@crawlee/core 4.0.0-beta.104 → 4.0.0-beta.106

package/storages/request_queue.js

@@ -2,12 +2,13 @@ import { inspect } from 'node:util';
 import { downloadListOfUrls, isAsyncIterable, isIterable, sleep } from '@crawlee/utils';
 import ow from 'ow';
 import { LruCache } from '@apify/datastructures';
+import { tryCancel } from '@apify/timeout';
 import { Configuration } from '../configuration.js';
 import { getObjectType } from '../debug.js';
 import { chunkedAsyncIterable, peekableAsyncIterable } from '../iterables.js';
 import { Request } from '../request.js';
 import { serviceLocator } from '../service_locator.js';
-import { checkStorageAccess } from './access_checking.js';
+import { activeStorageTransaction, rejectOperationInTransaction, withDirectStorageAccess } from './transaction.js';
 import { StorageStatsTracker } from './storage_stats.js';
 import { resolveStorageIdentifier } from './storage_instance_manager.js';
 import { getRequestId, purgeDefaultStorages } from './utils.js';
@@ -55,26 +56,28 @@ export class RequestQueue {
     id;
     name;
     backend;
-    proxyConfiguration;
+    #proxyConfiguration;
     log;
+    // kept as TS-private: request_queue tests read this cache directly
     requestCache;
     /**
      * 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.
      */
-    requestSeenCache;
-    queuePausedForMigration = false;
+    #requestSeenCache;
+    #queuePausedForMigration = false;
+    // kept as TS-private: packages/core/test request-queue tests write this counter directly
     inProgressRequestBatchCount = 0;
     /**
      * The largest expected request-processing time (in seconds) seen so far via
      * {@link setExpectedRequestProcessingTimeSecs}. Used to ensure that value is only ever raised, never
      * lowered, before being forwarded to the storage backend.
      */
-    expectedRequestProcessingSecs = 0;
-    httpClient;
-    events;
-    statsTracker = new StorageStatsTracker({
+    #expectedRequestProcessingSecs = 0;
+    #httpClient;
+    #events;
+    #statsTracker = new StorageStatsTracker({
         writeCount: 0,
         headItemReadCount: 0,
     });
@@ -83,7 +86,7 @@ export class RequestQueue {
      * queue-head reads issued to the underlying storage backend). Counted per backend call.
      */
     get stats() {
-        return this.statsTracker.current;
+        return this.#statsTracker.current;
     }
     /**
      * @internal
@@ -91,14 +94,14 @@ export class RequestQueue {
     constructor(options) {
         this.id = options.metadata.id;
         this.name = options.metadata.name;
-        this.events = serviceLocator.getEventManager();
+        this.#events = serviceLocator.getEventManager();
         this.backend = options.backend;
-        this.proxyConfiguration = options.proxyConfiguration;
+        this.#proxyConfiguration = options.proxyConfiguration;
         this.requestCache = new LruCache({ maxLength: MAX_CACHED_REQUESTS });
-        this.requestSeenCache = new RequestDeduplicationCache();
+        this.#requestSeenCache = new RequestDeduplicationCache();
         this.log = serviceLocator.getLogger().child({ prefix: `RequestQueue(${this.id}, ${this.name ?? 'no-name'})` });
-        this.events.on("migrating" /* EventType.MIGRATING */, async () => {
-            this.queuePausedForMigration = true;
+        this.#events.on("migrating" /* EventType.MIGRATING */, async () => {
+            this.#queuePausedForMigration = true;
         });
     }
     /**
@@ -134,7 +137,7 @@ export class RequestQueue {
      * @param [options] Request queue operation options.
      */
     async addRequest(requestLike, options = {}) {
-        checkStorageAccess();
+        const transaction = activeStorageTransaction();
         ow(requestLike, ow.object);
         ow(options, ow.object.exactShape({
             forefront: ow.optional.boolean,
@@ -150,10 +153,14 @@ export class RequestQueue {
             id: ow.undefined,
         }));
         const request = requestLike instanceof Request ? requestLike : new Request(requestLike);
+        if (transaction?.policy.requestQueue === 'deferred') {
+            return this.addRequestDeferred(transaction, request, forefront);
+        }
         const cacheKey = getRequestId(request.uniqueKey);
         const cachedInfo = this.requestCache.get(cacheKey);
         if (cachedInfo) {
             request.id = cachedInfo.id;
+            this.recordRequestJournalEntry(transaction, [request], forefront, true);
             return {
                 wasAlreadyPresent: true,
                 // We may assume that if request is in local cache then also the information if the
@@ -164,17 +171,163 @@ export class RequestQueue {
                 forefront,
             };
         }
-        this.statsTracker.add('writeCount');
+        this.#statsTracker.add('writeCount');
         const { processedRequests } = await this.backend.addBatchOfRequests([request], { forefront });
+        this.recordRequestJournalEntry(transaction, [request], forefront, true);
         const queueOperationInfo = {
             ...processedRequests[0],
             uniqueKey: request.uniqueKey,
             forefront,
         };
         this.cacheRequest(cacheKey, queueOperationInfo);
-        this.requestSeenCache.add(cacheKey, request.id);
+        this.#requestSeenCache.add(cacheKey, request.id);
         return queueOperationInfo;
     }
+    /**
+     * Journals an addition for introspection only; these entries are never replayed. A no-op unless the
+     * transaction is open, so detached and outliving writers stay out of the journal.
+     */
+    recordRequestJournalEntry(transaction, requests, forefront, writeThrough) {
+        if (!transaction?.isActive || requests.length === 0)
+            return;
+        transaction.recordJournalEntry({
+            type: 'requestQueue',
+            participant: this,
+            requests: requests.map((request) => ({
+                url: request.url,
+                uniqueKey: request.uniqueKey,
+                label: request.label,
+            })),
+            forefront,
+            writeThrough,
+        });
+    }
+    /**
+     * The requests buffered by the given transaction for this queue, keyed by `uniqueKey` — a dedup
+     * index derived from the transaction journal.
+     */
+    bufferedRequests(transaction) {
+        const buffered = new Map();
+        // Only `deferred` records snapshots, so scanning the journal under `writeThrough` never finds any.
+        if (transaction.policy.requestQueue !== 'deferred')
+            return buffered;
+        for (const entry of transaction.journal) {
+            if (entry.type !== 'requestQueue' || entry.participant !== this)
+                continue;
+            for (const request of entry.requests) {
+                if (request.snapshot !== undefined)
+                    buffered.set(request.uniqueKey, request.snapshot);
+            }
+        }
+        return buffered;
+    }
+    /**
+     * Adds a request under the `deferred` policy: journaled now, really added by the commit replay.
+     * A new request's `requestId` is the local `uniqueKey` hash and is **provisional** — never write it
+     * to `request.id` or the dedup caches. Dedup is cheapest-first: buffer, caches, then a backend probe.
+     */
+    async addRequestDeferred(transaction, request, forefront, buffered = this.bufferedRequests(transaction)) {
+        // This transaction's own buffered adds; the shared caches never see them (provisional ids).
+        if (buffered.has(request.uniqueKey)) {
+            this.recordRequestJournalEntry(transaction, [request], forefront, false);
+            return {
+                wasAlreadyPresent: true,
+                wasAlreadyHandled: false,
+                requestId: getRequestId(request.uniqueKey),
+                uniqueKey: request.uniqueKey,
+                forefront,
+            };
+        }
+        // The caches hold real backend ids. Only *writing* provisional ids to them would be wrong;
+        // reading saves a probe. Same lookup as the write-through path.
+        const cacheKey = getRequestId(request.uniqueKey);
+        const cachedInfo = this.requestCache.get(cacheKey);
+        const knownRequestId = cachedInfo?.id ?? this.#requestSeenCache.get(cacheKey);
+        if (knownRequestId) {
+            this.recordRequestJournalEntry(transaction, [request], forefront, false);
+            return {
+                wasAlreadyPresent: true,
+                // The dedup cache doesn't track the handled state; only the full record does.
+                wasAlreadyHandled: cachedInfo?.isHandled ?? false,
+                requestId: knownRequestId,
+                uniqueKey: request.uniqueKey,
+                forefront,
+            };
+        }
+        // The caches are bounded, so a miss is not proof of absence - probe for an accurate answer.
+        const existing = await this.backend.getRequest(request.uniqueKey);
+        if (existing) {
+            this.recordRequestJournalEntry(transaction, [request], forefront, false);
+            return {
+                wasAlreadyPresent: true,
+                wasAlreadyHandled: existing.handledAt != null,
+                requestId: existing.id,
+                uniqueKey: request.uniqueKey,
+                forefront,
+            };
+        }
+        // The entry below *is* the write, so a transaction closed during the probe must not receive it -
+        // pass through instead, per the closed-transaction rule. Under `deferred` that can land an
+        // addition a rollback would have discarded; dedup bounds that cost, silent loss is unbounded.
+        if (!transaction.isActive) {
+            return await this.addRequest(request, { forefront });
+        }
+        const snapshot = JSON.parse(JSON.stringify(request));
+        // Strip-list, not allow-list: every user-facing field flows through, including ones added to
+        // `Request` in the future. The exceptions are `id` and `handledAt`, the two backend-owned
+        // lifecycle fields.
+        delete snapshot.id;
+        delete snapshot.handledAt;
+        transaction.recordJournalEntry({
+            type: 'requestQueue',
+            participant: this,
+            requests: [{ url: request.url, uniqueKey: request.uniqueKey, label: request.label, snapshot }],
+            forefront,
+            writeThrough: false,
+        });
+        buffered.set(request.uniqueKey, snapshot);
+        return {
+            wasAlreadyPresent: false,
+            wasAlreadyHandled: false,
+            requestId: getRequestId(request.uniqueKey),
+            uniqueKey: request.uniqueKey,
+            forefront,
+        };
+    }
+    /** @internal */
+    async commitJournalEntries(entries) {
+        // Replay through `backend.addBatchOfRequests`, *not* the batched frontend wrapper - the wrapper
+        // resolves after the first chunk and sleeps between the rest, neither of which commit may
+        // inherit. One call per `forefront` flag; the order of forefront additions is arbitrary anyway.
+        for (const forefront of [false, true]) {
+            const requests = entries.flatMap((entry) => entry.type === 'requestQueue' && entry.forefront === forefront
+                ? // Requests without a snapshot were deduplicated or written through; nothing to replay.
+                    entry.requests
+                        .filter((journaled) => journaled.snapshot !== undefined)
+                        .map((journaled) => new Request(journaled.snapshot))
+                : []);
+            if (requests.length === 0)
+                continue;
+            this.#statsTracker.add('writeCount');
+            const { processedRequests, unprocessedRequests } = await this.backend.addBatchOfRequests(requests, {
+                forefront,
+            });
+            // Only now, with the real backend-assigned ids, may the shared dedup caches be populated.
+            for (const processed of processedRequests) {
+                const cacheKey = getRequestId(processed.uniqueKey);
+                this.cacheRequest(cacheKey, { ...processed, forefront });
+                this.#requestSeenCache.add(cacheKey, processed.requestId);
+            }
+            if (unprocessedRequests.length > 0) {
+                // Warn and skip, rather than retry or fail. `unprocessedRequests` is what remains after
+                // the backend's own transient-error handling - a semantic rejection that retrying here
+                // would only re-poke. And failing the commit would let one malformed request hold the
+                // whole transaction hostage.
+                this.log.warning('Some requests were rejected by the request queue while committing a storage transaction and will be skipped. ' +
+                    "This usually means the request data is malformed (e.g. an invalid 'userData' shape).", { unprocessedRequests });
+            }
+        }
+    }
     /**
      * Adds requests to the queue in batches of 25. This method will wait till all the requests are added
      * to the queue before resolving. You should prefer using `queue.addRequestsBatched()` or `crawler.addRequests()`
@@ -190,7 +343,7 @@ export class RequestQueue {
      * @param [options] Request queue operation options.
      */
     async addRequests(requestsLike, options = {}) {
-        checkStorageAccess();
+        const transaction = activeStorageTransaction();
         ow(requestsLike, ow.object
             .is((value) => isIterable(value) || isAsyncIterable(value))
             .message((value) => `Expected an iterable or async iterable, got ${getObjectType(value)}`));
@@ -225,12 +378,20 @@ export class RequestQueue {
                 requests.push(requestLike instanceof Request ? requestLike : new Request(requestLike));
             }
         }
+        if (transaction?.policy.requestQueue === 'deferred') {
+            const buffered = this.bufferedRequests(transaction);
+            for (const request of requests) {
+                results.processedRequests.push(await this.addRequestDeferred(transaction, request, forefront, buffered));
+            }
+            return results;
+        }
+        this.recordRequestJournalEntry(transaction, requests, forefront, true);
         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);
-            const knownRequestId = cachedInfo?.id ?? this.requestSeenCache.get(cacheKey);
+            const knownRequestId = cachedInfo?.id ?? this.#requestSeenCache.get(cacheKey);
             if (knownRequestId) {
                 request.id = knownRequestId;
                 results.processedRequests.push({
@@ -249,7 +410,7 @@ export class RequestQueue {
         if (!requestsToAdd.size) {
             return results;
         }
-        this.statsTracker.add('writeCount');
+        this.#statsTracker.add('writeCount');
         const apiResults = await this.backend.addBatchOfRequests([...requestsToAdd.values()], { forefront });
         // Report unprocessed requests
         results.unprocessedRequests = apiResults.unprocessedRequests;
@@ -262,7 +423,7 @@ export class RequestQueue {
                 this.cacheRequest(cacheKey, { ...newRequest, forefront });
             }
             // Unlike `requestCache`, populate this on every batch (including background ones).
-            this.requestSeenCache.add(cacheKey, newRequest.requestId);
+            this.#requestSeenCache.add(cacheKey, newRequest.requestId);
         }
         return results;
     }
@@ -276,7 +437,8 @@ export class RequestQueue {
      * @param options Options for the request queue
      */
     async addRequestsBatched(requests, options = {}) {
-        checkStorageAccess();
+        const transaction = activeStorageTransaction();
+        const deferred = transaction?.policy.requestQueue === 'deferred';
         ow(requests, ow.object
             .is((value) => isIterable(value) || isAsyncIterable(value))
             .message((value) => `Expected an iterable or async iterable, got ${getObjectType(value)}`));
@@ -313,7 +475,9 @@ export class RequestQueue {
                 }
             }
         }
-        const { batchSize = 1000, waitBetweenBatchesMillis = 1000, maxNewRequests = undefined } = options;
+        const { batchSize = 1000, maxNewRequests = undefined } = options;
+        // Under `deferred` no chunk performs backend I/O, so pacing them would only stall the handler.
+        const waitBetweenBatchesMillis = deferred ? 0 : (options.waitBetweenBatchesMillis ?? 1000);
         let remainingBudget = maxNewRequests ?? Infinity;
         const requestsOverLimit = [];
         // If there's a limit on the number of added requests, do not send batches bigger than the limit
@@ -373,21 +537,38 @@ export class RequestQueue {
         if ((await chunksIterator.peek()) === undefined) {
             return buildResult(addedRequests, Promise.resolve([]), requestIterator);
         }
-        // eslint-disable-next-line no-async-promise-executor
-        const promise = new Promise(async (resolve) => {
+        const processRemainingChunks = async () => {
             const finalAddedRequests = [];
             for await (const requestChunk of chunks) {
                 finalAddedRequests.push(...(await processChunk(requestChunk, false)));
                 await sleep(waitBetweenBatchesMillis);
             }
-            resolve(finalAddedRequests);
+            return finalAddedRequests;
+        };
+        // maxNewRequests needs all batches to report skipped requests accurately; `deferred` needs them
+        // too - a writer that finishes after commit would have nowhere to put its journal entries.
+        const awaitsRemainingChunks = options.waitForAllRequestsToBeAdded || maxNewRequests !== undefined || deferred;
+        // eslint-disable-next-line no-async-promise-executor
+        const promise = new Promise(async (resolve) => {
+            if (awaitsRemainingChunks) {
+                // Awaited below, i.e. still within the caller's transaction scope, so the additions are
+                // journaled like the initial chunk - introspection must not depend on where the chunk
+                // boundary happened to fall.
+                resolve(await processRemainingChunks());
+            }
+            else {
+                // Nobody awaits this writer, so it outlives the transaction scope it inherits and must
+                // not record into a transaction that may already be closed. It writes directly - its
+                // write-through additions were never going to be rolled back anyway - which means the
+                // requests it adds are not journaled. See `StorageTransactionView.enqueuedUrls`.
+                resolve(await withDirectStorageAccess(processRemainingChunks));
+            }
         });
         this.inProgressRequestBatchCount += 1;
         void promise.finally(() => {
             this.inProgressRequestBatchCount -= 1;
         });
-        // When maxNewRequests is set, we must wait for all batches so we can accurately report skipped requests.
-        if (options.waitForAllRequestsToBeAdded || maxNewRequests !== undefined) {
+        if (awaitsRemainingChunks) {
             addedRequests.push(...(await promise));
         }
         return buildResult(addedRequests, promise, requestIterator);
@@ -399,8 +580,13 @@ export class RequestQueue {
      * @returns Returns the request object, or `null` if it was not found.
      */
     async getRequest(uniqueKey) {
-        checkStorageAccess();
+        const transaction = activeStorageTransaction();
         ow(uniqueKey, ow.string);
+        // Requests buffered by the active transaction (under the `deferred` write policy) are visible to it.
+        const buffered = transaction && this.bufferedRequests(transaction).get(uniqueKey);
+        if (buffered) {
+            return new Request(buffered);
+        }
         const requestOptions = await this.backend.getRequest(uniqueKey);
         if (!requestOptions)
             return null;
@@ -424,11 +610,11 @@ export class RequestQueue {
      *   Returns the request object or `null` if there are no more pending requests.
      */
     async fetchNextRequest() {
-        checkStorageAccess();
-        if (this.queuePausedForMigration) {
+        rejectOperationInTransaction('RequestQueue.fetchNextRequest()', 'it is part of the crawler request-processing bookkeeping, which a transaction must not affect.');
+        if (this.#queuePausedForMigration) {
             return null;
         }
-        this.statsTracker.add('headItemReadCount');
+        this.#statsTracker.add('headItemReadCount');
         const requestOptions = await this.backend.fetchNextRequest();
         if (!requestOptions)
             return null;
@@ -441,7 +627,7 @@ export class RequestQueue {
      * Handled requests will never again be returned by the `fetchNextRequest` function.
      */
     async markRequestAsHandled(request) {
-        checkStorageAccess();
+        rejectOperationInTransaction('RequestQueue.markRequestAsHandled()', 'it is part of the crawler request-processing bookkeeping, which a transaction must not affect.');
         ow(request, ow.object.partialShape({
             id: ow.string,
             uniqueKey: ow.string,
@@ -449,7 +635,7 @@ export class RequestQueue {
         }));
         const forefront = this.requestCache.get(getRequestId(request.uniqueKey))?.forefront ?? false;
         const handledAt = request.handledAt ?? new Date().toISOString();
-        this.statsTracker.add('writeCount');
+        this.#statsTracker.add('writeCount');
         const processedRequest = await this.backend.markRequestAsHandled({
             ...request,
             handledAt,
@@ -474,7 +660,7 @@ export class RequestQueue {
      * For example, this lets you store the number of retries or error messages for the request.
      */
     async reclaimRequest(request, options = {}) {
-        checkStorageAccess();
+        rejectOperationInTransaction('RequestQueue.reclaimRequest()', 'it is part of the crawler request-processing bookkeeping, which a transaction must not affect.');
         ow(request, ow.object.partialShape({
             id: ow.string,
             uniqueKey: ow.string,
@@ -483,7 +669,7 @@ export class RequestQueue {
             forefront: ow.optional.boolean,
         }));
         const { forefront = false } = options;
-        this.statsTracker.add('writeCount');
+        this.#statsTracker.add('writeCount');
         const processedRequest = await this.backend.reclaimRequest(request, { forefront });
         // The request was not in progress — nothing to reclaim.
         if (!processedRequest) {
@@ -508,7 +694,11 @@ export class RequestQueue {
      * {@link RequestQueue.isFinished}.
      */
     async isEmpty() {
-        checkStorageAccess();
+        const transaction = activeStorageTransaction();
+        // Requests buffered by the active transaction count as pending from its point of view.
+        if (transaction && this.bufferedRequests(transaction).size > 0) {
+            return false;
+        }
         return this.backend.isEmpty();
     }
     /**
@@ -520,11 +710,15 @@ export class RequestQueue {
      * a false negative, but it shall never return a false positive.
      */
     async isFinished() {
-        checkStorageAccess();
+        const transaction = activeStorageTransaction();
         // We are not finished if we're still adding new requests in the background.
         if (this.inProgressRequestBatchCount > 0) {
             return false;
         }
+        // Requests buffered by the active transaction count as pending from its point of view.
+        if (transaction && this.bufferedRequests(transaction).size > 0) {
+            return false;
+        }
         return this.backend.isFinished();
     }
     /**
@@ -538,10 +732,10 @@ export class RequestQueue {
      * short the reservation of a long-lived one and have its in-flight request stolen.
      */
     async setExpectedRequestProcessingTimeSecs(secs) {
-        if (secs <= this.expectedRequestProcessingSecs) {
+        if (secs <= this.#expectedRequestProcessingSecs) {
             return;
         }
-        this.expectedRequestProcessingSecs = secs;
+        this.#expectedRequestProcessingSecs = secs;
         await this.backend.setExpectedRequestProcessingTimeSecs?.(secs);
     }
     /**
@@ -564,7 +758,7 @@ export class RequestQueue {
      * depending on the mode of operation.
      */
     async drop() {
-        checkStorageAccess();
+        rejectOperationInTransaction('RequestQueue.drop()');
         await this.backend.drop();
         serviceLocator.getStorageInstanceManager().removeFromCache(this);
     }
@@ -573,16 +767,16 @@ export class RequestQueue {
      * so it can be reused (e.g. across multiple `crawler.run()` calls).
      */
     async purge() {
-        checkStorageAccess();
+        rejectOperationInTransaction('RequestQueue.purge()');
         await this.backend.purge();
         // Reset in-memory bookkeeping so the queue behaves as if freshly opened.
         this.requestCache.clear();
-        this.requestSeenCache.clear();
+        this.#requestSeenCache.clear();
         this.inProgressRequestBatchCount = 0;
         // Reset the expected-processing-time high-water mark too, otherwise the monotonic-raise guard
         // in `setExpectedRequestProcessingTimeSecs` would let a value raised in an earlier run leak into a
         // later one and silently swallow a lower hint (the queue is meant to be reusable across runs).
-        this.expectedRequestProcessingSecs = 0;
+        this.#expectedRequestProcessingSecs = 0;
     }
     /**
      * @inheritdoc
@@ -630,8 +824,17 @@ export class RequestQueue {
      * @throws If the underlying storage no longer exists (e.g. it was deleted externally).
      */
     async getInfo() {
-        checkStorageAccess();
-        return this.backend.getMetadata();
+        const transaction = activeStorageTransaction();
+        const metadata = await this.backend.getMetadata();
+        const bufferedCount = transaction ? this.bufferedRequests(transaction).size : 0;
+        if (bufferedCount > 0) {
+            return {
+                ...metadata,
+                totalRequestCount: metadata.totalRequestCount + bufferedCount,
+                pendingRequestCount: metadata.pendingRequestCount + bufferedCount,
+            };
+        }
+        return metadata;
     }
     /**
      * Fetches URLs from requestsFromUrl and returns them in format of list of requests
@@ -641,10 +844,10 @@ export class RequestQueue {
         // Download remote resource and parse URLs.
         let urlsArr;
         try {
-            urlsArr = await this._downloadListOfUrls({
+            urlsArr = await this.downloadListOfUrls({
                 url: requestsFromUrl,
                 urlRegExp: regex,
-                proxyUrl: (await this.proxyConfiguration?.newProxyInfo())?.url,
+                proxyUrl: (await this.#proxyConfiguration?.newProxyInfo())?.url,
             });
         }
         catch (err) {
@@ -676,10 +879,10 @@ export class RequestQueue {
     /**
      * @internal wraps public utility for mocking purposes
      */
-    async _downloadListOfUrls(options) {
+    async downloadListOfUrls(options) {
         return downloadListOfUrls({
             ...options,
-            httpClient: this.httpClient,
+            httpClient: this.#httpClient,
         });
     }
     /**
@@ -700,7 +903,7 @@ export class RequestQueue {
      * @param [options] Open Request Queue options.
      */
     static async open(identifier, options = {}) {
-        checkStorageAccess();
+        tryCancel();
         ow(options, ow.object.exactShape({
             configuration: ow.optional.object.instanceOf(Configuration),
             storageBackend: ow.optional.object,
@@ -718,8 +921,8 @@ export class RequestQueue {
             backendOpener: () => storageBackend.createRequestQueueBackend(resolved),
             backendCacheKey: storageBackend.getStorageBackendCacheKey?.() ?? storageBackend.constructor.name,
         });
-        queue.proxyConfiguration = options.proxyConfiguration;
-        queue.httpClient = options.httpClient;
+        queue.#proxyConfiguration = options.proxyConfiguration;
+        queue.#httpClient = options.httpClient;
         return queue;
     }
 }

package/storages/sitemap_request_loader.d.ts

@@ -80,55 +80,12 @@ export interface SitemapRequestLoaderOptions extends UrlConstraints {
  * The loading of the sitemap is performed in the background so that crawling can start before the sitemap is fully loaded.
  */
 export declare class SitemapRequestLoader implements IRequestLoader {
+    #private;
     /**
      * Set of URLs that were returned by `fetchNextRequest()` and not marked as handled yet.
      * @internal
      */
     inProgress: Set<string>;
-    /**
-     * Map of returned Request objects that have not been marked as handled yet.
-     *
-     * We use this to persist custom user fields on the in-progress requests.
-     */
-    private requestData;
-    /**
-     * Object for keeping track of the sitemap parsing progress.
-     */
-    private sitemapParsingProgress;
-    /**
-     * Object stream of URLs parsed from the sitemaps.
-     * Using `highWaterMark`, this can manage the speed of the sitemap loading.
-     *
-     * Fetch the next URL to be processed using `fetchNextRequest()`.
-     */
-    private urlQueueStream;
-    /**
-     * Indicates whether the request list sitemap loading was aborted.
-     *
-     * If the loading was aborted before the sitemaps were fully loaded, the request list might be missing some URLs.
-     * The `isSitemapFullyLoaded` method can be used to check if the sitemaps were fully loaded.
-     *
-     * If the loading is aborted and all the requests are handled, `isFinished()` will return `true`.
-     */
-    private abortLoading;
-    /** Number of URLs that were marked as handled */
-    private handledUrlCount;
-    private persistStateKey?;
-    private store?;
-    private closed;
-    /**
-     * Proxy URL to be used for sitemap loading.
-     */
-    private proxyUrl?;
-    /**
-     * Logger instance.
-     */
-    private log;
-    private urlExcludePatternObjects;
-    private urlPatternObjects;
-    /** EventManager used to handle persistence */
-    private events;
-    private persistenceOptions;
     /** @internal */
     private constructor();
     /**