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

package/storages/sitemap_request_loader.js

@@ -25,11 +25,11 @@ 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 +49,7 @@ export class SitemapRequestLoader {
      *
      * Fetch the next URL to be processed using `fetchNextRequest()`.
      */
-    urlQueueStream;
+    #urlQueueStream;
     /**
      * Indicates whether the request list sitemap loading was aborted.
      *
@@ -58,25 +58,25 @@ 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;
     /**
      * Logger instance.
      */
-    log;
-    urlExcludePatternObjects = [];
-    urlPatternObjects = [];
+    #log;
+    #urlExcludePatternObjects = [];
+    #urlPatternObjects = [];
     /** EventManager used to handle persistence */
-    events;
-    persistenceOptions;
+    #events;
+    #persistenceOptions;
     /** @internal */
     constructor(options) {
         const urlPatternValidator = ow.any(ow.string, ow.regExp, ow.object.hasKeys('glob'), ow.object.hasKeys('regexp'));
@@ -94,19 +94,19 @@ export class SitemapRequestLoader {
             httpClient: ow.optional.object,
         }));
         const { include, exclude } = options;
-        this.log = serviceLocator.getLogger().child({ prefix: 'SitemapRequestLoader' });
+        this.#log = serviceLocator.getLogger().child({ prefix: 'SitemapRequestLoader' });
         if (exclude?.length) {
-            this.urlExcludePatternObjects.push(...constructUrlPatternObjects(exclude));
+            this.#urlExcludePatternObjects.push(...constructUrlPatternObjects(exclude));
         }
         if (include?.length) {
-            this.urlPatternObjects.push(...constructUrlPatternObjects(include));
+            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 = 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.persistState = this.persistState.bind(this);
     }
     /**
@@ -139,8 +139,8 @@ export class SitemapRequestLoader {
      * @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.
@@ -149,13 +149,13 @@ 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();
                 });
             }
@@ -172,21 +172,21 @@ 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');
         });
     }
     /**
@@ -195,8 +195,8 @@ 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.
@@ -204,34 +204,34 @@ 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,
                 })) {
                     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.
@@ -250,15 +250,15 @@ 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("persistState" /* 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;
@@ -269,14 +269,14 @@ 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
@@ -293,31 +293,31 @@ 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;
             }
@@ -325,48 +325,48 @@ 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 newStream = this.createNewStream(this.#urlQueueStream.readableHighWaterMark);
         for (const url of urlQueue) {
             newStream.push(url);
         }
-        if (this.urlQueueStream.writableEnded) {
+        if (this.#urlQueueStream.writableEnded) {
             newStream.end();
         }
-        this.urlQueueStream = newStream;
-        await this.store.setValue(this.persistStateKey, {
+        this.#urlQueueStream = newStream;
+        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
@@ -377,11 +377,11 @@ 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 }));
         }
         this.inProgress.add(nextUrl);
-        return this.requestData.get(nextUrl);
+        return this.#requestData.get(nextUrl);
     }
     /**
      * @inheritDoc
@@ -400,20 +400,20 @@ 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("persistState" /* 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.
      *

package/storages/storage_instance_manager.js

@@ -120,8 +120,8 @@ class StorageCache {
  * assigns a reserved default alias.
  */
 export class StorageInstanceManager {
-    cache = new StorageCache();
-    openerLocks = new Map();
+    #cache = new StorageCache();
+    #openerLocks = new Map();
     /**
      * Open (or retrieve from cache) a storage instance.
      *
@@ -142,46 +142,46 @@ export class StorageInstanceManager {
         }
         // Fast-path cache check (no lock).
         if (alias !== undefined) {
-            const cached = this.cache.get(cls, { alias, backendCacheKey });
+            const cached = this.#cache.get(cls, { alias, backendCacheKey });
             if (cached)
                 return cached;
         }
         else if (id) {
-            const cached = this.cache.get(cls, { id, backendCacheKey });
+            const cached = this.#cache.get(cls, { id, backendCacheKey });
             if (cached)
                 return cached;
         }
         else if (name) {
-            const cached = this.cache.get(cls, { name, backendCacheKey });
+            const cached = this.#cache.get(cls, { name, backendCacheKey });
             if (cached)
                 return cached;
         }
         const identifierKey = id ?? name ?? alias ?? DEFAULT_STORAGE_ALIAS;
         const lockKey = `${cls.name}:${identifierKey}:${backendCacheKey}`;
-        if (!this.openerLocks.has(lockKey)) {
-            this.openerLocks.set(lockKey, new AsyncQueue());
+        if (!this.#openerLocks.has(lockKey)) {
+            this.#openerLocks.set(lockKey, new AsyncQueue());
         }
-        const queue = this.openerLocks.get(lockKey);
+        const queue = this.#openerLocks.get(lockKey);
         await queue.wait();
         try {
             // Double-check cache under lock (another caller may have filled it while we waited).
             if (alias !== undefined) {
-                const cached = this.cache.get(cls, { alias, backendCacheKey });
+                const cached = this.#cache.get(cls, { alias, backendCacheKey });
                 if (cached)
                     return cached;
             }
             else if (id) {
-                const cached = this.cache.get(cls, { id, backendCacheKey });
+                const cached = this.#cache.get(cls, { id, backendCacheKey });
                 if (cached)
                     return cached;
             }
             else if (name) {
-                const cached = this.cache.get(cls, { name, backendCacheKey });
+                const cached = this.#cache.get(cls, { name, backendCacheKey });
                 if (cached)
                     return cached;
             }
             // Prevent the same string from being used as both a name and an alias.
-            this.cache.checkNameAliasConflict(cls, { name, alias, backendCacheKey });
+            this.#cache.checkNameAliasConflict(cls, { name, alias, backendCacheKey });
             // Cache miss — create the sub-backend and storage instance.
             const subBackend = await backendOpener();
             const storageInfo = await subBackend.getMetadata();
@@ -189,7 +189,7 @@ export class StorageInstanceManager {
             // we just fetched (so `id`/`name` etc. are available synchronously) along with the backend.
             const instance = new cls({ metadata: storageInfo, backend: subBackend });
             // Atomic cache writes (no awaits between these).
-            this.cache.set(cls, instance, backendCacheKey, alias);
+            this.#cache.set(cls, instance, backendCacheKey, alias);
             return instance;
         }
         finally {
@@ -197,7 +197,7 @@ export class StorageInstanceManager {
             // Clean up idle locks so the map doesn't grow unboundedly
             // (mirrors crawlee-python's WeakValueDictionary behaviour).
             if (queue.remaining === 0) {
-                this.openerLocks.delete(lockKey);
+                this.#openerLocks.delete(lockKey);
             }
         }
     }
@@ -205,7 +205,7 @@ export class StorageInstanceManager {
      * Remove a storage instance from the cache (called from `storage.drop()`).
      */
     removeFromCache(instance) {
-        this.cache.removeFromCache(instance);
+        this.#cache.removeFromCache(instance);
     }
     /**
      * Clear the entire cache. Also calls `clearCache()` on any cached KeyValueStore
@@ -213,12 +213,12 @@ export class StorageInstanceManager {
      * Called during service locator reset.
      */
     clearCache() {
-        for (const instance of this.cache.allValues()) {
+        for (const instance of this.#cache.allValues()) {
             if ('clearCache' in instance && typeof instance.clearCache === 'function') {
                 instance.clearCache();
             }
         }
-        this.cache.clear();
+        this.#cache.clear();
     }
 }
 /**

package/storages/storage_stats.d.ts

@@ -39,7 +39,7 @@ export interface RequestQueueStats {
  * the buckets that make sense for it.
  */
 export declare class StorageStatsTracker<T extends Record<keyof T, number>> {
-    private readonly counters;
+    #private;
     constructor(initial: T);
     /** Increment a counter bucket by `by` (default `1`). */
     add(key: keyof T, by?: number): void;

package/storages/storage_stats.js

@@ -14,16 +14,16 @@
  * the buckets that make sense for it.
  */
 export class StorageStatsTracker {
-    counters;
+    #counters;
     constructor(initial) {
-        this.counters = { ...initial };
+        this.#counters = { ...initial };
     }
     /** Increment a counter bucket by `by` (default `1`). */
     add(key, by = 1) {
-        this.counters[key] += by;
+        this.#counters[key] += by;
     }
     /** Return a snapshot of the current counters. The returned object is a copy and safe to keep. */
     get current() {
-        return { ...this.counters };
+        return { ...this.#counters };
     }
 }