@xberg-io/crawlberg-wasm 1.10.3 → 1.11.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -192,21 +192,31 @@ export const wasmcitationresult_new: (a: number, b: number, c: number, d: number
192
192
  export const wasmcitationresult_references: (a: number) => [number, number];
193
193
  export const wasmcitationresult_set_content: (a: number, b: number, c: number) => void;
194
194
  export const wasmcitationresult_set_references: (a: number, b: number, c: number) => void;
195
+ export const wasmcontentconfig_bullets: (a: number) => [number, number];
196
+ export const wasmcontentconfig_compactTables: (a: number) => number;
195
197
  export const wasmcontentconfig_copyForTransfer: (a: number) => number;
196
198
  export const wasmcontentconfig_default: () => number;
197
199
  export const wasmcontentconfig_excludeSelectors: (a: number) => [number, number];
198
200
  export const wasmcontentconfig_extractMetadata: (a: number) => number;
199
201
  export const wasmcontentconfig_includeDocumentStructure: (a: number) => number;
202
+ export const wasmcontentconfig_keepHidden: (a: number) => number;
203
+ export const wasmcontentconfig_keepInlineImagesIn: (a: number) => [number, number];
204
+ export const wasmcontentconfig_listIndentWidth: (a: number) => number;
200
205
  export const wasmcontentconfig_maxDepth: (a: number) => number;
201
- export const wasmcontentconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number) => number;
206
+ export const wasmcontentconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number) => number;
202
207
  export const wasmcontentconfig_outputFormat: (a: number) => [number, number];
203
208
  export const wasmcontentconfig_preprocessingPreset: (a: number) => [number, number];
204
209
  export const wasmcontentconfig_preserveTags: (a: number) => [number, number];
205
210
  export const wasmcontentconfig_removeForms: (a: number) => number;
206
211
  export const wasmcontentconfig_removeNavigation: (a: number) => number;
212
+ export const wasmcontentconfig_set_bullets: (a: number, b: number, c: number) => void;
213
+ export const wasmcontentconfig_set_compactTables: (a: number, b: number) => void;
207
214
  export const wasmcontentconfig_set_excludeSelectors: (a: number, b: number, c: number) => void;
208
215
  export const wasmcontentconfig_set_extractMetadata: (a: number, b: number) => void;
209
216
  export const wasmcontentconfig_set_includeDocumentStructure: (a: number, b: number) => void;
217
+ export const wasmcontentconfig_set_keepHidden: (a: number, b: number) => void;
218
+ export const wasmcontentconfig_set_keepInlineImagesIn: (a: number, b: number, c: number) => void;
219
+ export const wasmcontentconfig_set_listIndentWidth: (a: number, b: number) => void;
210
220
  export const wasmcontentconfig_set_maxDepth: (a: number, b: number) => void;
211
221
  export const wasmcontentconfig_set_outputFormat: (a: number, b: number, c: number) => void;
212
222
  export const wasmcontentconfig_set_preprocessingPreset: (a: number, b: number, c: number) => void;
@@ -214,11 +224,15 @@ export const wasmcontentconfig_set_preserveTags: (a: number, b: number, c: numbe
214
224
  export const wasmcontentconfig_set_removeForms: (a: number, b: number) => void;
215
225
  export const wasmcontentconfig_set_removeNavigation: (a: number, b: number) => void;
216
226
  export const wasmcontentconfig_set_skipImages: (a: number, b: number) => void;
227
+ export const wasmcontentconfig_set_stripNewlines: (a: number, b: number) => void;
217
228
  export const wasmcontentconfig_set_stripTags: (a: number, b: number, c: number) => void;
229
+ export const wasmcontentconfig_set_strongEmSymbol: (a: number, b: number, c: number) => void;
218
230
  export const wasmcontentconfig_set_wrap: (a: number, b: number) => void;
219
231
  export const wasmcontentconfig_set_wrapWidth: (a: number, b: number) => void;
220
232
  export const wasmcontentconfig_skipImages: (a: number) => number;
233
+ export const wasmcontentconfig_stripNewlines: (a: number) => number;
221
234
  export const wasmcontentconfig_stripTags: (a: number) => [number, number];
235
+ export const wasmcontentconfig_strongEmSymbol: (a: number) => [number, number];
222
236
  export const wasmcontentconfig_wrap: (a: number) => number;
223
237
  export const wasmcontentconfig_wrapWidth: (a: number) => number;
224
238
  export const wasmcookieinfo_copyForTransfer: (a: number) => number;
@@ -370,10 +384,11 @@ export const wasmcrawlpageresult_jsonLd: (a: number) => [number, number];
370
384
  export const wasmcrawlpageresult_links: (a: number) => [number, number];
371
385
  export const wasmcrawlpageresult_markdown: (a: number) => number;
372
386
  export const wasmcrawlpageresult_metadata: (a: number) => number;
373
- export const wasmcrawlpageresult_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number) => number;
387
+ export const wasmcrawlpageresult_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number) => number;
374
388
  export const wasmcrawlpageresult_nofollowDetected: (a: number) => number;
375
389
  export const wasmcrawlpageresult_noindexDetected: (a: number) => number;
376
390
  export const wasmcrawlpageresult_normalizedUrl: (a: number) => [number, number];
391
+ export const wasmcrawlpageresult_originalUrl: (a: number) => [number, number];
377
392
  export const wasmcrawlpageresult_redirectCount: (a: number) => number;
378
393
  export const wasmcrawlpageresult_set_bodySize: (a: number, b: number) => void;
379
394
  export const wasmcrawlpageresult_set_browserUsed: (a: number, b: number) => void;
@@ -395,6 +410,7 @@ export const wasmcrawlpageresult_set_metadata: (a: number, b: number) => void;
395
410
  export const wasmcrawlpageresult_set_nofollowDetected: (a: number, b: number) => void;
396
411
  export const wasmcrawlpageresult_set_noindexDetected: (a: number, b: number) => void;
397
412
  export const wasmcrawlpageresult_set_normalizedUrl: (a: number, b: number, c: number) => void;
413
+ export const wasmcrawlpageresult_set_originalUrl: (a: number, b: number, c: number) => void;
398
414
  export const wasmcrawlpageresult_set_redirectCount: (a: number, b: number) => void;
399
415
  export const wasmcrawlpageresult_set_ssrfRefusedUrls: (a: number, b: number, c: number) => void;
400
416
  export const wasmcrawlpageresult_set_statusCode: (a: number, b: number) => void;
@@ -553,11 +569,13 @@ export const wasmjsonldentry_set_schemaType: (a: number, b: number, c: number) =
553
569
  export const wasmlinkinfo_copyForTransfer: (a: number) => number;
554
570
  export const wasmlinkinfo_default: () => number;
555
571
  export const wasmlinkinfo_linkType: (a: number) => [number, number];
556
- export const wasmlinkinfo_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number) => number;
572
+ export const wasmlinkinfo_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number) => number;
557
573
  export const wasmlinkinfo_nofollow: (a: number) => number;
574
+ export const wasmlinkinfo_originalUrl: (a: number) => [number, number];
558
575
  export const wasmlinkinfo_rel: (a: number) => [number, number];
559
576
  export const wasmlinkinfo_set_linkType: (a: number, b: number) => void;
560
577
  export const wasmlinkinfo_set_nofollow: (a: number, b: number) => void;
578
+ export const wasmlinkinfo_set_originalUrl: (a: number, b: number, c: number) => void;
561
579
  export const wasmlinkinfo_set_rel: (a: number, b: number, c: number) => void;
562
580
  export const wasmlinkinfo_set_text: (a: number, b: number, c: number) => void;
563
581
  export const wasmlinkinfo_set_url: (a: number, b: number, c: number) => void;
@@ -812,9 +830,9 @@ export const wasmssrfpolicy_set_denyPrivate: (a: number, b: number) => void;
812
830
  export const wasmssrfpolicy_set_denylist: (a: number, b: any) => void;
813
831
  export const wasmssrfpolicy_set_maxRedirects: (a: number, b: number) => void;
814
832
  export const wasmssrfpolicy_set_schemeAllowlist: (a: number, b: number, c: number) => void;
815
- export const wasm_bindgen__convert__closures_____invoke__h2eedacf566f3748c: (a: number, b: number, c: any) => [number, number];
816
- export const wasm_bindgen__convert__closures_____invoke__h8ad11faaa2905452: (a: number, b: number, c: any, d: any) => void;
817
- export const wasm_bindgen__convert__closures_____invoke__h3a7ccac3e94bbb5c: (a: number, b: number) => void;
833
+ export const wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined_______true_: (a: number, b: number, c: any, d: any) => void;
834
+ export const wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___wasm_bindgen_ceaac9754e66b2fe___JsValue__core_608f92abc48d28da___result__Result_____wasm_bindgen_ceaac9754e66b2fe___JsError___true_: (a: number, b: number, c: any) => [number, number];
835
+ export const wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke_______true_: (a: number, b: number) => void;
818
836
  export const __wbindgen_malloc: (a: number, b: number) => number;
819
837
  export const __wbindgen_realloc: (a: number, b: number, c: number, d: number) => number;
820
838
  export const __wbindgen_exn_store: (a: number) => void;
@@ -1,5 +1,5 @@
1
1
  <!-- This file is auto-generated by alef — DO NOT EDIT. -->
2
- <!-- alef:hash:3e2f4a488c2c856faef7619c7da3856ab9b8cdad4534264dea7b27a2ebf8bf35 -->
2
+ <!-- alef:hash:317aa594b0d20c83a18e8cdb5342b90b90f56e94e51ba85fc80ebb44ab9c1331 -->
3
3
  <!-- To regenerate: alef readme -->
4
4
  <!-- To verify freshness: alef verify -->
5
5
 
@@ -101,14 +101,14 @@ pnpm add @xberg-io/crawlberg-wasm
101
101
 
102
102
  ## Agent plugin
103
103
 
104
- The `crawlberg` plugin is available via the `xberg-io/plugins` marketplace.
104
+ The `crawlberg` plugin is available via the `xberg-io/crawlberg` marketplace.
105
105
 
106
106
  ```text
107
- /plugin marketplace add xberg-io/plugins
108
- /plugin install crawlberg@xberg
107
+ /plugin marketplace add xberg-io/crawlberg
108
+ /plugin install crawlberg@crawlberg
109
109
  ```
110
110
 
111
- Works with Claude Code, Codex, Cursor, Gemini CLI, Factory Droid, GitHub Copilot CLI, and opencode. See [the marketplace README](https://github.com/xberg-io/plugins) for harness-specific install instructions.
111
+ See [the plugin guide](https://github.com/xberg-io/crawlberg/blob/main/plugin/README.md) for all supported runtimes, CLI prerequisites and standalone public skills.
112
112
 
113
113
  ## Quick Start
114
114
 
@@ -427,10 +427,15 @@ export class WasmContentConfig {
427
427
  */
428
428
  copyForTransfer(): WasmContentConfig;
429
429
  static default(): WasmContentConfig;
430
- 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);
430
+ 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);
431
+ bullets: string;
432
+ compactTables: boolean;
431
433
  excludeSelectors: string[];
432
434
  extractMetadata: boolean;
433
435
  includeDocumentStructure: boolean;
436
+ keepHidden: boolean;
437
+ keepInlineImagesIn: string[];
438
+ listIndentWidth: number;
434
439
  get maxDepth(): number | undefined;
435
440
  set maxDepth(value: number | null | undefined);
436
441
  outputFormat: string;
@@ -439,7 +444,9 @@ export class WasmContentConfig {
439
444
  removeForms: boolean;
440
445
  removeNavigation: boolean;
441
446
  skipImages: boolean;
447
+ stripNewlines: boolean;
442
448
  stripTags: string[];
449
+ strongEmSymbol: string;
443
450
  wrapWidth: number;
444
451
  wrap: boolean;
445
452
  }
@@ -452,7 +459,10 @@ export enum WasmContentFilterKind {
452
459
  }
453
460
 
454
461
  /**
455
- * Information about an HTTP cookie received from a response.
462
+ * Information about a cookie that a crawled page set.
463
+ *
464
+ * A crawl result gives `domain` and `path` as the browser stores the cookie, so a cookie has
465
+ * the same fields whether an HTTP fetch or the browser received it.
456
466
  */
457
467
  export class WasmCookieInfo {
458
468
  free(): void;
@@ -655,7 +665,7 @@ export class WasmCrawlPageResult {
655
665
  * wasm-bindgen cannot borrow class values nested in an array. To keep those handles,
656
666
  * copy each retained element with `copyForTransfer()` before passing the array.
657
667
  */
658
- constructor(url?: string | null, normalizedUrl?: string | null, statusCode?: number | null, contentType?: string | null, html?: string | null, bodySize?: number | null, metadata?: WasmPageMetadata | null, links?: WasmLinkInfo[] | null, images?: WasmImageInfo[] | null, feeds?: WasmFeedInfo[] | null, jsonLd?: WasmJsonLdEntry[] | null, depth?: number | null, stayedOnDomain?: boolean | null, wasSkipped?: boolean | null, isPdf?: boolean | null, browserUsed?: boolean | null, finalUrl?: string | null, redirectCount?: number | null, noindexDetected?: boolean | null, nofollowDetected?: boolean | null, ssrfRefusedUrls?: string[] | null, detectedCharset?: string | null, markdown?: WasmMarkdownResult | null, extractedData?: any | null, extractionMeta?: WasmExtractionMeta | null, downloadedDocument?: WasmDownloadedDocument | null);
668
+ constructor(url?: string | null, normalizedUrl?: string | null, statusCode?: number | null, contentType?: string | null, html?: string | null, bodySize?: number | null, metadata?: WasmPageMetadata | null, links?: WasmLinkInfo[] | null, images?: WasmImageInfo[] | null, feeds?: WasmFeedInfo[] | null, jsonLd?: WasmJsonLdEntry[] | null, depth?: number | null, stayedOnDomain?: boolean | null, wasSkipped?: boolean | null, isPdf?: boolean | null, browserUsed?: boolean | null, finalUrl?: string | null, redirectCount?: number | null, noindexDetected?: boolean | null, nofollowDetected?: boolean | null, ssrfRefusedUrls?: string[] | null, detectedCharset?: string | null, markdown?: WasmMarkdownResult | null, extractedData?: any | null, extractionMeta?: WasmExtractionMeta | null, downloadedDocument?: WasmDownloadedDocument | null, originalUrl?: string | null);
659
669
  bodySize: number;
660
670
  browserUsed: boolean;
661
671
  contentType: string;
@@ -729,6 +739,7 @@ export class WasmCrawlPageResult {
729
739
  nofollowDetected: boolean;
730
740
  noindexDetected: boolean;
731
741
  normalizedUrl: string;
742
+ originalUrl: string;
732
743
  redirectCount: number;
733
744
  ssrfRefusedUrls: string[];
734
745
  statusCode: number;
@@ -1126,10 +1137,11 @@ export class WasmLinkInfo {
1126
1137
  */
1127
1138
  copyForTransfer(): WasmLinkInfo;
1128
1139
  static default(): WasmLinkInfo;
1129
- constructor(url?: string | null, text?: string | null, linkType?: WasmLinkType | null, nofollow?: boolean | null, rel?: string | null);
1140
+ constructor(url?: string | null, text?: string | null, linkType?: WasmLinkType | null, nofollow?: boolean | null, rel?: string | null, originalUrl?: string | null);
1130
1141
  get linkType(): string;
1131
1142
  set linkType(value: WasmLinkType);
1132
1143
  nofollow: boolean;
1144
+ originalUrl: string;
1133
1145
  get rel(): string | undefined;
1134
1146
  set rel(value: string | null | undefined);
1135
1147
  text: string;
@@ -1807,6 +1807,28 @@ class WasmContentConfig {
1807
1807
  const ptr = this.__destroy_into_raw();
1808
1808
  wasm.__wbg_wasmcontentconfig_free(ptr, 0);
1809
1809
  }
1810
+ /**
1811
+ * @returns {string}
1812
+ */
1813
+ get bullets() {
1814
+ let deferred1_0;
1815
+ let deferred1_1;
1816
+ try {
1817
+ const ret = wasm.wasmcontentconfig_bullets(this.__wbg_ptr);
1818
+ deferred1_0 = ret[0];
1819
+ deferred1_1 = ret[1];
1820
+ return getStringFromWasm0(ret[0], ret[1]);
1821
+ } finally {
1822
+ wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
1823
+ }
1824
+ }
1825
+ /**
1826
+ * @returns {boolean}
1827
+ */
1828
+ get compactTables() {
1829
+ const ret = wasm.wasmcontentconfig_compactTables(this.__wbg_ptr);
1830
+ return ret !== 0;
1831
+ }
1810
1832
  /**
1811
1833
  * Return an independent handle for an ownership-taking API.
1812
1834
  *
@@ -1848,6 +1870,29 @@ class WasmContentConfig {
1848
1870
  const ret = wasm.wasmcontentconfig_includeDocumentStructure(this.__wbg_ptr);
1849
1871
  return ret !== 0;
1850
1872
  }
1873
+ /**
1874
+ * @returns {boolean}
1875
+ */
1876
+ get keepHidden() {
1877
+ const ret = wasm.wasmcontentconfig_keepHidden(this.__wbg_ptr);
1878
+ return ret !== 0;
1879
+ }
1880
+ /**
1881
+ * @returns {string[]}
1882
+ */
1883
+ get keepInlineImagesIn() {
1884
+ const ret = wasm.wasmcontentconfig_keepInlineImagesIn(this.__wbg_ptr);
1885
+ var v1 = getArrayJsValueFromWasm0(ret[0], ret[1]);
1886
+ wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
1887
+ return v1;
1888
+ }
1889
+ /**
1890
+ * @returns {number}
1891
+ */
1892
+ get listIndentWidth() {
1893
+ const ret = wasm.wasmcontentconfig_listIndentWidth(this.__wbg_ptr);
1894
+ return ret >>> 0;
1895
+ }
1851
1896
  /**
1852
1897
  * @returns {number | undefined}
1853
1898
  */
@@ -1869,8 +1914,15 @@ class WasmContentConfig {
1869
1914
  * @param {boolean | null} [includeDocumentStructure]
1870
1915
  * @param {boolean | null} [extractMetadata]
1871
1916
  * @param {number | null} [maxDepth]
1872
- */
1873
- constructor(outputFormat, preprocessingPreset, removeNavigation, removeForms, stripTags, preserveTags, excludeSelectors, skipImages, wrap, wrapWidth, includeDocumentStructure, extractMetadata, maxDepth) {
1917
+ * @param {boolean | null} [stripNewlines]
1918
+ * @param {string | null} [bullets]
1919
+ * @param {number | null} [listIndentWidth]
1920
+ * @param {string | null} [strongEmSymbol]
1921
+ * @param {boolean | null} [compactTables]
1922
+ * @param {string[] | null} [keepInlineImagesIn]
1923
+ * @param {boolean | null} [keepHidden]
1924
+ */
1925
+ constructor(outputFormat, preprocessingPreset, removeNavigation, removeForms, stripTags, preserveTags, excludeSelectors, skipImages, wrap, wrapWidth, includeDocumentStructure, extractMetadata, maxDepth, stripNewlines, bullets, listIndentWidth, strongEmSymbol, compactTables, keepInlineImagesIn, keepHidden) {
1874
1926
  var ptr0 = isLikeNone(outputFormat) ? 0 : passStringToWasm0(outputFormat, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
1875
1927
  var len0 = WASM_VECTOR_LEN;
1876
1928
  var ptr1 = isLikeNone(preprocessingPreset) ? 0 : passStringToWasm0(preprocessingPreset, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
@@ -1881,7 +1933,13 @@ class WasmContentConfig {
1881
1933
  var len3 = WASM_VECTOR_LEN;
1882
1934
  var ptr4 = isLikeNone(excludeSelectors) ? 0 : passArrayJsValueToWasm0(excludeSelectors, wasm.__wbindgen_malloc);
1883
1935
  var len4 = WASM_VECTOR_LEN;
1884
- const ret = wasm.wasmcontentconfig_new(ptr0, len0, ptr1, len1, isLikeNone(removeNavigation) ? 0xFFFFFF : removeNavigation ? 1 : 0, isLikeNone(removeForms) ? 0xFFFFFF : removeForms ? 1 : 0, ptr2, len2, ptr3, len3, ptr4, len4, isLikeNone(skipImages) ? 0xFFFFFF : skipImages ? 1 : 0, isLikeNone(wrap) ? 0xFFFFFF : wrap ? 1 : 0, isLikeNone(wrapWidth) ? Number.MAX_SAFE_INTEGER : (wrapWidth) >>> 0, isLikeNone(includeDocumentStructure) ? 0xFFFFFF : includeDocumentStructure ? 1 : 0, isLikeNone(extractMetadata) ? 0xFFFFFF : extractMetadata ? 1 : 0, isLikeNone(maxDepth) ? Number.MAX_SAFE_INTEGER : (maxDepth) >>> 0);
1936
+ var ptr5 = isLikeNone(bullets) ? 0 : passStringToWasm0(bullets, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
1937
+ var len5 = WASM_VECTOR_LEN;
1938
+ var ptr6 = isLikeNone(strongEmSymbol) ? 0 : passStringToWasm0(strongEmSymbol, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
1939
+ var len6 = WASM_VECTOR_LEN;
1940
+ var ptr7 = isLikeNone(keepInlineImagesIn) ? 0 : passArrayJsValueToWasm0(keepInlineImagesIn, wasm.__wbindgen_malloc);
1941
+ var len7 = WASM_VECTOR_LEN;
1942
+ const ret = wasm.wasmcontentconfig_new(ptr0, len0, ptr1, len1, isLikeNone(removeNavigation) ? 0xFFFFFF : removeNavigation ? 1 : 0, isLikeNone(removeForms) ? 0xFFFFFF : removeForms ? 1 : 0, ptr2, len2, ptr3, len3, ptr4, len4, isLikeNone(skipImages) ? 0xFFFFFF : skipImages ? 1 : 0, isLikeNone(wrap) ? 0xFFFFFF : wrap ? 1 : 0, isLikeNone(wrapWidth) ? Number.MAX_SAFE_INTEGER : (wrapWidth) >>> 0, isLikeNone(includeDocumentStructure) ? 0xFFFFFF : includeDocumentStructure ? 1 : 0, isLikeNone(extractMetadata) ? 0xFFFFFF : extractMetadata ? 1 : 0, isLikeNone(maxDepth) ? Number.MAX_SAFE_INTEGER : (maxDepth) >>> 0, isLikeNone(stripNewlines) ? 0xFFFFFF : stripNewlines ? 1 : 0, ptr5, len5, isLikeNone(listIndentWidth) ? Number.MAX_SAFE_INTEGER : (listIndentWidth) >>> 0, ptr6, len6, isLikeNone(compactTables) ? 0xFFFFFF : compactTables ? 1 : 0, ptr7, len7, isLikeNone(keepHidden) ? 0xFFFFFF : keepHidden ? 1 : 0);
1885
1943
  this.__wbg_ptr = ret;
1886
1944
  WasmContentConfigFinalization.register(this, this.__wbg_ptr, this);
1887
1945
  return this;
@@ -1939,6 +1997,20 @@ class WasmContentConfig {
1939
1997
  const ret = wasm.wasmcontentconfig_removeNavigation(this.__wbg_ptr);
1940
1998
  return ret !== 0;
1941
1999
  }
2000
+ /**
2001
+ * @param {string} value
2002
+ */
2003
+ set bullets(value) {
2004
+ const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2005
+ const len0 = WASM_VECTOR_LEN;
2006
+ wasm.wasmcontentconfig_set_bullets(this.__wbg_ptr, ptr0, len0);
2007
+ }
2008
+ /**
2009
+ * @param {boolean} value
2010
+ */
2011
+ set compactTables(value) {
2012
+ wasm.wasmcontentconfig_set_compactTables(this.__wbg_ptr, value);
2013
+ }
1942
2014
  /**
1943
2015
  * @param {string[]} value
1944
2016
  */
@@ -1959,6 +2031,26 @@ class WasmContentConfig {
1959
2031
  set includeDocumentStructure(value) {
1960
2032
  wasm.wasmcontentconfig_set_includeDocumentStructure(this.__wbg_ptr, value);
1961
2033
  }
2034
+ /**
2035
+ * @param {boolean} value
2036
+ */
2037
+ set keepHidden(value) {
2038
+ wasm.wasmcontentconfig_set_keepHidden(this.__wbg_ptr, value);
2039
+ }
2040
+ /**
2041
+ * @param {string[]} value
2042
+ */
2043
+ set keepInlineImagesIn(value) {
2044
+ const ptr0 = passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc);
2045
+ const len0 = WASM_VECTOR_LEN;
2046
+ wasm.wasmcontentconfig_set_keepInlineImagesIn(this.__wbg_ptr, ptr0, len0);
2047
+ }
2048
+ /**
2049
+ * @param {number} value
2050
+ */
2051
+ set listIndentWidth(value) {
2052
+ wasm.wasmcontentconfig_set_listIndentWidth(this.__wbg_ptr, value);
2053
+ }
1962
2054
  /**
1963
2055
  * @param {number | null} [value]
1964
2056
  */
@@ -2007,6 +2099,12 @@ class WasmContentConfig {
2007
2099
  set skipImages(value) {
2008
2100
  wasm.wasmcontentconfig_set_skipImages(this.__wbg_ptr, value);
2009
2101
  }
2102
+ /**
2103
+ * @param {boolean} value
2104
+ */
2105
+ set stripNewlines(value) {
2106
+ wasm.wasmcontentconfig_set_stripNewlines(this.__wbg_ptr, value);
2107
+ }
2010
2108
  /**
2011
2109
  * @param {string[]} value
2012
2110
  */
@@ -2015,6 +2113,14 @@ class WasmContentConfig {
2015
2113
  const len0 = WASM_VECTOR_LEN;
2016
2114
  wasm.wasmcontentconfig_set_stripTags(this.__wbg_ptr, ptr0, len0);
2017
2115
  }
2116
+ /**
2117
+ * @param {string} value
2118
+ */
2119
+ set strongEmSymbol(value) {
2120
+ const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2121
+ const len0 = WASM_VECTOR_LEN;
2122
+ wasm.wasmcontentconfig_set_strongEmSymbol(this.__wbg_ptr, ptr0, len0);
2123
+ }
2018
2124
  /**
2019
2125
  * @param {number} value
2020
2126
  */
@@ -2034,6 +2140,13 @@ class WasmContentConfig {
2034
2140
  const ret = wasm.wasmcontentconfig_skipImages(this.__wbg_ptr);
2035
2141
  return ret !== 0;
2036
2142
  }
2143
+ /**
2144
+ * @returns {boolean}
2145
+ */
2146
+ get stripNewlines() {
2147
+ const ret = wasm.wasmcontentconfig_stripNewlines(this.__wbg_ptr);
2148
+ return ret !== 0;
2149
+ }
2037
2150
  /**
2038
2151
  * @returns {string[]}
2039
2152
  */
@@ -2043,6 +2156,21 @@ class WasmContentConfig {
2043
2156
  wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
2044
2157
  return v1;
2045
2158
  }
2159
+ /**
2160
+ * @returns {string}
2161
+ */
2162
+ get strongEmSymbol() {
2163
+ let deferred1_0;
2164
+ let deferred1_1;
2165
+ try {
2166
+ const ret = wasm.wasmcontentconfig_strongEmSymbol(this.__wbg_ptr);
2167
+ deferred1_0 = ret[0];
2168
+ deferred1_1 = ret[1];
2169
+ return getStringFromWasm0(ret[0], ret[1]);
2170
+ } finally {
2171
+ wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
2172
+ }
2173
+ }
2046
2174
  /**
2047
2175
  * @returns {number}
2048
2176
  */
@@ -2071,7 +2199,10 @@ const WasmContentFilterKind = Object.freeze({
2071
2199
  exports.WasmContentFilterKind = WasmContentFilterKind;
2072
2200
 
2073
2201
  /**
2074
- * Information about an HTTP cookie received from a response.
2202
+ * Information about a cookie that a crawled page set.
2203
+ *
2204
+ * A crawl result gives `domain` and `path` as the browser stores the cookie, so a cookie has
2205
+ * the same fields whether an HTTP fetch or the browser received it.
2075
2206
  */
2076
2207
  class WasmCookieInfo {
2077
2208
  static __wrap(ptr) {
@@ -3538,8 +3669,9 @@ class WasmCrawlPageResult {
3538
3669
  * @param {any | null} [extractedData]
3539
3670
  * @param {WasmExtractionMeta | null} [extractionMeta]
3540
3671
  * @param {WasmDownloadedDocument | null} [downloadedDocument]
3672
+ * @param {string | null} [originalUrl]
3541
3673
  */
3542
- constructor(url, normalizedUrl, statusCode, contentType, html, bodySize, metadata, links, images, feeds, jsonLd, depth, stayedOnDomain, wasSkipped, isPdf, browserUsed, finalUrl, redirectCount, noindexDetected, nofollowDetected, ssrfRefusedUrls, detectedCharset, markdown, extractedData, extractionMeta, downloadedDocument) {
3674
+ constructor(url, normalizedUrl, statusCode, contentType, html, bodySize, metadata, links, images, feeds, jsonLd, depth, stayedOnDomain, wasSkipped, isPdf, browserUsed, finalUrl, redirectCount, noindexDetected, nofollowDetected, ssrfRefusedUrls, detectedCharset, markdown, extractedData, extractionMeta, downloadedDocument, originalUrl) {
3543
3675
  var ptr0 = isLikeNone(url) ? 0 : passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
3544
3676
  var len0 = WASM_VECTOR_LEN;
3545
3677
  var ptr1 = isLikeNone(normalizedUrl) ? 0 : passStringToWasm0(normalizedUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
@@ -3582,7 +3714,9 @@ class WasmCrawlPageResult {
3582
3714
  _assertClass(downloadedDocument, WasmDownloadedDocument);
3583
3715
  ptr14 = downloadedDocument.__destroy_into_raw();
3584
3716
  }
3585
- const ret = wasm.wasmcrawlpageresult_new(ptr0, len0, ptr1, len1, isLikeNone(statusCode) ? 0xFFFFFF : statusCode, ptr2, len2, ptr3, len3, isLikeNone(bodySize) ? Number.MAX_SAFE_INTEGER : (bodySize) >>> 0, ptr4, ptr5, len5, ptr6, len6, ptr7, len7, ptr8, len8, isLikeNone(depth) ? Number.MAX_SAFE_INTEGER : (depth) >>> 0, isLikeNone(stayedOnDomain) ? 0xFFFFFF : stayedOnDomain ? 1 : 0, isLikeNone(wasSkipped) ? 0xFFFFFF : wasSkipped ? 1 : 0, isLikeNone(isPdf) ? 0xFFFFFF : isPdf ? 1 : 0, isLikeNone(browserUsed) ? 0xFFFFFF : browserUsed ? 1 : 0, ptr9, len9, isLikeNone(redirectCount) ? Number.MAX_SAFE_INTEGER : (redirectCount) >>> 0, isLikeNone(noindexDetected) ? 0xFFFFFF : noindexDetected ? 1 : 0, isLikeNone(nofollowDetected) ? 0xFFFFFF : nofollowDetected ? 1 : 0, ptr10, len10, ptr11, len11, ptr12, isLikeNone(extractedData) ? 0 : addToExternrefTable0(extractedData), ptr13, ptr14);
3717
+ var ptr15 = isLikeNone(originalUrl) ? 0 : passStringToWasm0(originalUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
3718
+ var len15 = WASM_VECTOR_LEN;
3719
+ const ret = wasm.wasmcrawlpageresult_new(ptr0, len0, ptr1, len1, isLikeNone(statusCode) ? 0xFFFFFF : statusCode, ptr2, len2, ptr3, len3, isLikeNone(bodySize) ? Number.MAX_SAFE_INTEGER : (bodySize) >>> 0, ptr4, ptr5, len5, ptr6, len6, ptr7, len7, ptr8, len8, isLikeNone(depth) ? Number.MAX_SAFE_INTEGER : (depth) >>> 0, isLikeNone(stayedOnDomain) ? 0xFFFFFF : stayedOnDomain ? 1 : 0, isLikeNone(wasSkipped) ? 0xFFFFFF : wasSkipped ? 1 : 0, isLikeNone(isPdf) ? 0xFFFFFF : isPdf ? 1 : 0, isLikeNone(browserUsed) ? 0xFFFFFF : browserUsed ? 1 : 0, ptr9, len9, isLikeNone(redirectCount) ? Number.MAX_SAFE_INTEGER : (redirectCount) >>> 0, isLikeNone(noindexDetected) ? 0xFFFFFF : noindexDetected ? 1 : 0, isLikeNone(nofollowDetected) ? 0xFFFFFF : nofollowDetected ? 1 : 0, ptr10, len10, ptr11, len11, ptr12, isLikeNone(extractedData) ? 0 : addToExternrefTable0(extractedData), ptr13, ptr14, ptr15, len15);
3586
3720
  this.__wbg_ptr = ret;
3587
3721
  WasmCrawlPageResultFinalization.register(this, this.__wbg_ptr, this);
3588
3722
  return this;
@@ -3616,6 +3750,21 @@ class WasmCrawlPageResult {
3616
3750
  wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
3617
3751
  }
3618
3752
  }
3753
+ /**
3754
+ * @returns {string}
3755
+ */
3756
+ get originalUrl() {
3757
+ let deferred1_0;
3758
+ let deferred1_1;
3759
+ try {
3760
+ const ret = wasm.wasmcrawlpageresult_originalUrl(this.__wbg_ptr);
3761
+ deferred1_0 = ret[0];
3762
+ deferred1_1 = ret[1];
3763
+ return getStringFromWasm0(ret[0], ret[1]);
3764
+ } finally {
3765
+ wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
3766
+ }
3767
+ }
3619
3768
  /**
3620
3769
  * @returns {number}
3621
3770
  */
@@ -3777,6 +3926,14 @@ class WasmCrawlPageResult {
3777
3926
  const len0 = WASM_VECTOR_LEN;
3778
3927
  wasm.wasmcrawlpageresult_set_normalizedUrl(this.__wbg_ptr, ptr0, len0);
3779
3928
  }
3929
+ /**
3930
+ * @param {string} value
3931
+ */
3932
+ set originalUrl(value) {
3933
+ const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
3934
+ const len0 = WASM_VECTOR_LEN;
3935
+ wasm.wasmcrawlpageresult_set_originalUrl(this.__wbg_ptr, ptr0, len0);
3936
+ }
3780
3937
  /**
3781
3938
  * @param {number} value
3782
3939
  */
@@ -5813,15 +5970,18 @@ class WasmLinkInfo {
5813
5970
  * @param {WasmLinkType | null} [linkType]
5814
5971
  * @param {boolean | null} [nofollow]
5815
5972
  * @param {string | null} [rel]
5973
+ * @param {string | null} [originalUrl]
5816
5974
  */
5817
- constructor(url, text, linkType, nofollow, rel) {
5975
+ constructor(url, text, linkType, nofollow, rel, originalUrl) {
5818
5976
  var ptr0 = isLikeNone(url) ? 0 : passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
5819
5977
  var len0 = WASM_VECTOR_LEN;
5820
5978
  var ptr1 = isLikeNone(text) ? 0 : passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
5821
5979
  var len1 = WASM_VECTOR_LEN;
5822
5980
  var ptr2 = isLikeNone(rel) ? 0 : passStringToWasm0(rel, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
5823
5981
  var len2 = WASM_VECTOR_LEN;
5824
- const ret = wasm.wasmlinkinfo_new(ptr0, len0, ptr1, len1, isLikeNone(linkType) ? 4 : linkType, isLikeNone(nofollow) ? 0xFFFFFF : nofollow ? 1 : 0, ptr2, len2);
5982
+ var ptr3 = isLikeNone(originalUrl) ? 0 : passStringToWasm0(originalUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
5983
+ var len3 = WASM_VECTOR_LEN;
5984
+ const ret = wasm.wasmlinkinfo_new(ptr0, len0, ptr1, len1, isLikeNone(linkType) ? 4 : linkType, isLikeNone(nofollow) ? 0xFFFFFF : nofollow ? 1 : 0, ptr2, len2, ptr3, len3);
5825
5985
  this.__wbg_ptr = ret;
5826
5986
  WasmLinkInfoFinalization.register(this, this.__wbg_ptr, this);
5827
5987
  return this;
@@ -5833,6 +5993,21 @@ class WasmLinkInfo {
5833
5993
  const ret = wasm.wasmlinkinfo_nofollow(this.__wbg_ptr);
5834
5994
  return ret !== 0;
5835
5995
  }
5996
+ /**
5997
+ * @returns {string}
5998
+ */
5999
+ get originalUrl() {
6000
+ let deferred1_0;
6001
+ let deferred1_1;
6002
+ try {
6003
+ const ret = wasm.wasmlinkinfo_originalUrl(this.__wbg_ptr);
6004
+ deferred1_0 = ret[0];
6005
+ deferred1_1 = ret[1];
6006
+ return getStringFromWasm0(ret[0], ret[1]);
6007
+ } finally {
6008
+ wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
6009
+ }
6010
+ }
5836
6011
  /**
5837
6012
  * @returns {string | undefined}
5838
6013
  */
@@ -5857,6 +6032,14 @@ class WasmLinkInfo {
5857
6032
  set nofollow(value) {
5858
6033
  wasm.wasmlinkinfo_set_nofollow(this.__wbg_ptr, value);
5859
6034
  }
6035
+ /**
6036
+ * @param {string} value
6037
+ */
6038
+ set originalUrl(value) {
6039
+ const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
6040
+ const len0 = WASM_VECTOR_LEN;
6041
+ wasm.wasmlinkinfo_set_originalUrl(this.__wbg_ptr, ptr0, len0);
6042
+ }
5860
6043
  /**
5861
6044
  * @param {string | null} [value]
5862
6045
  */
@@ -9127,7 +9310,7 @@ function __wbg_get_imports() {
9127
9310
  const a = state0.a;
9128
9311
  state0.a = 0;
9129
9312
  try {
9130
- return wasm_bindgen__convert__closures_____invoke__h8ad11faaa2905452(a, state0.b, arg0, arg1);
9313
+ return wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined_______true_(a, state0.b, arg0, arg1);
9131
9314
  } finally {
9132
9315
  state0.a = a;
9133
9316
  }
@@ -9421,13 +9604,13 @@ function __wbg_get_imports() {
9421
9604
  return ret;
9422
9605
  },
9423
9606
  __wbindgen_generic_0000000000000001: function(arg0, arg1) {
9424
- // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1331, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
9425
- const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h2eedacf566f3748c);
9607
+ // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1467, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
9608
+ const ret = makeMutClosure(arg0, arg1, wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___wasm_bindgen_ceaac9754e66b2fe___JsValue__core_608f92abc48d28da___result__Result_____wasm_bindgen_ceaac9754e66b2fe___JsError___true_);
9426
9609
  return ret;
9427
9610
  },
9428
9611
  __wbindgen_generic_0000000000000002: function(arg0, arg1) {
9429
- // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1222, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
9430
- const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h3a7ccac3e94bbb5c);
9612
+ // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1309, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
9613
+ const ret = makeMutClosure(arg0, arg1, wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke_______true_);
9431
9614
  return ret;
9432
9615
  },
9433
9616
  __wbindgen_generic_0000000000000003: function(arg0) {
@@ -9466,19 +9649,19 @@ function __wbg_get_imports() {
9466
9649
  };
9467
9650
  }
9468
9651
 
9469
- function wasm_bindgen__convert__closures_____invoke__h3a7ccac3e94bbb5c(arg0, arg1) {
9470
- wasm.wasm_bindgen__convert__closures_____invoke__h3a7ccac3e94bbb5c(arg0, arg1);
9652
+ function wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke_______true_(arg0, arg1) {
9653
+ wasm.wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke_______true_(arg0, arg1);
9471
9654
  }
9472
9655
 
9473
- function wasm_bindgen__convert__closures_____invoke__h2eedacf566f3748c(arg0, arg1, arg2) {
9474
- const ret = wasm.wasm_bindgen__convert__closures_____invoke__h2eedacf566f3748c(arg0, arg1, arg2);
9656
+ function wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___wasm_bindgen_ceaac9754e66b2fe___JsValue__core_608f92abc48d28da___result__Result_____wasm_bindgen_ceaac9754e66b2fe___JsError___true_(arg0, arg1, arg2) {
9657
+ const ret = wasm.wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___wasm_bindgen_ceaac9754e66b2fe___JsValue__core_608f92abc48d28da___result__Result_____wasm_bindgen_ceaac9754e66b2fe___JsError___true_(arg0, arg1, arg2);
9475
9658
  if (ret[1]) {
9476
9659
  throw takeFromExternrefTable0(ret[0]);
9477
9660
  }
9478
9661
  }
9479
9662
 
9480
- function wasm_bindgen__convert__closures_____invoke__h8ad11faaa2905452(arg0, arg1, arg2, arg3) {
9481
- wasm.wasm_bindgen__convert__closures_____invoke__h8ad11faaa2905452(arg0, arg1, arg2, arg3);
9663
+ function wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined_______true_(arg0, arg1, arg2, arg3) {
9664
+ wasm.wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined_______true_(arg0, arg1, arg2, arg3);
9482
9665
  }
9483
9666
 
9484
9667
 
Binary file