@crawlee/basic 4.0.0-beta.2 → 4.0.0-beta.21

package/internals/basic-crawler.js

@@ -1,7 +1,7 @@
 import { writeFile } from 'node:fs/promises';
 import { dirname } from 'node:path';
-import { AutoscaledPool, Configuration, CriticalError, Dataset, enqueueLinks, EnqueueStrategy, GotScrapingHttpClient, KeyValueStore, mergeCookies, NonRetryableError, purgeDefaultStorages, RequestProvider, RequestQueue, RequestQueueV1, RequestState, RetryRequestError, Router, SessionError, SessionPool, Statistics, validators, } from '@crawlee/core';
-import { RobotsTxtFile, ROTATE_PROXY_ERRORS } from '@crawlee/utils';
+import { AutoscaledPool, Configuration, ContextPipeline, ContextPipelineCleanupError, ContextPipelineInitializationError, ContextPipelineInterruptedError, CriticalError, Dataset, enqueueLinks, EnqueueStrategy, GotScrapingHttpClient, KeyValueStore, mergeCookies, NonRetryableError, purgeDefaultStorages, RequestHandlerError, RequestListAdapter, RequestManagerTandem, RequestProvider, RequestQueue, RequestQueueV1, RequestState, RetryRequestError, Router, SessionError, SessionPool, Statistics, validators, } from '@crawlee/core';
+import { getObjectType, isAsyncIterable, isIterable, RobotsTxtFile, ROTATE_PROXY_ERRORS } from '@crawlee/utils';
 import { stringify } from 'csv-stringify/sync';
 import { ensureDir, writeJSON } from 'fs-extra/esm';
 import ow from 'ow';
@@ -88,6 +88,11 @@ const SAFE_MIGRATION_WAIT_MILLIS = 20000;
 export class BasicCrawler {
     config;
     static CRAWLEE_STATE_KEY = 'CRAWLEE_STATE';
+    /**
+     * Tracks crawler instances that accessed shared state without having an explicit id.
+     * Used to detect and warn about multiple crawlers sharing the same state.
+     */
+    static useStateCrawlerIds = new Set();
     /**
      * A reference to the underlying {@link Statistics} class that collects and logs run statistics for requests.
      */
@@ -103,6 +108,10 @@ export class BasicCrawler {
      * Only available if used by the crawler.
      */
     requestQueue;
+    /**
+     * The main request-handling component of the crawler. It's initialized during the crawler startup.
+     */
+    requestManager;
     /**
      * A reference to the underlying {@link SessionPool} class that manages the crawler's {@link Session|sessions}.
      * Only available if used by the crawler.
@@ -116,11 +125,24 @@ export class BasicCrawler {
      * or to abort it by calling {@link AutoscaledPool.abort|`autoscaledPool.abort()`}.
      */
     autoscaledPool;
+    /**
+     * A reference to the underlying {@link ProxyConfiguration} class that manages the crawler's proxies.
+     * Only available if used by the crawler.
+     */
+    proxyConfiguration;
     /**
      * Default {@link Router} instance that will be used if we don't specify any {@link BasicCrawlerOptions.requestHandler|`requestHandler`}.
      * See {@link Router.addHandler|`router.addHandler()`} and {@link Router.addDefaultHandler|`router.addDefaultHandler()`}.
      */
     router = Router.create();
+    contextPipelineBuilder;
+    _contextPipeline;
+    get contextPipeline() {
+        if (this._contextPipeline === undefined) {
+            this._contextPipeline = this.contextPipelineBuilder();
+        }
+        return this._contextPipeline;
+    }
     running = false;
     hasFinishedBefore = false;
     log;
@@ -130,15 +152,16 @@ export class BasicCrawler {
     requestHandlerTimeoutMillis;
     internalTimeoutMillis;
     maxRequestRetries;
+    maxCrawlDepth;
     sameDomainDelayMillis;
     domainAccessedTime;
     maxSessionRotations;
-    handledRequestsCount;
+    maxRequestsPerCrawl;
+    handledRequestsCount = 0;
     statusMessageLoggingInterval;
     statusMessageCallback;
     sessionPoolOptions;
     useSessionPool;
-    crawlingContexts = new Map();
     autoscaledPoolOptions;
     events;
     httpClient;
@@ -146,10 +169,16 @@ export class BasicCrawler {
     respectRobotsTxtFile;
     onSkippedRequest;
     _closeEvents;
+    shouldLogMaxProcessedRequestsExceeded = true;
+    shouldLogMaxEnqueuedRequestsExceeded = true;
     experiments;
     robotsTxtFileCache;
     _experimentWarnings = {};
+    crawlerId;
+    hasExplicitId;
     static optionsShape = {
+        contextPipelineBuilder: ow.optional.object,
+        extendContext: ow.optional.function,
         requestList: ow.optional.object.validate(validators.requestList),
         requestQueue: ow.optional.object.validate(validators.requestQueue),
         // Subclasses override this function instead of passing it
@@ -163,13 +192,15 @@ export class BasicCrawler {
         sameDomainDelaySecs: ow.optional.number,
         maxSessionRotations: ow.optional.number,
         maxRequestsPerCrawl: ow.optional.number,
+        maxCrawlDepth: ow.optional.number,
         autoscaledPoolOptions: ow.optional.object,
         sessionPoolOptions: ow.optional.object,
         useSessionPool: ow.optional.boolean,
+        proxyConfiguration: ow.optional.object.validate(validators.proxyConfiguration),
         statusMessageLoggingInterval: ow.optional.number,
         statusMessageCallback: ow.optional.function,
         retryOnBlocked: ow.optional.boolean,
-        respectRobotsTxtFile: ow.optional.boolean,
+        respectRobotsTxtFile: ow.optional.any(ow.boolean, ow.object),
         onSkippedRequest: ow.optional.function,
         httpClient: ow.optional.object,
         // AutoscaledPool shorthands
@@ -181,21 +212,64 @@ export class BasicCrawler {
         log: ow.optional.object,
         experiments: ow.optional.object,
         statisticsOptions: ow.optional.object,
+        id: ow.optional.string,
     };
     /**
      * All `BasicCrawler` parameters are passed via an options object.
      */
-    constructor(options = {}, config = Configuration.getGlobalConfig()) {
+    constructor(options = {}, // cast because the constructor logic handles missing `contextPipelineBuilder` - the type is just for DX
+    config = Configuration.getGlobalConfig()) {
         this.config = config;
         ow(options, 'BasicCrawlerOptions', ow.object.exactShape(BasicCrawler.optionsShape));
-        const { requestList, requestQueue, maxRequestRetries = 3, sameDomainDelaySecs = 0, maxSessionRotations = 10, maxRequestsPerCrawl, autoscaledPoolOptions = {}, keepAlive, sessionPoolOptions = {}, useSessionPool = true, 
+        const { requestList, requestQueue, requestManager, maxRequestRetries = 3, sameDomainDelaySecs = 0, maxSessionRotations = 10, maxRequestsPerCrawl, maxCrawlDepth, autoscaledPoolOptions = {}, keepAlive, sessionPoolOptions = {}, useSessionPool = true, proxyConfiguration, 
         // AutoscaledPool shorthands
         minConcurrency, maxConcurrency, maxRequestsPerMinute, retryOnBlocked = false, respectRobotsTxtFile = false, onSkippedRequest, requestHandler, requestHandlerTimeoutSecs, errorHandler, failedRequestHandler, statusMessageLoggingInterval = 10, statusMessageCallback, statisticsOptions, httpClient, 
         // internal
-        log = defaultLog.child({ prefix: this.constructor.name }), experiments = {}, } = options;
-        this.requestList = requestList;
-        this.requestQueue = requestQueue;
+        log = defaultLog.child({ prefix: this.constructor.name }), experiments = {}, id, } = options;
+        // Store whether the user explicitly provided an ID
+        this.hasExplicitId = id !== undefined;
+        // Store the user-provided ID, or generate a unique one for tracking purposes (not for state key)
+        this.crawlerId = id ?? cryptoRandomObjectId();
+        // Store the builder so that it can be run when the contextPipeline is needed.
+        // Invoking it immediately would cause problems with parent constructor call order.
+        this.contextPipelineBuilder = () => {
+            let contextPipeline = (options.contextPipelineBuilder?.() ??
+                ContextPipeline.create()); // Thanks to the RequireContextPipeline, contextPipeline will only be undefined if InitialContextType is CrawlingContext
+            if (options.extendContext !== undefined) {
+                contextPipeline = contextPipeline.compose({
+                    action: async (context) => await options.extendContext(context),
+                });
+            }
+            contextPipeline = contextPipeline.compose({
+                action: async (context) => {
+                    const { request } = context;
+                    if (!this.requestMatchesEnqueueStrategy(request)) {
+                        // eslint-disable-next-line dot-notation
+                        const message = `Skipping request ${request.id} (starting url: ${request.url} -> loaded url: ${request.loadedUrl}) because it does not match the enqueue strategy (${request['enqueueStrategy']}).`;
+                        this.log.debug(message);
+                        request.noRetry = true;
+                        request.state = RequestState.SKIPPED;
+                        await this.handleSkippedRequest({ url: request.url, reason: 'redirect' });
+                        throw new ContextPipelineInterruptedError(message);
+                    }
+                    return context;
+                },
+            });
+            return contextPipeline;
+        };
+        if (requestManager !== undefined) {
+            if (requestList !== undefined || requestQueue !== undefined) {
+                throw new Error('The `requestManager` option cannot be used in conjunction with `requestList` and/or `requestQueue`');
+            }
+            this.requestManager = requestManager;
+            this.requestQueue = requestManager; // TODO(v4) - the cast is not fully legitimate here, but it's fine for internal usage by the BasicCrawler
+        }
+        else {
+            this.requestList = requestList;
+            this.requestQueue = requestQueue;
+        }
         this.httpClient = httpClient ?? new GotScrapingHttpClient();
+        this.proxyConfiguration = proxyConfiguration;
         this.log = log;
         this.statusMessageLoggingInterval = statusMessageLoggingInterval;
         this.statusMessageCallback = statusMessageCallback;
@@ -203,7 +277,7 @@ export class BasicCrawler {
         this.domainAccessedTime = new Map();
         this.experiments = experiments;
         this.robotsTxtFileCache = new LruCache({ maxLength: 1000 });
-        // FIXME any
+        this.handleSkippedRequest = this.handleSkippedRequest.bind(this);
         this.requestHandler = requestHandler ?? this.router;
         this.failedRequestHandler = failedRequestHandler;
         this.errorHandler = errorHandler;
@@ -228,13 +302,14 @@ export class BasicCrawler {
             this.requestQueue.requestLockSecs = Math.max(this.requestHandlerTimeoutMillis / 1000 + 5, 60);
         }
         this.maxRequestRetries = maxRequestRetries;
+        this.maxCrawlDepth = maxCrawlDepth;
         this.sameDomainDelayMillis = sameDomainDelaySecs * 1000;
         this.maxSessionRotations = maxSessionRotations;
-        this.handledRequestsCount = 0;
         this.stats = new Statistics({
             logMessage: `${log.getOptions().prefix} request statistics:`,
             log,
             config,
+            ...(this.hasExplicitId ? { id: this.crawlerId } : {}),
             ...statisticsOptions,
         });
         this.sessionPoolOptions = {
@@ -248,7 +323,6 @@ export class BasicCrawler {
             }
         }
         this.useSessionPool = useSessionPool;
-        this.crawlingContexts = new Map();
         const maxSignedInteger = 2 ** 31 - 1;
         if (this.requestHandlerTimeoutMillis > maxSignedInteger) {
             log.warning(`requestHandlerTimeoutMillis ${this.requestHandlerTimeoutMillis}` +
@@ -256,8 +330,8 @@ export class BasicCrawler {
             this.requestHandlerTimeoutMillis = maxSignedInteger;
         }
         this.internalTimeoutMillis = Math.min(this.internalTimeoutMillis, maxSignedInteger);
-        let shouldLogMaxPagesExceeded = true;
-        const isMaxPagesExceeded = () => maxRequestsPerCrawl && maxRequestsPerCrawl <= this.handledRequestsCount;
+        this.maxRequestsPerCrawl = maxRequestsPerCrawl;
+        const isMaxPagesExceeded = () => this.maxRequestsPerCrawl && this.maxRequestsPerCrawl <= this.handledRequestsCount;
         // eslint-disable-next-line prefer-const
         let { isFinishedFunction, isTaskReadyFunction } = autoscaledPoolOptions;
         // override even if `isFinishedFunction` provided by user - `keepAlive` has higher priority
@@ -271,10 +345,10 @@ export class BasicCrawler {
             runTaskFunction: this._runTaskFunction.bind(this),
             isTaskReadyFunction: async () => {
                 if (isMaxPagesExceeded()) {
-                    if (shouldLogMaxPagesExceeded) {
+                    if (this.shouldLogMaxProcessedRequestsExceeded) {
                         log.info('Crawler reached the maxRequestsPerCrawl limit of ' +
-                            `${maxRequestsPerCrawl} requests and will shut down soon. Requests that are in progress will be allowed to finish.`);
-                        shouldLogMaxPagesExceeded = false;
+                            `${this.maxRequestsPerCrawl} requests and will shut down soon. Requests that are in progress will be allowed to finish.`);
+                        this.shouldLogMaxProcessedRequestsExceeded = false;
                     }
                     return false;
                 }
@@ -282,7 +356,7 @@ export class BasicCrawler {
             },
             isFinishedFunction: async () => {
                 if (isMaxPagesExceeded()) {
-                    log.info(`Earlier, the crawler reached the maxRequestsPerCrawl limit of ${maxRequestsPerCrawl} requests ` +
+                    log.info(`Earlier, the crawler reached the maxRequestsPerCrawl limit of ${this.maxRequestsPerCrawl} requests ` +
                         'and all requests that were in progress at that time have now finished. ' +
                         `In total, the crawler processed ${this.handledRequestsCount} requests and will shut down.`);
                     return true;
@@ -311,14 +385,6 @@ export class BasicCrawler {
     isProxyError(error) {
         return ROTATE_PROXY_ERRORS.some((x) => this._getMessageFromError(error)?.includes(x));
     }
-    /**
-     * Checks whether the given crawling context is getting blocked by anti-bot protection using several heuristics.
-     * Returns `false` if the request is not blocked, otherwise returns a string with a description of the block reason.
-     * @param _crawlingContext The crawling context to check.
-     */
-    async isRequestBlocked(_crawlingContext) {
-        throw new Error('the "isRequestBlocked" method is not implemented in this crawler.');
-    }
     /**
      * This method is periodically called by the crawler, every `statusMessageLoggingInterval` seconds.
      */
@@ -350,7 +416,7 @@ export class BasicCrawler {
                 message = `Experiencing problems, ${this.stats.state.requestsFailed - previousState.requestsFailed || this.stats.state.requestsFailed} failed requests in the past ${this.statusMessageLoggingInterval} seconds.`;
             }
             else {
-                const total = this.requestQueue?.getTotalCount() || this.requestList?.length();
+                const total = this.requestManager?.getTotalCount();
                 message = `Crawled ${this.stats.state.requestsFinished}${total ? `/${total}` : ''} pages, ${this.stats.state.requestsFailed} failed requests, desired concurrency ${this.autoscaledPool?.desiredConcurrency ?? 0}.`;
             }
             if (this.statusMessageCallback) {
@@ -390,20 +456,30 @@ export class BasicCrawler {
             if (this.requestQueue?.name === 'default' && purgeRequestQueue) {
                 await this.requestQueue.drop();
                 this.requestQueue = await this._getRequestQueue();
+                this.requestManager = undefined;
+                await this.initializeRequestManager();
+                this.handledRequestsCount = 0; // This would've been reset by this._init() further down below, but at that point `handledRequestsCount` could prevent `addRequests` from adding the initial requests
             }
             this.stats.reset();
             await this.stats.resetStore();
             await this.sessionPool?.resetStore();
         }
         this.running = true;
-        await purgeDefaultStorages({ onlyPurgeOnce: true });
+        this.shouldLogMaxProcessedRequestsExceeded = true;
+        this.shouldLogMaxEnqueuedRequestsExceeded = true;
+        await purgeDefaultStorages({
+            onlyPurgeOnce: true,
+            client: this.config.getStorageClient(),
+            config: this.config,
+        });
         if (requests) {
             await this.addRequests(requests, addRequestsOptions);
         }
         await this._init();
         await this.stats.startCapturing();
         const periodicLogger = this.getPeriodicLogger();
-        await this.setStatusMessage('Starting the crawler.', { level: 'INFO' });
+        // Don't await, we don't want to block the execution
+        void this.setStatusMessage('Starting the crawler.', { level: 'INFO' });
         const sigintHandler = async () => {
             this.log.warning('Pausing... Press CTRL+C again to force exit. To resume, do: CRAWLEE_PURGE_ON_START=0 npm start');
             await this._pauseOnMigration();
@@ -452,7 +528,8 @@ export class BasicCrawler {
                 finished = true;
             }
             periodicLogger.stop();
-            await this.setStatusMessage(`Finished! Total ${this.stats.state.requestsFinished + this.stats.state.requestsFailed} requests: ${this.stats.state.requestsFinished} succeeded, ${this.stats.state.requestsFailed} failed.`, { isStatusMessageTerminal: true, level: 'INFO' });
+            // Don't await, we don't want to block the execution
+            void this.setStatusMessage(`Finished! Total ${this.stats.state.requestsFinished + this.stats.state.requestsFailed} requests: ${this.stats.state.requestsFinished} succeeded, ${this.stats.state.requestsFailed} failed.`, { isStatusMessageTerminal: true, level: 'INFO' });
             this.running = false;
             this.hasFinishedBefore = true;
         }
@@ -462,6 +539,8 @@ export class BasicCrawler {
      * Gracefully stops the current run of the crawler.
      *
      * All the tasks active at the time of calling this method will be allowed to finish.
+     *
+     * To stop the crawler immediately, use {@link BasicCrawler.teardown|`crawler.teardown()`} instead.
      */
     stop(message = 'The crawler has been gracefully stopped.') {
         // Gracefully starve the this.autoscaledPool, so it doesn't start new tasks. Resolves once the pool is cleared.
@@ -478,13 +557,57 @@ export class BasicCrawler {
         if (!this.requestQueue && this.requestList) {
             this.log.warningOnce('When using RequestList and RequestQueue at the same time, you should instantiate both explicitly and provide them in the crawler options, to ensure correctly handled restarts of the crawler.');
         }
-        this.requestQueue ??= await this._getRequestQueue();
+        if (!this.requestQueue) {
+            this.requestQueue = await this._getRequestQueue();
+            this.requestManager = undefined;
+        }
+        if (!this.requestManager) {
+            this.requestManager =
+                this.requestList === undefined
+                    ? this.requestQueue
+                    : new RequestManagerTandem(this.requestList, this.requestQueue);
+        }
         return this.requestQueue;
     }
     async useState(defaultValue = {}) {
         const kvs = await KeyValueStore.open(null, { config: this.config });
+        if (this.hasExplicitId) {
+            const stateKey = `${BasicCrawler.CRAWLEE_STATE_KEY}_${this.crawlerId}`;
+            return kvs.getAutoSavedValue(stateKey, defaultValue);
+        }
+        BasicCrawler.useStateCrawlerIds.add(this.crawlerId);
+        if (BasicCrawler.useStateCrawlerIds.size > 1) {
+            defaultLog.warningOnce('Multiple crawler instances are calling useState() without an explicit `id` option. \n' +
+                'This means they will share the same state object, which is likely unintended. \n' +
+                'To fix this, provide a unique `id` option to each crawler instance. \n' +
+                'Example: new BasicCrawler({ id: "my-crawler-1", ... })');
+        }
         return kvs.getAutoSavedValue(BasicCrawler.CRAWLEE_STATE_KEY, defaultValue);
     }
+    get pendingRequestCountApproximation() {
+        return this.requestManager?.getPendingCount() ?? 0;
+    }
+    calculateEnqueuedRequestLimit(explicitLimit) {
+        if (this.maxRequestsPerCrawl === undefined) {
+            return explicitLimit;
+        }
+        const limit = Math.max(0, this.maxRequestsPerCrawl - this.handledRequestsCount - this.pendingRequestCountApproximation);
+        return Math.min(limit, explicitLimit ?? Infinity);
+    }
+    async handleSkippedRequest(options) {
+        if (options.reason === 'limit' && this.shouldLogMaxEnqueuedRequestsExceeded) {
+            this.log.info('The number of requests enqueued by the crawler reached the maxRequestsPerCrawl limit of ' +
+                `${this.maxRequestsPerCrawl} requests and no further requests will be added.`);
+            this.shouldLogMaxEnqueuedRequestsExceeded = false;
+        }
+        if (options.reason === 'enqueueLimit') {
+            const enqueuedRequestLimit = this.calculateEnqueuedRequestLimit();
+            if (enqueuedRequestLimit === undefined || enqueuedRequestLimit !== 0) {
+                this.log.info('The number of requests enqueued by the crawler reached the enqueueLinks limit.');
+            }
+        }
+        await this.onSkippedRequest?.(options);
+    }
     /**
      * Adds requests to the queue in batches. By default, it will resolve after the initial batch is added, and continue
      * adding the rest in background. You can configure the batch size via `batchSize` option and the sleep time in between
@@ -497,33 +620,57 @@ export class BasicCrawler {
      * @param options Options for the request queue
      */
     async addRequests(requests, options = {}) {
-        const requestQueue = await this.getRequestQueue();
-        if (!this.respectRobotsTxtFile) {
-            return requestQueue.addRequestsBatched(requests, options);
-        }
-        const allowedRequests = [];
-        const skipped = new Set();
-        for (const request of requests) {
-            const url = typeof request === 'string' ? request : request.url;
-            if (await this.isAllowedBasedOnRobotsTxtFile(url)) {
-                allowedRequests.push(request);
-            }
-            else {
-                skipped.add(url);
-                await this.onSkippedRequest?.({ url, reason: 'robotsTxt' });
+        await this.getRequestQueue();
+        const requestLimit = this.calculateEnqueuedRequestLimit();
+        const skippedBecauseOfRobots = new Set();
+        const skippedBecauseOfLimit = new Set();
+        const skippedBecauseOfMaxCrawlDepth = new Set();
+        const isAllowedBasedOnRobotsTxtFile = this.isAllowedBasedOnRobotsTxtFile.bind(this);
+        const maxCrawlDepth = this.maxCrawlDepth;
+        ow(requests, ow.object
+            .is((value) => isIterable(value) || isAsyncIterable(value))
+            .message((value) => `Expected an iterable or async iterable, got ${getObjectType(value)}`));
+        async function* filteredRequests() {
+            let yieldedRequestCount = 0;
+            for await (const request of requests) {
+                const url = typeof request === 'string' ? request : request.url;
+                if (requestLimit !== undefined && yieldedRequestCount >= requestLimit) {
+                    skippedBecauseOfLimit.add(url);
+                    continue;
+                }
+                if (maxCrawlDepth !== undefined && request.crawlDepth > maxCrawlDepth) {
+                    skippedBecauseOfMaxCrawlDepth.add(url);
+                    continue;
+                }
+                if (await isAllowedBasedOnRobotsTxtFile(url)) {
+                    yield request;
+                    yieldedRequestCount += 1;
+                }
+                else {
+                    skippedBecauseOfRobots.add(url);
+                }
             }
         }
-        if (skipped.size > 0) {
+        const result = await this.requestManager.addRequestsBatched(filteredRequests(), options);
+        if (skippedBecauseOfRobots.size > 0) {
             this.log.warning(`Some requests were skipped because they were disallowed based on the robots.txt file`, {
-                skipped: [...skipped],
+                skipped: [...skippedBecauseOfRobots],
             });
-            if (this.onSkippedRequest) {
-                await Promise.all([...skipped].map((url) => {
-                    return this.onSkippedRequest({ url, reason: 'robotsTxt' });
-                }));
-            }
         }
-        return requestQueue.addRequestsBatched(allowedRequests, options);
+        if (skippedBecauseOfRobots.size > 0 ||
+            skippedBecauseOfLimit.size > 0 ||
+            skippedBecauseOfMaxCrawlDepth.size > 0) {
+            await Promise.all([...skippedBecauseOfRobots]
+                .map((url) => {
+                return this.handleSkippedRequest({ url, reason: 'robotsTxt' });
+            })
+                .concat([...skippedBecauseOfLimit].map((url) => {
+                return this.handleSkippedRequest({ url, reason: 'limit' });
+            }), [...skippedBecauseOfMaxCrawlDepth].map((url) => {
+                return this.handleSkippedRequest({ url, reason: 'depth' });
+            })));
+        }
+        return result;
     }
     /**
      * Pushes data to the specified {@link Dataset}, or the default crawler {@link Dataset} by calling {@link Dataset.pushData}.
@@ -563,7 +710,21 @@ export class BasicCrawler {
         const dataset = await this.getDataset();
         const items = await dataset.export(options);
         if (format === 'csv') {
-            const value = stringify([Object.keys(items[0]), ...items.map((item) => Object.values(item))]);
+            let value;
+            if (items.length === 0) {
+                value = '';
+            }
+            else {
+                const keys = options?.collectAllKeys
+                    ? Array.from(new Set(items.flatMap(Object.keys)))
+                    : Object.keys(items[0]);
+                value = stringify([
+                    keys,
+                    ...items.map((item) => {
+                        return keys.map((k) => item[k]);
+                    }),
+                ]);
+            }
             await ensureDir(dirname(path));
             await writeFile(path, value);
             this.log.info(`Export to ${path} finished!`);
@@ -575,6 +736,9 @@ export class BasicCrawler {
         }
         return items;
     }
+    /**
+     * Initializes the crawler.
+     */
     async _init() {
         if (!this.events.isInitialized()) {
             await this.events.init();
@@ -589,10 +753,13 @@ export class BasicCrawler {
             // Assuming there are not more than 20 browsers running at once;
             this.sessionPool.setMaxListeners(20);
         }
+        await this.initializeRequestManager();
         await this._loadHandledRequestCount();
     }
-    async _runRequestHandler(crawlingContext) {
-        await this.requestHandler(crawlingContext);
+    async runRequestHandler(crawlingContext) {
+        await this.contextPipeline.call(crawlingContext, async (finalContext) => {
+            await addTimeoutToPromise(async () => this.requestHandler(finalContext), this.requestHandlerTimeoutMillis, `requestHandler timed out after ${this.requestHandlerTimeoutMillis / 1000} seconds (${finalContext.request.id}).`);
+        });
     }
     /**
      * Handles blocked request
@@ -608,7 +775,8 @@ export class BasicCrawler {
             return true;
         }
         const robotsTxtFile = await this.getRobotsTxtFileForUrl(url);
-        return !robotsTxtFile || robotsTxtFile.isAllowed(url);
+        const userAgent = typeof this.respectRobotsTxtFile === 'object' ? this.respectRobotsTxtFile?.userAgent : '*';
+        return !robotsTxtFile || robotsTxtFile.isAllowed(url, userAgent);
     }
     async getRobotsTxtFileForUrl(url) {
         if (!this.respectRobotsTxtFile) {
@@ -662,36 +830,36 @@ export class BasicCrawler {
         await Promise.all([requestListPersistPromise, this.stats.persistState()]);
     }
     /**
-     * Fetches request from either RequestList or RequestQueue. If request comes from a RequestList
-     * and RequestQueue is present then enqueues it to the queue first.
+     * Initializes the RequestManager based on the configured requestList and requestQueue.
      */
-    async _fetchNextRequest() {
-        if (!this.requestList || (await this.requestList.isFinished())) {
-            return this.requestQueue?.fetchNextRequest();
-        }
-        const request = await this.requestList.fetchNextRequest();
-        if (!this.requestQueue)
-            return request;
-        if (!request)
-            return this.requestQueue.fetchNextRequest();
-        try {
-            await this.requestQueue.addRequest(request, { forefront: true });
+    async initializeRequestManager() {
+        if (this.requestManager !== undefined) {
+            return;
         }
-        catch (err) {
-            // If requestQueue.addRequest() fails here then we must reclaim it back to
-            // the RequestList because probably it's not yet in the queue!
-            this.log.error('Adding of request from the RequestList to the RequestQueue failed, reclaiming request back to the list.', { request });
-            await this.requestList.reclaimRequest(request);
-            return null;
+        if (this.requestList && this.requestQueue) {
+            // Create a RequestManagerTandem if both RequestList and RequestQueue are provided
+            this.requestManager = new RequestManagerTandem(this.requestList, this.requestQueue);
         }
-        await this.requestList.markRequestHandled(request);
-        return this.requestQueue.fetchNextRequest();
+        else if (this.requestQueue) {
+            // Use RequestQueue directly if only it is provided
+            this.requestManager = this.requestQueue;
+        }
+        else if (this.requestList) {
+            // Use RequestList directly if only it is provided
+            // Make it compatible with the IRequestManager interface
+            this.requestManager = new RequestListAdapter(this.requestList);
+        }
+        // If neither RequestList nor RequestQueue is provided, leave the requestManager uninitialized until `getRequestQueue` is called
     }
     /**
-     * Executed when `errorHandler` finishes or the request is successful.
-     * Can be used to clean up orphaned browser pages.
+     * Fetches the next request to process from the underlying request provider.
      */
-    async _cleanupContext(_crawlingContext) { }
+    async _fetchNextRequest() {
+        if (this.requestManager === undefined) {
+            throw new Error(`_fetchNextRequest called on an uninitialized crawler`);
+        }
+        return this.requestManager.fetchNextRequest();
+    }
     /**
      * Delays processing of the request based on the `sameDomainDelaySecs` option,
      * adding it back to the queue after the timeout passes. Returns `true` if the request
@@ -729,18 +897,21 @@ export class BasicCrawler {
      * then retries them in a case of an error, etc.
      */
     async _runTaskFunction() {
-        const source = this.requestQueue || this.requestList || (await this.getRequestQueue());
-        let request;
-        let session;
-        await this._timeoutAndRetry(async () => {
-            request = await this._fetchNextRequest();
-        }, this.internalTimeoutMillis, `Fetching next request timed out after ${this.internalTimeoutMillis / 1e3} seconds.`);
+        const source = this.requestManager;
+        if (!source)
+            throw new Error('Request provider is not initialized!');
+        const request = await this._timeoutAndRetry(this._fetchNextRequest.bind(this), this.internalTimeoutMillis, `Fetching next request timed out after ${this.internalTimeoutMillis / 1e3} seconds.`);
         tryCancel();
-        if (this.useSessionPool) {
-            await this._timeoutAndRetry(async () => {
-                session = await this.sessionPool.getSession();
-            }, this.internalTimeoutMillis, `Fetching session timed out after ${this.internalTimeoutMillis / 1e3} seconds.`);
-        }
+        const session = this.useSessionPool
+            ? await this._timeoutAndRetry(async () => {
+                return await this.sessionPool.newSession({
+                    proxyInfo: await this.proxyConfiguration?.newProxyInfo({
+                        request: request ?? undefined,
+                    }),
+                    maxUsageCount: 1,
+                });
+            }, this.internalTimeoutMillis, `Fetching session timed out after ${this.internalTimeoutMillis / 1e3} seconds.`)
+            : undefined;
         tryCancel();
         if (!request || this.delayRequest(request, source)) {
             return;
@@ -750,7 +921,7 @@ export class BasicCrawler {
             request.state = RequestState.SKIPPED;
             request.noRetry = true;
             await source.markRequestHandled(request);
-            await this.onSkippedRequest?.({
+            await this.handleSkippedRequest({
                 url: request.url,
                 reason: 'robotsTxt',
             });
@@ -760,36 +931,34 @@ export class BasicCrawler {
         request.loadedUrl = undefined;
         const statisticsId = request.id || request.uniqueKey;
         this.stats.startJob(statisticsId);
-        // Shared crawling context
-        // @ts-expect-error
-        // All missing properties (that extend CrawlingContext) are set dynamically,
-        // but TS does not know that, so otherwise it would throw when compiling.
+        const deferredCleanup = [];
         const crawlingContext = {
             id: cryptoRandomObjectId(10),
-            crawler: this,
             log: this.log,
             request,
             session,
+            proxyInfo: session?.proxyInfo,
             enqueueLinks: async (options) => {
-                return enqueueLinks({
-                    // specify the RQ first to allow overriding it
-                    requestQueue: await this.getRequestQueue(),
-                    robotsTxtFile: await this.getRobotsTxtFileForUrl(request.url),
-                    onSkippedRequest: this.onSkippedRequest,
-                    ...options,
-                });
+                const requestQueue = await this.getRequestQueue();
+                return await this.enqueueLinksWithCrawlDepth(options, request, requestQueue);
+            },
+            addRequests: async (requests, options = {}) => {
+                const newCrawlDepth = request.crawlDepth + 1;
+                const requestsGenerator = this.addCrawlDepthRequestGenerator(requests, newCrawlDepth);
+                await this.addRequests(requestsGenerator, options);
             },
-            addRequests: this.addRequests.bind(this),
             pushData: this.pushData.bind(this),
             useState: this.useState.bind(this),
-            sendRequest: createSendRequest(this.httpClient, request, session, () => crawlingContext.proxyInfo?.url),
+            sendRequest: createSendRequest(this.httpClient, request, session),
             getKeyValueStore: async (idOrName) => KeyValueStore.open(idOrName, { config: this.config }),
+            registerDeferredCleanup: (cleanup) => {
+                deferredCleanup.push(cleanup);
+            },
         };
-        this.crawlingContexts.set(crawlingContext.id, crawlingContext);
         let isRequestLocked = true;
         try {
             request.state = RequestState.REQUEST_HANDLER;
-            await addTimeoutToPromise(async () => this._runRequestHandler(crawlingContext), this.requestHandlerTimeoutMillis, `requestHandler timed out after ${this.requestHandlerTimeoutMillis / 1000} seconds (${request.id}).`);
+            await this.runRequestHandler(crawlingContext);
             await this._timeoutAndRetry(async () => source.markRequestHandled(request), this.internalTimeoutMillis, `Marking request ${request.url} (${request.id}) as handled timed out after ${this.internalTimeoutMillis / 1e3} seconds.`);
             isRequestLocked = false; // markRequestHandled succeeded and unlocked the request
             this.stats.finishJob(statisticsId, request.retryCount);
@@ -798,7 +967,8 @@ export class BasicCrawler {
             request.state = RequestState.DONE;
             crawlingContext.session?.markGood();
         }
-        catch (err) {
+        catch (rawError) {
+            const err = this.unwrapError(rawError);
             try {
                 request.state = RequestState.ERROR_HANDLER;
                 await addTimeoutToPromise(async () => this._requestFunctionErrorHandler(err, crawlingContext, source), this.internalTimeoutMillis, `Handling request failure of ${request.url} (${request.id}) timed out after ${this.internalTimeoutMillis / 1e3} seconds.`);
@@ -808,24 +978,24 @@ export class BasicCrawler {
                 request.state = RequestState.DONE;
             }
             catch (secondaryError) {
-                if (!secondaryError.triggeredFromUserHandler &&
+                const unwrappedSecondaryError = this.unwrapError(secondaryError);
+                if (!unwrappedSecondaryError.triggeredFromUserHandler &&
                     // avoid reprinting the same critical error multiple times, as it will be printed by Nodejs at the end anyway
-                    !(secondaryError instanceof CriticalError)) {
+                    !(unwrappedSecondaryError instanceof CriticalError)) {
                     const apifySpecific = process.env.APIFY_IS_AT_HOME
                         ? `This may have happened due to an internal error of Apify's API or due to a misconfigured crawler.`
                         : '';
-                    this.log.exception(secondaryError, 'An exception occurred during handling of failed request. ' +
+                    this.log.exception(unwrappedSecondaryError, 'An exception occurred during handling of failed request. ' +
                         `This places the crawler and its underlying storages into an unknown state and crawling will be terminated. ${apifySpecific}`);
                 }
                 request.state = RequestState.ERROR;
-                throw secondaryError;
+                throw unwrappedSecondaryError;
             }
             // decrease the session score if the request fails (but the error handler did not throw)
             crawlingContext.session?.markBad();
         }
         finally {
-            await this._cleanupContext(crawlingContext);
-            this.crawlingContexts.delete(crawlingContext.id);
+            await Promise.all(deferredCleanup.map((cleanup) => cleanup()));
             // Safety net - release the lock if nobody managed to do it before
             if (isRequestLocked && source instanceof RequestProvider) {
                 try {
@@ -838,19 +1008,60 @@ export class BasicCrawler {
         }
     }
     /**
-     * Run async callback with given timeout and retry.
+     * Wrapper around the crawling context's `enqueueLinks` method:
+     * - Injects `crawlDepth` to each request being added based on the crawling context request.
+     * - Provides defaults for the `enqueueLinks` options based on the crawler configuration.
+     *      - These options can be overridden by the user.
+     * @internal
+     */
+    async enqueueLinksWithCrawlDepth(options, request, requestQueue) {
+        const transformRequestFunctionWrapper = (newRequest) => {
+            newRequest.crawlDepth = request.crawlDepth + 1;
+            if (this.maxCrawlDepth !== undefined && newRequest.crawlDepth > this.maxCrawlDepth) {
+                newRequest.skippedReason = 'depth';
+                return false;
+            }
+            // After injecting the crawlDepth, we call the user-provided transform function, if there is one.
+            return options.transformRequestFunction?.(newRequest) ?? newRequest;
+        };
+        return await enqueueLinks({
+            requestQueue,
+            robotsTxtFile: await this.getRobotsTxtFileForUrl(request.url),
+            onSkippedRequest: this.handleSkippedRequest,
+            limit: this.calculateEnqueuedRequestLimit(options.limit),
+            // Allow user options to override defaults set above ⤴
+            ...options,
+            transformRequestFunction: transformRequestFunctionWrapper,
+        });
+    }
+    /**
+     * Generator function that yields requests injected with the given crawl depth.
+     * @internal
+     */
+    async *addCrawlDepthRequestGenerator(requests, newRequestDepth) {
+        for await (const request of requests) {
+            if (typeof request === 'string') {
+                yield { url: request, crawlDepth: newRequestDepth };
+            }
+            else {
+                request.crawlDepth ??= newRequestDepth;
+                yield request;
+            }
+        }
+    }
+    /**
+     * Run async callback with given timeout and retry. Returns the result of the callback.
      * @ignore
      */
     async _timeoutAndRetry(handler, timeout, error, maxRetries = 3, retried = 1) {
         try {
-            await addTimeoutToPromise(handler, timeout, error);
+            return await addTimeoutToPromise(handler, timeout, error);
         }
         catch (e) {
             if (retried <= maxRetries) {
                 // we retry on any error, not just timeout
                 this.log.warning(`${e.message} (retrying ${retried}/${maxRetries})`);
-                void this._timeoutAndRetry(handler, timeout, error, maxRetries, retried + 1);
-                return;
+                return this._timeoutAndRetry(handler, timeout, error, maxRetries, retried + 1);
             }
             throw e;
         }
@@ -859,24 +1070,13 @@ export class BasicCrawler {
      * Returns true if either RequestList or RequestQueue have a request ready for processing.
      */
     async _isTaskReadyFunction() {
-        // First check RequestList, since it's only in memory.
-        const isRequestListEmpty = this.requestList ? await this.requestList.isEmpty() : true;
-        // If RequestList is not empty, task is ready, no reason to check RequestQueue.
-        if (!isRequestListEmpty)
-            return true;
-        // If RequestQueue is not empty, task is ready, return true, otherwise false.
-        return this.requestQueue ? !(await this.requestQueue.isEmpty()) : false;
+        return this.requestManager !== undefined && !(await this.requestManager.isEmpty());
     }
     /**
      * Returns true if both RequestList and RequestQueue have all requests finished.
      */
     async _defaultIsFinishedFunction() {
-        const [isRequestListFinished, isRequestQueueFinished] = await Promise.all([
-            this.requestList ? this.requestList.isFinished() : true,
-            this.requestQueue ? this.requestQueue.isFinished() : true,
-        ]);
-        // If both are finished, return true, otherwise return false.
-        return isRequestListFinished && isRequestQueueFinished;
+        return !this.requestManager || (await this.requestManager.isFinished());
     }
     async _rotateSession(crawlingContext) {
         const { request } = crawlingContext;
@@ -884,6 +1084,18 @@ export class BasicCrawler {
         request.sessionRotationCount++;
         crawlingContext.session?.retire();
     }
+    /**
+     * Unwraps errors thrown by the context pipeline to get the actual user error.
+     * RequestHandlerError and ContextPipelineInitializationError wrap the actual error.
+     */
+    unwrapError(error) {
+        if (error instanceof RequestHandlerError ||
+            error instanceof ContextPipelineInitializationError ||
+            error instanceof ContextPipelineCleanupError) {
+            return this.unwrapError(error.cause);
+        }
+        return error;
+    }
     /**
      * Handles errors thrown by user provided requestHandler()
      */
@@ -896,7 +1108,8 @@ export class BasicCrawler {
         const shouldRetryRequest = this._canRequestBeRetried(request, error);
         if (shouldRetryRequest) {
             await this.stats.errorTrackerRetry.addAsync(error, crawlingContext);
-            await this.errorHandler?.(crawlingContext, error);
+            await this.errorHandler?.(crawlingContext, // valid cast - ExtendedContext transitively extends CrawlingContext
+            error);
             if (error instanceof SessionError) {
                 await this._rotateSession(crawlingContext);
             }
@@ -948,7 +1161,8 @@ export class BasicCrawler {
         const message = this._getMessageFromError(error, true);
         this.log.error(`Request failed and reached maximum retries. ${message}`, { id, url, method, uniqueKey });
         if (this.failedRequestHandler) {
-            await this.failedRequestHandler?.(crawlingContext, error);
+            await this.failedRequestHandler?.(crawlingContext, // valid cast - ExtendedContext transitively extends CrawlingContext
+            error);
         }
     }
     /**
@@ -986,19 +1200,11 @@ export class BasicCrawler {
         return request.retryCount < maxRequestRetries;
     }
     /**
-     * Updates handledRequestsCount from possibly stored counts,
-     * usually after worker migration. Since one of the stores
-     * needs to have priority when both are present,
-     * it is the request queue, because generally, the request
-     * list will first be dumped into the queue and then left
-     * empty.
+     * Updates handledRequestsCount from possibly stored counts, usually after worker migration.
      */
     async _loadHandledRequestCount() {
-        if (this.requestQueue) {
-            this.handledRequestsCount = await this.requestQueue.handledCount();
-        }
-        else if (this.requestList) {
-            this.handledRequestsCount = this.requestList.handledCount();
+        if (this.requestManager) {
+            this.handledRequestsCount = await this.requestManager.handledCount();
         }
     }
     async _executeHooks(hooks, ...args) {
@@ -1009,14 +1215,15 @@ export class BasicCrawler {
         }
     }
     /**
-     * Function for cleaning up after all request are processed.
-     * @ignore
+     * Stops the crawler immediately.
+     *
+     * This method doesn't wait for currently active requests to finish.
+     *
+     * To stop the crawler gracefully (waiting for all running requests to finish), use {@link BasicCrawler.stop|`crawler.stop()`} instead.
      */
     async teardown() {
         this.events.emit("persistState" /* EventType.PERSIST_STATE */, { isMigrating: false });
-        if (this.useSessionPool) {
-            await this.sessionPool.teardown();
-        }
+        await this.sessionPool?.teardown();
         if (this._closeEvents) {
             await this.events.close();
         }