@xberg-io/crawlberg-wasm 1.8.0 → 1.9.0
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 +12 -5
- package/pkg/bundler/crawlberg_wasm_bg.js +161 -44
- package/pkg/bundler/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/bundler/crawlberg_wasm_bg.wasm.d.ts +18 -6
- package/pkg/bundler/package.json +1 -1
- package/pkg/deno/crawlberg_wasm.d.ts +12 -5
- package/pkg/deno/crawlberg_wasm.js +161 -44
- package/pkg/deno/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/deno/crawlberg_wasm_bg.wasm.d.ts +18 -6
- package/pkg/nodejs/crawlberg_wasm.d.ts +12 -5
- package/pkg/nodejs/crawlberg_wasm.js +161 -44
- package/pkg/nodejs/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/nodejs/crawlberg_wasm_bg.wasm.d.ts +18 -6
- package/pkg/nodejs/package.json +1 -1
- package/pkg/web/crawlberg_wasm.d.ts +30 -11
- package/pkg/web/crawlberg_wasm.js +161 -44
- package/pkg/web/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/web/crawlberg_wasm_bg.wasm.d.ts +18 -6
- package/pkg/web/package.json +1 -1
package/package.json
CHANGED
|
@@ -153,11 +153,14 @@ 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, 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);
|
|
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, chromeArgs?: string[] | null, endpoint?: string | null, waitSelector?: string | null, extraWait?: bigint | null, proxy?: WasmProxyConfig | null, evalScript?: string | null, robotsUserAgent?: string | null, chromePath?: string | null);
|
|
157
157
|
get backend(): string;
|
|
158
158
|
set backend(value: WasmBrowserBackend);
|
|
159
159
|
blockUrlPatterns: string[];
|
|
160
160
|
captureNetworkEvents: boolean;
|
|
161
|
+
chromeArgs: string[];
|
|
162
|
+
get chromePath(): string | undefined;
|
|
163
|
+
set chromePath(value: string | null | undefined);
|
|
161
164
|
get endpoint(): string | undefined;
|
|
162
165
|
set endpoint(value: string | null | undefined);
|
|
163
166
|
get evalScript(): string | undefined;
|
|
@@ -302,7 +305,7 @@ export class WasmCrawlConfig {
|
|
|
302
305
|
free(): void;
|
|
303
306
|
[Symbol.dispose](): void;
|
|
304
307
|
static default(): WasmCrawlConfig;
|
|
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);
|
|
308
|
+
constructor(crawlStrategy?: WasmCrawlStrategyKind | null, respectRobotsTxt?: boolean | null, softHttpErrors?: boolean | null, stayOnDomain?: boolean | null, allowSubdomains?: boolean | null, includePaths?: string[] | null, excludePaths?: string[] | null, pathPatternsMatchQuery?: boolean | null, pathPatternsMatchUrl?: boolean | null, dedupIncludeQuery?: boolean | null, stripTrackingParams?: boolean | null, trackingParams?: string[] | null, customHeaders?: any | null, requestTimeout?: bigint | null, maxRedirects?: number | null, retryCount?: number | null, retryCodes?: Uint16Array | null, retryInitialDelayMs?: bigint | null, retryMaxDelayMs?: bigint | null, rateLimitJitterRatio?: number | null, cookiesEnabled?: boolean | null, removeTags?: string[] | null, content?: WasmContentConfig | null, downloadAssets?: boolean | null, assetTypes?: WasmAssetCategory[] | null, browser?: WasmBrowserConfig | null, userAgents?: string[] | null, captureScreenshot?: boolean | null, followDocumentUrls?: boolean | null, downloadDocuments?: boolean | null, documentMimeTypes?: string[] | null, saveBrowserProfile?: boolean | null, ssrf?: WasmSsrfPolicy | null, maxDepth?: number | null, maxPages?: number | null, maxLinksPerPage?: number | null, maxConcurrent?: number | null, contentFilter?: WasmContentFilterKind | null, bm25Query?: string | null, bm25Threshold?: number | null, userAgent?: string | null, rateLimitMs?: bigint | null, auth?: any | null, maxBodySize?: number | null, mapLimit?: number | null, mapSearch?: string | null, maxAssetSize?: number | null, proxy?: WasmProxyConfig | null, documentUrlDepth?: number | null, documentMaxSize?: number | null, documentOutputDir?: string | null, documentContentEncoding?: WasmDocumentContentEncoding | null, warcOutput?: string | null, browserProfile?: string | null, ssrfDenyPrivateExplicit?: boolean | null);
|
|
306
309
|
/**
|
|
307
310
|
* Validate the configuration, returning an error if any values are invalid.
|
|
308
311
|
*
|
|
@@ -362,6 +365,7 @@ export class WasmCrawlConfig {
|
|
|
362
365
|
set maxPages(value: number | null | undefined);
|
|
363
366
|
maxRedirects: number;
|
|
364
367
|
pathPatternsMatchQuery: boolean;
|
|
368
|
+
pathPatternsMatchUrl: boolean;
|
|
365
369
|
get proxy(): WasmProxyConfig | undefined;
|
|
366
370
|
set proxy(value: WasmProxyConfig | null | undefined);
|
|
367
371
|
rateLimitJitterRatio: number;
|
|
@@ -409,7 +413,7 @@ export class WasmCrawlPageResult {
|
|
|
409
413
|
free(): void;
|
|
410
414
|
[Symbol.dispose](): void;
|
|
411
415
|
static default(): WasmCrawlPageResult;
|
|
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);
|
|
416
|
+
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);
|
|
413
417
|
bodySize: number;
|
|
414
418
|
browserUsed: boolean;
|
|
415
419
|
contentType: string;
|
|
@@ -436,6 +440,7 @@ export class WasmCrawlPageResult {
|
|
|
436
440
|
noindexDetected: boolean;
|
|
437
441
|
normalizedUrl: string;
|
|
438
442
|
redirectCount: number;
|
|
443
|
+
ssrfRefusedUrls: string[];
|
|
439
444
|
statusCode: number;
|
|
440
445
|
stayedOnDomain: boolean;
|
|
441
446
|
url: string;
|
|
@@ -689,12 +694,13 @@ export class WasmInteractionResult {
|
|
|
689
694
|
free(): void;
|
|
690
695
|
[Symbol.dispose](): void;
|
|
691
696
|
static default(): WasmInteractionResult;
|
|
692
|
-
constructor(actionResults?: WasmActionResult[] | null, finalHtml?: string | null, finalUrl?: string | null, screenshotBase64?: string | null);
|
|
697
|
+
constructor(actionResults?: WasmActionResult[] | null, finalHtml?: string | null, finalUrl?: string | null, ssrfRefusedUrls?: string[] | null, screenshotBase64?: string | null);
|
|
693
698
|
actionResults: WasmActionResult[];
|
|
694
699
|
finalHtml: string;
|
|
695
700
|
finalUrl: string;
|
|
696
701
|
get screenshotBase64(): string | undefined;
|
|
697
702
|
set screenshotBase64(value: string | null | undefined);
|
|
703
|
+
ssrfRefusedUrls: string[];
|
|
698
704
|
}
|
|
699
705
|
|
|
700
706
|
/**
|
|
@@ -939,7 +945,7 @@ export class WasmScrapeResult {
|
|
|
939
945
|
free(): void;
|
|
940
946
|
[Symbol.dispose](): void;
|
|
941
947
|
static default(): WasmScrapeResult;
|
|
942
|
-
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);
|
|
948
|
+
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, ssrfRefusedUrls?: string[] | 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);
|
|
943
949
|
assets: WasmDownloadedAsset[];
|
|
944
950
|
authHeaderSent: boolean;
|
|
945
951
|
bodySize: number;
|
|
@@ -975,6 +981,7 @@ export class WasmScrapeResult {
|
|
|
975
981
|
set responseMeta(value: WasmResponseMeta | null | undefined);
|
|
976
982
|
get screenshotBase64(): string | undefined;
|
|
977
983
|
set screenshotBase64(value: string | null | undefined);
|
|
984
|
+
ssrfRefusedUrls: string[];
|
|
978
985
|
statusCode: number;
|
|
979
986
|
wasSkipped: boolean;
|
|
980
987
|
get xRobotsTag(): string | undefined;
|
|
@@ -959,6 +959,27 @@ export class WasmBrowserConfig {
|
|
|
959
959
|
const ret = wasm.wasmbrowserconfig_captureNetworkEvents(this.__wbg_ptr);
|
|
960
960
|
return ret !== 0;
|
|
961
961
|
}
|
|
962
|
+
/**
|
|
963
|
+
* @returns {string[]}
|
|
964
|
+
*/
|
|
965
|
+
get chromeArgs() {
|
|
966
|
+
const ret = wasm.wasmbrowserconfig_chromeArgs(this.__wbg_ptr);
|
|
967
|
+
var v1 = getArrayJsValueFromWasm0(ret[0], ret[1]);
|
|
968
|
+
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
969
|
+
return v1;
|
|
970
|
+
}
|
|
971
|
+
/**
|
|
972
|
+
* @returns {string | undefined}
|
|
973
|
+
*/
|
|
974
|
+
get chromePath() {
|
|
975
|
+
const ret = wasm.wasmbrowserconfig_chromePath(this.__wbg_ptr);
|
|
976
|
+
let v1;
|
|
977
|
+
if (ret[0] !== 0) {
|
|
978
|
+
v1 = getStringFromWasm0(ret[0], ret[1]);
|
|
979
|
+
wasm.__wbindgen_free(ret[0], ret[1] * 1, 1);
|
|
980
|
+
}
|
|
981
|
+
return v1;
|
|
982
|
+
}
|
|
962
983
|
/**
|
|
963
984
|
* @returns {WasmBrowserConfig}
|
|
964
985
|
*/
|
|
@@ -1022,30 +1043,36 @@ export class WasmBrowserConfig {
|
|
|
1022
1043
|
* @param {string[] | null} [blockUrlPatterns]
|
|
1023
1044
|
* @param {boolean | null} [captureNetworkEvents]
|
|
1024
1045
|
* @param {boolean | null} [sessionAffinity]
|
|
1046
|
+
* @param {string[] | null} [chromeArgs]
|
|
1025
1047
|
* @param {string | null} [endpoint]
|
|
1026
1048
|
* @param {string | null} [waitSelector]
|
|
1027
1049
|
* @param {bigint | null} [extraWait]
|
|
1028
1050
|
* @param {WasmProxyConfig | null} [proxy]
|
|
1029
1051
|
* @param {string | null} [evalScript]
|
|
1030
1052
|
* @param {string | null} [robotsUserAgent]
|
|
1053
|
+
* @param {string | null} [chromePath]
|
|
1031
1054
|
*/
|
|
1032
|
-
constructor(mode, backend, timeout, overallTimeout, shutdownTimeout, wait, blockUrlPatterns, captureNetworkEvents, sessionAffinity, endpoint, waitSelector, extraWait, proxy, evalScript, robotsUserAgent) {
|
|
1055
|
+
constructor(mode, backend, timeout, overallTimeout, shutdownTimeout, wait, blockUrlPatterns, captureNetworkEvents, sessionAffinity, chromeArgs, endpoint, waitSelector, extraWait, proxy, evalScript, robotsUserAgent, chromePath) {
|
|
1033
1056
|
var ptr0 = isLikeNone(blockUrlPatterns) ? 0 : passArrayJsValueToWasm0(blockUrlPatterns, wasm.__wbindgen_malloc);
|
|
1034
1057
|
var len0 = WASM_VECTOR_LEN;
|
|
1035
|
-
var ptr1 = isLikeNone(
|
|
1058
|
+
var ptr1 = isLikeNone(chromeArgs) ? 0 : passArrayJsValueToWasm0(chromeArgs, wasm.__wbindgen_malloc);
|
|
1036
1059
|
var len1 = WASM_VECTOR_LEN;
|
|
1037
|
-
var ptr2 = isLikeNone(
|
|
1060
|
+
var ptr2 = isLikeNone(endpoint) ? 0 : passStringToWasm0(endpoint, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1038
1061
|
var len2 = WASM_VECTOR_LEN;
|
|
1039
|
-
|
|
1062
|
+
var ptr3 = isLikeNone(waitSelector) ? 0 : passStringToWasm0(waitSelector, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1063
|
+
var len3 = WASM_VECTOR_LEN;
|
|
1064
|
+
let ptr4 = 0;
|
|
1040
1065
|
if (!isLikeNone(proxy)) {
|
|
1041
1066
|
_assertClass(proxy, WasmProxyConfig);
|
|
1042
|
-
|
|
1067
|
+
ptr4 = proxy.__destroy_into_raw();
|
|
1043
1068
|
}
|
|
1044
|
-
var
|
|
1045
|
-
var len4 = WASM_VECTOR_LEN;
|
|
1046
|
-
var ptr5 = isLikeNone(robotsUserAgent) ? 0 : passStringToWasm0(robotsUserAgent, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1069
|
+
var ptr5 = isLikeNone(evalScript) ? 0 : passStringToWasm0(evalScript, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1047
1070
|
var len5 = WASM_VECTOR_LEN;
|
|
1048
|
-
|
|
1071
|
+
var ptr6 = isLikeNone(robotsUserAgent) ? 0 : passStringToWasm0(robotsUserAgent, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1072
|
+
var len6 = WASM_VECTOR_LEN;
|
|
1073
|
+
var ptr7 = isLikeNone(chromePath) ? 0 : passStringToWasm0(chromePath, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1074
|
+
var len7 = WASM_VECTOR_LEN;
|
|
1075
|
+
const ret = wasm.wasmbrowserconfig_new(isLikeNone(mode) ? 4 : mode, isLikeNone(backend) ? 2 : backend, !isLikeNone(timeout), isLikeNone(timeout) ? BigInt(0) : timeout, !isLikeNone(overallTimeout), isLikeNone(overallTimeout) ? BigInt(0) : overallTimeout, !isLikeNone(shutdownTimeout), isLikeNone(shutdownTimeout) ? BigInt(0) : shutdownTimeout, isLikeNone(wait) ? 3 : wait, ptr0, len0, isLikeNone(captureNetworkEvents) ? 0xFFFFFF : captureNetworkEvents ? 1 : 0, isLikeNone(sessionAffinity) ? 0xFFFFFF : sessionAffinity ? 1 : 0, ptr1, len1, ptr2, len2, ptr3, len3, !isLikeNone(extraWait), isLikeNone(extraWait) ? BigInt(0) : extraWait, ptr4, ptr5, len5, ptr6, len6, ptr7, len7);
|
|
1049
1076
|
this.__wbg_ptr = ret;
|
|
1050
1077
|
WasmBrowserConfigFinalization.register(this, this.__wbg_ptr, this);
|
|
1051
1078
|
return this;
|
|
@@ -1103,6 +1130,22 @@ export class WasmBrowserConfig {
|
|
|
1103
1130
|
set captureNetworkEvents(value) {
|
|
1104
1131
|
wasm.wasmbrowserconfig_set_captureNetworkEvents(this.__wbg_ptr, value);
|
|
1105
1132
|
}
|
|
1133
|
+
/**
|
|
1134
|
+
* @param {string[]} value
|
|
1135
|
+
*/
|
|
1136
|
+
set chromeArgs(value) {
|
|
1137
|
+
const ptr0 = passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc);
|
|
1138
|
+
const len0 = WASM_VECTOR_LEN;
|
|
1139
|
+
wasm.wasmbrowserconfig_set_chromeArgs(this.__wbg_ptr, ptr0, len0);
|
|
1140
|
+
}
|
|
1141
|
+
/**
|
|
1142
|
+
* @param {string | null} [value]
|
|
1143
|
+
*/
|
|
1144
|
+
set chromePath(value) {
|
|
1145
|
+
var ptr0 = isLikeNone(value) ? 0 : passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1146
|
+
var len0 = WASM_VECTOR_LEN;
|
|
1147
|
+
wasm.wasmbrowserconfig_set_chromePath(this.__wbg_ptr, ptr0, len0);
|
|
1148
|
+
}
|
|
1106
1149
|
/**
|
|
1107
1150
|
* @param {string | null} [value]
|
|
1108
1151
|
*/
|
|
@@ -2271,6 +2314,7 @@ export class WasmCrawlConfig {
|
|
|
2271
2314
|
* @param {string[] | null} [includePaths]
|
|
2272
2315
|
* @param {string[] | null} [excludePaths]
|
|
2273
2316
|
* @param {boolean | null} [pathPatternsMatchQuery]
|
|
2317
|
+
* @param {boolean | null} [pathPatternsMatchUrl]
|
|
2274
2318
|
* @param {boolean | null} [dedupIncludeQuery]
|
|
2275
2319
|
* @param {boolean | null} [stripTrackingParams]
|
|
2276
2320
|
* @param {string[] | null} [trackingParams]
|
|
@@ -2318,7 +2362,7 @@ export class WasmCrawlConfig {
|
|
|
2318
2362
|
* @param {string | null} [browserProfile]
|
|
2319
2363
|
* @param {boolean | null} [ssrfDenyPrivateExplicit]
|
|
2320
2364
|
*/
|
|
2321
|
-
constructor(crawlStrategy, respectRobotsTxt, softHttpErrors, stayOnDomain, allowSubdomains, includePaths, excludePaths, pathPatternsMatchQuery, dedupIncludeQuery, stripTrackingParams, trackingParams, customHeaders, requestTimeout, maxRedirects, retryCount, retryCodes, retryInitialDelayMs, retryMaxDelayMs, rateLimitJitterRatio, cookiesEnabled, removeTags, content, downloadAssets, assetTypes, browser, userAgents, captureScreenshot, followDocumentUrls, downloadDocuments, documentMimeTypes, saveBrowserProfile, ssrf, maxDepth, maxPages, maxLinksPerPage, maxConcurrent, contentFilter, bm25Query, bm25Threshold, userAgent, rateLimitMs, auth, maxBodySize, mapLimit, mapSearch, maxAssetSize, proxy, documentUrlDepth, documentMaxSize, documentOutputDir, documentContentEncoding, warcOutput, browserProfile, ssrfDenyPrivateExplicit) {
|
|
2365
|
+
constructor(crawlStrategy, respectRobotsTxt, softHttpErrors, stayOnDomain, allowSubdomains, includePaths, excludePaths, pathPatternsMatchQuery, pathPatternsMatchUrl, dedupIncludeQuery, stripTrackingParams, trackingParams, customHeaders, requestTimeout, maxRedirects, retryCount, retryCodes, retryInitialDelayMs, retryMaxDelayMs, rateLimitJitterRatio, cookiesEnabled, removeTags, content, downloadAssets, assetTypes, browser, userAgents, captureScreenshot, followDocumentUrls, downloadDocuments, documentMimeTypes, saveBrowserProfile, ssrf, maxDepth, maxPages, maxLinksPerPage, maxConcurrent, contentFilter, bm25Query, bm25Threshold, userAgent, rateLimitMs, auth, maxBodySize, mapLimit, mapSearch, maxAssetSize, proxy, documentUrlDepth, documentMaxSize, documentOutputDir, documentContentEncoding, warcOutput, browserProfile, ssrfDenyPrivateExplicit) {
|
|
2322
2366
|
var ptr0 = isLikeNone(includePaths) ? 0 : passArrayJsValueToWasm0(includePaths, wasm.__wbindgen_malloc);
|
|
2323
2367
|
var len0 = WASM_VECTOR_LEN;
|
|
2324
2368
|
var ptr1 = isLikeNone(excludePaths) ? 0 : passArrayJsValueToWasm0(excludePaths, wasm.__wbindgen_malloc);
|
|
@@ -2367,7 +2411,7 @@ export class WasmCrawlConfig {
|
|
|
2367
2411
|
var len16 = WASM_VECTOR_LEN;
|
|
2368
2412
|
var ptr17 = isLikeNone(browserProfile) ? 0 : passStringToWasm0(browserProfile, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
2369
2413
|
var len17 = WASM_VECTOR_LEN;
|
|
2370
|
-
const ret = wasm.wasmcrawlconfig_new(isLikeNone(crawlStrategy) ? 4 : crawlStrategy, isLikeNone(respectRobotsTxt) ? 0xFFFFFF : respectRobotsTxt ? 1 : 0, isLikeNone(softHttpErrors) ? 0xFFFFFF : softHttpErrors ? 1 : 0, isLikeNone(stayOnDomain) ? 0xFFFFFF : stayOnDomain ? 1 : 0, isLikeNone(allowSubdomains) ? 0xFFFFFF : allowSubdomains ? 1 : 0, ptr0, len0, ptr1, len1, isLikeNone(pathPatternsMatchQuery) ? 0xFFFFFF : pathPatternsMatchQuery ? 1 : 0, isLikeNone(dedupIncludeQuery) ? 0xFFFFFF : dedupIncludeQuery ? 1 : 0, isLikeNone(stripTrackingParams) ? 0xFFFFFF : stripTrackingParams ? 1 : 0, ptr2, len2, isLikeNone(customHeaders) ? 0 : addToExternrefTable0(customHeaders), !isLikeNone(requestTimeout), isLikeNone(requestTimeout) ? BigInt(0) : requestTimeout, isLikeNone(maxRedirects) ? Number.MAX_SAFE_INTEGER : (maxRedirects) >>> 0, isLikeNone(retryCount) ? Number.MAX_SAFE_INTEGER : (retryCount) >>> 0, ptr3, len3, !isLikeNone(retryInitialDelayMs), isLikeNone(retryInitialDelayMs) ? BigInt(0) : retryInitialDelayMs, !isLikeNone(retryMaxDelayMs), isLikeNone(retryMaxDelayMs) ? BigInt(0) : retryMaxDelayMs, !isLikeNone(rateLimitJitterRatio), isLikeNone(rateLimitJitterRatio) ? 0 : rateLimitJitterRatio, isLikeNone(cookiesEnabled) ? 0xFFFFFF : cookiesEnabled ? 1 : 0, ptr4, len4, ptr5, isLikeNone(downloadAssets) ? 0xFFFFFF : downloadAssets ? 1 : 0, ptr6, len6, ptr7, ptr8, len8, isLikeNone(captureScreenshot) ? 0xFFFFFF : captureScreenshot ? 1 : 0, isLikeNone(followDocumentUrls) ? 0xFFFFFF : followDocumentUrls ? 1 : 0, isLikeNone(downloadDocuments) ? 0xFFFFFF : downloadDocuments ? 1 : 0, ptr9, len9, isLikeNone(saveBrowserProfile) ? 0xFFFFFF : saveBrowserProfile ? 1 : 0, ptr10, isLikeNone(maxDepth) ? Number.MAX_SAFE_INTEGER : (maxDepth) >>> 0, isLikeNone(maxPages) ? Number.MAX_SAFE_INTEGER : (maxPages) >>> 0, isLikeNone(maxLinksPerPage) ? Number.MAX_SAFE_INTEGER : (maxLinksPerPage) >>> 0, isLikeNone(maxConcurrent) ? Number.MAX_SAFE_INTEGER : (maxConcurrent) >>> 0, isLikeNone(contentFilter) ? 1 : contentFilter, ptr11, len11, !isLikeNone(bm25Threshold), isLikeNone(bm25Threshold) ? 0 : bm25Threshold, ptr12, len12, !isLikeNone(rateLimitMs), isLikeNone(rateLimitMs) ? BigInt(0) : rateLimitMs, isLikeNone(auth) ? 0 : addToExternrefTable0(auth), isLikeNone(maxBodySize) ? Number.MAX_SAFE_INTEGER : (maxBodySize) >>> 0, isLikeNone(mapLimit) ? Number.MAX_SAFE_INTEGER : (mapLimit) >>> 0, ptr13, len13, isLikeNone(maxAssetSize) ? Number.MAX_SAFE_INTEGER : (maxAssetSize) >>> 0, ptr14, isLikeNone(documentUrlDepth) ? Number.MAX_SAFE_INTEGER : (documentUrlDepth) >>> 0, isLikeNone(documentMaxSize) ? Number.MAX_SAFE_INTEGER : (documentMaxSize) >>> 0, ptr15, len15, isLikeNone(documentContentEncoding) ? 1 : documentContentEncoding, ptr16, len16, ptr17, len17, isLikeNone(ssrfDenyPrivateExplicit) ? 0xFFFFFF : ssrfDenyPrivateExplicit ? 1 : 0);
|
|
2414
|
+
const ret = wasm.wasmcrawlconfig_new(isLikeNone(crawlStrategy) ? 4 : crawlStrategy, isLikeNone(respectRobotsTxt) ? 0xFFFFFF : respectRobotsTxt ? 1 : 0, isLikeNone(softHttpErrors) ? 0xFFFFFF : softHttpErrors ? 1 : 0, isLikeNone(stayOnDomain) ? 0xFFFFFF : stayOnDomain ? 1 : 0, isLikeNone(allowSubdomains) ? 0xFFFFFF : allowSubdomains ? 1 : 0, ptr0, len0, ptr1, len1, isLikeNone(pathPatternsMatchQuery) ? 0xFFFFFF : pathPatternsMatchQuery ? 1 : 0, isLikeNone(pathPatternsMatchUrl) ? 0xFFFFFF : pathPatternsMatchUrl ? 1 : 0, isLikeNone(dedupIncludeQuery) ? 0xFFFFFF : dedupIncludeQuery ? 1 : 0, isLikeNone(stripTrackingParams) ? 0xFFFFFF : stripTrackingParams ? 1 : 0, ptr2, len2, isLikeNone(customHeaders) ? 0 : addToExternrefTable0(customHeaders), !isLikeNone(requestTimeout), isLikeNone(requestTimeout) ? BigInt(0) : requestTimeout, isLikeNone(maxRedirects) ? Number.MAX_SAFE_INTEGER : (maxRedirects) >>> 0, isLikeNone(retryCount) ? Number.MAX_SAFE_INTEGER : (retryCount) >>> 0, ptr3, len3, !isLikeNone(retryInitialDelayMs), isLikeNone(retryInitialDelayMs) ? BigInt(0) : retryInitialDelayMs, !isLikeNone(retryMaxDelayMs), isLikeNone(retryMaxDelayMs) ? BigInt(0) : retryMaxDelayMs, !isLikeNone(rateLimitJitterRatio), isLikeNone(rateLimitJitterRatio) ? 0 : rateLimitJitterRatio, isLikeNone(cookiesEnabled) ? 0xFFFFFF : cookiesEnabled ? 1 : 0, ptr4, len4, ptr5, isLikeNone(downloadAssets) ? 0xFFFFFF : downloadAssets ? 1 : 0, ptr6, len6, ptr7, ptr8, len8, isLikeNone(captureScreenshot) ? 0xFFFFFF : captureScreenshot ? 1 : 0, isLikeNone(followDocumentUrls) ? 0xFFFFFF : followDocumentUrls ? 1 : 0, isLikeNone(downloadDocuments) ? 0xFFFFFF : downloadDocuments ? 1 : 0, ptr9, len9, isLikeNone(saveBrowserProfile) ? 0xFFFFFF : saveBrowserProfile ? 1 : 0, ptr10, isLikeNone(maxDepth) ? Number.MAX_SAFE_INTEGER : (maxDepth) >>> 0, isLikeNone(maxPages) ? Number.MAX_SAFE_INTEGER : (maxPages) >>> 0, isLikeNone(maxLinksPerPage) ? Number.MAX_SAFE_INTEGER : (maxLinksPerPage) >>> 0, isLikeNone(maxConcurrent) ? Number.MAX_SAFE_INTEGER : (maxConcurrent) >>> 0, isLikeNone(contentFilter) ? 1 : contentFilter, ptr11, len11, !isLikeNone(bm25Threshold), isLikeNone(bm25Threshold) ? 0 : bm25Threshold, ptr12, len12, !isLikeNone(rateLimitMs), isLikeNone(rateLimitMs) ? BigInt(0) : rateLimitMs, isLikeNone(auth) ? 0 : addToExternrefTable0(auth), isLikeNone(maxBodySize) ? Number.MAX_SAFE_INTEGER : (maxBodySize) >>> 0, isLikeNone(mapLimit) ? Number.MAX_SAFE_INTEGER : (mapLimit) >>> 0, ptr13, len13, isLikeNone(maxAssetSize) ? Number.MAX_SAFE_INTEGER : (maxAssetSize) >>> 0, ptr14, isLikeNone(documentUrlDepth) ? Number.MAX_SAFE_INTEGER : (documentUrlDepth) >>> 0, isLikeNone(documentMaxSize) ? Number.MAX_SAFE_INTEGER : (documentMaxSize) >>> 0, ptr15, len15, isLikeNone(documentContentEncoding) ? 1 : documentContentEncoding, ptr16, len16, ptr17, len17, isLikeNone(ssrfDenyPrivateExplicit) ? 0xFFFFFF : ssrfDenyPrivateExplicit ? 1 : 0);
|
|
2371
2415
|
this.__wbg_ptr = ret;
|
|
2372
2416
|
WasmCrawlConfigFinalization.register(this, this.__wbg_ptr, this);
|
|
2373
2417
|
return this;
|
|
@@ -2379,6 +2423,13 @@ export class WasmCrawlConfig {
|
|
|
2379
2423
|
const ret = wasm.wasmcrawlconfig_pathPatternsMatchQuery(this.__wbg_ptr);
|
|
2380
2424
|
return ret !== 0;
|
|
2381
2425
|
}
|
|
2426
|
+
/**
|
|
2427
|
+
* @returns {boolean}
|
|
2428
|
+
*/
|
|
2429
|
+
get pathPatternsMatchUrl() {
|
|
2430
|
+
const ret = wasm.wasmcrawlconfig_pathPatternsMatchUrl(this.__wbg_ptr);
|
|
2431
|
+
return ret !== 0;
|
|
2432
|
+
}
|
|
2382
2433
|
/**
|
|
2383
2434
|
* @returns {WasmProxyConfig | undefined}
|
|
2384
2435
|
*/
|
|
@@ -2684,6 +2735,12 @@ export class WasmCrawlConfig {
|
|
|
2684
2735
|
set pathPatternsMatchQuery(value) {
|
|
2685
2736
|
wasm.wasmcrawlconfig_set_pathPatternsMatchQuery(this.__wbg_ptr, value);
|
|
2686
2737
|
}
|
|
2738
|
+
/**
|
|
2739
|
+
* @param {boolean} value
|
|
2740
|
+
*/
|
|
2741
|
+
set pathPatternsMatchUrl(value) {
|
|
2742
|
+
wasm.wasmcrawlconfig_set_pathPatternsMatchUrl(this.__wbg_ptr, value);
|
|
2743
|
+
}
|
|
2687
2744
|
/**
|
|
2688
2745
|
* @param {WasmProxyConfig | null} [value]
|
|
2689
2746
|
*/
|
|
@@ -3151,13 +3208,14 @@ export class WasmCrawlPageResult {
|
|
|
3151
3208
|
* @param {number | null} [redirectCount]
|
|
3152
3209
|
* @param {boolean | null} [noindexDetected]
|
|
3153
3210
|
* @param {boolean | null} [nofollowDetected]
|
|
3211
|
+
* @param {string[] | null} [ssrfRefusedUrls]
|
|
3154
3212
|
* @param {string | null} [detectedCharset]
|
|
3155
3213
|
* @param {WasmMarkdownResult | null} [markdown]
|
|
3156
3214
|
* @param {any | null} [extractedData]
|
|
3157
3215
|
* @param {WasmExtractionMeta | null} [extractionMeta]
|
|
3158
3216
|
* @param {WasmDownloadedDocument | null} [downloadedDocument]
|
|
3159
3217
|
*/
|
|
3160
|
-
constructor(url, normalizedUrl, statusCode, contentType, html, bodySize, metadata, links, images, feeds, jsonLd, depth, stayedOnDomain, wasSkipped, isPdf, browserUsed, finalUrl, redirectCount, noindexDetected, nofollowDetected, detectedCharset, markdown, extractedData, extractionMeta, downloadedDocument) {
|
|
3218
|
+
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) {
|
|
3161
3219
|
var ptr0 = isLikeNone(url) ? 0 : passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
3162
3220
|
var len0 = WASM_VECTOR_LEN;
|
|
3163
3221
|
var ptr1 = isLikeNone(normalizedUrl) ? 0 : passStringToWasm0(normalizedUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
@@ -3181,24 +3239,26 @@ export class WasmCrawlPageResult {
|
|
|
3181
3239
|
var len8 = WASM_VECTOR_LEN;
|
|
3182
3240
|
var ptr9 = isLikeNone(finalUrl) ? 0 : passStringToWasm0(finalUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
3183
3241
|
var len9 = WASM_VECTOR_LEN;
|
|
3184
|
-
var ptr10 = isLikeNone(
|
|
3242
|
+
var ptr10 = isLikeNone(ssrfRefusedUrls) ? 0 : passArrayJsValueToWasm0(ssrfRefusedUrls, wasm.__wbindgen_malloc);
|
|
3185
3243
|
var len10 = WASM_VECTOR_LEN;
|
|
3186
|
-
|
|
3244
|
+
var ptr11 = isLikeNone(detectedCharset) ? 0 : passStringToWasm0(detectedCharset, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
3245
|
+
var len11 = WASM_VECTOR_LEN;
|
|
3246
|
+
let ptr12 = 0;
|
|
3187
3247
|
if (!isLikeNone(markdown)) {
|
|
3188
3248
|
_assertClass(markdown, WasmMarkdownResult);
|
|
3189
|
-
|
|
3249
|
+
ptr12 = markdown.__destroy_into_raw();
|
|
3190
3250
|
}
|
|
3191
|
-
let
|
|
3251
|
+
let ptr13 = 0;
|
|
3192
3252
|
if (!isLikeNone(extractionMeta)) {
|
|
3193
3253
|
_assertClass(extractionMeta, WasmExtractionMeta);
|
|
3194
|
-
|
|
3254
|
+
ptr13 = extractionMeta.__destroy_into_raw();
|
|
3195
3255
|
}
|
|
3196
|
-
let
|
|
3256
|
+
let ptr14 = 0;
|
|
3197
3257
|
if (!isLikeNone(downloadedDocument)) {
|
|
3198
3258
|
_assertClass(downloadedDocument, WasmDownloadedDocument);
|
|
3199
|
-
|
|
3259
|
+
ptr14 = downloadedDocument.__destroy_into_raw();
|
|
3200
3260
|
}
|
|
3201
|
-
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, isLikeNone(extractedData) ? 0 : addToExternrefTable0(extractedData),
|
|
3261
|
+
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);
|
|
3202
3262
|
this.__wbg_ptr = ret;
|
|
3203
3263
|
WasmCrawlPageResultFinalization.register(this, this.__wbg_ptr, this);
|
|
3204
3264
|
return this;
|
|
@@ -3400,6 +3460,14 @@ export class WasmCrawlPageResult {
|
|
|
3400
3460
|
set redirectCount(value) {
|
|
3401
3461
|
wasm.wasmcrawlpageresult_set_redirectCount(this.__wbg_ptr, value);
|
|
3402
3462
|
}
|
|
3463
|
+
/**
|
|
3464
|
+
* @param {string[]} value
|
|
3465
|
+
*/
|
|
3466
|
+
set ssrfRefusedUrls(value) {
|
|
3467
|
+
const ptr0 = passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc);
|
|
3468
|
+
const len0 = WASM_VECTOR_LEN;
|
|
3469
|
+
wasm.wasmcrawlpageresult_set_ssrfRefusedUrls(this.__wbg_ptr, ptr0, len0);
|
|
3470
|
+
}
|
|
3403
3471
|
/**
|
|
3404
3472
|
* @param {number} value
|
|
3405
3473
|
*/
|
|
@@ -3426,6 +3494,15 @@ export class WasmCrawlPageResult {
|
|
|
3426
3494
|
set wasSkipped(value) {
|
|
3427
3495
|
wasm.wasmcrawlpageresult_set_wasSkipped(this.__wbg_ptr, value);
|
|
3428
3496
|
}
|
|
3497
|
+
/**
|
|
3498
|
+
* @returns {string[]}
|
|
3499
|
+
*/
|
|
3500
|
+
get ssrfRefusedUrls() {
|
|
3501
|
+
const ret = wasm.wasmcrawlpageresult_ssrfRefusedUrls(this.__wbg_ptr);
|
|
3502
|
+
var v1 = getArrayJsValueFromWasm0(ret[0], ret[1]);
|
|
3503
|
+
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
3504
|
+
return v1;
|
|
3505
|
+
}
|
|
3429
3506
|
/**
|
|
3430
3507
|
* @returns {number}
|
|
3431
3508
|
*/
|
|
@@ -4981,18 +5058,21 @@ export class WasmInteractionResult {
|
|
|
4981
5058
|
* @param {WasmActionResult[] | null} [actionResults]
|
|
4982
5059
|
* @param {string | null} [finalHtml]
|
|
4983
5060
|
* @param {string | null} [finalUrl]
|
|
5061
|
+
* @param {string[] | null} [ssrfRefusedUrls]
|
|
4984
5062
|
* @param {string | null} [screenshotBase64]
|
|
4985
5063
|
*/
|
|
4986
|
-
constructor(actionResults, finalHtml, finalUrl, screenshotBase64) {
|
|
5064
|
+
constructor(actionResults, finalHtml, finalUrl, ssrfRefusedUrls, screenshotBase64) {
|
|
4987
5065
|
var ptr0 = isLikeNone(actionResults) ? 0 : passArrayJsValueToWasm0(actionResults, wasm.__wbindgen_malloc);
|
|
4988
5066
|
var len0 = WASM_VECTOR_LEN;
|
|
4989
5067
|
var ptr1 = isLikeNone(finalHtml) ? 0 : passStringToWasm0(finalHtml, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
4990
5068
|
var len1 = WASM_VECTOR_LEN;
|
|
4991
5069
|
var ptr2 = isLikeNone(finalUrl) ? 0 : passStringToWasm0(finalUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
4992
5070
|
var len2 = WASM_VECTOR_LEN;
|
|
4993
|
-
var ptr3 = isLikeNone(
|
|
5071
|
+
var ptr3 = isLikeNone(ssrfRefusedUrls) ? 0 : passArrayJsValueToWasm0(ssrfRefusedUrls, wasm.__wbindgen_malloc);
|
|
4994
5072
|
var len3 = WASM_VECTOR_LEN;
|
|
4995
|
-
|
|
5073
|
+
var ptr4 = isLikeNone(screenshotBase64) ? 0 : passStringToWasm0(screenshotBase64, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
5074
|
+
var len4 = WASM_VECTOR_LEN;
|
|
5075
|
+
const ret = wasm.wasminteractionresult_new(ptr0, len0, ptr1, len1, ptr2, len2, ptr3, len3, ptr4, len4);
|
|
4996
5076
|
this.__wbg_ptr = ret;
|
|
4997
5077
|
WasmInteractionResultFinalization.register(this, this.__wbg_ptr, this);
|
|
4998
5078
|
return this;
|
|
@@ -5041,6 +5121,23 @@ export class WasmInteractionResult {
|
|
|
5041
5121
|
var len0 = WASM_VECTOR_LEN;
|
|
5042
5122
|
wasm.wasminteractionresult_set_screenshotBase64(this.__wbg_ptr, ptr0, len0);
|
|
5043
5123
|
}
|
|
5124
|
+
/**
|
|
5125
|
+
* @param {string[]} value
|
|
5126
|
+
*/
|
|
5127
|
+
set ssrfRefusedUrls(value) {
|
|
5128
|
+
const ptr0 = passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc);
|
|
5129
|
+
const len0 = WASM_VECTOR_LEN;
|
|
5130
|
+
wasm.wasminteractionresult_set_ssrfRefusedUrls(this.__wbg_ptr, ptr0, len0);
|
|
5131
|
+
}
|
|
5132
|
+
/**
|
|
5133
|
+
* @returns {string[]}
|
|
5134
|
+
*/
|
|
5135
|
+
get ssrfRefusedUrls() {
|
|
5136
|
+
const ret = wasm.wasminteractionresult_ssrfRefusedUrls(this.__wbg_ptr);
|
|
5137
|
+
var v1 = getArrayJsValueFromWasm0(ret[0], ret[1]);
|
|
5138
|
+
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
5139
|
+
return v1;
|
|
5140
|
+
}
|
|
5044
5141
|
}
|
|
5045
5142
|
if (Symbol.dispose) WasmInteractionResult.prototype[Symbol.dispose] = WasmInteractionResult.prototype.free;
|
|
5046
5143
|
|
|
@@ -7295,6 +7392,7 @@ export class WasmScrapeResult {
|
|
|
7295
7392
|
* @param {WasmDownloadedAsset[] | null} [assets]
|
|
7296
7393
|
* @param {boolean | null} [jsRenderHint]
|
|
7297
7394
|
* @param {boolean | null} [browserUsed]
|
|
7395
|
+
* @param {string[] | null} [ssrfRefusedUrls]
|
|
7298
7396
|
* @param {bigint | null} [crawlDelay]
|
|
7299
7397
|
* @param {string | null} [xRobotsTag]
|
|
7300
7398
|
* @param {string | null} [detectedCharset]
|
|
@@ -7306,7 +7404,7 @@ export class WasmScrapeResult {
|
|
|
7306
7404
|
* @param {WasmDownloadedDocument | null} [downloadedDocument]
|
|
7307
7405
|
* @param {WasmBrowserExtras | null} [browser]
|
|
7308
7406
|
*/
|
|
7309
|
-
constructor(statusCode, finalUrl, contentType, html, bodySize, metadata, links, images, feeds, jsonLd, isAllowed, noindexDetected, nofollowDetected, isPdf, wasSkipped, authHeaderSent, assets, jsRenderHint, browserUsed, crawlDelay, xRobotsTag, detectedCharset, responseMeta, markdown, extractedData, extractionMeta, screenshotBase64, downloadedDocument, browser) {
|
|
7407
|
+
constructor(statusCode, finalUrl, contentType, html, bodySize, metadata, links, images, feeds, jsonLd, isAllowed, noindexDetected, nofollowDetected, isPdf, wasSkipped, authHeaderSent, assets, jsRenderHint, browserUsed, ssrfRefusedUrls, crawlDelay, xRobotsTag, detectedCharset, responseMeta, markdown, extractedData, extractionMeta, screenshotBase64, downloadedDocument, browser) {
|
|
7310
7408
|
var ptr0 = isLikeNone(finalUrl) ? 0 : passStringToWasm0(finalUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
7311
7409
|
var len0 = WASM_VECTOR_LEN;
|
|
7312
7410
|
var ptr1 = isLikeNone(contentType) ? 0 : passStringToWasm0(contentType, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
@@ -7328,38 +7426,40 @@ export class WasmScrapeResult {
|
|
|
7328
7426
|
var len7 = WASM_VECTOR_LEN;
|
|
7329
7427
|
var ptr8 = isLikeNone(assets) ? 0 : passArrayJsValueToWasm0(assets, wasm.__wbindgen_malloc);
|
|
7330
7428
|
var len8 = WASM_VECTOR_LEN;
|
|
7331
|
-
var ptr9 = isLikeNone(
|
|
7429
|
+
var ptr9 = isLikeNone(ssrfRefusedUrls) ? 0 : passArrayJsValueToWasm0(ssrfRefusedUrls, wasm.__wbindgen_malloc);
|
|
7332
7430
|
var len9 = WASM_VECTOR_LEN;
|
|
7333
|
-
var ptr10 = isLikeNone(
|
|
7431
|
+
var ptr10 = isLikeNone(xRobotsTag) ? 0 : passStringToWasm0(xRobotsTag, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
7334
7432
|
var len10 = WASM_VECTOR_LEN;
|
|
7335
|
-
|
|
7433
|
+
var ptr11 = isLikeNone(detectedCharset) ? 0 : passStringToWasm0(detectedCharset, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
7434
|
+
var len11 = WASM_VECTOR_LEN;
|
|
7435
|
+
let ptr12 = 0;
|
|
7336
7436
|
if (!isLikeNone(responseMeta)) {
|
|
7337
7437
|
_assertClass(responseMeta, WasmResponseMeta);
|
|
7338
|
-
|
|
7438
|
+
ptr12 = responseMeta.__destroy_into_raw();
|
|
7339
7439
|
}
|
|
7340
|
-
let
|
|
7440
|
+
let ptr13 = 0;
|
|
7341
7441
|
if (!isLikeNone(markdown)) {
|
|
7342
7442
|
_assertClass(markdown, WasmMarkdownResult);
|
|
7343
|
-
|
|
7443
|
+
ptr13 = markdown.__destroy_into_raw();
|
|
7344
7444
|
}
|
|
7345
|
-
let
|
|
7445
|
+
let ptr14 = 0;
|
|
7346
7446
|
if (!isLikeNone(extractionMeta)) {
|
|
7347
7447
|
_assertClass(extractionMeta, WasmExtractionMeta);
|
|
7348
|
-
|
|
7448
|
+
ptr14 = extractionMeta.__destroy_into_raw();
|
|
7349
7449
|
}
|
|
7350
|
-
var
|
|
7351
|
-
var
|
|
7352
|
-
let
|
|
7450
|
+
var ptr15 = isLikeNone(screenshotBase64) ? 0 : passStringToWasm0(screenshotBase64, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
7451
|
+
var len15 = WASM_VECTOR_LEN;
|
|
7452
|
+
let ptr16 = 0;
|
|
7353
7453
|
if (!isLikeNone(downloadedDocument)) {
|
|
7354
7454
|
_assertClass(downloadedDocument, WasmDownloadedDocument);
|
|
7355
|
-
|
|
7455
|
+
ptr16 = downloadedDocument.__destroy_into_raw();
|
|
7356
7456
|
}
|
|
7357
|
-
let
|
|
7457
|
+
let ptr17 = 0;
|
|
7358
7458
|
if (!isLikeNone(browser)) {
|
|
7359
7459
|
_assertClass(browser, WasmBrowserExtras);
|
|
7360
|
-
|
|
7460
|
+
ptr17 = browser.__destroy_into_raw();
|
|
7361
7461
|
}
|
|
7362
|
-
const ret = wasm.wasmscraperesult_new(isLikeNone(statusCode) ? 0xFFFFFF : statusCode, ptr0, len0, ptr1, len1, ptr2, len2, isLikeNone(bodySize) ? Number.MAX_SAFE_INTEGER : (bodySize) >>> 0, ptr3, ptr4, len4, ptr5, len5, ptr6, len6, ptr7, len7, isLikeNone(isAllowed) ? 0xFFFFFF : isAllowed ? 1 : 0, isLikeNone(noindexDetected) ? 0xFFFFFF : noindexDetected ? 1 : 0, isLikeNone(nofollowDetected) ? 0xFFFFFF : nofollowDetected ? 1 : 0, isLikeNone(isPdf) ? 0xFFFFFF : isPdf ? 1 : 0, isLikeNone(wasSkipped) ? 0xFFFFFF : wasSkipped ? 1 : 0, isLikeNone(authHeaderSent) ? 0xFFFFFF : authHeaderSent ? 1 : 0, ptr8, len8, isLikeNone(jsRenderHint) ? 0xFFFFFF : jsRenderHint ? 1 : 0, isLikeNone(browserUsed) ? 0xFFFFFF : browserUsed ? 1 : 0, !isLikeNone(crawlDelay), isLikeNone(crawlDelay) ? BigInt(0) : crawlDelay,
|
|
7462
|
+
const ret = wasm.wasmscraperesult_new(isLikeNone(statusCode) ? 0xFFFFFF : statusCode, ptr0, len0, ptr1, len1, ptr2, len2, isLikeNone(bodySize) ? Number.MAX_SAFE_INTEGER : (bodySize) >>> 0, ptr3, ptr4, len4, ptr5, len5, ptr6, len6, ptr7, len7, isLikeNone(isAllowed) ? 0xFFFFFF : isAllowed ? 1 : 0, isLikeNone(noindexDetected) ? 0xFFFFFF : noindexDetected ? 1 : 0, isLikeNone(nofollowDetected) ? 0xFFFFFF : nofollowDetected ? 1 : 0, isLikeNone(isPdf) ? 0xFFFFFF : isPdf ? 1 : 0, isLikeNone(wasSkipped) ? 0xFFFFFF : wasSkipped ? 1 : 0, isLikeNone(authHeaderSent) ? 0xFFFFFF : authHeaderSent ? 1 : 0, ptr8, len8, isLikeNone(jsRenderHint) ? 0xFFFFFF : jsRenderHint ? 1 : 0, isLikeNone(browserUsed) ? 0xFFFFFF : browserUsed ? 1 : 0, ptr9, len9, !isLikeNone(crawlDelay), isLikeNone(crawlDelay) ? BigInt(0) : crawlDelay, ptr10, len10, ptr11, len11, ptr12, ptr13, isLikeNone(extractedData) ? 0 : addToExternrefTable0(extractedData), ptr14, ptr15, len15, ptr16, ptr17);
|
|
7363
7463
|
this.__wbg_ptr = ret;
|
|
7364
7464
|
WasmScrapeResultFinalization.register(this, this.__wbg_ptr, this);
|
|
7365
7465
|
return this;
|
|
@@ -7600,6 +7700,14 @@ export class WasmScrapeResult {
|
|
|
7600
7700
|
var len0 = WASM_VECTOR_LEN;
|
|
7601
7701
|
wasm.wasmscraperesult_set_screenshotBase64(this.__wbg_ptr, ptr0, len0);
|
|
7602
7702
|
}
|
|
7703
|
+
/**
|
|
7704
|
+
* @param {string[]} value
|
|
7705
|
+
*/
|
|
7706
|
+
set ssrfRefusedUrls(value) {
|
|
7707
|
+
const ptr0 = passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc);
|
|
7708
|
+
const len0 = WASM_VECTOR_LEN;
|
|
7709
|
+
wasm.wasmscraperesult_set_ssrfRefusedUrls(this.__wbg_ptr, ptr0, len0);
|
|
7710
|
+
}
|
|
7603
7711
|
/**
|
|
7604
7712
|
* @param {number} value
|
|
7605
7713
|
*/
|
|
@@ -7620,6 +7728,15 @@ export class WasmScrapeResult {
|
|
|
7620
7728
|
var len0 = WASM_VECTOR_LEN;
|
|
7621
7729
|
wasm.wasmscraperesult_set_xRobotsTag(this.__wbg_ptr, ptr0, len0);
|
|
7622
7730
|
}
|
|
7731
|
+
/**
|
|
7732
|
+
* @returns {string[]}
|
|
7733
|
+
*/
|
|
7734
|
+
get ssrfRefusedUrls() {
|
|
7735
|
+
const ret = wasm.wasmscraperesult_ssrfRefusedUrls(this.__wbg_ptr);
|
|
7736
|
+
var v1 = getArrayJsValueFromWasm0(ret[0], ret[1]);
|
|
7737
|
+
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
7738
|
+
return v1;
|
|
7739
|
+
}
|
|
7623
7740
|
/**
|
|
7624
7741
|
* @returns {number}
|
|
7625
7742
|
*/
|
|
@@ -8577,13 +8694,13 @@ export function __wbg_wasmsitemapurl_unwrap(arg0) {
|
|
|
8577
8694
|
return ret;
|
|
8578
8695
|
}
|
|
8579
8696
|
export function __wbindgen_generic_0000000000000001(arg0, arg1) {
|
|
8580
|
-
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx:
|
|
8697
|
+
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1352, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
|
|
8581
8698
|
const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h2eedacf566f3748c);
|
|
8582
8699
|
return ret;
|
|
8583
8700
|
}
|
|
8584
8701
|
export function __wbindgen_generic_0000000000000002(arg0, arg1) {
|
|
8585
|
-
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx:
|
|
8586
|
-
const ret = makeMutClosure(arg0, arg1,
|
|
8702
|
+
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1243, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
|
|
8703
|
+
const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h3a7ccac3e94bbb5c);
|
|
8587
8704
|
return ret;
|
|
8588
8705
|
}
|
|
8589
8706
|
export function __wbindgen_generic_0000000000000003(arg0) {
|
|
@@ -8615,8 +8732,8 @@ export function __wbindgen_init_externref_table() {
|
|
|
8615
8732
|
table.set(offset + 2, true);
|
|
8616
8733
|
table.set(offset + 3, false);
|
|
8617
8734
|
}
|
|
8618
|
-
function
|
|
8619
|
-
wasm.
|
|
8735
|
+
function wasm_bindgen__convert__closures_____invoke__h3a7ccac3e94bbb5c(arg0, arg1) {
|
|
8736
|
+
wasm.wasm_bindgen__convert__closures_____invoke__h3a7ccac3e94bbb5c(arg0, arg1);
|
|
8620
8737
|
}
|
|
8621
8738
|
|
|
8622
8739
|
function wasm_bindgen__convert__closures_____invoke__h2eedacf566f3748c(arg0, arg1, arg2) {
|
|
Binary file
|