@crawlee/core 4.0.0-beta.99 → 4.0.0-rc.0

package/storages/sitemap_request_loader.d.ts

@@ -1,43 +1,31 @@
-import type { BaseHttpClient } from '@crawlee/types';
-import { type ParseSitemapOptions } from '@crawlee/utils';
-import type { GlobInput, RegExpInput } from '../enqueue_links/shared.js';
+import type { BaseHttpClient } from '@crawlee/http-client';
+import { EnqueueStrategy, type ParseSitemapOptions } from '@crawlee/utils';
+import type { UrlPatternInput } from '../enqueue_links/shared.js';
 import { Request } from '../request.js';
 import type { IRequestLoader } from './request_loader.js';
 import type { IRequestManager } from './request_manager.js';
 interface UrlConstraints {
     /**
-     * An array of glob pattern strings or plain objects
-     * containing glob pattern strings matching the URLs to be enqueued.
+     * An array of URL patterns that URLs must match to be included.
      *
-     * The plain objects must include at least the `glob` property, which holds the glob pattern string.
+     * Accepts glob pattern strings, `{ glob: string }` objects, `RegExp` instances, or `{ regexp: RegExp }` objects.
      *
-     * The matching is always case-insensitive.
-     * If you need case-sensitive matching, use `regexps` property directly.
+     * Glob matching is always case-insensitive.
+     * If you need case-sensitive matching, use a `RegExp`.
      *
-     * If `globs` is an empty array or `undefined`, and `regexps` are also not defined, then the `SitemapRequestLoader`
+     * If `include` is an empty array or `undefined`, then the `SitemapRequestLoader`
      * includes all the URLs from the sitemap.
      */
-    globs?: readonly GlobInput[];
+    include?: readonly UrlPatternInput[];
     /**
-     * An array of glob pattern strings, regexp patterns or plain objects
-     * containing patterns matching URLs that will **never** be included.
+     * An array of URL patterns. Matching URLs will **not** be included.
      *
-     * The plain objects must include either the `glob` property or the `regexp` property.
+     * Accepts glob pattern strings, `{ glob: string }` objects, `RegExp` instances, or `{ regexp: RegExp }` objects.
      *
      * Glob matching is always case-insensitive.
-     * If you need case-sensitive matching, provide a regexp.
+     * If you need case-sensitive matching, use a `RegExp`.
      */
-    exclude?: readonly (GlobInput | RegExp)[];
-    /**
-     * An array of regular expressions or plain objects
-     * containing regular expressions matching the URLs to be enqueued.
-     *
-     * The plain objects must include at least the `regexp` property, which holds the regular expression.
-     *
-     * If `regexps` is an empty array or `undefined`, and `globs` are also not defined, then the `SitemapRequestLoader`
-     * includes all the URLs from the sitemap.
-     */
-    regexps?: readonly RegExpInput[];
+    exclude?: readonly UrlPatternInput[];
 }
 export interface SitemapRequestLoaderOptions extends UrlConstraints {
     /**
@@ -77,6 +65,13 @@ export interface SitemapRequestLoaderOptions extends UrlConstraints {
      * @default 200
      */
     maxBufferSize?: number;
+    /**
+     * Keep only sitemap-derived URLs matching this strategy relative to the parent sitemap URL; non-`http(s)`
+     * schemes are always dropped. The filtering stays enforced after navigation (e.g. across redirects).
+     * Pass `'all'` to disable host filtering.
+     * @default EnqueueStrategy.SameHostname
+     */
+    enqueueStrategy?: EnqueueStrategy | `${EnqueueStrategy}`;
     /**
      * Advanced options for the underlying `parseSitemap` call.
      */
@@ -92,55 +87,12 @@ 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();
     /**
@@ -156,7 +108,7 @@ export declare class SitemapRequestLoader implements IRequestLoader {
      */
     private matchesUrl;
     /**
-     * Checks whether the URL matches the `globs` / `regexps` / `exclude` provided in the `options`.
+     * Checks whether the URL matches the `include` / `exclude` patterns provided in the `options`.
      * @param url URL to be checked.
      * @returns `true` if the URL matches the patterns, `false` otherwise.
      */

package/storages/sitemap_request_loader.js

@@ -1,12 +1,28 @@
 import { Transform } from 'node:stream';
-import { parseSitemap } from '@crawlee/utils';
+import { EnqueueStrategy, parseSitemap } from '@crawlee/utils';
 import { minimatch } from 'minimatch';
-import ow from 'ow';
-import { constructGlobObjectsFromGlobs, constructRegExpObjectsFromRegExps } from '../enqueue_links/shared.js';
+import { z } from 'zod';
+import { constructUrlPatternObjects, urlPatternSchema } from '../enqueue_links/shared.js';
+import { EventType } from '../events/event_manager.js';
 import { Request } from '../request.js';
 import { serviceLocator } from '../service_locator.js';
+import { parseArgument, schemas } from '../validators.js';
 import { KeyValueStore } from './key_value_store.js';
 import { purgeDefaultStorages } from './utils.js';
+const sitemapRequestLoaderOptionsSchema = z.strictObject({
+    sitemapUrls: schemas.arrayOf(z.string(), 'strings'),
+    proxyUrl: z.string().optional(),
+    persistStateKey: z.string().optional(),
+    signal: z.unknown().optional(),
+    timeoutMillis: schemas.anyNumber.optional(),
+    maxBufferSize: schemas.anyNumber.default(200),
+    enqueueStrategy: z.enum(EnqueueStrategy).default(EnqueueStrategy.SameHostname),
+    parseSitemapOptions: z.looseObject({}).optional(),
+    include: schemas.arrayOf(urlPatternSchema, 'URL patterns').optional(),
+    exclude: schemas.arrayOf(urlPatternSchema, 'URL patterns').optional(),
+    persistenceOptions: z.looseObject({}).optional(),
+    httpClient: schemas.httpClient.optional(),
+});
 /** @internal */
 const STATE_PERSISTENCE_KEY = 'SITEMAP_REQUEST_LOADER_STATE';
 /**
@@ -25,11 +41,11 @@ export class SitemapRequestLoader {
      *
      * We use this to persist custom user fields on the in-progress requests.
      */
-    requestData = new Map();
+    #requestData = new Map();
     /**
      * Object for keeping track of the sitemap parsing progress.
      */
-    sitemapParsingProgress = {
+    #sitemapParsingProgress = {
         /**
          * URL of the sitemap that is currently being parsed. `null` if no sitemap is being parsed.
          */
@@ -49,7 +65,7 @@ export class SitemapRequestLoader {
      *
      * Fetch the next URL to be processed using `fetchNextRequest()`.
      */
-    urlQueueStream;
+    #urlQueueStream;
     /**
      * Indicates whether the request list sitemap loading was aborted.
      *
@@ -58,64 +74,46 @@ export class SitemapRequestLoader {
      *
      * If the loading is aborted and all the requests are handled, `isFinished()` will return `true`.
      */
-    abortLoading = false;
+    #abortLoading = false;
     /** Number of URLs that were marked as handled */
-    handledUrlCount = 0;
-    persistStateKey;
-    store;
-    closed = false;
+    #handledUrlCount = 0;
+    #persistStateKey;
+    #store;
+    #closed = false;
     /**
      * Proxy URL to be used for sitemap loading.
      */
-    proxyUrl;
+    #proxyUrl;
+    /**
+     * Enqueue strategy applied to sitemap-derived URLs and stamped onto the emitted `Request` objects.
+     */
+    #enqueueStrategy;
     /**
      * Logger instance.
      */
-    log;
-    urlExcludePatternObjects = [];
-    urlPatternObjects = [];
+    #log;
+    #urlExcludePatternObjects = [];
+    #urlPatternObjects = [];
     /** EventManager used to handle persistence */
-    events;
-    persistenceOptions;
+    #events;
+    #persistenceOptions;
     /** @internal */
     constructor(options) {
-        ow(options, ow.object.exactShape({
-            sitemapUrls: ow.array.ofType(ow.string),
-            proxyUrl: ow.optional.string,
-            persistStateKey: ow.optional.string,
-            signal: ow.optional.any(),
-            timeoutMillis: ow.optional.number,
-            maxBufferSize: ow.optional.number,
-            parseSitemapOptions: ow.optional.object,
-            globs: ow.optional.array.ofType(ow.any(ow.string, ow.object.hasKeys('glob'))),
-            exclude: ow.optional.array.ofType(ow.any(ow.string, ow.regExp, ow.object.hasKeys('glob'), ow.object.hasKeys('regexp'))),
-            regexps: ow.optional.array.ofType(ow.any(ow.regExp, ow.object.hasKeys('regexp'))),
-            persistenceOptions: ow.optional.object,
-        }));
-        const { globs, exclude, regexps } = options;
-        this.log = serviceLocator.getLogger().child({ prefix: 'SitemapRequestLoader' });
+        const { include, exclude, persistStateKey, persistenceOptions, proxyUrl, maxBufferSize, sitemapUrls, enqueueStrategy, } = parseArgument(options, sitemapRequestLoaderOptionsSchema, 'SitemapRequestLoaderOptions');
+        this.#log = serviceLocator.getLogger().child({ prefix: 'SitemapRequestLoader' });
         if (exclude?.length) {
-            for (const excl of exclude) {
-                if (typeof excl === 'string' || 'glob' in excl) {
-                    this.urlExcludePatternObjects.push(...constructGlobObjectsFromGlobs([excl]));
-                }
-                else if (excl instanceof RegExp || 'regexp' in excl) {
-                    this.urlExcludePatternObjects.push(...constructRegExpObjectsFromRegExps([excl]));
-                }
-            }
-        }
-        if (globs?.length) {
-            this.urlPatternObjects.push(...constructGlobObjectsFromGlobs(globs));
+            this.#urlExcludePatternObjects.push(...constructUrlPatternObjects(exclude));
         }
-        if (regexps?.length) {
-            this.urlPatternObjects.push(...constructRegExpObjectsFromRegExps(regexps));
+        if (include?.length) {
+            this.#urlPatternObjects.push(...constructUrlPatternObjects(include));
         }
-        this.persistStateKey = options.persistStateKey;
-        this.persistenceOptions = { enable: true, ...options.persistenceOptions };
-        this.proxyUrl = options.proxyUrl;
-        this.urlQueueStream = this.createNewStream(options.maxBufferSize ?? 200);
-        this.sitemapParsingProgress.pendingSitemapUrls = new Set(options.sitemapUrls);
-        this.events = serviceLocator.getEventManager();
+        this.#persistStateKey = persistStateKey;
+        this.#persistenceOptions = { enable: true, ...persistenceOptions };
+        this.#proxyUrl = proxyUrl;
+        this.#enqueueStrategy = enqueueStrategy;
+        this.#urlQueueStream = this.createNewStream(maxBufferSize);
+        this.#sitemapParsingProgress.pendingSitemapUrls = new Set(sitemapUrls);
+        this.#events = serviceLocator.getEventManager();
         this.persistState = this.persistState.bind(this);
     }
     /**
@@ -143,13 +141,13 @@ export class SitemapRequestLoader {
         };
     }
     /**
-     * Checks whether the URL matches the `globs` / `regexps` / `exclude` provided in the `options`.
+     * Checks whether the URL matches the `include` / `exclude` patterns provided in the `options`.
      * @param url URL to be checked.
      * @returns `true` if the URL matches the patterns, `false` otherwise.
      */
     isUrlMatchingPatterns(url) {
-        return (!this.urlExcludePatternObjects.some(this.matchesUrl(url)) &&
-            (this.urlPatternObjects.length === 0 || this.urlPatternObjects.some(this.matchesUrl(url))));
+        return (!this.#urlExcludePatternObjects.some(this.matchesUrl(url)) &&
+            (this.#urlPatternObjects.length === 0 || this.#urlPatternObjects.some(this.matchesUrl(url))));
     }
     /**
      * Adds a URL to the queue of parsed URLs.
@@ -158,13 +156,13 @@ export class SitemapRequestLoader {
      */
     async pushNextUrl(url) {
         return new Promise((resolve) => {
-            if (this.closed || (url && !this.isUrlMatchingPatterns(url))) {
+            if (this.#closed || (url && !this.isUrlMatchingPatterns(url))) {
                 resolve();
                 return;
             }
-            if (!this.urlQueueStream.push(url)) {
+            if (!this.#urlQueueStream.push(url)) {
                 // This doesn't work with the 'drain' event (it's not emitted for some reason).
-                this.urlQueueStream.once('readdata', () => {
+                this.#urlQueueStream.once('readdata', () => {
                     resolve();
                 });
             }
@@ -181,21 +179,21 @@ export class SitemapRequestLoader {
      */
     async readNextUrl() {
         return new Promise((resolve) => {
-            if (this.closed) {
+            if (this.#closed) {
                 resolve(null);
                 return;
             }
-            const result = this.urlQueueStream.read();
+            const result = this.#urlQueueStream.read();
             if (!result && !this.isSitemapFullyLoaded()) {
-                this.urlQueueStream.once('readable', () => {
-                    const nextUrl = this.urlQueueStream.read();
+                this.#urlQueueStream.once('readable', () => {
+                    const nextUrl = this.#urlQueueStream.read();
                     resolve(nextUrl);
                 });
             }
             else {
                 resolve(result);
             }
-            this.urlQueueStream.emit('readdata');
+            this.#urlQueueStream.emit('readdata');
         });
     }
     /**
@@ -204,8 +202,8 @@ export class SitemapRequestLoader {
      * If this is `false`, the background processing is either still in progress or was aborted.
      */
     isSitemapFullyLoaded() {
-        return (this.sitemapParsingProgress.inProgressSitemapUrl === null &&
-            this.sitemapParsingProgress.pendingSitemapUrls.size === 0);
+        return (this.#sitemapParsingProgress.inProgressSitemapUrl === null &&
+            this.#sitemapParsingProgress.pendingSitemapUrls.size === 0);
     }
     /**
      * Start processing the sitemaps and loading the URLs.
@@ -213,34 +211,35 @@ export class SitemapRequestLoader {
      * Resolves once all the sitemaps URLs have been fully loaded (sets `isSitemapFullyLoaded` to `true`).
      */
     async load({ parseSitemapOptions, }) {
-        while (!this.isSitemapFullyLoaded() && !this.abortLoading) {
-            const sitemapUrl = this.sitemapParsingProgress.inProgressSitemapUrl ??
-                this.sitemapParsingProgress.pendingSitemapUrls.values().next().value;
+        while (!this.isSitemapFullyLoaded() && !this.#abortLoading) {
+            const sitemapUrl = this.#sitemapParsingProgress.inProgressSitemapUrl ??
+                this.#sitemapParsingProgress.pendingSitemapUrls.values().next().value;
             try {
-                for await (const item of parseSitemap([{ type: 'url', url: sitemapUrl }], this.proxyUrl, {
+                for await (const item of parseSitemap([{ type: 'url', url: sitemapUrl }], this.#proxyUrl, {
                     ...parseSitemapOptions,
                     maxDepth: 0,
                     emitNestedSitemaps: true,
+                    enqueueStrategy: this.#enqueueStrategy,
                 })) {
                     if (!item.originSitemapUrl) {
                         // This is a nested sitemap
-                        this.sitemapParsingProgress.pendingSitemapUrls.add(item.loc);
+                        this.#sitemapParsingProgress.pendingSitemapUrls.add(item.loc);
                         continue;
                     }
-                    if (!this.sitemapParsingProgress.inProgressEntries.has(item.loc)) {
+                    if (!this.#sitemapParsingProgress.inProgressEntries.has(item.loc)) {
                         await this.pushNextUrl(item.loc);
-                        this.sitemapParsingProgress.inProgressEntries.add(item.loc);
+                        this.#sitemapParsingProgress.inProgressEntries.add(item.loc);
                     }
                 }
             }
             catch (e) {
-                this.log.error('Error loading sitemap contents:', e);
+                this.#log.error('Error loading sitemap contents:', e);
             }
-            this.sitemapParsingProgress.pendingSitemapUrls.delete(sitemapUrl);
-            this.sitemapParsingProgress.inProgressEntries.clear();
-            this.sitemapParsingProgress.inProgressSitemapUrl = null;
+            this.#sitemapParsingProgress.pendingSitemapUrls.delete(sitemapUrl);
+            this.#sitemapParsingProgress.inProgressEntries.clear();
+            this.#sitemapParsingProgress.inProgressSitemapUrl = null;
         }
-        this.urlQueueStream.end();
+        this.#urlQueueStream.end();
     }
     /**
      * Open a sitemap and start processing it.
@@ -259,15 +258,15 @@ export class SitemapRequestLoader {
         void requestList.load({
             parseSitemapOptions: { logger: serviceLocator.getLogger(), ...options.parseSitemapOptions, httpClient },
         });
-        if (requestList.persistenceOptions.enable) {
-            requestList.events.on("persistState" /* EventType.PERSIST_STATE */, requestList.persistState);
+        if (requestList.#persistenceOptions.enable) {
+            requestList.#events.on(EventType.PERSIST_STATE, requestList.persistState);
         }
         options?.signal?.addEventListener('abort', () => {
-            requestList.abortLoading = true;
+            requestList.#abortLoading = true;
         });
         if (options.timeoutMillis) {
             setTimeout(() => {
-                requestList.abortLoading = true;
+                requestList.#abortLoading = true;
             }, options.timeoutMillis);
         }
         return requestList;
@@ -278,14 +277,14 @@ export class SitemapRequestLoader {
     async getTotalCount() {
         // Total known so far = not-yet-fetched (still buffered in the stream) + in-progress (fetched but not
         // yet handled) + already handled.
-        return this.urlQueueStream.readableLength + this.inProgress.size + this.handledUrlCount;
+        return this.#urlQueueStream.readableLength + this.inProgress.size + this.#handledUrlCount;
     }
     /**
      * @inheritDoc
      */
     async getPendingCount() {
         // Pending = everything not yet handled = not-yet-fetched + in-progress.
-        return this.urlQueueStream.readableLength + this.inProgress.size;
+        return this.#urlQueueStream.readableLength + this.inProgress.size;
     }
     /**
      * Combines this list with a request manager (a {@link RequestQueue} by default) into a
@@ -302,31 +301,31 @@ export class SitemapRequestLoader {
      * @inheritDoc
      */
     async isFinished() {
-        return ((await this.isEmpty()) && this.inProgress.size === 0 && (this.isSitemapFullyLoaded() || this.abortLoading));
+        return ((await this.isEmpty()) && this.inProgress.size === 0 && (this.isSitemapFullyLoaded() || this.#abortLoading));
     }
     /**
      * @inheritDoc
      */
     async isEmpty() {
-        return this.urlQueueStream.readableLength === 0;
+        return this.#urlQueueStream.readableLength === 0;
     }
     /**
      * @inheritDoc
      */
     async getHandledCount() {
-        return this.handledUrlCount;
+        return this.#handledUrlCount;
     }
     /**
      * @inheritDoc
      */
     async persistState() {
-        if (this.persistStateKey === undefined) {
+        if (this.#persistStateKey === undefined) {
             return;
         }
-        this.store ??= await KeyValueStore.open();
+        this.#store ??= await KeyValueStore.open();
         const urlQueue = [];
-        while (this.urlQueueStream.readableLength > 0) {
-            const url = this.urlQueueStream.read();
+        while (this.#urlQueueStream.readableLength > 0) {
+            const url = this.#urlQueueStream.read();
             if (url === null) {
                 break;
             }
@@ -334,48 +333,55 @@ export class SitemapRequestLoader {
         }
         // Create a new stream, as we have read all the URLs from the current one.
         // Pushing the urls back to the original stream might not be possible if it has been ended.
-        const newStream = this.createNewStream(this.urlQueueStream.readableHighWaterMark);
+        const previousStream = this.#urlQueueStream;
+        const newStream = this.createNewStream(previousStream.readableHighWaterMark);
         for (const url of urlQueue) {
             newStream.push(url);
         }
-        if (this.urlQueueStream.writableEnded) {
+        if (previousStream.writableEnded) {
             newStream.end();
         }
-        this.urlQueueStream = newStream;
-        await this.store.setValue(this.persistStateKey, {
+        this.#urlQueueStream = newStream;
+        // A `pushNextUrl()` call may be blocked on backpressure, waiting for a `readdata` event on the
+        // previous stream. That event is only ever emitted by `readNextUrl()` on the current stream, so
+        // after the swap the waiter would never be notified and the background sitemap loading would hang.
+        // Re-emit `readdata` on the previous stream to release any such pending waiter (its URL has already
+        // been transferred to the new stream above).
+        previousStream.emit('readdata');
+        await this.#store.setValue(this.#persistStateKey, {
             sitemapParsingProgress: {
-                pendingSitemapUrls: Array.from(this.sitemapParsingProgress.pendingSitemapUrls),
-                inProgressSitemapUrl: this.sitemapParsingProgress.inProgressSitemapUrl,
-                inProgressEntries: Array.from(this.sitemapParsingProgress.inProgressEntries),
+                pendingSitemapUrls: Array.from(this.#sitemapParsingProgress.pendingSitemapUrls),
+                inProgressSitemapUrl: this.#sitemapParsingProgress.inProgressSitemapUrl,
+                inProgressEntries: Array.from(this.#sitemapParsingProgress.inProgressEntries),
             },
             // Re-queue in-progress requests to the front so they are retried if the state is restored.
             urlQueue: [...this.inProgress, ...urlQueue],
-            requestData: Array.from(this.requestData.entries()),
-            abortLoading: this.abortLoading,
-            closed: this.closed,
+            requestData: Array.from(this.#requestData.entries()),
+            abortLoading: this.#abortLoading,
+            closed: this.#closed,
         });
     }
     async restoreState() {
         await purgeDefaultStorages({ onlyPurgeOnce: true });
-        if (this.persistStateKey === undefined) {
+        if (this.#persistStateKey === undefined) {
             return;
         }
-        this.store ??= await KeyValueStore.open();
-        const state = await this.store.getValue(this.persistStateKey);
+        this.#store ??= await KeyValueStore.open();
+        const state = await this.#store.getValue(this.#persistStateKey);
         if (state === null) {
             return;
         }
-        this.sitemapParsingProgress = {
+        this.#sitemapParsingProgress = {
             pendingSitemapUrls: new Set(state.sitemapParsingProgress.pendingSitemapUrls),
             inProgressSitemapUrl: state.sitemapParsingProgress.inProgressSitemapUrl,
             inProgressEntries: new Set(state.sitemapParsingProgress.inProgressEntries),
         };
-        this.requestData = new Map(state.requestData ?? []);
+        this.#requestData = new Map(state.requestData ?? []);
         for (const url of state.urlQueue) {
-            this.urlQueueStream.push(url);
+            this.#urlQueueStream.push(url);
         }
-        this.abortLoading = state.abortLoading;
-        this.closed = state.closed;
+        this.#abortLoading = state.abortLoading;
+        this.#closed = state.closed;
     }
     /**
      * @inheritDoc
@@ -386,11 +392,11 @@ export class SitemapRequestLoader {
             return null;
         }
         // A restored in-progress request already has its Request data; don't overwrite it.
-        if (!this.requestData.has(nextUrl)) {
-            this.requestData.set(nextUrl, new Request({ url: nextUrl }));
+        if (!this.#requestData.has(nextUrl)) {
+            this.#requestData.set(nextUrl, new Request({ url: nextUrl, enqueueStrategy: this.#enqueueStrategy }));
         }
         this.inProgress.add(nextUrl);
-        return this.requestData.get(nextUrl);
+        return this.#requestData.get(nextUrl);
     }
     /**
      * @inheritDoc
@@ -409,20 +415,20 @@ export class SitemapRequestLoader {
      * Calling `fetchNextRequest()` after this method will always return `null`.
      */
     async teardown() {
-        this.closed = true;
-        this.abortLoading = true;
-        this.events.off("persistState" /* EventType.PERSIST_STATE */, this.persistState);
+        this.#closed = true;
+        this.#abortLoading = true;
+        this.#events.off(EventType.PERSIST_STATE, this.persistState);
         await this.persistState();
-        this.urlQueueStream.emit('readdata'); // unblocks the potentially waiting `pushNextUrl` call
+        this.#urlQueueStream.emit('readdata'); // unblocks the potentially waiting `pushNextUrl` call
     }
     /**
      * @inheritDoc
      */
     async markRequestAsHandled(request) {
-        this.handledUrlCount += 1;
+        this.#handledUrlCount += 1;
         this.ensureInProgress(request.url);
         this.inProgress.delete(request.url);
-        this.requestData.delete(request.url);
+        this.#requestData.delete(request.url);
     }
     ensureInProgress(url) {
         if (!this.inProgress.has(url)) {

package/storages/storage_instance_manager.d.ts

@@ -19,8 +19,7 @@ type Hashable = string;
  * assigns a reserved default alias.
  */
 export declare class StorageInstanceManager {
-    private readonly cache;
-    private readonly openerLocks;
+    #private;
     /**
      * Open (or retrieve from cache) a storage instance.
      *