@xberg-io/crawlberg-wasm 1.2.1 → 1.3.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1784,6 +1784,15 @@ class WasmContentConfig {
1784
1784
  if (Symbol.dispose) WasmContentConfig.prototype[Symbol.dispose] = WasmContentConfig.prototype.free;
1785
1785
  exports.WasmContentConfig = WasmContentConfig;
1786
1786
 
1787
+ /**
1788
+ * Content filter applied to each crawled page before it reaches the result.
1789
+ * @enum {0}
1790
+ */
1791
+ const WasmContentFilterKind = Object.freeze({
1792
+ Bm25: 0, "0": "Bm25",
1793
+ });
1794
+ exports.WasmContentFilterKind = WasmContentFilterKind;
1795
+
1787
1796
  /**
1788
1797
  * Information about an HTTP cookie received from a response.
1789
1798
  */
@@ -1970,6 +1979,25 @@ class WasmCrawlConfig {
1970
1979
  const ret = wasm.wasmcrawlconfig_auth(this.__wbg_ptr);
1971
1980
  return ret;
1972
1981
  }
1982
+ /**
1983
+ * @returns {string | undefined}
1984
+ */
1985
+ get bm25Query() {
1986
+ const ret = wasm.wasmcrawlconfig_bm25Query(this.__wbg_ptr);
1987
+ let v1;
1988
+ if (ret[0] !== 0) {
1989
+ v1 = getStringFromWasm0(ret[0], ret[1]);
1990
+ wasm.__wbindgen_free(ret[0], ret[1] * 1, 1);
1991
+ }
1992
+ return v1;
1993
+ }
1994
+ /**
1995
+ * @returns {number | undefined}
1996
+ */
1997
+ get bm25Threshold() {
1998
+ const ret = wasm.wasmcrawlconfig_bm25Threshold(this.__wbg_ptr);
1999
+ return ret[0] === 0 ? undefined : ret[1];
2000
+ }
1973
2001
  /**
1974
2002
  * @returns {WasmBrowserConfig}
1975
2003
  */
@@ -2003,6 +2031,18 @@ class WasmCrawlConfig {
2003
2031
  const ret = wasm.wasmcrawlconfig_content(this.__wbg_ptr);
2004
2032
  return WasmContentConfig.__wrap(ret);
2005
2033
  }
2034
+ /**
2035
+ * @returns {string | undefined}
2036
+ */
2037
+ get contentFilter() {
2038
+ const ret = wasm.wasmcrawlconfig_contentFilter(this.__wbg_ptr);
2039
+ let v1;
2040
+ if (ret[0] !== 0) {
2041
+ v1 = getStringFromWasm0(ret[0], ret[1]);
2042
+ wasm.__wbindgen_free(ret[0], ret[1] * 1, 1);
2043
+ }
2044
+ return v1;
2045
+ }
2006
2046
  /**
2007
2047
  * @returns {boolean}
2008
2048
  */
@@ -2010,6 +2050,21 @@ class WasmCrawlConfig {
2010
2050
  const ret = wasm.wasmcrawlconfig_cookiesEnabled(this.__wbg_ptr);
2011
2051
  return ret !== 0;
2012
2052
  }
2053
+ /**
2054
+ * @returns {string}
2055
+ */
2056
+ get crawlStrategy() {
2057
+ let deferred1_0;
2058
+ let deferred1_1;
2059
+ try {
2060
+ const ret = wasm.wasmcrawlconfig_crawlStrategy(this.__wbg_ptr);
2061
+ deferred1_0 = ret[0];
2062
+ deferred1_1 = ret[1];
2063
+ return getStringFromWasm0(ret[0], ret[1]);
2064
+ } finally {
2065
+ wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
2066
+ }
2067
+ }
2013
2068
  /**
2014
2069
  * @returns {any}
2015
2070
  */
@@ -2179,6 +2234,7 @@ class WasmCrawlConfig {
2179
2234
  return ret >>> 0;
2180
2235
  }
2181
2236
  /**
2237
+ * @param {WasmCrawlStrategyKind | null} [crawlStrategy]
2182
2238
  * @param {boolean | null} [respectRobotsTxt]
2183
2239
  * @param {boolean | null} [softHttpErrors]
2184
2240
  * @param {boolean | null} [stayOnDomain]
@@ -2207,6 +2263,9 @@ class WasmCrawlConfig {
2207
2263
  * @param {number | null} [maxPages]
2208
2264
  * @param {number | null} [maxLinksPerPage]
2209
2265
  * @param {number | null} [maxConcurrent]
2266
+ * @param {WasmContentFilterKind | null} [contentFilter]
2267
+ * @param {string | null} [bm25Query]
2268
+ * @param {number | null} [bm25Threshold]
2210
2269
  * @param {string | null} [userAgent]
2211
2270
  * @param {bigint | null} [rateLimitMs]
2212
2271
  * @param {any | null} [auth]
@@ -2223,7 +2282,7 @@ class WasmCrawlConfig {
2223
2282
  * @param {string | null} [browserProfile]
2224
2283
  * @param {boolean | null} [ssrfDenyPrivateExplicit]
2225
2284
  */
2226
- constructor(respectRobotsTxt, softHttpErrors, stayOnDomain, allowSubdomains, includePaths, excludePaths, customHeaders, requestTimeout, maxRedirects, retryCount, retryCodes, cookiesEnabled, removeTags, content, downloadAssets, assetTypes, browser, userAgents, captureScreenshot, followDocumentUrls, downloadDocuments, documentMimeTypes, saveBrowserProfile, ssrf, maxDepth, maxPages, maxLinksPerPage, maxConcurrent, userAgent, rateLimitMs, auth, maxBodySize, mapLimit, mapSearch, maxAssetSize, proxy, documentUrlDepth, documentMaxSize, documentOutputDir, documentContentEncoding, warcOutput, browserProfile, ssrfDenyPrivateExplicit) {
2285
+ constructor(crawlStrategy, respectRobotsTxt, softHttpErrors, stayOnDomain, allowSubdomains, includePaths, excludePaths, customHeaders, requestTimeout, maxRedirects, retryCount, retryCodes, 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) {
2227
2286
  var ptr0 = isLikeNone(includePaths) ? 0 : passArrayJsValueToWasm0(includePaths, wasm.__wbindgen_malloc);
2228
2287
  var len0 = WASM_VECTOR_LEN;
2229
2288
  var ptr1 = isLikeNone(excludePaths) ? 0 : passArrayJsValueToWasm0(excludePaths, wasm.__wbindgen_malloc);
@@ -2253,22 +2312,24 @@ class WasmCrawlConfig {
2253
2312
  _assertClass(ssrf, WasmSsrfPolicy);
2254
2313
  ptr9 = ssrf.__destroy_into_raw();
2255
2314
  }
2256
- var ptr10 = isLikeNone(userAgent) ? 0 : passStringToWasm0(userAgent, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2315
+ var ptr10 = isLikeNone(bm25Query) ? 0 : passStringToWasm0(bm25Query, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2257
2316
  var len10 = WASM_VECTOR_LEN;
2258
- var ptr11 = isLikeNone(mapSearch) ? 0 : passStringToWasm0(mapSearch, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2317
+ var ptr11 = isLikeNone(userAgent) ? 0 : passStringToWasm0(userAgent, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2259
2318
  var len11 = WASM_VECTOR_LEN;
2260
- let ptr12 = 0;
2319
+ var ptr12 = isLikeNone(mapSearch) ? 0 : passStringToWasm0(mapSearch, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2320
+ var len12 = WASM_VECTOR_LEN;
2321
+ let ptr13 = 0;
2261
2322
  if (!isLikeNone(proxy)) {
2262
2323
  _assertClass(proxy, WasmProxyConfig);
2263
- ptr12 = proxy.__destroy_into_raw();
2324
+ ptr13 = proxy.__destroy_into_raw();
2264
2325
  }
2265
- var ptr13 = isLikeNone(documentOutputDir) ? 0 : passStringToWasm0(documentOutputDir, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2266
- var len13 = WASM_VECTOR_LEN;
2267
- var ptr14 = isLikeNone(warcOutput) ? 0 : passStringToWasm0(warcOutput, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2326
+ var ptr14 = isLikeNone(documentOutputDir) ? 0 : passStringToWasm0(documentOutputDir, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2268
2327
  var len14 = WASM_VECTOR_LEN;
2269
- var ptr15 = isLikeNone(browserProfile) ? 0 : passStringToWasm0(browserProfile, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2328
+ var ptr15 = isLikeNone(warcOutput) ? 0 : passStringToWasm0(warcOutput, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2270
2329
  var len15 = WASM_VECTOR_LEN;
2271
- const ret = wasm.wasmcrawlconfig_new(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(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, ptr2, len2, isLikeNone(cookiesEnabled) ? 0xFFFFFF : cookiesEnabled ? 1 : 0, ptr3, len3, ptr4, isLikeNone(downloadAssets) ? 0xFFFFFF : downloadAssets ? 1 : 0, ptr5, len5, ptr6, ptr7, len7, isLikeNone(captureScreenshot) ? 0xFFFFFF : captureScreenshot ? 1 : 0, isLikeNone(followDocumentUrls) ? 0xFFFFFF : followDocumentUrls ? 1 : 0, isLikeNone(downloadDocuments) ? 0xFFFFFF : downloadDocuments ? 1 : 0, ptr8, len8, isLikeNone(saveBrowserProfile) ? 0xFFFFFF : saveBrowserProfile ? 1 : 0, ptr9, 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, ptr10, len10, !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, ptr11, len11, isLikeNone(maxAssetSize) ? Number.MAX_SAFE_INTEGER : (maxAssetSize) >>> 0, ptr12, isLikeNone(documentUrlDepth) ? Number.MAX_SAFE_INTEGER : (documentUrlDepth) >>> 0, isLikeNone(documentMaxSize) ? Number.MAX_SAFE_INTEGER : (documentMaxSize) >>> 0, ptr13, len13, isLikeNone(documentContentEncoding) ? 1 : documentContentEncoding, ptr14, len14, ptr15, len15, isLikeNone(ssrfDenyPrivateExplicit) ? 0xFFFFFF : ssrfDenyPrivateExplicit ? 1 : 0);
2330
+ var ptr16 = isLikeNone(browserProfile) ? 0 : passStringToWasm0(browserProfile, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2331
+ var len16 = WASM_VECTOR_LEN;
2332
+ 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(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, ptr2, len2, isLikeNone(cookiesEnabled) ? 0xFFFFFF : cookiesEnabled ? 1 : 0, ptr3, len3, ptr4, isLikeNone(downloadAssets) ? 0xFFFFFF : downloadAssets ? 1 : 0, ptr5, len5, ptr6, ptr7, len7, isLikeNone(captureScreenshot) ? 0xFFFFFF : captureScreenshot ? 1 : 0, isLikeNone(followDocumentUrls) ? 0xFFFFFF : followDocumentUrls ? 1 : 0, isLikeNone(downloadDocuments) ? 0xFFFFFF : downloadDocuments ? 1 : 0, ptr8, len8, isLikeNone(saveBrowserProfile) ? 0xFFFFFF : saveBrowserProfile ? 1 : 0, ptr9, 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, ptr10, len10, !isLikeNone(bm25Threshold), isLikeNone(bm25Threshold) ? 0 : bm25Threshold, ptr11, len11, !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, ptr12, len12, isLikeNone(maxAssetSize) ? Number.MAX_SAFE_INTEGER : (maxAssetSize) >>> 0, ptr13, isLikeNone(documentUrlDepth) ? Number.MAX_SAFE_INTEGER : (documentUrlDepth) >>> 0, isLikeNone(documentMaxSize) ? Number.MAX_SAFE_INTEGER : (documentMaxSize) >>> 0, ptr14, len14, isLikeNone(documentContentEncoding) ? 1 : documentContentEncoding, ptr15, len15, ptr16, len16, isLikeNone(ssrfDenyPrivateExplicit) ? 0xFFFFFF : ssrfDenyPrivateExplicit ? 1 : 0);
2272
2333
  this.__wbg_ptr = ret;
2273
2334
  WasmCrawlConfigFinalization.register(this, this.__wbg_ptr, this);
2274
2335
  return this;
@@ -2353,6 +2414,20 @@ class WasmCrawlConfig {
2353
2414
  set auth(value) {
2354
2415
  wasm.wasmcrawlconfig_set_auth(this.__wbg_ptr, isLikeNone(value) ? 0 : addToExternrefTable0(value));
2355
2416
  }
2417
+ /**
2418
+ * @param {string | null} [value]
2419
+ */
2420
+ set bm25Query(value) {
2421
+ var ptr0 = isLikeNone(value) ? 0 : passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
2422
+ var len0 = WASM_VECTOR_LEN;
2423
+ wasm.wasmcrawlconfig_set_bm25Query(this.__wbg_ptr, ptr0, len0);
2424
+ }
2425
+ /**
2426
+ * @param {number | null} [value]
2427
+ */
2428
+ set bm25Threshold(value) {
2429
+ wasm.wasmcrawlconfig_set_bm25Threshold(this.__wbg_ptr, !isLikeNone(value), isLikeNone(value) ? 0 : value);
2430
+ }
2356
2431
  /**
2357
2432
  * @param {WasmBrowserConfig} value
2358
2433
  */
@@ -2383,12 +2458,24 @@ class WasmCrawlConfig {
2383
2458
  var ptr0 = value.__destroy_into_raw();
2384
2459
  wasm.wasmcrawlconfig_set_content(this.__wbg_ptr, ptr0);
2385
2460
  }
2461
+ /**
2462
+ * @param {WasmContentFilterKind | null} [value]
2463
+ */
2464
+ set contentFilter(value) {
2465
+ wasm.wasmcrawlconfig_set_contentFilter(this.__wbg_ptr, isLikeNone(value) ? 1 : value);
2466
+ }
2386
2467
  /**
2387
2468
  * @param {boolean} value
2388
2469
  */
2389
2470
  set cookiesEnabled(value) {
2390
2471
  wasm.wasmcrawlconfig_set_cookiesEnabled(this.__wbg_ptr, value);
2391
2472
  }
2473
+ /**
2474
+ * @param {WasmCrawlStrategyKind} value
2475
+ */
2476
+ set crawlStrategy(value) {
2477
+ wasm.wasmcrawlconfig_set_crawlStrategy(this.__wbg_ptr, value);
2478
+ }
2392
2479
  /**
2393
2480
  * @param {any} value
2394
2481
  */
@@ -3383,6 +3470,22 @@ class WasmCrawlResult {
3383
3470
  if (Symbol.dispose) WasmCrawlResult.prototype[Symbol.dispose] = WasmCrawlResult.prototype.free;
3384
3471
  exports.WasmCrawlResult = WasmCrawlResult;
3385
3472
 
3473
+ /**
3474
+ * Traversal order for a crawl.
3475
+ *
3476
+ * Selects both the queue discipline and the selection strategy, because global order is a
3477
+ * property of the frontier: the engine hands its bounded selection window to the strategy, so
3478
+ * a strategy alone can only reorder URLs that have already been dequeued.
3479
+ * @enum {0 | 1 | 2 | 3}
3480
+ */
3481
+ const WasmCrawlStrategyKind = Object.freeze({
3482
+ Bfs: 0, "0": "Bfs",
3483
+ Dfs: 1, "1": "Dfs",
3484
+ BestFirst: 2, "2": "BestFirst",
3485
+ Adaptive: 3, "3": "Adaptive",
3486
+ });
3487
+ exports.WasmCrawlStrategyKind = WasmCrawlStrategyKind;
3488
+
3386
3489
  /**
3387
3490
  * Opt-in encoding applied to a downloaded document's bytes for callers who need the
3388
3491
  * content available in a serializable field rather than reading it from disk.
@@ -8011,7 +8114,7 @@ function __wbg_get_imports() {
8011
8114
  const a = state0.a;
8012
8115
  state0.a = 0;
8013
8116
  try {
8014
- return wasm_bindgen__convert__closures_____invoke__h06ea3d672ad56c9d(a, state0.b, arg0, arg1);
8117
+ return wasm_bindgen__convert__closures_____invoke__h072ef20b96c52fc3(a, state0.b, arg0, arg1);
8015
8118
  } finally {
8016
8119
  state0.a = a;
8017
8120
  }
@@ -8034,10 +8137,18 @@ function __wbg_get_imports() {
8034
8137
  const ret = arg0.next;
8035
8138
  return ret;
8036
8139
  },
8140
+ __wbg_now_e7c6795a7f81e10f: function(arg0) {
8141
+ const ret = arg0.now();
8142
+ return ret;
8143
+ },
8037
8144
  __wbg_parse_1cc93481b0865939: function() { return handleError(function (arg0, arg1) {
8038
8145
  const ret = JSON.parse(getStringFromWasm0(arg0, arg1));
8039
8146
  return ret;
8040
8147
  }, arguments); },
8148
+ __wbg_performance_3fcf6e32a7e1ed0a: function(arg0) {
8149
+ const ret = arg0.performance;
8150
+ return ret;
8151
+ },
8041
8152
  __wbg_prototypesetcall_de8e0d9553586985: function(arg0, arg1, arg2) {
8042
8153
  Uint8Array.prototype.set.call(getArrayU8FromWasm0(arg0, arg1), arg2);
8043
8154
  },
@@ -8291,13 +8402,13 @@ function __wbg_get_imports() {
8291
8402
  return ret;
8292
8403
  },
8293
8404
  __wbindgen_cast_0000000000000001: function(arg0, arg1) {
8294
- // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 992, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
8295
- const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__he29f1bb22c192c97);
8405
+ // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1069, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
8406
+ const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h2deefa3c9140ef43);
8296
8407
  return ret;
8297
8408
  },
8298
8409
  __wbindgen_cast_0000000000000002: function(arg0, arg1) {
8299
- // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 882, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
8300
- const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__hbe9f9227fe7d8a65);
8410
+ // Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 954, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
8411
+ const ret = makeMutClosure(arg0, arg1, wasm_bindgen__convert__closures_____invoke__h99e891e0bdc2d97f);
8301
8412
  return ret;
8302
8413
  },
8303
8414
  __wbindgen_cast_0000000000000003: function(arg0) {
@@ -8336,19 +8447,19 @@ function __wbg_get_imports() {
8336
8447
  };
8337
8448
  }
8338
8449
 
8339
- function wasm_bindgen__convert__closures_____invoke__hbe9f9227fe7d8a65(arg0, arg1) {
8340
- wasm.wasm_bindgen__convert__closures_____invoke__hbe9f9227fe7d8a65(arg0, arg1);
8450
+ function wasm_bindgen__convert__closures_____invoke__h99e891e0bdc2d97f(arg0, arg1) {
8451
+ wasm.wasm_bindgen__convert__closures_____invoke__h99e891e0bdc2d97f(arg0, arg1);
8341
8452
  }
8342
8453
 
8343
- function wasm_bindgen__convert__closures_____invoke__he29f1bb22c192c97(arg0, arg1, arg2) {
8344
- const ret = wasm.wasm_bindgen__convert__closures_____invoke__he29f1bb22c192c97(arg0, arg1, arg2);
8454
+ function wasm_bindgen__convert__closures_____invoke__h2deefa3c9140ef43(arg0, arg1, arg2) {
8455
+ const ret = wasm.wasm_bindgen__convert__closures_____invoke__h2deefa3c9140ef43(arg0, arg1, arg2);
8345
8456
  if (ret[1]) {
8346
8457
  throw takeFromExternrefTable0(ret[0]);
8347
8458
  }
8348
8459
  }
8349
8460
 
8350
- function wasm_bindgen__convert__closures_____invoke__h06ea3d672ad56c9d(arg0, arg1, arg2, arg3) {
8351
- wasm.wasm_bindgen__convert__closures_____invoke__h06ea3d672ad56c9d(arg0, arg1, arg2, arg3);
8461
+ function wasm_bindgen__convert__closures_____invoke__h072ef20b96c52fc3(arg0, arg1, arg2, arg3) {
8462
+ wasm.wasm_bindgen__convert__closures_____invoke__h072ef20b96c52fc3(arg0, arg1, arg2, arg3);
8352
8463
  }
8353
8464
 
8354
8465
 
Binary file
@@ -203,11 +203,15 @@ export const wasmcookieinfo_value: (a: number) => [number, number];
203
203
  export const wasmcrawlconfig_allowSubdomains: (a: number) => number;
204
204
  export const wasmcrawlconfig_assetTypes: (a: number) => [number, number];
205
205
  export const wasmcrawlconfig_auth: (a: number) => any;
206
+ export const wasmcrawlconfig_bm25Query: (a: number) => [number, number];
207
+ export const wasmcrawlconfig_bm25Threshold: (a: number) => [number, number];
206
208
  export const wasmcrawlconfig_browser: (a: number) => number;
207
209
  export const wasmcrawlconfig_browserProfile: (a: number) => [number, number];
208
210
  export const wasmcrawlconfig_captureScreenshot: (a: number) => number;
209
211
  export const wasmcrawlconfig_content: (a: number) => number;
212
+ export const wasmcrawlconfig_contentFilter: (a: number) => [number, number];
210
213
  export const wasmcrawlconfig_cookiesEnabled: (a: number) => number;
214
+ export const wasmcrawlconfig_crawlStrategy: (a: number) => [number, number];
211
215
  export const wasmcrawlconfig_customHeaders: (a: number) => any;
212
216
  export const wasmcrawlconfig_default: () => number;
213
217
  export const wasmcrawlconfig_documentContentEncoding: (a: number) => [number, number];
@@ -229,8 +233,9 @@ export const wasmcrawlconfig_maxDepth: (a: number) => number;
229
233
  export const wasmcrawlconfig_maxLinksPerPage: (a: number) => number;
230
234
  export const wasmcrawlconfig_maxPages: (a: number) => number;
231
235
  export const wasmcrawlconfig_maxRedirects: (a: number) => number;
232
- 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: bigint, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number, n1: bigint, 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) => number;
236
+ 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: bigint, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: bigint, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number, e2: number, f2: number, g2: number, h2: number, i2: number, j2: number, k2: number) => number;
233
237
  export const wasmcrawlconfig_proxy: (a: number) => number;
238
+ export const wasmcrawlconfig_rateLimitMs: (a: number) => [number, bigint];
234
239
  export const wasmcrawlconfig_removeTags: (a: number) => [number, number];
235
240
  export const wasmcrawlconfig_respectRobotsTxt: (a: number) => number;
236
241
  export const wasmcrawlconfig_retryCodes: (a: number) => [number, number];
@@ -239,11 +244,15 @@ export const wasmcrawlconfig_saveBrowserProfile: (a: number) => number;
239
244
  export const wasmcrawlconfig_set_allowSubdomains: (a: number, b: number) => void;
240
245
  export const wasmcrawlconfig_set_assetTypes: (a: number, b: number, c: number) => void;
241
246
  export const wasmcrawlconfig_set_auth: (a: number, b: number) => void;
247
+ export const wasmcrawlconfig_set_bm25Query: (a: number, b: number, c: number) => void;
248
+ export const wasmcrawlconfig_set_bm25Threshold: (a: number, b: number, c: number) => void;
242
249
  export const wasmcrawlconfig_set_browser: (a: number, b: number) => void;
243
250
  export const wasmcrawlconfig_set_browserProfile: (a: number, b: number, c: number) => void;
244
251
  export const wasmcrawlconfig_set_captureScreenshot: (a: number, b: number) => void;
245
252
  export const wasmcrawlconfig_set_content: (a: number, b: number) => void;
253
+ export const wasmcrawlconfig_set_contentFilter: (a: number, b: number) => void;
246
254
  export const wasmcrawlconfig_set_cookiesEnabled: (a: number, b: number) => void;
255
+ export const wasmcrawlconfig_set_crawlStrategy: (a: number, b: number) => void;
247
256
  export const wasmcrawlconfig_set_customHeaders: (a: number, b: any) => void;
248
257
  export const wasmcrawlconfig_set_documentContentEncoding: (a: number, b: number) => void;
249
258
  export const wasmcrawlconfig_set_documentMaxSize: (a: number, b: number) => void;
@@ -265,6 +274,7 @@ export const wasmcrawlconfig_set_maxLinksPerPage: (a: number, b: number) => void
265
274
  export const wasmcrawlconfig_set_maxPages: (a: number, b: number) => void;
266
275
  export const wasmcrawlconfig_set_maxRedirects: (a: number, b: number) => void;
267
276
  export const wasmcrawlconfig_set_proxy: (a: number, b: number) => void;
277
+ export const wasmcrawlconfig_set_rateLimitMs: (a: number, b: number, c: bigint) => void;
268
278
  export const wasmcrawlconfig_set_removeTags: (a: number, b: number, c: number) => void;
269
279
  export const wasmcrawlconfig_set_respectRobotsTxt: (a: number, b: number) => void;
270
280
  export const wasmcrawlconfig_set_retryCodes: (a: number, b: number, c: number) => void;
@@ -381,14 +391,10 @@ export const wasmdownloadeddocument_size: (a: number) => number;
381
391
  export const wasmdownloadeddocument_truncated: (a: number) => number;
382
392
  export const wasmdownloadeddocument_url: (a: number) => [number, number];
383
393
  export const wasmextractionmeta_chunksProcessed: (a: number) => number;
384
- export const wasmextractionmeta_completionTokens: (a: number) => [number, bigint];
385
- export const wasmextractionmeta_cost: (a: number) => [number, number];
386
394
  export const wasmextractionmeta_default: () => number;
387
395
  export const wasmextractionmeta_model: (a: number) => [number, number];
388
396
  export const wasmextractionmeta_new: (a: number, b: number, c: number, d: number, e: bigint, f: number, g: bigint, h: number, i: number) => number;
389
397
  export const wasmextractionmeta_set_chunksProcessed: (a: number, b: number) => void;
390
- export const wasmextractionmeta_set_completionTokens: (a: number, b: number, c: bigint) => void;
391
- export const wasmextractionmeta_set_cost: (a: number, b: number, c: number) => void;
392
398
  export const wasmextractionmeta_set_model: (a: number, b: number, c: number) => void;
393
399
  export const wasmfaviconinfo_mimeType: (a: number) => [number, number];
394
400
  export const wasmfaviconinfo_rel: (a: number) => [number, number];
@@ -682,11 +688,13 @@ export const wasmssrfpolicy_set_allowlist: (a: number, b: any) => void;
682
688
  export const wasmssrfpolicy_set_denyPrivate: (a: number, b: number) => void;
683
689
  export const wasmssrfpolicy_set_maxRedirects: (a: number, b: number) => void;
684
690
  export const wasmhostmatcher_new: () => number;
685
- export const wasmcrawlconfig_set_rateLimitMs: (a: number, b: number, c: bigint) => void;
691
+ export const wasmcrawlconfig_set_requestTimeout: (a: number, b: number, c: bigint) => void;
692
+ export const wasmextractionmeta_set_completionTokens: (a: number, b: number, c: bigint) => void;
686
693
  export const wasmextractionmeta_set_promptTokens: (a: number, b: number, c: bigint) => void;
687
694
  export const wasmpageaction_set_milliseconds: (a: number, b: number, c: bigint) => void;
688
- export const wasmcrawlconfig_rateLimitMs: (a: number) => [number, bigint];
689
695
  export const wasmcrawlconfig_requestTimeout: (a: number) => [number, bigint];
696
+ export const wasmextractionmeta_completionTokens: (a: number) => [number, bigint];
697
+ export const wasmextractionmeta_cost: (a: number) => [number, number];
690
698
  export const wasmextractionmeta_promptTokens: (a: number) => [number, bigint];
691
699
  export const wasmpageaction_amount: (a: number) => [number, bigint];
692
700
  export const wasmpageaction_milliseconds: (a: number) => [number, bigint];
@@ -708,13 +716,13 @@ export const wasmimageinfo_width: (a: number) => number;
708
716
  export const wasmpagemetadata_wordCount: (a: number) => number;
709
717
  export const wasmlinkinfo_nofollow: (a: number) => number;
710
718
  export const wasmauthconfig_new: () => number;
711
- export const wasmcrawlconfig_set_requestTimeout: (a: number, b: number, c: bigint) => void;
719
+ export const wasmextractionmeta_set_cost: (a: number, b: number, c: number) => void;
712
720
  export const wasmpageaction_set_amount: (a: number, b: number, c: bigint) => void;
713
721
  export const wasmscraperesult_set_crawlDelay: (a: number, b: number, c: bigint) => void;
714
722
  export const wasmpageaction_new: () => number;
715
- export const wasm_bindgen__convert__closures_____invoke__he29f1bb22c192c97: (a: number, b: number, c: any) => [number, number];
716
- export const wasm_bindgen__convert__closures_____invoke__h06ea3d672ad56c9d: (a: number, b: number, c: any, d: any) => void;
717
- export const wasm_bindgen__convert__closures_____invoke__hbe9f9227fe7d8a65: (a: number, b: number) => void;
723
+ export const wasm_bindgen__convert__closures_____invoke__h2deefa3c9140ef43: (a: number, b: number, c: any) => [number, number];
724
+ export const wasm_bindgen__convert__closures_____invoke__h072ef20b96c52fc3: (a: number, b: number, c: any, d: any) => void;
725
+ export const wasm_bindgen__convert__closures_____invoke__h99e891e0bdc2d97f: (a: number, b: number) => void;
718
726
  export const __wbindgen_malloc: (a: number, b: number) => number;
719
727
  export const __wbindgen_realloc: (a: number, b: number, c: number, d: number) => number;
720
728
  export const __wbindgen_exn_store: (a: number) => void;
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "crawlberg-wasm",
3
3
  "description": "High-performance web crawling engine",
4
- "version": "1.2.1",
4
+ "version": "1.3.2",
5
5
  "license": "MIT",
6
6
  "repository": {
7
7
  "type": "git",
@@ -267,6 +267,13 @@ export class WasmContentConfig {
267
267
  wrapWidth: number;
268
268
  }
269
269
 
270
+ /**
271
+ * Content filter applied to each crawled page before it reaches the result.
272
+ */
273
+ export enum WasmContentFilterKind {
274
+ Bm25 = 0,
275
+ }
276
+
270
277
  /**
271
278
  * Information about an HTTP cookie received from a response.
272
279
  */
@@ -290,7 +297,7 @@ export class WasmCrawlConfig {
290
297
  free(): void;
291
298
  [Symbol.dispose](): void;
292
299
  static default(): WasmCrawlConfig;
293
- constructor(respectRobotsTxt?: boolean | null, softHttpErrors?: boolean | null, stayOnDomain?: boolean | null, allowSubdomains?: boolean | null, includePaths?: string[] | null, excludePaths?: string[] | null, customHeaders?: any | null, requestTimeout?: bigint | null, maxRedirects?: number | null, retryCount?: number | null, retryCodes?: Uint16Array | null, cookiesEnabled?: boolean | null, removeTags?: string[] | null, content?: WasmContentConfig | null, downloadAssets?: boolean | null, assetTypes?: any[] | null, browser?: WasmBrowserConfig | null, userAgents?: string[] | null, captureScreenshot?: boolean | null, followDocumentUrls?: boolean | null, downloadDocuments?: boolean | null, documentMimeTypes?: string[] | null, saveBrowserProfile?: boolean | null, ssrf?: WasmSsrfPolicy | null, maxDepth?: number | null, maxPages?: number | null, maxLinksPerPage?: number | null, maxConcurrent?: number | null, userAgent?: string | null, rateLimitMs?: bigint | null, auth?: any | null, maxBodySize?: number | null, mapLimit?: number | null, mapSearch?: string | null, maxAssetSize?: number | null, proxy?: WasmProxyConfig | null, documentUrlDepth?: number | null, documentMaxSize?: number | null, documentOutputDir?: string | null, documentContentEncoding?: WasmDocumentContentEncoding | null, warcOutput?: string | null, browserProfile?: string | null, ssrfDenyPrivateExplicit?: boolean | null);
300
+ constructor(crawlStrategy?: WasmCrawlStrategyKind | null, respectRobotsTxt?: boolean | null, softHttpErrors?: boolean | null, stayOnDomain?: boolean | null, allowSubdomains?: boolean | null, includePaths?: string[] | null, excludePaths?: string[] | null, customHeaders?: any | null, requestTimeout?: bigint | null, maxRedirects?: number | null, retryCount?: number | null, retryCodes?: Uint16Array | null, cookiesEnabled?: boolean | null, removeTags?: string[] | null, content?: WasmContentConfig | null, downloadAssets?: boolean | null, assetTypes?: any[] | null, browser?: WasmBrowserConfig | null, userAgents?: string[] | null, captureScreenshot?: boolean | null, followDocumentUrls?: boolean | null, downloadDocuments?: boolean | null, documentMimeTypes?: string[] | null, saveBrowserProfile?: boolean | null, ssrf?: WasmSsrfPolicy | null, maxDepth?: number | null, maxPages?: number | null, maxLinksPerPage?: number | null, maxConcurrent?: number | null, contentFilter?: WasmContentFilterKind | null, bm25Query?: string | null, bm25Threshold?: number | null, userAgent?: string | null, rateLimitMs?: bigint | null, auth?: any | null, maxBodySize?: number | null, mapLimit?: number | null, mapSearch?: string | null, maxAssetSize?: number | null, proxy?: WasmProxyConfig | null, documentUrlDepth?: number | null, documentMaxSize?: number | null, documentOutputDir?: string | null, documentContentEncoding?: WasmDocumentContentEncoding | null, warcOutput?: string | null, browserProfile?: string | null, ssrfDenyPrivateExplicit?: boolean | null);
294
301
  /**
295
302
  * Validate the configuration, returning an error if any values are invalid.
296
303
  */
@@ -299,12 +306,20 @@ export class WasmCrawlConfig {
299
306
  assetTypes: string[];
300
307
  get auth(): any | undefined;
301
308
  set auth(value: any | null | undefined);
309
+ get bm25Query(): string | undefined;
310
+ set bm25Query(value: string | null | undefined);
311
+ get bm25Threshold(): number | undefined;
312
+ set bm25Threshold(value: number | null | undefined);
302
313
  browser: WasmBrowserConfig;
303
314
  get browserProfile(): string | undefined;
304
315
  set browserProfile(value: string | null | undefined);
305
316
  captureScreenshot: boolean;
306
317
  content: WasmContentConfig;
318
+ get contentFilter(): string | undefined;
319
+ set contentFilter(value: WasmContentFilterKind | null | undefined);
307
320
  cookiesEnabled: boolean;
321
+ get crawlStrategy(): string;
322
+ set crawlStrategy(value: WasmCrawlStrategyKind);
308
323
  customHeaders: any;
309
324
  get documentContentEncoding(): string | undefined;
310
325
  set documentContentEncoding(value: WasmDocumentContentEncoding | null | undefined);
@@ -438,6 +453,20 @@ export class WasmCrawlResult {
438
453
  wasSkipped: boolean;
439
454
  }
440
455
 
456
+ /**
457
+ * Traversal order for a crawl.
458
+ *
459
+ * Selects both the queue discipline and the selection strategy, because global order is a
460
+ * property of the frontier: the engine hands its bounded selection window to the strategy, so
461
+ * a strategy alone can only reorder URLs that have already been dequeued.
462
+ */
463
+ export enum WasmCrawlStrategyKind {
464
+ Bfs = 0,
465
+ Dfs = 1,
466
+ BestFirst = 2,
467
+ Adaptive = 3,
468
+ }
469
+
441
470
  /**
442
471
  * Opt-in encoding applied to a downloaded document's bytes for callers who need the
443
472
  * content available in a serializable field rather than reading it from disk.
@@ -1245,11 +1274,15 @@ export interface InitOutput {
1245
1274
  readonly wasmcrawlconfig_allowSubdomains: (a: number) => number;
1246
1275
  readonly wasmcrawlconfig_assetTypes: (a: number) => [number, number];
1247
1276
  readonly wasmcrawlconfig_auth: (a: number) => any;
1277
+ readonly wasmcrawlconfig_bm25Query: (a: number) => [number, number];
1278
+ readonly wasmcrawlconfig_bm25Threshold: (a: number) => [number, number];
1248
1279
  readonly wasmcrawlconfig_browser: (a: number) => number;
1249
1280
  readonly wasmcrawlconfig_browserProfile: (a: number) => [number, number];
1250
1281
  readonly wasmcrawlconfig_captureScreenshot: (a: number) => number;
1251
1282
  readonly wasmcrawlconfig_content: (a: number) => number;
1283
+ readonly wasmcrawlconfig_contentFilter: (a: number) => [number, number];
1252
1284
  readonly wasmcrawlconfig_cookiesEnabled: (a: number) => number;
1285
+ readonly wasmcrawlconfig_crawlStrategy: (a: number) => [number, number];
1253
1286
  readonly wasmcrawlconfig_customHeaders: (a: number) => any;
1254
1287
  readonly wasmcrawlconfig_default: () => number;
1255
1288
  readonly wasmcrawlconfig_documentContentEncoding: (a: number) => [number, number];
@@ -1271,8 +1304,9 @@ export interface InitOutput {
1271
1304
  readonly wasmcrawlconfig_maxLinksPerPage: (a: number) => number;
1272
1305
  readonly wasmcrawlconfig_maxPages: (a: number) => number;
1273
1306
  readonly wasmcrawlconfig_maxRedirects: (a: number) => number;
1274
- readonly wasmcrawlconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: bigint, l: number, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number, n1: bigint, 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) => number;
1307
+ readonly wasmcrawlconfig_new: (a: number, b: number, c: number, d: number, e: number, f: number, g: number, h: number, i: number, j: number, k: number, l: bigint, m: number, n: number, o: number, p: number, q: number, r: number, s: number, t: number, u: number, v: number, w: number, x: number, y: number, z: number, a1: number, b1: number, c1: number, d1: number, e1: number, f1: number, g1: number, h1: number, i1: number, j1: number, k1: number, l1: number, m1: number, n1: number, o1: number, p1: number, q1: number, r1: number, s1: number, t1: bigint, u1: number, v1: number, w1: number, x1: number, y1: number, z1: number, a2: number, b2: number, c2: number, d2: number, e2: number, f2: number, g2: number, h2: number, i2: number, j2: number, k2: number) => number;
1275
1308
  readonly wasmcrawlconfig_proxy: (a: number) => number;
1309
+ readonly wasmcrawlconfig_rateLimitMs: (a: number) => [number, bigint];
1276
1310
  readonly wasmcrawlconfig_removeTags: (a: number) => [number, number];
1277
1311
  readonly wasmcrawlconfig_respectRobotsTxt: (a: number) => number;
1278
1312
  readonly wasmcrawlconfig_retryCodes: (a: number) => [number, number];
@@ -1281,11 +1315,15 @@ export interface InitOutput {
1281
1315
  readonly wasmcrawlconfig_set_allowSubdomains: (a: number, b: number) => void;
1282
1316
  readonly wasmcrawlconfig_set_assetTypes: (a: number, b: number, c: number) => void;
1283
1317
  readonly wasmcrawlconfig_set_auth: (a: number, b: number) => void;
1318
+ readonly wasmcrawlconfig_set_bm25Query: (a: number, b: number, c: number) => void;
1319
+ readonly wasmcrawlconfig_set_bm25Threshold: (a: number, b: number, c: number) => void;
1284
1320
  readonly wasmcrawlconfig_set_browser: (a: number, b: number) => void;
1285
1321
  readonly wasmcrawlconfig_set_browserProfile: (a: number, b: number, c: number) => void;
1286
1322
  readonly wasmcrawlconfig_set_captureScreenshot: (a: number, b: number) => void;
1287
1323
  readonly wasmcrawlconfig_set_content: (a: number, b: number) => void;
1324
+ readonly wasmcrawlconfig_set_contentFilter: (a: number, b: number) => void;
1288
1325
  readonly wasmcrawlconfig_set_cookiesEnabled: (a: number, b: number) => void;
1326
+ readonly wasmcrawlconfig_set_crawlStrategy: (a: number, b: number) => void;
1289
1327
  readonly wasmcrawlconfig_set_customHeaders: (a: number, b: any) => void;
1290
1328
  readonly wasmcrawlconfig_set_documentContentEncoding: (a: number, b: number) => void;
1291
1329
  readonly wasmcrawlconfig_set_documentMaxSize: (a: number, b: number) => void;
@@ -1307,6 +1345,7 @@ export interface InitOutput {
1307
1345
  readonly wasmcrawlconfig_set_maxPages: (a: number, b: number) => void;
1308
1346
  readonly wasmcrawlconfig_set_maxRedirects: (a: number, b: number) => void;
1309
1347
  readonly wasmcrawlconfig_set_proxy: (a: number, b: number) => void;
1348
+ readonly wasmcrawlconfig_set_rateLimitMs: (a: number, b: number, c: bigint) => void;
1310
1349
  readonly wasmcrawlconfig_set_removeTags: (a: number, b: number, c: number) => void;
1311
1350
  readonly wasmcrawlconfig_set_respectRobotsTxt: (a: number, b: number) => void;
1312
1351
  readonly wasmcrawlconfig_set_retryCodes: (a: number, b: number, c: number) => void;
@@ -1423,14 +1462,10 @@ export interface InitOutput {
1423
1462
  readonly wasmdownloadeddocument_truncated: (a: number) => number;
1424
1463
  readonly wasmdownloadeddocument_url: (a: number) => [number, number];
1425
1464
  readonly wasmextractionmeta_chunksProcessed: (a: number) => number;
1426
- readonly wasmextractionmeta_completionTokens: (a: number) => [number, bigint];
1427
- readonly wasmextractionmeta_cost: (a: number) => [number, number];
1428
1465
  readonly wasmextractionmeta_default: () => number;
1429
1466
  readonly wasmextractionmeta_model: (a: number) => [number, number];
1430
1467
  readonly wasmextractionmeta_new: (a: number, b: number, c: number, d: number, e: bigint, f: number, g: bigint, h: number, i: number) => number;
1431
1468
  readonly wasmextractionmeta_set_chunksProcessed: (a: number, b: number) => void;
1432
- readonly wasmextractionmeta_set_completionTokens: (a: number, b: number, c: bigint) => void;
1433
- readonly wasmextractionmeta_set_cost: (a: number, b: number, c: number) => void;
1434
1469
  readonly wasmextractionmeta_set_model: (a: number, b: number, c: number) => void;
1435
1470
  readonly wasmfaviconinfo_mimeType: (a: number) => [number, number];
1436
1471
  readonly wasmfaviconinfo_rel: (a: number) => [number, number];
@@ -1724,11 +1759,13 @@ export interface InitOutput {
1724
1759
  readonly wasmssrfpolicy_set_denyPrivate: (a: number, b: number) => void;
1725
1760
  readonly wasmssrfpolicy_set_maxRedirects: (a: number, b: number) => void;
1726
1761
  readonly wasmhostmatcher_new: () => number;
1727
- readonly wasmcrawlconfig_set_rateLimitMs: (a: number, b: number, c: bigint) => void;
1762
+ readonly wasmcrawlconfig_set_requestTimeout: (a: number, b: number, c: bigint) => void;
1763
+ readonly wasmextractionmeta_set_completionTokens: (a: number, b: number, c: bigint) => void;
1728
1764
  readonly wasmextractionmeta_set_promptTokens: (a: number, b: number, c: bigint) => void;
1729
1765
  readonly wasmpageaction_set_milliseconds: (a: number, b: number, c: bigint) => void;
1730
- readonly wasmcrawlconfig_rateLimitMs: (a: number) => [number, bigint];
1731
1766
  readonly wasmcrawlconfig_requestTimeout: (a: number) => [number, bigint];
1767
+ readonly wasmextractionmeta_completionTokens: (a: number) => [number, bigint];
1768
+ readonly wasmextractionmeta_cost: (a: number) => [number, number];
1732
1769
  readonly wasmextractionmeta_promptTokens: (a: number) => [number, bigint];
1733
1770
  readonly wasmpageaction_amount: (a: number) => [number, bigint];
1734
1771
  readonly wasmpageaction_milliseconds: (a: number) => [number, bigint];
@@ -1750,13 +1787,13 @@ export interface InitOutput {
1750
1787
  readonly wasmpagemetadata_wordCount: (a: number) => number;
1751
1788
  readonly wasmlinkinfo_nofollow: (a: number) => number;
1752
1789
  readonly wasmauthconfig_new: () => number;
1753
- readonly wasmcrawlconfig_set_requestTimeout: (a: number, b: number, c: bigint) => void;
1790
+ readonly wasmextractionmeta_set_cost: (a: number, b: number, c: number) => void;
1754
1791
  readonly wasmpageaction_set_amount: (a: number, b: number, c: bigint) => void;
1755
1792
  readonly wasmscraperesult_set_crawlDelay: (a: number, b: number, c: bigint) => void;
1756
1793
  readonly wasmpageaction_new: () => number;
1757
- readonly wasm_bindgen__convert__closures_____invoke__he29f1bb22c192c97: (a: number, b: number, c: any) => [number, number];
1758
- readonly wasm_bindgen__convert__closures_____invoke__h06ea3d672ad56c9d: (a: number, b: number, c: any, d: any) => void;
1759
- readonly wasm_bindgen__convert__closures_____invoke__hbe9f9227fe7d8a65: (a: number, b: number) => void;
1794
+ readonly wasm_bindgen__convert__closures_____invoke__h2deefa3c9140ef43: (a: number, b: number, c: any) => [number, number];
1795
+ readonly wasm_bindgen__convert__closures_____invoke__h072ef20b96c52fc3: (a: number, b: number, c: any, d: any) => void;
1796
+ readonly wasm_bindgen__convert__closures_____invoke__h99e891e0bdc2d97f: (a: number, b: number) => void;
1760
1797
  readonly __wbindgen_malloc: (a: number, b: number) => number;
1761
1798
  readonly __wbindgen_realloc: (a: number, b: number, c: number, d: number) => number;
1762
1799
  readonly __wbindgen_exn_store: (a: number) => void;