@xberg-io/crawlberg-wasm 1.7.2 → 1.8.0
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.
- package/package.json +1 -1
- package/pkg/bundler/crawlberg_wasm.d.ts +20 -4
- package/pkg/bundler/crawlberg_wasm_bg.js +351 -137
- package/pkg/bundler/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/bundler/crawlberg_wasm_bg.wasm.d.ts +35 -7
- package/pkg/bundler/package.json +1 -1
- package/pkg/deno/crawlberg_wasm.d.ts +20 -4
- package/pkg/deno/crawlberg_wasm.js +351 -137
- package/pkg/deno/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/deno/crawlberg_wasm_bg.wasm.d.ts +35 -7
- package/pkg/nodejs/crawlberg_wasm.d.ts +20 -4
- package/pkg/nodejs/crawlberg_wasm.js +351 -137
- package/pkg/nodejs/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/nodejs/crawlberg_wasm_bg.wasm.d.ts +35 -7
- package/pkg/nodejs/package.json +1 -1
- package/pkg/web/crawlberg_wasm.d.ts +55 -11
- package/pkg/web/crawlberg_wasm.js +351 -137
- package/pkg/web/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/web/crawlberg_wasm_bg.wasm.d.ts +35 -7
- package/pkg/web/package.json +1 -1
|
Binary file
|
|
@@ -129,7 +129,8 @@ export const wasmbrowserconfig_endpoint: (a: number) => [number, number];
|
|
|
129
129
|
export const wasmbrowserconfig_evalScript: (a: number) => [number, number];
|
|
130
130
|
export const wasmbrowserconfig_extraWait: (a: number) => [number, bigint];
|
|
131
131
|
export const wasmbrowserconfig_mode: (a: number) => [number, number];
|
|
132
|
-
export const wasmbrowserconfig_new: (a: number, b: number, c: number, d: bigint, e: number, f:
|
|
132
|
+
export const wasmbrowserconfig_new: (a: number, b: number, c: number, d: bigint, e: number, f: bigint, g: number, h: bigint, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: bigint, t: number, u: number, v: number, w: number, x: number) => number;
|
|
133
|
+
export const wasmbrowserconfig_overallTimeout: (a: number) => [number, bigint];
|
|
133
134
|
export const wasmbrowserconfig_proxy: (a: number) => number;
|
|
134
135
|
export const wasmbrowserconfig_robotsUserAgent: (a: number) => [number, number];
|
|
135
136
|
export const wasmbrowserconfig_sessionAffinity: (a: number) => number;
|
|
@@ -140,12 +141,15 @@ export const wasmbrowserconfig_set_endpoint: (a: number, b: number, c: number) =
|
|
|
140
141
|
export const wasmbrowserconfig_set_evalScript: (a: number, b: number, c: number) => void;
|
|
141
142
|
export const wasmbrowserconfig_set_extraWait: (a: number, b: number, c: bigint) => void;
|
|
142
143
|
export const wasmbrowserconfig_set_mode: (a: number, b: number) => void;
|
|
144
|
+
export const wasmbrowserconfig_set_overallTimeout: (a: number, b: number, c: bigint) => void;
|
|
143
145
|
export const wasmbrowserconfig_set_proxy: (a: number, b: number) => void;
|
|
144
146
|
export const wasmbrowserconfig_set_robotsUserAgent: (a: number, b: number, c: number) => void;
|
|
145
147
|
export const wasmbrowserconfig_set_sessionAffinity: (a: number, b: number) => void;
|
|
148
|
+
export const wasmbrowserconfig_set_shutdownTimeout: (a: number, b: number, c: bigint) => void;
|
|
146
149
|
export const wasmbrowserconfig_set_timeout: (a: number, b: number, c: bigint) => void;
|
|
147
150
|
export const wasmbrowserconfig_set_wait: (a: number, b: number) => void;
|
|
148
151
|
export const wasmbrowserconfig_set_waitSelector: (a: number, b: number, c: number) => void;
|
|
152
|
+
export const wasmbrowserconfig_shutdownTimeout: (a: number) => [number, bigint];
|
|
149
153
|
export const wasmbrowserconfig_timeout: (a: number) => [number, bigint];
|
|
150
154
|
export const wasmbrowserconfig_wait: (a: number) => [number, number];
|
|
151
155
|
export const wasmbrowserconfig_waitSelector: (a: number) => [number, number];
|
|
@@ -173,15 +177,17 @@ export const wasmcitationresult_set_content: (a: number, b: number, c: number) =
|
|
|
173
177
|
export const wasmcitationresult_set_references: (a: number, b: number, c: number) => void;
|
|
174
178
|
export const wasmcontentconfig_default: () => number;
|
|
175
179
|
export const wasmcontentconfig_excludeSelectors: (a: number) => [number, number];
|
|
180
|
+
export const wasmcontentconfig_extractMetadata: (a: number) => number;
|
|
176
181
|
export const wasmcontentconfig_includeDocumentStructure: (a: number) => number;
|
|
177
182
|
export const wasmcontentconfig_maxDepth: (a: number) => number;
|
|
178
|
-
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) => number;
|
|
183
|
+
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;
|
|
179
184
|
export const wasmcontentconfig_outputFormat: (a: number) => [number, number];
|
|
180
185
|
export const wasmcontentconfig_preprocessingPreset: (a: number) => [number, number];
|
|
181
186
|
export const wasmcontentconfig_preserveTags: (a: number) => [number, number];
|
|
182
187
|
export const wasmcontentconfig_removeForms: (a: number) => number;
|
|
183
188
|
export const wasmcontentconfig_removeNavigation: (a: number) => number;
|
|
184
189
|
export const wasmcontentconfig_set_excludeSelectors: (a: number, b: number, c: number) => void;
|
|
190
|
+
export const wasmcontentconfig_set_extractMetadata: (a: number, b: number) => void;
|
|
185
191
|
export const wasmcontentconfig_set_includeDocumentStructure: (a: number, b: number) => void;
|
|
186
192
|
export const wasmcontentconfig_set_maxDepth: (a: number, b: number) => void;
|
|
187
193
|
export const wasmcontentconfig_set_outputFormat: (a: number, b: number, c: number) => void;
|
|
@@ -220,6 +226,7 @@ export const wasmcrawlconfig_contentFilter: (a: number) => [number, number];
|
|
|
220
226
|
export const wasmcrawlconfig_cookiesEnabled: (a: number) => number;
|
|
221
227
|
export const wasmcrawlconfig_crawlStrategy: (a: number) => [number, number];
|
|
222
228
|
export const wasmcrawlconfig_customHeaders: (a: number) => any;
|
|
229
|
+
export const wasmcrawlconfig_dedupIncludeQuery: (a: number) => number;
|
|
223
230
|
export const wasmcrawlconfig_default: () => number;
|
|
224
231
|
export const wasmcrawlconfig_documentContentEncoding: (a: number) => [number, number];
|
|
225
232
|
export const wasmcrawlconfig_documentMaxSize: (a: number) => number;
|
|
@@ -240,14 +247,18 @@ export const wasmcrawlconfig_maxDepth: (a: number) => number;
|
|
|
240
247
|
export const wasmcrawlconfig_maxLinksPerPage: (a: number) => number;
|
|
241
248
|
export const wasmcrawlconfig_maxPages: (a: number) => number;
|
|
242
249
|
export const wasmcrawlconfig_maxRedirects: (a: number) => number;
|
|
243
|
-
export const wasmcrawlconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l:
|
|
250
|
+
export const wasmcrawlconfig_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: bigint, r: number, s: number, t: number, u: number, v: number, w: bigint, x: number, y: bigint, 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, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: number, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number, e2: bigint, f2: number, g2: number, h2: number, i2: number, j2: number, k2: number, l2: number, m2: number, n2: number, o2: number, p2: number, q2: number, r2: number, s2: number, t2: number, u2: number, v2: number) => number;
|
|
251
|
+
export const wasmcrawlconfig_pathPatternsMatchQuery: (a: number) => number;
|
|
244
252
|
export const wasmcrawlconfig_proxy: (a: number) => number;
|
|
253
|
+
export const wasmcrawlconfig_rateLimitJitterRatio: (a: number) => number;
|
|
245
254
|
export const wasmcrawlconfig_rateLimitMs: (a: number) => [number, bigint];
|
|
246
255
|
export const wasmcrawlconfig_removeTags: (a: number) => [number, number];
|
|
247
256
|
export const wasmcrawlconfig_requestTimeout: (a: number) => [number, bigint];
|
|
248
257
|
export const wasmcrawlconfig_respectRobotsTxt: (a: number) => number;
|
|
249
258
|
export const wasmcrawlconfig_retryCodes: (a: number) => [number, number];
|
|
250
259
|
export const wasmcrawlconfig_retryCount: (a: number) => number;
|
|
260
|
+
export const wasmcrawlconfig_retryInitialDelayMs: (a: number) => bigint;
|
|
261
|
+
export const wasmcrawlconfig_retryMaxDelayMs: (a: number) => bigint;
|
|
251
262
|
export const wasmcrawlconfig_saveBrowserProfile: (a: number) => number;
|
|
252
263
|
export const wasmcrawlconfig_set_allowSubdomains: (a: number, b: number) => void;
|
|
253
264
|
export const wasmcrawlconfig_set_assetTypes: (a: number, b: number, c: number) => void;
|
|
@@ -262,6 +273,7 @@ export const wasmcrawlconfig_set_contentFilter: (a: number, b: number) => void;
|
|
|
262
273
|
export const wasmcrawlconfig_set_cookiesEnabled: (a: number, b: number) => void;
|
|
263
274
|
export const wasmcrawlconfig_set_crawlStrategy: (a: number, b: number) => void;
|
|
264
275
|
export const wasmcrawlconfig_set_customHeaders: (a: number, b: any) => void;
|
|
276
|
+
export const wasmcrawlconfig_set_dedupIncludeQuery: (a: number, b: number) => void;
|
|
265
277
|
export const wasmcrawlconfig_set_documentContentEncoding: (a: number, b: number) => void;
|
|
266
278
|
export const wasmcrawlconfig_set_documentMaxSize: (a: number, b: number) => void;
|
|
267
279
|
export const wasmcrawlconfig_set_documentMimeTypes: (a: number, b: number, c: number) => void;
|
|
@@ -281,18 +293,24 @@ export const wasmcrawlconfig_set_maxDepth: (a: number, b: number) => void;
|
|
|
281
293
|
export const wasmcrawlconfig_set_maxLinksPerPage: (a: number, b: number) => void;
|
|
282
294
|
export const wasmcrawlconfig_set_maxPages: (a: number, b: number) => void;
|
|
283
295
|
export const wasmcrawlconfig_set_maxRedirects: (a: number, b: number) => void;
|
|
296
|
+
export const wasmcrawlconfig_set_pathPatternsMatchQuery: (a: number, b: number) => void;
|
|
284
297
|
export const wasmcrawlconfig_set_proxy: (a: number, b: number) => void;
|
|
298
|
+
export const wasmcrawlconfig_set_rateLimitJitterRatio: (a: number, b: number) => void;
|
|
285
299
|
export const wasmcrawlconfig_set_rateLimitMs: (a: number, b: number, c: bigint) => void;
|
|
286
300
|
export const wasmcrawlconfig_set_removeTags: (a: number, b: number, c: number) => void;
|
|
287
301
|
export const wasmcrawlconfig_set_requestTimeout: (a: number, b: number, c: bigint) => void;
|
|
288
302
|
export const wasmcrawlconfig_set_respectRobotsTxt: (a: number, b: number) => void;
|
|
289
303
|
export const wasmcrawlconfig_set_retryCodes: (a: number, b: number, c: number) => void;
|
|
290
304
|
export const wasmcrawlconfig_set_retryCount: (a: number, b: number) => void;
|
|
305
|
+
export const wasmcrawlconfig_set_retryInitialDelayMs: (a: number, b: bigint) => void;
|
|
306
|
+
export const wasmcrawlconfig_set_retryMaxDelayMs: (a: number, b: bigint) => void;
|
|
291
307
|
export const wasmcrawlconfig_set_saveBrowserProfile: (a: number, b: number) => void;
|
|
292
308
|
export const wasmcrawlconfig_set_softHttpErrors: (a: number, b: number) => void;
|
|
293
309
|
export const wasmcrawlconfig_set_ssrf: (a: number, b: number) => void;
|
|
294
310
|
export const wasmcrawlconfig_set_ssrfDenyPrivateExplicit: (a: number, b: number) => void;
|
|
295
311
|
export const wasmcrawlconfig_set_stayOnDomain: (a: number, b: number) => void;
|
|
312
|
+
export const wasmcrawlconfig_set_stripTrackingParams: (a: number, b: number) => void;
|
|
313
|
+
export const wasmcrawlconfig_set_trackingParams: (a: number, b: number, c: number) => void;
|
|
296
314
|
export const wasmcrawlconfig_set_userAgent: (a: number, b: number, c: number) => void;
|
|
297
315
|
export const wasmcrawlconfig_set_userAgents: (a: number, b: number, c: number) => void;
|
|
298
316
|
export const wasmcrawlconfig_set_warcOutput: (a: number, b: number, c: number) => void;
|
|
@@ -300,6 +318,8 @@ export const wasmcrawlconfig_softHttpErrors: (a: number) => number;
|
|
|
300
318
|
export const wasmcrawlconfig_ssrf: (a: number) => number;
|
|
301
319
|
export const wasmcrawlconfig_ssrfDenyPrivateExplicit: (a: number) => number;
|
|
302
320
|
export const wasmcrawlconfig_stayOnDomain: (a: number) => number;
|
|
321
|
+
export const wasmcrawlconfig_stripTrackingParams: (a: number) => number;
|
|
322
|
+
export const wasmcrawlconfig_trackingParams: (a: number) => [number, number];
|
|
303
323
|
export const wasmcrawlconfig_userAgent: (a: number) => [number, number];
|
|
304
324
|
export const wasmcrawlconfig_userAgents: (a: number) => [number, number];
|
|
305
325
|
export const wasmcrawlconfig_validate: (a: number) => [number, number];
|
|
@@ -314,6 +334,7 @@ export const wasmcrawlpageresult_downloadedDocument: (a: number) => number;
|
|
|
314
334
|
export const wasmcrawlpageresult_extractedData: (a: number) => any;
|
|
315
335
|
export const wasmcrawlpageresult_extractionMeta: (a: number) => number;
|
|
316
336
|
export const wasmcrawlpageresult_feeds: (a: number) => [number, number];
|
|
337
|
+
export const wasmcrawlpageresult_finalUrl: (a: number) => [number, number];
|
|
317
338
|
export const wasmcrawlpageresult_html: (a: number) => [number, number];
|
|
318
339
|
export const wasmcrawlpageresult_images: (a: number) => [number, number];
|
|
319
340
|
export const wasmcrawlpageresult_isPdf: (a: number) => number;
|
|
@@ -321,8 +342,11 @@ export const wasmcrawlpageresult_jsonLd: (a: number) => [number, number];
|
|
|
321
342
|
export const wasmcrawlpageresult_links: (a: number) => [number, number];
|
|
322
343
|
export const wasmcrawlpageresult_markdown: (a: number) => number;
|
|
323
344
|
export const wasmcrawlpageresult_metadata: (a: number) => number;
|
|
324
|
-
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) => number;
|
|
345
|
+
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) => number;
|
|
346
|
+
export const wasmcrawlpageresult_nofollowDetected: (a: number) => number;
|
|
347
|
+
export const wasmcrawlpageresult_noindexDetected: (a: number) => number;
|
|
325
348
|
export const wasmcrawlpageresult_normalizedUrl: (a: number) => [number, number];
|
|
349
|
+
export const wasmcrawlpageresult_redirectCount: (a: number) => number;
|
|
326
350
|
export const wasmcrawlpageresult_set_bodySize: (a: number, b: number) => void;
|
|
327
351
|
export const wasmcrawlpageresult_set_browserUsed: (a: number, b: number) => void;
|
|
328
352
|
export const wasmcrawlpageresult_set_contentType: (a: number, b: number, c: number) => void;
|
|
@@ -332,6 +356,7 @@ export const wasmcrawlpageresult_set_downloadedDocument: (a: number, b: number)
|
|
|
332
356
|
export const wasmcrawlpageresult_set_extractedData: (a: number, b: number) => void;
|
|
333
357
|
export const wasmcrawlpageresult_set_extractionMeta: (a: number, b: number) => void;
|
|
334
358
|
export const wasmcrawlpageresult_set_feeds: (a: number, b: number, c: number) => void;
|
|
359
|
+
export const wasmcrawlpageresult_set_finalUrl: (a: number, b: number, c: number) => void;
|
|
335
360
|
export const wasmcrawlpageresult_set_html: (a: number, b: number, c: number) => void;
|
|
336
361
|
export const wasmcrawlpageresult_set_images: (a: number, b: number, c: number) => void;
|
|
337
362
|
export const wasmcrawlpageresult_set_isPdf: (a: number, b: number) => void;
|
|
@@ -339,7 +364,10 @@ export const wasmcrawlpageresult_set_jsonLd: (a: number, b: number, c: number) =
|
|
|
339
364
|
export const wasmcrawlpageresult_set_links: (a: number, b: number, c: number) => void;
|
|
340
365
|
export const wasmcrawlpageresult_set_markdown: (a: number, b: number) => void;
|
|
341
366
|
export const wasmcrawlpageresult_set_metadata: (a: number, b: number) => void;
|
|
367
|
+
export const wasmcrawlpageresult_set_nofollowDetected: (a: number, b: number) => void;
|
|
368
|
+
export const wasmcrawlpageresult_set_noindexDetected: (a: number, b: number) => void;
|
|
342
369
|
export const wasmcrawlpageresult_set_normalizedUrl: (a: number, b: number, c: number) => void;
|
|
370
|
+
export const wasmcrawlpageresult_set_redirectCount: (a: number, b: number) => void;
|
|
343
371
|
export const wasmcrawlpageresult_set_statusCode: (a: number, b: number) => void;
|
|
344
372
|
export const wasmcrawlpageresult_set_stayedOnDomain: (a: number, b: number) => void;
|
|
345
373
|
export const wasmcrawlpageresult_set_url: (a: number, b: number, c: number) => void;
|
|
@@ -722,9 +750,9 @@ export const wasmssrfpolicy_set_allowlist: (a: number, b: any) => void;
|
|
|
722
750
|
export const wasmssrfpolicy_set_denyPrivate: (a: number, b: number) => void;
|
|
723
751
|
export const wasmssrfpolicy_set_maxRedirects: (a: number, b: number) => void;
|
|
724
752
|
export const wasmssrfpolicy_set_schemeAllowlist: (a: number, b: number, c: number) => void;
|
|
725
|
-
export const
|
|
726
|
-
export const
|
|
727
|
-
export const
|
|
753
|
+
export const wasm_bindgen__convert__closures_____invoke__h2eedacf566f3748c: (a: number, b: number, c: any) => [number, number];
|
|
754
|
+
export const wasm_bindgen__convert__closures_____invoke__h8ad11faaa2905452: (a: number, b: number, c: any, d: any) => void;
|
|
755
|
+
export const wasm_bindgen__convert__closures_____invoke__h21c8118e1fe083d1: (a: number, b: number) => void;
|
|
728
756
|
export const __wbindgen_malloc: (a: number, b: number) => number;
|
|
729
757
|
export const __wbindgen_realloc: (a: number, b: number, c: number, d: number) => number;
|
|
730
758
|
export const __wbindgen_exn_store: (a: number) => void;
|
|
@@ -153,7 +153,7 @@ export class WasmBrowserConfig {
|
|
|
153
153
|
free(): void;
|
|
154
154
|
[Symbol.dispose](): void;
|
|
155
155
|
static default(): WasmBrowserConfig;
|
|
156
|
-
constructor(mode?: WasmBrowserMode | null, backend?: WasmBrowserBackend | null, timeout?: bigint | null, wait?: WasmBrowserWait | null, blockUrlPatterns?: string[] | null, captureNetworkEvents?: boolean | null, sessionAffinity?: boolean | null, endpoint?: string | null, waitSelector?: string | null, extraWait?: bigint | null, proxy?: WasmProxyConfig | null, evalScript?: string | null, robotsUserAgent?: string | null);
|
|
156
|
+
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, endpoint?: string | null, waitSelector?: string | null, extraWait?: bigint | null, proxy?: WasmProxyConfig | null, evalScript?: string | null, robotsUserAgent?: string | null);
|
|
157
157
|
get backend(): string;
|
|
158
158
|
set backend(value: WasmBrowserBackend);
|
|
159
159
|
blockUrlPatterns: string[];
|
|
@@ -166,11 +166,15 @@ export class WasmBrowserConfig {
|
|
|
166
166
|
set extraWait(value: bigint | null | undefined);
|
|
167
167
|
get mode(): string;
|
|
168
168
|
set mode(value: WasmBrowserMode);
|
|
169
|
+
get overallTimeout(): bigint | undefined;
|
|
170
|
+
set overallTimeout(value: bigint | null | undefined);
|
|
169
171
|
get proxy(): WasmProxyConfig | undefined;
|
|
170
172
|
set proxy(value: WasmProxyConfig | null | undefined);
|
|
171
173
|
get robotsUserAgent(): string | undefined;
|
|
172
174
|
set robotsUserAgent(value: string | null | undefined);
|
|
173
175
|
sessionAffinity: boolean;
|
|
176
|
+
get shutdownTimeout(): bigint | undefined;
|
|
177
|
+
set shutdownTimeout(value: bigint | null | undefined);
|
|
174
178
|
get timeout(): bigint | undefined;
|
|
175
179
|
set timeout(value: bigint | null | undefined);
|
|
176
180
|
get waitSelector(): string | undefined;
|
|
@@ -251,8 +255,9 @@ export class WasmContentConfig {
|
|
|
251
255
|
free(): void;
|
|
252
256
|
[Symbol.dispose](): void;
|
|
253
257
|
static default(): WasmContentConfig;
|
|
254
|
-
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, maxDepth?: number | null);
|
|
258
|
+
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);
|
|
255
259
|
excludeSelectors: string[];
|
|
260
|
+
extractMetadata: boolean;
|
|
256
261
|
includeDocumentStructure: boolean;
|
|
257
262
|
get maxDepth(): number | undefined;
|
|
258
263
|
set maxDepth(value: number | null | undefined);
|
|
@@ -297,7 +302,7 @@ export class WasmCrawlConfig {
|
|
|
297
302
|
free(): void;
|
|
298
303
|
[Symbol.dispose](): void;
|
|
299
304
|
static default(): WasmCrawlConfig;
|
|
300
|
-
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);
|
|
305
|
+
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, 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);
|
|
301
306
|
/**
|
|
302
307
|
* Validate the configuration, returning an error if any values are invalid.
|
|
303
308
|
*
|
|
@@ -324,6 +329,7 @@ export class WasmCrawlConfig {
|
|
|
324
329
|
get crawlStrategy(): string;
|
|
325
330
|
set crawlStrategy(value: WasmCrawlStrategyKind);
|
|
326
331
|
customHeaders: any;
|
|
332
|
+
dedupIncludeQuery: boolean;
|
|
327
333
|
get documentContentEncoding(): string | undefined;
|
|
328
334
|
set documentContentEncoding(value: WasmDocumentContentEncoding | null | undefined);
|
|
329
335
|
get documentMaxSize(): number | undefined;
|
|
@@ -355,8 +361,10 @@ export class WasmCrawlConfig {
|
|
|
355
361
|
get maxPages(): number | undefined;
|
|
356
362
|
set maxPages(value: number | null | undefined);
|
|
357
363
|
maxRedirects: number;
|
|
364
|
+
pathPatternsMatchQuery: boolean;
|
|
358
365
|
get proxy(): WasmProxyConfig | undefined;
|
|
359
366
|
set proxy(value: WasmProxyConfig | null | undefined);
|
|
367
|
+
rateLimitJitterRatio: number;
|
|
360
368
|
get rateLimitMs(): bigint | undefined;
|
|
361
369
|
set rateLimitMs(value: bigint | null | undefined);
|
|
362
370
|
removeTags: string[];
|
|
@@ -365,12 +373,16 @@ export class WasmCrawlConfig {
|
|
|
365
373
|
respectRobotsTxt: boolean;
|
|
366
374
|
retryCodes: Uint16Array;
|
|
367
375
|
retryCount: number;
|
|
376
|
+
retryInitialDelayMs: bigint;
|
|
377
|
+
retryMaxDelayMs: bigint;
|
|
368
378
|
saveBrowserProfile: boolean;
|
|
369
379
|
softHttpErrors: boolean;
|
|
370
380
|
get ssrfDenyPrivateExplicit(): boolean | undefined;
|
|
371
381
|
set ssrfDenyPrivateExplicit(value: boolean | null | undefined);
|
|
372
382
|
ssrf: WasmSsrfPolicy;
|
|
373
383
|
stayOnDomain: boolean;
|
|
384
|
+
stripTrackingParams: boolean;
|
|
385
|
+
trackingParams: string[];
|
|
374
386
|
get userAgent(): string | undefined;
|
|
375
387
|
set userAgent(value: string | null | undefined);
|
|
376
388
|
userAgents: string[];
|
|
@@ -397,7 +409,7 @@ export class WasmCrawlPageResult {
|
|
|
397
409
|
free(): void;
|
|
398
410
|
[Symbol.dispose](): void;
|
|
399
411
|
static default(): WasmCrawlPageResult;
|
|
400
|
-
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, detectedCharset?: string | null, markdown?: WasmMarkdownResult | null, extractedData?: any | null, extractionMeta?: WasmExtractionMeta | null, downloadedDocument?: WasmDownloadedDocument | null);
|
|
412
|
+
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, detectedCharset?: string | null, markdown?: WasmMarkdownResult | null, extractedData?: any | null, extractionMeta?: WasmExtractionMeta | null, downloadedDocument?: WasmDownloadedDocument | null);
|
|
401
413
|
bodySize: number;
|
|
402
414
|
browserUsed: boolean;
|
|
403
415
|
contentType: string;
|
|
@@ -411,6 +423,7 @@ export class WasmCrawlPageResult {
|
|
|
411
423
|
get extractionMeta(): WasmExtractionMeta | undefined;
|
|
412
424
|
set extractionMeta(value: WasmExtractionMeta | null | undefined);
|
|
413
425
|
feeds: WasmFeedInfo[];
|
|
426
|
+
finalUrl: string;
|
|
414
427
|
html: string;
|
|
415
428
|
images: WasmImageInfo[];
|
|
416
429
|
isPdf: boolean;
|
|
@@ -419,7 +432,10 @@ export class WasmCrawlPageResult {
|
|
|
419
432
|
get markdown(): WasmMarkdownResult | undefined;
|
|
420
433
|
set markdown(value: WasmMarkdownResult | null | undefined);
|
|
421
434
|
metadata: WasmPageMetadata;
|
|
435
|
+
nofollowDetected: boolean;
|
|
436
|
+
noindexDetected: boolean;
|
|
422
437
|
normalizedUrl: string;
|
|
438
|
+
redirectCount: number;
|
|
423
439
|
statusCode: number;
|
|
424
440
|
stayedOnDomain: boolean;
|
|
425
441
|
url: string;
|