@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
|
@@ -1114,6 +1114,13 @@ class WasmBrowserConfig {
|
|
|
1114
1114
|
const ret = wasm.wasmbrowserconfig_default();
|
|
1115
1115
|
return WasmBrowserConfig.__wrap(ret);
|
|
1116
1116
|
}
|
|
1117
|
+
/**
|
|
1118
|
+
* @returns {bigint | undefined}
|
|
1119
|
+
*/
|
|
1120
|
+
get domQuietCap() {
|
|
1121
|
+
const ret = wasm.wasmbrowserconfig_domQuietCap(this.__wbg_ptr);
|
|
1122
|
+
return ret[0] === 0 ? undefined : BigInt.asUintN(64, ret[1]);
|
|
1123
|
+
}
|
|
1117
1124
|
/**
|
|
1118
1125
|
* @returns {bigint | undefined}
|
|
1119
1126
|
*/
|
|
@@ -1192,8 +1199,9 @@ class WasmBrowserConfig {
|
|
|
1192
1199
|
* @param {string | null} [chromePath]
|
|
1193
1200
|
* @param {boolean | null} [scrollContainers]
|
|
1194
1201
|
* @param {bigint | null} [domQuietWindow]
|
|
1202
|
+
* @param {bigint | null} [domQuietCap]
|
|
1195
1203
|
*/
|
|
1196
|
-
constructor(mode, backend, timeout, overallTimeout, shutdownTimeout, wait, blockUrlPatterns, captureNetworkEvents, sessionAffinity, chromeArgs, endpoint, waitSelector, extraWait, proxy, evalScript, robotsUserAgent, chromePath, scrollContainers, domQuietWindow) {
|
|
1204
|
+
constructor(mode, backend, timeout, overallTimeout, shutdownTimeout, wait, blockUrlPatterns, captureNetworkEvents, sessionAffinity, chromeArgs, endpoint, waitSelector, extraWait, proxy, evalScript, robotsUserAgent, chromePath, scrollContainers, domQuietWindow, domQuietCap) {
|
|
1197
1205
|
var ptr0 = isLikeNone(blockUrlPatterns) ? 0 : passArrayJsValueToWasm0(blockUrlPatterns, wasm.__wbindgen_malloc);
|
|
1198
1206
|
var len0 = WASM_VECTOR_LEN;
|
|
1199
1207
|
var ptr1 = isLikeNone(chromeArgs) ? 0 : passArrayJsValueToWasm0(chromeArgs, wasm.__wbindgen_malloc);
|
|
@@ -1213,7 +1221,7 @@ class WasmBrowserConfig {
|
|
|
1213
1221
|
var len6 = WASM_VECTOR_LEN;
|
|
1214
1222
|
var ptr7 = isLikeNone(chromePath) ? 0 : passStringToWasm0(chromePath, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1215
1223
|
var len7 = WASM_VECTOR_LEN;
|
|
1216
|
-
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);
|
|
1224
|
+
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);
|
|
1217
1225
|
this.__wbg_ptr = ret;
|
|
1218
1226
|
WasmBrowserConfigFinalization.register(this, this.__wbg_ptr, this);
|
|
1219
1227
|
return this;
|
|
@@ -1298,6 +1306,12 @@ class WasmBrowserConfig {
|
|
|
1298
1306
|
var len0 = WASM_VECTOR_LEN;
|
|
1299
1307
|
wasm.wasmbrowserconfig_set_chromePath(this.__wbg_ptr, ptr0, len0);
|
|
1300
1308
|
}
|
|
1309
|
+
/**
|
|
1310
|
+
* @param {bigint | null} [value]
|
|
1311
|
+
*/
|
|
1312
|
+
set domQuietCap(value) {
|
|
1313
|
+
wasm.wasmbrowserconfig_set_domQuietCap(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? BigInt(0) : value);
|
|
1314
|
+
}
|
|
1301
1315
|
/**
|
|
1302
1316
|
* @param {bigint | null} [value]
|
|
1303
1317
|
*/
|
|
@@ -1875,6 +1889,20 @@ class WasmContentConfig {
|
|
|
1875
1889
|
const ret = wasm.wasmcontentconfig_default();
|
|
1876
1890
|
return WasmContentConfig.__wrap(ret);
|
|
1877
1891
|
}
|
|
1892
|
+
/**
|
|
1893
|
+
* @returns {boolean}
|
|
1894
|
+
*/
|
|
1895
|
+
get escapeAsterisks() {
|
|
1896
|
+
const ret = wasm.wasmcontentconfig_escapeAsterisks(this.__wbg_ptr);
|
|
1897
|
+
return ret !== 0;
|
|
1898
|
+
}
|
|
1899
|
+
/**
|
|
1900
|
+
* @returns {boolean}
|
|
1901
|
+
*/
|
|
1902
|
+
get escapeMisc() {
|
|
1903
|
+
const ret = wasm.wasmcontentconfig_escapeMisc(this.__wbg_ptr);
|
|
1904
|
+
return ret !== 0;
|
|
1905
|
+
}
|
|
1878
1906
|
/**
|
|
1879
1907
|
* @returns {string[]}
|
|
1880
1908
|
*/
|
|
@@ -1884,6 +1912,13 @@ class WasmContentConfig {
|
|
|
1884
1912
|
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
1885
1913
|
return v1;
|
|
1886
1914
|
}
|
|
1915
|
+
/**
|
|
1916
|
+
* @returns {boolean}
|
|
1917
|
+
*/
|
|
1918
|
+
get expandAbbreviations() {
|
|
1919
|
+
const ret = wasm.wasmcontentconfig_expandAbbreviations(this.__wbg_ptr);
|
|
1920
|
+
return ret !== 0;
|
|
1921
|
+
}
|
|
1887
1922
|
/**
|
|
1888
1923
|
* @returns {boolean}
|
|
1889
1924
|
*/
|
|
@@ -1898,6 +1933,13 @@ class WasmContentConfig {
|
|
|
1898
1933
|
const ret = wasm.wasmcontentconfig_includeDocumentStructure(this.__wbg_ptr);
|
|
1899
1934
|
return ret !== 0;
|
|
1900
1935
|
}
|
|
1936
|
+
/**
|
|
1937
|
+
* @returns {boolean}
|
|
1938
|
+
*/
|
|
1939
|
+
get includeIconLabels() {
|
|
1940
|
+
const ret = wasm.wasmcontentconfig_includeIconLabels(this.__wbg_ptr);
|
|
1941
|
+
return ret !== 0;
|
|
1942
|
+
}
|
|
1901
1943
|
/**
|
|
1902
1944
|
* @returns {boolean}
|
|
1903
1945
|
*/
|
|
@@ -1949,8 +1991,12 @@ class WasmContentConfig {
|
|
|
1949
1991
|
* @param {boolean | null} [compactTables]
|
|
1950
1992
|
* @param {string[] | null} [keepInlineImagesIn]
|
|
1951
1993
|
* @param {boolean | null} [keepHidden]
|
|
1994
|
+
* @param {boolean | null} [expandAbbreviations]
|
|
1995
|
+
* @param {boolean | null} [escapeMisc]
|
|
1996
|
+
* @param {boolean | null} [escapeAsterisks]
|
|
1997
|
+
* @param {boolean | null} [includeIconLabels]
|
|
1952
1998
|
*/
|
|
1953
|
-
constructor(outputFormat, preprocessingPreset, removeNavigation, removeForms, stripTags, preserveTags, excludeSelectors, skipImages, wrap, wrapWidth, includeDocumentStructure, extractMetadata, maxDepth, stripNewlines, bullets, listIndentWidth, strongEmSymbol, compactTables, keepInlineImagesIn, keepHidden) {
|
|
1999
|
+
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) {
|
|
1954
2000
|
var ptr0 = isLikeNone(outputFormat) ? 0 : passStringToWasm0(outputFormat, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1955
2001
|
var len0 = WASM_VECTOR_LEN;
|
|
1956
2002
|
var ptr1 = isLikeNone(preprocessingPreset) ? 0 : passStringToWasm0(preprocessingPreset, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
@@ -1967,7 +2013,7 @@ class WasmContentConfig {
|
|
|
1967
2013
|
var len6 = WASM_VECTOR_LEN;
|
|
1968
2014
|
var ptr7 = isLikeNone(keepInlineImagesIn) ? 0 : passArrayJsValueToWasm0(keepInlineImagesIn, wasm.__wbindgen_malloc);
|
|
1969
2015
|
var len7 = WASM_VECTOR_LEN;
|
|
1970
|
-
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);
|
|
2016
|
+
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);
|
|
1971
2017
|
this.__wbg_ptr = ret;
|
|
1972
2018
|
WasmContentConfigFinalization.register(this, this.__wbg_ptr, this);
|
|
1973
2019
|
return this;
|
|
@@ -2039,6 +2085,18 @@ class WasmContentConfig {
|
|
|
2039
2085
|
set compactTables(value) {
|
|
2040
2086
|
wasm.wasmcontentconfig_set_compactTables(this.__wbg_ptr, value);
|
|
2041
2087
|
}
|
|
2088
|
+
/**
|
|
2089
|
+
* @param {boolean} value
|
|
2090
|
+
*/
|
|
2091
|
+
set escapeAsterisks(value) {
|
|
2092
|
+
wasm.wasmcontentconfig_set_escapeAsterisks(this.__wbg_ptr, value);
|
|
2093
|
+
}
|
|
2094
|
+
/**
|
|
2095
|
+
* @param {boolean} value
|
|
2096
|
+
*/
|
|
2097
|
+
set escapeMisc(value) {
|
|
2098
|
+
wasm.wasmcontentconfig_set_escapeMisc(this.__wbg_ptr, value);
|
|
2099
|
+
}
|
|
2042
2100
|
/**
|
|
2043
2101
|
* @param {string[]} value
|
|
2044
2102
|
*/
|
|
@@ -2047,6 +2105,12 @@ class WasmContentConfig {
|
|
|
2047
2105
|
const len0 = WASM_VECTOR_LEN;
|
|
2048
2106
|
wasm.wasmcontentconfig_set_excludeSelectors(this.__wbg_ptr, ptr0, len0);
|
|
2049
2107
|
}
|
|
2108
|
+
/**
|
|
2109
|
+
* @param {boolean} value
|
|
2110
|
+
*/
|
|
2111
|
+
set expandAbbreviations(value) {
|
|
2112
|
+
wasm.wasmcontentconfig_set_expandAbbreviations(this.__wbg_ptr, value);
|
|
2113
|
+
}
|
|
2050
2114
|
/**
|
|
2051
2115
|
* @param {boolean} value
|
|
2052
2116
|
*/
|
|
@@ -2059,6 +2123,12 @@ class WasmContentConfig {
|
|
|
2059
2123
|
set includeDocumentStructure(value) {
|
|
2060
2124
|
wasm.wasmcontentconfig_set_includeDocumentStructure(this.__wbg_ptr, value);
|
|
2061
2125
|
}
|
|
2126
|
+
/**
|
|
2127
|
+
* @param {boolean} value
|
|
2128
|
+
*/
|
|
2129
|
+
set includeIconLabels(value) {
|
|
2130
|
+
wasm.wasmcontentconfig_set_includeIconLabels(this.__wbg_ptr, value);
|
|
2131
|
+
}
|
|
2062
2132
|
/**
|
|
2063
2133
|
* @param {boolean} value
|
|
2064
2134
|
*/
|
|
@@ -2778,8 +2848,9 @@ class WasmCrawlConfig {
|
|
|
2778
2848
|
* @param {string | null} [browserProfile]
|
|
2779
2849
|
* @param {boolean | null} [ssrfDenyPrivateExplicit]
|
|
2780
2850
|
* @param {boolean | null} [dangerAcceptInvalidCerts]
|
|
2851
|
+
* @param {string[] | null} [subdomainHeaderAllowlist]
|
|
2781
2852
|
*/
|
|
2782
|
-
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) {
|
|
2853
|
+
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) {
|
|
2783
2854
|
var ptr0 = isLikeNone(includePaths) ? 0 : passArrayJsValueToWasm0(includePaths, wasm.__wbindgen_malloc);
|
|
2784
2855
|
var len0 = WASM_VECTOR_LEN;
|
|
2785
2856
|
var ptr1 = isLikeNone(excludePaths) ? 0 : passArrayJsValueToWasm0(excludePaths, wasm.__wbindgen_malloc);
|
|
@@ -2828,7 +2899,9 @@ class WasmCrawlConfig {
|
|
|
2828
2899
|
var len16 = WASM_VECTOR_LEN;
|
|
2829
2900
|
var ptr17 = isLikeNone(browserProfile) ? 0 : passStringToWasm0(browserProfile, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
2830
2901
|
var len17 = WASM_VECTOR_LEN;
|
|
2831
|
-
|
|
2902
|
+
var ptr18 = isLikeNone(subdomainHeaderAllowlist) ? 0 : passArrayJsValueToWasm0(subdomainHeaderAllowlist, wasm.__wbindgen_malloc);
|
|
2903
|
+
var len18 = WASM_VECTOR_LEN;
|
|
2904
|
+
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);
|
|
2832
2905
|
this.__wbg_ptr = ret;
|
|
2833
2906
|
WasmCrawlConfigFinalization.register(this, this.__wbg_ptr, this);
|
|
2834
2907
|
return this;
|
|
@@ -3268,6 +3341,14 @@ class WasmCrawlConfig {
|
|
|
3268
3341
|
set stripTrackingParams(value) {
|
|
3269
3342
|
wasm.wasmcrawlconfig_set_stripTrackingParams(this.__wbg_ptr, value);
|
|
3270
3343
|
}
|
|
3344
|
+
/**
|
|
3345
|
+
* @param {string[]} value
|
|
3346
|
+
*/
|
|
3347
|
+
set subdomainHeaderAllowlist(value) {
|
|
3348
|
+
const ptr0 = passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc);
|
|
3349
|
+
const len0 = WASM_VECTOR_LEN;
|
|
3350
|
+
wasm.wasmcrawlconfig_set_subdomainHeaderAllowlist(this.__wbg_ptr, ptr0, len0);
|
|
3351
|
+
}
|
|
3271
3352
|
/**
|
|
3272
3353
|
* @param {string[]} value
|
|
3273
3354
|
*/
|
|
@@ -3339,6 +3420,15 @@ class WasmCrawlConfig {
|
|
|
3339
3420
|
const ret = wasm.wasmcrawlconfig_stripTrackingParams(this.__wbg_ptr);
|
|
3340
3421
|
return ret !== 0;
|
|
3341
3422
|
}
|
|
3423
|
+
/**
|
|
3424
|
+
* @returns {string[]}
|
|
3425
|
+
*/
|
|
3426
|
+
get subdomainHeaderAllowlist() {
|
|
3427
|
+
const ret = wasm.wasmcrawlconfig_subdomainHeaderAllowlist(this.__wbg_ptr);
|
|
3428
|
+
var v1 = getArrayJsValueFromWasm0(ret[0], ret[1]);
|
|
3429
|
+
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
3430
|
+
return v1;
|
|
3431
|
+
}
|
|
3342
3432
|
/**
|
|
3343
3433
|
* @returns {string[]}
|
|
3344
3434
|
*/
|
|
@@ -9359,6 +9449,10 @@ function __wbg_get_imports() {
|
|
|
9359
9449
|
const ret = clearTimeout(arg0);
|
|
9360
9450
|
return ret;
|
|
9361
9451
|
},
|
|
9452
|
+
__wbg_create_7febfb959f8c7046: function(arg0) {
|
|
9453
|
+
const ret = Object.create(arg0);
|
|
9454
|
+
return ret;
|
|
9455
|
+
},
|
|
9362
9456
|
__wbg_done_b41a1d26cdb37fb6: function(arg0) {
|
|
9363
9457
|
const ret = arg0.done;
|
|
9364
9458
|
return ret;
|
|
@@ -9379,6 +9473,10 @@ function __wbg_get_imports() {
|
|
|
9379
9473
|
const ret = arg0.fetch(arg1);
|
|
9380
9474
|
return ret;
|
|
9381
9475
|
},
|
|
9476
|
+
__wbg_for_beb5dc4e908dd8b8: function(arg0, arg1) {
|
|
9477
|
+
const ret = Symbol.for(getStringFromWasm0(arg0, arg1));
|
|
9478
|
+
return ret;
|
|
9479
|
+
},
|
|
9382
9480
|
__wbg_getRandomValues_a608c4436c19407a: function() { return handleError(function (arg0, arg1) {
|
|
9383
9481
|
globalThis.crypto.getRandomValues(getArrayU8FromWasm0(arg0, arg1));
|
|
9384
9482
|
}, arguments); },
|
|
@@ -9482,6 +9580,10 @@ function __wbg_get_imports() {
|
|
|
9482
9580
|
const ret = new AbortController();
|
|
9483
9581
|
return ret;
|
|
9484
9582
|
}, arguments); },
|
|
9583
|
+
__wbg_new_343a093a3c2ffb4e: function(arg0, arg1) {
|
|
9584
|
+
const ret = new Error(getStringFromWasm0(arg0, arg1));
|
|
9585
|
+
return ret;
|
|
9586
|
+
},
|
|
9485
9587
|
__wbg_new_617a8cdb8bb1130e: function() {
|
|
9486
9588
|
const ret = new Object();
|
|
9487
9589
|
return ret;
|
|
@@ -9562,6 +9664,10 @@ function __wbg_get_imports() {
|
|
|
9562
9664
|
const ret = setTimeout(arg0, arg1);
|
|
9563
9665
|
return ret;
|
|
9564
9666
|
},
|
|
9667
|
+
__wbg_set_145a351398b48c65: function() { return handleError(function (arg0, arg1, arg2) {
|
|
9668
|
+
const ret = Reflect.set(arg0, arg1, arg2);
|
|
9669
|
+
return ret;
|
|
9670
|
+
}, arguments); },
|
|
9565
9671
|
__wbg_set_6ae97e73113c4f0b: function(arg0, arg1, arg2) {
|
|
9566
9672
|
const ret = arg0.set(arg1, arg2);
|
|
9567
9673
|
return ret;
|
|
@@ -9799,12 +9905,12 @@ function __wbg_get_imports() {
|
|
|
9799
9905
|
return ret;
|
|
9800
9906
|
},
|
|
9801
9907
|
__wbindgen_generic_0000000000000001: function(arg0, arg1) {
|
|
9802
|
-
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx:
|
|
9908
|
+
// 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`.
|
|
9803
9909
|
const ret = makeMutClosure(arg0, arg1, wasm_bindgen_a37d5a31ee71bc5f___convert__closures_____invoke___wasm_bindgen_a37d5a31ee71bc5f___JsValue__core_37ee0ce0c4cd57d5___result__Result_____wasm_bindgen_a37d5a31ee71bc5f___JsError___true_);
|
|
9804
9910
|
return ret;
|
|
9805
9911
|
},
|
|
9806
9912
|
__wbindgen_generic_0000000000000002: function(arg0, arg1) {
|
|
9807
|
-
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx:
|
|
9913
|
+
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1351, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
|
|
9808
9914
|
const ret = makeMutClosure(arg0, arg1, wasm_bindgen_a37d5a31ee71bc5f___convert__closures_____invoke_______true_);
|
|
9809
9915
|
return ret;
|
|
9810
9916
|
},
|
|
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/nodejs/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);
|
|
@@ -1889,12 +1896,13 @@ export interface InitOutput {
|
|
|
1889
1896
|
readonly wasmbrowserconfig_clearProxy: (a: number) => void;
|
|
1890
1897
|
readonly wasmbrowserconfig_copyForTransfer: (a: number) => number;
|
|
1891
1898
|
readonly wasmbrowserconfig_default: () => number;
|
|
1899
|
+
readonly wasmbrowserconfig_domQuietCap: (a: number) => [number, bigint];
|
|
1892
1900
|
readonly wasmbrowserconfig_domQuietWindow: (a: number) => [number, bigint];
|
|
1893
1901
|
readonly wasmbrowserconfig_endpoint: (a: number) => [number, number];
|
|
1894
1902
|
readonly wasmbrowserconfig_evalScript: (a: number) => [number, number];
|
|
1895
1903
|
readonly wasmbrowserconfig_extraWait: (a: number) => [number, bigint];
|
|
1896
1904
|
readonly wasmbrowserconfig_mode: (a: number) => [number, number];
|
|
1897
|
-
readonly 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;
|
|
1905
|
+
readonly 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;
|
|
1898
1906
|
readonly wasmbrowserconfig_overallTimeout: (a: number) => [number, bigint];
|
|
1899
1907
|
readonly wasmbrowserconfig_proxy: (a: number) => number;
|
|
1900
1908
|
readonly wasmbrowserconfig_robotsUserAgent: (a: number) => [number, number];
|
|
@@ -1905,6 +1913,7 @@ export interface InitOutput {
|
|
|
1905
1913
|
readonly wasmbrowserconfig_set_captureNetworkEvents: (a: number, b: number) => void;
|
|
1906
1914
|
readonly wasmbrowserconfig_set_chromeArgs: (a: number, b: number, c: number) => void;
|
|
1907
1915
|
readonly wasmbrowserconfig_set_chromePath: (a: number, b: number, c: number) => void;
|
|
1916
|
+
readonly wasmbrowserconfig_set_domQuietCap: (a: number, b: number, c: bigint) => void;
|
|
1908
1917
|
readonly wasmbrowserconfig_set_domQuietWindow: (a: number, b: number, c: bigint) => void;
|
|
1909
1918
|
readonly wasmbrowserconfig_set_endpoint: (a: number, b: number, c: number) => void;
|
|
1910
1919
|
readonly wasmbrowserconfig_set_evalScript: (a: number, b: number, c: number) => void;
|
|
@@ -1952,14 +1961,18 @@ export interface InitOutput {
|
|
|
1952
1961
|
readonly wasmcontentconfig_compactTables: (a: number) => number;
|
|
1953
1962
|
readonly wasmcontentconfig_copyForTransfer: (a: number) => number;
|
|
1954
1963
|
readonly wasmcontentconfig_default: () => number;
|
|
1964
|
+
readonly wasmcontentconfig_escapeAsterisks: (a: number) => number;
|
|
1965
|
+
readonly wasmcontentconfig_escapeMisc: (a: number) => number;
|
|
1955
1966
|
readonly wasmcontentconfig_excludeSelectors: (a: number) => [number, number];
|
|
1967
|
+
readonly wasmcontentconfig_expandAbbreviations: (a: number) => number;
|
|
1956
1968
|
readonly wasmcontentconfig_extractMetadata: (a: number) => number;
|
|
1957
1969
|
readonly wasmcontentconfig_includeDocumentStructure: (a: number) => number;
|
|
1970
|
+
readonly wasmcontentconfig_includeIconLabels: (a: number) => number;
|
|
1958
1971
|
readonly wasmcontentconfig_keepHidden: (a: number) => number;
|
|
1959
1972
|
readonly wasmcontentconfig_keepInlineImagesIn: (a: number) => [number, number];
|
|
1960
1973
|
readonly wasmcontentconfig_listIndentWidth: (a: number) => number;
|
|
1961
1974
|
readonly wasmcontentconfig_maxDepth: (a: number) => number;
|
|
1962
|
-
readonly 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;
|
|
1975
|
+
readonly 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;
|
|
1963
1976
|
readonly wasmcontentconfig_outputFormat: (a: number) => [number, number];
|
|
1964
1977
|
readonly wasmcontentconfig_preprocessingPreset: (a: number) => [number, number];
|
|
1965
1978
|
readonly wasmcontentconfig_preserveTags: (a: number) => [number, number];
|
|
@@ -1967,9 +1980,13 @@ export interface InitOutput {
|
|
|
1967
1980
|
readonly wasmcontentconfig_removeNavigation: (a: number) => number;
|
|
1968
1981
|
readonly wasmcontentconfig_set_bullets: (a: number, b: number, c: number) => void;
|
|
1969
1982
|
readonly wasmcontentconfig_set_compactTables: (a: number, b: number) => void;
|
|
1983
|
+
readonly wasmcontentconfig_set_escapeAsterisks: (a: number, b: number) => void;
|
|
1984
|
+
readonly wasmcontentconfig_set_escapeMisc: (a: number, b: number) => void;
|
|
1970
1985
|
readonly wasmcontentconfig_set_excludeSelectors: (a: number, b: number, c: number) => void;
|
|
1986
|
+
readonly wasmcontentconfig_set_expandAbbreviations: (a: number, b: number) => void;
|
|
1971
1987
|
readonly wasmcontentconfig_set_extractMetadata: (a: number, b: number) => void;
|
|
1972
1988
|
readonly wasmcontentconfig_set_includeDocumentStructure: (a: number, b: number) => void;
|
|
1989
|
+
readonly wasmcontentconfig_set_includeIconLabels: (a: number, b: number) => void;
|
|
1973
1990
|
readonly wasmcontentconfig_set_keepHidden: (a: number, b: number) => void;
|
|
1974
1991
|
readonly wasmcontentconfig_set_keepInlineImagesIn: (a: number, b: number, c: number) => void;
|
|
1975
1992
|
readonly wasmcontentconfig_set_listIndentWidth: (a: number, b: number) => void;
|
|
@@ -2039,7 +2056,7 @@ export interface InitOutput {
|
|
|
2039
2056
|
readonly wasmcrawlconfig_maxLinksPerPage: (a: number) => number;
|
|
2040
2057
|
readonly wasmcrawlconfig_maxPages: (a: number) => number;
|
|
2041
2058
|
readonly wasmcrawlconfig_maxRedirects: (a: number) => number;
|
|
2042
|
-
readonly wasmcrawlconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: 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;
|
|
2059
|
+
readonly wasmcrawlconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: 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;
|
|
2043
2060
|
readonly wasmcrawlconfig_pathPatternsMatchQuery: (a: number) => number;
|
|
2044
2061
|
readonly wasmcrawlconfig_pathPatternsMatchUrl: (a: number) => number;
|
|
2045
2062
|
readonly wasmcrawlconfig_proxy: (a: number) => number;
|
|
@@ -2105,6 +2122,7 @@ export interface InitOutput {
|
|
|
2105
2122
|
readonly wasmcrawlconfig_set_ssrfDenyPrivateExplicit: (a: number, b: number) => void;
|
|
2106
2123
|
readonly wasmcrawlconfig_set_stayOnDomain: (a: number, b: number) => void;
|
|
2107
2124
|
readonly wasmcrawlconfig_set_stripTrackingParams: (a: number, b: number) => void;
|
|
2125
|
+
readonly wasmcrawlconfig_set_subdomainHeaderAllowlist: (a: number, b: number, c: number) => void;
|
|
2108
2126
|
readonly wasmcrawlconfig_set_trackingParams: (a: number, b: number, c: number) => void;
|
|
2109
2127
|
readonly wasmcrawlconfig_set_userAgent: (a: number, b: number, c: number) => void;
|
|
2110
2128
|
readonly wasmcrawlconfig_set_userAgents: (a: number, b: number, c: number) => void;
|
|
@@ -2114,6 +2132,7 @@ export interface InitOutput {
|
|
|
2114
2132
|
readonly wasmcrawlconfig_ssrfDenyPrivateExplicit: (a: number) => number;
|
|
2115
2133
|
readonly wasmcrawlconfig_stayOnDomain: (a: number) => number;
|
|
2116
2134
|
readonly wasmcrawlconfig_stripTrackingParams: (a: number) => number;
|
|
2135
|
+
readonly wasmcrawlconfig_subdomainHeaderAllowlist: (a: number) => [number, number];
|
|
2117
2136
|
readonly wasmcrawlconfig_trackingParams: (a: number) => [number, number];
|
|
2118
2137
|
readonly wasmcrawlconfig_userAgent: (a: number) => [number, number];
|
|
2119
2138
|
readonly wasmcrawlconfig_userAgents: (a: number) => [number, number];
|