@xberg-io/crawlberg-wasm 1.1.4 → 1.3.2
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 +11 -7
- package/pkg/bundler/crawlberg_wasm.d.ts +106 -5
- package/pkg/bundler/crawlberg_wasm.js +1 -1
- package/pkg/bundler/crawlberg_wasm_bg.js +676 -253
- package/pkg/bundler/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/bundler/crawlberg_wasm_bg.wasm.d.ts +52 -17
- package/pkg/bundler/package.json +1 -1
- package/pkg/deno/crawlberg_wasm.d.ts +106 -5
- package/pkg/deno/crawlberg_wasm.js +676 -253
- package/pkg/deno/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/deno/crawlberg_wasm_bg.wasm.d.ts +52 -17
- package/pkg/nodejs/crawlberg_wasm.d.ts +106 -5
- package/pkg/nodejs/crawlberg_wasm.js +680 -253
- package/pkg/nodejs/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/nodejs/crawlberg_wasm_bg.wasm.d.ts +52 -17
- package/pkg/nodejs/package.json +1 -1
- package/pkg/web/crawlberg_wasm.d.ts +158 -22
- package/pkg/web/crawlberg_wasm.js +681 -254
- package/pkg/web/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/web/crawlberg_wasm_bg.wasm.d.ts +52 -17
- package/pkg/web/package.json +1 -1
|
@@ -267,6 +267,13 @@ export class WasmContentConfig {
|
|
|
267
267
|
wrapWidth: number;
|
|
268
268
|
}
|
|
269
269
|
|
|
270
|
+
/**
|
|
271
|
+
* Content filter applied to each crawled page before it reaches the result.
|
|
272
|
+
*/
|
|
273
|
+
export enum WasmContentFilterKind {
|
|
274
|
+
Bm25 = 0,
|
|
275
|
+
}
|
|
276
|
+
|
|
270
277
|
/**
|
|
271
278
|
* Information about an HTTP cookie received from a response.
|
|
272
279
|
*/
|
|
@@ -290,7 +297,7 @@ export class WasmCrawlConfig {
|
|
|
290
297
|
free(): void;
|
|
291
298
|
[Symbol.dispose](): void;
|
|
292
299
|
static default(): WasmCrawlConfig;
|
|
293
|
-
constructor(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?: any[] | 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, maxConcurrent?: 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, warcOutput?: string | null, browserProfile?: string | null);
|
|
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?: any[] | 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);
|
|
294
301
|
/**
|
|
295
302
|
* Validate the configuration, returning an error if any values are invalid.
|
|
296
303
|
*/
|
|
@@ -299,16 +306,28 @@ export class WasmCrawlConfig {
|
|
|
299
306
|
assetTypes: string[];
|
|
300
307
|
get auth(): any | undefined;
|
|
301
308
|
set auth(value: any | null | undefined);
|
|
309
|
+
get bm25Query(): string | undefined;
|
|
310
|
+
set bm25Query(value: string | null | undefined);
|
|
311
|
+
get bm25Threshold(): number | undefined;
|
|
312
|
+
set bm25Threshold(value: number | null | undefined);
|
|
302
313
|
browser: WasmBrowserConfig;
|
|
303
314
|
get browserProfile(): string | undefined;
|
|
304
315
|
set browserProfile(value: string | null | undefined);
|
|
305
316
|
captureScreenshot: boolean;
|
|
306
317
|
content: WasmContentConfig;
|
|
318
|
+
get contentFilter(): string | undefined;
|
|
319
|
+
set contentFilter(value: WasmContentFilterKind | null | undefined);
|
|
307
320
|
cookiesEnabled: boolean;
|
|
321
|
+
get crawlStrategy(): string;
|
|
322
|
+
set crawlStrategy(value: WasmCrawlStrategyKind);
|
|
308
323
|
customHeaders: any;
|
|
324
|
+
get documentContentEncoding(): string | undefined;
|
|
325
|
+
set documentContentEncoding(value: WasmDocumentContentEncoding | null | undefined);
|
|
309
326
|
get documentMaxSize(): number | undefined;
|
|
310
327
|
set documentMaxSize(value: number | null | undefined);
|
|
311
328
|
documentMimeTypes: string[];
|
|
329
|
+
get documentOutputDir(): string | undefined;
|
|
330
|
+
set documentOutputDir(value: string | null | undefined);
|
|
312
331
|
get documentUrlDepth(): number | undefined;
|
|
313
332
|
set documentUrlDepth(value: number | null | undefined);
|
|
314
333
|
downloadAssets: boolean;
|
|
@@ -328,6 +347,8 @@ export class WasmCrawlConfig {
|
|
|
328
347
|
set maxConcurrent(value: number | null | undefined);
|
|
329
348
|
get maxDepth(): number | undefined;
|
|
330
349
|
set maxDepth(value: number | null | undefined);
|
|
350
|
+
get maxLinksPerPage(): number | undefined;
|
|
351
|
+
set maxLinksPerPage(value: number | null | undefined);
|
|
331
352
|
get maxPages(): number | undefined;
|
|
332
353
|
set maxPages(value: number | null | undefined);
|
|
333
354
|
maxRedirects: number;
|
|
@@ -344,6 +365,8 @@ export class WasmCrawlConfig {
|
|
|
344
365
|
saveBrowserProfile: boolean;
|
|
345
366
|
softHttpErrors: boolean;
|
|
346
367
|
ssrf: WasmSsrfPolicy;
|
|
368
|
+
get ssrfDenyPrivateExplicit(): boolean | undefined;
|
|
369
|
+
set ssrfDenyPrivateExplicit(value: boolean | null | undefined);
|
|
347
370
|
stayOnDomain: boolean;
|
|
348
371
|
get userAgent(): string | undefined;
|
|
349
372
|
set userAgent(value: string | null | undefined);
|
|
@@ -410,6 +433,13 @@ export class WasmCrawlResult {
|
|
|
410
433
|
constructor(pages?: WasmCrawlPageResult[] | null, finalUrl?: string | null, redirectCount?: number | null, wasSkipped?: boolean | null, cookies?: WasmCookieInfo[] | null, stayedOnDomain?: boolean | null, browserUsed?: boolean | null, error?: string | null);
|
|
411
434
|
/**
|
|
412
435
|
* Returns the count of unique normalized URLs encountered during crawling.
|
|
436
|
+
*
|
|
437
|
+
* Computed from `pages` (not the deprecated `normalized_urls` field) so it
|
|
438
|
+
* is correct across every binding that reconstructs `CrawlResult` from
|
|
439
|
+
* `pages` alone. In streaming mode `pages` is empty, so this returns 0 on
|
|
440
|
+
* the opaque-handle (C/Go/C#/Zig/Dart) path where it previously counted
|
|
441
|
+
* streamed pages — a known, accepted cost of making the other ten binding
|
|
442
|
+
* families correct.
|
|
413
443
|
*/
|
|
414
444
|
uniqueNormalizedUrls(): number;
|
|
415
445
|
browserUsed: boolean;
|
|
@@ -423,6 +453,32 @@ export class WasmCrawlResult {
|
|
|
423
453
|
wasSkipped: boolean;
|
|
424
454
|
}
|
|
425
455
|
|
|
456
|
+
/**
|
|
457
|
+
* Traversal order for a crawl.
|
|
458
|
+
*
|
|
459
|
+
* Selects both the queue discipline and the selection strategy, because global order is a
|
|
460
|
+
* property of the frontier: the engine hands its bounded selection window to the strategy, so
|
|
461
|
+
* a strategy alone can only reorder URLs that have already been dequeued.
|
|
462
|
+
*/
|
|
463
|
+
export enum WasmCrawlStrategyKind {
|
|
464
|
+
Bfs = 0,
|
|
465
|
+
Dfs = 1,
|
|
466
|
+
BestFirst = 2,
|
|
467
|
+
Adaptive = 3,
|
|
468
|
+
}
|
|
469
|
+
|
|
470
|
+
/**
|
|
471
|
+
* Opt-in encoding applied to a downloaded document's bytes for callers who need the
|
|
472
|
+
* content available in a serializable field rather than reading it from disk.
|
|
473
|
+
*
|
|
474
|
+
* `None` (the `CrawlConfig.document_content_encoding` default) produces neither — unlike
|
|
475
|
+
* screenshots, base64-encoding a document by default would duplicate an already
|
|
476
|
+
* up-to-`document_max_size` buffer (50 MB default) in memory per document.
|
|
477
|
+
*/
|
|
478
|
+
export enum WasmDocumentContentEncoding {
|
|
479
|
+
Base64 = 0,
|
|
480
|
+
}
|
|
481
|
+
|
|
426
482
|
/**
|
|
427
483
|
* A downloaded asset from a page.
|
|
428
484
|
*/
|
|
@@ -453,13 +509,18 @@ export class WasmDownloadedDocument {
|
|
|
453
509
|
free(): void;
|
|
454
510
|
[Symbol.dispose](): void;
|
|
455
511
|
static default(): WasmDownloadedDocument;
|
|
456
|
-
constructor(url?: string | null, mimeType?: string | null, size?: number | null, contentHash?: string | null, headers?: any | null, filename?: string | null);
|
|
512
|
+
constructor(url?: string | null, mimeType?: string | null, size?: number | null, contentHash?: string | null, headers?: any | null, truncated?: boolean | null, filename?: string | null, contentPath?: string | null, contentBase64?: string | null);
|
|
513
|
+
get contentBase64(): string | undefined;
|
|
514
|
+
set contentBase64(value: string | null | undefined);
|
|
457
515
|
contentHash: string;
|
|
516
|
+
get contentPath(): string | undefined;
|
|
517
|
+
set contentPath(value: string | null | undefined);
|
|
458
518
|
get filename(): string | undefined;
|
|
459
519
|
set filename(value: string | null | undefined);
|
|
460
520
|
headers: any;
|
|
461
521
|
mimeType: string;
|
|
462
522
|
size: number;
|
|
523
|
+
truncated: boolean;
|
|
463
524
|
url: string;
|
|
464
525
|
}
|
|
465
526
|
|
|
@@ -534,6 +595,33 @@ export class WasmHeadingInfo {
|
|
|
534
595
|
text: string;
|
|
535
596
|
}
|
|
536
597
|
|
|
598
|
+
/**
|
|
599
|
+
* Hostname/IP allowlist matcher for SSRF policy.
|
|
600
|
+
*
|
|
601
|
+
* Serializes as an internally-tagged object so each variant is distinguishable on the
|
|
602
|
+
* wire and round-trips losslessly:
|
|
603
|
+
*
|
|
604
|
+
* ```json
|
|
605
|
+
* {"type": "exact", "value": "api.example.com"}
|
|
606
|
+
* {"type": "suffix", "value": ".example.com"}
|
|
607
|
+
* {"type": "cidr", "value": "10.0.0.0/8"}
|
|
608
|
+
* ```
|
|
609
|
+
*
|
|
610
|
+
* A bare JSON string is still accepted on deserialization and resolves to `Exact`,
|
|
611
|
+
* preserving configs written against the previous untagged representation.
|
|
612
|
+
*
|
|
613
|
+
* `Exact`: HostMatcher.Exact
|
|
614
|
+
*/
|
|
615
|
+
export class WasmHostMatcher {
|
|
616
|
+
free(): void;
|
|
617
|
+
[Symbol.dispose](): void;
|
|
618
|
+
static default(): WasmHostMatcher;
|
|
619
|
+
constructor();
|
|
620
|
+
type: string;
|
|
621
|
+
get value(): string | undefined;
|
|
622
|
+
set value(value: string | null | undefined);
|
|
623
|
+
}
|
|
624
|
+
|
|
537
625
|
/**
|
|
538
626
|
* An hreflang alternate link entry.
|
|
539
627
|
*/
|
|
@@ -582,10 +670,12 @@ export class WasmInteractionResult {
|
|
|
582
670
|
free(): void;
|
|
583
671
|
[Symbol.dispose](): void;
|
|
584
672
|
static default(): WasmInteractionResult;
|
|
585
|
-
constructor(actionResults?: WasmActionResult[] | null, finalHtml?: string | null, finalUrl?: string | null);
|
|
673
|
+
constructor(actionResults?: WasmActionResult[] | null, finalHtml?: string | null, finalUrl?: string | null, screenshotBase64?: string | null);
|
|
586
674
|
actionResults: WasmActionResult[];
|
|
587
675
|
finalHtml: string;
|
|
588
676
|
finalUrl: string;
|
|
677
|
+
get screenshotBase64(): string | undefined;
|
|
678
|
+
set screenshotBase64(value: string | null | undefined);
|
|
589
679
|
}
|
|
590
680
|
|
|
591
681
|
/**
|
|
@@ -830,7 +920,7 @@ export class WasmScrapeResult {
|
|
|
830
920
|
free(): void;
|
|
831
921
|
[Symbol.dispose](): void;
|
|
832
922
|
static default(): WasmScrapeResult;
|
|
833
|
-
constructor(statusCode?: number | null, finalUrl?: string | 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, isAllowed?: boolean | null, noindexDetected?: boolean | null, nofollowDetected?: boolean | null, isPdf?: boolean | null, wasSkipped?: boolean | null, authHeaderSent?: boolean | null, assets?: WasmDownloadedAsset[] | null, jsRenderHint?: boolean | null, browserUsed?: boolean | null, crawlDelay?: bigint | null, xRobotsTag?: string | null, detectedCharset?: string | null, responseMeta?: WasmResponseMeta | null, markdown?: WasmMarkdownResult | null, extractedData?: any | null, extractionMeta?: WasmExtractionMeta | null, downloadedDocument?: WasmDownloadedDocument | null, browser?: WasmBrowserExtras | null);
|
|
923
|
+
constructor(statusCode?: number | null, finalUrl?: string | 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, isAllowed?: boolean | null, noindexDetected?: boolean | null, nofollowDetected?: boolean | null, isPdf?: boolean | null, wasSkipped?: boolean | null, authHeaderSent?: boolean | null, assets?: WasmDownloadedAsset[] | null, jsRenderHint?: boolean | null, browserUsed?: boolean | null, crawlDelay?: bigint | null, xRobotsTag?: string | null, detectedCharset?: string | null, responseMeta?: WasmResponseMeta | null, markdown?: WasmMarkdownResult | null, extractedData?: any | null, extractionMeta?: WasmExtractionMeta | null, screenshotBase64?: string | null, downloadedDocument?: WasmDownloadedDocument | null, browser?: WasmBrowserExtras | null);
|
|
834
924
|
assets: WasmDownloadedAsset[];
|
|
835
925
|
authHeaderSent: boolean;
|
|
836
926
|
bodySize: number;
|
|
@@ -864,6 +954,8 @@ export class WasmScrapeResult {
|
|
|
864
954
|
noindexDetected: boolean;
|
|
865
955
|
get responseMeta(): WasmResponseMeta | undefined;
|
|
866
956
|
set responseMeta(value: WasmResponseMeta | null | undefined);
|
|
957
|
+
get screenshotBase64(): string | undefined;
|
|
958
|
+
set screenshotBase64(value: string | null | undefined);
|
|
867
959
|
statusCode: number;
|
|
868
960
|
wasSkipped: boolean;
|
|
869
961
|
get xRobotsTag(): string | undefined;
|
|
@@ -913,9 +1005,18 @@ export class WasmSsrfPolicy {
|
|
|
913
1005
|
* - Outbound requests in a browser go through the fetch API, which enforces its own network policies.
|
|
914
1006
|
* - Rust-side SSRF checking is unenforceable and redundant in a wasm32 context.
|
|
915
1007
|
* - For testing and localhost access, the host's network sandbox is the enforcing boundary.
|
|
1008
|
+
*
|
|
1009
|
+
* **Node.js caveat:** `deny_private` (whatever its value) has no effect on hostname-based
|
|
1010
|
+
* requests under `wasm32`. There is no DNS resolution on this target, so `validate_url`
|
|
1011
|
+
* only ever checks a literal IP host; a domain name falls straight through to `Ok(())`. In a
|
|
1012
|
+
* browser this is covered by same-origin/CORS. Node's `fetch` enforces no CORS, so a Node
|
|
1013
|
+
* service embedding this wasm module can be driven to internal hosts by domain name even
|
|
1014
|
+
* though `deny_private = true`. Do not rely on this policy to stop that in Node — enforce
|
|
1015
|
+
* egress restrictions (network policy, firewall, proxy allowlist) outside the process.
|
|
916
1016
|
*/
|
|
917
1017
|
static fromEnv(): WasmSsrfPolicy;
|
|
918
|
-
constructor(denyPrivate?: boolean | null, maxRedirects?: number | null);
|
|
1018
|
+
constructor(denyPrivate?: boolean | null, allowlist?: any | null, maxRedirects?: number | null);
|
|
1019
|
+
allowlist: any;
|
|
919
1020
|
denyPrivate: boolean;
|
|
920
1021
|
maxRedirects: number;
|
|
921
1022
|
}
|
|
@@ -994,6 +1095,7 @@ export interface InitOutput {
|
|
|
994
1095
|
readonly __wbg_wasmfaviconinfo_free: (a: number, b: number) => void;
|
|
995
1096
|
readonly __wbg_wasmfeedinfo_free: (a: number, b: number) => void;
|
|
996
1097
|
readonly __wbg_wasmheadinginfo_free: (a: number, b: number) => void;
|
|
1098
|
+
readonly __wbg_wasmhostmatcher_free: (a: number, b: number) => void;
|
|
997
1099
|
readonly __wbg_wasmhreflangentry_free: (a: number, b: number) => void;
|
|
998
1100
|
readonly __wbg_wasmimageinfo_free: (a: number, b: number) => void;
|
|
999
1101
|
readonly __wbg_wasminteractionresult_free: (a: number, b: number) => void;
|
|
@@ -1172,15 +1274,21 @@ export interface InitOutput {
|
|
|
1172
1274
|
readonly wasmcrawlconfig_allowSubdomains: (a: number) => number;
|
|
1173
1275
|
readonly wasmcrawlconfig_assetTypes: (a: number) => [number, number];
|
|
1174
1276
|
readonly wasmcrawlconfig_auth: (a: number) => any;
|
|
1277
|
+
readonly wasmcrawlconfig_bm25Query: (a: number) => [number, number];
|
|
1278
|
+
readonly wasmcrawlconfig_bm25Threshold: (a: number) => [number, number];
|
|
1175
1279
|
readonly wasmcrawlconfig_browser: (a: number) => number;
|
|
1176
1280
|
readonly wasmcrawlconfig_browserProfile: (a: number) => [number, number];
|
|
1177
1281
|
readonly wasmcrawlconfig_captureScreenshot: (a: number) => number;
|
|
1178
1282
|
readonly wasmcrawlconfig_content: (a: number) => number;
|
|
1283
|
+
readonly wasmcrawlconfig_contentFilter: (a: number) => [number, number];
|
|
1179
1284
|
readonly wasmcrawlconfig_cookiesEnabled: (a: number) => number;
|
|
1285
|
+
readonly wasmcrawlconfig_crawlStrategy: (a: number) => [number, number];
|
|
1180
1286
|
readonly wasmcrawlconfig_customHeaders: (a: number) => any;
|
|
1181
1287
|
readonly wasmcrawlconfig_default: () => number;
|
|
1288
|
+
readonly wasmcrawlconfig_documentContentEncoding: (a: number) => [number, number];
|
|
1182
1289
|
readonly wasmcrawlconfig_documentMaxSize: (a: number) => number;
|
|
1183
1290
|
readonly wasmcrawlconfig_documentMimeTypes: (a: number) => [number, number];
|
|
1291
|
+
readonly wasmcrawlconfig_documentOutputDir: (a: number) => [number, number];
|
|
1184
1292
|
readonly wasmcrawlconfig_documentUrlDepth: (a: number) => number;
|
|
1185
1293
|
readonly wasmcrawlconfig_downloadAssets: (a: number) => number;
|
|
1186
1294
|
readonly wasmcrawlconfig_downloadDocuments: (a: number) => number;
|
|
@@ -1193,10 +1301,12 @@ export interface InitOutput {
|
|
|
1193
1301
|
readonly wasmcrawlconfig_maxBodySize: (a: number) => number;
|
|
1194
1302
|
readonly wasmcrawlconfig_maxConcurrent: (a: number) => number;
|
|
1195
1303
|
readonly wasmcrawlconfig_maxDepth: (a: number) => number;
|
|
1304
|
+
readonly wasmcrawlconfig_maxLinksPerPage: (a: number) => number;
|
|
1196
1305
|
readonly wasmcrawlconfig_maxPages: (a: number) => number;
|
|
1197
1306
|
readonly wasmcrawlconfig_maxRedirects: (a: number) => number;
|
|
1198
|
-
readonly wasmcrawlconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k:
|
|
1307
|
+
readonly 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: bigint, 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, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: bigint, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number, e2: number, f2: number, g2: number, h2: number, i2: number, j2: number, k2: number) => number;
|
|
1199
1308
|
readonly wasmcrawlconfig_proxy: (a: number) => number;
|
|
1309
|
+
readonly wasmcrawlconfig_rateLimitMs: (a: number) => [number, bigint];
|
|
1200
1310
|
readonly wasmcrawlconfig_removeTags: (a: number) => [number, number];
|
|
1201
1311
|
readonly wasmcrawlconfig_respectRobotsTxt: (a: number) => number;
|
|
1202
1312
|
readonly wasmcrawlconfig_retryCodes: (a: number) => [number, number];
|
|
@@ -1205,14 +1315,20 @@ export interface InitOutput {
|
|
|
1205
1315
|
readonly wasmcrawlconfig_set_allowSubdomains: (a: number, b: number) => void;
|
|
1206
1316
|
readonly wasmcrawlconfig_set_assetTypes: (a: number, b: number, c: number) => void;
|
|
1207
1317
|
readonly wasmcrawlconfig_set_auth: (a: number, b: number) => void;
|
|
1318
|
+
readonly wasmcrawlconfig_set_bm25Query: (a: number, b: number, c: number) => void;
|
|
1319
|
+
readonly wasmcrawlconfig_set_bm25Threshold: (a: number, b: number, c: number) => void;
|
|
1208
1320
|
readonly wasmcrawlconfig_set_browser: (a: number, b: number) => void;
|
|
1209
1321
|
readonly wasmcrawlconfig_set_browserProfile: (a: number, b: number, c: number) => void;
|
|
1210
1322
|
readonly wasmcrawlconfig_set_captureScreenshot: (a: number, b: number) => void;
|
|
1211
1323
|
readonly wasmcrawlconfig_set_content: (a: number, b: number) => void;
|
|
1324
|
+
readonly wasmcrawlconfig_set_contentFilter: (a: number, b: number) => void;
|
|
1212
1325
|
readonly wasmcrawlconfig_set_cookiesEnabled: (a: number, b: number) => void;
|
|
1326
|
+
readonly wasmcrawlconfig_set_crawlStrategy: (a: number, b: number) => void;
|
|
1213
1327
|
readonly wasmcrawlconfig_set_customHeaders: (a: number, b: any) => void;
|
|
1328
|
+
readonly wasmcrawlconfig_set_documentContentEncoding: (a: number, b: number) => void;
|
|
1214
1329
|
readonly wasmcrawlconfig_set_documentMaxSize: (a: number, b: number) => void;
|
|
1215
1330
|
readonly wasmcrawlconfig_set_documentMimeTypes: (a: number, b: number, c: number) => void;
|
|
1331
|
+
readonly wasmcrawlconfig_set_documentOutputDir: (a: number, b: number, c: number) => void;
|
|
1216
1332
|
readonly wasmcrawlconfig_set_documentUrlDepth: (a: number, b: number) => void;
|
|
1217
1333
|
readonly wasmcrawlconfig_set_downloadAssets: (a: number, b: number) => void;
|
|
1218
1334
|
readonly wasmcrawlconfig_set_downloadDocuments: (a: number, b: number) => void;
|
|
@@ -1225,9 +1341,11 @@ export interface InitOutput {
|
|
|
1225
1341
|
readonly wasmcrawlconfig_set_maxBodySize: (a: number, b: number) => void;
|
|
1226
1342
|
readonly wasmcrawlconfig_set_maxConcurrent: (a: number, b: number) => void;
|
|
1227
1343
|
readonly wasmcrawlconfig_set_maxDepth: (a: number, b: number) => void;
|
|
1344
|
+
readonly wasmcrawlconfig_set_maxLinksPerPage: (a: number, b: number) => void;
|
|
1228
1345
|
readonly wasmcrawlconfig_set_maxPages: (a: number, b: number) => void;
|
|
1229
1346
|
readonly wasmcrawlconfig_set_maxRedirects: (a: number, b: number) => void;
|
|
1230
1347
|
readonly wasmcrawlconfig_set_proxy: (a: number, b: number) => void;
|
|
1348
|
+
readonly wasmcrawlconfig_set_rateLimitMs: (a: number, b: number, c: bigint) => void;
|
|
1231
1349
|
readonly wasmcrawlconfig_set_removeTags: (a: number, b: number, c: number) => void;
|
|
1232
1350
|
readonly wasmcrawlconfig_set_respectRobotsTxt: (a: number, b: number) => void;
|
|
1233
1351
|
readonly wasmcrawlconfig_set_retryCodes: (a: number, b: number, c: number) => void;
|
|
@@ -1235,12 +1353,14 @@ export interface InitOutput {
|
|
|
1235
1353
|
readonly wasmcrawlconfig_set_saveBrowserProfile: (a: number, b: number) => void;
|
|
1236
1354
|
readonly wasmcrawlconfig_set_softHttpErrors: (a: number, b: number) => void;
|
|
1237
1355
|
readonly wasmcrawlconfig_set_ssrf: (a: number, b: number) => void;
|
|
1356
|
+
readonly wasmcrawlconfig_set_ssrfDenyPrivateExplicit: (a: number, b: number) => void;
|
|
1238
1357
|
readonly wasmcrawlconfig_set_stayOnDomain: (a: number, b: number) => void;
|
|
1239
1358
|
readonly wasmcrawlconfig_set_userAgent: (a: number, b: number, c: number) => void;
|
|
1240
1359
|
readonly wasmcrawlconfig_set_userAgents: (a: number, b: number, c: number) => void;
|
|
1241
1360
|
readonly wasmcrawlconfig_set_warcOutput: (a: number, b: number, c: number) => void;
|
|
1242
1361
|
readonly wasmcrawlconfig_softHttpErrors: (a: number) => number;
|
|
1243
1362
|
readonly wasmcrawlconfig_ssrf: (a: number) => number;
|
|
1363
|
+
readonly wasmcrawlconfig_ssrfDenyPrivateExplicit: (a: number) => number;
|
|
1244
1364
|
readonly wasmcrawlconfig_stayOnDomain: (a: number) => number;
|
|
1245
1365
|
readonly wasmcrawlconfig_userAgent: (a: number) => [number, number];
|
|
1246
1366
|
readonly wasmcrawlconfig_userAgents: (a: number) => [number, number];
|
|
@@ -1321,27 +1441,31 @@ export interface InitOutput {
|
|
|
1321
1441
|
readonly wasmdownloadedasset_set_mimeType: (a: number, b: number, c: number) => void;
|
|
1322
1442
|
readonly wasmdownloadedasset_set_url: (a: number, b: number, c: number) => void;
|
|
1323
1443
|
readonly wasmdownloadedasset_url: (a: number) => [number, number];
|
|
1444
|
+
readonly wasmdownloadeddocument_contentBase64: (a: number) => [number, number];
|
|
1324
1445
|
readonly wasmdownloadeddocument_contentHash: (a: number) => [number, number];
|
|
1446
|
+
readonly wasmdownloadeddocument_contentPath: (a: number) => [number, number];
|
|
1325
1447
|
readonly wasmdownloadeddocument_default: () => number;
|
|
1326
1448
|
readonly wasmdownloadeddocument_filename: (a: number) => [number, number];
|
|
1327
1449
|
readonly wasmdownloadeddocument_headers: (a: number) => any;
|
|
1328
1450
|
readonly wasmdownloadeddocument_mimeType: (a: number) => [number, number];
|
|
1329
|
-
readonly wasmdownloadeddocument_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number) => number;
|
|
1451
|
+
readonly wasmdownloadeddocument_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) => number;
|
|
1452
|
+
readonly wasmdownloadeddocument_set_contentBase64: (a: number, b: number, c: number) => void;
|
|
1330
1453
|
readonly wasmdownloadeddocument_set_contentHash: (a: number, b: number, c: number) => void;
|
|
1454
|
+
readonly wasmdownloadeddocument_set_contentPath: (a: number, b: number, c: number) => void;
|
|
1331
1455
|
readonly wasmdownloadeddocument_set_filename: (a: number, b: number, c: number) => void;
|
|
1332
1456
|
readonly wasmdownloadeddocument_set_headers: (a: number, b: any) => void;
|
|
1333
1457
|
readonly wasmdownloadeddocument_set_mimeType: (a: number, b: number, c: number) => void;
|
|
1458
|
+
readonly wasmdownloadeddocument_set_size: (a: number, b: number) => void;
|
|
1459
|
+
readonly wasmdownloadeddocument_set_truncated: (a: number, b: number) => void;
|
|
1334
1460
|
readonly wasmdownloadeddocument_set_url: (a: number, b: number, c: number) => void;
|
|
1461
|
+
readonly wasmdownloadeddocument_size: (a: number) => number;
|
|
1462
|
+
readonly wasmdownloadeddocument_truncated: (a: number) => number;
|
|
1335
1463
|
readonly wasmdownloadeddocument_url: (a: number) => [number, number];
|
|
1336
1464
|
readonly wasmextractionmeta_chunksProcessed: (a: number) => number;
|
|
1337
|
-
readonly wasmextractionmeta_completionTokens: (a: number) => [number, bigint];
|
|
1338
|
-
readonly wasmextractionmeta_cost: (a: number) => [number, number];
|
|
1339
1465
|
readonly wasmextractionmeta_default: () => number;
|
|
1340
1466
|
readonly wasmextractionmeta_model: (a: number) => [number, number];
|
|
1341
1467
|
readonly wasmextractionmeta_new: (a: number, b: number, c: number, d: number, e: bigint, f: number, g: bigint, h: number, i: number) => number;
|
|
1342
1468
|
readonly wasmextractionmeta_set_chunksProcessed: (a: number, b: number) => void;
|
|
1343
|
-
readonly wasmextractionmeta_set_completionTokens: (a: number, b: number, c: bigint) => void;
|
|
1344
|
-
readonly wasmextractionmeta_set_cost: (a: number, b: number, c: number) => void;
|
|
1345
1469
|
readonly wasmextractionmeta_set_model: (a: number, b: number, c: number) => void;
|
|
1346
1470
|
readonly wasmfaviconinfo_mimeType: (a: number) => [number, number];
|
|
1347
1471
|
readonly wasmfaviconinfo_rel: (a: number) => [number, number];
|
|
@@ -1365,6 +1489,11 @@ export interface InitOutput {
|
|
|
1365
1489
|
readonly wasmheadinginfo_set_level: (a: number, b: number) => void;
|
|
1366
1490
|
readonly wasmheadinginfo_set_text: (a: number, b: number, c: number) => void;
|
|
1367
1491
|
readonly wasmheadinginfo_text: (a: number) => [number, number];
|
|
1492
|
+
readonly wasmhostmatcher_default: () => number;
|
|
1493
|
+
readonly wasmhostmatcher_set_type: (a: number, b: number, c: number) => void;
|
|
1494
|
+
readonly wasmhostmatcher_set_value: (a: number, b: number, c: number) => void;
|
|
1495
|
+
readonly wasmhostmatcher_type: (a: number) => [number, number];
|
|
1496
|
+
readonly wasmhostmatcher_value: (a: number) => [number, number];
|
|
1368
1497
|
readonly wasmhreflangentry_default: () => number;
|
|
1369
1498
|
readonly wasmhreflangentry_lang: (a: number) => [number, number];
|
|
1370
1499
|
readonly wasmhreflangentry_new: (a: number, b: number, c: number, d: number) => number;
|
|
@@ -1385,10 +1514,12 @@ export interface InitOutput {
|
|
|
1385
1514
|
readonly wasminteractionresult_default: () => number;
|
|
1386
1515
|
readonly wasminteractionresult_finalHtml: (a: number) => [number, number];
|
|
1387
1516
|
readonly wasminteractionresult_finalUrl: (a: number) => [number, number];
|
|
1388
|
-
readonly wasminteractionresult_new: (a: number, b: number, c: number, d: number, e: number, f: number) => number;
|
|
1517
|
+
readonly wasminteractionresult_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number) => number;
|
|
1518
|
+
readonly wasminteractionresult_screenshotBase64: (a: number) => [number, number];
|
|
1389
1519
|
readonly wasminteractionresult_set_actionResults: (a: number, b: number, c: number) => void;
|
|
1390
1520
|
readonly wasminteractionresult_set_finalHtml: (a: number, b: number, c: number) => void;
|
|
1391
1521
|
readonly wasminteractionresult_set_finalUrl: (a: number, b: number, c: number) => void;
|
|
1522
|
+
readonly wasminteractionresult_set_screenshotBase64: (a: number, b: number, c: number) => void;
|
|
1392
1523
|
readonly wasmjsonldentry_default: () => number;
|
|
1393
1524
|
readonly wasmjsonldentry_name: (a: number) => [number, number];
|
|
1394
1525
|
readonly wasmjsonldentry_new: (a: number, b: number, c: number, d: number, e: number, f: number) => number;
|
|
@@ -1572,10 +1703,11 @@ export interface InitOutput {
|
|
|
1572
1703
|
readonly wasmscraperesult_links: (a: number) => [number, number];
|
|
1573
1704
|
readonly wasmscraperesult_markdown: (a: number) => number;
|
|
1574
1705
|
readonly wasmscraperesult_metadata: (a: number) => number;
|
|
1575
|
-
readonly wasmscraperesult_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: bigint, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number) => number;
|
|
1706
|
+
readonly wasmscraperesult_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: bigint, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number, n1: number, o1: number) => number;
|
|
1576
1707
|
readonly wasmscraperesult_nofollowDetected: (a: number) => number;
|
|
1577
1708
|
readonly wasmscraperesult_noindexDetected: (a: number) => number;
|
|
1578
1709
|
readonly wasmscraperesult_responseMeta: (a: number) => number;
|
|
1710
|
+
readonly wasmscraperesult_screenshotBase64: (a: number) => [number, number];
|
|
1579
1711
|
readonly wasmscraperesult_set_assets: (a: number, b: number, c: number) => void;
|
|
1580
1712
|
readonly wasmscraperesult_set_authHeaderSent: (a: number, b: number) => void;
|
|
1581
1713
|
readonly wasmscraperesult_set_bodySize: (a: number, b: number) => void;
|
|
@@ -1600,6 +1732,7 @@ export interface InitOutput {
|
|
|
1600
1732
|
readonly wasmscraperesult_set_nofollowDetected: (a: number, b: number) => void;
|
|
1601
1733
|
readonly wasmscraperesult_set_noindexDetected: (a: number, b: number) => void;
|
|
1602
1734
|
readonly wasmscraperesult_set_responseMeta: (a: number, b: number) => void;
|
|
1735
|
+
readonly wasmscraperesult_set_screenshotBase64: (a: number, b: number, c: number) => void;
|
|
1603
1736
|
readonly wasmscraperesult_set_statusCode: (a: number, b: number) => void;
|
|
1604
1737
|
readonly wasmscraperesult_set_wasSkipped: (a: number, b: number) => void;
|
|
1605
1738
|
readonly wasmscraperesult_set_xRobotsTag: (a: number, b: number, c: number) => void;
|
|
@@ -1616,18 +1749,23 @@ export interface InitOutput {
|
|
|
1616
1749
|
readonly wasmsitemapurl_set_priority: (a: number, b: number, c: number) => void;
|
|
1617
1750
|
readonly wasmsitemapurl_set_url: (a: number, b: number, c: number) => void;
|
|
1618
1751
|
readonly wasmsitemapurl_url: (a: number) => [number, number];
|
|
1752
|
+
readonly wasmssrfpolicy_allowlist: (a: number) => any;
|
|
1619
1753
|
readonly wasmssrfpolicy_default: () => number;
|
|
1620
1754
|
readonly wasmssrfpolicy_denyPrivate: (a: number) => number;
|
|
1621
1755
|
readonly wasmssrfpolicy_fromEnv: () => number;
|
|
1622
1756
|
readonly wasmssrfpolicy_maxRedirects: (a: number) => number;
|
|
1623
|
-
readonly wasmssrfpolicy_new: (a: number, b: number) => number;
|
|
1757
|
+
readonly wasmssrfpolicy_new: (a: number, b: number, c: number) => number;
|
|
1758
|
+
readonly wasmssrfpolicy_set_allowlist: (a: number, b: any) => void;
|
|
1624
1759
|
readonly wasmssrfpolicy_set_denyPrivate: (a: number, b: number) => void;
|
|
1625
1760
|
readonly wasmssrfpolicy_set_maxRedirects: (a: number, b: number) => void;
|
|
1626
|
-
readonly
|
|
1761
|
+
readonly wasmhostmatcher_new: () => number;
|
|
1762
|
+
readonly wasmcrawlconfig_set_requestTimeout: (a: number, b: number, c: bigint) => void;
|
|
1763
|
+
readonly wasmextractionmeta_set_completionTokens: (a: number, b: number, c: bigint) => void;
|
|
1627
1764
|
readonly wasmextractionmeta_set_promptTokens: (a: number, b: number, c: bigint) => void;
|
|
1628
1765
|
readonly wasmpageaction_set_milliseconds: (a: number, b: number, c: bigint) => void;
|
|
1629
|
-
readonly wasmcrawlconfig_rateLimitMs: (a: number) => [number, bigint];
|
|
1630
1766
|
readonly wasmcrawlconfig_requestTimeout: (a: number) => [number, bigint];
|
|
1767
|
+
readonly wasmextractionmeta_completionTokens: (a: number) => [number, bigint];
|
|
1768
|
+
readonly wasmextractionmeta_cost: (a: number) => [number, number];
|
|
1631
1769
|
readonly wasmextractionmeta_promptTokens: (a: number) => [number, bigint];
|
|
1632
1770
|
readonly wasmpageaction_amount: (a: number) => [number, bigint];
|
|
1633
1771
|
readonly wasmpageaction_milliseconds: (a: number) => [number, bigint];
|
|
@@ -1637,27 +1775,25 @@ export interface InitOutput {
|
|
|
1637
1775
|
readonly wasmbatchscraperesults_set_failedCount: (a: number, b: number) => void;
|
|
1638
1776
|
readonly wasmbatchscraperesults_set_totalCount: (a: number, b: number) => void;
|
|
1639
1777
|
readonly wasmdownloadedasset_set_size: (a: number, b: number) => void;
|
|
1640
|
-
readonly wasmdownloadeddocument_set_size: (a: number, b: number) => void;
|
|
1641
1778
|
readonly wasmimageinfo_set_width: (a: number, b: number) => void;
|
|
1642
1779
|
readonly wasmpagemetadata_set_wordCount: (a: number, b: number) => void;
|
|
1643
1780
|
readonly wasmbatchscraperesults_completedCount: (a: number) => number;
|
|
1644
1781
|
readonly wasmbatchscraperesults_failedCount: (a: number) => number;
|
|
1645
1782
|
readonly wasmbatchscraperesults_totalCount: (a: number) => number;
|
|
1646
1783
|
readonly wasmdownloadedasset_size: (a: number) => number;
|
|
1647
|
-
readonly wasmdownloadeddocument_size: (a: number) => number;
|
|
1648
1784
|
readonly wasmlinkinfo_set_nofollow: (a: number, b: number) => void;
|
|
1649
1785
|
readonly wasmfaviconinfo_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number) => number;
|
|
1650
1786
|
readonly wasmimageinfo_width: (a: number) => number;
|
|
1651
1787
|
readonly wasmpagemetadata_wordCount: (a: number) => number;
|
|
1652
1788
|
readonly wasmlinkinfo_nofollow: (a: number) => number;
|
|
1653
1789
|
readonly wasmauthconfig_new: () => number;
|
|
1654
|
-
readonly
|
|
1790
|
+
readonly wasmextractionmeta_set_cost: (a: number, b: number, c: number) => void;
|
|
1655
1791
|
readonly wasmpageaction_set_amount: (a: number, b: number, c: bigint) => void;
|
|
1656
1792
|
readonly wasmscraperesult_set_crawlDelay: (a: number, b: number, c: bigint) => void;
|
|
1657
1793
|
readonly wasmpageaction_new: () => number;
|
|
1658
|
-
readonly
|
|
1659
|
-
readonly
|
|
1660
|
-
readonly
|
|
1794
|
+
readonly wasm_bindgen__convert__closures_____invoke__h2deefa3c9140ef43: (a: number, b: number, c: any) => [number, number];
|
|
1795
|
+
readonly wasm_bindgen__convert__closures_____invoke__h072ef20b96c52fc3: (a: number, b: number, c: any, d: any) => void;
|
|
1796
|
+
readonly wasm_bindgen__convert__closures_____invoke__h99e891e0bdc2d97f: (a: number, b: number) => void;
|
|
1661
1797
|
readonly __wbindgen_malloc: (a: number, b: number) => number;
|
|
1662
1798
|
readonly __wbindgen_realloc: (a: number, b: number, c: number, d: number) => number;
|
|
1663
1799
|
readonly __wbindgen_exn_store: (a: number) => void;
|