@xberg-io/crawlberg-wasm 1.7.1 → 1.7.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xberg-io/crawlberg-wasm",
-  "version": "1.7.1",
+  "version": "1.7.2",
   "private": false,
   "description": "High-performance web crawling engine",
   "license": "MIT",

package/pkg/bundler/crawlberg_wasm.d.ts

@@ -300,6 +300,9 @@ export class WasmCrawlConfig {
     constructor(crawlStrategy?: WasmCrawlStrategyKind | null, respectRobotsTxt?: boolean | null, softHttpErrors?: boolean | null, stayOnDomain?: boolean | null, allowSubdomains?: boolean | null, includePaths?: string[] | null, excludePaths?: string[] | null, customHeaders?: any | null, requestTimeout?: bigint | null, maxRedirects?: number | null, retryCount?: number | null, retryCodes?: Uint16Array | 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);
     /**
      * Validate the configuration, returning an error if any values are invalid.
+     *
+     * Only the first violation encountered is reported, not every one, so fixing a rejected
+     * configuration can surface a further error on the next attempt.
      */
     validate(): void;
     allowSubdomains: boolean;

package/pkg/bundler/crawlberg_wasm_bg.js

@@ -2744,6 +2744,9 @@ export class WasmCrawlConfig {
     }
     /**
      * Validate the configuration, returning an error if any values are invalid.
+     *
+     * Only the first violation encountered is reported, not every one, so fixing a rejected
+     * configuration can surface a further error on the next attempt.
      */
     validate() {
         const ret = wasm.wasmcrawlconfig_validate(this.__wbg_ptr);
@@ -7713,10 +7716,9 @@ if (Symbol.dispose) WasmSsrfPolicy.prototype[Symbol.dispose] = WasmSsrfPolicy.pr
  */
 export function batchCrawl(engine, urls) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.batchCrawl(ptr0, ptr1, len1);
+    const ptr0 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.batchCrawl(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7728,10 +7730,9 @@ export function batchCrawl(engine, urls) {
  */
 export function batchScrape(engine, urls) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.batchScrape(ptr0, ptr1, len1);
+    const ptr0 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.batchScrape(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7743,10 +7744,9 @@ export function batchScrape(engine, urls) {
  */
 export function crawl(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.crawl(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.crawl(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7796,12 +7796,11 @@ export function generateCitations(markdown) {
  */
 export function interact(engine, url, actions) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ptr1 = passArrayJsValueToWasm0(actions, wasm.__wbindgen_malloc);
     const len1 = WASM_VECTOR_LEN;
-    const ptr2 = passArrayJsValueToWasm0(actions, wasm.__wbindgen_malloc);
-    const len2 = WASM_VECTOR_LEN;
-    const ret = wasm.interact(ptr0, ptr1, len1, ptr2, len2);
+    const ret = wasm.interact(engine.__wbg_ptr, ptr0, len0, ptr1, len1);
     return ret;
 }
 
@@ -7813,10 +7812,9 @@ export function interact(engine, url, actions) {
  */
 export function mapUrls(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.mapUrls(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.mapUrls(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7828,10 +7826,9 @@ export function mapUrls(engine, url) {
  */
 export function scrape(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.scrape(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.scrape(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 export function __wbg_Error_67e7344beaa85059(arg0, arg1) {
@@ -8366,12 +8363,12 @@ export function __wbg_wasmsitemapurl_unwrap(arg0) {
     return ret;
 }
 export function __wbindgen_generic_0000000000000001(arg0, arg1) {
-    // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1107, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
+    // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1096, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
     const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h5567083e9d00da74);
     return ret;
 }
 export function __wbindgen_generic_0000000000000002(arg0, arg1) {
-    // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1011, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
+    // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1000, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
     const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h00e67b488e823f1f);
     return ret;
 }

package/pkg/bundler/package.json

@@ -2,7 +2,7 @@
   "name": "crawlberg-wasm",
   "type": "module",
   "description": "High-performance web crawling engine",
-  "version": "1.7.1",
+  "version": "1.7.2",
   "license": "MIT",
   "repository": {
     "type": "git",

package/pkg/deno/crawlberg_wasm.d.ts

@@ -300,6 +300,9 @@ export class WasmCrawlConfig {
     constructor(crawlStrategy?: WasmCrawlStrategyKind | null, respectRobotsTxt?: boolean | null, softHttpErrors?: boolean | null, stayOnDomain?: boolean | null, allowSubdomains?: boolean | null, includePaths?: string[] | null, excludePaths?: string[] | null, customHeaders?: any | null, requestTimeout?: bigint | null, maxRedirects?: number | null, retryCount?: number | null, retryCodes?: Uint16Array | 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);
     /**
      * Validate the configuration, returning an error if any values are invalid.
+     *
+     * Only the first violation encountered is reported, not every one, so fixing a rejected
+     * configuration can surface a further error on the next attempt.
      */
     validate(): void;
     allowSubdomains: boolean;

package/pkg/deno/crawlberg_wasm.js

@@ -2746,6 +2746,9 @@ export class WasmCrawlConfig {
     }
     /**
      * Validate the configuration, returning an error if any values are invalid.
+     *
+     * Only the first violation encountered is reported, not every one, so fixing a rejected
+     * configuration can surface a further error on the next attempt.
      */
     validate() {
         const ret = wasm.wasmcrawlconfig_validate(this.__wbg_ptr);
@@ -7715,10 +7718,9 @@ if (Symbol.dispose) WasmSsrfPolicy.prototype[Symbol.dispose] = WasmSsrfPolicy.pr
  */
 export function batchCrawl(engine, urls) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.batchCrawl(ptr0, ptr1, len1);
+    const ptr0 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.batchCrawl(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7730,10 +7732,9 @@ export function batchCrawl(engine, urls) {
  */
 export function batchScrape(engine, urls) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.batchScrape(ptr0, ptr1, len1);
+    const ptr0 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.batchScrape(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7745,10 +7746,9 @@ export function batchScrape(engine, urls) {
  */
 export function crawl(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.crawl(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.crawl(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7798,12 +7798,11 @@ export function generateCitations(markdown) {
  */
 export function interact(engine, url, actions) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ptr1 = passArrayJsValueToWasm0(actions, wasm.__wbindgen_malloc);
     const len1 = WASM_VECTOR_LEN;
-    const ptr2 = passArrayJsValueToWasm0(actions, wasm.__wbindgen_malloc);
-    const len2 = WASM_VECTOR_LEN;
-    const ret = wasm.interact(ptr0, ptr1, len1, ptr2, len2);
+    const ret = wasm.interact(engine.__wbg_ptr, ptr0, len0, ptr1, len1);
     return ret;
 }
 
@@ -7815,10 +7814,9 @@ export function interact(engine, url, actions) {
  */
 export function mapUrls(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.mapUrls(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.mapUrls(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7830,10 +7828,9 @@ export function mapUrls(engine, url) {
  */
 export function scrape(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.scrape(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.scrape(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 function __wbg_get_imports() {
@@ -8371,12 +8368,12 @@ function __wbg_get_imports() {
             return ret;
         },
         __wbindgen_generic_0000000000000001: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1107, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1096, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h5567083e9d00da74);
             return ret;
         },
         __wbindgen_generic_0000000000000002: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1011, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1000, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h00e67b488e823f1f);
             return ret;
         },

package/pkg/nodejs/crawlberg_wasm.d.ts

@@ -300,6 +300,9 @@ export class WasmCrawlConfig {
     constructor(crawlStrategy?: WasmCrawlStrategyKind | null, respectRobotsTxt?: boolean | null, softHttpErrors?: boolean | null, stayOnDomain?: boolean | null, allowSubdomains?: boolean | null, includePaths?: string[] | null, excludePaths?: string[] | null, customHeaders?: any | null, requestTimeout?: bigint | null, maxRedirects?: number | null, retryCount?: number | null, retryCodes?: Uint16Array | 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);
     /**
      * Validate the configuration, returning an error if any values are invalid.
+     *
+     * Only the first violation encountered is reported, not every one, so fixing a rejected
+     * configuration can surface a further error on the next attempt.
      */
     validate(): void;
     allowSubdomains: boolean;

package/pkg/nodejs/crawlberg_wasm.js

@@ -2764,6 +2764,9 @@ class WasmCrawlConfig {
     }
     /**
      * Validate the configuration, returning an error if any values are invalid.
+     *
+     * Only the first violation encountered is reported, not every one, so fixing a rejected
+     * configuration can surface a further error on the next attempt.
      */
     validate() {
         const ret = wasm.wasmcrawlconfig_validate(this.__wbg_ptr);
@@ -7764,10 +7767,9 @@ exports.WasmSsrfPolicy = WasmSsrfPolicy;
  */
 function batchCrawl(engine, urls) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.batchCrawl(ptr0, ptr1, len1);
+    const ptr0 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.batchCrawl(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 exports.batchCrawl = batchCrawl;
@@ -7780,10 +7782,9 @@ exports.batchCrawl = batchCrawl;
  */
 function batchScrape(engine, urls) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.batchScrape(ptr0, ptr1, len1);
+    const ptr0 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.batchScrape(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 exports.batchScrape = batchScrape;
@@ -7796,10 +7797,9 @@ exports.batchScrape = batchScrape;
  */
 function crawl(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.crawl(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.crawl(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 exports.crawl = crawl;
@@ -7852,12 +7852,11 @@ exports.generateCitations = generateCitations;
  */
 function interact(engine, url, actions) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ptr1 = passArrayJsValueToWasm0(actions, wasm.__wbindgen_malloc);
     const len1 = WASM_VECTOR_LEN;
-    const ptr2 = passArrayJsValueToWasm0(actions, wasm.__wbindgen_malloc);
-    const len2 = WASM_VECTOR_LEN;
-    const ret = wasm.interact(ptr0, ptr1, len1, ptr2, len2);
+    const ret = wasm.interact(engine.__wbg_ptr, ptr0, len0, ptr1, len1);
     return ret;
 }
 exports.interact = interact;
@@ -7870,10 +7869,9 @@ exports.interact = interact;
  */
 function mapUrls(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.mapUrls(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.mapUrls(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 exports.mapUrls = mapUrls;
@@ -7886,10 +7884,9 @@ exports.mapUrls = mapUrls;
  */
 function scrape(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.scrape(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.scrape(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 exports.scrape = scrape;
@@ -8428,12 +8425,12 @@ function __wbg_get_imports() {
             return ret;
         },
         __wbindgen_generic_0000000000000001: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1107, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1096, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h5567083e9d00da74);
             return ret;
         },
         __wbindgen_generic_0000000000000002: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1011, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1000, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h00e67b488e823f1f);
             return ret;
         },

package/pkg/nodejs/package.json

@@ -1,7 +1,7 @@
 {
   "name": "crawlberg-wasm",
   "description": "High-performance web crawling engine",
-  "version": "1.7.1",
+  "version": "1.7.2",
   "license": "MIT",
   "repository": {
     "type": "git",

package/pkg/web/crawlberg_wasm.d.ts

@@ -300,6 +300,9 @@ export class WasmCrawlConfig {
     constructor(crawlStrategy?: WasmCrawlStrategyKind | null, respectRobotsTxt?: boolean | null, softHttpErrors?: boolean | null, stayOnDomain?: boolean | null, allowSubdomains?: boolean | null, includePaths?: string[] | null, excludePaths?: string[] | null, customHeaders?: any | null, requestTimeout?: bigint | null, maxRedirects?: number | null, retryCount?: number | null, retryCodes?: Uint16Array | 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);
     /**
      * Validate the configuration, returning an error if any values are invalid.
+     *
+     * Only the first violation encountered is reported, not every one, so fixing a rejected
+     * configuration can surface a further error on the next attempt.
      */
     validate(): void;
     allowSubdomains: boolean;

package/pkg/web/crawlberg_wasm.js

@@ -2746,6 +2746,9 @@ export class WasmCrawlConfig {
     }
     /**
      * Validate the configuration, returning an error if any values are invalid.
+     *
+     * Only the first violation encountered is reported, not every one, so fixing a rejected
+     * configuration can surface a further error on the next attempt.
      */
     validate() {
         const ret = wasm.wasmcrawlconfig_validate(this.__wbg_ptr);
@@ -7715,10 +7718,9 @@ if (Symbol.dispose) WasmSsrfPolicy.prototype[Symbol.dispose] = WasmSsrfPolicy.pr
  */
 export function batchCrawl(engine, urls) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.batchCrawl(ptr0, ptr1, len1);
+    const ptr0 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.batchCrawl(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7730,10 +7732,9 @@ export function batchCrawl(engine, urls) {
  */
 export function batchScrape(engine, urls) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.batchScrape(ptr0, ptr1, len1);
+    const ptr0 = passArrayJsValueToWasm0(urls, wasm.__wbindgen_malloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.batchScrape(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7745,10 +7746,9 @@ export function batchScrape(engine, urls) {
  */
 export function crawl(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.crawl(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.crawl(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7798,12 +7798,11 @@ export function generateCitations(markdown) {
  */
 export function interact(engine, url, actions) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ptr1 = passArrayJsValueToWasm0(actions, wasm.__wbindgen_malloc);
     const len1 = WASM_VECTOR_LEN;
-    const ptr2 = passArrayJsValueToWasm0(actions, wasm.__wbindgen_malloc);
-    const len2 = WASM_VECTOR_LEN;
-    const ret = wasm.interact(ptr0, ptr1, len1, ptr2, len2);
+    const ret = wasm.interact(engine.__wbg_ptr, ptr0, len0, ptr1, len1);
     return ret;
 }
 
@@ -7815,10 +7814,9 @@ export function interact(engine, url, actions) {
  */
 export function mapUrls(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.mapUrls(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.mapUrls(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 
@@ -7830,10 +7828,9 @@ export function mapUrls(engine, url) {
  */
 export function scrape(engine, url) {
     _assertClass(engine, WasmCrawlEngineHandle);
-    var ptr0 = engine.__destroy_into_raw();
-    const ptr1 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
-    const len1 = WASM_VECTOR_LEN;
-    const ret = wasm.scrape(ptr0, ptr1, len1);
+    const ptr0 = passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.scrape(engine.__wbg_ptr, ptr0, len0);
     return ret;
 }
 function __wbg_get_imports() {
@@ -8371,12 +8368,12 @@ function __wbg_get_imports() {
             return ret;
         },
         __wbindgen_generic_0000000000000001: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1107, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1096, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h5567083e9d00da74);
             return ret;
         },
         __wbindgen_generic_0000000000000002: function(arg0, arg1) {
-            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1011, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
+            // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1000, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
             const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h00e67b488e823f1f);
             return ret;
         },

package/pkg/web/package.json

@@ -2,7 +2,7 @@
   "name": "crawlberg-wasm",
   "type": "module",
   "description": "High-performance web crawling engine",
-  "version": "1.7.1",
+  "version": "1.7.2",
   "license": "MIT",
   "repository": {
     "type": "git",