@xberg-io/crawlberg-wasm 1.12.0 → 1.12.3
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 +10 -3
- package/pkg/bundler/crawlberg_wasm_bg.js +114 -8
- package/pkg/bundler/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/bundler/crawlberg_wasm_bg.wasm.d.ts +15 -3
- package/pkg/bundler/package.json +1 -1
- package/pkg/deno/crawlberg_wasm.d.ts +10 -3
- package/pkg/deno/crawlberg_wasm.js +114 -8
- package/pkg/deno/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/deno/crawlberg_wasm_bg.wasm.d.ts +15 -3
- package/pkg/nodejs/crawlberg_wasm.d.ts +10 -3
- package/pkg/nodejs/crawlberg_wasm.js +114 -8
- package/pkg/nodejs/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/nodejs/crawlberg_wasm_bg.wasm.d.ts +15 -3
- package/pkg/nodejs/package.json +1 -1
- package/pkg/web/crawlberg_wasm.d.ts +25 -6
- package/pkg/web/crawlberg_wasm.js +114 -8
- package/pkg/web/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/web/crawlberg_wasm_bg.wasm.d.ts +15 -3
- package/pkg/web/package.json +1 -1
package/package.json
CHANGED
|
@@ -260,7 +260,7 @@ export class WasmBrowserConfig {
|
|
|
260
260
|
* keep a handle you already hold, build with `default()` and assign the property instead:
|
|
261
261
|
* the generated setter borrows and leaves your handle alive.
|
|
262
262
|
*/
|
|
263
|
-
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, scrollContainers?: boolean | null, domQuietWindow?: bigint | null);
|
|
263
|
+
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, scrollContainers?: boolean | null, domQuietWindow?: bigint | null, domQuietCap?: bigint | null);
|
|
264
264
|
get backend(): string;
|
|
265
265
|
set backend(value: WasmBrowserBackend);
|
|
266
266
|
blockUrlPatterns: string[];
|
|
@@ -268,6 +268,8 @@ export class WasmBrowserConfig {
|
|
|
268
268
|
chromeArgs: string[];
|
|
269
269
|
get chromePath(): string | undefined;
|
|
270
270
|
set chromePath(value: string | null | undefined);
|
|
271
|
+
get domQuietCap(): bigint | undefined;
|
|
272
|
+
set domQuietCap(value: bigint | null | undefined);
|
|
271
273
|
get domQuietWindow(): bigint | undefined;
|
|
272
274
|
set domQuietWindow(value: bigint | null | undefined);
|
|
273
275
|
get endpoint(): string | undefined;
|
|
@@ -430,12 +432,16 @@ export class WasmContentConfig {
|
|
|
430
432
|
*/
|
|
431
433
|
copyForTransfer(): WasmContentConfig;
|
|
432
434
|
static default(): WasmContentConfig;
|
|
433
|
-
constructor(outputFormat?: string | null, preprocessingPreset?: string | null, removeNavigation?: boolean | null, removeForms?: boolean | null, stripTags?: string[] | null, preserveTags?: string[] | null, excludeSelectors?: string[] | null, skipImages?: boolean | null, wrap?: boolean | null, wrapWidth?: number | null, includeDocumentStructure?: boolean | null, extractMetadata?: boolean | null, maxDepth?: number | null, stripNewlines?: boolean | null, bullets?: string | null, listIndentWidth?: number | null, strongEmSymbol?: string | null, compactTables?: boolean | null, keepInlineImagesIn?: string[] | null, keepHidden?: boolean | null);
|
|
435
|
+
constructor(outputFormat?: string | null, preprocessingPreset?: string | null, removeNavigation?: boolean | null, removeForms?: boolean | null, stripTags?: string[] | null, preserveTags?: string[] | null, excludeSelectors?: string[] | null, skipImages?: boolean | null, wrap?: boolean | null, wrapWidth?: number | null, includeDocumentStructure?: boolean | null, extractMetadata?: boolean | null, maxDepth?: number | null, stripNewlines?: boolean | null, bullets?: string | null, listIndentWidth?: number | null, strongEmSymbol?: string | null, compactTables?: boolean | null, keepInlineImagesIn?: string[] | null, keepHidden?: boolean | null, expandAbbreviations?: boolean | null, escapeMisc?: boolean | null, escapeAsterisks?: boolean | null, includeIconLabels?: boolean | null);
|
|
434
436
|
bullets: string;
|
|
435
437
|
compactTables: boolean;
|
|
438
|
+
escapeAsterisks: boolean;
|
|
439
|
+
escapeMisc: boolean;
|
|
436
440
|
excludeSelectors: string[];
|
|
441
|
+
expandAbbreviations: boolean;
|
|
437
442
|
extractMetadata: boolean;
|
|
438
443
|
includeDocumentStructure: boolean;
|
|
444
|
+
includeIconLabels: boolean;
|
|
439
445
|
keepHidden: boolean;
|
|
440
446
|
keepInlineImagesIn: string[];
|
|
441
447
|
listIndentWidth: number;
|
|
@@ -509,7 +515,7 @@ export class WasmCrawlConfig {
|
|
|
509
515
|
* keep a handle you already hold, build with `default()` and assign the property instead:
|
|
510
516
|
* the generated setter borrows and leaves your handle alive.
|
|
511
517
|
*/
|
|
512
|
-
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, dangerAcceptInvalidCerts?: boolean | null);
|
|
518
|
+
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, dangerAcceptInvalidCerts?: boolean | null, subdomainHeaderAllowlist?: string[] | null);
|
|
513
519
|
/**
|
|
514
520
|
* Validate the configuration, returning an error if any values are invalid.
|
|
515
521
|
*
|
|
@@ -615,6 +621,7 @@ export class WasmCrawlConfig {
|
|
|
615
621
|
ssrf: WasmSsrfPolicy;
|
|
616
622
|
stayOnDomain: boolean;
|
|
617
623
|
stripTrackingParams: boolean;
|
|
624
|
+
subdomainHeaderAllowlist: string[];
|
|
618
625
|
trackingParams: string[];
|
|
619
626
|
get userAgent(): string | undefined;
|
|
620
627
|
set userAgent(value: string | null | undefined);
|
|
@@ -1103,6 +1103,13 @@ export class WasmBrowserConfig {
|
|
|
1103
1103
|
const ret = wasm.wasmbrowserconfig_default();
|
|
1104
1104
|
return WasmBrowserConfig.__wrap(ret);
|
|
1105
1105
|
}
|
|
1106
|
+
/**
|
|
1107
|
+
* @returns {bigint | undefined}
|
|
1108
|
+
*/
|
|
1109
|
+
get domQuietCap() {
|
|
1110
|
+
const ret = wasm.wasmbrowserconfig_domQuietCap(this.__wbg_ptr);
|
|
1111
|
+
return ret[0] === 0 ? undefined : BigInt.asUintN(64, ret[1]);
|
|
1112
|
+
}
|
|
1106
1113
|
/**
|
|
1107
1114
|
* @returns {bigint | undefined}
|
|
1108
1115
|
*/
|
|
@@ -1181,8 +1188,9 @@ export class WasmBrowserConfig {
|
|
|
1181
1188
|
* @param {string | null} [chromePath]
|
|
1182
1189
|
* @param {boolean | null} [scrollContainers]
|
|
1183
1190
|
* @param {bigint | null} [domQuietWindow]
|
|
1191
|
+
* @param {bigint | null} [domQuietCap]
|
|
1184
1192
|
*/
|
|
1185
|
-
constructor(mode, backend, timeout, overallTimeout, shutdownTimeout, wait, blockUrlPatterns, captureNetworkEvents, sessionAffinity, chromeArgs, endpoint, waitSelector, extraWait, proxy, evalScript, robotsUserAgent, chromePath, scrollContainers, domQuietWindow) {
|
|
1193
|
+
constructor(mode, backend, timeout, overallTimeout, shutdownTimeout, wait, blockUrlPatterns, captureNetworkEvents, sessionAffinity, chromeArgs, endpoint, waitSelector, extraWait, proxy, evalScript, robotsUserAgent, chromePath, scrollContainers, domQuietWindow, domQuietCap) {
|
|
1186
1194
|
var ptr0 = isLikeNone(blockUrlPatterns) ? 0 : passArrayJsValueToWasm0(blockUrlPatterns, wasm.__wbindgen_malloc);
|
|
1187
1195
|
var len0 = WASM_VECTOR_LEN;
|
|
1188
1196
|
var ptr1 = isLikeNone(chromeArgs) ? 0 : passArrayJsValueToWasm0(chromeArgs, wasm.__wbindgen_malloc);
|
|
@@ -1202,7 +1210,7 @@ export class WasmBrowserConfig {
|
|
|
1202
1210
|
var len6 = WASM_VECTOR_LEN;
|
|
1203
1211
|
var ptr7 = isLikeNone(chromePath) ? 0 : passStringToWasm0(chromePath, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1204
1212
|
var len7 = WASM_VECTOR_LEN;
|
|
1205
|
-
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, isLikeNone(scrollContainers) ? 0xFFFFFF : scrollContainers ? 1 : 0, !isLikeNone(domQuietWindow), isLikeNone(domQuietWindow) ? BigInt(0) : domQuietWindow);
|
|
1213
|
+
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, isLikeNone(scrollContainers) ? 0xFFFFFF : scrollContainers ? 1 : 0, !isLikeNone(domQuietWindow), isLikeNone(domQuietWindow) ? BigInt(0) : domQuietWindow, !isLikeNone(domQuietCap), isLikeNone(domQuietCap) ? BigInt(0) : domQuietCap);
|
|
1206
1214
|
this.__wbg_ptr = ret;
|
|
1207
1215
|
WasmBrowserConfigFinalization.register(this, this.__wbg_ptr, this);
|
|
1208
1216
|
return this;
|
|
@@ -1287,6 +1295,12 @@ export class WasmBrowserConfig {
|
|
|
1287
1295
|
var len0 = WASM_VECTOR_LEN;
|
|
1288
1296
|
wasm.wasmbrowserconfig_set_chromePath(this.__wbg_ptr, ptr0, len0);
|
|
1289
1297
|
}
|
|
1298
|
+
/**
|
|
1299
|
+
* @param {bigint | null} [value]
|
|
1300
|
+
*/
|
|
1301
|
+
set domQuietCap(value) {
|
|
1302
|
+
wasm.wasmbrowserconfig_set_domQuietCap(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? BigInt(0) : value);
|
|
1303
|
+
}
|
|
1290
1304
|
/**
|
|
1291
1305
|
* @param {bigint | null} [value]
|
|
1292
1306
|
*/
|
|
@@ -1858,6 +1872,20 @@ export class WasmContentConfig {
|
|
|
1858
1872
|
const ret = wasm.wasmcontentconfig_default();
|
|
1859
1873
|
return WasmContentConfig.__wrap(ret);
|
|
1860
1874
|
}
|
|
1875
|
+
/**
|
|
1876
|
+
* @returns {boolean}
|
|
1877
|
+
*/
|
|
1878
|
+
get escapeAsterisks() {
|
|
1879
|
+
const ret = wasm.wasmcontentconfig_escapeAsterisks(this.__wbg_ptr);
|
|
1880
|
+
return ret !== 0;
|
|
1881
|
+
}
|
|
1882
|
+
/**
|
|
1883
|
+
* @returns {boolean}
|
|
1884
|
+
*/
|
|
1885
|
+
get escapeMisc() {
|
|
1886
|
+
const ret = wasm.wasmcontentconfig_escapeMisc(this.__wbg_ptr);
|
|
1887
|
+
return ret !== 0;
|
|
1888
|
+
}
|
|
1861
1889
|
/**
|
|
1862
1890
|
* @returns {string[]}
|
|
1863
1891
|
*/
|
|
@@ -1867,6 +1895,13 @@ export class WasmContentConfig {
|
|
|
1867
1895
|
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
1868
1896
|
return v1;
|
|
1869
1897
|
}
|
|
1898
|
+
/**
|
|
1899
|
+
* @returns {boolean}
|
|
1900
|
+
*/
|
|
1901
|
+
get expandAbbreviations() {
|
|
1902
|
+
const ret = wasm.wasmcontentconfig_expandAbbreviations(this.__wbg_ptr);
|
|
1903
|
+
return ret !== 0;
|
|
1904
|
+
}
|
|
1870
1905
|
/**
|
|
1871
1906
|
* @returns {boolean}
|
|
1872
1907
|
*/
|
|
@@ -1881,6 +1916,13 @@ export class WasmContentConfig {
|
|
|
1881
1916
|
const ret = wasm.wasmcontentconfig_includeDocumentStructure(this.__wbg_ptr);
|
|
1882
1917
|
return ret !== 0;
|
|
1883
1918
|
}
|
|
1919
|
+
/**
|
|
1920
|
+
* @returns {boolean}
|
|
1921
|
+
*/
|
|
1922
|
+
get includeIconLabels() {
|
|
1923
|
+
const ret = wasm.wasmcontentconfig_includeIconLabels(this.__wbg_ptr);
|
|
1924
|
+
return ret !== 0;
|
|
1925
|
+
}
|
|
1884
1926
|
/**
|
|
1885
1927
|
* @returns {boolean}
|
|
1886
1928
|
*/
|
|
@@ -1932,8 +1974,12 @@ export class WasmContentConfig {
|
|
|
1932
1974
|
* @param {boolean | null} [compactTables]
|
|
1933
1975
|
* @param {string[] | null} [keepInlineImagesIn]
|
|
1934
1976
|
* @param {boolean | null} [keepHidden]
|
|
1977
|
+
* @param {boolean | null} [expandAbbreviations]
|
|
1978
|
+
* @param {boolean | null} [escapeMisc]
|
|
1979
|
+
* @param {boolean | null} [escapeAsterisks]
|
|
1980
|
+
* @param {boolean | null} [includeIconLabels]
|
|
1935
1981
|
*/
|
|
1936
|
-
constructor(outputFormat, preprocessingPreset, removeNavigation, removeForms, stripTags, preserveTags, excludeSelectors, skipImages, wrap, wrapWidth, includeDocumentStructure, extractMetadata, maxDepth, stripNewlines, bullets, listIndentWidth, strongEmSymbol, compactTables, keepInlineImagesIn, keepHidden) {
|
|
1982
|
+
constructor(outputFormat, preprocessingPreset, removeNavigation, removeForms, stripTags, preserveTags, excludeSelectors, skipImages, wrap, wrapWidth, includeDocumentStructure, extractMetadata, maxDepth, stripNewlines, bullets, listIndentWidth, strongEmSymbol, compactTables, keepInlineImagesIn, keepHidden, expandAbbreviations, escapeMisc, escapeAsterisks, includeIconLabels) {
|
|
1937
1983
|
var ptr0 = isLikeNone(outputFormat) ? 0 : passStringToWasm0(outputFormat, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1938
1984
|
var len0 = WASM_VECTOR_LEN;
|
|
1939
1985
|
var ptr1 = isLikeNone(preprocessingPreset) ? 0 : passStringToWasm0(preprocessingPreset, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
@@ -1950,7 +1996,7 @@ export class WasmContentConfig {
|
|
|
1950
1996
|
var len6 = WASM_VECTOR_LEN;
|
|
1951
1997
|
var ptr7 = isLikeNone(keepInlineImagesIn) ? 0 : passArrayJsValueToWasm0(keepInlineImagesIn, wasm.__wbindgen_malloc);
|
|
1952
1998
|
var len7 = WASM_VECTOR_LEN;
|
|
1953
|
-
const ret = wasm.wasmcontentconfig_new(ptr0, len0, ptr1, len1, isLikeNone(removeNavigation) ? 0xFFFFFF : removeNavigation ? 1 : 0, isLikeNone(removeForms) ? 0xFFFFFF : removeForms ? 1 : 0, ptr2, len2, ptr3, len3, ptr4, len4, isLikeNone(skipImages) ? 0xFFFFFF : skipImages ? 1 : 0, isLikeNone(wrap) ? 0xFFFFFF : wrap ? 1 : 0, isLikeNone(wrapWidth) ? Number.MAX_SAFE_INTEGER : (wrapWidth) >>> 0, isLikeNone(includeDocumentStructure) ? 0xFFFFFF : includeDocumentStructure ? 1 : 0, isLikeNone(extractMetadata) ? 0xFFFFFF : extractMetadata ? 1 : 0, isLikeNone(maxDepth) ? Number.MAX_SAFE_INTEGER : (maxDepth) >>> 0, isLikeNone(stripNewlines) ? 0xFFFFFF : stripNewlines ? 1 : 0, ptr5, len5, isLikeNone(listIndentWidth) ? Number.MAX_SAFE_INTEGER : (listIndentWidth) >>> 0, ptr6, len6, isLikeNone(compactTables) ? 0xFFFFFF : compactTables ? 1 : 0, ptr7, len7, isLikeNone(keepHidden) ? 0xFFFFFF : keepHidden ? 1 : 0);
|
|
1999
|
+
const ret = wasm.wasmcontentconfig_new(ptr0, len0, ptr1, len1, isLikeNone(removeNavigation) ? 0xFFFFFF : removeNavigation ? 1 : 0, isLikeNone(removeForms) ? 0xFFFFFF : removeForms ? 1 : 0, ptr2, len2, ptr3, len3, ptr4, len4, isLikeNone(skipImages) ? 0xFFFFFF : skipImages ? 1 : 0, isLikeNone(wrap) ? 0xFFFFFF : wrap ? 1 : 0, isLikeNone(wrapWidth) ? Number.MAX_SAFE_INTEGER : (wrapWidth) >>> 0, isLikeNone(includeDocumentStructure) ? 0xFFFFFF : includeDocumentStructure ? 1 : 0, isLikeNone(extractMetadata) ? 0xFFFFFF : extractMetadata ? 1 : 0, isLikeNone(maxDepth) ? Number.MAX_SAFE_INTEGER : (maxDepth) >>> 0, isLikeNone(stripNewlines) ? 0xFFFFFF : stripNewlines ? 1 : 0, ptr5, len5, isLikeNone(listIndentWidth) ? Number.MAX_SAFE_INTEGER : (listIndentWidth) >>> 0, ptr6, len6, isLikeNone(compactTables) ? 0xFFFFFF : compactTables ? 1 : 0, ptr7, len7, isLikeNone(keepHidden) ? 0xFFFFFF : keepHidden ? 1 : 0, isLikeNone(expandAbbreviations) ? 0xFFFFFF : expandAbbreviations ? 1 : 0, isLikeNone(escapeMisc) ? 0xFFFFFF : escapeMisc ? 1 : 0, isLikeNone(escapeAsterisks) ? 0xFFFFFF : escapeAsterisks ? 1 : 0, isLikeNone(includeIconLabels) ? 0xFFFFFF : includeIconLabels ? 1 : 0);
|
|
1954
2000
|
this.__wbg_ptr = ret;
|
|
1955
2001
|
WasmContentConfigFinalization.register(this, this.__wbg_ptr, this);
|
|
1956
2002
|
return this;
|
|
@@ -2022,6 +2068,18 @@ export class WasmContentConfig {
|
|
|
2022
2068
|
set compactTables(value) {
|
|
2023
2069
|
wasm.wasmcontentconfig_set_compactTables(this.__wbg_ptr, value);
|
|
2024
2070
|
}
|
|
2071
|
+
/**
|
|
2072
|
+
* @param {boolean} value
|
|
2073
|
+
*/
|
|
2074
|
+
set escapeAsterisks(value) {
|
|
2075
|
+
wasm.wasmcontentconfig_set_escapeAsterisks(this.__wbg_ptr, value);
|
|
2076
|
+
}
|
|
2077
|
+
/**
|
|
2078
|
+
* @param {boolean} value
|
|
2079
|
+
*/
|
|
2080
|
+
set escapeMisc(value) {
|
|
2081
|
+
wasm.wasmcontentconfig_set_escapeMisc(this.__wbg_ptr, value);
|
|
2082
|
+
}
|
|
2025
2083
|
/**
|
|
2026
2084
|
* @param {string[]} value
|
|
2027
2085
|
*/
|
|
@@ -2030,6 +2088,12 @@ export class WasmContentConfig {
|
|
|
2030
2088
|
const len0 = WASM_VECTOR_LEN;
|
|
2031
2089
|
wasm.wasmcontentconfig_set_excludeSelectors(this.__wbg_ptr, ptr0, len0);
|
|
2032
2090
|
}
|
|
2091
|
+
/**
|
|
2092
|
+
* @param {boolean} value
|
|
2093
|
+
*/
|
|
2094
|
+
set expandAbbreviations(value) {
|
|
2095
|
+
wasm.wasmcontentconfig_set_expandAbbreviations(this.__wbg_ptr, value);
|
|
2096
|
+
}
|
|
2033
2097
|
/**
|
|
2034
2098
|
* @param {boolean} value
|
|
2035
2099
|
*/
|
|
@@ -2042,6 +2106,12 @@ export class WasmContentConfig {
|
|
|
2042
2106
|
set includeDocumentStructure(value) {
|
|
2043
2107
|
wasm.wasmcontentconfig_set_includeDocumentStructure(this.__wbg_ptr, value);
|
|
2044
2108
|
}
|
|
2109
|
+
/**
|
|
2110
|
+
* @param {boolean} value
|
|
2111
|
+
*/
|
|
2112
|
+
set includeIconLabels(value) {
|
|
2113
|
+
wasm.wasmcontentconfig_set_includeIconLabels(this.__wbg_ptr, value);
|
|
2114
|
+
}
|
|
2045
2115
|
/**
|
|
2046
2116
|
* @param {boolean} value
|
|
2047
2117
|
*/
|
|
@@ -2758,8 +2828,9 @@ export class WasmCrawlConfig {
|
|
|
2758
2828
|
* @param {string | null} [browserProfile]
|
|
2759
2829
|
* @param {boolean | null} [ssrfDenyPrivateExplicit]
|
|
2760
2830
|
* @param {boolean | null} [dangerAcceptInvalidCerts]
|
|
2831
|
+
* @param {string[] | null} [subdomainHeaderAllowlist]
|
|
2761
2832
|
*/
|
|
2762
|
-
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, dangerAcceptInvalidCerts) {
|
|
2833
|
+
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, dangerAcceptInvalidCerts, subdomainHeaderAllowlist) {
|
|
2763
2834
|
var ptr0 = isLikeNone(includePaths) ? 0 : passArrayJsValueToWasm0(includePaths, wasm.__wbindgen_malloc);
|
|
2764
2835
|
var len0 = WASM_VECTOR_LEN;
|
|
2765
2836
|
var ptr1 = isLikeNone(excludePaths) ? 0 : passArrayJsValueToWasm0(excludePaths, wasm.__wbindgen_malloc);
|
|
@@ -2808,7 +2879,9 @@ export class WasmCrawlConfig {
|
|
|
2808
2879
|
var len16 = WASM_VECTOR_LEN;
|
|
2809
2880
|
var ptr17 = isLikeNone(browserProfile) ? 0 : passStringToWasm0(browserProfile, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
2810
2881
|
var len17 = WASM_VECTOR_LEN;
|
|
2811
|
-
|
|
2882
|
+
var ptr18 = isLikeNone(subdomainHeaderAllowlist) ? 0 : passArrayJsValueToWasm0(subdomainHeaderAllowlist, wasm.__wbindgen_malloc);
|
|
2883
|
+
var len18 = WASM_VECTOR_LEN;
|
|
2884
|
+
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, isLikeNone(dangerAcceptInvalidCerts) ? 0xFFFFFF : dangerAcceptInvalidCerts ? 1 : 0, ptr18, len18);
|
|
2812
2885
|
this.__wbg_ptr = ret;
|
|
2813
2886
|
WasmCrawlConfigFinalization.register(this, this.__wbg_ptr, this);
|
|
2814
2887
|
return this;
|
|
@@ -3248,6 +3321,14 @@ export class WasmCrawlConfig {
|
|
|
3248
3321
|
set stripTrackingParams(value) {
|
|
3249
3322
|
wasm.wasmcrawlconfig_set_stripTrackingParams(this.__wbg_ptr, value);
|
|
3250
3323
|
}
|
|
3324
|
+
/**
|
|
3325
|
+
* @param {string[]} value
|
|
3326
|
+
*/
|
|
3327
|
+
set subdomainHeaderAllowlist(value) {
|
|
3328
|
+
const ptr0 = passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc);
|
|
3329
|
+
const len0 = WASM_VECTOR_LEN;
|
|
3330
|
+
wasm.wasmcrawlconfig_set_subdomainHeaderAllowlist(this.__wbg_ptr, ptr0, len0);
|
|
3331
|
+
}
|
|
3251
3332
|
/**
|
|
3252
3333
|
* @param {string[]} value
|
|
3253
3334
|
*/
|
|
@@ -3319,6 +3400,15 @@ export class WasmCrawlConfig {
|
|
|
3319
3400
|
const ret = wasm.wasmcrawlconfig_stripTrackingParams(this.__wbg_ptr);
|
|
3320
3401
|
return ret !== 0;
|
|
3321
3402
|
}
|
|
3403
|
+
/**
|
|
3404
|
+
* @returns {string[]}
|
|
3405
|
+
*/
|
|
3406
|
+
get subdomainHeaderAllowlist() {
|
|
3407
|
+
const ret = wasm.wasmcrawlconfig_subdomainHeaderAllowlist(this.__wbg_ptr);
|
|
3408
|
+
var v1 = getArrayJsValueFromWasm0(ret[0], ret[1]);
|
|
3409
|
+
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
3410
|
+
return v1;
|
|
3411
|
+
}
|
|
3322
3412
|
/**
|
|
3323
3413
|
* @returns {string[]}
|
|
3324
3414
|
*/
|
|
@@ -9296,6 +9386,10 @@ export function __wbg_clearTimeout_1daad5c6ee4fcd3b(arg0) {
|
|
|
9296
9386
|
const ret = clearTimeout(arg0);
|
|
9297
9387
|
return ret;
|
|
9298
9388
|
}
|
|
9389
|
+
export function __wbg_create_7febfb959f8c7046(arg0) {
|
|
9390
|
+
const ret = Object.create(arg0);
|
|
9391
|
+
return ret;
|
|
9392
|
+
}
|
|
9299
9393
|
export function __wbg_done_b41a1d26cdb37fb6(arg0) {
|
|
9300
9394
|
const ret = arg0.done;
|
|
9301
9395
|
return ret;
|
|
@@ -9316,6 +9410,10 @@ export function __wbg_fetch_efc64baf325b084d(arg0, arg1) {
|
|
|
9316
9410
|
const ret = arg0.fetch(arg1);
|
|
9317
9411
|
return ret;
|
|
9318
9412
|
}
|
|
9413
|
+
export function __wbg_for_beb5dc4e908dd8b8(arg0, arg1) {
|
|
9414
|
+
const ret = Symbol.for(getStringFromWasm0(arg0, arg1));
|
|
9415
|
+
return ret;
|
|
9416
|
+
}
|
|
9319
9417
|
export function __wbg_getRandomValues_a608c4436c19407a() { return handleError(function (arg0, arg1) {
|
|
9320
9418
|
globalThis.crypto.getRandomValues(getArrayU8FromWasm0(arg0, arg1));
|
|
9321
9419
|
}, arguments); }
|
|
@@ -9419,6 +9517,10 @@ export function __wbg_new_2e41357094e99294() { return handleError(function () {
|
|
|
9419
9517
|
const ret = new AbortController();
|
|
9420
9518
|
return ret;
|
|
9421
9519
|
}, arguments); }
|
|
9520
|
+
export function __wbg_new_343a093a3c2ffb4e(arg0, arg1) {
|
|
9521
|
+
const ret = new Error(getStringFromWasm0(arg0, arg1));
|
|
9522
|
+
return ret;
|
|
9523
|
+
}
|
|
9422
9524
|
export function __wbg_new_617a8cdb8bb1130e() {
|
|
9423
9525
|
const ret = new Object();
|
|
9424
9526
|
return ret;
|
|
@@ -9499,6 +9601,10 @@ export function __wbg_setTimeout_b4a9096784635514(arg0, arg1) {
|
|
|
9499
9601
|
const ret = setTimeout(arg0, arg1);
|
|
9500
9602
|
return ret;
|
|
9501
9603
|
}
|
|
9604
|
+
export function __wbg_set_145a351398b48c65() { return handleError(function (arg0, arg1, arg2) {
|
|
9605
|
+
const ret = Reflect.set(arg0, arg1, arg2);
|
|
9606
|
+
return ret;
|
|
9607
|
+
}, arguments); }
|
|
9502
9608
|
export function __wbg_set_6ae97e73113c4f0b(arg0, arg1, arg2) {
|
|
9503
9609
|
const ret = arg0.set(arg1, arg2);
|
|
9504
9610
|
return ret;
|
|
@@ -9736,12 +9842,12 @@ export function __wbg_wasmsitemapurl_unwrap(arg0) {
|
|
|
9736
9842
|
return ret;
|
|
9737
9843
|
}
|
|
9738
9844
|
export function __wbindgen_generic_0000000000000001(arg0, arg1) {
|
|
9739
|
-
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx:
|
|
9845
|
+
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1509, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
|
|
9740
9846
|
const ret = makeMutClosure(arg0, arg1, wasm_bindgen_a37d5a31ee71bc5f___convert__closures_____invoke___wasm_bindgen_a37d5a31ee71bc5f___JsValue__core_37ee0ce0c4cd57d5___result__Result_____wasm_bindgen_a37d5a31ee71bc5f___JsError___true_);
|
|
9741
9847
|
return ret;
|
|
9742
9848
|
}
|
|
9743
9849
|
export function __wbindgen_generic_0000000000000002(arg0, arg1) {
|
|
9744
|
-
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx:
|
|
9850
|
+
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1351, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
|
|
9745
9851
|
const ret = makeMutClosure(arg0, arg1, wasm_bindgen_a37d5a31ee71bc5f___convert__closures_____invoke_______true_);
|
|
9746
9852
|
return ret;
|
|
9747
9853
|
}
|
|
Binary file
|
|
@@ -138,12 +138,13 @@ export const wasmbrowserconfig_chromePath: (a: number) => [number, number];
|
|
|
138
138
|
export const wasmbrowserconfig_clearProxy: (a: number) => void;
|
|
139
139
|
export const wasmbrowserconfig_copyForTransfer: (a: number) => number;
|
|
140
140
|
export const wasmbrowserconfig_default: () => number;
|
|
141
|
+
export const wasmbrowserconfig_domQuietCap: (a: number) => [number, bigint];
|
|
141
142
|
export const wasmbrowserconfig_domQuietWindow: (a: number) => [number, bigint];
|
|
142
143
|
export const wasmbrowserconfig_endpoint: (a: number) => [number, number];
|
|
143
144
|
export const wasmbrowserconfig_evalScript: (a: number) => [number, number];
|
|
144
145
|
export const wasmbrowserconfig_extraWait: (a: number) => [number, bigint];
|
|
145
146
|
export const wasmbrowserconfig_mode: (a: number) => [number, number];
|
|
146
|
-
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: number, t: number, u: bigint, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: bigint) => number;
|
|
147
|
+
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: number, t: number, u: bigint, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: bigint, f1: number, g1: bigint) => number;
|
|
147
148
|
export const wasmbrowserconfig_overallTimeout: (a: number) => [number, bigint];
|
|
148
149
|
export const wasmbrowserconfig_proxy: (a: number) => number;
|
|
149
150
|
export const wasmbrowserconfig_robotsUserAgent: (a: number) => [number, number];
|
|
@@ -154,6 +155,7 @@ export const wasmbrowserconfig_set_blockUrlPatterns: (a: number, b: number, c: n
|
|
|
154
155
|
export const wasmbrowserconfig_set_captureNetworkEvents: (a: number, b: number) => void;
|
|
155
156
|
export const wasmbrowserconfig_set_chromeArgs: (a: number, b: number, c: number) => void;
|
|
156
157
|
export const wasmbrowserconfig_set_chromePath: (a: number, b: number, c: number) => void;
|
|
158
|
+
export const wasmbrowserconfig_set_domQuietCap: (a: number, b: number, c: bigint) => void;
|
|
157
159
|
export const wasmbrowserconfig_set_domQuietWindow: (a: number, b: number, c: bigint) => void;
|
|
158
160
|
export const wasmbrowserconfig_set_endpoint: (a: number, b: number, c: number) => void;
|
|
159
161
|
export const wasmbrowserconfig_set_evalScript: (a: number, b: number, c: number) => void;
|
|
@@ -201,14 +203,18 @@ export const wasmcontentconfig_bullets: (a: number) => [number, number];
|
|
|
201
203
|
export const wasmcontentconfig_compactTables: (a: number) => number;
|
|
202
204
|
export const wasmcontentconfig_copyForTransfer: (a: number) => number;
|
|
203
205
|
export const wasmcontentconfig_default: () => number;
|
|
206
|
+
export const wasmcontentconfig_escapeAsterisks: (a: number) => number;
|
|
207
|
+
export const wasmcontentconfig_escapeMisc: (a: number) => number;
|
|
204
208
|
export const wasmcontentconfig_excludeSelectors: (a: number) => [number, number];
|
|
209
|
+
export const wasmcontentconfig_expandAbbreviations: (a: number) => number;
|
|
205
210
|
export const wasmcontentconfig_extractMetadata: (a: number) => number;
|
|
206
211
|
export const wasmcontentconfig_includeDocumentStructure: (a: number) => number;
|
|
212
|
+
export const wasmcontentconfig_includeIconLabels: (a: number) => number;
|
|
207
213
|
export const wasmcontentconfig_keepHidden: (a: number) => number;
|
|
208
214
|
export const wasmcontentconfig_keepInlineImagesIn: (a: number) => [number, number];
|
|
209
215
|
export const wasmcontentconfig_listIndentWidth: (a: number) => number;
|
|
210
216
|
export const wasmcontentconfig_maxDepth: (a: number) => number;
|
|
211
|
-
export const wasmcontentconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number) => number;
|
|
217
|
+
export const wasmcontentconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number) => number;
|
|
212
218
|
export const wasmcontentconfig_outputFormat: (a: number) => [number, number];
|
|
213
219
|
export const wasmcontentconfig_preprocessingPreset: (a: number) => [number, number];
|
|
214
220
|
export const wasmcontentconfig_preserveTags: (a: number) => [number, number];
|
|
@@ -216,9 +222,13 @@ export const wasmcontentconfig_removeForms: (a: number) => number;
|
|
|
216
222
|
export const wasmcontentconfig_removeNavigation: (a: number) => number;
|
|
217
223
|
export const wasmcontentconfig_set_bullets: (a: number, b: number, c: number) => void;
|
|
218
224
|
export const wasmcontentconfig_set_compactTables: (a: number, b: number) => void;
|
|
225
|
+
export const wasmcontentconfig_set_escapeAsterisks: (a: number, b: number) => void;
|
|
226
|
+
export const wasmcontentconfig_set_escapeMisc: (a: number, b: number) => void;
|
|
219
227
|
export const wasmcontentconfig_set_excludeSelectors: (a: number, b: number, c: number) => void;
|
|
228
|
+
export const wasmcontentconfig_set_expandAbbreviations: (a: number, b: number) => void;
|
|
220
229
|
export const wasmcontentconfig_set_extractMetadata: (a: number, b: number) => void;
|
|
221
230
|
export const wasmcontentconfig_set_includeDocumentStructure: (a: number, b: number) => void;
|
|
231
|
+
export const wasmcontentconfig_set_includeIconLabels: (a: number, b: number) => void;
|
|
222
232
|
export const wasmcontentconfig_set_keepHidden: (a: number, b: number) => void;
|
|
223
233
|
export const wasmcontentconfig_set_keepInlineImagesIn: (a: number, b: number, c: number) => void;
|
|
224
234
|
export const wasmcontentconfig_set_listIndentWidth: (a: number, b: number) => void;
|
|
@@ -288,7 +298,7 @@ export const wasmcrawlconfig_maxDepth: (a: number) => number;
|
|
|
288
298
|
export const wasmcrawlconfig_maxLinksPerPage: (a: number) => number;
|
|
289
299
|
export const wasmcrawlconfig_maxPages: (a: number) => number;
|
|
290
300
|
export const wasmcrawlconfig_maxRedirects: (a: number) => number;
|
|
291
|
-
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: number, r: bigint, s: number, t: number, u: number, v: number, w: number, x: bigint, y: number, z: bigint, 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: number, f2: bigint, 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, w2: number, x2: number) => number;
|
|
301
|
+
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: number, r: bigint, s: number, t: number, u: number, v: number, w: number, x: bigint, y: number, z: bigint, 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: number, f2: bigint, 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, w2: number, x2: number, y2: number, z2: number) => number;
|
|
292
302
|
export const wasmcrawlconfig_pathPatternsMatchQuery: (a: number) => number;
|
|
293
303
|
export const wasmcrawlconfig_pathPatternsMatchUrl: (a: number) => number;
|
|
294
304
|
export const wasmcrawlconfig_proxy: (a: number) => number;
|
|
@@ -354,6 +364,7 @@ export const wasmcrawlconfig_set_ssrf: (a: number, b: number) => void;
|
|
|
354
364
|
export const wasmcrawlconfig_set_ssrfDenyPrivateExplicit: (a: number, b: number) => void;
|
|
355
365
|
export const wasmcrawlconfig_set_stayOnDomain: (a: number, b: number) => void;
|
|
356
366
|
export const wasmcrawlconfig_set_stripTrackingParams: (a: number, b: number) => void;
|
|
367
|
+
export const wasmcrawlconfig_set_subdomainHeaderAllowlist: (a: number, b: number, c: number) => void;
|
|
357
368
|
export const wasmcrawlconfig_set_trackingParams: (a: number, b: number, c: number) => void;
|
|
358
369
|
export const wasmcrawlconfig_set_userAgent: (a: number, b: number, c: number) => void;
|
|
359
370
|
export const wasmcrawlconfig_set_userAgents: (a: number, b: number, c: number) => void;
|
|
@@ -363,6 +374,7 @@ export const wasmcrawlconfig_ssrf: (a: number) => number;
|
|
|
363
374
|
export const wasmcrawlconfig_ssrfDenyPrivateExplicit: (a: number) => number;
|
|
364
375
|
export const wasmcrawlconfig_stayOnDomain: (a: number) => number;
|
|
365
376
|
export const wasmcrawlconfig_stripTrackingParams: (a: number) => number;
|
|
377
|
+
export const wasmcrawlconfig_subdomainHeaderAllowlist: (a: number) => [number, number];
|
|
366
378
|
export const wasmcrawlconfig_trackingParams: (a: number) => [number, number];
|
|
367
379
|
export const wasmcrawlconfig_userAgent: (a: number) => [number, number];
|
|
368
380
|
export const wasmcrawlconfig_userAgents: (a: number) => [number, number];
|
package/pkg/bundler/package.json
CHANGED
|
@@ -260,7 +260,7 @@ export class WasmBrowserConfig {
|
|
|
260
260
|
* keep a handle you already hold, build with `default()` and assign the property instead:
|
|
261
261
|
* the generated setter borrows and leaves your handle alive.
|
|
262
262
|
*/
|
|
263
|
-
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, scrollContainers?: boolean | null, domQuietWindow?: bigint | null);
|
|
263
|
+
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, scrollContainers?: boolean | null, domQuietWindow?: bigint | null, domQuietCap?: bigint | null);
|
|
264
264
|
get backend(): string;
|
|
265
265
|
set backend(value: WasmBrowserBackend);
|
|
266
266
|
blockUrlPatterns: string[];
|
|
@@ -268,6 +268,8 @@ export class WasmBrowserConfig {
|
|
|
268
268
|
chromeArgs: string[];
|
|
269
269
|
get chromePath(): string | undefined;
|
|
270
270
|
set chromePath(value: string | null | undefined);
|
|
271
|
+
get domQuietCap(): bigint | undefined;
|
|
272
|
+
set domQuietCap(value: bigint | null | undefined);
|
|
271
273
|
get domQuietWindow(): bigint | undefined;
|
|
272
274
|
set domQuietWindow(value: bigint | null | undefined);
|
|
273
275
|
get endpoint(): string | undefined;
|
|
@@ -430,12 +432,16 @@ export class WasmContentConfig {
|
|
|
430
432
|
*/
|
|
431
433
|
copyForTransfer(): WasmContentConfig;
|
|
432
434
|
static default(): WasmContentConfig;
|
|
433
|
-
constructor(outputFormat?: string | null, preprocessingPreset?: string | null, removeNavigation?: boolean | null, removeForms?: boolean | null, stripTags?: string[] | null, preserveTags?: string[] | null, excludeSelectors?: string[] | null, skipImages?: boolean | null, wrap?: boolean | null, wrapWidth?: number | null, includeDocumentStructure?: boolean | null, extractMetadata?: boolean | null, maxDepth?: number | null, stripNewlines?: boolean | null, bullets?: string | null, listIndentWidth?: number | null, strongEmSymbol?: string | null, compactTables?: boolean | null, keepInlineImagesIn?: string[] | null, keepHidden?: boolean | null);
|
|
435
|
+
constructor(outputFormat?: string | null, preprocessingPreset?: string | null, removeNavigation?: boolean | null, removeForms?: boolean | null, stripTags?: string[] | null, preserveTags?: string[] | null, excludeSelectors?: string[] | null, skipImages?: boolean | null, wrap?: boolean | null, wrapWidth?: number | null, includeDocumentStructure?: boolean | null, extractMetadata?: boolean | null, maxDepth?: number | null, stripNewlines?: boolean | null, bullets?: string | null, listIndentWidth?: number | null, strongEmSymbol?: string | null, compactTables?: boolean | null, keepInlineImagesIn?: string[] | null, keepHidden?: boolean | null, expandAbbreviations?: boolean | null, escapeMisc?: boolean | null, escapeAsterisks?: boolean | null, includeIconLabels?: boolean | null);
|
|
434
436
|
bullets: string;
|
|
435
437
|
compactTables: boolean;
|
|
438
|
+
escapeAsterisks: boolean;
|
|
439
|
+
escapeMisc: boolean;
|
|
436
440
|
excludeSelectors: string[];
|
|
441
|
+
expandAbbreviations: boolean;
|
|
437
442
|
extractMetadata: boolean;
|
|
438
443
|
includeDocumentStructure: boolean;
|
|
444
|
+
includeIconLabels: boolean;
|
|
439
445
|
keepHidden: boolean;
|
|
440
446
|
keepInlineImagesIn: string[];
|
|
441
447
|
listIndentWidth: number;
|
|
@@ -509,7 +515,7 @@ export class WasmCrawlConfig {
|
|
|
509
515
|
* keep a handle you already hold, build with `default()` and assign the property instead:
|
|
510
516
|
* the generated setter borrows and leaves your handle alive.
|
|
511
517
|
*/
|
|
512
|
-
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, dangerAcceptInvalidCerts?: boolean | null);
|
|
518
|
+
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, dangerAcceptInvalidCerts?: boolean | null, subdomainHeaderAllowlist?: string[] | null);
|
|
513
519
|
/**
|
|
514
520
|
* Validate the configuration, returning an error if any values are invalid.
|
|
515
521
|
*
|
|
@@ -615,6 +621,7 @@ export class WasmCrawlConfig {
|
|
|
615
621
|
ssrf: WasmSsrfPolicy;
|
|
616
622
|
stayOnDomain: boolean;
|
|
617
623
|
stripTrackingParams: boolean;
|
|
624
|
+
subdomainHeaderAllowlist: string[];
|
|
618
625
|
trackingParams: string[];
|
|
619
626
|
get userAgent(): string | undefined;
|
|
620
627
|
set userAgent(value: string | null | undefined);
|