@xberg-io/crawlberg-wasm 1.11.1 → 1.12.2

package/package.json

@@ -10,7 +10,7 @@
     }
   ],
   "name": "@xberg-io/crawlberg-wasm",
-  "version": "1.11.1",
+  "version": "1.12.2",
   "private": false,
   "description": "High-performance web crawling engine",
   "license": "MIT",
@@ -45,8 +45,8 @@
     "node": ">= 22"
   },
   "devDependencies": {
-    "@vitest/coverage-v8": "^5.0.4",
-    "vitest": "^5.0.4"
+    "@vitest/coverage-v8": "^5.0.5",
+    "vitest": "^5.0.5"
   },
   "scripts": {
     "build": "wasm-pack build --target nodejs --out-dir pkg/nodejs",

package/pkg/bundler/crawlberg_wasm.d.ts

@@ -260,7 +260,7 @@ export class WasmBrowserConfig {
      * keep a handle you already hold, build with `default()` and assign the property instead:
      * the generated setter borrows and leaves your handle alive.
      */
-    constructor(mode?: WasmBrowserMode | null, backend?: WasmBrowserBackend | null, timeout?: bigint | null, overallTimeout?: bigint | null, shutdownTimeout?: bigint | null, wait?: WasmBrowserWait | null, blockUrlPatterns?: string[] | null, captureNetworkEvents?: boolean | null, sessionAffinity?: boolean | null, chromeArgs?: string[] | null, endpoint?: string | null, waitSelector?: string | null, extraWait?: bigint | null, proxy?: WasmProxyConfig | null, evalScript?: string | null, robotsUserAgent?: string | null, chromePath?: string | null);
+    constructor(mode?: WasmBrowserMode | null, backend?: WasmBrowserBackend | null, timeout?: bigint | null, overallTimeout?: bigint | null, shutdownTimeout?: bigint | null, wait?: WasmBrowserWait | null, blockUrlPatterns?: string[] | null, captureNetworkEvents?: boolean | null, sessionAffinity?: boolean | null, chromeArgs?: string[] | null, endpoint?: string | null, waitSelector?: string | null, extraWait?: bigint | null, proxy?: WasmProxyConfig | null, evalScript?: string | null, robotsUserAgent?: string | null, chromePath?: string | null, scrollContainers?: boolean | null, domQuietWindow?: bigint | null, domQuietCap?: bigint | null);
     get backend(): string;
     set backend(value: WasmBrowserBackend);
     blockUrlPatterns: string[];
@@ -268,6 +268,10 @@ export class WasmBrowserConfig {
     chromeArgs: string[];
     get chromePath(): string | undefined;
     set chromePath(value: string | null | undefined);
+    get domQuietCap(): bigint | undefined;
+    set domQuietCap(value: bigint | null | undefined);
+    get domQuietWindow(): bigint | undefined;
+    set domQuietWindow(value: bigint | null | undefined);
     get endpoint(): string | undefined;
     set endpoint(value: string | null | undefined);
     get evalScript(): string | undefined;
@@ -288,6 +292,7 @@ export class WasmBrowserConfig {
     set proxy(value: WasmProxyConfig);
     get robotsUserAgent(): string | undefined;
     set robotsUserAgent(value: string | null | undefined);
+    scrollContainers: boolean;
     sessionAffinity: boolean;
     get shutdownTimeout(): bigint | undefined;
     set shutdownTimeout(value: bigint | null | undefined);
@@ -427,12 +432,16 @@ export class WasmContentConfig {
      */
     copyForTransfer(): WasmContentConfig;
     static default(): WasmContentConfig;
-    constructor(outputFormat?: string | null, preprocessingPreset?: string | null, removeNavigation?: boolean | null, removeForms?: boolean | null, stripTags?: string[] | null, preserveTags?: string[] | null, excludeSelectors?: string[] | null, skipImages?: boolean | null, wrap?: boolean | null, wrapWidth?: number | null, includeDocumentStructure?: boolean | null, extractMetadata?: boolean | null, maxDepth?: number | null, stripNewlines?: boolean | null, bullets?: string | null, listIndentWidth?: number | null, strongEmSymbol?: string | null, compactTables?: boolean | null, keepInlineImagesIn?: string[] | null, keepHidden?: boolean | null);
+    constructor(outputFormat?: string | null, preprocessingPreset?: string | null, removeNavigation?: boolean | null, removeForms?: boolean | null, stripTags?: string[] | null, preserveTags?: string[] | null, excludeSelectors?: string[] | null, skipImages?: boolean | null, wrap?: boolean | null, wrapWidth?: number | null, includeDocumentStructure?: boolean | null, extractMetadata?: boolean | null, maxDepth?: number | null, stripNewlines?: boolean | null, bullets?: string | null, listIndentWidth?: number | null, strongEmSymbol?: string | null, compactTables?: boolean | null, keepInlineImagesIn?: string[] | null, keepHidden?: boolean | null, expandAbbreviations?: boolean | null, escapeMisc?: boolean | null, escapeAsterisks?: boolean | null, includeIconLabels?: boolean | null);
     bullets: string;
     compactTables: boolean;
+    escapeAsterisks: boolean;
+    escapeMisc: boolean;
     excludeSelectors: string[];
+    expandAbbreviations: boolean;
     extractMetadata: boolean;
     includeDocumentStructure: boolean;
+    includeIconLabels: boolean;
     keepHidden: boolean;
     keepInlineImagesIn: string[];
     listIndentWidth: number;
@@ -506,7 +515,7 @@ export class WasmCrawlConfig {
      * keep a handle you already hold, build with `default()` and assign the property instead:
      * the generated setter borrows and leaves your handle alive.
      */
-    constructor(crawlStrategy?: WasmCrawlStrategyKind | null, respectRobotsTxt?: boolean | null, softHttpErrors?: boolean | null, stayOnDomain?: boolean | null, allowSubdomains?: boolean | null, includePaths?: string[] | null, excludePaths?: string[] | null, pathPatternsMatchQuery?: boolean | null, pathPatternsMatchUrl?: boolean | null, dedupIncludeQuery?: boolean | null, stripTrackingParams?: boolean | null, trackingParams?: string[] | null, customHeaders?: any | null, requestTimeout?: bigint | null, maxRedirects?: number | null, retryCount?: number | null, retryCodes?: Uint16Array | null, retryInitialDelayMs?: bigint | null, retryMaxDelayMs?: bigint | null, rateLimitJitterRatio?: number | null, cookiesEnabled?: boolean | null, removeTags?: string[] | null, content?: WasmContentConfig | null, downloadAssets?: boolean | null, assetTypes?: WasmAssetCategory[] | null, browser?: WasmBrowserConfig | null, userAgents?: string[] | null, captureScreenshot?: boolean | null, followDocumentUrls?: boolean | null, downloadDocuments?: boolean | null, documentMimeTypes?: string[] | null, saveBrowserProfile?: boolean | null, ssrf?: WasmSsrfPolicy | null, maxDepth?: number | null, maxPages?: number | null, maxLinksPerPage?: number | null, maxConcurrent?: number | null, contentFilter?: WasmContentFilterKind | null, bm25Query?: string | null, bm25Threshold?: number | null, userAgent?: string | null, rateLimitMs?: bigint | null, auth?: any | null, maxBodySize?: number | null, mapLimit?: number | null, mapSearch?: string | null, maxAssetSize?: number | null, proxy?: WasmProxyConfig | null, documentUrlDepth?: number | null, documentMaxSize?: number | null, documentOutputDir?: string | null, documentContentEncoding?: WasmDocumentContentEncoding | null, warcOutput?: string | null, browserProfile?: string | null, ssrfDenyPrivateExplicit?: boolean | null);
+    constructor(crawlStrategy?: WasmCrawlStrategyKind | null, respectRobotsTxt?: boolean | null, softHttpErrors?: boolean | null, stayOnDomain?: boolean | null, allowSubdomains?: boolean | null, includePaths?: string[] | null, excludePaths?: string[] | null, pathPatternsMatchQuery?: boolean | null, pathPatternsMatchUrl?: boolean | null, dedupIncludeQuery?: boolean | null, stripTrackingParams?: boolean | null, trackingParams?: string[] | null, customHeaders?: any | null, requestTimeout?: bigint | null, maxRedirects?: number | null, retryCount?: number | null, retryCodes?: Uint16Array | null, retryInitialDelayMs?: bigint | null, retryMaxDelayMs?: bigint | null, rateLimitJitterRatio?: number | null, cookiesEnabled?: boolean | null, removeTags?: string[] | null, content?: WasmContentConfig | null, downloadAssets?: boolean | null, assetTypes?: WasmAssetCategory[] | null, browser?: WasmBrowserConfig | null, userAgents?: string[] | null, captureScreenshot?: boolean | null, followDocumentUrls?: boolean | null, downloadDocuments?: boolean | null, documentMimeTypes?: string[] | null, saveBrowserProfile?: boolean | null, ssrf?: WasmSsrfPolicy | null, maxDepth?: number | null, maxPages?: number | null, maxLinksPerPage?: number | null, maxConcurrent?: number | null, contentFilter?: WasmContentFilterKind | null, bm25Query?: string | null, bm25Threshold?: number | null, userAgent?: string | null, rateLimitMs?: bigint | null, auth?: any | null, maxBodySize?: number | null, mapLimit?: number | null, mapSearch?: string | null, maxAssetSize?: number | null, proxy?: WasmProxyConfig | null, documentUrlDepth?: number | null, documentMaxSize?: number | null, documentOutputDir?: string | null, documentContentEncoding?: WasmDocumentContentEncoding | null, warcOutput?: string | null, browserProfile?: string | null, ssrfDenyPrivateExplicit?: boolean | null, dangerAcceptInvalidCerts?: boolean | null, subdomainHeaderAllowlist?: string[] | null);
     /**
      * Validate the configuration, returning an error if any values are invalid.
      *
@@ -545,6 +554,7 @@ export class WasmCrawlConfig {
     get crawlStrategy(): string;
     set crawlStrategy(value: WasmCrawlStrategyKind);
     customHeaders: any;
+    dangerAcceptInvalidCerts: boolean;
     dedupIncludeQuery: boolean;
     get documentContentEncoding(): string | undefined;
     set documentContentEncoding(value: WasmDocumentContentEncoding | null | undefined);
@@ -611,6 +621,7 @@ export class WasmCrawlConfig {
     ssrf: WasmSsrfPolicy;
     stayOnDomain: boolean;
     stripTrackingParams: boolean;
+    subdomainHeaderAllowlist: string[];
     trackingParams: string[];
     get userAgent(): string | undefined;
     set userAgent(value: string | null | undefined);
@@ -754,6 +765,7 @@ export class WasmCrawlPageResult {
 export class WasmCrawlResult {
     free(): void;
     [Symbol.dispose](): void;
+    clearWafBlock(): void;
     /**
      * Return an independent handle for an ownership-taking API.
      *
@@ -763,11 +775,16 @@ export class WasmCrawlResult {
     copyForTransfer(): WasmCrawlResult;
     static default(): WasmCrawlResult;
     /**
+     * Takes ownership of `wafBlock`.
+     * wasm-bindgen lowers a by-value class argument through `__destroy_into_raw()`, so each
+     * of those handles is dead once this returns. Passing a freshly built value is fine. To
+     * keep a handle you already hold, build with `default()` and assign the property instead:
+     * the generated setter borrows and leaves your handle alive.
      * Takes ownership of every element in `pages`, `cookies`.
      * wasm-bindgen cannot borrow class values nested in an array. To keep those handles,
      * copy each retained element with `copyForTransfer()` before passing the array.
      */
-    constructor(pages?: WasmCrawlPageResult[] | null, finalUrl?: string | null, redirectCount?: number | null, wasSkipped?: boolean | null, cookies?: WasmCookieInfo[] | null, stayedOnDomain?: boolean | null, browserUsed?: boolean | null, error?: string | null);
+    constructor(pages?: WasmCrawlPageResult[] | null, finalUrl?: string | null, redirectCount?: number | null, wasSkipped?: boolean | null, cookies?: WasmCookieInfo[] | null, stayedOnDomain?: boolean | null, browserUsed?: boolean | null, error?: string | null, wafBlock?: WasmWafBlockDetails | null);
     /**
      * Returns the count of unique normalized URLs encountered during crawling.
      *
@@ -799,6 +816,14 @@ export class WasmCrawlResult {
     pages: WasmCrawlPageResult[];
     redirectCount: number;
     stayedOnDomain: boolean;
+    /**
+     * Returns a detached copy of `wafBlock`.
+     *
+     * Changes to the returned value do not update this object. To update `wafBlock`,
+     * read it, modify the copy, then assign the copy back to `wafBlock`.
+     */
+    get wafBlock(): WasmWafBlockDetails | undefined;
+    set wafBlock(value: WasmWafBlockDetails);
     wasSkipped: boolean;
 }
 
@@ -1661,6 +1686,28 @@ export class WasmSsrfPolicy {
     schemeAllowlist: string[];
 }
 
+/**
+ * Structured context for a WAF-blocked response. <!-- -->
+ */
+export class WasmWafBlockDetails {
+    free(): void;
+    [Symbol.dispose](): void;
+    /**
+     * Return an independent handle for an ownership-taking API.
+     *
+     * Use `copyForTransfer()` for each class value placed in an array when the original
+     * handle must remain usable.
+     */
+    copyForTransfer(): WasmWafBlockDetails;
+    static default(): WasmWafBlockDetails;
+    constructor(vendor?: string | null, status?: number | null, bodyPrefix?: string | null);
+    get bodyPrefix(): string | undefined;
+    set bodyPrefix(value: string | null | undefined);
+    get status(): number | undefined;
+    set status(value: number | null | undefined);
+    vendor: string;
+}
+
 /**
  * Crawl multiple seed URLs concurrently, each following links to configured depth.
  */

package/pkg/bundler/crawlberg_wasm.js

@@ -5,5 +5,5 @@ import { __wbg_set_wasm } from "./crawlberg_wasm_bg.js";
 __wbg_set_wasm(wasm);
 wasm.__wbindgen_start();
 export {
-    WasmActionResult, WasmArticleMetadata, WasmAssetCategory, WasmAuthConfig, WasmBatchCrawlResult, WasmBatchCrawlResults, WasmBatchScrapeResult, WasmBatchScrapeResults, WasmBrowserBackend, WasmBrowserConfig, WasmBrowserExtras, WasmBrowserMode, WasmBrowserWait, WasmCitationReference, WasmCitationResult, WasmContentConfig, WasmContentFilterKind, WasmCookieInfo, WasmCrawlConfig, WasmCrawlEngineHandle, WasmCrawlPageResult, WasmCrawlResult, WasmCrawlStrategyKind, WasmDocumentContentEncoding, WasmDownloadedAsset, WasmDownloadedDocument, WasmExtractionMeta, WasmFaviconInfo, WasmFeedInfo, WasmFeedType, WasmHeadingInfo, WasmHostMatcher, WasmHreflangEntry, WasmImageInfo, WasmImageSource, WasmInteractionResult, WasmJsonLdEntry, WasmLinkInfo, WasmLinkType, WasmMapResult, WasmMarkdownResult, WasmPageAction, WasmPageMetadata, WasmProxyConfig, WasmResponseMeta, WasmScrapeResult, WasmScrollDirection, WasmSitemapUrl, WasmSsrfPolicy, batchCrawl, batchScrape, crawl, createEngine, generateCitations, interact, mapUrls, scrape
+    WasmActionResult, WasmArticleMetadata, WasmAssetCategory, WasmAuthConfig, WasmBatchCrawlResult, WasmBatchCrawlResults, WasmBatchScrapeResult, WasmBatchScrapeResults, WasmBrowserBackend, WasmBrowserConfig, WasmBrowserExtras, WasmBrowserMode, WasmBrowserWait, WasmCitationReference, WasmCitationResult, WasmContentConfig, WasmContentFilterKind, WasmCookieInfo, WasmCrawlConfig, WasmCrawlEngineHandle, WasmCrawlPageResult, WasmCrawlResult, WasmCrawlStrategyKind, WasmDocumentContentEncoding, WasmDownloadedAsset, WasmDownloadedDocument, WasmExtractionMeta, WasmFaviconInfo, WasmFeedInfo, WasmFeedType, WasmHeadingInfo, WasmHostMatcher, WasmHreflangEntry, WasmImageInfo, WasmImageSource, WasmInteractionResult, WasmJsonLdEntry, WasmLinkInfo, WasmLinkType, WasmMapResult, WasmMarkdownResult, WasmPageAction, WasmPageMetadata, WasmProxyConfig, WasmResponseMeta, WasmScrapeResult, WasmScrollDirection, WasmSitemapUrl, WasmSsrfPolicy, WasmWafBlockDetails, batchCrawl, batchScrape, crawl, createEngine, generateCitations, interact, mapUrls, scrape
 } from "./crawlberg_wasm_bg.js";