@crawlee/utils 3.13.1-beta.8 → 3.13.1
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 +2 -2
- package/index.mjs +1 -0
- package/internals/cheerio.d.ts +1 -1
- package/internals/cheerio.d.ts.map +1 -1
- package/internals/cheerio.js.map +1 -1
- package/internals/debug.d.ts +1 -1
- package/internals/debug.d.ts.map +1 -1
- package/internals/memory-info.d.ts.map +1 -1
- package/internals/memory-info.js.map +1 -1
- package/internals/robots.d.ts +6 -5
- package/internals/robots.d.ts.map +1 -1
- package/internals/robots.js +12 -11
- package/internals/robots.js.map +1 -1
- package/internals/sitemap.d.ts +1 -1
- package/internals/sitemap.d.ts.map +1 -1
- package/internals/sitemap.js +5 -6
- package/internals/sitemap.js.map +1 -1
- package/internals/social.d.ts +31 -31
- package/internals/social.d.ts.map +1 -1
- package/internals/social.js +80 -80
- package/internals/social.js.map +1 -1
- package/internals/systemInfoV2/ps-tree.d.ts.map +1 -1
- package/internals/systemInfoV2/ps-tree.js +4 -4
- package/internals/systemInfoV2/ps-tree.js.map +1 -1
- package/package.json +3 -3
- package/tsconfig.build.tsbuildinfo +1 -1
package/README.md
CHANGED
|
@@ -28,7 +28,7 @@ Crawlee is available as the [`crawlee`](https://www.npmjs.com/package/crawlee) N
|
|
|
28
28
|
|
|
29
29
|
## Installation
|
|
30
30
|
|
|
31
|
-
We recommend visiting the [Introduction tutorial](https://crawlee.dev/docs/introduction) in Crawlee documentation for more information.
|
|
31
|
+
We recommend visiting the [Introduction tutorial](https://crawlee.dev/js/docs/introduction) in Crawlee documentation for more information.
|
|
32
32
|
|
|
33
33
|
> Crawlee requires **Node.js 16 or higher**.
|
|
34
34
|
|
|
@@ -78,7 +78,7 @@ const crawler = new PlaywrightCrawler({
|
|
|
78
78
|
await crawler.run(['https://crawlee.dev']);
|
|
79
79
|
```
|
|
80
80
|
|
|
81
|
-
By default, Crawlee stores data to `./storage` in the current working directory. You can override this directory via Crawlee configuration. For details, see [Configuration guide](https://crawlee.dev/docs/guides/configuration), [Request storage](https://crawlee.dev/docs/guides/request-storage) and [Result storage](https://crawlee.dev/docs/guides/result-storage).
|
|
81
|
+
By default, Crawlee stores data to `./storage` in the current working directory. You can override this directory via Crawlee configuration. For details, see [Configuration guide](https://crawlee.dev/js/docs/guides/configuration), [Request storage](https://crawlee.dev/js/docs/guides/request-storage) and [Result storage](https://crawlee.dev/js/docs/guides/result-storage).
|
|
82
82
|
|
|
83
83
|
### Installing pre-release versions
|
|
84
84
|
|
package/index.mjs
CHANGED
|
@@ -5,6 +5,7 @@ export const CLOUDFLARE_RETRY_CSS_SELECTORS = mod.CLOUDFLARE_RETRY_CSS_SELECTORS
|
|
|
5
5
|
export const RETRY_CSS_SELECTORS = mod.RETRY_CSS_SELECTORS;
|
|
6
6
|
export const ROTATE_PROXY_ERRORS = mod.ROTATE_PROXY_ERRORS;
|
|
7
7
|
export const RobotsFile = mod.RobotsFile;
|
|
8
|
+
export const RobotsTxtFile = mod.RobotsTxtFile;
|
|
8
9
|
export const Sitemap = mod.Sitemap;
|
|
9
10
|
export const URL_NO_COMMAS_REGEX = mod.URL_NO_COMMAS_REGEX;
|
|
10
11
|
export const URL_WITH_COMMAS_REGEX = mod.URL_WITH_COMMAS_REGEX;
|
package/internals/cheerio.d.ts
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"cheerio.d.ts","sourceRoot":"","sources":["../../src/internals/cheerio.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,
|
|
1
|
+
{"version":3,"file":"cheerio.d.ts","sourceRoot":"","sources":["../../src/internals/cheerio.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,MAAM,SAAS,CAAC;AAKhD,MAAM,MAAM,WAAW,GAAG,UAAU,CAAC,OAAO,IAAI,CAAC,CAAC;AAOlD;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2BG;AACH,wBAAgB,UAAU,CAAC,oBAAoB,EAAE,MAAM,GAAG,WAAW,GAAG,MAAM,CA2C7E;AAED;;;;;;;;GAQG;AACH,wBAAgB,sBAAsB,CAAC,CAAC,EAAE,UAAU,EAAE,QAAQ,SAAM,EAAE,OAAO,SAAK,GAAG,MAAM,EAAE,CAwB5F"}
|
package/internals/cheerio.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"cheerio.js","sourceRoot":"","sources":["../../src/internals/cheerio.ts"],"names":[],"mappings":";;AAyCA,gCA2CC;AAWD,wDAwBC;;AArHD,yDAAmC;AAEnC,iDAAgD;AAIhD,gIAAgI;AAChI,MAAM,eAAe,GAAG,uCAAuC,CAAC;AAChE,MAAM,gBAAgB,GAClB,sGAAsG,CAAC;AAE3G;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2BG;AACH,SAAgB,UAAU,CAAC,oBAA0C;IACjE,IAAI,CAAC,oBAAoB;QAAE,OAAO,EAAE,CAAC;IAErC,MAAM,CAAC,GACH,OAAO,oBAAoB,KAAK,UAAU;QACtC,CAAC,CAAC,oBAAoB;QACtB,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,oBAAoB,EAAE,EAAE,cAAc,EAAE,IAAI,EAAE,CAAC,CAAC;IACvE,IAAI,IAAI,GAAG,EAAE,CAAC;IAEd,MAAM,OAAO,GAAG,CAAC,KAAiB,EAAE,EAAE;QAClC,MAAM,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;QACrC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC;YAC3B,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;YACtB,IAAI,IAAI,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;gBACvB,qDAAqD;gBACrD,IAAI,KAAK,CAAC;gBACV,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM,CAAC,OAAO,KAAK,KAAK;oBAAE,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC;;oBAC/D,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;gBAC5C,+EAA+E;gBAC/E,IAAI,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC;oBAAE,KAAK,GAAG,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;gBAC9E,IAAI,IAAI,KAAK,CAAC;YAClB,CAAC;iBAAM,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,IAAI,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBACvE,qCAAqC;YACzC,CAAC;iBAAM,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC/B,IAAI,IAAI,IAAI,CAAC;YACjB,CAAC;iBAAM,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC/B,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBACvB,IAAI,IAAI,IAAI,CAAC;YACjB,CAAC;iBAAM,CAAC;gBACJ,2EAA2E;gBAC3E,MAAM,UAAU,GAAG,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;gBACvD,IAAI,UAAU,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC;oBAAE,IAAI,IAAI,IAAI,CAAC;gBACtD,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBACvB,IAAI,UAAU,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC;oBAAE,IAAI,IAAI,IAAI,CAAC;YACzD,CAAC;QACL,CAAC;IACL,CAAC,CAAC;IAEF,kFAAkF;IAClF,MAAM,KAAK,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC;IACxB,OAAO,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;IAE7C,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC;AACvB,CAAC;AAED;;;;;;;;GAQG;AACH,SAAgB,sBAAsB,CAAC,CAAa,EAAE,
|
|
1
|
+
{"version":3,"file":"cheerio.js","sourceRoot":"","sources":["../../src/internals/cheerio.ts"],"names":[],"mappings":";;AAyCA,gCA2CC;AAWD,wDAwBC;;AArHD,yDAAmC;AAEnC,iDAAgD;AAIhD,gIAAgI;AAChI,MAAM,eAAe,GAAG,uCAAuC,CAAC;AAChE,MAAM,gBAAgB,GAClB,sGAAsG,CAAC;AAE3G;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2BG;AACH,SAAgB,UAAU,CAAC,oBAA0C;IACjE,IAAI,CAAC,oBAAoB;QAAE,OAAO,EAAE,CAAC;IAErC,MAAM,CAAC,GACH,OAAO,oBAAoB,KAAK,UAAU;QACtC,CAAC,CAAC,oBAAoB;QACtB,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,oBAAoB,EAAE,EAAE,cAAc,EAAE,IAAI,EAAE,CAAC,CAAC;IACvE,IAAI,IAAI,GAAG,EAAE,CAAC;IAEd,MAAM,OAAO,GAAG,CAAC,KAAiB,EAAE,EAAE;QAClC,MAAM,GAAG,GAAG,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;QACrC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC;YAC3B,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;YACtB,IAAI,IAAI,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;gBACvB,qDAAqD;gBACrD,IAAI,KAAK,CAAC;gBACV,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM,CAAC,OAAO,KAAK,KAAK;oBAAE,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC;;oBAC/D,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;gBAC5C,+EAA+E;gBAC/E,IAAI,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC;oBAAE,KAAK,GAAG,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;gBAC9E,IAAI,IAAI,KAAK,CAAC;YAClB,CAAC;iBAAM,IAAI,IAAI,CAAC,IAAI,KAAK,SAAS,IAAI,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBACvE,qCAAqC;YACzC,CAAC;iBAAM,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC/B,IAAI,IAAI,IAAI,CAAC;YACjB,CAAC;iBAAM,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC/B,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBACvB,IAAI,IAAI,IAAI,CAAC;YACjB,CAAC;iBAAM,CAAC;gBACJ,2EAA2E;gBAC3E,MAAM,UAAU,GAAG,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;gBACvD,IAAI,UAAU,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC;oBAAE,IAAI,IAAI,IAAI,CAAC;gBACtD,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBACvB,IAAI,UAAU,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC;oBAAE,IAAI,IAAI,IAAI,CAAC;YACzD,CAAC;QACL,CAAC;IACL,CAAC,CAAC;IAEF,kFAAkF;IAClF,MAAM,KAAK,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC;IACxB,OAAO,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;IAE7C,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC;AACvB,CAAC;AAED;;;;;;;;GAQG;AACH,SAAgB,sBAAsB,CAAC,CAAa,EAAE,QAAQ,GAAG,GAAG,EAAE,OAAO,GAAG,EAAE;IAC9E,MAAM,IAAI,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IACpC,MAAM,eAAe,GAAG,IAAI,IAAI,IAAA,6BAAc,EAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAE9D,IAAI,eAAe,EAAE,CAAC;QAClB,OAAO,GAAG,eAAe,CAAC;IAC9B,CAAC;IAED,OAAO,CAAC,CAAC,QAAQ,CAAC;SACb,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;SACnC,GAAG,EAAE;SACL,MAAM,CAAC,OAAO,CAAC;SACf,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QACV,yHAAyH;QACzH,MAAM,cAAc,GAAG,qBAAqB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,6CAA6C;QACtG,IAAI,CAAC,cAAc,IAAI,CAAC,OAAO,EAAE,CAAC;YAC9B,MAAM,IAAI,KAAK,CACX,qBAAqB,IAAI,uCAAuC;gBAC5D,2DAA2D,CAClE,CAAC;QACN,CAAC;QACD,OAAO,OAAO,CAAC,CAAC,CAAC,IAAA,6BAAc,EAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;IAC1D,CAAC,CAAC;SACD,MAAM,CAAC,OAAO,CAAa,CAAC;AACrC,CAAC"}
|
package/internals/debug.d.ts
CHANGED
package/internals/debug.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"debug.d.ts","sourceRoot":"","sources":["../../src/internals/debug.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,WAAW,CAAC;AAEjD,OAAO,KAAK,EAAE,
|
|
1
|
+
{"version":3,"file":"debug.d.ts","sourceRoot":"","sources":["../../src/internals/debug.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,WAAW,CAAC;AAEjD,OAAO,KAAK,EAAE,kBAAkB,EAAE,UAAU,EAAE,MAAM,gBAAgB,CAAC;AAGrE,UAAU,mBAAmB;IACzB,MAAM,IAAI,MAAM,CAAC;CACpB;AAED,UAAU,OAAO,CAAC,QAAQ,SAAS,UAAU,GAAG,UAAU;IACtD,EAAE,CAAC,EAAE,MAAM,CAAC;IACZ,GAAG,EAAE,MAAM,CAAC;IACZ,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,SAAS,EAAE,MAAM,CAAC;IAClB,MAAM,EAAE,kBAAkB,CAAC;IAC3B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,OAAO,EAAE,OAAO,CAAC;IACjB,UAAU,EAAE,MAAM,CAAC;IACnB,aAAa,EAAE,MAAM,EAAE,CAAC;IACxB,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IACjC,QAAQ,EAAE,QAAQ,CAAC;IACnB,SAAS,CAAC,EAAE,MAAM,CAAC;CACtB;AAED;;;;;;;;GAQG;AACH,wBAAgB,sBAAsB,CAClC,OAAO,EAAE,OAAO,EAChB,QAAQ,GAAE,eAAe,GAAG,OAAO,CAAC,mBAAmB,CAAM,EAC7D,gBAAgB,GAAE,UAAe,GAClC,UAAU,CAmBZ"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"memory-info.d.ts","sourceRoot":"","sources":["../../src/internals/memory-info.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"memory-info.d.ts","sourceRoot":"","sources":["../../src/internals/memory-info.ts"],"names":[],"mappings":"AAwBA;;GAEG;AACH,MAAM,WAAW,UAAU;IACvB,wDAAwD;IACxD,UAAU,EAAE,MAAM,CAAC;IAEnB,uDAAuD;IACvD,SAAS,EAAE,MAAM,CAAC;IAElB,uDAAuD;IACvD,SAAS,EAAE,MAAM,CAAC;IAElB,wDAAwD;IACxD,gBAAgB,EAAE,MAAM,CAAC;IAEzB,8EAA8E;IAC9E,mBAAmB,EAAE,MAAM,CAAC;CAC/B;AAED;;;;;;;;GAQG;AACH,wBAAsB,aAAa,IAAI,OAAO,CAAC,UAAU,CAAC,CA+GzD"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"memory-info.js","sourceRoot":"","sources":["../../src/internals/memory-info.ts"],"names":[],"mappings":";;
|
|
1
|
+
{"version":3,"file":"memory-info.js","sourceRoot":"","sources":["../../src/internals/memory-info.ts"],"names":[],"mappings":";;AAqDA,sCA+GC;;AApKD,2DAA8C;AAC9C,+CAAoD;AACpD,qCAA4C;AAC5C,kEAA6B;AAI7B,6DAA6B;AAC7B,6DAA6D;AAC7D,qEAAoC;AAEpC,uCAAqC;AAErC,MAAM,iBAAiB,GAAG;IACtB,KAAK,EAAE;QACH,EAAE,EAAE,6CAA6C;QACjD,EAAE,EAAE,2BAA2B;KAClC;IACD,IAAI,EAAE;QACF,EAAE,EAAE,6CAA6C;QACjD,EAAE,EAAE,+BAA+B;KACtC;CACJ,CAAC;AAsBF;;;;;;;;GAQG;AACI,KAAK,UAAU,aAAa;IAC/B,MAAM,cAAc,GAAG,mBAAI,CAAC,SAAS,CAAC,iBAAM,CAAC,CAAC;IAE9C,2DAA2D;IAC3D,oCAAoC;IACpC,MAAM,mBAAmB,GAAG,OAAO,CAAC,QAAQ,KAAK,OAAO,IAAI,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,+BAA+B,CAAC;IAE1G,MAAM,WAAW,GAAG,CAAC,mBAAmB,IAAI,CAAC,MAAM,IAAA,kBAAQ,GAAE,CAAC,CAAC;IAE/D,IAAI,gBAAgB,GAAG,CAAC,CAAC,CAAC;IAC1B,IAAI,mBAAmB,GAAG,CAAC,CAAC;IAE5B,IAAI,mBAAmB,EAAE,CAAC;QACtB,oBAAoB;QACpB,gBAAgB,GAAG,OAAO,CAAC,WAAW,EAAE,CAAC,GAAG,CAAC;QAE7C,8CAA8C;QAC9C,MAAM,OAAO,GAAG,IAAA,6BAAQ,EAAC,mBAAmB,CAAC,CAAC,QAAQ,EAAE,CAAC;QACzD,MAAM,MAAM,GAAG,OAAO,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC;QACnE,uFAAuF;QACvF,qGAAqG;QACrG,mBAAmB,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,GAAG,IAAI,GAAG,gBAAgB,CAAC;IAChE,CAAC;SAAM,CAAC;QACJ,sCAAsC;QACtC,MAAM,SAAS,GAAG,MAAM,cAAc,CAAC,OAAO,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;QAE1D,SAAS,CAAC,OAAO,CAAC,CAAC,GAAuB,EAAE,EAAE;YAC1C,0EAA0E;YAC1E,IAAI,GAAG,CAAC,OAAO,KAAK,IAAI,IAAI,GAAG,CAAC,OAAO,KAAK,UAAU,EAAE,CAAC;gBACrD,OAAO;YACX,CAAC;YACD,MAAM,KAAK,GAAG,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;YACpC,yCAAyC;YACzC,IAAI,GAAG,CAAC,GAAG,KAAK,GAAG,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC;gBAC/B,gBAAgB,GAAG,KAAK,CAAC;gBACzB,OAAO;YACX,CAAC;YACD,mBAAmB,IAAI,KAAK,CAAC;QACjC,CAAC,CAAC,CAAC;IACP,CAAC;IAED,IAAI,UAAkB,CAAC;IACvB,IAAI,SAAiB,CAAC;IACtB,IAAI,SAAiB,CAAC;IAEtB,IAAI,mBAAmB,EAAE,CAAC;QACtB,sCAAsC;QACtC,UAAU,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,+BAAgC,EAAE,EAAE,CAAC,GAAG,OAAO,CAAC;QAClF,SAAS,GAAG,gBAAgB,GAAG,mBAAmB,CAAC;QACnD,SAAS,GAAG,UAAU,GAAG,SAAS,CAAC;QAEnC,aAAG,CAAC,KAAK,CAAC,kBAAkB,UAAU,SAAS,SAAS,aAAa,CAAC,CAAC;IAC3E,CAAC;SAAM,IAAI,WAAW,EAAE,CAAC;QACrB,oEAAoE;QAEpE,yCAAyC;QACzC,IAAI,cAAc,GAAyC,IAAI,CAAC;QAChE,IAAI,CAAC;YACD,uEAAuE;YACvE,MAAM,IAAA,iBAAM,EAAC,wBAAwB,CAAC,CAAC;QAC3C,CAAC;QAAC,MAAM,CAAC;YACL,cAAc,GAAG,IAAI,CAAC;QAC1B,CAAC;QAED,IAAI,CAAC;YACD,IAAI,CAAC,aAAa,EAAE,YAAY,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;gBAClD,IAAA,mBAAQ,EAAC,iBAAiB,CAAC,KAAK,CAAC,cAAc,CAAC,EAAE,MAAM,CAAC;gBACzD,IAAA,mBAAQ,EAAC,iBAAiB,CAAC,IAAI,CAAC,cAAc,CAAC,EAAE,MAAM,CAAC;aAC3D,CAAC,CAAC;YAEH,gHAAgH;YAChH,aAAa,GAAG,aAAa,CAAC,OAAO,CAAC,gBAAgB,EAAE,EAAE,CAAC,CAAC;YAC5D,YAAY,GAAG,YAAY,CAAC,OAAO,CAAC,gBAAgB,EAAE,EAAE,CAAC,CAAC;YAE1D,4DAA4D;YAC5D,oIAAoI;YACpI,IAAI,aAAa,KAAK,KAAK,EAAE,CAAC;gBAC1B,UAAU,GAAG,IAAA,kBAAQ,GAAE,CAAC;gBACxB,yFAAyF;gBACzF,8CAA8C;YAClD,CAAC;iBAAM,CAAC;gBACJ,UAAU,GAAG,QAAQ,CAAC,aAAa,EAAE,EAAE,CAAC,CAAC;gBACzC,MAAM,gCAAgC,GAAG,UAAU,GAAG,MAAM,CAAC,gBAAgB,CAAC;gBAC9E,IAAI,gCAAgC;oBAAE,UAAU,GAAG,IAAA,kBAAQ,GAAE,CAAC;YAClE,CAAC;YACD,SAAS,GAAG,QAAQ,CAAC,YAAY,EAAE,EAAE,CAAC,CAAC;YACvC,SAAS,GAAG,UAAU,GAAG,SAAS,CAAC;QACvC,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACX,0CAA0C;YAC1C,aAAG,CAAC,UAAU,CACV,+EAA+E;gBAC3E,mGAAmG;gBACnG,UAAW,GAAa,CAAC,OAAO,EAAE,CACzC,CAAC;YACF,UAAU,GAAG,IAAA,kBAAQ,GAAE,CAAC;YACxB,SAAS,GAAG,IAAA,iBAAO,GAAE,CAAC;YACtB,SAAS,GAAG,UAAU,GAAG,SAAS,CAAC;QACvC,CAAC;IACL,CAAC;SAAM,CAAC;QACJ,UAAU,GAAG,IAAA,kBAAQ,GAAE,CAAC;QACxB,SAAS,GAAG,IAAA,iBAAO,GAAE,CAAC;QACtB,SAAS,GAAG,UAAU,GAAG,SAAS,CAAC;IACvC,CAAC;IAED,OAAO;QACH,UAAU;QACV,SAAS;QACT,SAAS;QACT,gBAAgB;QAChB,mBAAmB;KACtB,CAAC;AACN,CAAC"}
|
package/internals/robots.d.ts
CHANGED
|
@@ -5,7 +5,7 @@ import { Sitemap } from './sitemap';
|
|
|
5
5
|
* **Example usage:**
|
|
6
6
|
* ```javascript
|
|
7
7
|
* // Load the robots.txt file
|
|
8
|
-
* const robots = await
|
|
8
|
+
* const robots = await RobotsTxtFile.find('https://crawlee.dev/js/docs/introduction/first-crawler');
|
|
9
9
|
*
|
|
10
10
|
* // Check if a URL should be crawled according to robots.txt
|
|
11
11
|
* const url = 'https://crawlee.dev/api/puppeteer-crawler/class/PuppeteerCrawler';
|
|
@@ -17,7 +17,7 @@ import { Sitemap } from './sitemap';
|
|
|
17
17
|
* await crawler.addRequests(await robots.parseUrlsFromSitemaps());
|
|
18
18
|
* ```
|
|
19
19
|
*/
|
|
20
|
-
export declare class
|
|
20
|
+
export declare class RobotsTxtFile {
|
|
21
21
|
private robots;
|
|
22
22
|
private proxyUrl?;
|
|
23
23
|
private constructor();
|
|
@@ -26,15 +26,15 @@ export declare class RobotsFile {
|
|
|
26
26
|
* @param url the URL to fetch robots.txt for
|
|
27
27
|
* @param [proxyUrl] a proxy to be used for fetching the robots.txt file
|
|
28
28
|
*/
|
|
29
|
-
static find(url: string, proxyUrl?: string): Promise<
|
|
29
|
+
static find(url: string, proxyUrl?: string): Promise<RobotsTxtFile>;
|
|
30
30
|
/**
|
|
31
31
|
* Allows providing the URL and robots.txt content explicitly instead of loading it from the target site.
|
|
32
32
|
* @param url the URL for robots.txt file
|
|
33
33
|
* @param content contents of robots.txt
|
|
34
34
|
* @param [proxyUrl] a proxy to be used for fetching the robots.txt file
|
|
35
35
|
*/
|
|
36
|
-
static from(url: string, content: string, proxyUrl?: string):
|
|
37
|
-
protected static load(url: string, proxyUrl?: string): Promise<
|
|
36
|
+
static from(url: string, content: string, proxyUrl?: string): RobotsTxtFile;
|
|
37
|
+
protected static load(url: string, proxyUrl?: string): Promise<RobotsTxtFile>;
|
|
38
38
|
/**
|
|
39
39
|
* Check if a URL should be crawled by robots.
|
|
40
40
|
* @param url the URL to check against the rules in robots.txt
|
|
@@ -54,4 +54,5 @@ export declare class RobotsFile {
|
|
|
54
54
|
*/
|
|
55
55
|
parseUrlsFromSitemaps(): Promise<string[]>;
|
|
56
56
|
}
|
|
57
|
+
export { RobotsTxtFile as RobotsFile };
|
|
57
58
|
//# sourceMappingURL=robots.d.ts.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"robots.d.ts","sourceRoot":"","sources":["../../src/internals/robots.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAIpC;;;;;;;;;;;;;;;;;GAiBG;AACH,qBAAa,
|
|
1
|
+
{"version":3,"file":"robots.d.ts","sourceRoot":"","sources":["../../src/internals/robots.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAIpC;;;;;;;;;;;;;;;;;GAiBG;AACH,qBAAa,aAAa;IAElB,OAAO,CAAC,MAAM;IACd,OAAO,CAAC,QAAQ,CAAC;IAFrB,OAAO;IAKP;;;;OAIG;WACU,IAAI,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,aAAa,CAAC;IAQzE;;;;;OAKG;IACH,MAAM,CAAC,IAAI,CAAC,GAAG,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,aAAa;qBAIpD,IAAI,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,aAAa,CAAC;IAgCnF;;;;OAIG;IACH,SAAS,CAAC,GAAG,EAAE,MAAM,EAAE,SAAS,SAAM,GAAG,OAAO;IAIhD;;OAEG;IACH,WAAW,IAAI,MAAM,EAAE;IAIvB;;OAEG;IACG,aAAa,IAAI,OAAO,CAAC,OAAO,CAAC;IAIvC;;OAEG;IACG,qBAAqB,IAAI,OAAO,CAAC,MAAM,EAAE,CAAC;CAGnD;AAGD,OAAO,EAAE,aAAa,IAAI,UAAU,EAAE,CAAC"}
|
package/internals/robots.js
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
"use strict";
|
|
2
2
|
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
-
exports.RobotsFile = void 0;
|
|
3
|
+
exports.RobotsFile = exports.RobotsTxtFile = void 0;
|
|
4
4
|
const tslib_1 = require("tslib");
|
|
5
5
|
const robots_parser_1 = tslib_1.__importDefault(require("robots-parser"));
|
|
6
6
|
const gotScraping_1 = require("./gotScraping");
|
|
@@ -12,7 +12,7 @@ let HTTPError;
|
|
|
12
12
|
* **Example usage:**
|
|
13
13
|
* ```javascript
|
|
14
14
|
* // Load the robots.txt file
|
|
15
|
-
* const robots = await
|
|
15
|
+
* const robots = await RobotsTxtFile.find('https://crawlee.dev/js/docs/introduction/first-crawler');
|
|
16
16
|
*
|
|
17
17
|
* // Check if a URL should be crawled according to robots.txt
|
|
18
18
|
* const url = 'https://crawlee.dev/api/puppeteer-crawler/class/PuppeteerCrawler';
|
|
@@ -24,7 +24,7 @@ let HTTPError;
|
|
|
24
24
|
* await crawler.addRequests(await robots.parseUrlsFromSitemaps());
|
|
25
25
|
* ```
|
|
26
26
|
*/
|
|
27
|
-
class
|
|
27
|
+
class RobotsTxtFile {
|
|
28
28
|
constructor(robots, proxyUrl) {
|
|
29
29
|
Object.defineProperty(this, "robots", {
|
|
30
30
|
enumerable: true,
|
|
@@ -45,10 +45,10 @@ class RobotsFile {
|
|
|
45
45
|
* @param [proxyUrl] a proxy to be used for fetching the robots.txt file
|
|
46
46
|
*/
|
|
47
47
|
static async find(url, proxyUrl) {
|
|
48
|
-
const
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
return
|
|
48
|
+
const robotsTxtFileUrl = new URL(url);
|
|
49
|
+
robotsTxtFileUrl.pathname = '/robots.txt';
|
|
50
|
+
robotsTxtFileUrl.search = '';
|
|
51
|
+
return RobotsTxtFile.load(robotsTxtFileUrl.toString(), proxyUrl);
|
|
52
52
|
}
|
|
53
53
|
/**
|
|
54
54
|
* Allows providing the URL and robots.txt content explicitly instead of loading it from the target site.
|
|
@@ -57,7 +57,7 @@ class RobotsFile {
|
|
|
57
57
|
* @param [proxyUrl] a proxy to be used for fetching the robots.txt file
|
|
58
58
|
*/
|
|
59
59
|
static from(url, content, proxyUrl) {
|
|
60
|
-
return new
|
|
60
|
+
return new RobotsTxtFile((0, robots_parser_1.default)(url, content), proxyUrl);
|
|
61
61
|
}
|
|
62
62
|
static async load(url, proxyUrl) {
|
|
63
63
|
if (!HTTPError) {
|
|
@@ -70,11 +70,11 @@ class RobotsFile {
|
|
|
70
70
|
method: 'GET',
|
|
71
71
|
responseType: 'text',
|
|
72
72
|
});
|
|
73
|
-
return new
|
|
73
|
+
return new RobotsTxtFile((0, robots_parser_1.default)(url.toString(), response.body), proxyUrl);
|
|
74
74
|
}
|
|
75
75
|
catch (e) {
|
|
76
76
|
if (e instanceof HTTPError && e.response.statusCode === 404) {
|
|
77
|
-
return new
|
|
77
|
+
return new RobotsTxtFile({
|
|
78
78
|
isAllowed() {
|
|
79
79
|
return true;
|
|
80
80
|
},
|
|
@@ -113,5 +113,6 @@ class RobotsFile {
|
|
|
113
113
|
return (await this.parseSitemaps()).urls;
|
|
114
114
|
}
|
|
115
115
|
}
|
|
116
|
-
exports.
|
|
116
|
+
exports.RobotsTxtFile = RobotsTxtFile;
|
|
117
|
+
exports.RobotsFile = RobotsTxtFile;
|
|
117
118
|
//# sourceMappingURL=robots.js.map
|
package/internals/robots.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"robots.js","sourceRoot":"","sources":["../../src/internals/robots.ts"],"names":[],"mappings":";;;;AAGA,0EAAyC;AAEzC,+CAA4C;AAC5C,uCAAoC;AAEpC,IAAI,SAAgC,CAAC;AAErC;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAa,
|
|
1
|
+
{"version":3,"file":"robots.js","sourceRoot":"","sources":["../../src/internals/robots.ts"],"names":[],"mappings":";;;;AAGA,0EAAyC;AAEzC,+CAA4C;AAC5C,uCAAoC;AAEpC,IAAI,SAAgC,CAAC;AAErC;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAa,aAAa;IACtB,YACY,MAAgD,EAChD,QAAiB;QADzB;;;;mBAAQ,MAAM;WAA0C;QACxD;;;;mBAAQ,QAAQ;WAAS;IAC1B,CAAC;IAEJ;;;;OAIG;IACH,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,GAAW,EAAE,QAAiB;QAC5C,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;QACtC,gBAAgB,CAAC,QAAQ,GAAG,aAAa,CAAC;QAC1C,gBAAgB,CAAC,MAAM,GAAG,EAAE,CAAC;QAE7B,OAAO,aAAa,CAAC,IAAI,CAAC,gBAAgB,CAAC,QAAQ,EAAE,EAAE,QAAQ,CAAC,CAAC;IACrE,CAAC;IAED;;;;;OAKG;IACH,MAAM,CAAC,IAAI,CAAC,GAAW,EAAE,OAAe,EAAE,QAAiB;QACvD,OAAO,IAAI,aAAa,CAAC,IAAA,uBAAY,EAAC,GAAG,EAAE,OAAO,CAAC,EAAE,QAAQ,CAAC,CAAC;IACnE,CAAC;IAES,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,GAAW,EAAE,QAAiB;QACtD,IAAI,CAAC,SAAS,EAAE,CAAC;YACb,SAAS,GAAG,CAAC,MAAM,MAAM,CAAC,cAAc,CAAC,CAAC,CAAC,SAAS,CAAC;QACzD,CAAC;QAED,IAAI,CAAC;YACD,MAAM,QAAQ,GAAG,MAAM,IAAA,yBAAW,EAAC;gBAC/B,GAAG;gBACH,QAAQ;gBACR,MAAM,EAAE,KAAK;gBACb,YAAY,EAAE,MAAM;aACvB,CAAC,CAAC;YAEH,OAAO,IAAI,aAAa,CAAC,IAAA,uBAAY,EAAC,GAAG,CAAC,QAAQ,EAAE,EAAE,QAAQ,CAAC,IAAI,CAAC,EAAE,QAAQ,CAAC,CAAC;QACpF,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACT,IAAI,CAAC,YAAY,SAAS,IAAI,CAAC,CAAC,QAAQ,CAAC,UAAU,KAAK,GAAG,EAAE,CAAC;gBAC1D,OAAO,IAAI,aAAa,CACpB;oBACI,SAAS;wBACL,OAAO,IAAI,CAAC;oBAChB,CAAC;oBACD,WAAW;wBACP,OAAO,EAAE,CAAC;oBACd,CAAC;iBACJ,EACD,QAAQ,CACX,CAAC;YACN,CAAC;YACD,MAAM,CAAC,CAAC;QACZ,CAAC;IACL,CAAC;IAED;;;;OAIG;IACH,SAAS,CAAC,GAAW,EAAE,SAAS,GAAG,GAAG;QAClC,OAAO,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,GAAG,EAAE,SAAS,CAAC,IAAI,IAAI,CAAC,CAAC,+FAA+F;IACzJ,CAAC;IAED;;OAEG;IACH,WAAW;QACP,OAAO,IAAI,CAAC,MAAM,CAAC,WAAW,EAAE,CAAC;IACrC,CAAC;IAED;;OAEG;IACH,KAAK,CAAC,aAAa;QACf,OAAO,iBAAO,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,WAAW,EAAE,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;IAClE,CAAC;IAED;;OAEG;IACH,KAAK,CAAC,qBAAqB;QACvB,OAAO,CAAC,MAAM,IAAI,CAAC,aAAa,EAAE,CAAC,CAAC,IAAI,CAAC;IAC7C,CAAC;CACJ;AA1FD,sCA0FC;AAGyB,mCAAU"}
|
package/internals/sitemap.d.ts
CHANGED
|
@@ -58,7 +58,7 @@ export declare class Sitemap {
|
|
|
58
58
|
constructor(urls: string[]);
|
|
59
59
|
/**
|
|
60
60
|
* Try to load sitemap from the most common locations - `/sitemap.xml` and `/sitemap.txt`.
|
|
61
|
-
* For loading based on `Sitemap` entries in `robots.txt`, the {@apilink
|
|
61
|
+
* For loading based on `Sitemap` entries in `robots.txt`, the {@apilink RobotsTxtFile} class should be used.
|
|
62
62
|
* @param url The domain URL to fetch the sitemap for.
|
|
63
63
|
* @param proxyUrl A proxy to be used for fetching the sitemap file.
|
|
64
64
|
*/
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sitemap.d.ts","sourceRoot":"","sources":["../../src/internals/sitemap.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"sitemap.d.ts","sourceRoot":"","sources":["../../src/internals/sitemap.ts"],"names":[],"mappings":"AAOA,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,cAAc,CAAC;AAM3C,UAAU,cAAc;IACpB,GAAG,EAAE,MAAM,CAAC;IACZ,OAAO,CAAC,EAAE,IAAI,CAAC;IACf,UAAU,CAAC,EAAE,QAAQ,GAAG,QAAQ,GAAG,OAAO,GAAG,QAAQ,GAAG,SAAS,GAAG,QAAQ,GAAG,OAAO,CAAC;IACvF,QAAQ,CAAC,EAAE,MAAM,CAAC;CACrB;AAED,MAAM,MAAM,UAAU,GAAG,cAAc,GAAG;IACtC,gBAAgB,EAAE,MAAM,CAAC;CAC5B,CAAC;AAEF,UAAU,aAAa;IACnB,GAAG,EAAE,MAAM,CAAC;IACZ,gBAAgB,EAAE,IAAI,CAAC;CAC1B;AAED,KAAK,aAAa,GAAG,CAAC;IAAE,IAAI,EAAE,KAAK,CAAC;IAAC,GAAG,EAAE,MAAM,CAAA;CAAE,GAAG;IAAE,IAAI,EAAE,KAAK,CAAC;IAAC,OAAO,EAAE,MAAM,CAAA;CAAE,CAAC,GAAG;IAAE,KAAK,CAAC,EAAE,MAAM,CAAA;CAAE,CAAC;AA8I5G,MAAM,WAAW,mBAAmB;IAChC;;OAEG;IACH,kBAAkB,CAAC,EAAE,IAAI,GAAG,KAAK,CAAC;IAClC;;OAEG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB;;OAEG;IACH,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,CAAC;CAC5B;AAED,wBAAuB,YAAY,CAAC,CAAC,SAAS,mBAAmB,EAC7D,cAAc,EAAE,aAAa,EAAE,EAC/B,QAAQ,CAAC,EAAE,MAAM,EACjB,OAAO,CAAC,EAAE,CAAC,GACZ,aAAa,CAAC,CAAC,CAAC,oBAAoB,CAAC,SAAS,IAAI,GAAG,UAAU,GAAG,aAAa,GAAG,UAAU,CAAC,CAgJ/F;AAED;;;;;;;;;;;GAWG;AACH,qBAAa,OAAO;IACJ,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE;gBAAd,IAAI,EAAE,MAAM,EAAE;IAEnC;;;;;OAKG;WACU,cAAc,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC;IAe7E;;;;OAIG;WACU,IAAI,CACb,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,EACvB,QAAQ,CAAC,EAAE,MAAM,EACjB,mBAAmB,CAAC,EAAE,mBAAmB,GAC1C,OAAO,CAAC,OAAO,CAAC;IAQnB;;;;OAIG;WACU,aAAa,CAAC,OAAO,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC;qBAIzD,KAAK,CACxB,OAAO,EAAE,aAAa,EAAE,EACxB,QAAQ,CAAC,EAAE,MAAM,EACjB,mBAAmB,CAAC,EAAE,mBAAmB,GAC1C,OAAO,CAAC,OAAO,CAAC;CAatB"}
|
package/internals/sitemap.js
CHANGED
|
@@ -7,9 +7,9 @@ const node_crypto_1 = require("node:crypto");
|
|
|
7
7
|
const node_stream_1 = require("node:stream");
|
|
8
8
|
const node_string_decoder_1 = require("node:string_decoder");
|
|
9
9
|
const node_zlib_1 = require("node:zlib");
|
|
10
|
-
const log_1 = tslib_1.__importDefault(require("@apify/log"));
|
|
11
10
|
const sax_1 = tslib_1.__importDefault(require("sax"));
|
|
12
11
|
const whatwg_mimetype_1 = tslib_1.__importDefault(require("whatwg-mimetype"));
|
|
12
|
+
const log_1 = tslib_1.__importDefault(require("@apify/log"));
|
|
13
13
|
class SitemapTxtParser extends node_stream_1.Transform {
|
|
14
14
|
constructor() {
|
|
15
15
|
super({
|
|
@@ -171,7 +171,7 @@ async function* parseSitemap(initialSources, proxyUrl, options) {
|
|
|
171
171
|
try {
|
|
172
172
|
mimeType = new whatwg_mimetype_1.default(contentType);
|
|
173
173
|
}
|
|
174
|
-
catch
|
|
174
|
+
catch {
|
|
175
175
|
mimeType = null;
|
|
176
176
|
}
|
|
177
177
|
if (mimeType?.isXML() || url?.pathname.endsWith('.xml')) {
|
|
@@ -202,7 +202,7 @@ async function* parseSitemap(initialSources, proxyUrl, options) {
|
|
|
202
202
|
method: 'GET',
|
|
203
203
|
timeout: networkTimeouts,
|
|
204
204
|
headers: {
|
|
205
|
-
|
|
205
|
+
accept: 'text/plain, application/xhtml+xml, application/xml;q=0.9, */*;q=0.8',
|
|
206
206
|
},
|
|
207
207
|
});
|
|
208
208
|
request.on('response', () => resolve(request));
|
|
@@ -257,7 +257,6 @@ async function* parseSitemap(initialSources, proxyUrl, options) {
|
|
|
257
257
|
if (item.type === 'sitemapUrl' && !visitedSitemapUrls.has(item.url)) {
|
|
258
258
|
sources.push({ type: 'url', url: item.url, depth: (source.depth ?? 0) + 1 });
|
|
259
259
|
if (emitNestedSitemaps) {
|
|
260
|
-
// @ts-ignore
|
|
261
260
|
yield { loc: item.url, originSitemapUrl: null };
|
|
262
261
|
}
|
|
263
262
|
}
|
|
@@ -295,7 +294,7 @@ class Sitemap {
|
|
|
295
294
|
}
|
|
296
295
|
/**
|
|
297
296
|
* Try to load sitemap from the most common locations - `/sitemap.xml` and `/sitemap.txt`.
|
|
298
|
-
* For loading based on `Sitemap` entries in `robots.txt`, the {@apilink
|
|
297
|
+
* For loading based on `Sitemap` entries in `robots.txt`, the {@apilink RobotsTxtFile} class should be used.
|
|
299
298
|
* @param url The domain URL to fetch the sitemap for.
|
|
300
299
|
* @param proxyUrl A proxy to be used for fetching the sitemap file.
|
|
301
300
|
*/
|
|
@@ -332,7 +331,7 @@ class Sitemap {
|
|
|
332
331
|
urls.push(item.loc);
|
|
333
332
|
}
|
|
334
333
|
}
|
|
335
|
-
catch
|
|
334
|
+
catch {
|
|
336
335
|
return new Sitemap([]);
|
|
337
336
|
}
|
|
338
337
|
return new Sitemap(urls);
|
package/internals/sitemap.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sitemap.js","sourceRoot":"","sources":["../../src/internals/sitemap.ts"],"names":[],"mappings":";;;AA6LA,oCAqJC;;AAlVD,6CAAyC;AAEzC,6CAAyE;AACzE,6DAAoD;AACpD,yCAAyC;AAEzC,6DAA6B;AAG7B,sDAAsB;AACtB,8EAAuC;AAqBvC,MAAM,gBAAiB,SAAQ,uBAAS;IAIpC;QACI,KAAK,CAAC;YACF,kBAAkB,EAAE,IAAI;YACxB,SAAS,EAAE,CAAC,KAAK,EAAE,SAAS,EAAE,QAAQ,EAAE,EAAE;gBACtC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,KAAK,CAAC,CAAC;gBACrD,QAAQ,EAAE,CAAC;YACf,CAAC;YACD,KAAK,EAAE,CAAC,QAAQ,EAAE,EAAE;gBAChB,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,IAAI,CAAC,CAAC;gBAC7C,QAAQ,EAAE,CAAC;YACf,CAAC;SACJ,CAAC,CAAC;QAdC;;;;mBAAyB,IAAI,mCAAa,CAAC,MAAM,CAAC;WAAC;QACnD;;;;mBAAiB,EAAE;WAAC;IAc5B,CAAC;IAEO,aAAa,CAAC,KAAa,EAAE,QAAiB;QAClD,IAAI,CAAC,MAAM,IAAI,KAAK,CAAC;QAErB,IAAI,QAAQ,IAAI,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;YACzC,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM;iBACpB,KAAK,CAAC,IAAI,CAAC;iBACX,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;iBAC1B,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;YAEvC,IAAI,QAAQ,EAAE,CAAC;gBACX,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;oBACtB,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,EAAwB,CAAC,CAAC;gBAC/D,CAAC;gBAED,IAAI,CAAC,MAAM,GAAG,EAAE,CAAC;YACrB,CAAC;iBAAM,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC1B,KAAK,MAAM,GAAG,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBACnC,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,EAAwB,CAAC,CAAC;gBAC/D,CAAC;gBAED,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAE,CAAC;YAChC,CAAC;QACL,CAAC;IACL,CAAC;CACJ;AAED,MAAM,gBAAiB,SAAQ,uBAAS;IAQpC;QACI,KAAK,CAAC;YACF,kBAAkB,EAAE,IAAI;YACxB,SAAS,EAAE,CAAC,KAAK,EAAE,SAAS,EAAE,QAAQ,EAAE,EAAE;gBACtC,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC;gBAC7C,QAAQ,EAAE,CAAC;YACf,CAAC;YACD,KAAK,EAAE,CAAC,QAAQ,EAAE,EAAE;gBAChB,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;gBAChC,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;oBAClB,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;gBAC5B,CAAC;gBAED,IAAI,CAAC,MAAM,CAAC,GAAG,EAAE,CAAC;gBAClB,QAAQ,EAAE,CAAC;YACf,CAAC;SACJ,CAAC,CAAC;QAvBC;;;;mBAAyB,IAAI,mCAAa,CAAC,MAAM,CAAC;WAAC;QACnD;;;;mBAAS,IAAI,aAAG,CAAC,SAAS,CAAC,IAAI,CAAC;WAAC;QAEjC;;;;;WAAwC;QACxC;;;;mBAA6D,SAAS;WAAC;QACvE;;;;mBAA2B,EAAE;WAAC;QAoBlC,IAAI,CAAC,MAAM,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAClD,IAAI,CAAC,MAAM,CAAC,UAAU,GAAG,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAEpD,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC5C,IAAI,CAAC,MAAM,CAAC,OAAO,GAAG,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAE7C,IAAI,CAAC,MAAM,CAAC,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAClD,CAAC;IAEO,SAAS,CAAC,IAAgC;QAC9C,IAAI,IAAI,CAAC,WAAW,KAAK,SAAS,EAAE,CAAC;YACjC,IACI,IAAI,CAAC,IAAI,KAAK,KAAK;gBACnB,IAAI,CAAC,IAAI,KAAK,SAAS;gBACvB,IAAI,CAAC,IAAI,KAAK,UAAU;gBACxB,IAAI,CAAC,IAAI,KAAK,YAAY,EAC5B,CAAC;gBACC,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,IAAI,CAAC;YAChC,CAAC;QACL,CAAC;QACD,IAAI,IAAI,CAAC,IAAI,KAAK,QAAQ,EAAE,CAAC;YACzB,IAAI,CAAC,WAAW,GAAG,QAAQ,CAAC;QAChC,CAAC;QACD,IAAI,IAAI,CAAC,IAAI,KAAK,cAAc,EAAE,CAAC;YAC/B,IAAI,CAAC,WAAW,GAAG,cAAc,CAAC;QACtC,CAAC;IACL,CAAC;IAEO,UAAU,CAAC,IAAY;QAC3B,IAAI,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,UAAU,IAAI,IAAI,KAAK,YAAY,EAAE,CAAC;YACvF,IAAI,CAAC,UAAU,GAAG,SAAS,CAAC;QAChC,CAAC;QAED,IAAI,IAAI,KAAK,KAAK,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,KAAK,SAAS,EAAE,CAAC;YAC/C,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,EAAwB,CAAC,CAAC;YACjF,IAAI,CAAC,GAAG,GAAG,EAAE,CAAC;QAClB,CAAC;IACL,CAAC;IAEO,MAAM,CAAC,IAAY;QACvB,IAAI,IAAI,CAAC,UAAU,KAAK,KAAK,EAAE,CAAC;YAC5B,IAAI,IAAI,CAAC,WAAW,KAAK,cAAc,EAAE,CAAC;gBACtC,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,EAAE,IAAI,CAAC,IAAI,EAAE,EAAwB,CAAC,CAAC;YAC9E,CAAC;YAED,IAAI,IAAI,CAAC,WAAW,KAAK,QAAQ,EAAE,CAAC;gBAChC,IAAI,CAAC,GAAG,KAAR,IAAI,CAAC,GAAG,GAAK,EAAE,EAAC;gBAChB,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;YAC/B,CAAC;QACL,CAAC;QAED,IAAI,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;QAEnB,IAAI,IAAI,CAAC,UAAU,KAAK,SAAS,EAAE,CAAC;YAChC,IAAI,CAAC,GAAG,CAAC,OAAO,GAAG,IAAI,IAAI,CAAC,IAAI,CAAC,CAAC;QACtC,CAAC;QAED,IAAI,IAAI,CAAC,UAAU,KAAK,UAAU,EAAE,CAAC;YACjC,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC;QACrC,CAAC;QAED,IAAI,IAAI,CAAC,UAAU,KAAK,YAAY,EAAE,CAAC;YACnC,IAAI,CAAC,QAAQ,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,QAAQ,EAAE,OAAO,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;gBACvF,IAAI,CAAC,GAAG,CAAC,UAAU,GAAG,IAAgC,CAAC;YAC3D,CAAC;QACL,CAAC;IACL,CAAC;CACJ;AAqBM,KAAK,SAAS,CAAC,CAAC,YAAY,CAC/B,cAA+B,EAC/B,QAAiB,EACjB,OAAW;IAEX,MAAM,EAAE,WAAW,EAAE,GAAG,MAAM,MAAM,CAAC,cAAc,CAAC,CAAC;IACrD,MAAM,EAAE,cAAc,EAAE,GAAG,MAAM,MAAM,CAAC,WAAW,CAAC,CAAC;IACrD,MAAM,EAAE,kBAAkB,GAAG,KAAK,EAAE,QAAQ,GAAG,QAAQ,EAAE,cAAc,GAAG,CAAC,EAAE,eAAe,EAAE,GAAG,OAAO,IAAI,EAAE,CAAC;IAE/G,MAAM,OAAO,GAAG,CAAC,GAAG,cAAc,CAAC,CAAC;IACpC,MAAM,kBAAkB,GAAG,IAAI,GAAG,EAAU,CAAC;IAE7C,MAAM,YAAY,GAAG,CAAC,cAAsB,EAAE,EAAE,GAAS,EAAU,EAAE;QACjE,IAAI,QAAyB,CAAC;QAE9B,IAAI,CAAC;YACD,QAAQ,GAAG,IAAI,yBAAQ,CAAC,WAAW,CAAC,CAAC;QACzC,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACT,QAAQ,GAAG,IAAI,CAAC;QACpB,CAAC;QAED,IAAI,QAAQ,EAAE,KAAK,EAAE,IAAI,GAAG,EAAE,QAAQ,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;YACtD,OAAO,IAAI,gBAAgB,EAAE,CAAC;QAClC,CAAC;QAED,IAAI,QAAQ,EAAE,OAAO,KAAK,YAAY,IAAI,GAAG,EAAE,QAAQ,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;YACvE,OAAO,IAAI,gBAAgB,EAAE,CAAC;QAClC,CAAC;QAED,MAAM,IAAI,KAAK,CAAC,mDAAmD,WAAW,WAAW,GAAG,EAAE,QAAQ,EAAE,GAAG,CAAC,CAAC;IACjH,CAAC,CAAC;IAEF,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACxB,MAAM,MAAM,GAAG,OAAO,CAAC,KAAK,EAAG,CAAC;QAEhC,IAAI,CAAC,MAAM,EAAE,KAAK,IAAI,CAAC,CAAC,GAAG,QAAQ,EAAE,CAAC;YAClC,aAAG,CAAC,KAAK,CACL,oBAAoB,MAAM,CAAC,IAAI,KAAK,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,iCAAiC,QAAQ,GAAG,CAC1G,CAAC;YACF,SAAS;QACb,CAAC;QAED,IAAI,KAAK,GAAsC,IAAI,CAAC;QAEpD,IAAI,MAAM,CAAC,IAAI,KAAK,KAAK,EAAE,CAAC;YACxB,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YACvC,kBAAkB,CAAC,GAAG,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,CAAC;YAC9C,IAAI,WAAW,GAAG,cAAc,GAAG,CAAC,CAAC;YAErC,OAAO,WAAW,EAAE,GAAG,CAAC,EAAE,CAAC;gBACvB,IAAI,CAAC;oBACD,MAAM,aAAa,GAAG,MAAM,IAAI,OAAO,CACnC,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;wBAChB,MAAM,OAAO,GAAG,WAAW,CAAC,MAAM,CAAC;4BAC/B,GAAG,EAAE,UAAU;4BACf,QAAQ;4BACR,MAAM,EAAE,KAAK;4BACb,OAAO,EAAE,eAAe;4BACxB,OAAO,EAAE;gCACL,QAAQ,EAAE,qEAAqE;6BAClF;yBACJ,CAAC,CAAC;wBACH,OAAO,CAAC,EAAE,CAAC,UAAU,EAAE,GAAG,EAAE,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC;wBAC/C,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;oBAChC,CAAC,CACJ,CAAC;oBAEF,IAAI,KAAK,GAAiB,IAAI,CAAC;oBAE/B,IAAI,aAAa,CAAC,QAAS,CAAC,UAAU,IAAI,GAAG,IAAI,aAAa,CAAC,QAAS,CAAC,UAAU,GAAG,GAAG,EAAE,CAAC;wBACxF,IAAI,WAAW,GAAG,aAAa,CAAC,QAAS,CAAC,OAAO,CAAC,cAAc,CAAC,CAAC;wBAElE,MAAM,cAAc,GAAG,MAAM,cAAc,CAAC,aAAa,CAAC,CAAC;wBAC3D,IAAI,cAAc,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;4BACxC,WAAW,GAAG,cAAc,CAAC,QAAQ,CAAC,IAAI,CAAC;wBAC/C,CAAC;wBAED,IAAI,SAAS,GAAG,KAAK,CAAC;wBAEtB,IACI,WAAW,KAAK,SAAS;4BACrB,CAAC,CAAC,WAAW,KAAK,kBAAkB;4BACpC,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,EAC3C,CAAC;4BACC,SAAS,GAAG,IAAI,CAAC;4BAEjB,IAAI,UAAU,CAAC,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;gCACtC,UAAU,CAAC,QAAQ,GAAG,UAAU,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,EAAE,UAAU,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;4BAC3F,CAAC;wBACL,CAAC;wBAED,KAAK,GAAG,IAAA,sBAAQ,EACZ,cAAc,EACd,SAAS,CAAC,CAAC,CAAC,IAAA,wBAAY,GAAE,CAAC,CAAC,CAAC,IAAI,yBAAW,EAAE,EAC9C,YAAY,CAAC,WAAW,EAAE,UAAU,CAAC,EACrC,CAAC,CAAC,EAAE,EAAE;4BACF,IAAI,CAAC,KAAK,SAAS,EAAE,CAAC;gCAClB,KAAK,GAAG,CAAC,CAAC;4BACd,CAAC;wBACL,CAAC,CACJ,CAAC;oBACN,CAAC;yBAAM,CAAC;wBACJ,KAAK,GAAG,IAAI,KAAK,CACb,4BAA4B,UAAU,kBAAkB,aAAa,CAAC,QAAS,CAAC,UAAU,EAAE,CAC/F,CAAC;oBACN,CAAC;oBAED,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;wBACjB,MAAM,KAAK,CAAC;oBAChB,CAAC;oBACD,MAAM;gBACV,CAAC;gBAAC,OAAO,CAAC,EAAE,CAAC;oBACT,aAAG,CAAC,OAAO,CACP,8BAA8B,UAAU,KAAK,WAAW,KAAK,CAAC,CAAC,CAAC,CAAC,kBAAkB,CAAC,CAAC,CAAC,aAAa,KAAK,CAAC,GAAG,CAC/G,CAAC;gBACN,CAAC;YACL,CAAC;QACL,CAAC;aAAM,IAAI,MAAM,CAAC,IAAI,KAAK,KAAK,EAAE,CAAC;YAC/B,KAAK,GAAG,IAAA,sBAAQ,EAAC,sBAAQ,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,EAAE,YAAY,CAAC,UAAU,CAAC,EAAE,CAAC,KAAK,EAAE,EAAE;gBAClF,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;oBACtB,aAAG,CAAC,OAAO,CAAC,8BAA8B,KAAK,EAAE,CAAC,CAAC;gBACvD,CAAC;YACL,CAAC,CAAC,CAAC;QACP,CAAC;QAED,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;YACjB,SAAS;QACb,CAAC;QAED,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YAC7B,IAAI,IAAI,CAAC,IAAI,KAAK,YAAY,IAAI,CAAC,kBAAkB,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC;gBAClE,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,EAAE,KAAK,EAAE,CAAC,MAAM,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;gBAC7E,IAAI,kBAAkB,EAAE,CAAC;oBACrB,aAAa;oBACb,MAAM,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,EAAE,gBAAgB,EAAE,IAAI,EAAE,CAAC;gBACpD,CAAC;YACL,CAAC;YAED,IAAI,IAAI,CAAC,IAAI,KAAK,KAAK,EAAE,CAAC;gBACtB,MAAM;oBACF,GAAG,IAAI;oBACP,gBAAgB,EACZ,MAAM,CAAC,IAAI,KAAK,KAAK;wBACjB,CAAC,CAAC,MAAM,CAAC,GAAG;wBACZ,CAAC,CAAC,SAAS,IAAA,wBAAU,EAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE;iBACpF,CAAC;YACN,CAAC;QACL,CAAC;IACL,CAAC;AACL,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAa,OAAO;IAChB,YAAqB,IAAc;QAAvB;;;;mBAAS,IAAI;WAAU;IAAG,CAAC;IAEvC;;;;;OAKG;IACH,MAAM,CAAC,KAAK,CAAC,cAAc,CAAC,GAAW,EAAE,QAAiB;QACtD,MAAM,WAAW,GAAa,EAAE,CAAC;QAEjC,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;QAChC,UAAU,CAAC,MAAM,GAAG,EAAE,CAAC;QAEvB,UAAU,CAAC,QAAQ,GAAG,cAAc,CAAC;QACrC,WAAW,CAAC,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,CAAC;QAExC,UAAU,CAAC,QAAQ,GAAG,cAAc,CAAC;QACrC,WAAW,CAAC,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,CAAC;QAExC,OAAO,OAAO,CAAC,IAAI,CAAC,WAAW,EAAE,QAAQ,CAAC,CAAC;IAC/C,CAAC;IAED;;;;OAIG;IACH,MAAM,CAAC,KAAK,CAAC,IAAI,CACb,IAAuB,EACvB,QAAiB,EACjB,mBAAyC;QAEzC,OAAO,MAAM,IAAI,CAAC,KAAK,CACnB,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,CAAC,CAAC,EAC1E,QAAQ,EACR,mBAAmB,CACtB,CAAC;IACN,CAAC;IAED;;;;OAIG;IACH,MAAM,CAAC,KAAK,CAAC,aAAa,CAAC,OAAe,EAAE,QAAiB;QACzD,OAAO,MAAM,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC,EAAE,QAAQ,CAAC,CAAC;IAClE,CAAC;IAES,MAAM,CAAC,KAAK,CAAC,KAAK,CACxB,OAAwB,EACxB,QAAiB,EACjB,mBAAyC;QAEzC,MAAM,IAAI,GAAa,EAAE,CAAC;QAE1B,IAAI,CAAC;YACD,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,YAAY,CAAC,OAAO,EAAE,QAAQ,EAAE,mBAAmB,CAAC,EAAE,CAAC;gBAC5E,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACxB,CAAC;QACL,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACT,OAAO,IAAI,OAAO,CAAC,EAAE,CAAC,CAAC;QAC3B,CAAC;QAED,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC;IAC7B,CAAC;CACJ;AAnED,0BAmEC"}
|
|
1
|
+
{"version":3,"file":"sitemap.js","sourceRoot":"","sources":["../../src/internals/sitemap.ts"],"names":[],"mappings":";;;AA8LA,oCAoJC;;AAlVD,6CAAyC;AAEzC,6CAAyE;AACzE,6DAAoD;AACpD,yCAAyC;AAIzC,sDAAsB;AACtB,8EAAuC;AAEvC,6DAA6B;AAqB7B,MAAM,gBAAiB,SAAQ,uBAAS;IAIpC;QACI,KAAK,CAAC;YACF,kBAAkB,EAAE,IAAI;YACxB,SAAS,EAAE,CAAC,KAAK,EAAE,SAAS,EAAE,QAAQ,EAAE,EAAE;gBACtC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,KAAK,CAAC,CAAC;gBACrD,QAAQ,EAAE,CAAC;YACf,CAAC;YACD,KAAK,EAAE,CAAC,QAAQ,EAAE,EAAE;gBAChB,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,IAAI,CAAC,CAAC;gBAC7C,QAAQ,EAAE,CAAC;YACf,CAAC;SACJ,CAAC,CAAC;QAdC;;;;mBAAyB,IAAI,mCAAa,CAAC,MAAM,CAAC;WAAC;QACnD;;;;mBAAS,EAAE;WAAC;IAcpB,CAAC;IAEO,aAAa,CAAC,KAAa,EAAE,QAAiB;QAClD,IAAI,CAAC,MAAM,IAAI,KAAK,CAAC;QAErB,IAAI,QAAQ,IAAI,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;YACzC,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM;iBACpB,KAAK,CAAC,IAAI,CAAC;iBACX,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;iBAC1B,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;YAEvC,IAAI,QAAQ,EAAE,CAAC;gBACX,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;oBACtB,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,EAAwB,CAAC,CAAC;gBAC/D,CAAC;gBAED,IAAI,CAAC,MAAM,GAAG,EAAE,CAAC;YACrB,CAAC;iBAAM,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC1B,KAAK,MAAM,GAAG,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBACnC,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,EAAwB,CAAC,CAAC;gBAC/D,CAAC;gBAED,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAE,CAAC;YAChC,CAAC;QACL,CAAC;IACL,CAAC;CACJ;AAED,MAAM,gBAAiB,SAAQ,uBAAS;IAQpC;QACI,KAAK,CAAC;YACF,kBAAkB,EAAE,IAAI;YACxB,SAAS,EAAE,CAAC,KAAK,EAAE,SAAS,EAAE,QAAQ,EAAE,EAAE;gBACtC,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC;gBAC7C,QAAQ,EAAE,CAAC;YACf,CAAC;YACD,KAAK,EAAE,CAAC,QAAQ,EAAE,EAAE;gBAChB,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;gBAChC,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;oBAClB,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;gBAC5B,CAAC;gBAED,IAAI,CAAC,MAAM,CAAC,GAAG,EAAE,CAAC;gBAClB,QAAQ,EAAE,CAAC;YACf,CAAC;SACJ,CAAC,CAAC;QAvBC;;;;mBAAyB,IAAI,mCAAa,CAAC,MAAM,CAAC;WAAC;QACnD;;;;mBAAS,IAAI,aAAG,CAAC,SAAS,CAAC,IAAI,CAAC;WAAC;QAEjC;;;;;WAAwC;QACxC;;;;mBAA6D,SAAS;WAAC;QACvE;;;;mBAA2B,EAAE;WAAC;QAoBlC,IAAI,CAAC,MAAM,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAClD,IAAI,CAAC,MAAM,CAAC,UAAU,GAAG,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAEpD,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC5C,IAAI,CAAC,MAAM,CAAC,OAAO,GAAG,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAE7C,IAAI,CAAC,MAAM,CAAC,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAClD,CAAC;IAEO,SAAS,CAAC,IAAgC;QAC9C,IAAI,IAAI,CAAC,WAAW,KAAK,SAAS,EAAE,CAAC;YACjC,IACI,IAAI,CAAC,IAAI,KAAK,KAAK;gBACnB,IAAI,CAAC,IAAI,KAAK,SAAS;gBACvB,IAAI,CAAC,IAAI,KAAK,UAAU;gBACxB,IAAI,CAAC,IAAI,KAAK,YAAY,EAC5B,CAAC;gBACC,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,IAAI,CAAC;YAChC,CAAC;QACL,CAAC;QACD,IAAI,IAAI,CAAC,IAAI,KAAK,QAAQ,EAAE,CAAC;YACzB,IAAI,CAAC,WAAW,GAAG,QAAQ,CAAC;QAChC,CAAC;QACD,IAAI,IAAI,CAAC,IAAI,KAAK,cAAc,EAAE,CAAC;YAC/B,IAAI,CAAC,WAAW,GAAG,cAAc,CAAC;QACtC,CAAC;IACL,CAAC;IAEO,UAAU,CAAC,IAAY;QAC3B,IAAI,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,UAAU,IAAI,IAAI,KAAK,YAAY,EAAE,CAAC;YACvF,IAAI,CAAC,UAAU,GAAG,SAAS,CAAC;QAChC,CAAC;QAED,IAAI,IAAI,KAAK,KAAK,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,KAAK,SAAS,EAAE,CAAC;YAC/C,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,EAAwB,CAAC,CAAC;YACjF,IAAI,CAAC,GAAG,GAAG,EAAE,CAAC;QAClB,CAAC;IACL,CAAC;IAEO,MAAM,CAAC,IAAY;QACvB,IAAI,IAAI,CAAC,UAAU,KAAK,KAAK,EAAE,CAAC;YAC5B,IAAI,IAAI,CAAC,WAAW,KAAK,cAAc,EAAE,CAAC;gBACtC,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,YAAY,EAAE,GAAG,EAAE,IAAI,CAAC,IAAI,EAAE,EAAwB,CAAC,CAAC;YAC9E,CAAC;YAED,IAAI,IAAI,CAAC,WAAW,KAAK,QAAQ,EAAE,CAAC;gBAChC,IAAI,CAAC,GAAG,KAAR,IAAI,CAAC,GAAG,GAAK,EAAE,EAAC;gBAChB,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;YAC/B,CAAC;QACL,CAAC;QAED,IAAI,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;QAEnB,IAAI,IAAI,CAAC,UAAU,KAAK,SAAS,EAAE,CAAC;YAChC,IAAI,CAAC,GAAG,CAAC,OAAO,GAAG,IAAI,IAAI,CAAC,IAAI,CAAC,CAAC;QACtC,CAAC;QAED,IAAI,IAAI,CAAC,UAAU,KAAK,UAAU,EAAE,CAAC;YACjC,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC;QACrC,CAAC;QAED,IAAI,IAAI,CAAC,UAAU,KAAK,YAAY,EAAE,CAAC;YACnC,IAAI,CAAC,QAAQ,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,QAAQ,EAAE,OAAO,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;gBACvF,IAAI,CAAC,GAAG,CAAC,UAAU,GAAG,IAAgC,CAAC;YAC3D,CAAC;QACL,CAAC;IACL,CAAC;CACJ;AAqBM,KAAK,SAAS,CAAC,CAAC,YAAY,CAC/B,cAA+B,EAC/B,QAAiB,EACjB,OAAW;IAEX,MAAM,EAAE,WAAW,EAAE,GAAG,MAAM,MAAM,CAAC,cAAc,CAAC,CAAC;IACrD,MAAM,EAAE,cAAc,EAAE,GAAG,MAAM,MAAM,CAAC,WAAW,CAAC,CAAC;IACrD,MAAM,EAAE,kBAAkB,GAAG,KAAK,EAAE,QAAQ,GAAG,QAAQ,EAAE,cAAc,GAAG,CAAC,EAAE,eAAe,EAAE,GAAG,OAAO,IAAI,EAAE,CAAC;IAE/G,MAAM,OAAO,GAAG,CAAC,GAAG,cAAc,CAAC,CAAC;IACpC,MAAM,kBAAkB,GAAG,IAAI,GAAG,EAAU,CAAC;IAE7C,MAAM,YAAY,GAAG,CAAC,WAAW,GAAG,EAAE,EAAE,GAAS,EAAU,EAAE;QACzD,IAAI,QAAyB,CAAC;QAE9B,IAAI,CAAC;YACD,QAAQ,GAAG,IAAI,yBAAQ,CAAC,WAAW,CAAC,CAAC;QACzC,CAAC;QAAC,MAAM,CAAC;YACL,QAAQ,GAAG,IAAI,CAAC;QACpB,CAAC;QAED,IAAI,QAAQ,EAAE,KAAK,EAAE,IAAI,GAAG,EAAE,QAAQ,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;YACtD,OAAO,IAAI,gBAAgB,EAAE,CAAC;QAClC,CAAC;QAED,IAAI,QAAQ,EAAE,OAAO,KAAK,YAAY,IAAI,GAAG,EAAE,QAAQ,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;YACvE,OAAO,IAAI,gBAAgB,EAAE,CAAC;QAClC,CAAC;QAED,MAAM,IAAI,KAAK,CAAC,mDAAmD,WAAW,WAAW,GAAG,EAAE,QAAQ,EAAE,GAAG,CAAC,CAAC;IACjH,CAAC,CAAC;IAEF,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACxB,MAAM,MAAM,GAAG,OAAO,CAAC,KAAK,EAAG,CAAC;QAEhC,IAAI,CAAC,MAAM,EAAE,KAAK,IAAI,CAAC,CAAC,GAAG,QAAQ,EAAE,CAAC;YAClC,aAAG,CAAC,KAAK,CACL,oBAAoB,MAAM,CAAC,IAAI,KAAK,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,iCAAiC,QAAQ,GAAG,CAC1G,CAAC;YACF,SAAS;QACb,CAAC;QAED,IAAI,KAAK,GAAsC,IAAI,CAAC;QAEpD,IAAI,MAAM,CAAC,IAAI,KAAK,KAAK,EAAE,CAAC;YACxB,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YACvC,kBAAkB,CAAC,GAAG,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,CAAC;YAC9C,IAAI,WAAW,GAAG,cAAc,GAAG,CAAC,CAAC;YAErC,OAAO,WAAW,EAAE,GAAG,CAAC,EAAE,CAAC;gBACvB,IAAI,CAAC;oBACD,MAAM,aAAa,GAAG,MAAM,IAAI,OAAO,CACnC,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;wBAChB,MAAM,OAAO,GAAG,WAAW,CAAC,MAAM,CAAC;4BAC/B,GAAG,EAAE,UAAU;4BACf,QAAQ;4BACR,MAAM,EAAE,KAAK;4BACb,OAAO,EAAE,eAAe;4BACxB,OAAO,EAAE;gCACL,MAAM,EAAE,qEAAqE;6BAChF;yBACJ,CAAC,CAAC;wBACH,OAAO,CAAC,EAAE,CAAC,UAAU,EAAE,GAAG,EAAE,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC;wBAC/C,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;oBAChC,CAAC,CACJ,CAAC;oBAEF,IAAI,KAAK,GAAiB,IAAI,CAAC;oBAE/B,IAAI,aAAa,CAAC,QAAS,CAAC,UAAU,IAAI,GAAG,IAAI,aAAa,CAAC,QAAS,CAAC,UAAU,GAAG,GAAG,EAAE,CAAC;wBACxF,IAAI,WAAW,GAAG,aAAa,CAAC,QAAS,CAAC,OAAO,CAAC,cAAc,CAAC,CAAC;wBAElE,MAAM,cAAc,GAAG,MAAM,cAAc,CAAC,aAAa,CAAC,CAAC;wBAC3D,IAAI,cAAc,CAAC,QAAQ,KAAK,SAAS,EAAE,CAAC;4BACxC,WAAW,GAAG,cAAc,CAAC,QAAQ,CAAC,IAAI,CAAC;wBAC/C,CAAC;wBAED,IAAI,SAAS,GAAG,KAAK,CAAC;wBAEtB,IACI,WAAW,KAAK,SAAS;4BACrB,CAAC,CAAC,WAAW,KAAK,kBAAkB;4BACpC,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,EAC3C,CAAC;4BACC,SAAS,GAAG,IAAI,CAAC;4BAEjB,IAAI,UAAU,CAAC,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;gCACtC,UAAU,CAAC,QAAQ,GAAG,UAAU,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,EAAE,UAAU,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;4BAC3F,CAAC;wBACL,CAAC;wBAED,KAAK,GAAG,IAAA,sBAAQ,EACZ,cAAc,EACd,SAAS,CAAC,CAAC,CAAC,IAAA,wBAAY,GAAE,CAAC,CAAC,CAAC,IAAI,yBAAW,EAAE,EAC9C,YAAY,CAAC,WAAW,EAAE,UAAU,CAAC,EACrC,CAAC,CAAC,EAAE,EAAE;4BACF,IAAI,CAAC,KAAK,SAAS,EAAE,CAAC;gCAClB,KAAK,GAAG,CAAC,CAAC;4BACd,CAAC;wBACL,CAAC,CACJ,CAAC;oBACN,CAAC;yBAAM,CAAC;wBACJ,KAAK,GAAG,IAAI,KAAK,CACb,4BAA4B,UAAU,kBAAkB,aAAa,CAAC,QAAS,CAAC,UAAU,EAAE,CAC/F,CAAC;oBACN,CAAC;oBAED,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;wBACjB,MAAM,KAAK,CAAC;oBAChB,CAAC;oBACD,MAAM;gBACV,CAAC;gBAAC,OAAO,CAAC,EAAE,CAAC;oBACT,aAAG,CAAC,OAAO,CACP,8BAA8B,UAAU,KAAK,WAAW,KAAK,CAAC,CAAC,CAAC,CAAC,kBAAkB,CAAC,CAAC,CAAC,aAAa,KAAK,CAAC,GAAG,CAC/G,CAAC;gBACN,CAAC;YACL,CAAC;QACL,CAAC;aAAM,IAAI,MAAM,CAAC,IAAI,KAAK,KAAK,EAAE,CAAC;YAC/B,KAAK,GAAG,IAAA,sBAAQ,EAAC,sBAAQ,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,EAAE,YAAY,CAAC,UAAU,CAAC,EAAE,CAAC,KAAK,EAAE,EAAE;gBAClF,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;oBACtB,aAAG,CAAC,OAAO,CAAC,8BAA8B,KAAK,EAAE,CAAC,CAAC;gBACvD,CAAC;YACL,CAAC,CAAC,CAAC;QACP,CAAC;QAED,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;YACjB,SAAS;QACb,CAAC;QAED,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YAC7B,IAAI,IAAI,CAAC,IAAI,KAAK,YAAY,IAAI,CAAC,kBAAkB,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC;gBAClE,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,EAAE,KAAK,EAAE,CAAC,MAAM,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;gBAC7E,IAAI,kBAAkB,EAAE,CAAC;oBACrB,MAAM,EAAE,GAAG,EAAE,IAAI,CAAC,GAAG,EAAE,gBAAgB,EAAE,IAAI,EAAS,CAAC;gBAC3D,CAAC;YACL,CAAC;YAED,IAAI,IAAI,CAAC,IAAI,KAAK,KAAK,EAAE,CAAC;gBACtB,MAAM;oBACF,GAAG,IAAI;oBACP,gBAAgB,EACZ,MAAM,CAAC,IAAI,KAAK,KAAK;wBACjB,CAAC,CAAC,MAAM,CAAC,GAAG;wBACZ,CAAC,CAAC,SAAS,IAAA,wBAAU,EAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE;iBACpF,CAAC;YACN,CAAC;QACL,CAAC;IACL,CAAC;AACL,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAa,OAAO;IAChB,YAAqB,IAAc;QAAvB;;;;mBAAS,IAAI;WAAU;IAAG,CAAC;IAEvC;;;;;OAKG;IACH,MAAM,CAAC,KAAK,CAAC,cAAc,CAAC,GAAW,EAAE,QAAiB;QACtD,MAAM,WAAW,GAAa,EAAE,CAAC;QAEjC,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;QAChC,UAAU,CAAC,MAAM,GAAG,EAAE,CAAC;QAEvB,UAAU,CAAC,QAAQ,GAAG,cAAc,CAAC;QACrC,WAAW,CAAC,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,CAAC;QAExC,UAAU,CAAC,QAAQ,GAAG,cAAc,CAAC;QACrC,WAAW,CAAC,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,CAAC;QAExC,OAAO,OAAO,CAAC,IAAI,CAAC,WAAW,EAAE,QAAQ,CAAC,CAAC;IAC/C,CAAC;IAED;;;;OAIG;IACH,MAAM,CAAC,KAAK,CAAC,IAAI,CACb,IAAuB,EACvB,QAAiB,EACjB,mBAAyC;QAEzC,OAAO,MAAM,IAAI,CAAC,KAAK,CACnB,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,CAAC,CAAC,EAC1E,QAAQ,EACR,mBAAmB,CACtB,CAAC;IACN,CAAC;IAED;;;;OAIG;IACH,MAAM,CAAC,KAAK,CAAC,aAAa,CAAC,OAAe,EAAE,QAAiB;QACzD,OAAO,MAAM,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC,EAAE,QAAQ,CAAC,CAAC;IAClE,CAAC;IAES,MAAM,CAAC,KAAK,CAAC,KAAK,CACxB,OAAwB,EACxB,QAAiB,EACjB,mBAAyC;QAEzC,MAAM,IAAI,GAAa,EAAE,CAAC;QAE1B,IAAI,CAAC;YACD,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,YAAY,CAAC,OAAO,EAAE,QAAQ,EAAE,mBAAmB,CAAC,EAAE,CAAC;gBAC5E,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACxB,CAAC;QACL,CAAC;QAAC,MAAM,CAAC;YACL,OAAO,IAAI,OAAO,CAAC,EAAE,CAAC,CAAC;QAC3B,CAAC;QAED,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC;IAC7B,CAAC;CACJ;AAnED,0BAmEC"}
|
package/internals/social.d.ts
CHANGED
|
@@ -58,37 +58,6 @@ export interface SocialHandles {
|
|
|
58
58
|
pinterests: string[];
|
|
59
59
|
discords: string[];
|
|
60
60
|
}
|
|
61
|
-
/**
|
|
62
|
-
* The function attempts to extract emails, phone numbers and social profile URLs from a HTML document,
|
|
63
|
-
* specifically LinkedIn, Twitter, Instagram and Facebook profile URLs.
|
|
64
|
-
* The function removes duplicates from the resulting arrays and sorts the items alphabetically.
|
|
65
|
-
*
|
|
66
|
-
* Note that the `phones` field contains phone numbers extracted from the special phone links
|
|
67
|
-
* such as `[call us](tel:+1234556789)` (see {@apilink phonesFromUrls})
|
|
68
|
-
* and potentially other sources with high certainty, while `phonesUncertain` contains phone numbers
|
|
69
|
-
* extracted from the plain text, which might be very inaccurate.
|
|
70
|
-
*
|
|
71
|
-
* **Example usage:**
|
|
72
|
-
* ```typescript
|
|
73
|
-
* import { launchPuppeteer, social } from 'crawlee';
|
|
74
|
-
*
|
|
75
|
-
* const browser = await launchPuppeteer();
|
|
76
|
-
* const page = await browser.newPage();
|
|
77
|
-
* await page.goto('http://www.example.com');
|
|
78
|
-
* const html = await page.content();
|
|
79
|
-
*
|
|
80
|
-
* const result = social.parseHandlesFromHtml(html);
|
|
81
|
-
* console.log('Social handles:');
|
|
82
|
-
* console.dir(result);
|
|
83
|
-
* ```
|
|
84
|
-
*
|
|
85
|
-
* @param html HTML text
|
|
86
|
-
* @param [data] Optional object which will receive the `text` and `$` properties
|
|
87
|
-
* that contain text content of the HTML and `cheerio` object, respectively. This is an optimization
|
|
88
|
-
* so that the caller doesn't need to parse the HTML document again, if needed.
|
|
89
|
-
* @return An object with the social handles.
|
|
90
|
-
*/
|
|
91
|
-
export declare function parseHandlesFromHtml(html: string, data?: Record<string, unknown> | null): SocialHandles;
|
|
92
61
|
/**
|
|
93
62
|
* Regular expression to exactly match a single LinkedIn profile URL.
|
|
94
63
|
* It has the following form: `/^...$/i` and matches URLs such as:
|
|
@@ -489,4 +458,35 @@ export declare const DISCORD_REGEX: RegExp;
|
|
|
489
458
|
* ```
|
|
490
459
|
*/
|
|
491
460
|
export declare const DISCORD_REGEX_GLOBAL: RegExp;
|
|
461
|
+
/**
|
|
462
|
+
* The function attempts to extract emails, phone numbers and social profile URLs from a HTML document,
|
|
463
|
+
* specifically LinkedIn, Twitter, Instagram and Facebook profile URLs.
|
|
464
|
+
* The function removes duplicates from the resulting arrays and sorts the items alphabetically.
|
|
465
|
+
*
|
|
466
|
+
* Note that the `phones` field contains phone numbers extracted from the special phone links
|
|
467
|
+
* such as `[call us](tel:+1234556789)` (see {@apilink phonesFromUrls})
|
|
468
|
+
* and potentially other sources with high certainty, while `phonesUncertain` contains phone numbers
|
|
469
|
+
* extracted from the plain text, which might be very inaccurate.
|
|
470
|
+
*
|
|
471
|
+
* **Example usage:**
|
|
472
|
+
* ```typescript
|
|
473
|
+
* import { launchPuppeteer, social } from 'crawlee';
|
|
474
|
+
*
|
|
475
|
+
* const browser = await launchPuppeteer();
|
|
476
|
+
* const page = await browser.newPage();
|
|
477
|
+
* await page.goto('http://www.example.com');
|
|
478
|
+
* const html = await page.content();
|
|
479
|
+
*
|
|
480
|
+
* const result = social.parseHandlesFromHtml(html);
|
|
481
|
+
* console.log('Social handles:');
|
|
482
|
+
* console.dir(result);
|
|
483
|
+
* ```
|
|
484
|
+
*
|
|
485
|
+
* @param html HTML text
|
|
486
|
+
* @param [data] Optional object which will receive the `text` and `$` properties
|
|
487
|
+
* that contain text content of the HTML and `cheerio` object, respectively. This is an optimization
|
|
488
|
+
* so that the caller doesn't need to parse the HTML document again, if needed.
|
|
489
|
+
* @return An object with the social handles.
|
|
490
|
+
*/
|
|
491
|
+
export declare function parseHandlesFromHtml(html: string, data?: Record<string, unknown> | null): SocialHandles;
|
|
492
492
|
//# sourceMappingURL=social.d.ts.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"social.d.ts","sourceRoot":"","sources":["../../src/internals/social.ts"],"names":[],"mappings":"AAQA;;;GAGG;AACH,eAAO,MAAM,WAAW,QAA6C,CAAC;AAEtE;;;GAGG;AACH,eAAO,MAAM,kBAAkB,QAAuC,CAAC;AAIvE;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAGrD;AAED;;;;;;;GAOG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,EAAE,GAAG,MAAM,EAAE,CAavD;AA4DD;;;;;;;GAOG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAiBrD;AAED;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,EAAE,GAAG,MAAM,EAAE,CAYvD;AAqCD;;GAEG;AACH,MAAM,WAAW,aAAa;IAC1B,MAAM,EAAE,MAAM,EAAE,CAAC;IACjB,MAAM,EAAE,MAAM,EAAE,CAAC;IACjB,eAAe,EAAE,MAAM,EAAE,CAAC;IAC1B,SAAS,EAAE,MAAM,EAAE,CAAC;IACpB,QAAQ,EAAE,MAAM,EAAE,CAAC;IACnB,UAAU,EAAE,MAAM,EAAE,CAAC;IACrB,SAAS,EAAE,MAAM,EAAE,CAAC;IACpB,QAAQ,EAAE,MAAM,EAAE,CAAC;IACnB,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,UAAU,EAAE,MAAM,EAAE,CAAC;IACrB,QAAQ,EAAE,MAAM,EAAE,CAAC;CACtB;AAED
|
|
1
|
+
{"version":3,"file":"social.d.ts","sourceRoot":"","sources":["../../src/internals/social.ts"],"names":[],"mappings":"AAQA;;;GAGG;AACH,eAAO,MAAM,WAAW,QAA6C,CAAC;AAEtE;;;GAGG;AACH,eAAO,MAAM,kBAAkB,QAAuC,CAAC;AAIvE;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAGrD;AAED;;;;;;;GAOG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,EAAE,GAAG,MAAM,EAAE,CAavD;AA4DD;;;;;;;GAOG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAiBrD;AAED;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,EAAE,GAAG,MAAM,EAAE,CAYvD;AAqCD;;GAEG;AACH,MAAM,WAAW,aAAa;IAC1B,MAAM,EAAE,MAAM,EAAE,CAAC;IACjB,MAAM,EAAE,MAAM,EAAE,CAAC;IACjB,eAAe,EAAE,MAAM,EAAE,CAAC;IAC1B,SAAS,EAAE,MAAM,EAAE,CAAC;IACpB,QAAQ,EAAE,MAAM,EAAE,CAAC;IACnB,UAAU,EAAE,MAAM,EAAE,CAAC;IACrB,SAAS,EAAE,MAAM,EAAE,CAAC;IACpB,QAAQ,EAAE,MAAM,EAAE,CAAC;IACnB,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,UAAU,EAAE,MAAM,EAAE,CAAC;IACrB,QAAQ,EAAE,MAAM,EAAE,CAAC;CACtB;AAED;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AACH,eAAO,MAAM,cAAc,QAAgD,CAAC;AAE5E;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2BG;AACH,eAAO,MAAM,qBAAqB,QAA0C,CAAC;AAE7E;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AACH,eAAO,MAAM,eAAe,QAAiD,CAAC;AAE9E;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiCG;AACH,eAAO,MAAM,sBAAsB,QAA2C,CAAC;AAE/E;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,eAAO,MAAM,aAAa,QAA+C,CAAC;AAE1E;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AACH,eAAO,MAAM,oBAAoB,QAAyC,CAAC;AAE3E;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AACH,eAAO,MAAM,cAAc,QAAgD,CAAC;AAE5E;;;;;;;;;;;;;;;;;;;;;;;;;;GA0BG;AACH,eAAO,MAAM,qBAAqB,QAA0C,CAAC;AAE7E;;;;;;;;;;;;;;;;;;;;;GAqBG;AACH,eAAO,MAAM,aAAa,QAA+C,CAAC;AAE1E;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,eAAO,MAAM,oBAAoB,QAAyC,CAAC;AAC3E;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,eAAO,MAAM,YAAY,QAA8C,CAAC;AAExE;;;;;;;;;;;;;;;;;;;GAmBG;AACH,eAAO,MAAM,mBAAmB,QAAwC,CAAC;AAEzE;;;;;;;;;;;;;;;;;;;;;GAqBG;AACH,eAAO,MAAM,eAAe,QAAiD,CAAC;AAE9E;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,eAAO,MAAM,sBAAsB,QAA2C,CAAC;AAE/E;;;;;;;;;;;;;;;;;;;GAmBG;AACH,eAAO,MAAM,aAAa,QAA+C,CAAC;AAE1E;;;;;;;;;;;;;;;;;;GAkBG;AACH,eAAO,MAAM,oBAAoB,QAAyC,CAAC;AAE3E;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AACH,wBAAgB,oBAAoB,CAAC,IAAI,EAAE,MAAM,EAAE,IAAI,GAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAW,GAAG,aAAa,CAqD7G"}
|
package/internals/social.js
CHANGED
|
@@ -166,86 +166,6 @@ const YOUTUBE_REGEX_STRING = '(?<!\\w)(?:https?:\\/\\/)?(?:youtu\\.be\\/|(?:www\
|
|
|
166
166
|
const TIKTOK_REGEX_STRING = '(?<!\\w)(?:http(?:s)?:\\/\\/)?(?:(?:www|m)\\.)?(?:tiktok\\.com)\\/(((?:(?:v|embed|trending)(?:\\?shareId=|\\/))[0-9]{2,50}(?![0-9]))|(?:@)[a-z0-9\\-_\\.]+((?:\\/video\\/)[0-9]{2,50}(?![0-9]))?)(?:\\/)?';
|
|
167
167
|
const PINTEREST_REGEX_STRING = '(?<!\\w)(?:http(?:s)?:\\/\\/)?(?:(?:(?:(?:www\\.)?pinterest(?:\\.com|(?:\\.[a-z]{2}){1,2}))|(?:[a-z]{2})\\.pinterest\\.com)(?:\\/))((pin\\/[0-9]{2,50})|((?!pin)[a-z0-9\\-_\\.]+(\\/[a-z0-9\\-_\\.]+)?))(?:\\/)?';
|
|
168
168
|
const DISCORD_REGEX_STRING = '(?<!\\w)(?:https?:\\/\\/)?(?:www\\.)?((?:(?:(?:canary|ptb).)?(?:discord|discordapp)\\.com\\/channels(?:\\/)[0-9]{2,50}(\\/[0-9]{2,50})*)|(?:(?:(?:canary|ptb).)?(?:discord\\.(?:com|me|li|gg|io)|discordapp\\.com)(?:\\/invite)?)\\/(?!channels)[a-z0-9\\-_]{2,50})(?:\\/)?';
|
|
169
|
-
/**
|
|
170
|
-
* The function attempts to extract emails, phone numbers and social profile URLs from a HTML document,
|
|
171
|
-
* specifically LinkedIn, Twitter, Instagram and Facebook profile URLs.
|
|
172
|
-
* The function removes duplicates from the resulting arrays and sorts the items alphabetically.
|
|
173
|
-
*
|
|
174
|
-
* Note that the `phones` field contains phone numbers extracted from the special phone links
|
|
175
|
-
* such as `[call us](tel:+1234556789)` (see {@apilink phonesFromUrls})
|
|
176
|
-
* and potentially other sources with high certainty, while `phonesUncertain` contains phone numbers
|
|
177
|
-
* extracted from the plain text, which might be very inaccurate.
|
|
178
|
-
*
|
|
179
|
-
* **Example usage:**
|
|
180
|
-
* ```typescript
|
|
181
|
-
* import { launchPuppeteer, social } from 'crawlee';
|
|
182
|
-
*
|
|
183
|
-
* const browser = await launchPuppeteer();
|
|
184
|
-
* const page = await browser.newPage();
|
|
185
|
-
* await page.goto('http://www.example.com');
|
|
186
|
-
* const html = await page.content();
|
|
187
|
-
*
|
|
188
|
-
* const result = social.parseHandlesFromHtml(html);
|
|
189
|
-
* console.log('Social handles:');
|
|
190
|
-
* console.dir(result);
|
|
191
|
-
* ```
|
|
192
|
-
*
|
|
193
|
-
* @param html HTML text
|
|
194
|
-
* @param [data] Optional object which will receive the `text` and `$` properties
|
|
195
|
-
* that contain text content of the HTML and `cheerio` object, respectively. This is an optimization
|
|
196
|
-
* so that the caller doesn't need to parse the HTML document again, if needed.
|
|
197
|
-
* @return An object with the social handles.
|
|
198
|
-
*/
|
|
199
|
-
function parseHandlesFromHtml(html, data = null) {
|
|
200
|
-
const result = {
|
|
201
|
-
emails: [],
|
|
202
|
-
phones: [],
|
|
203
|
-
phonesUncertain: [],
|
|
204
|
-
linkedIns: [],
|
|
205
|
-
twitters: [],
|
|
206
|
-
instagrams: [],
|
|
207
|
-
facebooks: [],
|
|
208
|
-
youtubes: [],
|
|
209
|
-
tiktoks: [],
|
|
210
|
-
pinterests: [],
|
|
211
|
-
discords: [],
|
|
212
|
-
};
|
|
213
|
-
if (typeof html !== 'string')
|
|
214
|
-
return result;
|
|
215
|
-
const $ = cheerio.load(html, { decodeEntities: true });
|
|
216
|
-
if (data)
|
|
217
|
-
data.$ = $;
|
|
218
|
-
const text = (0, cheerio_1.htmlToText)($);
|
|
219
|
-
if (data)
|
|
220
|
-
data.text = text;
|
|
221
|
-
// Find all <a> links with href tag
|
|
222
|
-
const linkUrls = [];
|
|
223
|
-
$('a[href]').each((_index, elem) => {
|
|
224
|
-
if (elem)
|
|
225
|
-
linkUrls.push($(elem).attr('href'));
|
|
226
|
-
});
|
|
227
|
-
result.emails = emailsFromUrls(linkUrls).concat(emailsFromText(text));
|
|
228
|
-
result.phones = phonesFromUrls(linkUrls);
|
|
229
|
-
result.phonesUncertain = phonesFromText(text);
|
|
230
|
-
// Note that these regexps extract just the base profile path. For example for
|
|
231
|
-
// https://www.linkedin.com/in/carl-newman-123456a/detail/recent-activity/
|
|
232
|
-
// they match just:
|
|
233
|
-
// https://www.linkedin.com/in/carl-newman-123456a
|
|
234
|
-
result.linkedIns = html.match(exports.LINKEDIN_REGEX_GLOBAL) || [];
|
|
235
|
-
result.twitters = html.match(exports.TWITTER_REGEX_GLOBAL) || [];
|
|
236
|
-
result.instagrams = html.match(exports.INSTAGRAM_REGEX_GLOBAL) || [];
|
|
237
|
-
result.facebooks = html.match(exports.FACEBOOK_REGEX_GLOBAL) || [];
|
|
238
|
-
result.youtubes = html.match(exports.YOUTUBE_REGEX_GLOBAL) || [];
|
|
239
|
-
result.tiktoks = html.match(exports.TIKTOK_REGEX_GLOBAL) || [];
|
|
240
|
-
result.pinterests = html.match(exports.PINTEREST_REGEX_GLOBAL) || [];
|
|
241
|
-
result.discords = html.match(exports.DISCORD_REGEX_GLOBAL) || [];
|
|
242
|
-
// Sort and deduplicate handles
|
|
243
|
-
for (const key of Object.keys(result)) {
|
|
244
|
-
result[key].sort();
|
|
245
|
-
result[key] = [...new Set(result[key])].sort();
|
|
246
|
-
}
|
|
247
|
-
return result;
|
|
248
|
-
}
|
|
249
169
|
/**
|
|
250
170
|
* Regular expression to exactly match a single LinkedIn profile URL.
|
|
251
171
|
* It has the following form: `/^...$/i` and matches URLs such as:
|
|
@@ -646,4 +566,84 @@ exports.DISCORD_REGEX = new RegExp(`^${DISCORD_REGEX_STRING}$`, 'i');
|
|
|
646
566
|
* ```
|
|
647
567
|
*/
|
|
648
568
|
exports.DISCORD_REGEX_GLOBAL = new RegExp(DISCORD_REGEX_STRING, 'ig');
|
|
569
|
+
/**
|
|
570
|
+
* The function attempts to extract emails, phone numbers and social profile URLs from a HTML document,
|
|
571
|
+
* specifically LinkedIn, Twitter, Instagram and Facebook profile URLs.
|
|
572
|
+
* The function removes duplicates from the resulting arrays and sorts the items alphabetically.
|
|
573
|
+
*
|
|
574
|
+
* Note that the `phones` field contains phone numbers extracted from the special phone links
|
|
575
|
+
* such as `[call us](tel:+1234556789)` (see {@apilink phonesFromUrls})
|
|
576
|
+
* and potentially other sources with high certainty, while `phonesUncertain` contains phone numbers
|
|
577
|
+
* extracted from the plain text, which might be very inaccurate.
|
|
578
|
+
*
|
|
579
|
+
* **Example usage:**
|
|
580
|
+
* ```typescript
|
|
581
|
+
* import { launchPuppeteer, social } from 'crawlee';
|
|
582
|
+
*
|
|
583
|
+
* const browser = await launchPuppeteer();
|
|
584
|
+
* const page = await browser.newPage();
|
|
585
|
+
* await page.goto('http://www.example.com');
|
|
586
|
+
* const html = await page.content();
|
|
587
|
+
*
|
|
588
|
+
* const result = social.parseHandlesFromHtml(html);
|
|
589
|
+
* console.log('Social handles:');
|
|
590
|
+
* console.dir(result);
|
|
591
|
+
* ```
|
|
592
|
+
*
|
|
593
|
+
* @param html HTML text
|
|
594
|
+
* @param [data] Optional object which will receive the `text` and `$` properties
|
|
595
|
+
* that contain text content of the HTML and `cheerio` object, respectively. This is an optimization
|
|
596
|
+
* so that the caller doesn't need to parse the HTML document again, if needed.
|
|
597
|
+
* @return An object with the social handles.
|
|
598
|
+
*/
|
|
599
|
+
function parseHandlesFromHtml(html, data = null) {
|
|
600
|
+
const result = {
|
|
601
|
+
emails: [],
|
|
602
|
+
phones: [],
|
|
603
|
+
phonesUncertain: [],
|
|
604
|
+
linkedIns: [],
|
|
605
|
+
twitters: [],
|
|
606
|
+
instagrams: [],
|
|
607
|
+
facebooks: [],
|
|
608
|
+
youtubes: [],
|
|
609
|
+
tiktoks: [],
|
|
610
|
+
pinterests: [],
|
|
611
|
+
discords: [],
|
|
612
|
+
};
|
|
613
|
+
if (typeof html !== 'string')
|
|
614
|
+
return result;
|
|
615
|
+
const $ = cheerio.load(html, { decodeEntities: true });
|
|
616
|
+
if (data)
|
|
617
|
+
data.$ = $;
|
|
618
|
+
const text = (0, cheerio_1.htmlToText)($);
|
|
619
|
+
if (data)
|
|
620
|
+
data.text = text;
|
|
621
|
+
// Find all <a> links with href tag
|
|
622
|
+
const linkUrls = [];
|
|
623
|
+
$('a[href]').each((_index, elem) => {
|
|
624
|
+
if (elem)
|
|
625
|
+
linkUrls.push($(elem).attr('href'));
|
|
626
|
+
});
|
|
627
|
+
result.emails = emailsFromUrls(linkUrls).concat(emailsFromText(text));
|
|
628
|
+
result.phones = phonesFromUrls(linkUrls);
|
|
629
|
+
result.phonesUncertain = phonesFromText(text);
|
|
630
|
+
// Note that these regexps extract just the base profile path. For example for
|
|
631
|
+
// https://www.linkedin.com/in/carl-newman-123456a/detail/recent-activity/
|
|
632
|
+
// they match just:
|
|
633
|
+
// https://www.linkedin.com/in/carl-newman-123456a
|
|
634
|
+
result.linkedIns = html.match(exports.LINKEDIN_REGEX_GLOBAL) || [];
|
|
635
|
+
result.twitters = html.match(exports.TWITTER_REGEX_GLOBAL) || [];
|
|
636
|
+
result.instagrams = html.match(exports.INSTAGRAM_REGEX_GLOBAL) || [];
|
|
637
|
+
result.facebooks = html.match(exports.FACEBOOK_REGEX_GLOBAL) || [];
|
|
638
|
+
result.youtubes = html.match(exports.YOUTUBE_REGEX_GLOBAL) || [];
|
|
639
|
+
result.tiktoks = html.match(exports.TIKTOK_REGEX_GLOBAL) || [];
|
|
640
|
+
result.pinterests = html.match(exports.PINTEREST_REGEX_GLOBAL) || [];
|
|
641
|
+
result.discords = html.match(exports.DISCORD_REGEX_GLOBAL) || [];
|
|
642
|
+
// Sort and deduplicate handles
|
|
643
|
+
for (const key of Object.keys(result)) {
|
|
644
|
+
result[key].sort();
|
|
645
|
+
result[key] = [...new Set(result[key])].sort();
|
|
646
|
+
}
|
|
647
|
+
return result;
|
|
648
|
+
}
|
|
649
649
|
//# sourceMappingURL=social.js.map
|