@staticn0va/wigolo 0.1.0 → 0.1.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (1003) hide show
  1. package/LICENSE +1 -1
  2. package/README.md +146 -227
  3. package/SKILL.md +382 -0
  4. package/assets/blocks/claude-code/CLAUDE.md.block +20 -0
  5. package/assets/blocks/claude-code/wigolo-command.md +40 -0
  6. package/assets/blocks/cursor/wigolo.mdc +46 -0
  7. package/assets/blocks/gemini-cli/GEMINI.md.block +18 -0
  8. package/assets/blocks/vscode/copilot-instructions.md.block +18 -0
  9. package/assets/skills/wigolo/SKILL.md +50 -0
  10. package/assets/skills/wigolo/rules/cache-first.md +30 -0
  11. package/assets/skills/wigolo/rules/synthesis.md +43 -0
  12. package/assets/skills/wigolo-agent/SKILL.md +73 -0
  13. package/assets/skills/wigolo-crawl/SKILL.md +60 -0
  14. package/assets/skills/wigolo-extract/SKILL.md +59 -0
  15. package/assets/skills/wigolo-fetch/SKILL.md +65 -0
  16. package/assets/skills/wigolo-find-similar/SKILL.md +72 -0
  17. package/assets/skills/wigolo-research/SKILL.md +77 -0
  18. package/assets/skills/wigolo-search/SKILL.md +78 -0
  19. package/dist/agent/executor.d.ts +33 -0
  20. package/dist/agent/executor.d.ts.map +1 -0
  21. package/dist/agent/executor.js +233 -0
  22. package/dist/agent/executor.js.map +1 -0
  23. package/dist/agent/pipeline.d.ts +5 -0
  24. package/dist/agent/pipeline.d.ts.map +1 -0
  25. package/dist/agent/pipeline.js +238 -0
  26. package/dist/agent/pipeline.js.map +1 -0
  27. package/dist/agent/planner.d.ts +13 -0
  28. package/dist/agent/planner.d.ts.map +1 -0
  29. package/dist/agent/planner.js +271 -0
  30. package/dist/agent/planner.js.map +1 -0
  31. package/dist/agent/relevance.d.ts +15 -0
  32. package/dist/agent/relevance.d.ts.map +1 -0
  33. package/dist/agent/relevance.js +60 -0
  34. package/dist/agent/relevance.js.map +1 -0
  35. package/dist/cache/backfill-embeddings.d.ts +23 -0
  36. package/dist/cache/backfill-embeddings.d.ts.map +1 -0
  37. package/dist/cache/backfill-embeddings.js +105 -0
  38. package/dist/cache/backfill-embeddings.js.map +1 -0
  39. package/dist/cache/change-detector.d.ts +7 -0
  40. package/dist/cache/change-detector.d.ts.map +1 -0
  41. package/dist/cache/change-detector.js +43 -0
  42. package/dist/cache/change-detector.js.map +1 -0
  43. package/dist/cache/db.d.ts +1 -0
  44. package/dist/cache/db.d.ts.map +1 -1
  45. package/dist/cache/db.js +94 -22
  46. package/dist/cache/db.js.map +1 -1
  47. package/dist/cache/diff-summary.d.ts +2 -0
  48. package/dist/cache/diff-summary.d.ts.map +1 -0
  49. package/dist/cache/diff-summary.js +82 -0
  50. package/dist/cache/diff-summary.js.map +1 -0
  51. package/dist/cache/migrations/runner.d.ts +29 -0
  52. package/dist/cache/migrations/runner.d.ts.map +1 -0
  53. package/dist/cache/migrations/runner.js +147 -0
  54. package/dist/cache/migrations/runner.js.map +1 -0
  55. package/dist/cache/sqlite-vec-store.d.ts +42 -0
  56. package/dist/cache/sqlite-vec-store.d.ts.map +1 -0
  57. package/dist/cache/sqlite-vec-store.js +176 -0
  58. package/dist/cache/sqlite-vec-store.js.map +1 -0
  59. package/dist/cache/store.d.ts +47 -1
  60. package/dist/cache/store.d.ts.map +1 -1
  61. package/dist/cache/store.js +364 -168
  62. package/dist/cache/store.js.map +1 -1
  63. package/dist/cli/agents/antigravity.d.ts +20 -0
  64. package/dist/cli/agents/antigravity.d.ts.map +1 -0
  65. package/dist/cli/agents/antigravity.js +49 -0
  66. package/dist/cli/agents/antigravity.js.map +1 -0
  67. package/dist/cli/agents/claude-code.d.ts +25 -0
  68. package/dist/cli/agents/claude-code.d.ts.map +1 -0
  69. package/dist/cli/agents/claude-code.js +111 -0
  70. package/dist/cli/agents/claude-code.js.map +1 -0
  71. package/dist/cli/agents/cursor.d.ts +21 -0
  72. package/dist/cli/agents/cursor.d.ts.map +1 -0
  73. package/dist/cli/agents/cursor.js +58 -0
  74. package/dist/cli/agents/cursor.js.map +1 -0
  75. package/dist/cli/agents/gemini-cli.d.ts +21 -0
  76. package/dist/cli/agents/gemini-cli.d.ts.map +1 -0
  77. package/dist/cli/agents/gemini-cli.js +55 -0
  78. package/dist/cli/agents/gemini-cli.js.map +1 -0
  79. package/dist/cli/agents/registry.d.ts +21 -0
  80. package/dist/cli/agents/registry.d.ts.map +1 -0
  81. package/dist/cli/agents/registry.js +27 -0
  82. package/dist/cli/agents/registry.js.map +1 -0
  83. package/dist/cli/agents/utils.d.ts +26 -0
  84. package/dist/cli/agents/utils.d.ts.map +1 -0
  85. package/dist/cli/agents/utils.js +136 -0
  86. package/dist/cli/agents/utils.js.map +1 -0
  87. package/dist/cli/agents/vscode.d.ts +21 -0
  88. package/dist/cli/agents/vscode.d.ts.map +1 -0
  89. package/dist/cli/agents/vscode.js +62 -0
  90. package/dist/cli/agents/vscode.js.map +1 -0
  91. package/dist/cli/auth.d.ts +2 -0
  92. package/dist/cli/auth.d.ts.map +1 -0
  93. package/dist/cli/auth.js +94 -0
  94. package/dist/cli/auth.js.map +1 -0
  95. package/dist/cli/backfill.d.ts +2 -0
  96. package/dist/cli/backfill.d.ts.map +1 -0
  97. package/dist/cli/backfill.js +58 -0
  98. package/dist/cli/backfill.js.map +1 -0
  99. package/dist/cli/daemon.d.ts +6 -1
  100. package/dist/cli/daemon.d.ts.map +1 -1
  101. package/dist/cli/daemon.js +61 -3
  102. package/dist/cli/daemon.js.map +1 -1
  103. package/dist/cli/doctor.d.ts +8 -0
  104. package/dist/cli/doctor.d.ts.map +1 -0
  105. package/dist/cli/doctor.js +344 -0
  106. package/dist/cli/doctor.js.map +1 -0
  107. package/dist/cli/health.d.ts +1 -1
  108. package/dist/cli/health.d.ts.map +1 -1
  109. package/dist/cli/health.js +42 -3
  110. package/dist/cli/health.js.map +1 -1
  111. package/dist/cli/help.d.ts +6 -0
  112. package/dist/cli/help.d.ts.map +1 -0
  113. package/dist/cli/help.js +63 -0
  114. package/dist/cli/help.js.map +1 -0
  115. package/dist/cli/index.d.ts +1 -1
  116. package/dist/cli/index.d.ts.map +1 -1
  117. package/dist/cli/index.js +35 -7
  118. package/dist/cli/index.js.map +1 -1
  119. package/dist/cli/init.d.ts +2 -0
  120. package/dist/cli/init.d.ts.map +1 -0
  121. package/dist/cli/init.js +201 -0
  122. package/dist/cli/init.js.map +1 -0
  123. package/dist/cli/plugin.d.ts +5 -0
  124. package/dist/cli/plugin.d.ts.map +1 -0
  125. package/dist/cli/plugin.js +185 -0
  126. package/dist/cli/plugin.js.map +1 -0
  127. package/dist/cli/setup-mcp.d.ts +2 -0
  128. package/dist/cli/setup-mcp.d.ts.map +1 -0
  129. package/dist/cli/setup-mcp.js +114 -0
  130. package/dist/cli/setup-mcp.js.map +1 -0
  131. package/dist/cli/shell.d.ts +2 -0
  132. package/dist/cli/shell.d.ts.map +1 -0
  133. package/dist/cli/shell.js +86 -0
  134. package/dist/cli/shell.js.map +1 -0
  135. package/dist/cli/shutdown.d.ts +2 -0
  136. package/dist/cli/shutdown.d.ts.map +1 -0
  137. package/dist/cli/shutdown.js +26 -0
  138. package/dist/cli/shutdown.js.map +1 -0
  139. package/dist/cli/status.d.ts +2 -0
  140. package/dist/cli/status.d.ts.map +1 -0
  141. package/dist/cli/status.js +31 -0
  142. package/dist/cli/status.js.map +1 -0
  143. package/dist/cli/telemetry.d.ts +10 -0
  144. package/dist/cli/telemetry.d.ts.map +1 -0
  145. package/dist/cli/telemetry.js +56 -0
  146. package/dist/cli/telemetry.js.map +1 -0
  147. package/dist/cli/tui/agents-types.d.ts +28 -0
  148. package/dist/cli/tui/agents-types.d.ts.map +1 -0
  149. package/dist/cli/tui/agents-types.js +1 -0
  150. package/dist/cli/tui/agents-types.js.map +1 -0
  151. package/dist/cli/tui/agents.d.ts +11 -0
  152. package/dist/cli/tui/agents.d.ts.map +1 -0
  153. package/dist/cli/tui/agents.js +93 -0
  154. package/dist/cli/tui/agents.js.map +1 -0
  155. package/dist/cli/tui/banner.d.ts +3 -0
  156. package/dist/cli/tui/banner.d.ts.map +1 -0
  157. package/dist/cli/tui/banner.js +30 -0
  158. package/dist/cli/tui/banner.js.map +1 -0
  159. package/dist/cli/tui/components/AgentSelect.d.ts +13 -0
  160. package/dist/cli/tui/components/AgentSelect.d.ts.map +1 -0
  161. package/dist/cli/tui/components/AgentSelect.js +116 -0
  162. package/dist/cli/tui/components/AgentSelect.js.map +1 -0
  163. package/dist/cli/tui/components/Banner.d.ts +6 -0
  164. package/dist/cli/tui/components/Banner.d.ts.map +1 -0
  165. package/dist/cli/tui/components/Banner.js +25 -0
  166. package/dist/cli/tui/components/Banner.js.map +1 -0
  167. package/dist/cli/tui/components/BrowserSelect.d.ts +7 -0
  168. package/dist/cli/tui/components/BrowserSelect.d.ts.map +1 -0
  169. package/dist/cli/tui/components/BrowserSelect.js +19 -0
  170. package/dist/cli/tui/components/BrowserSelect.js.map +1 -0
  171. package/dist/cli/tui/components/InstallProgress.d.ts +9 -0
  172. package/dist/cli/tui/components/InstallProgress.d.ts.map +1 -0
  173. package/dist/cli/tui/components/InstallProgress.js +67 -0
  174. package/dist/cli/tui/components/InstallProgress.js.map +1 -0
  175. package/dist/cli/tui/components/SkillInstall.d.ts +14 -0
  176. package/dist/cli/tui/components/SkillInstall.d.ts.map +1 -0
  177. package/dist/cli/tui/components/SkillInstall.js +94 -0
  178. package/dist/cli/tui/components/SkillInstall.js.map +1 -0
  179. package/dist/cli/tui/components/Summary.d.ts +22 -0
  180. package/dist/cli/tui/components/Summary.d.ts.map +1 -0
  181. package/dist/cli/tui/components/Summary.js +135 -0
  182. package/dist/cli/tui/components/Summary.js.map +1 -0
  183. package/dist/cli/tui/components/SystemCheck.d.ts +8 -0
  184. package/dist/cli/tui/components/SystemCheck.d.ts.map +1 -0
  185. package/dist/cli/tui/components/SystemCheck.js +71 -0
  186. package/dist/cli/tui/components/SystemCheck.js.map +1 -0
  187. package/dist/cli/tui/components/Verification.d.ts +8 -0
  188. package/dist/cli/tui/components/Verification.d.ts.map +1 -0
  189. package/dist/cli/tui/components/Verification.js +63 -0
  190. package/dist/cli/tui/components/Verification.js.map +1 -0
  191. package/dist/cli/tui/config-writer-cli.d.ts +12 -0
  192. package/dist/cli/tui/config-writer-cli.d.ts.map +1 -0
  193. package/dist/cli/tui/config-writer-cli.js +39 -0
  194. package/dist/cli/tui/config-writer-cli.js.map +1 -0
  195. package/dist/cli/tui/config-writer-json.d.ts +16 -0
  196. package/dist/cli/tui/config-writer-json.d.ts.map +1 -0
  197. package/dist/cli/tui/config-writer-json.js +86 -0
  198. package/dist/cli/tui/config-writer-json.js.map +1 -0
  199. package/dist/cli/tui/config-writer-toml.d.ts +16 -0
  200. package/dist/cli/tui/config-writer-toml.d.ts.map +1 -0
  201. package/dist/cli/tui/config-writer-toml.js +83 -0
  202. package/dist/cli/tui/config-writer-toml.js.map +1 -0
  203. package/dist/cli/tui/config-writer.d.ts +25 -0
  204. package/dist/cli/tui/config-writer.d.ts.map +1 -0
  205. package/dist/cli/tui/config-writer.js +101 -0
  206. package/dist/cli/tui/config-writer.js.map +1 -0
  207. package/dist/cli/tui/detect-helpers.d.ts +6 -0
  208. package/dist/cli/tui/detect-helpers.d.ts.map +1 -0
  209. package/dist/cli/tui/detect-helpers.js +45 -0
  210. package/dist/cli/tui/detect-helpers.js.map +1 -0
  211. package/dist/cli/tui/extras-prompt.d.ts +7 -0
  212. package/dist/cli/tui/extras-prompt.d.ts.map +1 -0
  213. package/dist/cli/tui/extras-prompt.js +42 -0
  214. package/dist/cli/tui/extras-prompt.js.map +1 -0
  215. package/dist/cli/tui/flags-types.d.ts +19 -0
  216. package/dist/cli/tui/flags-types.d.ts.map +1 -0
  217. package/dist/cli/tui/flags-types.js +23 -0
  218. package/dist/cli/tui/flags-types.js.map +1 -0
  219. package/dist/cli/tui/flags.d.ts +5 -0
  220. package/dist/cli/tui/flags.d.ts.map +1 -0
  221. package/dist/cli/tui/flags.js +132 -0
  222. package/dist/cli/tui/flags.js.map +1 -0
  223. package/dist/cli/tui/format.d.ts +14 -0
  224. package/dist/cli/tui/format.d.ts.map +1 -0
  225. package/dist/cli/tui/format.js +37 -0
  226. package/dist/cli/tui/format.js.map +1 -0
  227. package/dist/cli/tui/hooks/useAgentDetect.d.ts +6 -0
  228. package/dist/cli/tui/hooks/useAgentDetect.d.ts.map +1 -0
  229. package/dist/cli/tui/hooks/useAgentDetect.js +19 -0
  230. package/dist/cli/tui/hooks/useAgentDetect.js.map +1 -0
  231. package/dist/cli/tui/hooks/useInstall.d.ts +14 -0
  232. package/dist/cli/tui/hooks/useInstall.d.ts.map +1 -0
  233. package/dist/cli/tui/hooks/useInstall.js +90 -0
  234. package/dist/cli/tui/hooks/useInstall.js.map +1 -0
  235. package/dist/cli/tui/hooks/useSystemCheck.d.ts +13 -0
  236. package/dist/cli/tui/hooks/useSystemCheck.d.ts.map +1 -0
  237. package/dist/cli/tui/hooks/useSystemCheck.js +95 -0
  238. package/dist/cli/tui/hooks/useSystemCheck.js.map +1 -0
  239. package/dist/cli/tui/hooks/useVerify.d.ts +14 -0
  240. package/dist/cli/tui/hooks/useVerify.d.ts.map +1 -0
  241. package/dist/cli/tui/hooks/useVerify.js +71 -0
  242. package/dist/cli/tui/hooks/useVerify.js.map +1 -0
  243. package/dist/cli/tui/ink-init.d.ts +2 -0
  244. package/dist/cli/tui/ink-init.d.ts.map +1 -0
  245. package/dist/cli/tui/ink-init.js +198 -0
  246. package/dist/cli/tui/ink-init.js.map +1 -0
  247. package/dist/cli/tui/reporter-auto.d.ts +7 -0
  248. package/dist/cli/tui/reporter-auto.d.ts.map +1 -0
  249. package/dist/cli/tui/reporter-auto.js +15 -0
  250. package/dist/cli/tui/reporter-auto.js.map +1 -0
  251. package/dist/cli/tui/reporter.d.ts +26 -0
  252. package/dist/cli/tui/reporter.d.ts.map +1 -0
  253. package/dist/cli/tui/reporter.js +32 -0
  254. package/dist/cli/tui/reporter.js.map +1 -0
  255. package/dist/cli/tui/run-command.d.ts +14 -0
  256. package/dist/cli/tui/run-command.d.ts.map +1 -0
  257. package/dist/cli/tui/run-command.js +72 -0
  258. package/dist/cli/tui/run-command.js.map +1 -0
  259. package/dist/cli/tui/select-agents.d.ts +6 -0
  260. package/dist/cli/tui/select-agents.d.ts.map +1 -0
  261. package/dist/cli/tui/select-agents.js +32 -0
  262. package/dist/cli/tui/select-agents.js.map +1 -0
  263. package/dist/cli/tui/status-agents.d.ts +11 -0
  264. package/dist/cli/tui/status-agents.d.ts.map +1 -0
  265. package/dist/cli/tui/status-agents.js +53 -0
  266. package/dist/cli/tui/status-agents.js.map +1 -0
  267. package/dist/cli/tui/status-cache.d.ts +6 -0
  268. package/dist/cli/tui/status-cache.d.ts.map +1 -0
  269. package/dist/cli/tui/status-cache.js +39 -0
  270. package/dist/cli/tui/status-cache.js.map +1 -0
  271. package/dist/cli/tui/status-format.d.ts +14 -0
  272. package/dist/cli/tui/status-format.d.ts.map +1 -0
  273. package/dist/cli/tui/status-format.js +41 -0
  274. package/dist/cli/tui/status-format.js.map +1 -0
  275. package/dist/cli/tui/status-python.d.ts +6 -0
  276. package/dist/cli/tui/status-python.d.ts.map +1 -0
  277. package/dist/cli/tui/status-python.js +30 -0
  278. package/dist/cli/tui/status-python.js.map +1 -0
  279. package/dist/cli/tui/system-check.d.ts +24 -0
  280. package/dist/cli/tui/system-check.d.ts.map +1 -0
  281. package/dist/cli/tui/system-check.js +103 -0
  282. package/dist/cli/tui/system-check.js.map +1 -0
  283. package/dist/cli/tui/tui-reporter.d.ts +19 -0
  284. package/dist/cli/tui/tui-reporter.d.ts.map +1 -0
  285. package/dist/cli/tui/tui-reporter.js +95 -0
  286. package/dist/cli/tui/tui-reporter.js.map +1 -0
  287. package/dist/cli/tui/utils/config-writer.d.ts +3 -0
  288. package/dist/cli/tui/utils/config-writer.d.ts.map +1 -0
  289. package/dist/cli/tui/utils/config-writer.js +22 -0
  290. package/dist/cli/tui/utils/config-writer.js.map +1 -0
  291. package/dist/cli/tui/utils/suppress-logs.d.ts +3 -0
  292. package/dist/cli/tui/utils/suppress-logs.d.ts.map +1 -0
  293. package/dist/cli/tui/utils/suppress-logs.js +11 -0
  294. package/dist/cli/tui/utils/suppress-logs.js.map +1 -0
  295. package/dist/cli/tui/verify-suggestions.d.ts +5 -0
  296. package/dist/cli/tui/verify-suggestions.d.ts.map +1 -0
  297. package/dist/cli/tui/verify-suggestions.js +20 -0
  298. package/dist/cli/tui/verify-suggestions.js.map +1 -0
  299. package/dist/cli/tui/verify.d.ts +14 -0
  300. package/dist/cli/tui/verify.d.ts.map +1 -0
  301. package/dist/cli/tui/verify.js +101 -0
  302. package/dist/cli/tui/verify.js.map +1 -0
  303. package/dist/cli/tui/version.d.ts +2 -0
  304. package/dist/cli/tui/version.d.ts.map +1 -0
  305. package/dist/cli/tui/version.js +14 -0
  306. package/dist/cli/tui/version.js.map +1 -0
  307. package/dist/cli/uninstall.d.ts +2 -0
  308. package/dist/cli/uninstall.d.ts.map +1 -0
  309. package/dist/cli/uninstall.js +57 -0
  310. package/dist/cli/uninstall.js.map +1 -0
  311. package/dist/cli/warmup.d.ts +10 -2
  312. package/dist/cli/warmup.d.ts.map +1 -1
  313. package/dist/cli/warmup.js +226 -93
  314. package/dist/cli/warmup.js.map +1 -1
  315. package/dist/config.d.ts +28 -2
  316. package/dist/config.d.ts.map +1 -1
  317. package/dist/config.js +106 -56
  318. package/dist/config.js.map +1 -1
  319. package/dist/crawl/crawler.d.ts +6 -0
  320. package/dist/crawl/crawler.d.ts.map +1 -1
  321. package/dist/crawl/crawler.js +210 -209
  322. package/dist/crawl/crawler.js.map +1 -1
  323. package/dist/crawl/dedup.d.ts +1 -0
  324. package/dist/crawl/dedup.d.ts.map +1 -1
  325. package/dist/crawl/dedup.js +124 -81
  326. package/dist/crawl/dedup.js.map +1 -1
  327. package/dist/crawl/etag-incremental.d.ts +43 -0
  328. package/dist/crawl/etag-incremental.d.ts.map +1 -0
  329. package/dist/crawl/etag-incremental.js +94 -0
  330. package/dist/crawl/etag-incremental.js.map +1 -0
  331. package/dist/crawl/index-to-vec.d.ts +10 -0
  332. package/dist/crawl/index-to-vec.d.ts.map +1 -0
  333. package/dist/crawl/index-to-vec.js +44 -0
  334. package/dist/crawl/index-to-vec.js.map +1 -0
  335. package/dist/crawl/mapper.js +136 -164
  336. package/dist/crawl/mapper.js.map +1 -1
  337. package/dist/crawl/rate-limiter.js +63 -66
  338. package/dist/crawl/rate-limiter.js.map +1 -1
  339. package/dist/crawl/robots.js +58 -57
  340. package/dist/crawl/robots.js.map +1 -1
  341. package/dist/crawl/sitemap-first.d.ts +12 -0
  342. package/dist/crawl/sitemap-first.d.ts.map +1 -0
  343. package/dist/crawl/sitemap-first.js +47 -0
  344. package/dist/crawl/sitemap-first.js.map +1 -0
  345. package/dist/crawl/sitemap.js +33 -32
  346. package/dist/crawl/sitemap.js.map +1 -1
  347. package/dist/crawl/url-utils.d.ts +1 -0
  348. package/dist/crawl/url-utils.d.ts.map +1 -1
  349. package/dist/crawl/url-utils.js +49 -37
  350. package/dist/crawl/url-utils.js.map +1 -1
  351. package/dist/daemon/health-check.d.ts +16 -0
  352. package/dist/daemon/health-check.d.ts.map +1 -0
  353. package/dist/daemon/health-check.js +33 -0
  354. package/dist/daemon/health-check.js.map +1 -0
  355. package/dist/daemon/http-server.d.ts +26 -0
  356. package/dist/daemon/http-server.d.ts.map +1 -0
  357. package/dist/daemon/http-server.js +275 -0
  358. package/dist/daemon/http-server.js.map +1 -0
  359. package/dist/daemon/proxy.d.ts +10 -0
  360. package/dist/daemon/proxy.d.ts.map +1 -0
  361. package/dist/daemon/proxy.js +93 -0
  362. package/dist/daemon/proxy.js.map +1 -0
  363. package/dist/embedding/embed.d.ts +59 -0
  364. package/dist/embedding/embed.d.ts.map +1 -0
  365. package/dist/embedding/embed.js +233 -0
  366. package/dist/embedding/embed.js.map +1 -0
  367. package/dist/embedding/fastembed-provider.d.ts +19 -0
  368. package/dist/embedding/fastembed-provider.d.ts.map +1 -0
  369. package/dist/embedding/fastembed-provider.js +51 -0
  370. package/dist/embedding/fastembed-provider.js.map +1 -0
  371. package/dist/embedding/key-terms.d.ts +12 -0
  372. package/dist/embedding/key-terms.d.ts.map +1 -0
  373. package/dist/embedding/key-terms.js +234 -0
  374. package/dist/embedding/key-terms.js.map +1 -0
  375. package/dist/extraction/boilerplate.d.ts +15 -0
  376. package/dist/extraction/boilerplate.d.ts.map +1 -0
  377. package/dist/extraction/boilerplate.js +52 -0
  378. package/dist/extraction/boilerplate.js.map +1 -0
  379. package/dist/extraction/defuddle.d.ts.map +1 -1
  380. package/dist/extraction/defuddle.js +27 -23
  381. package/dist/extraction/defuddle.js.map +1 -1
  382. package/dist/extraction/extract.d.ts.map +1 -1
  383. package/dist/extraction/extract.js +76 -76
  384. package/dist/extraction/extract.js.map +1 -1
  385. package/dist/extraction/jsonld.js +50 -54
  386. package/dist/extraction/jsonld.js.map +1 -1
  387. package/dist/extraction/lang-hints.d.ts +2 -0
  388. package/dist/extraction/lang-hints.d.ts.map +1 -0
  389. package/dist/extraction/lang-hints.js +30 -0
  390. package/dist/extraction/lang-hints.js.map +1 -0
  391. package/dist/extraction/llm-fallback.d.ts +17 -0
  392. package/dist/extraction/llm-fallback.d.ts.map +1 -0
  393. package/dist/extraction/llm-fallback.js +130 -0
  394. package/dist/extraction/llm-fallback.js.map +1 -0
  395. package/dist/extraction/markdown-sanitize.d.ts +2 -0
  396. package/dist/extraction/markdown-sanitize.d.ts.map +1 -0
  397. package/dist/extraction/markdown-sanitize.js +151 -0
  398. package/dist/extraction/markdown-sanitize.js.map +1 -0
  399. package/dist/extraction/markdown.d.ts +11 -0
  400. package/dist/extraction/markdown.d.ts.map +1 -1
  401. package/dist/extraction/markdown.js +195 -91
  402. package/dist/extraction/markdown.js.map +1 -1
  403. package/dist/extraction/pipeline.d.ts +8 -0
  404. package/dist/extraction/pipeline.d.ts.map +1 -1
  405. package/dist/extraction/pipeline.js +57 -91
  406. package/dist/extraction/pipeline.js.map +1 -1
  407. package/dist/extraction/readability.d.ts +1 -1
  408. package/dist/extraction/readability.d.ts.map +1 -1
  409. package/dist/extraction/readability.js +28 -29
  410. package/dist/extraction/readability.js.map +1 -1
  411. package/dist/extraction/schema.d.ts +12 -0
  412. package/dist/extraction/schema.d.ts.map +1 -1
  413. package/dist/extraction/schema.js +135 -72
  414. package/dist/extraction/schema.js.map +1 -1
  415. package/dist/extraction/site-extractors/docs-generic.d.ts.map +1 -1
  416. package/dist/extraction/site-extractors/docs-generic.js +81 -91
  417. package/dist/extraction/site-extractors/docs-generic.js.map +1 -1
  418. package/dist/extraction/site-extractors/github.d.ts.map +1 -1
  419. package/dist/extraction/site-extractors/github.js +87 -95
  420. package/dist/extraction/site-extractors/github.js.map +1 -1
  421. package/dist/extraction/site-extractors/mdn.d.ts.map +1 -1
  422. package/dist/extraction/site-extractors/mdn.js +46 -54
  423. package/dist/extraction/site-extractors/mdn.js.map +1 -1
  424. package/dist/extraction/site-extractors/stackoverflow.d.ts.map +1 -1
  425. package/dist/extraction/site-extractors/stackoverflow.js +71 -80
  426. package/dist/extraction/site-extractors/stackoverflow.js.map +1 -1
  427. package/dist/extraction/structured-data.d.ts +4 -0
  428. package/dist/extraction/structured-data.d.ts.map +1 -0
  429. package/dist/extraction/structured-data.js +173 -0
  430. package/dist/extraction/structured-data.js.map +1 -0
  431. package/dist/extraction/structured.d.ts +4 -0
  432. package/dist/extraction/structured.d.ts.map +1 -0
  433. package/dist/extraction/structured.js +163 -0
  434. package/dist/extraction/structured.js.map +1 -0
  435. package/dist/extraction/v1/classifier.d.ts +3 -0
  436. package/dist/extraction/v1/classifier.d.ts.map +1 -0
  437. package/dist/extraction/v1/classifier.js +110 -0
  438. package/dist/extraction/v1/classifier.js.map +1 -0
  439. package/dist/extraction/v1/extract-provider.d.ts +16 -0
  440. package/dist/extraction/v1/extract-provider.d.ts.map +1 -0
  441. package/dist/extraction/v1/extract-provider.js +43 -0
  442. package/dist/extraction/v1/extract-provider.js.map +1 -0
  443. package/dist/extraction/v1/local-llm.d.ts +8 -0
  444. package/dist/extraction/v1/local-llm.d.ts.map +1 -0
  445. package/dist/extraction/v1/local-llm.js +34 -0
  446. package/dist/extraction/v1/local-llm.js.map +1 -0
  447. package/dist/extraction/v1/news.d.ts +3 -0
  448. package/dist/extraction/v1/news.d.ts.map +1 -0
  449. package/dist/extraction/v1/news.js +61 -0
  450. package/dist/extraction/v1/news.js.map +1 -0
  451. package/dist/extraction/v1/product.d.ts +3 -0
  452. package/dist/extraction/v1/product.d.ts.map +1 -0
  453. package/dist/extraction/v1/product.js +166 -0
  454. package/dist/extraction/v1/product.js.map +1 -0
  455. package/dist/extraction/v1/recipe.d.ts +3 -0
  456. package/dist/extraction/v1/recipe.d.ts.map +1 -0
  457. package/dist/extraction/v1/recipe.js +136 -0
  458. package/dist/extraction/v1/recipe.js.map +1 -0
  459. package/dist/extraction/v1/routed.d.ts +17 -0
  460. package/dist/extraction/v1/routed.d.ts.map +1 -0
  461. package/dist/extraction/v1/routed.js +68 -0
  462. package/dist/extraction/v1/routed.js.map +1 -0
  463. package/dist/extraction/v1/schemas/Article.d.ts +11 -0
  464. package/dist/extraction/v1/schemas/Article.d.ts.map +1 -0
  465. package/dist/extraction/v1/schemas/Article.js +23 -0
  466. package/dist/extraction/v1/schemas/Article.js.map +1 -0
  467. package/dist/extraction/v1/schemas/CodeSnippet.d.ts +9 -0
  468. package/dist/extraction/v1/schemas/CodeSnippet.d.ts.map +1 -0
  469. package/dist/extraction/v1/schemas/CodeSnippet.js +90 -0
  470. package/dist/extraction/v1/schemas/CodeSnippet.js.map +1 -0
  471. package/dist/extraction/v1/schemas/EventListing.d.ts +10 -0
  472. package/dist/extraction/v1/schemas/EventListing.d.ts.map +1 -0
  473. package/dist/extraction/v1/schemas/EventListing.js +122 -0
  474. package/dist/extraction/v1/schemas/EventListing.js.map +1 -0
  475. package/dist/extraction/v1/schemas/Paper.d.ts +10 -0
  476. package/dist/extraction/v1/schemas/Paper.d.ts.map +1 -0
  477. package/dist/extraction/v1/schemas/Paper.js +156 -0
  478. package/dist/extraction/v1/schemas/Paper.js.map +1 -0
  479. package/dist/extraction/v1/schemas/Product.d.ts +17 -0
  480. package/dist/extraction/v1/schemas/Product.d.ts.map +1 -0
  481. package/dist/extraction/v1/schemas/Product.js +149 -0
  482. package/dist/extraction/v1/schemas/Product.js.map +1 -0
  483. package/dist/extraction/v1/schemas/Recipe.d.ts +14 -0
  484. package/dist/extraction/v1/schemas/Recipe.d.ts.map +1 -0
  485. package/dist/extraction/v1/schemas/Recipe.js +160 -0
  486. package/dist/extraction/v1/schemas/Recipe.js.map +1 -0
  487. package/dist/extraction/v1/schemas/index.d.ts +13 -0
  488. package/dist/extraction/v1/schemas/index.d.ts.map +1 -0
  489. package/dist/extraction/v1/schemas/index.js +44 -0
  490. package/dist/extraction/v1/schemas/index.js.map +1 -0
  491. package/dist/extraction/v1/site-extractors.d.ts +5 -0
  492. package/dist/extraction/v1/site-extractors.d.ts.map +1 -0
  493. package/dist/extraction/v1/site-extractors.js +31 -0
  494. package/dist/extraction/v1/site-extractors.js.map +1 -0
  495. package/dist/fetch/action-executor.d.ts +28 -0
  496. package/dist/fetch/action-executor.d.ts.map +1 -0
  497. package/dist/fetch/action-executor.js +88 -0
  498. package/dist/fetch/action-executor.js.map +1 -0
  499. package/dist/fetch/auth.d.ts +2 -1
  500. package/dist/fetch/auth.d.ts.map +1 -1
  501. package/dist/fetch/auth.js +56 -26
  502. package/dist/fetch/auth.js.map +1 -1
  503. package/dist/fetch/browser-pool.d.ts +30 -11
  504. package/dist/fetch/browser-pool.d.ts.map +1 -1
  505. package/dist/fetch/browser-pool.js +303 -127
  506. package/dist/fetch/browser-pool.js.map +1 -1
  507. package/dist/fetch/browser-selector.d.ts +17 -0
  508. package/dist/fetch/browser-selector.d.ts.map +1 -0
  509. package/dist/fetch/browser-selector.js +72 -0
  510. package/dist/fetch/browser-selector.js.map +1 -0
  511. package/dist/fetch/browser-types.d.ts +3 -0
  512. package/dist/fetch/browser-types.d.ts.map +1 -0
  513. package/dist/fetch/browser-types.js +45 -0
  514. package/dist/fetch/browser-types.js.map +1 -0
  515. package/dist/fetch/cdp-client.d.ts +9 -0
  516. package/dist/fetch/cdp-client.d.ts.map +1 -0
  517. package/dist/fetch/cdp-client.js +89 -0
  518. package/dist/fetch/cdp-client.js.map +1 -0
  519. package/dist/fetch/content-check.js +39 -46
  520. package/dist/fetch/content-check.js.map +1 -1
  521. package/dist/fetch/error-describe.d.ts +7 -0
  522. package/dist/fetch/error-describe.d.ts.map +1 -0
  523. package/dist/fetch/error-describe.js +37 -0
  524. package/dist/fetch/error-describe.js.map +1 -0
  525. package/dist/fetch/http-client.d.ts +4 -0
  526. package/dist/fetch/http-client.d.ts.map +1 -1
  527. package/dist/fetch/http-client.js +147 -128
  528. package/dist/fetch/http-client.js.map +1 -1
  529. package/dist/fetch/lightpanda.d.ts +28 -0
  530. package/dist/fetch/lightpanda.d.ts.map +1 -0
  531. package/dist/fetch/lightpanda.js +174 -0
  532. package/dist/fetch/lightpanda.js.map +1 -0
  533. package/dist/fetch/playwright-tier.d.ts +19 -0
  534. package/dist/fetch/playwright-tier.d.ts.map +1 -0
  535. package/dist/fetch/playwright-tier.js +76 -0
  536. package/dist/fetch/playwright-tier.js.map +1 -0
  537. package/dist/fetch/router.d.ts +49 -3
  538. package/dist/fetch/router.d.ts.map +1 -1
  539. package/dist/fetch/router.js +187 -81
  540. package/dist/fetch/router.js.map +1 -1
  541. package/dist/index.js +102 -17
  542. package/dist/index.js.map +1 -1
  543. package/dist/instructions.d.ts +31 -0
  544. package/dist/instructions.d.ts.map +1 -0
  545. package/dist/instructions.js +245 -0
  546. package/dist/instructions.js.map +1 -0
  547. package/dist/integrations/cloud/llm/anthropic.d.ts +3 -0
  548. package/dist/integrations/cloud/llm/anthropic.d.ts.map +1 -0
  549. package/dist/integrations/cloud/llm/anthropic.js +41 -0
  550. package/dist/integrations/cloud/llm/anthropic.js.map +1 -0
  551. package/dist/integrations/cloud/llm/cache.d.ts +5 -0
  552. package/dist/integrations/cloud/llm/cache.d.ts.map +1 -0
  553. package/dist/integrations/cloud/llm/cache.js +49 -0
  554. package/dist/integrations/cloud/llm/cache.js.map +1 -0
  555. package/dist/integrations/cloud/llm/gemini.d.ts +3 -0
  556. package/dist/integrations/cloud/llm/gemini.d.ts.map +1 -0
  557. package/dist/integrations/cloud/llm/gemini.js +37 -0
  558. package/dist/integrations/cloud/llm/gemini.js.map +1 -0
  559. package/dist/integrations/cloud/llm/groq.d.ts +3 -0
  560. package/dist/integrations/cloud/llm/groq.d.ts.map +1 -0
  561. package/dist/integrations/cloud/llm/groq.js +74 -0
  562. package/dist/integrations/cloud/llm/groq.js.map +1 -0
  563. package/dist/integrations/cloud/llm/hash.d.ts +3 -0
  564. package/dist/integrations/cloud/llm/hash.d.ts.map +1 -0
  565. package/dist/integrations/cloud/llm/hash.js +26 -0
  566. package/dist/integrations/cloud/llm/hash.js.map +1 -0
  567. package/dist/integrations/cloud/llm/model-select.d.ts +5 -0
  568. package/dist/integrations/cloud/llm/model-select.d.ts.map +1 -0
  569. package/dist/integrations/cloud/llm/model-select.js +32 -0
  570. package/dist/integrations/cloud/llm/model-select.js.map +1 -0
  571. package/dist/integrations/cloud/llm/openai.d.ts +3 -0
  572. package/dist/integrations/cloud/llm/openai.d.ts.map +1 -0
  573. package/dist/integrations/cloud/llm/openai.js +43 -0
  574. package/dist/integrations/cloud/llm/openai.js.map +1 -0
  575. package/dist/integrations/cloud/llm/run.d.ts +27 -0
  576. package/dist/integrations/cloud/llm/run.d.ts.map +1 -0
  577. package/dist/integrations/cloud/llm/run.js +99 -0
  578. package/dist/integrations/cloud/llm/run.js.map +1 -0
  579. package/dist/integrations/cloud/llm/select.d.ts +5 -0
  580. package/dist/integrations/cloud/llm/select.d.ts.map +1 -0
  581. package/dist/integrations/cloud/llm/select.js +30 -0
  582. package/dist/integrations/cloud/llm/select.js.map +1 -0
  583. package/dist/integrations/cloud/llm/text-adapters.d.ts +19 -0
  584. package/dist/integrations/cloud/llm/text-adapters.d.ts.map +1 -0
  585. package/dist/integrations/cloud/llm/text-adapters.js +103 -0
  586. package/dist/integrations/cloud/llm/text-adapters.js.map +1 -0
  587. package/dist/integrations/cloud/llm/types.d.ts +24 -0
  588. package/dist/integrations/cloud/llm/types.d.ts.map +1 -0
  589. package/dist/integrations/cloud/llm/types.js +1 -0
  590. package/dist/integrations/cloud/llm/types.js.map +1 -0
  591. package/dist/integrations/cloud/llm/validate.d.ts +6 -0
  592. package/dist/integrations/cloud/llm/validate.d.ts.map +1 -0
  593. package/dist/integrations/cloud/llm/validate.js +63 -0
  594. package/dist/integrations/cloud/llm/validate.js.map +1 -0
  595. package/dist/logger.d.ts +4 -1
  596. package/dist/logger.d.ts.map +1 -1
  597. package/dist/logger.js +71 -30
  598. package/dist/logger.js.map +1 -1
  599. package/dist/pdf-parse.d.js +1 -0
  600. package/dist/pdf-parse.d.js.map +1 -0
  601. package/dist/plugins/loader.d.ts +20 -0
  602. package/dist/plugins/loader.d.ts.map +1 -0
  603. package/dist/plugins/loader.js +157 -0
  604. package/dist/plugins/loader.js.map +1 -0
  605. package/dist/plugins/registry.d.ts +26 -0
  606. package/dist/plugins/registry.d.ts.map +1 -0
  607. package/dist/plugins/registry.js +71 -0
  608. package/dist/plugins/registry.js.map +1 -0
  609. package/dist/plugins/validate.d.ts +9 -0
  610. package/dist/plugins/validate.d.ts.map +1 -0
  611. package/dist/plugins/validate.js +79 -0
  612. package/dist/plugins/validate.js.map +1 -0
  613. package/dist/providers/embed-provider.d.ts +11 -0
  614. package/dist/providers/embed-provider.d.ts.map +1 -0
  615. package/dist/providers/embed-provider.js +24 -0
  616. package/dist/providers/embed-provider.js.map +1 -0
  617. package/dist/providers/extract-provider.d.ts +23 -0
  618. package/dist/providers/extract-provider.d.ts.map +1 -0
  619. package/dist/providers/extract-provider.js +25 -0
  620. package/dist/providers/extract-provider.js.map +1 -0
  621. package/dist/providers/rerank-provider.d.ts +17 -0
  622. package/dist/providers/rerank-provider.d.ts.map +1 -0
  623. package/dist/providers/rerank-provider.js +41 -0
  624. package/dist/providers/rerank-provider.js.map +1 -0
  625. package/dist/providers/search-provider.d.ts +25 -0
  626. package/dist/providers/search-provider.d.ts.map +1 -0
  627. package/dist/providers/search-provider.js +44 -0
  628. package/dist/providers/search-provider.js.map +1 -0
  629. package/dist/providers/vector-store.d.ts +27 -0
  630. package/dist/providers/vector-store.d.ts.map +1 -0
  631. package/dist/providers/vector-store.js +27 -0
  632. package/dist/providers/vector-store.js.map +1 -0
  633. package/dist/python-env.d.ts +9 -0
  634. package/dist/python-env.d.ts.map +1 -0
  635. package/dist/python-env.js +13 -0
  636. package/dist/python-env.js.map +1 -0
  637. package/dist/repl/commands/agent.d.ts +5 -0
  638. package/dist/repl/commands/agent.d.ts.map +1 -0
  639. package/dist/repl/commands/agent.js +62 -0
  640. package/dist/repl/commands/agent.js.map +1 -0
  641. package/dist/repl/commands/cache.d.ts +4 -0
  642. package/dist/repl/commands/cache.d.ts.map +1 -0
  643. package/dist/repl/commands/cache.js +43 -0
  644. package/dist/repl/commands/cache.js.map +1 -0
  645. package/dist/repl/commands/crawl.d.ts +7 -0
  646. package/dist/repl/commands/crawl.d.ts.map +1 -0
  647. package/dist/repl/commands/crawl.js +44 -0
  648. package/dist/repl/commands/crawl.js.map +1 -0
  649. package/dist/repl/commands/extract.d.ts +5 -0
  650. package/dist/repl/commands/extract.d.ts.map +1 -0
  651. package/dist/repl/commands/extract.js +47 -0
  652. package/dist/repl/commands/extract.js.map +1 -0
  653. package/dist/repl/commands/fetch.d.ts +5 -0
  654. package/dist/repl/commands/fetch.d.ts.map +1 -0
  655. package/dist/repl/commands/fetch.js +67 -0
  656. package/dist/repl/commands/fetch.js.map +1 -0
  657. package/dist/repl/commands/find-similar.d.ts +5 -0
  658. package/dist/repl/commands/find-similar.d.ts.map +1 -0
  659. package/dist/repl/commands/find-similar.js +74 -0
  660. package/dist/repl/commands/find-similar.js.map +1 -0
  661. package/dist/repl/commands/research.d.ts +5 -0
  662. package/dist/repl/commands/research.d.ts.map +1 -0
  663. package/dist/repl/commands/research.js +65 -0
  664. package/dist/repl/commands/research.js.map +1 -0
  665. package/dist/repl/commands/search.d.ts +5 -0
  666. package/dist/repl/commands/search.d.ts.map +1 -0
  667. package/dist/repl/commands/search.js +74 -0
  668. package/dist/repl/commands/search.js.map +1 -0
  669. package/dist/repl/commands/types.d.ts +9 -0
  670. package/dist/repl/commands/types.d.ts.map +1 -0
  671. package/dist/repl/commands/types.js +1 -0
  672. package/dist/repl/commands/types.js.map +1 -0
  673. package/dist/repl/formatters.d.ts +13 -0
  674. package/dist/repl/formatters.d.ts.map +1 -0
  675. package/dist/repl/formatters.js +283 -0
  676. package/dist/repl/formatters.js.map +1 -0
  677. package/dist/repl/parser.d.ts +9 -0
  678. package/dist/repl/parser.d.ts.map +1 -0
  679. package/dist/repl/parser.js +86 -0
  680. package/dist/repl/parser.js.map +1 -0
  681. package/dist/repl/shell.d.ts +8 -0
  682. package/dist/repl/shell.d.ts.map +1 -0
  683. package/dist/repl/shell.js +184 -0
  684. package/dist/repl/shell.js.map +1 -0
  685. package/dist/research/branch-exploration.d.ts +14 -0
  686. package/dist/research/branch-exploration.d.ts.map +1 -0
  687. package/dist/research/branch-exploration.js +100 -0
  688. package/dist/research/branch-exploration.js.map +1 -0
  689. package/dist/research/brief.d.ts +6 -0
  690. package/dist/research/brief.d.ts.map +1 -0
  691. package/dist/research/brief.js +246 -0
  692. package/dist/research/brief.js.map +1 -0
  693. package/dist/research/citation-graph.d.ts +9 -0
  694. package/dist/research/citation-graph.d.ts.map +1 -0
  695. package/dist/research/citation-graph.js +114 -0
  696. package/dist/research/citation-graph.js.map +1 -0
  697. package/dist/research/decompose.d.ts +14 -0
  698. package/dist/research/decompose.d.ts.map +1 -0
  699. package/dist/research/decompose.js +439 -0
  700. package/dist/research/decompose.js.map +1 -0
  701. package/dist/research/pipeline.d.ts +5 -0
  702. package/dist/research/pipeline.d.ts.map +1 -0
  703. package/dist/research/pipeline.js +269 -0
  704. package/dist/research/pipeline.js.map +1 -0
  705. package/dist/research/synthesis-local.d.ts +19 -0
  706. package/dist/research/synthesis-local.d.ts.map +1 -0
  707. package/dist/research/synthesis-local.js +62 -0
  708. package/dist/research/synthesis-local.js.map +1 -0
  709. package/dist/research/synthesize.d.ts +10 -0
  710. package/dist/research/synthesize.d.ts.map +1 -0
  711. package/dist/research/synthesize.js +137 -0
  712. package/dist/research/synthesize.js.map +1 -0
  713. package/dist/search/answer-synthesis.d.ts +33 -0
  714. package/dist/search/answer-synthesis.d.ts.map +1 -0
  715. package/dist/search/answer-synthesis.js +244 -0
  716. package/dist/search/answer-synthesis.js.map +1 -0
  717. package/dist/search/context-formatter.d.ts +3 -0
  718. package/dist/search/context-formatter.d.ts.map +1 -0
  719. package/dist/search/context-formatter.js +56 -0
  720. package/dist/search/context-formatter.js.map +1 -0
  721. package/dist/search/dedup.d.ts +1 -0
  722. package/dist/search/dedup.d.ts.map +1 -1
  723. package/dist/search/dedup.js +40 -32
  724. package/dist/search/dedup.js.map +1 -1
  725. package/dist/search/engines/arxiv.d.ts +7 -0
  726. package/dist/search/engines/arxiv.d.ts.map +1 -0
  727. package/dist/search/engines/arxiv.js +70 -0
  728. package/dist/search/engines/arxiv.js.map +1 -0
  729. package/dist/search/engines/bing-news.d.ts +7 -0
  730. package/dist/search/engines/bing-news.d.ts.map +1 -0
  731. package/dist/search/engines/bing-news.js +97 -0
  732. package/dist/search/engines/bing-news.js.map +1 -0
  733. package/dist/search/engines/bing.d.ts +1 -0
  734. package/dist/search/engines/bing.d.ts.map +1 -1
  735. package/dist/search/engines/bing.js +100 -44
  736. package/dist/search/engines/bing.js.map +1 -1
  737. package/dist/search/engines/devdocs.d.ts +6 -0
  738. package/dist/search/engines/devdocs.d.ts.map +1 -0
  739. package/dist/search/engines/devdocs.js +56 -0
  740. package/dist/search/engines/devdocs.js.map +1 -0
  741. package/dist/search/engines/duckduckgo.d.ts.map +1 -1
  742. package/dist/search/engines/duckduckgo.js +56 -44
  743. package/dist/search/engines/duckduckgo.js.map +1 -1
  744. package/dist/search/engines/github-code.d.ts +7 -0
  745. package/dist/search/engines/github-code.d.ts.map +1 -0
  746. package/dist/search/engines/github-code.js +55 -0
  747. package/dist/search/engines/github-code.js.map +1 -0
  748. package/dist/search/engines/hn-algolia.d.ts +7 -0
  749. package/dist/search/engines/hn-algolia.d.ts.map +1 -0
  750. package/dist/search/engines/hn-algolia.js +76 -0
  751. package/dist/search/engines/hn-algolia.js.map +1 -0
  752. package/dist/search/engines/lobsters.d.ts +7 -0
  753. package/dist/search/engines/lobsters.d.ts.map +1 -0
  754. package/dist/search/engines/lobsters.js +83 -0
  755. package/dist/search/engines/lobsters.js.map +1 -0
  756. package/dist/search/engines/mdn.d.ts +7 -0
  757. package/dist/search/engines/mdn.d.ts.map +1 -0
  758. package/dist/search/engines/mdn.js +48 -0
  759. package/dist/search/engines/mdn.js.map +1 -0
  760. package/dist/search/engines/semantic-scholar.d.ts +7 -0
  761. package/dist/search/engines/semantic-scholar.d.ts.map +1 -0
  762. package/dist/search/engines/semantic-scholar.js +69 -0
  763. package/dist/search/engines/semantic-scholar.js.map +1 -0
  764. package/dist/search/engines/stackoverflow.d.ts +7 -0
  765. package/dist/search/engines/stackoverflow.d.ts.map +1 -0
  766. package/dist/search/engines/stackoverflow.js +73 -0
  767. package/dist/search/engines/stackoverflow.js.map +1 -0
  768. package/dist/search/engines/startpage.d.ts.map +1 -1
  769. package/dist/search/engines/startpage.js +65 -46
  770. package/dist/search/engines/startpage.js.map +1 -1
  771. package/dist/search/evidence.d.ts +25 -0
  772. package/dist/search/evidence.d.ts.map +1 -0
  773. package/dist/search/evidence.js +220 -0
  774. package/dist/search/evidence.js.map +1 -0
  775. package/dist/search/filters.d.ts.map +1 -1
  776. package/dist/search/filters.js +58 -54
  777. package/dist/search/filters.js.map +1 -1
  778. package/dist/search/find-similar/crawl-rank.d.ts +9 -0
  779. package/dist/search/find-similar/crawl-rank.d.ts.map +1 -0
  780. package/dist/search/find-similar/crawl-rank.js +272 -0
  781. package/dist/search/find-similar/crawl-rank.js.map +1 -0
  782. package/dist/search/find-similar/mode.d.ts +4 -0
  783. package/dist/search/find-similar/mode.d.ts.map +1 -0
  784. package/dist/search/find-similar/mode.js +12 -0
  785. package/dist/search/find-similar/mode.js.map +1 -0
  786. package/dist/search/find-similar.d.ts +5 -0
  787. package/dist/search/find-similar.d.ts.map +1 -0
  788. package/dist/search/find-similar.js +509 -0
  789. package/dist/search/find-similar.js.map +1 -0
  790. package/dist/search/highlights.d.ts +19 -0
  791. package/dist/search/highlights.d.ts.map +1 -0
  792. package/dist/search/highlights.js +167 -0
  793. package/dist/search/highlights.js.map +1 -0
  794. package/dist/search/language-filter.d.ts +29 -0
  795. package/dist/search/language-filter.d.ts.map +1 -0
  796. package/dist/search/language-filter.js +126 -0
  797. package/dist/search/language-filter.js.map +1 -0
  798. package/dist/search/legacy/searxng-orchestrator.d.ts +4 -0
  799. package/dist/search/legacy/searxng-orchestrator.d.ts.map +1 -0
  800. package/dist/search/legacy/searxng-orchestrator.js +501 -0
  801. package/dist/search/legacy/searxng-orchestrator.js.map +1 -0
  802. package/dist/search/legacy/searxng-provider.d.ts +7 -0
  803. package/dist/search/legacy/searxng-provider.d.ts.map +1 -0
  804. package/dist/search/legacy/searxng-provider.js +11 -0
  805. package/dist/search/legacy/searxng-provider.js.map +1 -0
  806. package/dist/search/multi-query.d.ts +25 -0
  807. package/dist/search/multi-query.d.ts.map +1 -0
  808. package/dist/search/multi-query.js +228 -0
  809. package/dist/search/multi-query.js.map +1 -0
  810. package/dist/search/query.js +32 -34
  811. package/dist/search/query.js.map +1 -1
  812. package/dist/search/rerank.d.ts +3 -1
  813. package/dist/search/rerank.d.ts.map +1 -1
  814. package/dist/search/rerank.js +44 -35
  815. package/dist/search/rerank.js.map +1 -1
  816. package/dist/search/reranker/authority-boost.d.ts +3 -0
  817. package/dist/search/reranker/authority-boost.d.ts.map +1 -0
  818. package/dist/search/reranker/authority-boost.js +179 -0
  819. package/dist/search/reranker/authority-boost.js.map +1 -0
  820. package/dist/search/reranker/consensus-boost.d.ts +3 -0
  821. package/dist/search/reranker/consensus-boost.d.ts.map +1 -0
  822. package/dist/search/reranker/consensus-boost.js +27 -0
  823. package/dist/search/reranker/consensus-boost.js.map +1 -0
  824. package/dist/search/reranker/recency-boost.d.ts +3 -0
  825. package/dist/search/reranker/recency-boost.d.ts.map +1 -0
  826. package/dist/search/reranker/recency-boost.js +13 -0
  827. package/dist/search/reranker/recency-boost.js.map +1 -0
  828. package/dist/search/reranker/recency.d.ts +3 -0
  829. package/dist/search/reranker/recency.d.ts.map +1 -0
  830. package/dist/search/reranker/recency.js +23 -0
  831. package/dist/search/reranker/recency.js.map +1 -0
  832. package/dist/search/reranker/transformers-rerank-provider.d.ts +13 -0
  833. package/dist/search/reranker/transformers-rerank-provider.d.ts.map +1 -0
  834. package/dist/search/reranker/transformers-rerank-provider.js +94 -0
  835. package/dist/search/reranker/transformers-rerank-provider.js.map +1 -0
  836. package/dist/search/rrf.d.ts +17 -0
  837. package/dist/search/rrf.d.ts.map +1 -0
  838. package/dist/search/rrf.js +39 -0
  839. package/dist/search/rrf.js.map +1 -0
  840. package/dist/search/sampling.d.ts +25 -0
  841. package/dist/search/sampling.d.ts.map +1 -0
  842. package/dist/search/sampling.js +52 -0
  843. package/dist/search/sampling.js.map +1 -0
  844. package/dist/search/searxng.d.ts.map +1 -1
  845. package/dist/search/searxng.js +69 -79
  846. package/dist/search/searxng.js.map +1 -1
  847. package/dist/search/tokens.d.ts +3 -0
  848. package/dist/search/tokens.d.ts.map +1 -0
  849. package/dist/search/tokens.js +39 -0
  850. package/dist/search/tokens.js.map +1 -0
  851. package/dist/search/truncate.d.ts +6 -0
  852. package/dist/search/truncate.d.ts.map +1 -0
  853. package/dist/search/truncate.js +26 -0
  854. package/dist/search/truncate.js.map +1 -0
  855. package/dist/search/url-unwrap.d.ts +3 -0
  856. package/dist/search/url-unwrap.d.ts.map +1 -0
  857. package/dist/search/url-unwrap.js +43 -0
  858. package/dist/search/url-unwrap.js.map +1 -0
  859. package/dist/search/v1/context-rank.d.ts +13 -0
  860. package/dist/search/v1/context-rank.d.ts.map +1 -0
  861. package/dist/search/v1/context-rank.js +74 -0
  862. package/dist/search/v1/context-rank.js.map +1 -0
  863. package/dist/search/v1/engine-base.d.ts +27 -0
  864. package/dist/search/v1/engine-base.d.ts.map +1 -0
  865. package/dist/search/v1/engine-base.js +110 -0
  866. package/dist/search/v1/engine-base.js.map +1 -0
  867. package/dist/search/v1/intent-router.d.ts +22 -0
  868. package/dist/search/v1/intent-router.d.ts.map +1 -0
  869. package/dist/search/v1/intent-router.js +138 -0
  870. package/dist/search/v1/intent-router.js.map +1 -0
  871. package/dist/search/v1/orchestrator.d.ts +24 -0
  872. package/dist/search/v1/orchestrator.d.ts.map +1 -0
  873. package/dist/search/v1/orchestrator.js +163 -0
  874. package/dist/search/v1/orchestrator.js.map +1 -0
  875. package/dist/search/v1/recency-boost.d.ts +9 -0
  876. package/dist/search/v1/recency-boost.d.ts.map +1 -0
  877. package/dist/search/v1/recency-boost.js +37 -0
  878. package/dist/search/v1/recency-boost.js.map +1 -0
  879. package/dist/search/v1/recent-cache-dedup.d.ts +6 -0
  880. package/dist/search/v1/recent-cache-dedup.d.ts.map +1 -0
  881. package/dist/search/v1/recent-cache-dedup.js +85 -0
  882. package/dist/search/v1/recent-cache-dedup.js.map +1 -0
  883. package/dist/search/v1/rss/feed-config.d.ts +21 -0
  884. package/dist/search/v1/rss/feed-config.d.ts.map +1 -0
  885. package/dist/search/v1/rss/feed-config.js +90 -0
  886. package/dist/search/v1/rss/feed-config.js.map +1 -0
  887. package/dist/search/v1/rss/feed-parser.d.ts +14 -0
  888. package/dist/search/v1/rss/feed-parser.d.ts.map +1 -0
  889. package/dist/search/v1/rss/feed-parser.js +104 -0
  890. package/dist/search/v1/rss/feed-parser.js.map +1 -0
  891. package/dist/search/v1/rss/feed-poller.d.ts +22 -0
  892. package/dist/search/v1/rss/feed-poller.d.ts.map +1 -0
  893. package/dist/search/v1/rss/feed-poller.js +102 -0
  894. package/dist/search/v1/rss/feed-poller.js.map +1 -0
  895. package/dist/search/v1/rss/feed-store.d.ts +30 -0
  896. package/dist/search/v1/rss/feed-store.d.ts.map +1 -0
  897. package/dist/search/v1/rss/feed-store.js +134 -0
  898. package/dist/search/v1/rss/feed-store.js.map +1 -0
  899. package/dist/search/v1/rss/rss-engine.d.ts +6 -0
  900. package/dist/search/v1/rss/rss-engine.d.ts.map +1 -0
  901. package/dist/search/v1/rss/rss-engine.js +28 -0
  902. package/dist/search/v1/rss/rss-engine.js.map +1 -0
  903. package/dist/search/v1/v1-provider.d.ts +7 -0
  904. package/dist/search/v1/v1-provider.d.ts.map +1 -0
  905. package/dist/search/v1/v1-provider.js +68 -0
  906. package/dist/search/v1/v1-provider.js.map +1 -0
  907. package/dist/search/v1/verticals/code.d.ts +4 -0
  908. package/dist/search/v1/verticals/code.d.ts.map +1 -0
  909. package/dist/search/v1/verticals/code.js +20 -0
  910. package/dist/search/v1/verticals/code.js.map +1 -0
  911. package/dist/search/v1/verticals/docs.d.ts +4 -0
  912. package/dist/search/v1/verticals/docs.d.ts.map +1 -0
  913. package/dist/search/v1/verticals/docs.js +20 -0
  914. package/dist/search/v1/verticals/docs.js.map +1 -0
  915. package/dist/search/v1/verticals/general.d.ts +4 -0
  916. package/dist/search/v1/verticals/general.d.ts.map +1 -0
  917. package/dist/search/v1/verticals/general.js +22 -0
  918. package/dist/search/v1/verticals/general.js.map +1 -0
  919. package/dist/search/v1/verticals/news.d.ts +10 -0
  920. package/dist/search/v1/verticals/news.d.ts.map +1 -0
  921. package/dist/search/v1/verticals/news.js +52 -0
  922. package/dist/search/v1/verticals/news.js.map +1 -0
  923. package/dist/search/v1/verticals/papers.d.ts +4 -0
  924. package/dist/search/v1/verticals/papers.d.ts.map +1 -0
  925. package/dist/search/v1/verticals/papers.js +23 -0
  926. package/dist/search/v1/verticals/papers.js.map +1 -0
  927. package/dist/search/validator.js +31 -31
  928. package/dist/search/validator.js.map +1 -1
  929. package/dist/searxng/bootstrap.d.ts +30 -0
  930. package/dist/searxng/bootstrap.d.ts.map +1 -1
  931. package/dist/searxng/bootstrap.js +223 -85
  932. package/dist/searxng/bootstrap.js.map +1 -1
  933. package/dist/searxng/docker.d.ts.map +1 -1
  934. package/dist/searxng/docker.js +69 -60
  935. package/dist/searxng/docker.js.map +1 -1
  936. package/dist/searxng/process.d.ts +13 -1
  937. package/dist/searxng/process.d.ts.map +1 -1
  938. package/dist/searxng/process.js +231 -164
  939. package/dist/searxng/process.js.map +1 -1
  940. package/dist/server/backend-status.d.ts +13 -0
  941. package/dist/server/backend-status.d.ts.map +1 -0
  942. package/dist/server/backend-status.js +40 -0
  943. package/dist/server/backend-status.js.map +1 -0
  944. package/dist/server/tool-schemas.d.ts +549 -0
  945. package/dist/server/tool-schemas.d.ts.map +1 -0
  946. package/dist/server/tool-schemas.js +464 -0
  947. package/dist/server/tool-schemas.js.map +1 -0
  948. package/dist/server/warmup-on-start.d.ts +9 -0
  949. package/dist/server/warmup-on-start.d.ts.map +1 -0
  950. package/dist/server/warmup-on-start.js +55 -0
  951. package/dist/server/warmup-on-start.js.map +1 -0
  952. package/dist/server.d.ts +17 -0
  953. package/dist/server.d.ts.map +1 -1
  954. package/dist/server.js +454 -297
  955. package/dist/server.js.map +1 -1
  956. package/dist/tools/agent.d.ts +5 -0
  957. package/dist/tools/agent.d.ts.map +1 -0
  958. package/dist/tools/agent.js +128 -0
  959. package/dist/tools/agent.js.map +1 -0
  960. package/dist/tools/cache.d.ts +2 -1
  961. package/dist/tools/cache.d.ts.map +1 -1
  962. package/dist/tools/cache.js +177 -44
  963. package/dist/tools/cache.js.map +1 -1
  964. package/dist/tools/crawl.d.ts.map +1 -1
  965. package/dist/tools/crawl.js +171 -88
  966. package/dist/tools/crawl.js.map +1 -1
  967. package/dist/tools/extract.d.ts +2 -2
  968. package/dist/tools/extract.d.ts.map +1 -1
  969. package/dist/tools/extract.js +175 -59
  970. package/dist/tools/extract.js.map +1 -1
  971. package/dist/tools/fetch.d.ts +2 -2
  972. package/dist/tools/fetch.d.ts.map +1 -1
  973. package/dist/tools/fetch.js +174 -68
  974. package/dist/tools/fetch.js.map +1 -1
  975. package/dist/tools/find-similar.d.ts +5 -0
  976. package/dist/tools/find-similar.d.ts.map +1 -0
  977. package/dist/tools/find-similar.js +127 -0
  978. package/dist/tools/find-similar.js.map +1 -0
  979. package/dist/tools/research.d.ts +5 -0
  980. package/dist/tools/research.d.ts.map +1 -0
  981. package/dist/tools/research.js +107 -0
  982. package/dist/tools/research.js.map +1 -0
  983. package/dist/tools/search.d.ts +10 -2
  984. package/dist/tools/search.d.ts.map +1 -1
  985. package/dist/tools/search.js +13 -158
  986. package/dist/tools/search.js.map +1 -1
  987. package/dist/types.d.ts +350 -7
  988. package/dist/types.d.ts.map +1 -1
  989. package/dist/types.js +6 -1
  990. package/dist/types.js.map +1 -1
  991. package/dist/util/mode.d.ts +4 -0
  992. package/dist/util/mode.d.ts.map +1 -0
  993. package/dist/util/mode.js +34 -0
  994. package/dist/util/mode.js.map +1 -0
  995. package/package.json +78 -8
  996. package/dist/extraction/trafilatura.d.ts +0 -6
  997. package/dist/extraction/trafilatura.d.ts.map +0 -1
  998. package/dist/extraction/trafilatura.js +0 -105
  999. package/dist/extraction/trafilatura.js.map +0 -1
  1000. package/dist/search/flashrank.d.ts +0 -12
  1001. package/dist/search/flashrank.d.ts.map +0 -1
  1002. package/dist/search/flashrank.js +0 -63
  1003. package/dist/search/flashrank.js.map +0 -1
@@ -0,0 +1 @@
1
+ {"version":3,"sources":["../../src/fetch/browser-types.ts"],"sourcesContent":["import type { BrowserType } from '../types.js';\n\nconst VALID_BROWSER_TYPES: ReadonlySet<string> = new Set(['chromium', 'firefox', 'webkit', 'lightpanda']);\n\nconst DEFAULT_BROWSER_TYPES: BrowserType[] = ['chromium'];\n\nfunction warn(msg: string, data: Record<string, unknown>): void {\n const line = JSON.stringify({ ts: new Date().toISOString(), level: 'warn', msg, module: 'fetch', data });\n process.stderr.write(line + '\\n');\n}\n\nexport function parseBrowserTypes(input: string | undefined | null): BrowserType[] {\n if (!input || typeof input !== 'string') {\n return [...DEFAULT_BROWSER_TYPES];\n }\n\n const trimmed = input.trim();\n if (trimmed.length === 0) {\n return [...DEFAULT_BROWSER_TYPES];\n }\n\n const parts = trimmed\n .split(',')\n .map(s => s.trim())\n .filter(s => s.length > 0);\n\n if (parts.length === 0) {\n return [...DEFAULT_BROWSER_TYPES];\n }\n\n const seen = new Set<string>();\n const valid: BrowserType[] = [];\n const invalid: string[] = [];\n\n for (const part of parts) {\n if (!VALID_BROWSER_TYPES.has(part)) {\n invalid.push(part);\n continue;\n }\n if (seen.has(part)) {\n continue;\n }\n seen.add(part);\n valid.push(part as BrowserType);\n }\n\n if (invalid.length > 0) {\n warn('ignored invalid browser types', { invalid, valid: [...valid] });\n }\n\n if (valid.length === 0) {\n warn('no valid browser types found, falling back to chromium', { input });\n return [...DEFAULT_BROWSER_TYPES];\n }\n\n return valid;\n}\n"],"mappings":"AAEA,MAAM,sBAA2C,oBAAI,IAAI,CAAC,YAAY,WAAW,UAAU,YAAY,CAAC;AAExG,MAAM,wBAAuC,CAAC,UAAU;AAExD,SAAS,KAAK,KAAa,MAAqC;AAC9D,QAAM,OAAO,KAAK,UAAU,EAAE,KAAI,oBAAI,KAAK,GAAE,YAAY,GAAG,OAAO,QAAQ,KAAK,QAAQ,SAAS,KAAK,CAAC;AACvG,UAAQ,OAAO,MAAM,OAAO,IAAI;AAClC;AAEO,SAAS,kBAAkB,OAAiD;AACjF,MAAI,CAAC,SAAS,OAAO,UAAU,UAAU;AACvC,WAAO,CAAC,GAAG,qBAAqB;AAAA,EAClC;AAEA,QAAM,UAAU,MAAM,KAAK;AAC3B,MAAI,QAAQ,WAAW,GAAG;AACxB,WAAO,CAAC,GAAG,qBAAqB;AAAA,EAClC;AAEA,QAAM,QAAQ,QACX,MAAM,GAAG,EACT,IAAI,OAAK,EAAE,KAAK,CAAC,EACjB,OAAO,OAAK,EAAE,SAAS,CAAC;AAE3B,MAAI,MAAM,WAAW,GAAG;AACtB,WAAO,CAAC,GAAG,qBAAqB;AAAA,EAClC;AAEA,QAAM,OAAO,oBAAI,IAAY;AAC7B,QAAM,QAAuB,CAAC;AAC9B,QAAM,UAAoB,CAAC;AAE3B,aAAW,QAAQ,OAAO;AACxB,QAAI,CAAC,oBAAoB,IAAI,IAAI,GAAG;AAClC,cAAQ,KAAK,IAAI;AACjB;AAAA,IACF;AACA,QAAI,KAAK,IAAI,IAAI,GAAG;AAClB;AAAA,IACF;AACA,SAAK,IAAI,IAAI;AACb,UAAM,KAAK,IAAmB;AAAA,EAChC;AAEA,MAAI,QAAQ,SAAS,GAAG;AACtB,SAAK,iCAAiC,EAAE,SAAS,OAAO,CAAC,GAAG,KAAK,EAAE,CAAC;AAAA,EACtE;AAEA,MAAI,MAAM,WAAW,GAAG;AACtB,SAAK,0DAA0D,EAAE,MAAM,CAAC;AACxE,WAAO,CAAC,GAAG,qBAAqB;AAAA,EAClC;AAEA,SAAO;AACT;","names":[]}
@@ -0,0 +1,9 @@
1
+ import type { CDPSession } from '../types.js';
2
+ export interface DiscoverOptions {
3
+ timeoutMs?: number;
4
+ filterPages?: boolean;
5
+ }
6
+ export declare function parseCDPResponse(raw: string): CDPSession[];
7
+ export declare function discoverSessions(cdpUrl: string, options?: DiscoverOptions): Promise<CDPSession[]>;
8
+ export declare function isCDPReachable(cdpUrl: string, timeoutMs?: number): Promise<boolean>;
9
+ //# sourceMappingURL=cdp-client.d.ts.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"cdp-client.d.ts","sourceRoot":"","sources":["../../src/fetch/cdp-client.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AAM9C,MAAM,WAAW,eAAe;IAC9B,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,WAAW,CAAC,EAAE,OAAO,CAAC;CACvB;AAiCD,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,MAAM,GAAG,UAAU,EAAE,CA2B1D;AAED,wBAAsB,gBAAgB,CACpC,MAAM,EAAE,MAAM,EACd,OAAO,CAAC,EAAE,eAAe,GACxB,OAAO,CAAC,UAAU,EAAE,CAAC,CA6BvB;AAED,wBAAsB,cAAc,CAAC,MAAM,EAAE,MAAM,EAAE,SAAS,SAAqB,GAAG,OAAO,CAAC,OAAO,CAAC,CAQrG"}
@@ -0,0 +1,89 @@
1
+ import * as http from "node:http";
2
+ import { createLogger } from "../logger.js";
3
+ const log = createLogger("fetch");
4
+ const DEFAULT_TIMEOUT_MS = 3e3;
5
+ function httpGet(url, timeoutMs) {
6
+ return new Promise((resolve, reject) => {
7
+ const req = http.get(url, { timeout: timeoutMs }, (res) => {
8
+ let body = "";
9
+ res.on("data", (chunk) => {
10
+ body += chunk;
11
+ });
12
+ res.on("end", () => {
13
+ resolve({ statusCode: res.statusCode ?? 0, body });
14
+ });
15
+ });
16
+ req.on("error", (err) => {
17
+ reject(err);
18
+ });
19
+ req.on("timeout", () => {
20
+ req.destroy();
21
+ reject(new Error(`CDP request timed out after ${timeoutMs}ms`));
22
+ });
23
+ });
24
+ }
25
+ function parseCDPResponse(raw) {
26
+ try {
27
+ const parsed = JSON.parse(raw);
28
+ if (!Array.isArray(parsed)) {
29
+ log.debug("CDP response is not an array");
30
+ return [];
31
+ }
32
+ return parsed.filter((entry) => {
33
+ if (!entry || typeof entry !== "object") return false;
34
+ if (!entry.id || typeof entry.id !== "string") return false;
35
+ if (!entry.webSocketDebuggerUrl || typeof entry.webSocketDebuggerUrl !== "string") return false;
36
+ return true;
37
+ }).map((entry) => ({
38
+ id: entry.id,
39
+ url: entry.url ?? "",
40
+ title: entry.title ?? "",
41
+ webSocketDebuggerUrl: entry.webSocketDebuggerUrl,
42
+ type: entry.type,
43
+ devtoolsFrontendUrl: entry.devtoolsFrontendUrl
44
+ }));
45
+ } catch (err) {
46
+ log.debug("failed to parse CDP response", { error: err instanceof Error ? err.message : String(err) });
47
+ return [];
48
+ }
49
+ }
50
+ async function discoverSessions(cdpUrl, options) {
51
+ const timeoutMs = options?.timeoutMs ?? DEFAULT_TIMEOUT_MS;
52
+ const filterPages = options?.filterPages ?? false;
53
+ const jsonUrl = cdpUrl.endsWith("/") ? `${cdpUrl}json` : `${cdpUrl}/json`;
54
+ try {
55
+ log.debug("discovering CDP sessions", { url: jsonUrl, timeoutMs });
56
+ const { statusCode, body } = await httpGet(jsonUrl, timeoutMs);
57
+ if (statusCode !== 200) {
58
+ log.warn("CDP endpoint returned non-200", { statusCode, url: jsonUrl });
59
+ return [];
60
+ }
61
+ let sessions = parseCDPResponse(body);
62
+ if (filterPages) {
63
+ sessions = sessions.filter((s) => s.type === "page" || s.type === void 0);
64
+ }
65
+ log.info("CDP sessions discovered", { count: sessions.length, url: jsonUrl });
66
+ return sessions;
67
+ } catch (err) {
68
+ log.debug("CDP discovery failed", {
69
+ url: jsonUrl,
70
+ error: err instanceof Error ? err.message : String(err)
71
+ });
72
+ return [];
73
+ }
74
+ }
75
+ async function isCDPReachable(cdpUrl, timeoutMs = DEFAULT_TIMEOUT_MS) {
76
+ const jsonUrl = cdpUrl.endsWith("/") ? `${cdpUrl}json` : `${cdpUrl}/json`;
77
+ try {
78
+ const { statusCode } = await httpGet(jsonUrl, timeoutMs);
79
+ return statusCode === 200;
80
+ } catch {
81
+ return false;
82
+ }
83
+ }
84
+ export {
85
+ discoverSessions,
86
+ isCDPReachable,
87
+ parseCDPResponse
88
+ };
89
+ //# sourceMappingURL=cdp-client.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"sources":["../../src/fetch/cdp-client.ts"],"sourcesContent":["import * as http from 'node:http';\nimport { createLogger } from '../logger.js';\nimport type { CDPSession } from '../types.js';\n\nconst log = createLogger('fetch');\n\nconst DEFAULT_TIMEOUT_MS = 3000;\n\nexport interface DiscoverOptions {\n timeoutMs?: number;\n filterPages?: boolean;\n}\n\nfunction httpGet(url: string, timeoutMs: number): Promise<{ statusCode: number; body: string }> {\n return new Promise((resolve, reject) => {\n const req = http.get(url, { timeout: timeoutMs }, (res) => {\n let body = '';\n res.on('data', (chunk: string) => { body += chunk; });\n res.on('end', () => {\n resolve({ statusCode: res.statusCode ?? 0, body });\n });\n });\n\n req.on('error', (err) => {\n reject(err);\n });\n\n req.on('timeout', () => {\n req.destroy();\n reject(new Error(`CDP request timed out after ${timeoutMs}ms`));\n });\n });\n}\n\ninterface RawCDPEntry {\n id?: string;\n url?: string;\n title?: string;\n webSocketDebuggerUrl?: string;\n type?: string;\n devtoolsFrontendUrl?: string;\n description?: string;\n}\n\nexport function parseCDPResponse(raw: string): CDPSession[] {\n try {\n const parsed = JSON.parse(raw);\n if (!Array.isArray(parsed)) {\n log.debug('CDP response is not an array');\n return [];\n }\n\n return parsed\n .filter((entry: RawCDPEntry) => {\n if (!entry || typeof entry !== 'object') return false;\n if (!entry.id || typeof entry.id !== 'string') return false;\n if (!entry.webSocketDebuggerUrl || typeof entry.webSocketDebuggerUrl !== 'string') return false;\n return true;\n })\n .map((entry: RawCDPEntry) => ({\n id: entry.id!,\n url: entry.url ?? '',\n title: entry.title ?? '',\n webSocketDebuggerUrl: entry.webSocketDebuggerUrl!,\n type: entry.type,\n devtoolsFrontendUrl: entry.devtoolsFrontendUrl,\n }));\n } catch (err) {\n log.debug('failed to parse CDP response', { error: err instanceof Error ? err.message : String(err) });\n return [];\n }\n}\n\nexport async function discoverSessions(\n cdpUrl: string,\n options?: DiscoverOptions,\n): Promise<CDPSession[]> {\n const timeoutMs = options?.timeoutMs ?? DEFAULT_TIMEOUT_MS;\n const filterPages = options?.filterPages ?? false;\n const jsonUrl = cdpUrl.endsWith('/') ? `${cdpUrl}json` : `${cdpUrl}/json`;\n\n try {\n log.debug('discovering CDP sessions', { url: jsonUrl, timeoutMs });\n const { statusCode, body } = await httpGet(jsonUrl, timeoutMs);\n\n if (statusCode !== 200) {\n log.warn('CDP endpoint returned non-200', { statusCode, url: jsonUrl });\n return [];\n }\n\n let sessions = parseCDPResponse(body);\n\n if (filterPages) {\n sessions = sessions.filter(s => s.type === 'page' || s.type === undefined);\n }\n\n log.info('CDP sessions discovered', { count: sessions.length, url: jsonUrl });\n return sessions;\n } catch (err) {\n log.debug('CDP discovery failed', {\n url: jsonUrl,\n error: err instanceof Error ? err.message : String(err),\n });\n return [];\n }\n}\n\nexport async function isCDPReachable(cdpUrl: string, timeoutMs = DEFAULT_TIMEOUT_MS): Promise<boolean> {\n const jsonUrl = cdpUrl.endsWith('/') ? `${cdpUrl}json` : `${cdpUrl}/json`;\n try {\n const { statusCode } = await httpGet(jsonUrl, timeoutMs);\n return statusCode === 200;\n } catch {\n return false;\n }\n}\n"],"mappings":"AAAA,YAAY,UAAU;AACtB,SAAS,oBAAoB;AAG7B,MAAM,MAAM,aAAa,OAAO;AAEhC,MAAM,qBAAqB;AAO3B,SAAS,QAAQ,KAAa,WAAkE;AAC9F,SAAO,IAAI,QAAQ,CAAC,SAAS,WAAW;AACtC,UAAM,MAAM,KAAK,IAAI,KAAK,EAAE,SAAS,UAAU,GAAG,CAAC,QAAQ;AACzD,UAAI,OAAO;AACX,UAAI,GAAG,QAAQ,CAAC,UAAkB;AAAE,gBAAQ;AAAA,MAAO,CAAC;AACpD,UAAI,GAAG,OAAO,MAAM;AAClB,gBAAQ,EAAE,YAAY,IAAI,cAAc,GAAG,KAAK,CAAC;AAAA,MACnD,CAAC;AAAA,IACH,CAAC;AAED,QAAI,GAAG,SAAS,CAAC,QAAQ;AACvB,aAAO,GAAG;AAAA,IACZ,CAAC;AAED,QAAI,GAAG,WAAW,MAAM;AACtB,UAAI,QAAQ;AACZ,aAAO,IAAI,MAAM,+BAA+B,SAAS,IAAI,CAAC;AAAA,IAChE,CAAC;AAAA,EACH,CAAC;AACH;AAYO,SAAS,iBAAiB,KAA2B;AAC1D,MAAI;AACF,UAAM,SAAS,KAAK,MAAM,GAAG;AAC7B,QAAI,CAAC,MAAM,QAAQ,MAAM,GAAG;AAC1B,UAAI,MAAM,8BAA8B;AACxC,aAAO,CAAC;AAAA,IACV;AAEA,WAAO,OACJ,OAAO,CAAC,UAAuB;AAC9B,UAAI,CAAC,SAAS,OAAO,UAAU,SAAU,QAAO;AAChD,UAAI,CAAC,MAAM,MAAM,OAAO,MAAM,OAAO,SAAU,QAAO;AACtD,UAAI,CAAC,MAAM,wBAAwB,OAAO,MAAM,yBAAyB,SAAU,QAAO;AAC1F,aAAO;AAAA,IACT,CAAC,EACA,IAAI,CAAC,WAAwB;AAAA,MAC5B,IAAI,MAAM;AAAA,MACV,KAAK,MAAM,OAAO;AAAA,MAClB,OAAO,MAAM,SAAS;AAAA,MACtB,sBAAsB,MAAM;AAAA,MAC5B,MAAM,MAAM;AAAA,MACZ,qBAAqB,MAAM;AAAA,IAC7B,EAAE;AAAA,EACN,SAAS,KAAK;AACZ,QAAI,MAAM,gCAAgC,EAAE,OAAO,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,EAAE,CAAC;AACrG,WAAO,CAAC;AAAA,EACV;AACF;AAEA,eAAsB,iBACpB,QACA,SACuB;AACvB,QAAM,YAAY,SAAS,aAAa;AACxC,QAAM,cAAc,SAAS,eAAe;AAC5C,QAAM,UAAU,OAAO,SAAS,GAAG,IAAI,GAAG,MAAM,SAAS,GAAG,MAAM;AAElE,MAAI;AACF,QAAI,MAAM,4BAA4B,EAAE,KAAK,SAAS,UAAU,CAAC;AACjE,UAAM,EAAE,YAAY,KAAK,IAAI,MAAM,QAAQ,SAAS,SAAS;AAE7D,QAAI,eAAe,KAAK;AACtB,UAAI,KAAK,iCAAiC,EAAE,YAAY,KAAK,QAAQ,CAAC;AACtE,aAAO,CAAC;AAAA,IACV;AAEA,QAAI,WAAW,iBAAiB,IAAI;AAEpC,QAAI,aAAa;AACf,iBAAW,SAAS,OAAO,OAAK,EAAE,SAAS,UAAU,EAAE,SAAS,MAAS;AAAA,IAC3E;AAEA,QAAI,KAAK,2BAA2B,EAAE,OAAO,SAAS,QAAQ,KAAK,QAAQ,CAAC;AAC5E,WAAO;AAAA,EACT,SAAS,KAAK;AACZ,QAAI,MAAM,wBAAwB;AAAA,MAChC,KAAK;AAAA,MACL,OAAO,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;AAAA,IACxD,CAAC;AACD,WAAO,CAAC;AAAA,EACV;AACF;AAEA,eAAsB,eAAe,QAAgB,YAAY,oBAAsC;AACrG,QAAM,UAAU,OAAO,SAAS,GAAG,IAAI,GAAG,MAAM,SAAS,GAAG,MAAM;AAClE,MAAI;AACF,UAAM,EAAE,WAAW,IAAI,MAAM,QAAQ,SAAS,SAAS;AACvD,WAAO,eAAe;AAAA,EACxB,QAAQ;AACN,WAAO;AAAA,EACT;AACF;","names":[]}
@@ -1,62 +1,55 @@
1
1
  const VISIBLE_TEXT_THRESHOLD = 200;
2
2
  const SCRIPT_RATIO_THRESHOLD = 0.8;
3
3
  function stripScriptsAndStyles(html) {
4
- return html
5
- .replace(/<script[\s\S]*?<\/script>/gi, '')
6
- .replace(/<style[\s\S]*?<\/style>/gi, '');
4
+ return html.replace(/<script[\s\S]*?<\/script>/gi, "").replace(/<style[\s\S]*?<\/style>/gi, "");
7
5
  }
8
6
  function extractVisibleText(html) {
9
- const stripped = stripScriptsAndStyles(html);
10
- const noTags = stripped.replace(/<[^>]+>/g, ' ');
11
- return noTags.replace(/\s+/g, ' ').trim();
7
+ const stripped = stripScriptsAndStyles(html);
8
+ const noTags = stripped.replace(/<[^>]+>/g, " ");
9
+ return noTags.replace(/\s+/g, " ").trim();
12
10
  }
13
11
  function hasSpaShellIndicator(html) {
14
- const spaPatterns = [
15
- /<div[^>]+id=["']root["'][^>]*>\s*<\/div>/i,
16
- /<div[^>]+id=["']app["'][^>]*>\s*<\/div>/i,
17
- /<div[^>]+id=["']__next["'][^>]*>\s*<\/div>/i,
18
- ];
19
- return spaPatterns.some((pattern) => pattern.test(html));
12
+ const spaPatterns = [
13
+ /<div[^>]+id=["']root["'][^>]*>\s*<\/div>/i,
14
+ /<div[^>]+id=["']app["'][^>]*>\s*<\/div>/i,
15
+ /<div[^>]+id=["']__next["'][^>]*>\s*<\/div>/i
16
+ ];
17
+ return spaPatterns.some((pattern) => pattern.test(html));
20
18
  }
21
19
  function hasNextData(html) {
22
- if (!/__NEXT_DATA__/.test(html))
23
- return false;
24
- const withoutScripts = stripScriptsAndStyles(html);
25
- const visibleText = withoutScripts.replace(/<[^>]+>/g, ' ').replace(/\s+/g, ' ').trim();
26
- return visibleText.length < VISIBLE_TEXT_THRESHOLD;
20
+ if (!/__NEXT_DATA__/.test(html)) return false;
21
+ const withoutScripts = stripScriptsAndStyles(html);
22
+ const visibleText = withoutScripts.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
23
+ return visibleText.length < VISIBLE_TEXT_THRESHOLD;
27
24
  }
28
25
  function hasNoscriptRequired(html) {
29
- const noscriptMatches = html.match(/<noscript[^>]*>([\s\S]*?)<\/noscript>/gi);
30
- if (!noscriptMatches)
31
- return false;
32
- return noscriptMatches.some((tag) => {
33
- const inner = tag.replace(/<[^>]+>/g, '').toLowerCase();
34
- return inner.includes('javascript') || inner.includes('enable');
35
- });
26
+ const noscriptMatches = html.match(/<noscript[^>]*>([\s\S]*?)<\/noscript>/gi);
27
+ if (!noscriptMatches) return false;
28
+ return noscriptMatches.some((tag) => {
29
+ const inner = tag.replace(/<[^>]+>/g, "").toLowerCase();
30
+ return inner.includes("javascript") || inner.includes("enable");
31
+ });
36
32
  }
37
33
  function hasHighScriptRatio(html) {
38
- const bodyMatch = html.match(/<body[^>]*>([\s\S]*?)<\/body>/i);
39
- const bodyContent = bodyMatch ? bodyMatch[1] : html;
40
- const scriptMatches = bodyContent.match(/<script[\s\S]*?<\/script>/gi) ?? [];
41
- const scriptText = scriptMatches.join('');
42
- const scriptLen = scriptText.length;
43
- const totalLen = bodyContent.length;
44
- if (totalLen === 0)
45
- return false;
46
- return scriptLen / totalLen > SCRIPT_RATIO_THRESHOLD;
34
+ const bodyMatch = html.match(/<body[^>]*>([\s\S]*?)<\/body>/i);
35
+ const bodyContent = bodyMatch ? bodyMatch[1] : html;
36
+ const scriptMatches = bodyContent.match(/<script[\s\S]*?<\/script>/gi) ?? [];
37
+ const scriptText = scriptMatches.join("");
38
+ const scriptLen = scriptText.length;
39
+ const totalLen = bodyContent.length;
40
+ if (totalLen === 0) return false;
41
+ return scriptLen / totalLen > SCRIPT_RATIO_THRESHOLD;
47
42
  }
48
- export function contentAppearsEmpty(html) {
49
- const visibleText = extractVisibleText(html);
50
- if (visibleText.length < VISIBLE_TEXT_THRESHOLD)
51
- return true;
52
- if (hasSpaShellIndicator(html))
53
- return true;
54
- if (hasNextData(html))
55
- return true;
56
- if (hasNoscriptRequired(html))
57
- return true;
58
- if (hasHighScriptRatio(html))
59
- return true;
60
- return false;
43
+ function contentAppearsEmpty(html) {
44
+ const visibleText = extractVisibleText(html);
45
+ if (visibleText.length < VISIBLE_TEXT_THRESHOLD) return true;
46
+ if (hasSpaShellIndicator(html)) return true;
47
+ if (hasNextData(html)) return true;
48
+ if (hasNoscriptRequired(html)) return true;
49
+ if (hasHighScriptRatio(html)) return true;
50
+ return false;
61
51
  }
52
+ export {
53
+ contentAppearsEmpty
54
+ };
62
55
  //# sourceMappingURL=content-check.js.map
@@ -1 +1 @@
1
- {"version":3,"file":"content-check.js","sourceRoot":"","sources":["../../src/fetch/content-check.ts"],"names":[],"mappings":"AAAA,MAAM,sBAAsB,GAAG,GAAG,CAAC;AACnC,MAAM,sBAAsB,GAAG,GAAG,CAAC;AAEnC,SAAS,qBAAqB,CAAC,IAAY;IACzC,OAAO,IAAI;SACR,OAAO,CAAC,6BAA6B,EAAE,EAAE,CAAC;SAC1C,OAAO,CAAC,2BAA2B,EAAE,EAAE,CAAC,CAAC;AAC9C,CAAC;AAED,SAAS,kBAAkB,CAAC,IAAY;IACtC,MAAM,QAAQ,GAAG,qBAAqB,CAAC,IAAI,CAAC,CAAC;IAC7C,MAAM,MAAM,GAAG,QAAQ,CAAC,OAAO,CAAC,UAAU,EAAE,GAAG,CAAC,CAAC;IACjD,OAAO,MAAM,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;AAC5C,CAAC;AAED,SAAS,oBAAoB,CAAC,IAAY;IACxC,MAAM,WAAW,GAAG;QAClB,2CAA2C;QAC3C,0CAA0C;QAC1C,6CAA6C;KAC9C,CAAC;IACF,OAAO,WAAW,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;AAC3D,CAAC;AAED,SAAS,WAAW,CAAC,IAAY;IAC/B,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IAC9C,MAAM,cAAc,GAAG,qBAAqB,CAAC,IAAI,CAAC,CAAC;IACnD,MAAM,WAAW,GAAG,cAAc,CAAC,OAAO,CAAC,UAAU,EAAE,GAAG,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;IACxF,OAAO,WAAW,CAAC,MAAM,GAAG,sBAAsB,CAAC;AACrD,CAAC;AAED,SAAS,mBAAmB,CAAC,IAAY;IACvC,MAAM,eAAe,GAAG,IAAI,CAAC,KAAK,CAAC,yCAAyC,CAAC,CAAC;IAC9E,IAAI,CAAC,eAAe;QAAE,OAAO,KAAK,CAAC;IACnC,OAAO,eAAe,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE;QAClC,MAAM,KAAK,GAAG,GAAG,CAAC,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC,WAAW,EAAE,CAAC;QACxD,OAAO,KAAK,CAAC,QAAQ,CAAC,YAAY,CAAC,IAAI,KAAK,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC;IAClE,CAAC,CAAC,CAAC;AACL,CAAC;AAED,SAAS,kBAAkB,CAAC,IAAY;IACtC,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,gCAAgC,CAAC,CAAC;IAC/D,MAAM,WAAW,GAAG,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;IAEpD,MAAM,aAAa,GAAG,WAAW,CAAC,KAAK,CAAC,6BAA6B,CAAC,IAAI,EAAE,CAAC;IAC7E,MAAM,UAAU,GAAG,aAAa,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAE1C,MAAM,SAAS,GAAG,UAAU,CAAC,MAAM,CAAC;IACpC,MAAM,QAAQ,GAAG,WAAW,CAAC,MAAM,CAAC;IAEpC,IAAI,QAAQ,KAAK,CAAC;QAAE,OAAO,KAAK,CAAC;IACjC,OAAO,SAAS,GAAG,QAAQ,GAAG,sBAAsB,CAAC;AACvD,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,IAAY;IAC9C,MAAM,WAAW,GAAG,kBAAkB,CAAC,IAAI,CAAC,CAAC;IAC7C,IAAI,WAAW,CAAC,MAAM,GAAG,sBAAsB;QAAE,OAAO,IAAI,CAAC;IAE7D,IAAI,oBAAoB,CAAC,IAAI,CAAC;QAAE,OAAO,IAAI,CAAC;IAC5C,IAAI,WAAW,CAAC,IAAI,CAAC;QAAE,OAAO,IAAI,CAAC;IACnC,IAAI,mBAAmB,CAAC,IAAI,CAAC;QAAE,OAAO,IAAI,CAAC;IAC3C,IAAI,kBAAkB,CAAC,IAAI,CAAC;QAAE,OAAO,IAAI,CAAC;IAE1C,OAAO,KAAK,CAAC;AACf,CAAC"}
1
+ {"version":3,"sources":["../../src/fetch/content-check.ts"],"sourcesContent":["const VISIBLE_TEXT_THRESHOLD = 200;\nconst SCRIPT_RATIO_THRESHOLD = 0.8;\n\nfunction stripScriptsAndStyles(html: string): string {\n return html\n .replace(/<script[\\s\\S]*?<\\/script>/gi, '')\n .replace(/<style[\\s\\S]*?<\\/style>/gi, '');\n}\n\nfunction extractVisibleText(html: string): string {\n const stripped = stripScriptsAndStyles(html);\n const noTags = stripped.replace(/<[^>]+>/g, ' ');\n return noTags.replace(/\\s+/g, ' ').trim();\n}\n\nfunction hasSpaShellIndicator(html: string): boolean {\n const spaPatterns = [\n /<div[^>]+id=[\"']root[\"'][^>]*>\\s*<\\/div>/i,\n /<div[^>]+id=[\"']app[\"'][^>]*>\\s*<\\/div>/i,\n /<div[^>]+id=[\"']__next[\"'][^>]*>\\s*<\\/div>/i,\n ];\n return spaPatterns.some((pattern) => pattern.test(html));\n}\n\nfunction hasNextData(html: string): boolean {\n if (!/__NEXT_DATA__/.test(html)) return false;\n const withoutScripts = stripScriptsAndStyles(html);\n const visibleText = withoutScripts.replace(/<[^>]+>/g, ' ').replace(/\\s+/g, ' ').trim();\n return visibleText.length < VISIBLE_TEXT_THRESHOLD;\n}\n\nfunction hasNoscriptRequired(html: string): boolean {\n const noscriptMatches = html.match(/<noscript[^>]*>([\\s\\S]*?)<\\/noscript>/gi);\n if (!noscriptMatches) return false;\n return noscriptMatches.some((tag) => {\n const inner = tag.replace(/<[^>]+>/g, '').toLowerCase();\n return inner.includes('javascript') || inner.includes('enable');\n });\n}\n\nfunction hasHighScriptRatio(html: string): boolean {\n const bodyMatch = html.match(/<body[^>]*>([\\s\\S]*?)<\\/body>/i);\n const bodyContent = bodyMatch ? bodyMatch[1] : html;\n\n const scriptMatches = bodyContent.match(/<script[\\s\\S]*?<\\/script>/gi) ?? [];\n const scriptText = scriptMatches.join('');\n\n const scriptLen = scriptText.length;\n const totalLen = bodyContent.length;\n\n if (totalLen === 0) return false;\n return scriptLen / totalLen > SCRIPT_RATIO_THRESHOLD;\n}\n\nexport function contentAppearsEmpty(html: string): boolean {\n const visibleText = extractVisibleText(html);\n if (visibleText.length < VISIBLE_TEXT_THRESHOLD) return true;\n\n if (hasSpaShellIndicator(html)) return true;\n if (hasNextData(html)) return true;\n if (hasNoscriptRequired(html)) return true;\n if (hasHighScriptRatio(html)) return true;\n\n return false;\n}\n"],"mappings":"AAAA,MAAM,yBAAyB;AAC/B,MAAM,yBAAyB;AAE/B,SAAS,sBAAsB,MAAsB;AACnD,SAAO,KACJ,QAAQ,+BAA+B,EAAE,EACzC,QAAQ,6BAA6B,EAAE;AAC5C;AAEA,SAAS,mBAAmB,MAAsB;AAChD,QAAM,WAAW,sBAAsB,IAAI;AAC3C,QAAM,SAAS,SAAS,QAAQ,YAAY,GAAG;AAC/C,SAAO,OAAO,QAAQ,QAAQ,GAAG,EAAE,KAAK;AAC1C;AAEA,SAAS,qBAAqB,MAAuB;AACnD,QAAM,cAAc;AAAA,IAClB;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACA,SAAO,YAAY,KAAK,CAAC,YAAY,QAAQ,KAAK,IAAI,CAAC;AACzD;AAEA,SAAS,YAAY,MAAuB;AAC1C,MAAI,CAAC,gBAAgB,KAAK,IAAI,EAAG,QAAO;AACxC,QAAM,iBAAiB,sBAAsB,IAAI;AACjD,QAAM,cAAc,eAAe,QAAQ,YAAY,GAAG,EAAE,QAAQ,QAAQ,GAAG,EAAE,KAAK;AACtF,SAAO,YAAY,SAAS;AAC9B;AAEA,SAAS,oBAAoB,MAAuB;AAClD,QAAM,kBAAkB,KAAK,MAAM,yCAAyC;AAC5E,MAAI,CAAC,gBAAiB,QAAO;AAC7B,SAAO,gBAAgB,KAAK,CAAC,QAAQ;AACnC,UAAM,QAAQ,IAAI,QAAQ,YAAY,EAAE,EAAE,YAAY;AACtD,WAAO,MAAM,SAAS,YAAY,KAAK,MAAM,SAAS,QAAQ;AAAA,EAChE,CAAC;AACH;AAEA,SAAS,mBAAmB,MAAuB;AACjD,QAAM,YAAY,KAAK,MAAM,gCAAgC;AAC7D,QAAM,cAAc,YAAY,UAAU,CAAC,IAAI;AAE/C,QAAM,gBAAgB,YAAY,MAAM,6BAA6B,KAAK,CAAC;AAC3E,QAAM,aAAa,cAAc,KAAK,EAAE;AAExC,QAAM,YAAY,WAAW;AAC7B,QAAM,WAAW,YAAY;AAE7B,MAAI,aAAa,EAAG,QAAO;AAC3B,SAAO,YAAY,WAAW;AAChC;AAEO,SAAS,oBAAoB,MAAuB;AACzD,QAAM,cAAc,mBAAmB,IAAI;AAC3C,MAAI,YAAY,SAAS,uBAAwB,QAAO;AAExD,MAAI,qBAAqB,IAAI,EAAG,QAAO;AACvC,MAAI,YAAY,IAAI,EAAG,QAAO;AAC9B,MAAI,oBAAoB,IAAI,EAAG,QAAO;AACtC,MAAI,mBAAmB,IAAI,EAAG,QAAO;AAErC,SAAO;AACT;","names":[]}
@@ -0,0 +1,7 @@
1
+ interface DescribedError {
2
+ reason: string;
3
+ hint?: string;
4
+ }
5
+ export declare function describeFetchError(err: unknown): DescribedError;
6
+ export {};
7
+ //# sourceMappingURL=error-describe.d.ts.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"error-describe.d.ts","sourceRoot":"","sources":["../../src/fetch/error-describe.ts"],"names":[],"mappings":"AAKA,UAAU,cAAc;IACtB,MAAM,EAAE,MAAM,CAAC;IACf,IAAI,CAAC,EAAE,MAAM,CAAC;CACf;AAwBD,wBAAgB,kBAAkB,CAAC,GAAG,EAAE,OAAO,GAAG,cAAc,CAa/D"}
@@ -0,0 +1,37 @@
1
+ const CODE_DESCRIPTIONS = {
2
+ ENOTFOUND: { reason: "DNS resolution failed (ENOTFOUND)", hint: "Check the domain name; the host could not be resolved" },
3
+ ECONNREFUSED: { reason: "Connection refused (ECONNREFUSED)", hint: "Target host rejected the connection \u2014 server may be down or port closed" },
4
+ ECONNRESET: { reason: "Connection reset (ECONNRESET)", hint: "Remote peer closed the connection mid-request \u2014 retry may succeed" },
5
+ ETIMEDOUT: { reason: "Connection timed out (ETIMEDOUT)", hint: "Increase timeoutMs or check network reachability" },
6
+ EAI_AGAIN: { reason: "DNS lookup temporarily failed (EAI_AGAIN)", hint: "DNS resolver issue \u2014 retry shortly" },
7
+ EHOSTUNREACH: { reason: "Host unreachable (EHOSTUNREACH)" },
8
+ ENETUNREACH: { reason: "Network unreachable (ENETUNREACH)" },
9
+ CERT_HAS_EXPIRED: { reason: "TLS certificate expired" },
10
+ UNABLE_TO_VERIFY_LEAF_SIGNATURE: { reason: "TLS chain verification failed" },
11
+ SELF_SIGNED_CERT_IN_CHAIN: { reason: "Self-signed TLS certificate in chain" }
12
+ };
13
+ function extractCode(err, depth = 0) {
14
+ if (!err || depth > 5) return null;
15
+ if (typeof err !== "object") return null;
16
+ const e = err;
17
+ if (typeof e.code === "string" && e.code) return e.code;
18
+ if (e.cause) return extractCode(e.cause, depth + 1);
19
+ return null;
20
+ }
21
+ function describeFetchError(err) {
22
+ const code = extractCode(err);
23
+ if (code && CODE_DESCRIPTIONS[code]) return CODE_DESCRIPTIONS[code];
24
+ if (err instanceof Error) {
25
+ if (err.name === "TimeoutError" || err.name === "AbortError") {
26
+ return { reason: "Request timed out", hint: "Increase timeoutMs or check network reachability" };
27
+ }
28
+ const msg = err.message;
29
+ if (code) return { reason: `${msg} (${code})` };
30
+ return { reason: msg || "fetch failed" };
31
+ }
32
+ return { reason: String(err) };
33
+ }
34
+ export {
35
+ describeFetchError
36
+ };
37
+ //# sourceMappingURL=error-describe.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"sources":["../../src/fetch/error-describe.ts"],"sourcesContent":["// Translate a thrown fetch error into a stable, specific reason string.\n// Node's undici surfaces \"fetch failed\" on the outer TypeError while the\n// actual code (ENOTFOUND/ECONNREFUSED/etc.) hides on err.cause. Drill the\n// chain so callers see what actually broke instead of a generic phrase.\n\ninterface DescribedError {\n reason: string;\n hint?: string;\n}\n\nconst CODE_DESCRIPTIONS: Record<string, { reason: string; hint?: string }> = {\n ENOTFOUND: { reason: 'DNS resolution failed (ENOTFOUND)', hint: 'Check the domain name; the host could not be resolved' },\n ECONNREFUSED: { reason: 'Connection refused (ECONNREFUSED)', hint: 'Target host rejected the connection — server may be down or port closed' },\n ECONNRESET: { reason: 'Connection reset (ECONNRESET)', hint: 'Remote peer closed the connection mid-request — retry may succeed' },\n ETIMEDOUT: { reason: 'Connection timed out (ETIMEDOUT)', hint: 'Increase timeoutMs or check network reachability' },\n EAI_AGAIN: { reason: 'DNS lookup temporarily failed (EAI_AGAIN)', hint: 'DNS resolver issue — retry shortly' },\n EHOSTUNREACH: { reason: 'Host unreachable (EHOSTUNREACH)' },\n ENETUNREACH: { reason: 'Network unreachable (ENETUNREACH)' },\n CERT_HAS_EXPIRED: { reason: 'TLS certificate expired' },\n UNABLE_TO_VERIFY_LEAF_SIGNATURE: { reason: 'TLS chain verification failed' },\n SELF_SIGNED_CERT_IN_CHAIN: { reason: 'Self-signed TLS certificate in chain' },\n};\n\nfunction extractCode(err: unknown, depth = 0): string | null {\n if (!err || depth > 5) return null;\n if (typeof err !== 'object') return null;\n const e = err as { code?: unknown; cause?: unknown };\n if (typeof e.code === 'string' && e.code) return e.code;\n if (e.cause) return extractCode(e.cause, depth + 1);\n return null;\n}\n\nexport function describeFetchError(err: unknown): DescribedError {\n const code = extractCode(err);\n if (code && CODE_DESCRIPTIONS[code]) return CODE_DESCRIPTIONS[code];\n\n if (err instanceof Error) {\n if (err.name === 'TimeoutError' || err.name === 'AbortError') {\n return { reason: 'Request timed out', hint: 'Increase timeoutMs or check network reachability' };\n }\n const msg = err.message;\n if (code) return { reason: `${msg} (${code})` };\n return { reason: msg || 'fetch failed' };\n }\n return { reason: String(err) };\n}\n"],"mappings":"AAUA,MAAM,oBAAuE;AAAA,EAC3E,WAAW,EAAE,QAAQ,qCAAqC,MAAM,wDAAwD;AAAA,EACxH,cAAc,EAAE,QAAQ,qCAAqC,MAAM,+EAA0E;AAAA,EAC7I,YAAY,EAAE,QAAQ,iCAAiC,MAAM,yEAAoE;AAAA,EACjI,WAAW,EAAE,QAAQ,oCAAoC,MAAM,mDAAmD;AAAA,EAClH,WAAW,EAAE,QAAQ,6CAA6C,MAAM,0CAAqC;AAAA,EAC7G,cAAc,EAAE,QAAQ,kCAAkC;AAAA,EAC1D,aAAa,EAAE,QAAQ,oCAAoC;AAAA,EAC3D,kBAAkB,EAAE,QAAQ,0BAA0B;AAAA,EACtD,iCAAiC,EAAE,QAAQ,gCAAgC;AAAA,EAC3E,2BAA2B,EAAE,QAAQ,uCAAuC;AAC9E;AAEA,SAAS,YAAY,KAAc,QAAQ,GAAkB;AAC3D,MAAI,CAAC,OAAO,QAAQ,EAAG,QAAO;AAC9B,MAAI,OAAO,QAAQ,SAAU,QAAO;AACpC,QAAM,IAAI;AACV,MAAI,OAAO,EAAE,SAAS,YAAY,EAAE,KAAM,QAAO,EAAE;AACnD,MAAI,EAAE,MAAO,QAAO,YAAY,EAAE,OAAO,QAAQ,CAAC;AAClD,SAAO;AACT;AAEO,SAAS,mBAAmB,KAA8B;AAC/D,QAAM,OAAO,YAAY,GAAG;AAC5B,MAAI,QAAQ,kBAAkB,IAAI,EAAG,QAAO,kBAAkB,IAAI;AAElE,MAAI,eAAe,OAAO;AACxB,QAAI,IAAI,SAAS,kBAAkB,IAAI,SAAS,cAAc;AAC5D,aAAO,EAAE,QAAQ,qBAAqB,MAAM,mDAAmD;AAAA,IACjG;AACA,UAAM,MAAM,IAAI;AAChB,QAAI,KAAM,QAAO,EAAE,QAAQ,GAAG,GAAG,KAAK,IAAI,IAAI;AAC9C,WAAO,EAAE,QAAQ,OAAO,eAAe;AAAA,EACzC;AACA,SAAO,EAAE,QAAQ,OAAO,GAAG,EAAE;AAC/B;","names":[]}
@@ -1,6 +1,10 @@
1
1
  export interface HttpFetchOptions {
2
2
  headers?: Record<string, string>;
3
3
  timeoutMs?: number;
4
+ conditionalHeaders?: {
5
+ ifNoneMatch?: string;
6
+ ifModifiedSince?: string;
7
+ };
4
8
  }
5
9
  export interface HttpFetchResult {
6
10
  url: string;
@@ -1 +1 @@
1
- {"version":3,"file":"http-client.d.ts","sourceRoot":"","sources":["../../src/fetch/http-client.ts"],"names":[],"mappings":"AAGA,MAAM,WAAW,gBAAgB;IAC/B,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IACjC,SAAS,CAAC,EAAE,MAAM,CAAC;CACpB;AAED,MAAM,WAAW,eAAe;IAC9B,GAAG,EAAE,MAAM,CAAC;IACZ,QAAQ,EAAE,MAAM,CAAC;IACjB,IAAI,EAAE,MAAM,CAAC;IACb,WAAW,EAAE,MAAM,CAAC;IACpB,UAAU,EAAE,MAAM,CAAC;IACnB,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAChC,SAAS,CAAC,EAAE,MAAM,CAAC;CACpB;AAmCD,wBAAsB,SAAS,CAAC,GAAG,EAAE,MAAM,EAAE,OAAO,GAAE,gBAAqB,GAAG,OAAO,CAAC,eAAe,CAAC,CAyCrG"}
1
+ {"version":3,"file":"http-client.d.ts","sourceRoot":"","sources":["../../src/fetch/http-client.ts"],"names":[],"mappings":"AAGA,MAAM,WAAW,gBAAgB;IAC/B,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IACjC,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,kBAAkB,CAAC,EAAE;QACnB,WAAW,CAAC,EAAE,MAAM,CAAC;QACrB,eAAe,CAAC,EAAE,MAAM,CAAC;KAC1B,CAAC;CACH;AAED,MAAM,WAAW,eAAe;IAC9B,GAAG,EAAE,MAAM,CAAC;IACZ,QAAQ,EAAE,MAAM,CAAC;IACjB,IAAI,EAAE,MAAM,CAAC;IACb,WAAW,EAAE,MAAM,CAAC;IACpB,UAAU,EAAE,MAAM,CAAC;IACnB,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAChC,SAAS,CAAC,EAAE,MAAM,CAAC;CACpB;AAmCD,wBAAsB,SAAS,CAAC,GAAG,EAAE,MAAM,EAAE,OAAO,GAAE,gBAAqB,GAAG,OAAO,CAAC,eAAe,CAAC,CAyCrG"}
@@ -1,146 +1,165 @@
1
- import { getConfig } from '../config.js';
2
- import { createLogger } from '../logger.js';
3
- const RETRYABLE_STATUSES = new Set([429, 502, 503]);
4
- const RETRYABLE_ERROR_CODES = new Set(['ECONNRESET', 'ETIMEDOUT', 'ECONNREFUSED']);
5
- const REDIRECT_STATUSES = new Set([301, 302, 307, 308]);
1
+ import { getConfig } from "../config.js";
2
+ import { createLogger } from "../logger.js";
3
+ const RETRYABLE_STATUSES = /* @__PURE__ */ new Set([429, 502, 503]);
4
+ const RETRYABLE_ERROR_CODES = /* @__PURE__ */ new Set(["ECONNRESET", "ETIMEDOUT", "ECONNREFUSED"]);
5
+ const REDIRECT_STATUSES = /* @__PURE__ */ new Set([301, 302, 307, 308]);
6
6
  const DEFAULT_USER_AGENTS = [
7
- 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36',
8
- 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36',
9
- 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36',
7
+ "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36",
8
+ "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36",
9
+ "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/125.0.0.0 Safari/537.36"
10
10
  ];
11
11
  function getRotatingUserAgent(config) {
12
- if (config.userAgent)
13
- return config.userAgent;
14
- return DEFAULT_USER_AGENTS[Math.floor(Math.random() * DEFAULT_USER_AGENTS.length)];
12
+ if (config.userAgent) return config.userAgent;
13
+ return DEFAULT_USER_AGENTS[Math.floor(Math.random() * DEFAULT_USER_AGENTS.length)];
15
14
  }
16
15
  function isRetryableError(err) {
17
- if (err instanceof Error) {
18
- const code = err.code;
19
- if (code && RETRYABLE_ERROR_CODES.has(code))
20
- return true;
21
- // AbortSignal timeout throws DOMException with name TimeoutError
22
- if (err.name === 'TimeoutError')
23
- return true;
24
- }
25
- return false;
16
+ if (err instanceof Error) {
17
+ const code = err.code;
18
+ if (code && RETRYABLE_ERROR_CODES.has(code)) return true;
19
+ if (err.name === "TimeoutError") return true;
20
+ }
21
+ return false;
26
22
  }
27
23
  function backoffMs(attempt) {
28
- return 500 * Math.pow(2, attempt) + Math.random() * 500;
24
+ return 500 * Math.pow(2, attempt) + Math.random() * 500;
29
25
  }
30
26
  function sleep(ms) {
31
- return new Promise((resolve) => setTimeout(resolve, ms));
27
+ return new Promise((resolve) => setTimeout(resolve, ms));
32
28
  }
33
- export async function httpFetch(url, options = {}) {
34
- const config = getConfig();
35
- const logger = createLogger('fetch');
36
- const maxRetries = config.fetchMaxRetries;
37
- const timeoutMs = options.timeoutMs ?? config.fetchTimeoutMs;
38
- const maxRedirects = config.maxRedirects;
39
- let lastError;
40
- for (let attempt = 0; attempt <= maxRetries; attempt++) {
41
- if (attempt > 0) {
42
- const delay = backoffMs(attempt - 1);
43
- logger.debug('retrying after backoff', { attempt, delayMs: delay, url });
44
- await sleep(delay);
45
- }
46
- try {
47
- const result = await fetchWithRedirects(url, options, timeoutMs, maxRedirects, logger);
48
- return result;
49
- }
50
- catch (err) {
51
- lastError = err;
52
- if (err instanceof HttpFetchError && !err.retryable) {
53
- throw err;
54
- }
55
- const retryable = err instanceof HttpFetchError ? err.retryable : isRetryableError(err);
56
- if (!retryable || attempt >= maxRetries) {
57
- throw err;
58
- }
59
- logger.warn('fetch failed, will retry', {
60
- attempt,
61
- url,
62
- error: err instanceof Error ? err.message : String(err),
63
- });
64
- }
29
+ async function httpFetch(url, options = {}) {
30
+ const config = getConfig();
31
+ const logger = createLogger("fetch");
32
+ const maxRetries = config.fetchMaxRetries;
33
+ const timeoutMs = options.timeoutMs ?? config.fetchTimeoutMs;
34
+ const maxRedirects = config.maxRedirects;
35
+ let lastError;
36
+ for (let attempt = 0; attempt <= maxRetries; attempt++) {
37
+ if (attempt > 0) {
38
+ const delay = backoffMs(attempt - 1);
39
+ logger.debug("retrying after backoff", { attempt, delayMs: delay, url });
40
+ await sleep(delay);
41
+ }
42
+ try {
43
+ const result = await fetchWithRedirects(url, options, timeoutMs, maxRedirects, logger);
44
+ return result;
45
+ } catch (err) {
46
+ lastError = err;
47
+ if (err instanceof HttpFetchError && !err.retryable) {
48
+ throw err;
49
+ }
50
+ const retryable = err instanceof HttpFetchError ? err.retryable : isRetryableError(err);
51
+ if (!retryable || attempt >= maxRetries) {
52
+ throw err;
53
+ }
54
+ logger.warn("fetch failed, will retry", {
55
+ attempt,
56
+ url,
57
+ error: err instanceof Error ? err.message : String(err)
58
+ });
65
59
  }
66
- throw lastError;
60
+ }
61
+ throw lastError;
67
62
  }
68
63
  class HttpFetchError extends Error {
69
- retryable;
70
- constructor(message, retryable) {
71
- super(message);
72
- this.retryable = retryable;
73
- this.name = 'HttpFetchError';
74
- }
64
+ constructor(message, retryable) {
65
+ super(message);
66
+ this.retryable = retryable;
67
+ this.name = "HttpFetchError";
68
+ }
69
+ retryable;
75
70
  }
76
71
  async function fetchWithRedirects(originalUrl, options, timeoutMs, maxRedirects, logger) {
77
- const visited = new Set();
78
- let currentUrl = originalUrl;
79
- let redirectCount = 0;
80
- while (true) {
81
- if (visited.has(currentUrl)) {
82
- throw new HttpFetchError(`Redirect loop detected at ${currentUrl}`, false);
83
- }
84
- visited.add(currentUrl);
85
- logger.debug('fetching', { url: currentUrl, attempt: redirectCount });
86
- const signal = AbortSignal.timeout(timeoutMs);
87
- let response;
88
- try {
89
- const ua = getRotatingUserAgent(getConfig());
90
- const mergedHeaders = { 'User-Agent': ua, ...options.headers };
91
- response = await fetch(currentUrl, {
92
- headers: mergedHeaders,
93
- redirect: 'manual',
94
- signal,
95
- });
96
- }
97
- catch (err) {
98
- const isTimeout = err instanceof Error && err.name === 'TimeoutError';
99
- const isConnErr = err instanceof Error && RETRYABLE_ERROR_CODES.has(err.code ?? '');
100
- const retryable = isTimeout || isConnErr;
101
- throw Object.assign(err instanceof Error ? err : new Error(String(err)), { retryable });
102
- }
103
- if (REDIRECT_STATUSES.has(response.status)) {
104
- const location = response.headers.get('location');
105
- if (!location) {
106
- throw new HttpFetchError(`Redirect with no location header at ${currentUrl}`, false);
107
- }
108
- redirectCount++;
109
- if (redirectCount > maxRedirects) {
110
- throw new HttpFetchError(`Too many redirects (>${maxRedirects}) from ${originalUrl}`, false);
111
- }
112
- // Resolve relative redirects
113
- currentUrl = new URL(location, currentUrl).toString();
114
- continue;
115
- }
116
- if (RETRYABLE_STATUSES.has(response.status)) {
117
- throw new HttpFetchError(`HTTP ${response.status} from ${currentUrl}`, true);
118
- }
119
- const contentType = response.headers.get('content-type') ?? '';
120
- const headers = {};
121
- response.headers.forEach((value, key) => {
122
- headers[key] = value;
123
- });
124
- const isPdf = contentType.includes('application/pdf');
125
- let html;
126
- let rawBuffer;
127
- if (isPdf) {
128
- const arrayBuf = await response.arrayBuffer();
129
- rawBuffer = Buffer.from(arrayBuf);
130
- html = '';
131
- }
132
- else {
133
- html = await response.text();
134
- }
135
- return {
136
- url: originalUrl,
137
- finalUrl: currentUrl,
138
- html,
139
- contentType,
140
- statusCode: response.status,
141
- headers,
142
- rawBuffer,
143
- };
72
+ const visited = /* @__PURE__ */ new Set();
73
+ let currentUrl = originalUrl;
74
+ let redirectCount = 0;
75
+ while (true) {
76
+ if (visited.has(currentUrl)) {
77
+ throw new HttpFetchError(`Redirect loop detected at ${currentUrl}`, false);
78
+ }
79
+ visited.add(currentUrl);
80
+ logger.debug("fetching", { url: currentUrl, attempt: redirectCount });
81
+ const signal = AbortSignal.timeout(timeoutMs);
82
+ let response;
83
+ try {
84
+ const ua = getRotatingUserAgent(getConfig());
85
+ const mergedHeaders = { "User-Agent": ua, ...options.headers };
86
+ if (options.conditionalHeaders?.ifNoneMatch) {
87
+ mergedHeaders["If-None-Match"] = options.conditionalHeaders.ifNoneMatch;
88
+ }
89
+ if (options.conditionalHeaders?.ifModifiedSince) {
90
+ mergedHeaders["If-Modified-Since"] = options.conditionalHeaders.ifModifiedSince;
91
+ }
92
+ response = await fetch(currentUrl, {
93
+ headers: mergedHeaders,
94
+ redirect: "manual",
95
+ signal
96
+ });
97
+ } catch (err) {
98
+ const isTimeout = err instanceof Error && err.name === "TimeoutError";
99
+ const isConnErr = err instanceof Error && RETRYABLE_ERROR_CODES.has(err.code ?? "");
100
+ const retryable = isTimeout || isConnErr;
101
+ throw Object.assign(err instanceof Error ? err : new Error(String(err)), { retryable });
102
+ }
103
+ if (response.status === 304) {
104
+ const headers2 = {};
105
+ response.headers.forEach((value, key) => {
106
+ headers2[key] = value;
107
+ });
108
+ try {
109
+ await response.arrayBuffer();
110
+ } catch {
111
+ }
112
+ return {
113
+ url: originalUrl,
114
+ finalUrl: currentUrl,
115
+ html: "",
116
+ contentType: response.headers.get("content-type") ?? "",
117
+ statusCode: 304,
118
+ headers: headers2
119
+ };
120
+ }
121
+ if (REDIRECT_STATUSES.has(response.status)) {
122
+ const location = response.headers.get("location");
123
+ if (!location) {
124
+ throw new HttpFetchError(`Redirect with no location header at ${currentUrl}`, false);
125
+ }
126
+ redirectCount++;
127
+ if (redirectCount > maxRedirects) {
128
+ throw new HttpFetchError(`Too many redirects (>${maxRedirects}) from ${originalUrl}`, false);
129
+ }
130
+ currentUrl = new URL(location, currentUrl).toString();
131
+ continue;
132
+ }
133
+ if (RETRYABLE_STATUSES.has(response.status)) {
134
+ throw new HttpFetchError(`HTTP ${response.status} from ${currentUrl}`, true);
135
+ }
136
+ const contentType = response.headers.get("content-type") ?? "";
137
+ const headers = {};
138
+ response.headers.forEach((value, key) => {
139
+ headers[key] = value;
140
+ });
141
+ const isPdf = contentType.includes("application/pdf");
142
+ let html;
143
+ let rawBuffer;
144
+ if (isPdf) {
145
+ const arrayBuf = await response.arrayBuffer();
146
+ rawBuffer = Buffer.from(arrayBuf);
147
+ html = "";
148
+ } else {
149
+ html = await response.text();
144
150
  }
151
+ return {
152
+ url: originalUrl,
153
+ finalUrl: currentUrl,
154
+ html,
155
+ contentType,
156
+ statusCode: response.status,
157
+ headers,
158
+ rawBuffer
159
+ };
160
+ }
145
161
  }
162
+ export {
163
+ httpFetch
164
+ };
146
165
  //# sourceMappingURL=http-client.js.map