@xberg-io/crawlberg-wasm 1.10.3 → 1.11.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +5 -5
- package/package.json +23 -8
- package/pkg/bundler/README.md +5 -5
- package/pkg/bundler/crawlberg_wasm.d.ts +16 -4
- package/pkg/bundler/crawlberg_wasm_bg.js +202 -19
- package/pkg/bundler/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/bundler/crawlberg_wasm_bg.wasm.d.ts +24 -6
- package/pkg/bundler/package.json +5 -1
- package/pkg/deno/README.md +5 -5
- package/pkg/deno/crawlberg_wasm.d.ts +16 -4
- package/pkg/deno/crawlberg_wasm.js +202 -19
- package/pkg/deno/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/deno/crawlberg_wasm_bg.wasm.d.ts +24 -6
- package/pkg/nodejs/README.md +5 -5
- package/pkg/nodejs/crawlberg_wasm.d.ts +16 -4
- package/pkg/nodejs/crawlberg_wasm.js +202 -19
- package/pkg/nodejs/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/nodejs/crawlberg_wasm_bg.wasm.d.ts +24 -6
- package/pkg/nodejs/package.json +5 -1
- package/pkg/web/README.md +5 -5
- package/pkg/web/crawlberg_wasm.d.ts +40 -10
- package/pkg/web/crawlberg_wasm.js +202 -19
- package/pkg/web/crawlberg_wasm_bg.wasm +0 -0
- package/pkg/web/crawlberg_wasm_bg.wasm.d.ts +24 -6
- package/pkg/web/package.json +5 -1
package/README.md
CHANGED
|
@@ -1,5 +1,5 @@
|
|
|
1
1
|
<!-- This file is auto-generated by alef — DO NOT EDIT. -->
|
|
2
|
-
<!-- alef:hash:
|
|
2
|
+
<!-- alef:hash:317aa594b0d20c83a18e8cdb5342b90b90f56e94e51ba85fc80ebb44ab9c1331 -->
|
|
3
3
|
<!-- To regenerate: alef readme -->
|
|
4
4
|
<!-- To verify freshness: alef verify -->
|
|
5
5
|
|
|
@@ -101,14 +101,14 @@ pnpm add @xberg-io/crawlberg-wasm
|
|
|
101
101
|
|
|
102
102
|
## Agent plugin
|
|
103
103
|
|
|
104
|
-
The `crawlberg` plugin is available via the `xberg-io/
|
|
104
|
+
The `crawlberg` plugin is available via the `xberg-io/crawlberg` marketplace.
|
|
105
105
|
|
|
106
106
|
```text
|
|
107
|
-
/plugin marketplace add xberg-io/
|
|
108
|
-
/plugin install crawlberg@
|
|
107
|
+
/plugin marketplace add xberg-io/crawlberg
|
|
108
|
+
/plugin install crawlberg@crawlberg
|
|
109
109
|
```
|
|
110
110
|
|
|
111
|
-
|
|
111
|
+
See [the plugin guide](https://github.com/xberg-io/crawlberg/blob/main/plugin/README.md) for all supported runtimes, CLI prerequisites and standalone public skills.
|
|
112
112
|
|
|
113
113
|
## Quick Start
|
|
114
114
|
|
package/package.json
CHANGED
|
@@ -1,6 +1,16 @@
|
|
|
1
1
|
{
|
|
2
|
+
"author": {
|
|
3
|
+
"name": "Na'aman Hirschfeld",
|
|
4
|
+
"email": "naaman@xberg.io"
|
|
5
|
+
},
|
|
6
|
+
"contributors": [
|
|
7
|
+
{
|
|
8
|
+
"name": "Tobias Perlstein",
|
|
9
|
+
"email": "tobias@xberg.io"
|
|
10
|
+
}
|
|
11
|
+
],
|
|
2
12
|
"name": "@xberg-io/crawlberg-wasm",
|
|
3
|
-
"version": "1.
|
|
13
|
+
"version": "1.11.0",
|
|
4
14
|
"private": false,
|
|
5
15
|
"description": "High-performance web crawling engine",
|
|
6
16
|
"license": "MIT",
|
|
@@ -13,7 +23,12 @@
|
|
|
13
23
|
"access": "public"
|
|
14
24
|
},
|
|
15
25
|
"type": "module",
|
|
16
|
-
"files": [
|
|
26
|
+
"files": [
|
|
27
|
+
"pkg",
|
|
28
|
+
"*.wasm",
|
|
29
|
+
"*.d.ts",
|
|
30
|
+
"README.md"
|
|
31
|
+
],
|
|
17
32
|
"main": "pkg/nodejs/crawlberg_wasm.js",
|
|
18
33
|
"module": "pkg/web/crawlberg_wasm.js",
|
|
19
34
|
"types": "pkg/nodejs/crawlberg_wasm.d.ts",
|
|
@@ -29,6 +44,10 @@
|
|
|
29
44
|
"engines": {
|
|
30
45
|
"node": ">= 22"
|
|
31
46
|
},
|
|
47
|
+
"devDependencies": {
|
|
48
|
+
"@vitest/coverage-v8": "^5.0.4",
|
|
49
|
+
"vitest": "^5.0.4"
|
|
50
|
+
},
|
|
32
51
|
"scripts": {
|
|
33
52
|
"build": "wasm-pack build --target nodejs --out-dir pkg/nodejs",
|
|
34
53
|
"build:ci": "wasm-pack build --release --target nodejs --out-dir pkg/nodejs",
|
|
@@ -36,14 +55,10 @@
|
|
|
36
55
|
"build:wasm:bundler": "wasm-pack build --release --target bundler --out-dir pkg/bundler",
|
|
37
56
|
"build:wasm:nodejs": "wasm-pack build --release --target nodejs --out-dir pkg/nodejs",
|
|
38
57
|
"build:wasm:deno": "wasm-pack build --release --target deno --out-dir pkg/deno",
|
|
39
|
-
"build:all": "
|
|
58
|
+
"build:all": "pnpm run build:wasm:web && pnpm run build:wasm:bundler && pnpm run build:wasm:nodejs && pnpm run build:wasm:deno && find pkg -name .gitignore -delete",
|
|
40
59
|
"test": "vitest run",
|
|
41
60
|
"test:watch": "vitest watch",
|
|
42
61
|
"test:coverage": "vitest run --coverage",
|
|
43
62
|
"clean": "rm -rf pkg dist"
|
|
44
|
-
},
|
|
45
|
-
"devDependencies": {
|
|
46
|
-
"vitest": "^5.0.0",
|
|
47
|
-
"@vitest/coverage-v8": "^5.0.0"
|
|
48
63
|
}
|
|
49
|
-
}
|
|
64
|
+
}
|
package/pkg/bundler/README.md
CHANGED
|
@@ -1,5 +1,5 @@
|
|
|
1
1
|
<!-- This file is auto-generated by alef — DO NOT EDIT. -->
|
|
2
|
-
<!-- alef:hash:
|
|
2
|
+
<!-- alef:hash:317aa594b0d20c83a18e8cdb5342b90b90f56e94e51ba85fc80ebb44ab9c1331 -->
|
|
3
3
|
<!-- To regenerate: alef readme -->
|
|
4
4
|
<!-- To verify freshness: alef verify -->
|
|
5
5
|
|
|
@@ -101,14 +101,14 @@ pnpm add @xberg-io/crawlberg-wasm
|
|
|
101
101
|
|
|
102
102
|
## Agent plugin
|
|
103
103
|
|
|
104
|
-
The `crawlberg` plugin is available via the `xberg-io/
|
|
104
|
+
The `crawlberg` plugin is available via the `xberg-io/crawlberg` marketplace.
|
|
105
105
|
|
|
106
106
|
```text
|
|
107
|
-
/plugin marketplace add xberg-io/
|
|
108
|
-
/plugin install crawlberg@
|
|
107
|
+
/plugin marketplace add xberg-io/crawlberg
|
|
108
|
+
/plugin install crawlberg@crawlberg
|
|
109
109
|
```
|
|
110
110
|
|
|
111
|
-
|
|
111
|
+
See [the plugin guide](https://github.com/xberg-io/crawlberg/blob/main/plugin/README.md) for all supported runtimes, CLI prerequisites and standalone public skills.
|
|
112
112
|
|
|
113
113
|
## Quick Start
|
|
114
114
|
|
|
@@ -427,10 +427,15 @@ export class WasmContentConfig {
|
|
|
427
427
|
*/
|
|
428
428
|
copyForTransfer(): WasmContentConfig;
|
|
429
429
|
static default(): WasmContentConfig;
|
|
430
|
-
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);
|
|
430
|
+
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);
|
|
431
|
+
bullets: string;
|
|
432
|
+
compactTables: boolean;
|
|
431
433
|
excludeSelectors: string[];
|
|
432
434
|
extractMetadata: boolean;
|
|
433
435
|
includeDocumentStructure: boolean;
|
|
436
|
+
keepHidden: boolean;
|
|
437
|
+
keepInlineImagesIn: string[];
|
|
438
|
+
listIndentWidth: number;
|
|
434
439
|
get maxDepth(): number | undefined;
|
|
435
440
|
set maxDepth(value: number | null | undefined);
|
|
436
441
|
outputFormat: string;
|
|
@@ -439,7 +444,9 @@ export class WasmContentConfig {
|
|
|
439
444
|
removeForms: boolean;
|
|
440
445
|
removeNavigation: boolean;
|
|
441
446
|
skipImages: boolean;
|
|
447
|
+
stripNewlines: boolean;
|
|
442
448
|
stripTags: string[];
|
|
449
|
+
strongEmSymbol: string;
|
|
443
450
|
wrapWidth: number;
|
|
444
451
|
wrap: boolean;
|
|
445
452
|
}
|
|
@@ -452,7 +459,10 @@ export enum WasmContentFilterKind {
|
|
|
452
459
|
}
|
|
453
460
|
|
|
454
461
|
/**
|
|
455
|
-
* Information about
|
|
462
|
+
* Information about a cookie that a crawled page set.
|
|
463
|
+
*
|
|
464
|
+
* A crawl result gives `domain` and `path` as the browser stores the cookie, so a cookie has
|
|
465
|
+
* the same fields whether an HTTP fetch or the browser received it.
|
|
456
466
|
*/
|
|
457
467
|
export class WasmCookieInfo {
|
|
458
468
|
free(): void;
|
|
@@ -655,7 +665,7 @@ export class WasmCrawlPageResult {
|
|
|
655
665
|
* wasm-bindgen cannot borrow class values nested in an array. To keep those handles,
|
|
656
666
|
* copy each retained element with `copyForTransfer()` before passing the array.
|
|
657
667
|
*/
|
|
658
|
-
constructor(url?: string | null, normalizedUrl?: string | null, statusCode?: number | null, contentType?: string | null, html?: string | null, bodySize?: number | null, metadata?: WasmPageMetadata | null, links?: WasmLinkInfo[] | null, images?: WasmImageInfo[] | null, feeds?: WasmFeedInfo[] | null, jsonLd?: WasmJsonLdEntry[] | null, depth?: number | null, stayedOnDomain?: boolean | null, wasSkipped?: boolean | null, isPdf?: boolean | null, browserUsed?: boolean | null, finalUrl?: string | null, redirectCount?: number | null, noindexDetected?: boolean | null, nofollowDetected?: boolean | null, ssrfRefusedUrls?: string[] | null, detectedCharset?: string | null, markdown?: WasmMarkdownResult | null, extractedData?: any | null, extractionMeta?: WasmExtractionMeta | null, downloadedDocument?: WasmDownloadedDocument | null);
|
|
668
|
+
constructor(url?: string | null, normalizedUrl?: string | null, statusCode?: number | null, contentType?: string | null, html?: string | null, bodySize?: number | null, metadata?: WasmPageMetadata | null, links?: WasmLinkInfo[] | null, images?: WasmImageInfo[] | null, feeds?: WasmFeedInfo[] | null, jsonLd?: WasmJsonLdEntry[] | null, depth?: number | null, stayedOnDomain?: boolean | null, wasSkipped?: boolean | null, isPdf?: boolean | null, browserUsed?: boolean | null, finalUrl?: string | null, redirectCount?: number | null, noindexDetected?: boolean | null, nofollowDetected?: boolean | null, ssrfRefusedUrls?: string[] | null, detectedCharset?: string | null, markdown?: WasmMarkdownResult | null, extractedData?: any | null, extractionMeta?: WasmExtractionMeta | null, downloadedDocument?: WasmDownloadedDocument | null, originalUrl?: string | null);
|
|
659
669
|
bodySize: number;
|
|
660
670
|
browserUsed: boolean;
|
|
661
671
|
contentType: string;
|
|
@@ -729,6 +739,7 @@ export class WasmCrawlPageResult {
|
|
|
729
739
|
nofollowDetected: boolean;
|
|
730
740
|
noindexDetected: boolean;
|
|
731
741
|
normalizedUrl: string;
|
|
742
|
+
originalUrl: string;
|
|
732
743
|
redirectCount: number;
|
|
733
744
|
ssrfRefusedUrls: string[];
|
|
734
745
|
statusCode: number;
|
|
@@ -1126,10 +1137,11 @@ export class WasmLinkInfo {
|
|
|
1126
1137
|
*/
|
|
1127
1138
|
copyForTransfer(): WasmLinkInfo;
|
|
1128
1139
|
static default(): WasmLinkInfo;
|
|
1129
|
-
constructor(url?: string | null, text?: string | null, linkType?: WasmLinkType | null, nofollow?: boolean | null, rel?: string | null);
|
|
1140
|
+
constructor(url?: string | null, text?: string | null, linkType?: WasmLinkType | null, nofollow?: boolean | null, rel?: string | null, originalUrl?: string | null);
|
|
1130
1141
|
get linkType(): string;
|
|
1131
1142
|
set linkType(value: WasmLinkType);
|
|
1132
1143
|
nofollow: boolean;
|
|
1144
|
+
originalUrl: string;
|
|
1133
1145
|
get rel(): string | undefined;
|
|
1134
1146
|
set rel(value: string | null | undefined);
|
|
1135
1147
|
text: string;
|
|
@@ -1790,6 +1790,28 @@ export class WasmContentConfig {
|
|
|
1790
1790
|
const ptr = this.__destroy_into_raw();
|
|
1791
1791
|
wasm.__wbg_wasmcontentconfig_free(ptr, 0);
|
|
1792
1792
|
}
|
|
1793
|
+
/**
|
|
1794
|
+
* @returns {string}
|
|
1795
|
+
*/
|
|
1796
|
+
get bullets() {
|
|
1797
|
+
let deferred1_0;
|
|
1798
|
+
let deferred1_1;
|
|
1799
|
+
try {
|
|
1800
|
+
const ret = wasm.wasmcontentconfig_bullets(this.__wbg_ptr);
|
|
1801
|
+
deferred1_0 = ret[0];
|
|
1802
|
+
deferred1_1 = ret[1];
|
|
1803
|
+
return getStringFromWasm0(ret[0], ret[1]);
|
|
1804
|
+
} finally {
|
|
1805
|
+
wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
|
|
1806
|
+
}
|
|
1807
|
+
}
|
|
1808
|
+
/**
|
|
1809
|
+
* @returns {boolean}
|
|
1810
|
+
*/
|
|
1811
|
+
get compactTables() {
|
|
1812
|
+
const ret = wasm.wasmcontentconfig_compactTables(this.__wbg_ptr);
|
|
1813
|
+
return ret !== 0;
|
|
1814
|
+
}
|
|
1793
1815
|
/**
|
|
1794
1816
|
* Return an independent handle for an ownership-taking API.
|
|
1795
1817
|
*
|
|
@@ -1831,6 +1853,29 @@ export class WasmContentConfig {
|
|
|
1831
1853
|
const ret = wasm.wasmcontentconfig_includeDocumentStructure(this.__wbg_ptr);
|
|
1832
1854
|
return ret !== 0;
|
|
1833
1855
|
}
|
|
1856
|
+
/**
|
|
1857
|
+
* @returns {boolean}
|
|
1858
|
+
*/
|
|
1859
|
+
get keepHidden() {
|
|
1860
|
+
const ret = wasm.wasmcontentconfig_keepHidden(this.__wbg_ptr);
|
|
1861
|
+
return ret !== 0;
|
|
1862
|
+
}
|
|
1863
|
+
/**
|
|
1864
|
+
* @returns {string[]}
|
|
1865
|
+
*/
|
|
1866
|
+
get keepInlineImagesIn() {
|
|
1867
|
+
const ret = wasm.wasmcontentconfig_keepInlineImagesIn(this.__wbg_ptr);
|
|
1868
|
+
var v1 = getArrayJsValueFromWasm0(ret[0], ret[1]);
|
|
1869
|
+
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
1870
|
+
return v1;
|
|
1871
|
+
}
|
|
1872
|
+
/**
|
|
1873
|
+
* @returns {number}
|
|
1874
|
+
*/
|
|
1875
|
+
get listIndentWidth() {
|
|
1876
|
+
const ret = wasm.wasmcontentconfig_listIndentWidth(this.__wbg_ptr);
|
|
1877
|
+
return ret >>> 0;
|
|
1878
|
+
}
|
|
1834
1879
|
/**
|
|
1835
1880
|
* @returns {number | undefined}
|
|
1836
1881
|
*/
|
|
@@ -1852,8 +1897,15 @@ export class WasmContentConfig {
|
|
|
1852
1897
|
* @param {boolean | null} [includeDocumentStructure]
|
|
1853
1898
|
* @param {boolean | null} [extractMetadata]
|
|
1854
1899
|
* @param {number | null} [maxDepth]
|
|
1855
|
-
|
|
1856
|
-
|
|
1900
|
+
* @param {boolean | null} [stripNewlines]
|
|
1901
|
+
* @param {string | null} [bullets]
|
|
1902
|
+
* @param {number | null} [listIndentWidth]
|
|
1903
|
+
* @param {string | null} [strongEmSymbol]
|
|
1904
|
+
* @param {boolean | null} [compactTables]
|
|
1905
|
+
* @param {string[] | null} [keepInlineImagesIn]
|
|
1906
|
+
* @param {boolean | null} [keepHidden]
|
|
1907
|
+
*/
|
|
1908
|
+
constructor(outputFormat, preprocessingPreset, removeNavigation, removeForms, stripTags, preserveTags, excludeSelectors, skipImages, wrap, wrapWidth, includeDocumentStructure, extractMetadata, maxDepth, stripNewlines, bullets, listIndentWidth, strongEmSymbol, compactTables, keepInlineImagesIn, keepHidden) {
|
|
1857
1909
|
var ptr0 = isLikeNone(outputFormat) ? 0 : passStringToWasm0(outputFormat, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1858
1910
|
var len0 = WASM_VECTOR_LEN;
|
|
1859
1911
|
var ptr1 = isLikeNone(preprocessingPreset) ? 0 : passStringToWasm0(preprocessingPreset, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
@@ -1864,7 +1916,13 @@ export class WasmContentConfig {
|
|
|
1864
1916
|
var len3 = WASM_VECTOR_LEN;
|
|
1865
1917
|
var ptr4 = isLikeNone(excludeSelectors) ? 0 : passArrayJsValueToWasm0(excludeSelectors, wasm.__wbindgen_malloc);
|
|
1866
1918
|
var len4 = WASM_VECTOR_LEN;
|
|
1867
|
-
|
|
1919
|
+
var ptr5 = isLikeNone(bullets) ? 0 : passStringToWasm0(bullets, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1920
|
+
var len5 = WASM_VECTOR_LEN;
|
|
1921
|
+
var ptr6 = isLikeNone(strongEmSymbol) ? 0 : passStringToWasm0(strongEmSymbol, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1922
|
+
var len6 = WASM_VECTOR_LEN;
|
|
1923
|
+
var ptr7 = isLikeNone(keepInlineImagesIn) ? 0 : passArrayJsValueToWasm0(keepInlineImagesIn, wasm.__wbindgen_malloc);
|
|
1924
|
+
var len7 = WASM_VECTOR_LEN;
|
|
1925
|
+
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);
|
|
1868
1926
|
this.__wbg_ptr = ret;
|
|
1869
1927
|
WasmContentConfigFinalization.register(this, this.__wbg_ptr, this);
|
|
1870
1928
|
return this;
|
|
@@ -1922,6 +1980,20 @@ export class WasmContentConfig {
|
|
|
1922
1980
|
const ret = wasm.wasmcontentconfig_removeNavigation(this.__wbg_ptr);
|
|
1923
1981
|
return ret !== 0;
|
|
1924
1982
|
}
|
|
1983
|
+
/**
|
|
1984
|
+
* @param {string} value
|
|
1985
|
+
*/
|
|
1986
|
+
set bullets(value) {
|
|
1987
|
+
const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
1988
|
+
const len0 = WASM_VECTOR_LEN;
|
|
1989
|
+
wasm.wasmcontentconfig_set_bullets(this.__wbg_ptr, ptr0, len0);
|
|
1990
|
+
}
|
|
1991
|
+
/**
|
|
1992
|
+
* @param {boolean} value
|
|
1993
|
+
*/
|
|
1994
|
+
set compactTables(value) {
|
|
1995
|
+
wasm.wasmcontentconfig_set_compactTables(this.__wbg_ptr, value);
|
|
1996
|
+
}
|
|
1925
1997
|
/**
|
|
1926
1998
|
* @param {string[]} value
|
|
1927
1999
|
*/
|
|
@@ -1942,6 +2014,26 @@ export class WasmContentConfig {
|
|
|
1942
2014
|
set includeDocumentStructure(value) {
|
|
1943
2015
|
wasm.wasmcontentconfig_set_includeDocumentStructure(this.__wbg_ptr, value);
|
|
1944
2016
|
}
|
|
2017
|
+
/**
|
|
2018
|
+
* @param {boolean} value
|
|
2019
|
+
*/
|
|
2020
|
+
set keepHidden(value) {
|
|
2021
|
+
wasm.wasmcontentconfig_set_keepHidden(this.__wbg_ptr, value);
|
|
2022
|
+
}
|
|
2023
|
+
/**
|
|
2024
|
+
* @param {string[]} value
|
|
2025
|
+
*/
|
|
2026
|
+
set keepInlineImagesIn(value) {
|
|
2027
|
+
const ptr0 = passArrayJsValueToWasm0(value, wasm.__wbindgen_malloc);
|
|
2028
|
+
const len0 = WASM_VECTOR_LEN;
|
|
2029
|
+
wasm.wasmcontentconfig_set_keepInlineImagesIn(this.__wbg_ptr, ptr0, len0);
|
|
2030
|
+
}
|
|
2031
|
+
/**
|
|
2032
|
+
* @param {number} value
|
|
2033
|
+
*/
|
|
2034
|
+
set listIndentWidth(value) {
|
|
2035
|
+
wasm.wasmcontentconfig_set_listIndentWidth(this.__wbg_ptr, value);
|
|
2036
|
+
}
|
|
1945
2037
|
/**
|
|
1946
2038
|
* @param {number | null} [value]
|
|
1947
2039
|
*/
|
|
@@ -1990,6 +2082,12 @@ export class WasmContentConfig {
|
|
|
1990
2082
|
set skipImages(value) {
|
|
1991
2083
|
wasm.wasmcontentconfig_set_skipImages(this.__wbg_ptr, value);
|
|
1992
2084
|
}
|
|
2085
|
+
/**
|
|
2086
|
+
* @param {boolean} value
|
|
2087
|
+
*/
|
|
2088
|
+
set stripNewlines(value) {
|
|
2089
|
+
wasm.wasmcontentconfig_set_stripNewlines(this.__wbg_ptr, value);
|
|
2090
|
+
}
|
|
1993
2091
|
/**
|
|
1994
2092
|
* @param {string[]} value
|
|
1995
2093
|
*/
|
|
@@ -1998,6 +2096,14 @@ export class WasmContentConfig {
|
|
|
1998
2096
|
const len0 = WASM_VECTOR_LEN;
|
|
1999
2097
|
wasm.wasmcontentconfig_set_stripTags(this.__wbg_ptr, ptr0, len0);
|
|
2000
2098
|
}
|
|
2099
|
+
/**
|
|
2100
|
+
* @param {string} value
|
|
2101
|
+
*/
|
|
2102
|
+
set strongEmSymbol(value) {
|
|
2103
|
+
const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
2104
|
+
const len0 = WASM_VECTOR_LEN;
|
|
2105
|
+
wasm.wasmcontentconfig_set_strongEmSymbol(this.__wbg_ptr, ptr0, len0);
|
|
2106
|
+
}
|
|
2001
2107
|
/**
|
|
2002
2108
|
* @param {number} value
|
|
2003
2109
|
*/
|
|
@@ -2017,6 +2123,13 @@ export class WasmContentConfig {
|
|
|
2017
2123
|
const ret = wasm.wasmcontentconfig_skipImages(this.__wbg_ptr);
|
|
2018
2124
|
return ret !== 0;
|
|
2019
2125
|
}
|
|
2126
|
+
/**
|
|
2127
|
+
* @returns {boolean}
|
|
2128
|
+
*/
|
|
2129
|
+
get stripNewlines() {
|
|
2130
|
+
const ret = wasm.wasmcontentconfig_stripNewlines(this.__wbg_ptr);
|
|
2131
|
+
return ret !== 0;
|
|
2132
|
+
}
|
|
2020
2133
|
/**
|
|
2021
2134
|
* @returns {string[]}
|
|
2022
2135
|
*/
|
|
@@ -2026,6 +2139,21 @@ export class WasmContentConfig {
|
|
|
2026
2139
|
wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
|
|
2027
2140
|
return v1;
|
|
2028
2141
|
}
|
|
2142
|
+
/**
|
|
2143
|
+
* @returns {string}
|
|
2144
|
+
*/
|
|
2145
|
+
get strongEmSymbol() {
|
|
2146
|
+
let deferred1_0;
|
|
2147
|
+
let deferred1_1;
|
|
2148
|
+
try {
|
|
2149
|
+
const ret = wasm.wasmcontentconfig_strongEmSymbol(this.__wbg_ptr);
|
|
2150
|
+
deferred1_0 = ret[0];
|
|
2151
|
+
deferred1_1 = ret[1];
|
|
2152
|
+
return getStringFromWasm0(ret[0], ret[1]);
|
|
2153
|
+
} finally {
|
|
2154
|
+
wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
|
|
2155
|
+
}
|
|
2156
|
+
}
|
|
2029
2157
|
/**
|
|
2030
2158
|
* @returns {number}
|
|
2031
2159
|
*/
|
|
@@ -2052,7 +2180,10 @@ export const WasmContentFilterKind = Object.freeze({
|
|
|
2052
2180
|
});
|
|
2053
2181
|
|
|
2054
2182
|
/**
|
|
2055
|
-
* Information about
|
|
2183
|
+
* Information about a cookie that a crawled page set.
|
|
2184
|
+
*
|
|
2185
|
+
* A crawl result gives `domain` and `path` as the browser stores the cookie, so a cookie has
|
|
2186
|
+
* the same fields whether an HTTP fetch or the browser received it.
|
|
2056
2187
|
*/
|
|
2057
2188
|
export class WasmCookieInfo {
|
|
2058
2189
|
static __wrap(ptr) {
|
|
@@ -3516,8 +3647,9 @@ export class WasmCrawlPageResult {
|
|
|
3516
3647
|
* @param {any | null} [extractedData]
|
|
3517
3648
|
* @param {WasmExtractionMeta | null} [extractionMeta]
|
|
3518
3649
|
* @param {WasmDownloadedDocument | null} [downloadedDocument]
|
|
3650
|
+
* @param {string | null} [originalUrl]
|
|
3519
3651
|
*/
|
|
3520
|
-
constructor(url, normalizedUrl, statusCode, contentType, html, bodySize, metadata, links, images, feeds, jsonLd, depth, stayedOnDomain, wasSkipped, isPdf, browserUsed, finalUrl, redirectCount, noindexDetected, nofollowDetected, ssrfRefusedUrls, detectedCharset, markdown, extractedData, extractionMeta, downloadedDocument) {
|
|
3652
|
+
constructor(url, normalizedUrl, statusCode, contentType, html, bodySize, metadata, links, images, feeds, jsonLd, depth, stayedOnDomain, wasSkipped, isPdf, browserUsed, finalUrl, redirectCount, noindexDetected, nofollowDetected, ssrfRefusedUrls, detectedCharset, markdown, extractedData, extractionMeta, downloadedDocument, originalUrl) {
|
|
3521
3653
|
var ptr0 = isLikeNone(url) ? 0 : passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
3522
3654
|
var len0 = WASM_VECTOR_LEN;
|
|
3523
3655
|
var ptr1 = isLikeNone(normalizedUrl) ? 0 : passStringToWasm0(normalizedUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
@@ -3560,7 +3692,9 @@ export class WasmCrawlPageResult {
|
|
|
3560
3692
|
_assertClass(downloadedDocument, WasmDownloadedDocument);
|
|
3561
3693
|
ptr14 = downloadedDocument.__destroy_into_raw();
|
|
3562
3694
|
}
|
|
3563
|
-
|
|
3695
|
+
var ptr15 = isLikeNone(originalUrl) ? 0 : passStringToWasm0(originalUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
3696
|
+
var len15 = WASM_VECTOR_LEN;
|
|
3697
|
+
const ret = wasm.wasmcrawlpageresult_new(ptr0, len0, ptr1, len1, isLikeNone(statusCode) ? 0xFFFFFF : statusCode, ptr2, len2, ptr3, len3, isLikeNone(bodySize) ? Number.MAX_SAFE_INTEGER : (bodySize) >>> 0, ptr4, ptr5, len5, ptr6, len6, ptr7, len7, ptr8, len8, isLikeNone(depth) ? Number.MAX_SAFE_INTEGER : (depth) >>> 0, isLikeNone(stayedOnDomain) ? 0xFFFFFF : stayedOnDomain ? 1 : 0, isLikeNone(wasSkipped) ? 0xFFFFFF : wasSkipped ? 1 : 0, isLikeNone(isPdf) ? 0xFFFFFF : isPdf ? 1 : 0, isLikeNone(browserUsed) ? 0xFFFFFF : browserUsed ? 1 : 0, ptr9, len9, isLikeNone(redirectCount) ? Number.MAX_SAFE_INTEGER : (redirectCount) >>> 0, isLikeNone(noindexDetected) ? 0xFFFFFF : noindexDetected ? 1 : 0, isLikeNone(nofollowDetected) ? 0xFFFFFF : nofollowDetected ? 1 : 0, ptr10, len10, ptr11, len11, ptr12, isLikeNone(extractedData) ? 0 : addToExternrefTable0(extractedData), ptr13, ptr14, ptr15, len15);
|
|
3564
3698
|
this.__wbg_ptr = ret;
|
|
3565
3699
|
WasmCrawlPageResultFinalization.register(this, this.__wbg_ptr, this);
|
|
3566
3700
|
return this;
|
|
@@ -3594,6 +3728,21 @@ export class WasmCrawlPageResult {
|
|
|
3594
3728
|
wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
|
|
3595
3729
|
}
|
|
3596
3730
|
}
|
|
3731
|
+
/**
|
|
3732
|
+
* @returns {string}
|
|
3733
|
+
*/
|
|
3734
|
+
get originalUrl() {
|
|
3735
|
+
let deferred1_0;
|
|
3736
|
+
let deferred1_1;
|
|
3737
|
+
try {
|
|
3738
|
+
const ret = wasm.wasmcrawlpageresult_originalUrl(this.__wbg_ptr);
|
|
3739
|
+
deferred1_0 = ret[0];
|
|
3740
|
+
deferred1_1 = ret[1];
|
|
3741
|
+
return getStringFromWasm0(ret[0], ret[1]);
|
|
3742
|
+
} finally {
|
|
3743
|
+
wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
|
|
3744
|
+
}
|
|
3745
|
+
}
|
|
3597
3746
|
/**
|
|
3598
3747
|
* @returns {number}
|
|
3599
3748
|
*/
|
|
@@ -3755,6 +3904,14 @@ export class WasmCrawlPageResult {
|
|
|
3755
3904
|
const len0 = WASM_VECTOR_LEN;
|
|
3756
3905
|
wasm.wasmcrawlpageresult_set_normalizedUrl(this.__wbg_ptr, ptr0, len0);
|
|
3757
3906
|
}
|
|
3907
|
+
/**
|
|
3908
|
+
* @param {string} value
|
|
3909
|
+
*/
|
|
3910
|
+
set originalUrl(value) {
|
|
3911
|
+
const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
3912
|
+
const len0 = WASM_VECTOR_LEN;
|
|
3913
|
+
wasm.wasmcrawlpageresult_set_originalUrl(this.__wbg_ptr, ptr0, len0);
|
|
3914
|
+
}
|
|
3758
3915
|
/**
|
|
3759
3916
|
* @param {number} value
|
|
3760
3917
|
*/
|
|
@@ -5774,15 +5931,18 @@ export class WasmLinkInfo {
|
|
|
5774
5931
|
* @param {WasmLinkType | null} [linkType]
|
|
5775
5932
|
* @param {boolean | null} [nofollow]
|
|
5776
5933
|
* @param {string | null} [rel]
|
|
5934
|
+
* @param {string | null} [originalUrl]
|
|
5777
5935
|
*/
|
|
5778
|
-
constructor(url, text, linkType, nofollow, rel) {
|
|
5936
|
+
constructor(url, text, linkType, nofollow, rel, originalUrl) {
|
|
5779
5937
|
var ptr0 = isLikeNone(url) ? 0 : passStringToWasm0(url, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
5780
5938
|
var len0 = WASM_VECTOR_LEN;
|
|
5781
5939
|
var ptr1 = isLikeNone(text) ? 0 : passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
5782
5940
|
var len1 = WASM_VECTOR_LEN;
|
|
5783
5941
|
var ptr2 = isLikeNone(rel) ? 0 : passStringToWasm0(rel, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
5784
5942
|
var len2 = WASM_VECTOR_LEN;
|
|
5785
|
-
|
|
5943
|
+
var ptr3 = isLikeNone(originalUrl) ? 0 : passStringToWasm0(originalUrl, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
5944
|
+
var len3 = WASM_VECTOR_LEN;
|
|
5945
|
+
const ret = wasm.wasmlinkinfo_new(ptr0, len0, ptr1, len1, isLikeNone(linkType) ? 4 : linkType, isLikeNone(nofollow) ? 0xFFFFFF : nofollow ? 1 : 0, ptr2, len2, ptr3, len3);
|
|
5786
5946
|
this.__wbg_ptr = ret;
|
|
5787
5947
|
WasmLinkInfoFinalization.register(this, this.__wbg_ptr, this);
|
|
5788
5948
|
return this;
|
|
@@ -5794,6 +5954,21 @@ export class WasmLinkInfo {
|
|
|
5794
5954
|
const ret = wasm.wasmlinkinfo_nofollow(this.__wbg_ptr);
|
|
5795
5955
|
return ret !== 0;
|
|
5796
5956
|
}
|
|
5957
|
+
/**
|
|
5958
|
+
* @returns {string}
|
|
5959
|
+
*/
|
|
5960
|
+
get originalUrl() {
|
|
5961
|
+
let deferred1_0;
|
|
5962
|
+
let deferred1_1;
|
|
5963
|
+
try {
|
|
5964
|
+
const ret = wasm.wasmlinkinfo_originalUrl(this.__wbg_ptr);
|
|
5965
|
+
deferred1_0 = ret[0];
|
|
5966
|
+
deferred1_1 = ret[1];
|
|
5967
|
+
return getStringFromWasm0(ret[0], ret[1]);
|
|
5968
|
+
} finally {
|
|
5969
|
+
wasm.__wbindgen_free(deferred1_0, deferred1_1, 1);
|
|
5970
|
+
}
|
|
5971
|
+
}
|
|
5797
5972
|
/**
|
|
5798
5973
|
* @returns {string | undefined}
|
|
5799
5974
|
*/
|
|
@@ -5818,6 +5993,14 @@ export class WasmLinkInfo {
|
|
|
5818
5993
|
set nofollow(value) {
|
|
5819
5994
|
wasm.wasmlinkinfo_set_nofollow(this.__wbg_ptr, value);
|
|
5820
5995
|
}
|
|
5996
|
+
/**
|
|
5997
|
+
* @param {string} value
|
|
5998
|
+
*/
|
|
5999
|
+
set originalUrl(value) {
|
|
6000
|
+
const ptr0 = passStringToWasm0(value, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
|
|
6001
|
+
const len0 = WASM_VECTOR_LEN;
|
|
6002
|
+
wasm.wasmlinkinfo_set_originalUrl(this.__wbg_ptr, ptr0, len0);
|
|
6003
|
+
}
|
|
5821
6004
|
/**
|
|
5822
6005
|
* @param {string | null} [value]
|
|
5823
6006
|
*/
|
|
@@ -9065,7 +9248,7 @@ export function __wbg_new_typed_b01cb72a8af741a3(arg0, arg1) {
|
|
|
9065
9248
|
const a = state0.a;
|
|
9066
9249
|
state0.a = 0;
|
|
9067
9250
|
try {
|
|
9068
|
-
return
|
|
9251
|
+
return wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined_______true_(a, state0.b, arg0, arg1);
|
|
9069
9252
|
} finally {
|
|
9070
9253
|
state0.a = a;
|
|
9071
9254
|
}
|
|
@@ -9359,13 +9542,13 @@ export function __wbg_wasmsitemapurl_unwrap(arg0) {
|
|
|
9359
9542
|
return ret;
|
|
9360
9543
|
}
|
|
9361
9544
|
export function __wbindgen_generic_0000000000000001(arg0, arg1) {
|
|
9362
|
-
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx:
|
|
9363
|
-
const ret = makeMutClosure(arg0, arg1,
|
|
9545
|
+
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [Externref], shim_idx: 1467, ret: Result(Unit), inner_ret: Some(Result(Unit)) }, mutable: true }) -> Externref`.
|
|
9546
|
+
const ret = makeMutClosure(arg0, arg1, wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___wasm_bindgen_ceaac9754e66b2fe___JsValue__core_608f92abc48d28da___result__Result_____wasm_bindgen_ceaac9754e66b2fe___JsError___true_);
|
|
9364
9547
|
return ret;
|
|
9365
9548
|
}
|
|
9366
9549
|
export function __wbindgen_generic_0000000000000002(arg0, arg1) {
|
|
9367
|
-
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx:
|
|
9368
|
-
const ret = makeMutClosure(arg0, arg1,
|
|
9550
|
+
// Cast intrinsic for `Closure(Closure { owned: true, function: Function { arguments: [], shim_idx: 1309, ret: Unit, inner_ret: Some(Unit) }, mutable: true }) -> Externref`.
|
|
9551
|
+
const ret = makeMutClosure(arg0, arg1, wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke_______true_);
|
|
9369
9552
|
return ret;
|
|
9370
9553
|
}
|
|
9371
9554
|
export function __wbindgen_generic_0000000000000003(arg0) {
|
|
@@ -9397,19 +9580,19 @@ export function __wbindgen_init_externref_table() {
|
|
|
9397
9580
|
table.set(offset + 2, true);
|
|
9398
9581
|
table.set(offset + 3, false);
|
|
9399
9582
|
}
|
|
9400
|
-
function
|
|
9401
|
-
wasm.
|
|
9583
|
+
function wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke_______true_(arg0, arg1) {
|
|
9584
|
+
wasm.wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke_______true_(arg0, arg1);
|
|
9402
9585
|
}
|
|
9403
9586
|
|
|
9404
|
-
function
|
|
9405
|
-
const ret = wasm.
|
|
9587
|
+
function wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___wasm_bindgen_ceaac9754e66b2fe___JsValue__core_608f92abc48d28da___result__Result_____wasm_bindgen_ceaac9754e66b2fe___JsError___true_(arg0, arg1, arg2) {
|
|
9588
|
+
const ret = wasm.wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___wasm_bindgen_ceaac9754e66b2fe___JsValue__core_608f92abc48d28da___result__Result_____wasm_bindgen_ceaac9754e66b2fe___JsError___true_(arg0, arg1, arg2);
|
|
9406
9589
|
if (ret[1]) {
|
|
9407
9590
|
throw takeFromExternrefTable0(ret[0]);
|
|
9408
9591
|
}
|
|
9409
9592
|
}
|
|
9410
9593
|
|
|
9411
|
-
function
|
|
9412
|
-
wasm.
|
|
9594
|
+
function wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined_______true_(arg0, arg1, arg2, arg3) {
|
|
9595
|
+
wasm.wasm_bindgen_ceaac9754e66b2fe___convert__closures_____invoke___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined___js_sys_6b40a5061b778857___Function_fn_wasm_bindgen_ceaac9754e66b2fe___JsValue_____wasm_bindgen_ceaac9754e66b2fe___sys__Undefined_______true_(arg0, arg1, arg2, arg3);
|
|
9413
9596
|
}
|
|
9414
9597
|
|
|
9415
9598
|
|
|
Binary file
|