@detiq/crawler 0.1.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (412) hide show
  1. package/Dockerfile +49 -0
  2. package/README.md +11 -0
  3. package/dist/accessibility-auditor.d.ts +29 -0
  4. package/dist/accessibility-auditor.js +57 -0
  5. package/dist/accessibility-auditor.js.map +7 -0
  6. package/dist/ai-login.d.ts +32 -0
  7. package/dist/ai-login.js +177 -0
  8. package/dist/ai-login.js.map +7 -0
  9. package/dist/auth-token-rotator.d.ts +47 -0
  10. package/dist/auth-token-rotator.js +58 -0
  11. package/dist/auth-token-rotator.js.map +7 -0
  12. package/dist/auth.d.ts +20 -0
  13. package/dist/auth.js +329 -0
  14. package/dist/auth.js.map +7 -0
  15. package/dist/browser-extension.d.ts +37 -0
  16. package/dist/browser-extension.js +113 -0
  17. package/dist/browser-extension.js.map +7 -0
  18. package/dist/captcha-solver.d.ts +16 -0
  19. package/dist/captcha-solver.js +105 -0
  20. package/dist/captcha-solver.js.map +7 -0
  21. package/dist/color-contrast.d.ts +30 -0
  22. package/dist/color-contrast.js +78 -0
  23. package/dist/color-contrast.js.map +7 -0
  24. package/dist/content-chunker.d.ts +18 -0
  25. package/dist/content-chunker.js +152 -0
  26. package/dist/content-chunker.js.map +7 -0
  27. package/dist/content-dedup.d.ts +23 -0
  28. package/dist/content-dedup.js +60 -0
  29. package/dist/content-dedup.js.map +7 -0
  30. package/dist/cookie-consent.d.ts +25 -0
  31. package/dist/cookie-consent.js +78 -0
  32. package/dist/cookie-consent.js.map +7 -0
  33. package/dist/cookie-jar.d.ts +20 -0
  34. package/dist/cookie-jar.js +61 -0
  35. package/dist/cookie-jar.js.map +7 -0
  36. package/dist/crawler.d.ts +179 -0
  37. package/dist/crawler.js +3672 -0
  38. package/dist/crawler.js.map +7 -0
  39. package/dist/custom-js-injector.d.ts +42 -0
  40. package/dist/custom-js-injector.js +50 -0
  41. package/dist/custom-js-injector.js.map +7 -0
  42. package/dist/dark-mode-capture.d.ts +23 -0
  43. package/dist/dark-mode-capture.js +75 -0
  44. package/dist/dark-mode-capture.js.map +7 -0
  45. package/dist/dead-link-detector.d.ts +37 -0
  46. package/dist/dead-link-detector.js +50 -0
  47. package/dist/dead-link-detector.js.map +7 -0
  48. package/dist/element-hash.d.ts +9 -0
  49. package/dist/element-hash.js +9 -0
  50. package/dist/element-hash.js.map +7 -0
  51. package/dist/email-otp.d.ts +42 -0
  52. package/dist/email-otp.js +177 -0
  53. package/dist/email-otp.js.map +7 -0
  54. package/dist/engine-router.d.ts +47 -0
  55. package/dist/engine-router.js +94 -0
  56. package/dist/engine-router.js.map +7 -0
  57. package/dist/export/bulk-export.d.ts +26 -0
  58. package/dist/export/bulk-export.js +76 -0
  59. package/dist/export/bulk-export.js.map +7 -0
  60. package/dist/export/vector-export.d.ts +76 -0
  61. package/dist/export/vector-export.js +85 -0
  62. package/dist/export/vector-export.js.map +7 -0
  63. package/dist/extractors/advanced.d.ts +54 -0
  64. package/dist/extractors/advanced.js +245 -0
  65. package/dist/extractors/advanced.js.map +7 -0
  66. package/dist/extractors/artifact-envelope.d.ts +63 -0
  67. package/dist/extractors/artifact-envelope.js +63 -0
  68. package/dist/extractors/artifact-envelope.js.map +7 -0
  69. package/dist/failure-analysis-agent.d.ts +16 -0
  70. package/dist/failure-analysis-agent.js +40 -0
  71. package/dist/failure-analysis-agent.js.map +7 -0
  72. package/dist/feature-flag-detector.d.ts +30 -0
  73. package/dist/feature-flag-detector.js +55 -0
  74. package/dist/feature-flag-detector.js.map +7 -0
  75. package/dist/form-detector.d.ts +34 -0
  76. package/dist/form-detector.js +58 -0
  77. package/dist/form-detector.js.map +7 -0
  78. package/dist/frontier.d.ts +32 -0
  79. package/dist/frontier.js +83 -0
  80. package/dist/frontier.js.map +7 -0
  81. package/dist/gap-pass.d.ts +35 -0
  82. package/dist/gap-pass.js +81 -0
  83. package/dist/gap-pass.js.map +7 -0
  84. package/dist/graphql-detector.d.ts +39 -0
  85. package/dist/graphql-detector.js +86 -0
  86. package/dist/graphql-detector.js.map +7 -0
  87. package/dist/http-engine.d.ts +80 -0
  88. package/dist/http-engine.js +304 -0
  89. package/dist/http-engine.js.map +7 -0
  90. package/dist/incremental-diff.d.ts +59 -0
  91. package/dist/incremental-diff.js +61 -0
  92. package/dist/incremental-diff.js.map +7 -0
  93. package/dist/index.d.ts +35 -0
  94. package/dist/index.js +61 -0
  95. package/dist/index.js.map +7 -0
  96. package/dist/input-fuzzer.d.ts +30 -0
  97. package/dist/input-fuzzer.js +105 -0
  98. package/dist/input-fuzzer.js.map +7 -0
  99. package/dist/interactive-probe.d.ts +104 -0
  100. package/dist/interactive-probe.js +442 -0
  101. package/dist/interactive-probe.js.map +7 -0
  102. package/dist/job-queue.d.ts +171 -0
  103. package/dist/job-queue.js +2256 -0
  104. package/dist/job-queue.js.map +7 -0
  105. package/dist/js-coverage.d.ts +29 -0
  106. package/dist/js-coverage.js +56 -0
  107. package/dist/js-coverage.js.map +7 -0
  108. package/dist/js-router-detector.d.ts +17 -0
  109. package/dist/js-router-detector.js +90 -0
  110. package/dist/js-router-detector.js.map +7 -0
  111. package/dist/language-detector.d.ts +22 -0
  112. package/dist/language-detector.js +57 -0
  113. package/dist/language-detector.js.map +7 -0
  114. package/dist/link-classifier.d.ts +23 -0
  115. package/dist/link-classifier.js +40 -0
  116. package/dist/link-classifier.js.map +7 -0
  117. package/dist/live-view.d.ts +36 -0
  118. package/dist/live-view.js +100 -0
  119. package/dist/live-view.js.map +7 -0
  120. package/dist/logger.d.ts +2 -0
  121. package/dist/logger.js +10 -0
  122. package/dist/logger.js.map +7 -0
  123. package/dist/markdown-extract.d.ts +8 -0
  124. package/dist/markdown-extract.js +54 -0
  125. package/dist/markdown-extract.js.map +7 -0
  126. package/dist/native-crawler.d.ts +69 -0
  127. package/dist/native-crawler.js +839 -0
  128. package/dist/native-crawler.js.map +7 -0
  129. package/dist/network-observer.d.ts +49 -0
  130. package/dist/network-observer.js +149 -0
  131. package/dist/network-observer.js.map +7 -0
  132. package/dist/network-throttle.d.ts +25 -0
  133. package/dist/network-throttle.js +34 -0
  134. package/dist/network-throttle.js.map +7 -0
  135. package/dist/openapi-generator.d.ts +49 -0
  136. package/dist/openapi-generator.js +89 -0
  137. package/dist/openapi-generator.js.map +7 -0
  138. package/dist/pdf-extractor.d.ts +28 -0
  139. package/dist/pdf-extractor.js +81 -0
  140. package/dist/pdf-extractor.js.map +7 -0
  141. package/dist/performance-metrics.d.ts +53 -0
  142. package/dist/performance-metrics.js +83 -0
  143. package/dist/performance-metrics.js.map +7 -0
  144. package/dist/phash.d.ts +38 -0
  145. package/dist/phash.js +115 -0
  146. package/dist/phash.js.map +7 -0
  147. package/dist/pii-scrubber.d.ts +21 -0
  148. package/dist/pii-scrubber.js +75 -0
  149. package/dist/pii-scrubber.js.map +7 -0
  150. package/dist/playwright-recorder.d.ts +39 -0
  151. package/dist/playwright-recorder.js +123 -0
  152. package/dist/playwright-recorder.js.map +7 -0
  153. package/dist/policies/crawl-safety.d.ts +9 -0
  154. package/dist/policies/crawl-safety.js +37 -0
  155. package/dist/policies/crawl-safety.js.map +7 -0
  156. package/dist/profiles/domain-profile.d.ts +19 -0
  157. package/dist/profiles/domain-profile.js +140 -0
  158. package/dist/profiles/domain-profile.js.map +7 -0
  159. package/dist/proxy.d.ts +50 -0
  160. package/dist/proxy.js +165 -0
  161. package/dist/proxy.js.map +7 -0
  162. package/dist/rate-limiter.d.ts +31 -0
  163. package/dist/rate-limiter.js +64 -0
  164. package/dist/rate-limiter.js.map +7 -0
  165. package/dist/redirect-tracker.d.ts +37 -0
  166. package/dist/redirect-tracker.js +65 -0
  167. package/dist/redirect-tracker.js.map +7 -0
  168. package/dist/redis-frontier.d.ts +48 -0
  169. package/dist/redis-frontier.js +126 -0
  170. package/dist/redis-frontier.js.map +7 -0
  171. package/dist/redis-queue.d.ts +48 -0
  172. package/dist/redis-queue.js +165 -0
  173. package/dist/redis-queue.js.map +7 -0
  174. package/dist/response-cache.d.ts +38 -0
  175. package/dist/response-cache.js +140 -0
  176. package/dist/response-cache.js.map +7 -0
  177. package/dist/responsive-detector.d.ts +29 -0
  178. package/dist/responsive-detector.js +46 -0
  179. package/dist/responsive-detector.js.map +7 -0
  180. package/dist/retry-policy.d.ts +20 -0
  181. package/dist/retry-policy.js +54 -0
  182. package/dist/retry-policy.js.map +7 -0
  183. package/dist/robots-txt.d.ts +9 -0
  184. package/dist/robots-txt.js +77 -0
  185. package/dist/robots-txt.js.map +7 -0
  186. package/dist/screen-reader.d.ts +16 -0
  187. package/dist/screen-reader.js +46 -0
  188. package/dist/screen-reader.js.map +7 -0
  189. package/dist/screenshot-masker.d.ts +28 -0
  190. package/dist/screenshot-masker.js +62 -0
  191. package/dist/screenshot-masker.js.map +7 -0
  192. package/dist/seo-checker.d.ts +33 -0
  193. package/dist/seo-checker.js +91 -0
  194. package/dist/seo-checker.js.map +7 -0
  195. package/dist/session-capture.d.ts +35 -0
  196. package/dist/session-capture.js +267 -0
  197. package/dist/session-capture.js.map +7 -0
  198. package/dist/session-export.d.ts +49 -0
  199. package/dist/session-export.js +117 -0
  200. package/dist/session-export.js.map +7 -0
  201. package/dist/sitemap-writer.d.ts +21 -0
  202. package/dist/sitemap-writer.js +47 -0
  203. package/dist/sitemap-writer.js.map +7 -0
  204. package/dist/sitemap.d.ts +51 -0
  205. package/dist/sitemap.js +152 -0
  206. package/dist/sitemap.js.map +7 -0
  207. package/dist/skyvern.d.ts +56 -0
  208. package/dist/skyvern.js +129 -0
  209. package/dist/skyvern.js.map +7 -0
  210. package/dist/sqs-queue.d.ts +2 -0
  211. package/dist/sqs-queue.js +47 -0
  212. package/dist/sqs-queue.js.map +7 -0
  213. package/dist/sso-token.d.ts +29 -0
  214. package/dist/sso-token.js +76 -0
  215. package/dist/sso-token.js.map +7 -0
  216. package/dist/stealth/human-behavior.d.ts +20 -0
  217. package/dist/stealth/human-behavior.js +168 -0
  218. package/dist/stealth/human-behavior.js.map +7 -0
  219. package/dist/telemetry.d.ts +34 -0
  220. package/dist/telemetry.js +77 -0
  221. package/dist/telemetry.js.map +7 -0
  222. package/dist/test-executor.d.ts +67 -0
  223. package/dist/test-executor.js +311 -0
  224. package/dist/test-executor.js.map +7 -0
  225. package/dist/tls-fingerprint.d.ts +50 -0
  226. package/dist/tls-fingerprint.js +81 -0
  227. package/dist/tls-fingerprint.js.map +7 -0
  228. package/dist/url-pattern-completer.d.ts +28 -0
  229. package/dist/url-pattern-completer.js +87 -0
  230. package/dist/url-pattern-completer.js.map +7 -0
  231. package/dist/visual-diff-verdict.d.ts +40 -0
  232. package/dist/visual-diff-verdict.js +25 -0
  233. package/dist/visual-diff-verdict.js.map +7 -0
  234. package/dist/webhook-notifier.d.ts +30 -0
  235. package/dist/webhook-notifier.js +76 -0
  236. package/dist/webhook-notifier.js.map +7 -0
  237. package/dist/websocket-detector.d.ts +24 -0
  238. package/dist/websocket-detector.js +60 -0
  239. package/dist/websocket-detector.js.map +7 -0
  240. package/dist/worker.d.ts +1 -0
  241. package/dist/worker.js +33 -0
  242. package/dist/worker.js.map +7 -0
  243. package/package.json +45 -0
  244. package/src/accessibility-auditor.ts +102 -0
  245. package/src/ai-login.ts +218 -0
  246. package/src/auth-token-rotator.ts +89 -0
  247. package/src/auth.ts +396 -0
  248. package/src/browser-extension.ts +186 -0
  249. package/src/captcha-solver.ts +143 -0
  250. package/src/color-contrast.ts +116 -0
  251. package/src/content-chunker.ts +170 -0
  252. package/src/content-dedup.ts +99 -0
  253. package/src/cookie-consent.ts +106 -0
  254. package/src/cookie-jar.ts +91 -0
  255. package/src/crawler.ts +4305 -0
  256. package/src/custom-js-injector.ts +93 -0
  257. package/src/dark-mode-capture.ts +110 -0
  258. package/src/dead-link-detector.ts +76 -0
  259. package/src/element-hash.ts +16 -0
  260. package/src/email-otp.ts +252 -0
  261. package/src/engine-router.ts +141 -0
  262. package/src/export/bulk-export.ts +85 -0
  263. package/src/export/vector-export.ts +160 -0
  264. package/src/extractors/advanced.ts +283 -0
  265. package/src/extractors/artifact-envelope.ts +111 -0
  266. package/src/failure-analysis-agent.ts +49 -0
  267. package/src/feature-flag-detector.ts +97 -0
  268. package/src/form-detector.ts +103 -0
  269. package/src/frontier.test.ts +64 -0
  270. package/src/frontier.ts +112 -0
  271. package/src/gap-pass.ts +122 -0
  272. package/src/graphql-detector.ts +129 -0
  273. package/src/http-engine.ts +433 -0
  274. package/src/incremental-diff.ts +121 -0
  275. package/src/index.ts +35 -0
  276. package/src/input-fuzzer.ts +129 -0
  277. package/src/interactive-probe.ts +623 -0
  278. package/src/job-queue.test.ts +120 -0
  279. package/src/job-queue.ts +2701 -0
  280. package/src/js-coverage.ts +87 -0
  281. package/src/js-router-detector.ts +112 -0
  282. package/src/language-detector.ts +82 -0
  283. package/src/link-classifier.ts +65 -0
  284. package/src/live-view.ts +141 -0
  285. package/src/logger.ts +9 -0
  286. package/src/markdown-extract.test.ts +73 -0
  287. package/src/markdown-extract.ts +76 -0
  288. package/src/native-crawler.ts +1106 -0
  289. package/src/network-observer.ts +204 -0
  290. package/src/network-throttle.ts +69 -0
  291. package/src/openapi-generator.ts +135 -0
  292. package/src/pdf-extractor.ts +123 -0
  293. package/src/performance-metrics.ts +134 -0
  294. package/src/phash.ts +151 -0
  295. package/src/pii-scrubber.ts +99 -0
  296. package/src/playwright-recorder.ts +170 -0
  297. package/src/policies/crawl-safety.ts +46 -0
  298. package/src/profiles/domain-profile.ts +167 -0
  299. package/src/proxy.ts +238 -0
  300. package/src/rate-limiter.ts +87 -0
  301. package/src/redirect-tracker.ts +97 -0
  302. package/src/redis-frontier.ts +143 -0
  303. package/src/redis-queue.ts +208 -0
  304. package/src/response-cache.ts +150 -0
  305. package/src/responsive-detector.ts +92 -0
  306. package/src/retry-policy.ts +87 -0
  307. package/src/robots-txt.ts +96 -0
  308. package/src/screen-reader.ts +89 -0
  309. package/src/screenshot-masker.ts +98 -0
  310. package/src/seo-checker.ts +134 -0
  311. package/src/session-capture.ts +325 -0
  312. package/src/session-export.test.ts +75 -0
  313. package/src/session-export.ts +172 -0
  314. package/src/sitemap-writer.ts +65 -0
  315. package/src/sitemap.test.ts +131 -0
  316. package/src/sitemap.ts +224 -0
  317. package/src/skyvern.ts +199 -0
  318. package/src/sqs-queue.ts +49 -0
  319. package/src/sso-token.ts +120 -0
  320. package/src/stealth/human-behavior.ts +191 -0
  321. package/src/telemetry.ts +122 -0
  322. package/src/test-executor.test.ts +72 -0
  323. package/src/test-executor.ts +424 -0
  324. package/src/tls-fingerprint.ts +111 -0
  325. package/src/url-pattern-completer.ts +114 -0
  326. package/src/visual-diff-verdict.test.ts +80 -0
  327. package/src/visual-diff-verdict.ts +57 -0
  328. package/src/webhook-notifier.ts +109 -0
  329. package/src/websocket-detector.ts +83 -0
  330. package/src/worker.ts +48 -0
  331. package/tsconfig.json +13 -0
  332. package/types/accessibility-auditor.d.ts +29 -0
  333. package/types/ai-login.d.ts +32 -0
  334. package/types/auth-token-rotator.d.ts +47 -0
  335. package/types/auth.d.ts +20 -0
  336. package/types/browser-extension.d.ts +37 -0
  337. package/types/bull-worker.d.ts +6 -0
  338. package/types/captcha-solver.d.ts +16 -0
  339. package/types/color-contrast.d.ts +30 -0
  340. package/types/content-chunker.d.ts +18 -0
  341. package/types/content-dedup.d.ts +23 -0
  342. package/types/cookie-consent.d.ts +25 -0
  343. package/types/cookie-jar.d.ts +20 -0
  344. package/types/crawler.d.ts +179 -0
  345. package/types/custom-js-injector.d.ts +42 -0
  346. package/types/dark-mode-capture.d.ts +23 -0
  347. package/types/dead-link-detector.d.ts +37 -0
  348. package/types/element-hash.d.ts +9 -0
  349. package/types/email-otp.d.ts +42 -0
  350. package/types/engine-router.d.ts +47 -0
  351. package/types/export/bulk-export.d.ts +26 -0
  352. package/types/export/vector-export.d.ts +76 -0
  353. package/types/extractors/advanced.d.ts +54 -0
  354. package/types/extractors/artifact-envelope.d.ts +63 -0
  355. package/types/failure-analysis-agent.d.ts +16 -0
  356. package/types/feature-flag-detector.d.ts +30 -0
  357. package/types/form-detector.d.ts +34 -0
  358. package/types/frontier.d.ts +32 -0
  359. package/types/gap-pass.d.ts +35 -0
  360. package/types/graphql-detector.d.ts +39 -0
  361. package/types/http-engine.d.ts +80 -0
  362. package/types/incremental-diff.d.ts +59 -0
  363. package/types/index.d.ts +35 -0
  364. package/types/input-fuzzer.d.ts +30 -0
  365. package/types/interactive-probe.d.ts +104 -0
  366. package/types/job-queue.d.ts +171 -0
  367. package/types/js-coverage.d.ts +29 -0
  368. package/types/js-router-detector.d.ts +17 -0
  369. package/types/language-detector.d.ts +22 -0
  370. package/types/link-classifier.d.ts +23 -0
  371. package/types/live-view.d.ts +36 -0
  372. package/types/logger.d.ts +2 -0
  373. package/types/markdown-extract.d.ts +8 -0
  374. package/types/native-crawler.d.ts +69 -0
  375. package/types/network-observer.d.ts +49 -0
  376. package/types/network-throttle.d.ts +25 -0
  377. package/types/openapi-generator.d.ts +49 -0
  378. package/types/pdf-extractor.d.ts +28 -0
  379. package/types/performance-metrics.d.ts +53 -0
  380. package/types/phash.d.ts +38 -0
  381. package/types/pii-scrubber.d.ts +21 -0
  382. package/types/playwright-recorder.d.ts +39 -0
  383. package/types/policies/crawl-safety.d.ts +9 -0
  384. package/types/profiles/domain-profile.d.ts +19 -0
  385. package/types/proxy.d.ts +50 -0
  386. package/types/rate-limiter.d.ts +31 -0
  387. package/types/redirect-tracker.d.ts +37 -0
  388. package/types/redis-frontier.d.ts +48 -0
  389. package/types/redis-queue.d.ts +48 -0
  390. package/types/response-cache.d.ts +38 -0
  391. package/types/responsive-detector.d.ts +29 -0
  392. package/types/retry-policy.d.ts +20 -0
  393. package/types/robots-txt.d.ts +9 -0
  394. package/types/screen-reader.d.ts +16 -0
  395. package/types/screenshot-masker.d.ts +28 -0
  396. package/types/seo-checker.d.ts +33 -0
  397. package/types/session-capture.d.ts +38 -0
  398. package/types/session-export.d.ts +49 -0
  399. package/types/sitemap-writer.d.ts +21 -0
  400. package/types/sitemap.d.ts +51 -0
  401. package/types/skyvern.d.ts +56 -0
  402. package/types/sqs-queue.d.ts +2 -0
  403. package/types/sso-token.d.ts +29 -0
  404. package/types/stealth/human-behavior.d.ts +20 -0
  405. package/types/telemetry.d.ts +34 -0
  406. package/types/test-executor.d.ts +67 -0
  407. package/types/tls-fingerprint.d.ts +50 -0
  408. package/types/url-pattern-completer.d.ts +28 -0
  409. package/types/visual-diff-verdict.d.ts +40 -0
  410. package/types/webhook-notifier.d.ts +30 -0
  411. package/types/websocket-detector.d.ts +24 -0
  412. package/types/worker.d.ts +1 -0
@@ -0,0 +1,91 @@
1
+ /**
2
+ * Cookie jar — persist and restore Playwright storageState cookies between
3
+ * crawl segments and restarts. Enables continuous auth across session boundaries
4
+ * (e.g. segment 1 logs in, segment 2–N inherit the session without re-login).
5
+ *
6
+ * Storage: JSON file per project in the crawl cache dir.
7
+ * Format: Playwright's storageState shape (cookies + origins).
8
+ */
9
+
10
+ import * as fs from 'fs/promises';
11
+ import * as path from 'path';
12
+
13
+ export interface CookieJarEntry {
14
+ projectId: string;
15
+ tenantId: string;
16
+ storageState: string; // JSON-stringified Playwright StorageState
17
+ savedAt: string; // ISO timestamp
18
+ domain: string; // hostname for quick lookup
19
+ }
20
+
21
+ function jarPath(cacheDir: string, projectId: string): string {
22
+ return path.join(cacheDir, `cookie-jar-${projectId}.json`);
23
+ }
24
+
25
+ export async function saveCookieJar(
26
+ cacheDir: string,
27
+ projectId: string,
28
+ tenantId: string,
29
+ storageState: string,
30
+ domain: string,
31
+ ): Promise<void> {
32
+ await fs.mkdir(cacheDir, { recursive: true });
33
+ const entry: CookieJarEntry = {
34
+ projectId,
35
+ tenantId,
36
+ storageState,
37
+ savedAt: new Date().toISOString(),
38
+ domain,
39
+ };
40
+ await fs.writeFile(jarPath(cacheDir, projectId), JSON.stringify(entry, null, 2), 'utf8');
41
+ }
42
+
43
+ export async function loadCookieJar(
44
+ cacheDir: string,
45
+ projectId: string,
46
+ maxAgeMs = 8 * 60 * 60 * 1000, // 8 hours
47
+ ): Promise<string | null> {
48
+ try {
49
+ const raw = await fs.readFile(jarPath(cacheDir, projectId), 'utf8');
50
+ const entry: CookieJarEntry = JSON.parse(raw);
51
+ const age = Date.now() - new Date(entry.savedAt).getTime();
52
+ if (age > maxAgeMs) return null; // expired
53
+ return entry.storageState;
54
+ } catch {
55
+ return null;
56
+ }
57
+ }
58
+
59
+ export async function clearCookieJar(cacheDir: string, projectId: string): Promise<void> {
60
+ try {
61
+ await fs.unlink(jarPath(cacheDir, projectId));
62
+ } catch { /* already gone */ }
63
+ }
64
+
65
+ /** Merge two storageState JSON strings — second takes priority for same-domain cookies. */
66
+ export function mergeStorageStates(base: string, overlay: string): string {
67
+ try {
68
+ const a = JSON.parse(base);
69
+ const b = JSON.parse(overlay);
70
+
71
+ // Cookies: overlay replaces same-name+domain+path cookies from base
72
+ const cookies = [...(a.cookies ?? [])];
73
+ for (const ck of (b.cookies ?? [])) {
74
+ const idx = cookies.findIndex(
75
+ (c: any) => c.name === ck.name && c.domain === ck.domain && c.path === ck.path,
76
+ );
77
+ if (idx >= 0) cookies[idx] = ck;
78
+ else cookies.push(ck);
79
+ }
80
+
81
+ // Origins: merge origins, overlay wins on same-origin
82
+ const originsMap = new Map<string, any>();
83
+ for (const o of [...(a.origins ?? []), ...(b.origins ?? [])]) {
84
+ originsMap.set(o.origin, o);
85
+ }
86
+
87
+ return JSON.stringify({ cookies, origins: [...originsMap.values()] });
88
+ } catch {
89
+ return overlay || base;
90
+ }
91
+ }