueueLinks'> = CrawlingContext> extends Router<Context> {
+/**
+ * The crawling context received by a route handler, with `request.userData` narrowed to `UserData`.
+ */
+export type RouterHandlerContext<Context, UserData extends Dictionary> = Omit<Context, 'request'> & {
+    request: LoadedRequest<Request<UserData>>;
+};
+/**
+ * The set of labels accepted by {@link Router.addHandler}. When the router declares a concrete
+ * route map (e.g. `{ PRODUCT: ...; CATEGORY: ... }`), only those labels (plus symbols) are
+ * allowed — unknown labels become a compile-time error. When the map is left open (the default
+ * `Record<string, ...>`), any string or symbol label is accepted, preserving the original behaviour.
+ */
+export type RouterLabel<Routes extends Record<keyof Routes, Dictionary>> = string extends keyof Routes ? string | symbol : (keyof Routes & string) | symbol;
+export interface RouterHandler<Context extends Omit<RestrictedCrawlingContext, 'enqueueLinks'> = CrawlingContext, Routes extends Record<keyof Routes, Dictionary> = Record<string, GetUserDataFromRequest<Context['request']>>> extends Router<Context, Routes> {
     (ctx: Context): Awaitable<void>;
 }
 export type GetUserDataFromRequest<T> = T extends Request<infer Y> ? Y : never;
-export type RouterRoutes<Context, UserData extends Dictionary> = {
-    [label in string | symbol]: (ctx: Omit<Context, 'request'> & {
-        request: Request<UserData>;
+export type RouterRoutes<Context, Routes extends Record<keyof Routes, Dictionary>> = {
+    [Label in keyof Routes]: (ctx: Omit<Context, 'request'> & {
+        request: Request<Routes[Label]>;
     }) => Awaitable<void>;
 };
 /**
@@ -75,8 +88,30 @@ export type RouterRoutes<Context, UserData extends Dictionary> = {
  *    ctx.log.info('...');
  * });
  * ```
+ *
+ * To get `request.userData` typed per label, declare a route map and pass it as the second
+ * type argument. The label passed to {@link Router.addHandler} then drives the type of
+ * `request.userData`, and unknown labels are rejected at compile time:
+ *
+ * ```ts
+ * import { createCheerioRouter, CheerioCrawlingContext } from 'crawlee';
+ *
+ * interface Routes {
+ *     PRODUCT: { sku: string; price: number };
+ *     CATEGORY: { categoryId: string };
+ * }
+ *
+ * const router = createCheerioRouter<CheerioCrawlingContext, Routes>();
+ *
+ * router.addHandler('PRODUCT', async ({ request }) => {
+ *     request.userData.sku;   // string
+ *     request.userData.price; // number
+ * });
+ *
+ * router.addHandler('TYPO', async () => {}); // compile error: not a known label
+ * ```
  */
-export declare class Router<Context extends Omit<RestrictedCrawlingContext, 'enqueueLinks'>> {
+export declare class Router<Context extends Omit<RestrictedCrawlingContext, 'enqueueLinks'>, Routes extends Record<keyof Routes, Dictionary> = Record<string, GetUserDataFromRequest<Context['request']>>> {
     private readonly routes;
     private readonly middlewares;
     /**
@@ -85,17 +120,22 @@ export declare class Router<Context extends Omit<RestrictedCrawlingContext, 'enq
      */
     protected constructor();
     /**
-     * Registers new route handler for given label.
+     * Registers new route handler for given label. When the router declares a route map, the
+     * `label` is restricted to the declared labels and `request.userData` is typed accordingly.
+     */
+    addHandler<Label extends keyof Routes & string>(label: Label, handler: (ctx: RouterHandlerContext<Context, Routes[Label]>) => Awaitable<void>): void;
+    /**
+     * Registers new route handler for given label, explicitly typing `request.userData` via the
+     * `UserData` type argument. Useful when the router has no declared route map (the open default)
+     * and you want to type a single handler, or to register a handler under a `symbol` label.
      */
-    addHandler<UserData extends Dictionary = GetUserDataFromRequest<Context['request']>>(label: string | symbol, handler: (ctx: Omit<Context, 'request'> & {
-        request: LoadedRequest<Request<UserData>>;
-    }) => Awaitable<void>): void;
+    addHandler<UserData extends Dictionary = GetUserDataFromRequest<Context['request']>>(label: RouterLabel<Routes>, handler: (ctx: RouterHandlerContext<Context, UserData>) => Awaitable<void>): void;
     /**
-     * Registers default route handler.
+     * Registers default route handler. As a fallback it can receive any request (including labels not
+     * declared in the route map), so `request.userData` defaults to the context's `userData` type
+     * (loosely typed by default). Pass an explicit `UserData` type argument to narrow it.
      */
-    addDefaultHandler<UserData extends Dictionary = GetUserDataFromRequest<Context['request']>>(handler: (ctx: Omit<Context, 'request'> & {
-        request: LoadedRequest<Request<UserData>>;
-    }) => Awaitable<void>): void;
+    addDefaultHandler<UserData extends Dictionary = GetUserDataFromRequest<Context['request']>>(handler: (ctx: RouterHandlerContext<Context, UserData>) => Awaitable<void>): void;
     /**
      * Registers a middleware that will be fired before the matching route handler.
      * Multiple middlewares can be registered, they will be fired in the same order.
@@ -129,5 +169,6 @@ export declare class Router<Context extends Omit<RestrictedCrawlingContext, 'enq
      * await crawler.run();
      * ```
      */
-    static create<Context extends Omit<RestrictedCrawlingContext, 'enqueueLinks'> = CrawlingContext, UserData extends Dictionary = GetUserDataFromRequest<Context['request']>>(routes?: RouterRoutes<Context, UserData>): RouterHandler<Context>;
+    static create<Context extends Omit<RestrictedCrawlingContext, 'enqueueLinks'> = CrawlingContext, Routes extends Record<keyof Routes, Dictionary> = Record<string, GetUserDataFromRequest<Context['request']>>>(routes?: RouterRoutes<Context, Routes>): RouterHandler<Context, Routes>;
+    static create<Context extends Omit<RestrictedCrawlingContext, 'enqueueLinks'> = CrawlingContext, UserData extends Dictionary = GetUserDataFromRequest<Context['request']>>(routes?: RouterRoutes<Context, Record<string, UserData>>): RouterHandler<Context, Record<string, UserData>>;
 }

package/router.js

@@ -67,6 +67,28 @@ const defaultRoute = Symbol('default-route');
  *    ctx.log.info('...');
  * });
  * ```
+ *
+ * To get `request.userData` typed per label, declare a route map and pass it as the second
+ * type argument. The label passed to {@link Router.addHandler} then drives the type of
+ * `request.userData`, and unknown labels are rejected at compile time:
+ *
+ * ```ts
+ * import { createCheerioRouter, CheerioCrawlingContext } from 'crawlee';
+ *
+ * interface Routes {
+ *     PRODUCT: { sku: string; price: number };
+ *     CATEGORY: { categoryId: string };
+ * }
+ *
+ * const router = createCheerioRouter<CheerioCrawlingContext, Routes>();
+ *
+ * router.addHandler('PRODUCT', async ({ request }) => {
+ *     request.userData.sku;   // string
+ *     request.userData.price; // number
+ * });
+ *
+ * router.addHandler('TYPO', async () => {}); // compile error: not a known label
+ * ```
  */
 class Router {
     /**
@@ -87,15 +109,14 @@ class Router {
             value: []
         });
     }
-    /**
-     * Registers new route handler for given label.
-     */
     addHandler(label, handler) {
         this.validate(label);
         this.routes.set(label, handler);
     }
     /**
-     * Registers default route handler.
+     * Registers default route handler. As a fallback it can receive any request (including labels not
+     * declared in the route map), so `request.userData` defaults to the context's `userData` type
+     * (loosely typed by default). Pass an explicit `UserData` type argument to narrow it.
      */
     addDefaultHandler(handler) {
         this.validate(defaultRoute);
@@ -133,26 +154,6 @@ class Router {
             throw new Error(message);
         }
     }
-    /**
-     * Creates new router instance. This instance can then serve as a `requestHandler` of your crawler.
-     *
-     * ```ts
-     * import { Router, CheerioCrawler, CheerioCrawlingContext } from 'crawlee';
-     *
-     * const router = Router.create<CheerioCrawlingContext>();
-     * router.addHandler('label-a', async (ctx) => {
-     *    ctx.log.info('...');
-     * });
-     * router.addDefaultHandler(async (ctx) => {
-     *    ctx.log.info('...');
-     * });
-     *
-     * const crawler = new CheerioCrawler({
-     *     requestHandler: router,
-     * });
-     * await crawler.run();
-     * ```
-     */
     static create(routes) {
         const router = new Router();
         const obj = Object.create(Function.prototype);

package/storages/request_dedup_cache.d.ts

@@ -0,0 +1,23 @@
+/**
+ * A fixed-size, direct-mapped cache for `uniqueKey`-based request deduplication.
+ *
+ * `RequestProvider.requestCache` only remembers the first batch of requests, so repeated
+ * `addRequestsBatched()` calls with overlapping URLs re-submit already-enqueued requests
+ * (https://github.com/apify/crawlee/issues/3120). This is a separate, cheaper cache we can populate on
+ * every batch: a fixed number of slots indexed by a hash of the request's cache key, storing the
+ * server-assigned `requestId`. Memory is capped by the slot count regardless of the working set size;
+ * a hash collision just overwrites a slot, causing an occasional cache miss (a harmless re-submission)
+ * but never a false hit — so a genuinely new request is never dropped.
+ *
+ * @internal
+ */
+export declare class RequestDeduplicationCache {
+    private readonly size;
+    private keys;
+    private ids;
+    constructor(size?: number);
+    get(cacheKey: string): string | null;
+    add(cacheKey: string, requestId: string): void;
+    clear(): void;
+    private indexOf;
+}

package/storages/request_dedup_cache.js

@@ -0,0 +1,66 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.RequestDeduplicationCache = void 0;
+/**
+ * A fixed-size, direct-mapped cache for `uniqueKey`-based request deduplication.
+ *
+ * `RequestProvider.requestCache` only remembers the first batch of requests, so repeated
+ * `addRequestsBatched()` calls with overlapping URLs re-submit already-enqueued requests
+ * (https://github.com/apify/crawlee/issues/3120). This is a separate, cheaper cache we can populate on
+ * every batch: a fixed number of slots indexed by a hash of the request's cache key, storing the
+ * server-assigned `requestId`. Memory is capped by the slot count regardless of the working set size;
+ * a hash collision just overwrites a slot, causing an occasional cache miss (a harmless re-submission)
+ * but never a false hit — so a genuinely new request is never dropped.
+ *
+ * @internal
+ */
+class RequestDeduplicationCache {
+    // The slot count is the same for every queue, so it's a fixed default rather than a per-consumer option.
+    constructor(size = 1000000) {
+        Object.defineProperty(this, "size", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: size
+        });
+        Object.defineProperty(this, "keys", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: void 0
+        });
+        Object.defineProperty(this, "ids", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: void 0
+        });
+        this.keys = new Array(size);
+        this.ids = new Array(size);
+    }
+    get(cacheKey) {
+        const index = this.indexOf(cacheKey);
+        return this.keys[index] === cacheKey ? this.ids[index] : null;
+    }
+    add(cacheKey, requestId) {
+        const index = this.indexOf(cacheKey);
+        this.keys[index] = cacheKey;
+        this.ids[index] = requestId;
+    }
+    clear() {
+        this.keys = new Array(this.size);
+        this.ids = new Array(this.size);
+    }
+    // A cheap FNV-1a hash of the cache key — avoids pulling in a dedicated hashing dependency.
+    indexOf(cacheKey) {
+        /* eslint-disable no-bitwise */
+        let hash = 0x811c9dc5;
+        for (let i = 0; i < cacheKey.length; i++) {
+            hash ^= cacheKey.charCodeAt(i);
+            hash = Math.imul(hash, 0x01000193);
+        }
+        return (hash >>> 0) % this.size;
+        /* eslint-enable no-bitwise */
+    }
+}
+exports.RequestDeduplicationCache = RequestDeduplicationCache;

package/storages/request_list.js

@@ -324,9 +324,11 @@ class RequestList {
                 const sourcesFromFunction = await this.sourcesFunction();
                 const sourcesFromFunctionCount = sourcesFromFunction.length;
                 for (let i = 0; i < sourcesFromFunctionCount; i++) {
-                    const source = sourcesFromFunction.shift();
+                    const source = sourcesFromFunction[i];
+                    delete sourcesFromFunction[i];
                     this._addRequest(source);
                 }
+                sourcesFromFunction.length = 0;
             }
             catch (e) {
                 const err = e;

package/storages/request_provider.d.ts

@@ -5,6 +5,7 @@ import { Configuration } from '../configuration';
 import type { ProxyConfiguration } from '../proxy_configuration';
 import type { InternalSource, RequestOptions, Source } from '../request';
 import { Request } from '../request';
+import { RequestDeduplicationCache } from './request_dedup_cache';
 import type { IStorage, StorageManagerOptions } from './storage_manager';
 export type RequestsLike = AsyncIterable<Source | string> | Iterable<Source | string> | (Source | string)[];
 /**
@@ -74,6 +75,12 @@ export declare abstract class RequestProvider implements IStorage, IRequestManag
     private initialHandledCount;
     protected queueHeadIds: ListDictionary<string>;
     protected requestCache: LruCache<RequestLruItem>;
+    /**
+     * Remembers the `requestId` of every request already submitted to the client — including background
+     * batches that `requestCache` skips — so overlapping URL sets aren't re-submitted.
+     * See {@link RequestDeduplicationCache} for why this is a separate, cheaper cache.
+     */
+    protected requestSeenCache: RequestDeduplicationCache;
     protected recentlyHandledRequestsCache: LruCache<boolean>;
     protected queuePausedForMigration: boolean;
     protected lastActivity: Date;

package/storages/request_provider.js

@@ -11,8 +11,14 @@ const configuration_1 = require("../configuration");
 const log_1 = require("../log");
 const request_1 = require("../request");
 const access_checking_1 = require("./access_checking");
+const request_dedup_cache_1 = require("./request_dedup_cache");
 const storage_manager_1 = require("./storage_manager");
 const utils_2 = require("./utils");
+/**
+ * Maximum number of consecutive batch-add attempts that make no progress before the remaining
+ * unprocessed requests are skipped, so permanently rejected requests don't retry forever.
+ */
+const MAX_UNPROCESSED_REQUESTS_RETRIES = 3;
 class RequestProvider {
     constructor(options, config = configuration_1.Configuration.getGlobalConfig()) {
         Object.defineProperty(this, "config", {
@@ -113,6 +119,17 @@ class RequestProvider {
             writable: true,
             value: void 0
         });
+        /**
+         * Remembers the `requestId` of every request already submitted to the client — including background
+         * batches that `requestCache` skips — so overlapping URL sets aren't re-submitted.
+         * See {@link RequestDeduplicationCache} for why this is a separate, cheaper cache.
+         */
+        Object.defineProperty(this, "requestSeenCache", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: void 0
+        });
         Object.defineProperty(this, "recentlyHandledRequestsCache", {
             enumerable: true,
             configurable: true,
@@ -151,6 +168,7 @@ class RequestProvider {
         });
         this.proxyConfiguration = options.proxyConfiguration;
         this.requestCache = new datastructures_1.LruCache({ maxLength: options.requestCacheMaxSize });
+        this.requestSeenCache = new request_dedup_cache_1.RequestDeduplicationCache();
         this.recentlyHandledRequestsCache = new datastructures_1.LruCache({ maxLength: options.recentlyHandledRequestsMaxSize });
         this.log = log_1.log.child({ prefix: `${options.logPrefix}(${this.id}, ${this.name ?? 'no-name'})` });
         const eventManager = config.getEventManager();
@@ -227,6 +245,7 @@ class RequestProvider {
         };
         const { requestId, wasAlreadyPresent } = queueOperationInfo;
         this._cacheRequest(cacheKey, queueOperationInfo);
+        this.requestSeenCache.add(cacheKey, requestId);
         if (!wasAlreadyPresent && !this.recentlyHandledRequestsCache.get(requestId)) {
             this.assumedTotalCount++;
             // Performance optimization: add request straight to head if possible
@@ -288,16 +307,17 @@ class RequestProvider {
         const requestsToAdd = new Map();
         for (const request of requests) {
             const cacheKey = getCachedRequestId(request.uniqueKey);
+            // Prefer the full `requestCache` record; fall back to the dedup cache for background batches it skips.
             const cachedInfo = this.requestCache.get(cacheKey);
-            if (cachedInfo) {
-                request.id = cachedInfo.id;
+            const knownRequestId = cachedInfo?.id ?? this.requestSeenCache.get(cacheKey);
+            if (knownRequestId) {
+                request.id = knownRequestId;
                 results.processedRequests.push({
                     wasAlreadyPresent: true,
-                    // We may assume that if request is in local cache then also the information if the
-                    // request was already handled is there because just one client should be using one queue.
-                    wasAlreadyHandled: cachedInfo.isHandled,
-                    requestId: cachedInfo.id,
-                    uniqueKey: cachedInfo.uniqueKey,
+                    // The dedup cache doesn't track the handled state; only the full record does.
+                    wasAlreadyHandled: cachedInfo?.isHandled ?? false,
+                    requestId: knownRequestId,
+                    uniqueKey: request.uniqueKey,
                 });
             }
             else if (!requestsToAdd.has(request.uniqueKey)) {
@@ -320,6 +340,8 @@ class RequestProvider {
             if (cache) {
                 this._cacheRequest(cacheKey, { ...newRequest, forefront });
             }
+            // Unlike `requestCache`, populate this on every batch (including background ones).
+            this.requestSeenCache.add(cacheKey, requestId);
             if (!wasAlreadyPresent && !this.recentlyHandledRequestsCache.get(requestId)) {
                 this.assumedTotalCount++;
                 // Performance optimization: add request straight to head if possible
@@ -385,13 +407,20 @@ class RequestProvider {
         const requestIterator = generateRequests();
         const chunks = (0, utils_1.peekableAsyncIterable)((0, utils_1.chunkedAsyncIterable)(requestIterator, effectiveChunkSize));
         const chunksIterator = chunks[Symbol.asyncIterator]();
-        const attemptToAddToQueueAndAddAnyUnprocessed = async (providedRequests, cache = true) => {
+        const attemptToAddToQueueAndAddAnyUnprocessed = async (providedRequests, cache = true, unsuccessfulAttempts = 0) => {
             const resultsToReturn = [];
             const apiResult = await this.addRequests(providedRequests, { forefront: options.forefront, cache });
             resultsToReturn.push(...apiResult.processedRequests);
             if (apiResult.unprocessedRequests.length) {
+                // Count attempts that make no progress, so permanently rejected requests (e.g. a malformed
+                // `userData` shape causing a 400) don't loop forever. Any progress resets the counter.
+                const attempts = apiResult.processedRequests.length ? 0 : unsuccessfulAttempts + 1;
+                if (attempts >= MAX_UNPROCESSED_REQUESTS_RETRIES) {
+                    this.log.warning(`Some requests were consistently rejected by the request queue and will be skipped after ${MAX_UNPROCESSED_REQUESTS_RETRIES} attempts. This usually means the request data is malformed (e.g. an invalid 'userData' shape).`, { unprocessedRequests: apiResult.unprocessedRequests });
+                    return resultsToReturn;
+                }
                 await (0, utils_1.sleep)(waitBetweenBatchesMillis);
-                resultsToReturn.push(...(await attemptToAddToQueueAndAddAnyUnprocessed(providedRequests.filter((r) => !apiResult.processedRequests.some((pr) => pr.uniqueKey === r.uniqueKey)), false)));
+                resultsToReturn.push(...(await attemptToAddToQueueAndAddAnyUnprocessed(providedRequests.filter((r) => !apiResult.processedRequests.some((pr) => pr.uniqueKey === r.uniqueKey)), false, attempts)));
             }
             return resultsToReturn;
         };
@@ -546,6 +575,7 @@ class RequestProvider {
         this.assumedTotalCount = 0;
         this.assumedHandledCount = 0;
         this.requestCache.clear();
+        this.requestSeenCache.clear();
     }
     /**
      * Caches information about request to beware of unneeded addRequest() calls.