hydra-crawler 2.8.4 → 3.0.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (468) hide show
  1. package/dist/apis/{autocomplete.api.d.ts → autocomplete.api.d.mts} +4 -4
  2. package/dist/apis/autocomplete.api.mjs +14 -0
  3. package/dist/apis/autocomplete.api.mjs.map +1 -0
  4. package/dist/apis/{bugs.api.d.ts → bugs.api.d.mts} +4 -4
  5. package/dist/apis/bugs.api.mjs +18 -0
  6. package/dist/apis/bugs.api.mjs.map +1 -0
  7. package/dist/apis/{crawl.api.d.ts → crawl.api.d.mts} +4 -4
  8. package/dist/apis/crawl.api.mjs +17 -0
  9. package/dist/apis/crawl.api.mjs.map +1 -0
  10. package/dist/apis/{domains.api.d.ts → domains.api.d.mts} +4 -4
  11. package/dist/apis/{domains.api.js → domains.api.mjs} +10 -18
  12. package/dist/apis/domains.api.mjs.map +1 -0
  13. package/dist/apis/{images.api.d.ts → images.api.d.mts} +4 -4
  14. package/dist/apis/images.api.mjs +19 -0
  15. package/dist/apis/images.api.mjs.map +1 -0
  16. package/dist/apis/{statistics.api.d.ts → statistics.api.d.mts} +4 -4
  17. package/dist/apis/statistics.api.mjs +33 -0
  18. package/dist/apis/statistics.api.mjs.map +1 -0
  19. package/dist/apis/{test.api.d.ts → test.api.d.mts} +3 -3
  20. package/dist/apis/test.api.mjs +10 -0
  21. package/dist/apis/test.api.mjs.map +1 -0
  22. package/dist/apis/{urls.api.d.ts → urls.api.d.mts} +4 -4
  23. package/dist/apis/urls.api.mjs +21 -0
  24. package/dist/apis/urls.api.mjs.map +1 -0
  25. package/dist/apps/{cleanup.app.d.ts → cleanup.app.d.mts} +5 -5
  26. package/dist/apps/cleanup.app.mjs +129 -0
  27. package/dist/apps/cleanup.app.mjs.map +1 -0
  28. package/dist/apps/{cross-populate-export.app.d.ts → cross-populate-export.app.d.mts} +3 -3
  29. package/dist/apps/cross-populate-export.app.mjs +61 -0
  30. package/dist/apps/cross-populate-export.app.mjs.map +1 -0
  31. package/dist/apps/{cross-populate-import.app.d.ts → cross-populate-import.app.d.mts} +3 -3
  32. package/dist/apps/cross-populate-import.app.mjs +61 -0
  33. package/dist/apps/cross-populate-import.app.mjs.map +1 -0
  34. package/dist/apps/{denylist.app.d.ts → denylist.app.d.mts} +5 -5
  35. package/dist/apps/denylist.app.mjs +122 -0
  36. package/dist/apps/denylist.app.mjs.map +1 -0
  37. package/dist/apps/{expire.app.d.ts → expire.app.d.mts} +8 -7
  38. package/dist/apps/expire.app.mjs +51 -0
  39. package/dist/apps/expire.app.mjs.map +1 -0
  40. package/dist/apps/{export-domain-urls.d.ts → export-domain-urls.d.mts} +3 -3
  41. package/dist/apps/export-domain-urls.mjs +85 -0
  42. package/dist/apps/export-domain-urls.mjs.map +1 -0
  43. package/dist/apps/{extract-text.app.d.ts → extract-text.app.d.mts} +2 -2
  44. package/dist/apps/extract-text.app.mjs +43 -0
  45. package/dist/apps/extract-text.app.mjs.map +1 -0
  46. package/dist/apps/{hydra.app.d.ts → hydra.app.d.mts} +10 -10
  47. package/dist/apps/hydra.app.mjs +222 -0
  48. package/dist/apps/hydra.app.mjs.map +1 -0
  49. package/dist/apps/{import.app.d.ts → import.app.d.mts} +3 -3
  50. package/dist/apps/import.app.mjs +48 -0
  51. package/dist/apps/import.app.mjs.map +1 -0
  52. package/dist/apps/{internal-hydra-common.app.d.ts → internal-hydra-common.app.d.mts} +6 -6
  53. package/dist/apps/{internal-hydra-common.app.js → internal-hydra-common.app.mjs} +1 -1
  54. package/dist/apps/internal-hydra-common.app.mjs.map +1 -0
  55. package/dist/apps/{move-to-archive.app.d.ts → move-to-archive.app.d.mts} +2 -2
  56. package/dist/apps/move-to-archive.app.mjs +31 -0
  57. package/dist/apps/move-to-archive.app.mjs.map +1 -0
  58. package/dist/apps/{prune-archive.app.d.ts → prune-archive.app.d.mts} +2 -2
  59. package/dist/apps/prune-archive.app.mjs +40 -0
  60. package/dist/apps/prune-archive.app.mjs.map +1 -0
  61. package/dist/apps/{query.app.d.ts → query.app.d.mts} +6 -6
  62. package/dist/apps/query.app.mjs +69 -0
  63. package/dist/apps/query.app.mjs.map +1 -0
  64. package/dist/apps/{reattempt.app.d.ts → reattempt.app.d.mts} +6 -6
  65. package/dist/apps/reattempt.app.mjs +74 -0
  66. package/dist/apps/reattempt.app.mjs.map +1 -0
  67. package/dist/apps/{requeue-domain.app.d.ts → requeue-domain.app.d.mts} +3 -3
  68. package/dist/apps/requeue-domain.app.mjs +52 -0
  69. package/dist/apps/requeue-domain.app.mjs.map +1 -0
  70. package/dist/apps/{seed.app.d.ts → seed.app.d.mts} +5 -5
  71. package/dist/apps/seed.app.mjs +59 -0
  72. package/dist/apps/seed.app.mjs.map +1 -0
  73. package/dist/apps/{startup.app.d.ts → startup.app.d.mts} +3 -3
  74. package/dist/apps/startup.app.mjs +55 -0
  75. package/dist/apps/startup.app.mjs.map +1 -0
  76. package/dist/apps/{unarchive-urls.app.d.ts → unarchive-urls.app.d.mts} +5 -5
  77. package/dist/apps/unarchive-urls.app.mjs +71 -0
  78. package/dist/apps/unarchive-urls.app.mjs.map +1 -0
  79. package/dist/classes/{cleaner.d.ts → cleaner.d.mts} +2 -2
  80. package/dist/classes/cleaner.mjs +257 -0
  81. package/dist/classes/cleaner.mjs.map +1 -0
  82. package/dist/classes/{crawler.d.ts → crawler.d.mts} +10 -12
  83. package/dist/classes/crawler.mjs +418 -0
  84. package/dist/classes/crawler.mjs.map +1 -0
  85. package/dist/classes/{dns.js → dns.mjs} +4 -4
  86. package/dist/classes/dns.mjs.map +1 -0
  87. package/dist/classes/{expirer.d.ts → expirer.d.mts} +2 -2
  88. package/dist/classes/expirer.mjs +117 -0
  89. package/dist/classes/expirer.mjs.map +1 -0
  90. package/dist/classes/{expiry.d.ts → expiry.d.mts} +1 -1
  91. package/dist/classes/{expiry.js → expiry.mjs} +7 -9
  92. package/dist/classes/expiry.mjs.map +1 -0
  93. package/dist/classes/{lists.d.ts → lists.d.mts} +2 -2
  94. package/dist/classes/{lists.js → lists.mjs} +4 -4
  95. package/dist/classes/lists.mjs.map +1 -0
  96. package/dist/classes/{robot.d.ts → robot.d.mts} +3 -3
  97. package/dist/classes/robot.mjs +74 -0
  98. package/dist/classes/robot.mjs.map +1 -0
  99. package/dist/classes/{tracker.d.ts → tracker.d.mts} +3 -3
  100. package/dist/classes/tracker.mjs +101 -0
  101. package/dist/classes/tracker.mjs.map +1 -0
  102. package/dist/{cli.js → cli.mjs} +46 -58
  103. package/dist/cli.mjs.map +1 -0
  104. package/dist/enums/{eavailable-strategy.js → eavailable-strategy.mjs} +1 -1
  105. package/dist/enums/eavailable-strategy.mjs.map +1 -0
  106. package/dist/enums/{elist.js → elist.mjs} +1 -1
  107. package/dist/enums/elist.mjs.map +1 -0
  108. package/dist/enums/{eserver.js → eserver.mjs} +1 -1
  109. package/dist/enums/eserver.mjs.map +1 -0
  110. package/dist/enums/{ex-powered-by.js → ex-powered-by.mjs} +1 -1
  111. package/dist/enums/ex-powered-by.mjs.map +1 -0
  112. package/dist/helpers/{matcher.d.ts → matcher.d.mts} +1 -1
  113. package/dist/helpers/{matcher.js → matcher.mjs} +1 -1
  114. package/dist/helpers/matcher.mjs.map +1 -0
  115. package/dist/helpers/{random.d.ts → random.d.mts} +1 -1
  116. package/dist/helpers/{random.js → random.mjs} +1 -1
  117. package/dist/helpers/random.mjs.map +1 -0
  118. package/dist/helpers/{utf-decoder.d.ts → utf-decoder.d.mts} +0 -1
  119. package/dist/helpers/{utf-decoder.js → utf-decoder.mjs} +3 -3
  120. package/dist/helpers/utf-decoder.mjs.map +1 -0
  121. package/dist/interfaces/{iexpiry.d.ts → iexpiry.d.mts} +1 -1
  122. package/dist/interfaces/{iexpiry.js → iexpiry.mjs} +3 -3
  123. package/dist/interfaces/iexpiry.mjs.map +1 -0
  124. package/dist/interfaces/{imatch.js → imatch.mjs} +2 -2
  125. package/dist/interfaces/imatch.mjs.map +1 -0
  126. package/dist/interfaces/{iparser-config.js → iparser-config.mjs} +2 -2
  127. package/dist/interfaces/iparser-config.mjs.map +1 -0
  128. package/dist/interfaces/{iparser.d.ts → iparser.d.mts} +1 -1
  129. package/dist/interfaces/iparser.mjs +2 -0
  130. package/dist/interfaces/iparser.mjs.map +1 -0
  131. package/dist/interfaces/{irequest-outcome.d.ts → irequest-outcome.d.mts} +0 -2
  132. package/dist/interfaces/irequest-outcome.mjs +2 -0
  133. package/dist/interfaces/irequest-outcome.mjs.map +1 -0
  134. package/dist/interfaces/iserver.mjs +2 -0
  135. package/dist/interfaces/iserver.mjs.map +1 -0
  136. package/dist/parsers/accessibility-metrics.parser.d.mts +11 -0
  137. package/dist/parsers/accessibility-metrics.parser.mjs +30 -0
  138. package/dist/parsers/accessibility-metrics.parser.mjs.map +1 -0
  139. package/dist/parsers/asp-error.parser.d.mts +12 -0
  140. package/dist/parsers/asp-error.parser.mjs +38 -0
  141. package/dist/parsers/asp-error.parser.mjs.map +1 -0
  142. package/dist/parsers/{bad-words.parser.d.ts → bad-words.parser.d.mts} +6 -6
  143. package/dist/parsers/bad-words.parser.mjs +17 -0
  144. package/dist/parsers/bad-words.parser.mjs.map +1 -0
  145. package/dist/parsers/complex-english.parser.d.mts +15 -0
  146. package/dist/parsers/complex-english.parser.mjs +52 -0
  147. package/dist/parsers/complex-english.parser.mjs.map +1 -0
  148. package/dist/parsers/data.parser.d.mts +13 -0
  149. package/dist/parsers/{data.parser.js → data.parser.mjs} +8 -7
  150. package/dist/parsers/data.parser.mjs.map +1 -0
  151. package/dist/parsers/{dictionary.parser.d.ts → dictionary.parser.d.mts} +6 -6
  152. package/dist/parsers/dictionary.parser.mjs +63 -0
  153. package/dist/parsers/dictionary.parser.mjs.map +1 -0
  154. package/dist/parsers/html.parser.d.mts +13 -0
  155. package/dist/parsers/{html.parser.js → html.parser.mjs} +4 -3
  156. package/dist/parsers/html.parser.mjs.map +1 -0
  157. package/dist/parsers/hyperlinks.parser.d.mts +20 -0
  158. package/dist/parsers/hyperlinks.parser.mjs +104 -0
  159. package/dist/parsers/hyperlinks.parser.mjs.map +1 -0
  160. package/dist/parsers/image-tags.parser.d.mts +19 -0
  161. package/dist/parsers/image-tags.parser.mjs +42 -0
  162. package/dist/parsers/image-tags.parser.mjs.map +1 -0
  163. package/dist/parsers/{interest.parser.d.ts → interest.parser.d.mts} +7 -7
  164. package/dist/parsers/interest.parser.mjs +60 -0
  165. package/dist/parsers/interest.parser.mjs.map +1 -0
  166. package/dist/parsers/jpeg.parser.d.mts +11 -0
  167. package/dist/parsers/jpeg.parser.mjs +29 -0
  168. package/dist/parsers/jpeg.parser.mjs.map +1 -0
  169. package/dist/parsers/{llama-guard.parser.d.ts → llama-guard.parser.d.mts} +7 -7
  170. package/dist/parsers/llama-guard.parser.mjs +56 -0
  171. package/dist/parsers/llama-guard.parser.mjs.map +1 -0
  172. package/dist/parsers/{offence.parser.d.ts → offence.parser.d.mts} +7 -7
  173. package/dist/parsers/offence.parser.mjs +60 -0
  174. package/dist/parsers/offence.parser.mjs.map +1 -0
  175. package/dist/parsers/{ollama.parser.d.ts → ollama.parser.d.mts} +6 -6
  176. package/dist/parsers/ollama.parser.mjs +43 -0
  177. package/dist/parsers/ollama.parser.mjs.map +1 -0
  178. package/dist/parsers/{paragraphs.parser.d.ts → paragraphs.parser.d.mts} +5 -5
  179. package/dist/parsers/paragraphs.parser.mjs +38 -0
  180. package/dist/parsers/paragraphs.parser.mjs.map +1 -0
  181. package/dist/parsers/{parser.d.ts → parser.d.mts} +6 -6
  182. package/dist/parsers/parser.mjs +45 -0
  183. package/dist/parsers/parser.mjs.map +1 -0
  184. package/dist/parsers/php-error.parser.d.mts +12 -0
  185. package/dist/parsers/php-error.parser.mjs +42 -0
  186. package/dist/parsers/php-error.parser.mjs.map +1 -0
  187. package/dist/parsers/{phrase.parser.d.ts → phrase.parser.d.mts} +3 -3
  188. package/dist/parsers/phrase.parser.mjs +15 -0
  189. package/dist/parsers/phrase.parser.mjs.map +1 -0
  190. package/dist/parsers/{regex.parser.d.ts → regex.parser.d.mts} +3 -3
  191. package/dist/parsers/regex.parser.mjs +29 -0
  192. package/dist/parsers/regex.parser.mjs.map +1 -0
  193. package/dist/parsers/server.parser.d.mts +11 -0
  194. package/dist/parsers/server.parser.mjs +57 -0
  195. package/dist/parsers/server.parser.mjs.map +1 -0
  196. package/dist/parsers/{spelling.parser.d.ts → spelling.parser.d.mts} +6 -6
  197. package/dist/parsers/spelling.parser.mjs +17 -0
  198. package/dist/parsers/spelling.parser.mjs.map +1 -0
  199. package/dist/parsers/string.parser.d.mts +8 -0
  200. package/dist/parsers/{string.parser.js → string.parser.mjs} +5 -4
  201. package/dist/parsers/string.parser.mjs.map +1 -0
  202. package/dist/parsers/{text.parser.d.ts → text.parser.d.mts} +3 -3
  203. package/dist/parsers/text.parser.mjs +30 -0
  204. package/dist/parsers/text.parser.mjs.map +1 -0
  205. package/dist/parsers/{words.parser.d.ts → words.parser.d.mts} +3 -3
  206. package/dist/parsers/words.parser.mjs +29 -0
  207. package/dist/parsers/words.parser.mjs.map +1 -0
  208. package/dist/queries/complex-english.query.d.mts +2 -0
  209. package/dist/queries/{complex-english.query.js → complex-english.query.mjs} +18 -27
  210. package/dist/queries/complex-english.query.mjs.map +1 -0
  211. package/dist/queries/flash-content.query.d.mts +2 -0
  212. package/dist/queries/flash-content.query.mjs +80 -0
  213. package/dist/queries/flash-content.query.mjs.map +1 -0
  214. package/dist/queries/linking-to-domains.query.d.mts +2 -0
  215. package/dist/queries/linking-to-domains.query.mjs +128 -0
  216. package/dist/queries/linking-to-domains.query.mjs.map +1 -0
  217. package/dist/queries/llamaguard-unsafe-content.query.d.mts +2 -0
  218. package/dist/queries/llamaguard-unsafe-content.query.mjs +90 -0
  219. package/dist/queries/llamaguard-unsafe-content.query.mjs.map +1 -0
  220. package/dist/queries/readability-score.query.d.mts +2 -0
  221. package/dist/queries/{readability-score.query.js → readability-score.query.mjs} +11 -20
  222. package/dist/queries/readability-score.query.mjs.map +1 -0
  223. package/dist/servers/{crawl.server.d.ts → crawl.server.d.mts} +10 -10
  224. package/dist/servers/crawl.server.mjs +192 -0
  225. package/dist/servers/crawl.server.mjs.map +1 -0
  226. package/dist/servers/{express.server.d.ts → express.server.d.mts} +2 -3
  227. package/dist/servers/express.server.mjs +13 -0
  228. package/dist/servers/express.server.mjs.map +1 -0
  229. package/dist/servers/{maintenance.server.d.ts → maintenance.server.d.mts} +5 -5
  230. package/dist/servers/maintenance.server.mjs +97 -0
  231. package/dist/servers/maintenance.server.mjs.map +1 -0
  232. package/dist/servers/{rest.server.d.ts → rest.server.d.mts} +4 -4
  233. package/dist/servers/rest.server.mjs +62 -0
  234. package/dist/servers/rest.server.mjs.map +1 -0
  235. package/dist/servers/{socket-io.server.d.ts → socket-io.server.d.mts} +4 -4
  236. package/dist/servers/socket-io.server.mjs +22 -0
  237. package/dist/servers/socket-io.server.mjs.map +1 -0
  238. package/dist/services/{database.service.d.ts → database.service.d.mts} +12 -12
  239. package/dist/services/database.service.mjs +645 -0
  240. package/dist/services/database.service.mjs.map +1 -0
  241. package/dist/services/{ollama-rest.service.d.ts → ollama-rest.service.d.mts} +2 -2
  242. package/dist/services/ollama-rest.service.mjs +27 -0
  243. package/dist/services/ollama-rest.service.mjs.map +1 -0
  244. package/dist/services/{rig-llama-guard.service.d.ts → rig-llama-guard.service.d.mts} +2 -2
  245. package/dist/services/{rig-llama-guard.service.js → rig-llama-guard.service.mjs} +33 -43
  246. package/dist/services/rig-llama-guard.service.mjs.map +1 -0
  247. package/dist/services/{rig-queue-length.service.d.ts → rig-queue-length.service.d.mts} +1 -1
  248. package/dist/services/rig-queue-length.service.mjs +12 -0
  249. package/dist/services/rig-queue-length.service.mjs.map +1 -0
  250. package/dist/services/{rig-subjectivity-scale.service.d.ts → rig-subjectivity-scale.service.d.mts} +2 -2
  251. package/dist/services/rig-subjectivity-scale.service.mjs +82 -0
  252. package/dist/services/rig-subjectivity-scale.service.mjs.map +1 -0
  253. package/dist/services/{rig-ticket-rest.service.d.ts → rig-ticket-rest.service.d.mts} +3 -3
  254. package/dist/services/rig-ticket-rest.service.mjs +15 -0
  255. package/dist/services/rig-ticket-rest.service.mjs.map +1 -0
  256. package/dist/services/{rig-ticketed-promise.service.d.ts → rig-ticketed-promise.service.d.mts} +4 -6
  257. package/dist/services/{rig-ticketed-promise.service.js → rig-ticketed-promise.service.mjs} +7 -6
  258. package/dist/services/rig-ticketed-promise.service.mjs.map +1 -0
  259. package/dist/types/{tcrawl-config.js → tcrawl-config.mjs} +2 -2
  260. package/dist/types/tcrawl-config.mjs.map +1 -0
  261. package/dist/types/{thydra-config.js → thydra-config.mjs} +2 -2
  262. package/dist/types/thydra-config.mjs.map +1 -0
  263. package/dist/types/tparser-ctor.d.mts +7 -0
  264. package/dist/types/tparser-ctor.mjs +2 -0
  265. package/dist/types/tparser-ctor.mjs.map +1 -0
  266. package/dist/types/tquery.d.mts +7 -0
  267. package/dist/types/tquery.mjs +2 -0
  268. package/dist/types/tquery.mjs.map +1 -0
  269. package/dist/types/tqueue-length.mjs +2 -0
  270. package/dist/types/tqueue-length.mjs.map +1 -0
  271. package/dist/types/{trobots-config.js → trobots-config.mjs} +2 -2
  272. package/dist/types/trobots-config.mjs.map +1 -0
  273. package/package.json +37 -43
  274. package/dist/apis/autocomplete.api.js +0 -22
  275. package/dist/apis/autocomplete.api.js.map +0 -1
  276. package/dist/apis/bugs.api.js +0 -26
  277. package/dist/apis/bugs.api.js.map +0 -1
  278. package/dist/apis/crawl.api.js +0 -25
  279. package/dist/apis/crawl.api.js.map +0 -1
  280. package/dist/apis/domains.api.js.map +0 -1
  281. package/dist/apis/images.api.js +0 -27
  282. package/dist/apis/images.api.js.map +0 -1
  283. package/dist/apis/statistics.api.js +0 -41
  284. package/dist/apis/statistics.api.js.map +0 -1
  285. package/dist/apis/test.api.js +0 -19
  286. package/dist/apis/test.api.js.map +0 -1
  287. package/dist/apis/urls.api.js +0 -29
  288. package/dist/apis/urls.api.js.map +0 -1
  289. package/dist/apps/cleanup.app.js +0 -151
  290. package/dist/apps/cleanup.app.js.map +0 -1
  291. package/dist/apps/cross-populate-export.app.js +0 -75
  292. package/dist/apps/cross-populate-export.app.js.map +0 -1
  293. package/dist/apps/cross-populate-import.app.js +0 -100
  294. package/dist/apps/cross-populate-import.app.js.map +0 -1
  295. package/dist/apps/denylist.app.js +0 -132
  296. package/dist/apps/denylist.app.js.map +0 -1
  297. package/dist/apps/expire.app.js +0 -63
  298. package/dist/apps/expire.app.js.map +0 -1
  299. package/dist/apps/export-domain-urls.js +0 -99
  300. package/dist/apps/export-domain-urls.js.map +0 -1
  301. package/dist/apps/extract-text.app.js +0 -55
  302. package/dist/apps/extract-text.app.js.map +0 -1
  303. package/dist/apps/hydra.app.js +0 -218
  304. package/dist/apps/hydra.app.js.map +0 -1
  305. package/dist/apps/import.app.js +0 -57
  306. package/dist/apps/import.app.js.map +0 -1
  307. package/dist/apps/internal-hydra-common.app.js.map +0 -1
  308. package/dist/apps/move-to-archive.app.js +0 -46
  309. package/dist/apps/move-to-archive.app.js.map +0 -1
  310. package/dist/apps/prune-archive.app.js +0 -55
  311. package/dist/apps/prune-archive.app.js.map +0 -1
  312. package/dist/apps/query.app.js +0 -80
  313. package/dist/apps/query.app.js.map +0 -1
  314. package/dist/apps/reattempt.app.js +0 -83
  315. package/dist/apps/reattempt.app.js.map +0 -1
  316. package/dist/apps/requeue-domain.app.js +0 -64
  317. package/dist/apps/requeue-domain.app.js.map +0 -1
  318. package/dist/apps/seed.app.js +0 -69
  319. package/dist/apps/seed.app.js.map +0 -1
  320. package/dist/apps/startup.app.js +0 -64
  321. package/dist/apps/startup.app.js.map +0 -1
  322. package/dist/apps/unarchive-urls.app.js +0 -83
  323. package/dist/apps/unarchive-urls.app.js.map +0 -1
  324. package/dist/classes/cleaner.js +0 -266
  325. package/dist/classes/cleaner.js.map +0 -1
  326. package/dist/classes/crawler.js +0 -422
  327. package/dist/classes/crawler.js.map +0 -1
  328. package/dist/classes/dns.js.map +0 -1
  329. package/dist/classes/expirer.js +0 -121
  330. package/dist/classes/expirer.js.map +0 -1
  331. package/dist/classes/expiry.js.map +0 -1
  332. package/dist/classes/lists.js.map +0 -1
  333. package/dist/classes/robot.js +0 -82
  334. package/dist/classes/robot.js.map +0 -1
  335. package/dist/classes/tracker.js +0 -120
  336. package/dist/classes/tracker.js.map +0 -1
  337. package/dist/cli.js.map +0 -1
  338. package/dist/enums/eavailable-strategy.js.map +0 -1
  339. package/dist/enums/elist.js.map +0 -1
  340. package/dist/enums/eserver.js.map +0 -1
  341. package/dist/enums/ex-powered-by.js.map +0 -1
  342. package/dist/helpers/matcher.js.map +0 -1
  343. package/dist/helpers/random.js.map +0 -1
  344. package/dist/helpers/utf-decoder.js.map +0 -1
  345. package/dist/interfaces/iexpiry.js.map +0 -1
  346. package/dist/interfaces/imatch.js.map +0 -1
  347. package/dist/interfaces/iparser-config.js.map +0 -1
  348. package/dist/interfaces/iparser.js +0 -2
  349. package/dist/interfaces/iparser.js.map +0 -1
  350. package/dist/interfaces/irequest-outcome.js +0 -2
  351. package/dist/interfaces/irequest-outcome.js.map +0 -1
  352. package/dist/interfaces/iserver.js +0 -2
  353. package/dist/interfaces/iserver.js.map +0 -1
  354. package/dist/parsers/accessibility-metrics.parser.d.ts +0 -11
  355. package/dist/parsers/accessibility-metrics.parser.js +0 -40
  356. package/dist/parsers/accessibility-metrics.parser.js.map +0 -1
  357. package/dist/parsers/asp-error.parser.d.ts +0 -12
  358. package/dist/parsers/asp-error.parser.js +0 -48
  359. package/dist/parsers/asp-error.parser.js.map +0 -1
  360. package/dist/parsers/bad-words.parser.js +0 -27
  361. package/dist/parsers/bad-words.parser.js.map +0 -1
  362. package/dist/parsers/complex-english.parser.d.ts +0 -15
  363. package/dist/parsers/complex-english.parser.js +0 -61
  364. package/dist/parsers/complex-english.parser.js.map +0 -1
  365. package/dist/parsers/data.parser.d.ts +0 -14
  366. package/dist/parsers/data.parser.js.map +0 -1
  367. package/dist/parsers/dictionary.parser.js +0 -73
  368. package/dist/parsers/dictionary.parser.js.map +0 -1
  369. package/dist/parsers/html.parser.d.ts +0 -13
  370. package/dist/parsers/html.parser.js.map +0 -1
  371. package/dist/parsers/hyperlinks.parser.d.ts +0 -20
  372. package/dist/parsers/hyperlinks.parser.js +0 -115
  373. package/dist/parsers/hyperlinks.parser.js.map +0 -1
  374. package/dist/parsers/image-tags.parser.d.ts +0 -19
  375. package/dist/parsers/image-tags.parser.js +0 -52
  376. package/dist/parsers/image-tags.parser.js.map +0 -1
  377. package/dist/parsers/interest.parser.js +0 -69
  378. package/dist/parsers/interest.parser.js.map +0 -1
  379. package/dist/parsers/jpeg.parser.d.ts +0 -11
  380. package/dist/parsers/jpeg.parser.js +0 -39
  381. package/dist/parsers/jpeg.parser.js.map +0 -1
  382. package/dist/parsers/llama-guard.parser.js +0 -65
  383. package/dist/parsers/llama-guard.parser.js.map +0 -1
  384. package/dist/parsers/offence.parser.js +0 -69
  385. package/dist/parsers/offence.parser.js.map +0 -1
  386. package/dist/parsers/ollama.parser.js +0 -51
  387. package/dist/parsers/ollama.parser.js.map +0 -1
  388. package/dist/parsers/paragraphs.parser.js +0 -49
  389. package/dist/parsers/paragraphs.parser.js.map +0 -1
  390. package/dist/parsers/parser.js +0 -57
  391. package/dist/parsers/parser.js.map +0 -1
  392. package/dist/parsers/php-error.parser.d.ts +0 -12
  393. package/dist/parsers/php-error.parser.js +0 -52
  394. package/dist/parsers/php-error.parser.js.map +0 -1
  395. package/dist/parsers/phrase.parser.js +0 -26
  396. package/dist/parsers/phrase.parser.js.map +0 -1
  397. package/dist/parsers/regex.parser.js +0 -43
  398. package/dist/parsers/regex.parser.js.map +0 -1
  399. package/dist/parsers/server.parser.d.ts +0 -11
  400. package/dist/parsers/server.parser.js +0 -67
  401. package/dist/parsers/server.parser.js.map +0 -1
  402. package/dist/parsers/spelling.parser.js +0 -27
  403. package/dist/parsers/spelling.parser.js.map +0 -1
  404. package/dist/parsers/string.parser.d.ts +0 -8
  405. package/dist/parsers/string.parser.js.map +0 -1
  406. package/dist/parsers/text.parser.js +0 -41
  407. package/dist/parsers/text.parser.js.map +0 -1
  408. package/dist/parsers/words.parser.js +0 -40
  409. package/dist/parsers/words.parser.js.map +0 -1
  410. package/dist/queries/complex-english.query.d.ts +0 -2
  411. package/dist/queries/complex-english.query.js.map +0 -1
  412. package/dist/queries/flash-content.query.d.ts +0 -2
  413. package/dist/queries/flash-content.query.js +0 -82
  414. package/dist/queries/flash-content.query.js.map +0 -1
  415. package/dist/queries/linking-to-domains.query.d.ts +0 -2
  416. package/dist/queries/linking-to-domains.query.js +0 -130
  417. package/dist/queries/linking-to-domains.query.js.map +0 -1
  418. package/dist/queries/llamaguard-unsafe-content.query.d.ts +0 -2
  419. package/dist/queries/llamaguard-unsafe-content.query.js +0 -92
  420. package/dist/queries/llamaguard-unsafe-content.query.js.map +0 -1
  421. package/dist/queries/readability-score.query.d.ts +0 -2
  422. package/dist/queries/readability-score.query.js.map +0 -1
  423. package/dist/servers/crawl.server.js +0 -198
  424. package/dist/servers/crawl.server.js.map +0 -1
  425. package/dist/servers/express.server.js +0 -13
  426. package/dist/servers/express.server.js.map +0 -1
  427. package/dist/servers/maintenance.server.js +0 -103
  428. package/dist/servers/maintenance.server.js.map +0 -1
  429. package/dist/servers/rest.server.js +0 -46
  430. package/dist/servers/rest.server.js.map +0 -1
  431. package/dist/servers/socket-io.server.js +0 -61
  432. package/dist/servers/socket-io.server.js.map +0 -1
  433. package/dist/services/database.service.js +0 -795
  434. package/dist/services/database.service.js.map +0 -1
  435. package/dist/services/ollama-rest.service.js +0 -46
  436. package/dist/services/ollama-rest.service.js.map +0 -1
  437. package/dist/services/rig-llama-guard.service.js.map +0 -1
  438. package/dist/services/rig-queue-length.service.js +0 -22
  439. package/dist/services/rig-queue-length.service.js.map +0 -1
  440. package/dist/services/rig-subjectivity-scale.service.js +0 -96
  441. package/dist/services/rig-subjectivity-scale.service.js.map +0 -1
  442. package/dist/services/rig-ticket-rest.service.js +0 -29
  443. package/dist/services/rig-ticket-rest.service.js.map +0 -1
  444. package/dist/services/rig-ticketed-promise.service.js.map +0 -1
  445. package/dist/types/tcrawl-config.js.map +0 -1
  446. package/dist/types/thydra-config.js.map +0 -1
  447. package/dist/types/tparser-ctor.d.ts +0 -7
  448. package/dist/types/tparser-ctor.js +0 -2
  449. package/dist/types/tparser-ctor.js.map +0 -1
  450. package/dist/types/tquery.d.ts +0 -7
  451. package/dist/types/tquery.js +0 -2
  452. package/dist/types/tquery.js.map +0 -1
  453. package/dist/types/tqueue-length.js +0 -2
  454. package/dist/types/tqueue-length.js.map +0 -1
  455. package/dist/types/trobots-config.js.map +0 -1
  456. /package/dist/classes/{dns.d.ts → dns.d.mts} +0 -0
  457. /package/dist/{cli.d.ts → cli.d.mts} +0 -0
  458. /package/dist/enums/{eavailable-strategy.d.ts → eavailable-strategy.d.mts} +0 -0
  459. /package/dist/enums/{elist.d.ts → elist.d.mts} +0 -0
  460. /package/dist/enums/{eserver.d.ts → eserver.d.mts} +0 -0
  461. /package/dist/enums/{ex-powered-by.d.ts → ex-powered-by.d.mts} +0 -0
  462. /package/dist/interfaces/{imatch.d.ts → imatch.d.mts} +0 -0
  463. /package/dist/interfaces/{iparser-config.d.ts → iparser-config.d.mts} +0 -0
  464. /package/dist/interfaces/{iserver.d.ts → iserver.d.mts} +0 -0
  465. /package/dist/types/{tcrawl-config.d.ts → tcrawl-config.d.mts} +0 -0
  466. /package/dist/types/{thydra-config.d.ts → thydra-config.d.mts} +0 -0
  467. /package/dist/types/{tqueue-length.d.ts → tqueue-length.d.mts} +0 -0
  468. /package/dist/types/{trobots-config.d.ts → trobots-config.d.mts} +0 -0
@@ -0,0 +1,52 @@
1
+ import { EStatus } from 'hydra-crawler-esm-assets';
2
+ import { commonsLogDoing } from 'nodecommons-esm-log';
3
+ import { CommonsApp } from 'nodecommons-esm-app';
4
+ export class RequeueDomainApp extends CommonsApp {
5
+ domain;
6
+ done200Only;
7
+ databaseService;
8
+ constructor(domain, done200Only) {
9
+ super('hydra-crawler');
10
+ this.domain = domain;
11
+ this.done200Only = done200Only;
12
+ }
13
+ getAppName() {
14
+ return 'Hydra - Requeue domain';
15
+ }
16
+ setDatabaseService(databaseService) {
17
+ this.databaseService = databaseService;
18
+ }
19
+ async init() {
20
+ if (!this.databaseService)
21
+ throw new Error('Database service has not been set yet');
22
+ const log = commonsLogDoing('requeue-domain', 'Connecting to database');
23
+ await this.databaseService.init();
24
+ log.success();
25
+ await super.init();
26
+ }
27
+ async run() {
28
+ if (!this.databaseService)
29
+ throw new Error('Database service has not been set');
30
+ const conditions = {
31
+ domain: this.domain,
32
+ status: { $in: [] }
33
+ };
34
+ let log;
35
+ if (this.done200Only) {
36
+ conditions.status.$in = [EStatus.DONE];
37
+ conditions.statusCode = 200;
38
+ log = commonsLogDoing('requeue-domain', `Marking all 200 DONE as QUEUED for ${this.domain}`);
39
+ }
40
+ else {
41
+ conditions.status.$in = [
42
+ EStatus.DONE,
43
+ EStatus.FAILED
44
+ ];
45
+ log = commonsLogDoing('requeue-domain', `Marking all DONE and FAILED as QUEUED for ${this.domain}`);
46
+ }
47
+ await this.databaseService.getUrls()
48
+ .updateMany(conditions, { $set: { status: EStatus.QUEUED } });
49
+ log.success();
50
+ }
51
+ }
52
+ //# sourceMappingURL=requeue-domain.app.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"requeue-domain.app.mjs","sourceRoot":"","sources":["../../src/apps/requeue-domain.app.mts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAEnD,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAMjD,MAAM,OAAO,gBAAiB,SAAQ,UAAU;IAIrC;IACA;IAJF,eAAe,CAA4B;IAEnD,YACU,MAAc,EACd,WAAoB;QAE7B,KAAK,CAAC,eAAe,CAAC,CAAC;QAHd,WAAM,GAAN,MAAM,CAAQ;QACd,gBAAW,GAAX,WAAW,CAAS;IAG9B,CAAC;IAEM,UAAU;QAChB,OAAO,wBAAwB,CAAC;IACjC,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,gBAAgB,EAAE,wBAAwB,CAAC,CAAC;QAC5F,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,MAAM,UAAU,GAMZ;YACF,MAAM,EAAE,IAAI,CAAC,MAAM;YACnB,MAAM,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE;SACpB,CAAC;QAEF,IAAI,GAAuB,CAAC;QAC5B,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;YACtB,UAAU,CAAC,MAAM,CAAC,GAAG,GAAG,CAAE,OAAO,CAAC,IAAI,CAAE,CAAC;YACzC,UAAU,CAAC,UAAU,GAAG,GAAG,CAAC;YAE5B,GAAG,GAAG,eAAe,CAAC,gBAAgB,EAAE,sCAAsC,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;QAC9F,CAAC;aAAM,CAAC;YACP,UAAU,CAAC,MAAM,CAAC,GAAG,GAAG;gBACtB,OAAO,CAAC,IAAI;gBACZ,OAAO,CAAC,MAAM;aACf,CAAC;YAEF,GAAG,GAAG,eAAe,CAAC,gBAAgB,EAAE,6CAA6C,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;QACrG,CAAC;QAED,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;aACjC,UAAU,CACT,UAAU,EACR,EAAE,IAAI,EAAE,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,EAAE,CACvC,CAAC;QAEJ,GAAG,CAAC,OAAO,EAAE,CAAC;IACf,CAAC;CACD"}
@@ -1,8 +1,8 @@
1
- import { CommonsApp } from 'nodecommons-es-app';
2
- import { DatabaseService } from '../services/database.service';
3
- import { IMatch } from '../interfaces/imatch';
4
- import { EList } from '../enums/elist';
5
- import { IInternalHydraCommonListApp } from './internal-hydra-common.app';
1
+ import { CommonsApp } from 'nodecommons-esm-app';
2
+ import { DatabaseService } from '../services/database.service.mjs';
3
+ import { IMatch } from '../interfaces/imatch.mjs';
4
+ import { EList } from '../enums/elist.mjs';
5
+ import { IInternalHydraCommonListApp } from './internal-hydra-common.app.mjs';
6
6
  export declare class SeedApp extends CommonsApp implements IInternalHydraCommonListApp {
7
7
  private databaseService;
8
8
  private lists;
@@ -0,0 +1,59 @@
1
+ import { commonsLogDoing } from 'nodecommons-esm-log';
2
+ import { CommonsApp } from 'nodecommons-esm-app';
3
+ import { Lists } from '../classes/lists.mjs';
4
+ import { EList } from '../enums/elist.mjs';
5
+ export class SeedApp extends CommonsApp {
6
+ databaseService;
7
+ lists;
8
+ constructor() {
9
+ super('hydra-crawler');
10
+ this.lists = new Lists();
11
+ }
12
+ getAppName() {
13
+ return 'Hydra - Seed';
14
+ }
15
+ setDatabaseService(databaseService) {
16
+ this.databaseService = databaseService;
17
+ }
18
+ addToList(list, entries) {
19
+ this.lists.add(list, entries);
20
+ }
21
+ async init() {
22
+ if (!this.databaseService)
23
+ throw new Error('Database service has not been set yet');
24
+ const log = commonsLogDoing('seed', 'Connecting to database');
25
+ await this.databaseService.init();
26
+ log.success();
27
+ await super.init();
28
+ }
29
+ async run() {
30
+ if (!this.databaseService)
31
+ throw new Error('Database service has not been set');
32
+ const unique = [];
33
+ { // scope
34
+ const log = commonsLogDoing('seed', 'Building allowlist hostname seeds');
35
+ const hostnames = this.lists.listHostnames(EList.ALLOW);
36
+ for (const hostname of hostnames) {
37
+ if (!unique.includes(hostname))
38
+ unique.push(`http://${hostname}/`);
39
+ if (!unique.includes(`www.${hostname}`))
40
+ unique.push(`http://www.${hostname}/`);
41
+ log.progress(unique.length);
42
+ }
43
+ log.result(unique.length);
44
+ }
45
+ { // scope
46
+ const log = commonsLogDoing('seed', 'Seeding into queue');
47
+ let tally = 0;
48
+ for (const url of unique) {
49
+ const outcome = await this.databaseService.queue(url);
50
+ if (outcome) {
51
+ tally++;
52
+ log.progress(tally);
53
+ }
54
+ }
55
+ log.result(tally);
56
+ }
57
+ }
58
+ }
59
+ //# sourceMappingURL=seed.app.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"seed.app.mjs","sourceRoot":"","sources":["../../src/apps/seed.app.mts"],"names":[],"mappings":"AAAA,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAEjD,OAAO,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAM7C,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAC;AAI3C,MAAM,OAAO,OAAQ,SAAQ,UAAU;IAC9B,eAAe,CAA4B;IAE3C,KAAK,CAAQ;IAErB;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;QAEvB,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;IAC1B,CAAC;IAEM,UAAU;QAChB,OAAO,cAAc,CAAC;IACvB,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,MAAM,EAAE,wBAAwB,CAAC,CAAC;QAClF,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,MAAM,MAAM,GAAa,EAAE,CAAC;QAC5B,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,MAAM,EAAE,mCAAmC,CAAC,CAAC;YAC7F,MAAM,SAAS,GAAa,IAAI,CAAC,KAAK,CAAC,aAAa,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;YAElE,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;gBAClC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC;oBAAE,MAAM,CAAC,IAAI,CAAC,UAAU,QAAQ,GAAG,CAAC,CAAC;gBACnE,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,OAAO,QAAQ,EAAE,CAAC;oBAAE,MAAM,CAAC,IAAI,CAAC,cAAc,QAAQ,GAAG,CAAC,CAAC;gBAChF,GAAG,CAAC,QAAQ,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;YAC7B,CAAC;YAED,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;QAC3B,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,MAAM,EAAE,oBAAoB,CAAC,CAAC;YAC9E,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,KAAK,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;gBAC1B,MAAM,OAAO,GAAY,MAAM,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBAC/D,IAAI,OAAO,EAAE,CAAC;oBACb,KAAK,EAAE,CAAC;oBACR,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;gBACrB,CAAC;YACF,CAAC;YACD,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACnB,CAAC;IACF,CAAC;CACD"}
@@ -1,6 +1,6 @@
1
- import { CommonsApp } from 'nodecommons-es-app';
2
- import { DatabaseService } from '../services/database.service';
3
- import { IInternalHydraCommonDbApp } from './internal-hydra-common.app';
1
+ import { CommonsApp } from 'nodecommons-esm-app';
2
+ import { DatabaseService } from '../services/database.service.mjs';
3
+ import { IInternalHydraCommonDbApp } from './internal-hydra-common.app.mjs';
4
4
  export declare class StartupApp extends CommonsApp implements IInternalHydraCommonDbApp {
5
5
  private databaseService;
6
6
  constructor();
@@ -0,0 +1,55 @@
1
+ import { commonsLogDoing } from 'nodecommons-esm-log';
2
+ import { CommonsApp } from 'nodecommons-esm-app';
3
+ // load the tables into Mongo's memory
4
+ export class StartupApp extends CommonsApp {
5
+ databaseService;
6
+ constructor() {
7
+ super('hydra-crawler');
8
+ }
9
+ getAppName() {
10
+ return 'Hydra - Startup';
11
+ }
12
+ setDatabaseService(databaseService) {
13
+ this.databaseService = databaseService;
14
+ }
15
+ async init() {
16
+ if (!this.databaseService)
17
+ throw new Error('Database service has not been set yet');
18
+ const log = commonsLogDoing('startup', 'Connecting to database');
19
+ await this.databaseService.init();
20
+ log.success();
21
+ await super.init();
22
+ }
23
+ async run() {
24
+ if (!this.databaseService)
25
+ throw new Error('Database service has not been set');
26
+ const dbo = this.databaseService.getRawDatabase();
27
+ for (const collection of 'domains,urls,links'.split(',')) {
28
+ const log = commonsLogDoing('startup', `Loading ${collection} collection into memory (simple enumeration)`);
29
+ const tally = await dbo.collection(collection).find({}).count();
30
+ log.result(tally);
31
+ }
32
+ { // scope
33
+ const log = commonsLogDoing('startup', 'Loading domains collection into memory (regex enumeration)');
34
+ await this.databaseService.getDomains().find({
35
+ domain: /[0-9]temp[0-9]/
36
+ }).count();
37
+ log.success();
38
+ }
39
+ { // scope
40
+ const log = commonsLogDoing('startup', 'Loading urls collection into memory (regex enumeration)');
41
+ await this.databaseService.getUrls().find({
42
+ url: /[0-9]temp[0-9]/
43
+ }).count();
44
+ log.success();
45
+ }
46
+ { // scope
47
+ const log = commonsLogDoing('startup', 'Loading links collection into memory (regex enumeration)');
48
+ await this.databaseService.getLinks().find({
49
+ outgoing: /[0-9]temp[0-9]/
50
+ }).count();
51
+ log.success();
52
+ }
53
+ }
54
+ }
55
+ //# sourceMappingURL=startup.app.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"startup.app.mjs","sourceRoot":"","sources":["../../src/apps/startup.app.mts"],"names":[],"mappings":"AAEA,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAMjD,sCAAsC;AAEtC,MAAM,OAAO,UAAW,SAAQ,UAAU;IACjC,eAAe,CAA4B;IAEnD;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;IACxB,CAAC;IAEM,UAAU;QAChB,OAAO,iBAAiB,CAAC;IAC1B,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,wBAAwB,CAAC,CAAC;QACrF,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,MAAM,GAAG,GAAe,IAAI,CAAC,eAAe,CAAC,cAAc,EAAE,CAAC;QAE9D,KAAK,MAAM,UAAU,IAAI,oBAAoB,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;YAC1D,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,WAAW,UAAU,8CAA8C,CAAC,CAAC;YAChI,MAAM,KAAK,GAAW,MAAM,GAAG,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,KAAK,EAAE,CAAC;YACxE,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACnB,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,4DAA4D,CAAC,CAAC;YACzH,MAAM,IAAI,CAAC,eAAe,CAAC,UAAU,EAAE,CAAC,IAAI,CAAC;gBAC3C,MAAM,EAAE,gBAAgB;aACzB,CAAC,CAAC,KAAK,EAAE,CAAC;YACX,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,yDAAyD,CAAC,CAAC;YACtH,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CAAC;gBACxC,GAAG,EAAE,gBAAgB;aACtB,CAAC,CAAC,KAAK,EAAE,CAAC;YACX,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,0DAA0D,CAAC,CAAC;YACvH,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;gBACzC,QAAQ,EAAE,gBAAgB;aAC3B,CAAC,CAAC,KAAK,EAAE,CAAC;YACX,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;IACF,CAAC;CACD"}
@@ -1,8 +1,8 @@
1
- import { CommonsApp } from 'nodecommons-es-app';
2
- import { DatabaseService } from '../services/database.service';
3
- import { IMatch } from '../interfaces/imatch';
4
- import { EList } from '../enums/elist';
5
- import { IInternalHydraCommonListApp } from './internal-hydra-common.app';
1
+ import { CommonsApp } from 'nodecommons-esm-app';
2
+ import { DatabaseService } from '../services/database.service.mjs';
3
+ import { IMatch } from '../interfaces/imatch.mjs';
4
+ import { EList } from '../enums/elist.mjs';
5
+ import { IInternalHydraCommonListApp } from './internal-hydra-common.app.mjs';
6
6
  export declare class UnarchiveUrlsApp extends CommonsApp implements IInternalHydraCommonListApp {
7
7
  private databaseService;
8
8
  private lists;
@@ -0,0 +1,71 @@
1
+ import { EStatus } from 'hydra-crawler-esm-assets';
2
+ import { commonsLogDoing } from 'nodecommons-esm-log';
3
+ import { CommonsApp } from 'nodecommons-esm-app';
4
+ import { Lists } from '../classes/lists.mjs';
5
+ import { EList } from '../enums/elist.mjs';
6
+ // This only unarchives the status in the urls collection. It doesn't move entries in the archives collection back into urls
7
+ export class UnarchiveUrlsApp extends CommonsApp {
8
+ databaseService;
9
+ lists;
10
+ constructor() {
11
+ super('hydra-crawler');
12
+ this.lists = new Lists();
13
+ }
14
+ getAppName() {
15
+ return 'Hydra - Unarchive';
16
+ }
17
+ setDatabaseService(databaseService) {
18
+ this.databaseService = databaseService;
19
+ }
20
+ addToList(list, entries) {
21
+ this.lists.add(list, entries);
22
+ }
23
+ async init() {
24
+ if (!this.databaseService)
25
+ throw new Error('Database service has not been set yet');
26
+ const log = commonsLogDoing('unarchive-urls', 'Connecting to database');
27
+ await this.databaseService.init();
28
+ log.success();
29
+ await super.init();
30
+ }
31
+ async run() {
32
+ if (!this.databaseService)
33
+ throw new Error('Database service has not been set');
34
+ const urls = [];
35
+ { // scope
36
+ const log = commonsLogDoing('unarchive-urls', 'Enumerating ARCHIVED URLS');
37
+ const results = this.databaseService.getUrls()
38
+ .find({ status: EStatus.ARCHIVED }, {});
39
+ let tally = 0;
40
+ let found = 0;
41
+ while (true) {
42
+ if ((tally % 100) === 0)
43
+ log.progress(`${tally} ${found}`);
44
+ tally++;
45
+ const row = await results.next();
46
+ if (row === null)
47
+ break;
48
+ if (!this.lists.match(EList.ALLOW, row.url))
49
+ continue;
50
+ urls.push(row.url);
51
+ found++;
52
+ }
53
+ log.result(found);
54
+ }
55
+ { // scope
56
+ const log = commonsLogDoing('unarchive-urls', 'Re-QUEUEING matched URLs');
57
+ let tally = 0;
58
+ while (urls.length > 0) {
59
+ log.progress(tally);
60
+ tally += 100;
61
+ const batch = urls.slice(0, 100);
62
+ await this.databaseService.getUrls().updateMany({ url: { $in: batch } }, {
63
+ $set: { status: EStatus.QUEUED }
64
+ });
65
+ urls.splice(0, 100);
66
+ }
67
+ log.success();
68
+ }
69
+ }
70
+ }
71
+ //# sourceMappingURL=unarchive-urls.app.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"unarchive-urls.app.mjs","sourceRoot":"","sources":["../../src/apps/unarchive-urls.app.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAGnD,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAEjD,OAAO,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAM7C,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAC;AAI3C,4HAA4H;AAE5H,MAAM,OAAO,gBAAiB,SAAQ,UAAU;IACvC,eAAe,CAA4B;IAE3C,KAAK,CAAQ;IAErB;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;QAEvB,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;IAC1B,CAAC;IAEM,UAAU;QAChB,OAAO,mBAAmB,CAAC;IAC5B,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,gBAAgB,EAAE,wBAAwB,CAAC,CAAC;QAC5F,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,MAAM,IAAI,GAAa,EAAE,CAAC;QAC1B,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,gBAAgB,EAAE,2BAA2B,CAAC,CAAC;YAE/F,MAAM,OAAO,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBACrE,IAAI,CACH,EAAE,MAAM,EAAE,OAAO,CAAC,QAAQ,EAAE,EAC5B,EAAE,CACH,CAAC;YAEJ,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,OAAO,IAAI,EAAE,CAAC;gBACb,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,IAAI,KAAK,EAAE,CAAC,CAAC;gBAC3D,KAAK,EAAE,CAAC;gBAER,MAAM,GAAG,GAAc,MAAM,OAAO,CAAC,IAAI,EAAE,CAAC;gBAC5C,IAAI,GAAG,KAAK,IAAI;oBAAE,MAAM;gBAExB,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,GAAG,CAAC;oBAAE,SAAS;gBAEtD,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;gBAEnB,KAAK,EAAE,CAAC;YACT,CAAC;YACD,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACnB,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,gBAAgB,EAAE,0BAA0B,CAAC,CAAC;YAC9F,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACxB,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;gBACpB,KAAK,IAAI,GAAG,CAAC;gBAEb,MAAM,KAAK,GAAa,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;gBAE3C,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,UAAU,CAC7C,EAAE,GAAG,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,EACvB;oBACE,IAAI,EAAE,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE;iBACjC,CACF,CAAC;gBAEF,IAAI,CAAC,MAAM,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;YACrB,CAAC;YACD,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;IACF,CAAC;CACD"}
@@ -1,5 +1,5 @@
1
- import { Lists } from '../classes/lists';
2
- import { DatabaseService } from '../services/database.service';
1
+ import { Lists } from '../classes/lists.mjs';
2
+ import { DatabaseService } from '../services/database.service.mjs';
3
3
  export declare class Cleaner {
4
4
  private lists;
5
5
  private databaseService;
@@ -0,0 +1,257 @@
1
+ import { commonsTypeHasPropertyNumber } from 'tscommons-esm-core';
2
+ import { EStatus } from 'hydra-crawler-esm-assets';
3
+ import { commonsLogDebug, commonsLogDoing } from 'nodecommons-esm-log';
4
+ import { EList } from '../enums/elist.mjs';
5
+ export class Cleaner {
6
+ lists;
7
+ databaseService;
8
+ constructor(lists, databaseService) {
9
+ this.lists = lists;
10
+ this.databaseService = databaseService;
11
+ }
12
+ async detectStatusOrphans(statuses, hardLimit) {
13
+ if (!this.databaseService)
14
+ throw new Error('Database service has not been set yet');
15
+ let tally = 0;
16
+ let found = 0;
17
+ const log = commonsLogDoing('cleaner', `Detecting ${statuses.join(', ')} orphan urls`);
18
+ while (true) {
19
+ const urls = this.databaseService.getUrls()
20
+ .find({
21
+ $and: [
22
+ { status: { $in: statuses } },
23
+ { $or: [
24
+ { orphan: null },
25
+ { orphan: false }
26
+ ] }
27
+ ]
28
+ }, {})
29
+ .sort({ _id: 1 })
30
+ .skip(tally);
31
+ try {
32
+ while (true) {
33
+ tally++;
34
+ if ((tally % 100) === 0)
35
+ log.progress(`${tally} urls, ${found} orphans`);
36
+ if (hardLimit && tally >= hardLimit)
37
+ break;
38
+ const row = await urls.next();
39
+ if (row === null)
40
+ break;
41
+ const incoming = await this.databaseService.getLinks().findOne({ outgoing: row.url }, { limit: 1 });
42
+ if (incoming)
43
+ continue;
44
+ // eslint-disable-next-line @typescript-eslint/no-unsafe-assignment
45
+ await this.databaseService.getUrls().updateOne({ _id: row['_id'] }, { $set: { orphan: true } });
46
+ found++;
47
+ }
48
+ break;
49
+ }
50
+ catch (err) {
51
+ if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
52
+ throw err;
53
+ commonsLogDebug('Code 43 encountered. Continuing');
54
+ }
55
+ }
56
+ log.result(`${tally} urls, ${found} orphans`);
57
+ }
58
+ async detectNonAllowlistOrphans(statuses, hardLimit) {
59
+ let tally = 0;
60
+ let found = 0;
61
+ const log = commonsLogDoing('cleaner', `Detecting non-allowlist ${statuses.join(', ')} orphan urls`);
62
+ while (true) {
63
+ const urls = this.databaseService.getUrls()
64
+ .find({
65
+ $and: [
66
+ { status: { $in: statuses } },
67
+ { $or: [
68
+ { orphan: null },
69
+ { orphan: false }
70
+ ] }
71
+ ]
72
+ }, {})
73
+ .sort({ _id: 1 })
74
+ .skip(tally);
75
+ try {
76
+ while (true) {
77
+ tally++;
78
+ if ((tally % 100) === 0)
79
+ log.progress(`${tally} urls, ${found} orphans`);
80
+ if (hardLimit && tally >= hardLimit)
81
+ break;
82
+ const row = await urls.next();
83
+ if (row === null)
84
+ break;
85
+ if (this.lists.match(EList.ALLOW, row.url))
86
+ continue;
87
+ const incoming = await this.databaseService.getLinks().findOne({ outgoing: row.url }, { limit: 1 });
88
+ if (incoming)
89
+ continue;
90
+ // eslint-disable-next-line @typescript-eslint/no-unsafe-assignment
91
+ await this.databaseService.getUrls().updateOne({ _id: row['_id'] }, { $set: { orphan: true } });
92
+ found++;
93
+ }
94
+ break;
95
+ }
96
+ catch (err) {
97
+ if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
98
+ throw err;
99
+ commonsLogDebug('Code 43 encountered. Continuing');
100
+ }
101
+ }
102
+ log.result(`${tally} urls, ${found} orphans`);
103
+ }
104
+ async detectStatusCodeOrphans(gt, lt, hardLimit) {
105
+ let tally = 0;
106
+ let found = 0;
107
+ const log = commonsLogDoing('cleaner', `Detecting DONE status code between ${gt}<${lt} orphan urls`);
108
+ while (true) {
109
+ const urls = this.databaseService.getUrls()
110
+ .find({
111
+ $and: [
112
+ { status: EStatus.DONE },
113
+ { statusCode: { $gt: gt } },
114
+ { statusCode: { $lt: lt } },
115
+ { $or: [
116
+ { orphan: null },
117
+ { orphan: false }
118
+ ] }
119
+ ]
120
+ }, {})
121
+ .sort({ _id: 1 })
122
+ .skip(tally);
123
+ try {
124
+ while (true) {
125
+ tally++;
126
+ if ((tally % 100) === 0)
127
+ log.progress(`${tally} urls, ${found} orphans`);
128
+ if (hardLimit && tally >= hardLimit)
129
+ break;
130
+ const row = await urls.next();
131
+ if (row === null)
132
+ break;
133
+ const incoming = await this.databaseService.getLinks().findOne({ outgoing: row.url }, { limit: 1 });
134
+ if (incoming)
135
+ continue;
136
+ // eslint-disable-next-line @typescript-eslint/no-unsafe-assignment
137
+ await this.databaseService.getUrls().updateOne({ _id: row['_id'] }, { $set: { orphan: true } });
138
+ found++;
139
+ }
140
+ break;
141
+ }
142
+ catch (err) {
143
+ if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
144
+ throw err;
145
+ commonsLogDebug('Code 43 encountered. Continuing');
146
+ }
147
+ }
148
+ log.result(`${tally} urls, ${found} orphans`);
149
+ }
150
+ async purgeOrphanUrls(hardLimit) {
151
+ await this.detectStatusOrphans([
152
+ EStatus.DENY,
153
+ EStatus.FAILED,
154
+ EStatus.DEAD,
155
+ EStatus.DISALLOWED
156
+ ], hardLimit);
157
+ await this.detectNonAllowlistOrphans([
158
+ EStatus.QUEUED,
159
+ EStatus.DONE
160
+ ], hardLimit);
161
+ await this.detectStatusCodeOrphans(300, 310, hardLimit);
162
+ await this.detectStatusCodeOrphans(400, 500, hardLimit);
163
+ let tally = 0;
164
+ { // scope
165
+ const log = commonsLogDoing('clealer', 'Removing orphan outgoing links');
166
+ while (true) {
167
+ const urls2 = this.databaseService.getUrls()
168
+ .find({ orphan: true }, {})
169
+ .sort({ _id: 1 })
170
+ .skip(tally);
171
+ try {
172
+ while (true) {
173
+ tally++;
174
+ if ((tally % 100) === 0)
175
+ log.progress(tally);
176
+ const row = await urls2.next();
177
+ if (row === null)
178
+ break;
179
+ await this.databaseService.getLinks().deleteMany({ url: row.url });
180
+ }
181
+ break;
182
+ }
183
+ catch (err) {
184
+ if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
185
+ throw err;
186
+ commonsLogDebug('Code 43 encountered. Continuing');
187
+ }
188
+ }
189
+ log.result(tally);
190
+ }
191
+ { // scope
192
+ const log = commonsLogDoing('clealer', 'Archiving orphans');
193
+ await this.databaseService.getUrls().updateMany({ orphan: true }, {
194
+ $set: { status: EStatus.ARCHIVED },
195
+ $unset: {
196
+ orphan: true,
197
+ headers: true,
198
+ server: true,
199
+ hash: true,
200
+ hashSet: true,
201
+ ttl: true,
202
+ reason: true,
203
+ links: true,
204
+ title: true
205
+ }
206
+ });
207
+ log.success();
208
+ }
209
+ }
210
+ async purgeEmptyDomains() {
211
+ let tally = 0;
212
+ let found = 0;
213
+ const domains = [];
214
+ { // scope
215
+ const log = commonsLogDoing('clealer', 'Enumerating domains');
216
+ const results = this.databaseService.getDomains().find({}, {});
217
+ try {
218
+ tally = 0;
219
+ while (true) {
220
+ tally++;
221
+ if ((tally % 100) === 0)
222
+ log.progress(tally);
223
+ const row = await results.next();
224
+ if (row === null)
225
+ break;
226
+ domains.push(row.domain);
227
+ }
228
+ }
229
+ catch (err) {
230
+ if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
231
+ throw err;
232
+ }
233
+ log.result(tally);
234
+ }
235
+ { // scope
236
+ tally = 0;
237
+ found = 0;
238
+ const log = commonsLogDoing('clealer', 'Detecting and removing empty domains');
239
+ for (const domain of domains) {
240
+ tally++;
241
+ if ((tally % 10) === 0)
242
+ log.progress(`${tally} domains, ${found} removed`);
243
+ const urls = await this.databaseService.getUrls()
244
+ .findOne({
245
+ domain: domain,
246
+ status: { $ne: EStatus.ARCHIVED }
247
+ }, { limit: 1 });
248
+ if (urls)
249
+ continue;
250
+ await this.databaseService.getDomains().deleteOne({ domain: domain });
251
+ found++;
252
+ }
253
+ log.result(`${tally} domains, ${found} removed`);
254
+ }
255
+ }
256
+ }
257
+ //# sourceMappingURL=cleaner.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"cleaner.mjs","sourceRoot":"","sources":["../../src/classes/cleaner.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,4BAA4B,EAAmB,MAAM,oBAAoB,CAAC;AAGnF,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAGnD,OAAO,EAAE,eAAe,EAAE,eAAe,EAAsB,MAAM,qBAAqB,CAAC;AAM3F,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAC;AAE3C,MAAM,OAAO,OAAO;IAET;IACA;IAFV,YACU,KAAY,EACZ,eAAgC;QADhC,UAAK,GAAL,KAAK,CAAO;QACZ,oBAAe,GAAf,eAAe,CAAiB;IACvC,CAAC;IAEI,KAAK,CAAC,mBAAmB,CAC/B,QAAmB,EACnB,SAAkB;QAEnB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,IAAI,KAAK,GAAW,CAAC,CAAC;QAEtB,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,aAAa,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QAE3G,OAAO,IAAI,EAAE,CAAC;YACb,MAAM,IAAI,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBAClE,IAAI,CACH;gBACE,IAAI,EAAE;oBACJ,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE,QAAQ,EAAE,EAAE;oBAC7B,EAAE,GAAG,EAAE;4BACL,EAAE,MAAM,EAAE,IAAI,EAAE;4BAChB,EAAE,MAAM,EAAE,KAAK,EAAE;yBAClB,EAAE;iBACJ;aACF,EACD,EAAE,CACH;iBACA,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;iBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;YAEf,IAAI,CAAC;gBACJ,OAAO,IAAI,EAAE,CAAC;oBACb,KAAK,EAAE,CAAC;oBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;wBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;oBAEzE,IAAI,SAAS,IAAI,KAAK,IAAI,SAAS;wBAAE,MAAM;oBAE3C,MAAM,GAAG,GAAc,MAAM,IAAI,CAAC,IAAI,EAAE,CAAC;oBACzC,IAAI,GAAG,KAAK,IAAI;wBAAE,MAAM;oBAExB,MAAM,QAAQ,GAAe,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,OAAO,CACxE,EAAE,QAAQ,EAAE,GAAG,CAAC,GAAG,EAAE,EACrB,EAAE,KAAK,EAAE,CAAC,EAAE,CACb,CAAC;oBACF,IAAI,QAAQ;wBAAE,SAAS;oBAEvB,mEAAmE;oBACnE,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,SAAS,CAAC,EAAE,GAAG,EAAG,GAAuB,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,CAAC;oBACrH,KAAK,EAAE,CAAC;gBACT,CAAC;gBAED,MAAM;YACP,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACd,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;oBAAE,MAAM,GAAG,CAAC;gBAC7E,eAAe,CAAC,iCAAiC,CAAC,CAAC;YACpD,CAAC;QACF,CAAC;QACD,GAAG,CAAC,MAAM,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;IAC/C,CAAC;IAEO,KAAK,CAAC,yBAAyB,CACrC,QAAmB,EACnB,SAAkB;QAEnB,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,IAAI,KAAK,GAAW,CAAC,CAAC;QAEtB,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,2BAA2B,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QAEzH,OAAO,IAAI,EAAE,CAAC;YACb,MAAM,IAAI,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBAClE,IAAI,CACH;gBACE,IAAI,EAAE;oBACJ,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE,QAAQ,EAAE,EAAE;oBAC7B,EAAE,GAAG,EAAE;4BACL,EAAE,MAAM,EAAE,IAAI,EAAE;4BAChB,EAAE,MAAM,EAAE,KAAK,EAAE;yBAClB,EAAE;iBACJ;aACF,EACD,EAAE,CACH;iBACA,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;iBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;YAEf,IAAI,CAAC;gBACJ,OAAO,IAAI,EAAE,CAAC;oBACb,KAAK,EAAE,CAAC;oBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;wBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;oBAEzE,IAAI,SAAS,IAAI,KAAK,IAAI,SAAS;wBAAE,MAAM;oBAE3C,MAAM,GAAG,GAAc,MAAM,IAAI,CAAC,IAAI,EAAE,CAAC;oBACzC,IAAI,GAAG,KAAK,IAAI;wBAAE,MAAM;oBAExB,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,GAAG,CAAC;wBAAE,SAAS;oBAErD,MAAM,QAAQ,GAAe,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,OAAO,CACxE,EAAE,QAAQ,EAAE,GAAG,CAAC,GAAG,EAAE,EACrB,EAAE,KAAK,EAAE,CAAC,EAAE,CACb,CAAC;oBACF,IAAI,QAAQ;wBAAE,SAAS;oBAEvB,mEAAmE;oBACnE,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,SAAS,CAAC,EAAE,GAAG,EAAG,GAAuB,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,CAAC;oBACrH,KAAK,EAAE,CAAC;gBACT,CAAC;gBAED,MAAM;YACP,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACd,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;oBAAE,MAAM,GAAG,CAAC;gBAC7E,eAAe,CAAC,iCAAiC,CAAC,CAAC;YACpD,CAAC;QACF,CAAC;QACD,GAAG,CAAC,MAAM,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;IAC/C,CAAC;IAEO,KAAK,CAAC,uBAAuB,CACnC,EAAU,EACV,EAAU,EACV,SAAkB;QAEnB,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,IAAI,KAAK,GAAW,CAAC,CAAC;QAEtB,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,sCAAsC,EAAE,IAAI,EAAE,cAAc,CAAC,CAAC;QAEzH,OAAO,IAAI,EAAE,CAAC;YACb,MAAM,IAAI,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBAClE,IAAI,CACH;gBACE,IAAI,EAAE;oBACJ,EAAE,MAAM,EAAE,OAAO,CAAC,IAAI,EAAE;oBACxB,EAAE,UAAU,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,EAAE;oBAC3B,EAAE,UAAU,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,EAAE;oBAC3B,EAAE,GAAG,EAAE;4BACL,EAAE,MAAM,EAAE,IAAI,EAAE;4BAChB,EAAE,MAAM,EAAE,KAAK,EAAE;yBAClB,EAAE;iBACJ;aACF,EACD,EAAE,CACH;iBACA,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;iBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;YAEf,IAAI,CAAC;gBACJ,OAAO,IAAI,EAAE,CAAC;oBACb,KAAK,EAAE,CAAC;oBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;wBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;oBAEzE,IAAI,SAAS,IAAI,KAAK,IAAI,SAAS;wBAAE,MAAM;oBAE3C,MAAM,GAAG,GAAc,MAAM,IAAI,CAAC,IAAI,EAAE,CAAC;oBACzC,IAAI,GAAG,KAAK,IAAI;wBAAE,MAAM;oBAExB,MAAM,QAAQ,GAAe,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,OAAO,CACxE,EAAE,QAAQ,EAAE,GAAG,CAAC,GAAG,EAAE,EACrB,EAAE,KAAK,EAAE,CAAC,EAAE,CACb,CAAC;oBACF,IAAI,QAAQ;wBAAE,SAAS;oBAEvB,mEAAmE;oBACnE,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,SAAS,CAAC,EAAE,GAAG,EAAG,GAAuB,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,CAAC;oBACrH,KAAK,EAAE,CAAC;gBACT,CAAC;gBAED,MAAM;YACP,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACd,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;oBAAE,MAAM,GAAG,CAAC;gBAC7E,eAAe,CAAC,iCAAiC,CAAC,CAAC;YACpD,CAAC;QACF,CAAC;QAED,GAAG,CAAC,MAAM,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;IAC/C,CAAC;IAEM,KAAK,CAAC,eAAe,CAAC,SAAkB;QAC9C,MAAM,IAAI,CAAC,mBAAmB,CAAC;YAC7B,OAAO,CAAC,IAAI;YACZ,OAAO,CAAC,MAAM;YACd,OAAO,CAAC,IAAI;YACZ,OAAO,CAAC,UAAU;SACnB,EAAE,SAAS,CAAC,CAAC;QAEd,MAAM,IAAI,CAAC,yBAAyB,CAAC;YACnC,OAAO,CAAC,MAAM;YACd,OAAO,CAAC,IAAI;SACb,EAAE,SAAS,CAAC,CAAC;QAEd,MAAM,IAAI,CAAC,uBAAuB,CAAC,GAAG,EAAE,GAAG,EAAE,SAAS,CAAC,CAAC;QACxD,MAAM,IAAI,CAAC,uBAAuB,CAAC,GAAG,EAAE,GAAG,EAAE,SAAS,CAAC,CAAC;QAExD,IAAI,KAAK,GAAW,CAAC,CAAC;QAEtB,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,gCAAgC,CAAC,CAAC;YAC7F,OAAO,IAAI,EAAE,CAAC;gBACb,MAAM,KAAK,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;qBACnE,IAAI,CAAO,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE,EAAE,CAAC;qBAChC,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;qBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;gBAEf,IAAI,CAAC;oBACJ,OAAO,IAAI,EAAE,CAAC;wBACb,KAAK,EAAE,CAAC;wBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;4BAAE,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;wBAE7C,MAAM,GAAG,GAAc,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;wBAC1C,IAAI,GAAG,KAAK,IAAI;4BAAE,MAAM;wBAExB,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC;oBACpE,CAAC;oBAED,MAAM;gBACP,CAAC;gBAAC,OAAO,GAAG,EAAE,CAAC;oBACd,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;wBAAE,MAAM,GAAG,CAAC;oBAC7E,eAAe,CAAC,iCAAiC,CAAC,CAAC;gBACpD,CAAC;YACF,CAAC;YACD,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACnB,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,mBAAmB,CAAC,CAAC;YAChF,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,UAAU,CAC7C,EAAE,MAAM,EAAE,IAAI,EAAE,EAChB;gBACE,IAAI,EAAE,EAAE,MAAM,EAAE,OAAO,CAAC,QAAQ,EAAE;gBAClC,MAAM,EAAE;oBACN,MAAM,EAAE,IAAI;oBACZ,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,IAAI;oBACZ,IAAI,EAAE,IAAI;oBACV,OAAO,EAAE,IAAI;oBACb,GAAG,EAAE,IAAI;oBACT,MAAM,EAAE,IAAI;oBACZ,KAAK,EAAE,IAAI;oBACX,KAAK,EAAE,IAAI;iBACZ;aACF,CACF,CAAC;YACF,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;IACF,CAAC;IAEM,KAAK,CAAC,iBAAiB;QAC7B,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,IAAI,KAAK,GAAW,CAAC,CAAC;QAEtB,MAAM,OAAO,GAAa,EAAE,CAAC;QAC7B,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,qBAAqB,CAAC,CAAC;YAElF,MAAM,OAAO,GAAgC,IAAI,CAAC,eAAe,CAAC,UAAU,EAAE,CAAC,IAAI,CAAU,EAAE,EAAE,EAAE,CAAC,CAAC;YAErG,IAAI,CAAC;gBACJ,KAAK,GAAG,CAAC,CAAC;gBAEV,OAAO,IAAI,EAAE,CAAC;oBACb,KAAK,EAAE,CAAC;oBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;wBAAE,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;oBAE7C,MAAM,GAAG,GAAiB,MAAM,OAAO,CAAC,IAAI,EAAE,CAAC;oBAC/C,IAAI,GAAG,KAAK,IAAI;wBAAE,MAAM;oBAExB,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;gBAC1B,CAAC;YACF,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACd,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;oBAAE,MAAM,GAAG,CAAC;YAC9E,CAAC;YAED,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACnB,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,KAAK,GAAG,CAAC,CAAC;YAAC,KAAK,GAAG,CAAC,CAAC;YACrB,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,sCAAsC,CAAC,CAAC;YACnG,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;gBAC9B,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,KAAK,CAAC;oBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,aAAa,KAAK,UAAU,CAAC,CAAC;gBAE3E,MAAM,IAAI,GAAc,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;qBACzD,OAAO,CAAO;oBACb,MAAM,EAAE,MAAM;oBACd,MAAM,EAAE,EAAE,GAAG,EAAE,OAAO,CAAC,QAAQ,EAAE;iBAClC,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;gBACnB,IAAI,IAAI;oBAAE,SAAS;gBAEnB,MAAM,IAAI,CAAC,eAAe,CAAC,UAAU,EAAE,CAAC,SAAS,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC,CAAC;gBAEtE,KAAK,EAAE,CAAC;YACT,CAAC;YACD,GAAG,CAAC,MAAM,CAAC,GAAG,KAAK,aAAa,KAAK,UAAU,CAAC,CAAC;QAClD,CAAC;IACF,CAAC;CAED"}
@@ -1,16 +1,14 @@
1
- /// <reference types="node" />
2
- /// <reference types="node" />
3
1
  import * as http from 'http';
4
2
  import * as https from 'https';
5
- import { TKeyObject } from 'tscommons-es-core';
6
- import { Lists } from '../classes/lists';
7
- import { Tracker } from '../classes/tracker';
8
- import { DatabaseService } from '../services/database.service';
9
- import { IRequestOutcome } from '../interfaces/irequest-outcome';
10
- import { IParserConfig } from '../interfaces/iparser-config';
11
- import { TCrawlConfig } from '../types/tcrawl-config';
12
- import { TRobotsConfig } from '../types/trobots-config';
13
- import { TParserCtor } from '../types/tparser-ctor';
3
+ import { TPropertyObject } from 'tscommons-esm-core';
4
+ import { Lists } from '../classes/lists.mjs';
5
+ import { Tracker } from '../classes/tracker.mjs';
6
+ import { DatabaseService } from '../services/database.service.mjs';
7
+ import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
8
+ import { IParserConfig } from '../interfaces/iparser-config.mjs';
9
+ import { TCrawlConfig } from '../types/tcrawl-config.mjs';
10
+ import { TRobotsConfig } from '../types/trobots-config.mjs';
11
+ import { TParserCtor } from '../types/tparser-ctor.mjs';
14
12
  export declare class Crawler {
15
13
  private domain;
16
14
  private database;
@@ -25,7 +23,7 @@ export declare class Crawler {
25
23
  private static pruneHeaders;
26
24
  private isPaused;
27
25
  private isAborted;
28
- constructor(domain: string, database: DatabaseService, crawlConfig: TCrawlConfig, parsersConfig: TKeyObject<IParserConfig>, robotsConfig: TRobotsConfig, parsers: TParserCtor[], lists: Lists, tracker?: Tracker | undefined);
26
+ constructor(domain: string, database: DatabaseService, crawlConfig: TCrawlConfig, parsersConfig: TPropertyObject<IParserConfig>, robotsConfig: TRobotsConfig, parsers: TParserCtor[], lists: Lists, tracker?: Tracker | undefined);
29
27
  private abort;
30
28
  pause(): void;
31
29
  resume(): void;