hydra-crawler 2.8.3 → 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 -141
  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
@@ -1,83 +0,0 @@
1
- var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
2
- function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
3
- return new (P || (P = Promise))(function (resolve, reject) {
4
- function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
5
- function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
6
- function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
7
- step((generator = generator.apply(thisArg, _arguments || [])).next());
8
- });
9
- };
10
- import { EStatus } from 'hydra-crawler-ts-assets';
11
- import { commonsOutputDoing, commonsOutputFound, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
12
- import { CommonsApp } from 'nodecommons-es-app';
13
- import { Lists } from '../classes/lists';
14
- import { EList } from '../enums/elist';
15
- // This only unarchives the status in the urls collection. It doesn't move entries in the archives collection back into urls
16
- export class UnarchiveUrlsApp extends CommonsApp {
17
- constructor() {
18
- super('hydra-crawler');
19
- this.lists = new Lists();
20
- }
21
- getAppName() {
22
- return 'Hydra - Unarchive';
23
- }
24
- setDatabaseService(databaseService) {
25
- this.databaseService = databaseService;
26
- }
27
- addToList(list, entries) {
28
- this.lists.add(list, entries);
29
- }
30
- init() {
31
- const _super = Object.create(null, {
32
- init: { get: () => super.init }
33
- });
34
- return __awaiter(this, void 0, void 0, function* () {
35
- if (!this.databaseService)
36
- throw new Error('Database service has not been set yet');
37
- commonsOutputDoing('Connecting to database');
38
- yield this.databaseService.init();
39
- commonsOutputSuccess();
40
- yield _super.init.call(this);
41
- });
42
- }
43
- run() {
44
- return __awaiter(this, void 0, void 0, function* () {
45
- if (!this.databaseService)
46
- throw new Error('Database service has not been set');
47
- let tally = 0;
48
- let found = 0;
49
- commonsOutputDoing('Enumerating ARCHIVED URLS');
50
- const results = this.databaseService.getUrls()
51
- .find({ status: EStatus.ARCHIVED }, {});
52
- const urls = [];
53
- tally = 0;
54
- found = 0;
55
- while (true) {
56
- if ((tally % 100) === 0)
57
- commonsOutputFound(tally, found);
58
- tally++;
59
- const row = yield results.next();
60
- if (row === null)
61
- break;
62
- if (!this.lists.match(EList.ALLOW, row.url))
63
- continue;
64
- urls.push(row.url);
65
- found++;
66
- }
67
- commonsOutputResult(found);
68
- commonsOutputDoing('Re-QUEUEING matched URLs');
69
- tally = 0;
70
- while (urls.length > 0) {
71
- commonsOutputProgress(tally);
72
- tally += 100;
73
- const batch = urls.slice(0, 100);
74
- yield this.databaseService.getUrls().updateMany({ url: { $in: batch } }, {
75
- $set: { status: EStatus.QUEUED }
76
- });
77
- urls.splice(0, 100);
78
- }
79
- commonsOutputSuccess();
80
- });
81
- }
82
- }
83
- //# sourceMappingURL=unarchive-urls.app.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"unarchive-urls.app.js","sourceRoot":"","sources":["../../src/apps/unarchive-urls.app.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,OAAO,EAAE,MAAM,yBAAyB,CAAC;AAGlD,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAC9I,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAEhD,OAAO,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAC;AAMzC,OAAO,EAAE,KAAK,EAAE,MAAM,gBAAgB,CAAC;AAIvC,4HAA4H;AAE5H,MAAM,OAAO,gBAAiB,SAAQ,UAAU;IAK/C;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;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,kBAAkB,CAAC,2BAA2B,CAAC,CAAC;YAEhD,MAAM,OAAO,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBACrE,IAAI,CACH,EAAE,MAAM,EAAE,OAAO,CAAC,QAAQ,EAAE,EAC5B,EAAE,CACH,CAAC;YAEJ,MAAM,IAAI,GAAa,EAAE,CAAC;YAC1B,KAAK,GAAG,CAAC,CAAC;YAAC,KAAK,GAAG,CAAC,CAAC;YAErB,OAAO,IAAI,EAAE;gBACZ,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,kBAAkB,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC;gBAC1D,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;aACR;YACD,mBAAmB,CAAC,KAAK,CAAC,CAAC;YAE3B,kBAAkB,CAAC,0BAA0B,CAAC,CAAC;YAC/C,KAAK,GAAG,CAAC,CAAC;YACV,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE;gBACvB,qBAAqB,CAAC,KAAK,CAAC,CAAC;gBAC7B,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;aACpB;YACD,oBAAoB,EAAE,CAAC;QACxB,CAAC;KAAA;CACD"}
@@ -1,266 +0,0 @@
1
- var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
2
- function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
3
- return new (P || (P = Promise))(function (resolve, reject) {
4
- function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
5
- function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
6
- function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
7
- step((generator = generator.apply(thisArg, _arguments || [])).next());
8
- });
9
- };
10
- import { commonsTypeHasPropertyNumber } from 'tscommons-es-core';
11
- import { EStatus } from 'hydra-crawler-ts-assets';
12
- import { commonsOutputDebug, commonsOutputDoing, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
13
- import { EList } from '../enums/elist';
14
- export class Cleaner {
15
- constructor(lists, databaseService) {
16
- this.lists = lists;
17
- this.databaseService = databaseService;
18
- }
19
- detectStatusOrphans(statuses, hardLimit) {
20
- return __awaiter(this, void 0, void 0, function* () {
21
- if (!this.databaseService)
22
- throw new Error('Database service has not been set yet');
23
- let tally = 0;
24
- let found = 0;
25
- commonsOutputDoing(`Detecting ${statuses.join(', ')} orphan urls`);
26
- while (true) {
27
- const urls = this.databaseService.getUrls()
28
- .find({
29
- $and: [
30
- { status: { $in: statuses } },
31
- { $or: [
32
- { orphan: null },
33
- { orphan: false }
34
- ] }
35
- ]
36
- }, {})
37
- .sort({ _id: 1 })
38
- .skip(tally);
39
- try {
40
- while (true) {
41
- tally++;
42
- if ((tally % 100) === 0)
43
- commonsOutputProgress(`${tally} urls, ${found} orphans`);
44
- if (hardLimit && tally >= hardLimit)
45
- break;
46
- const row = yield urls.next();
47
- if (row === null)
48
- break;
49
- const incoming = yield this.databaseService.getLinks().findOne({ outgoing: row.url }, { limit: 1 });
50
- if (incoming)
51
- continue;
52
- // eslint-disable-next-line @typescript-eslint/no-unsafe-assignment
53
- yield this.databaseService.getUrls().updateOne({ _id: row['_id'] }, { $set: { orphan: true } });
54
- found++;
55
- }
56
- break;
57
- }
58
- catch (err) {
59
- if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
60
- throw err;
61
- commonsOutputDebug('Code 43 encountered. Continuing');
62
- }
63
- }
64
- commonsOutputResult(`${tally} urls, ${found} orphans`);
65
- });
66
- }
67
- detectNonAllowlistOrphans(statuses, hardLimit) {
68
- return __awaiter(this, void 0, void 0, function* () {
69
- let tally = 0;
70
- let found = 0;
71
- commonsOutputDoing(`Detecting non-allowlist ${statuses.join(', ')} orphan urls`);
72
- while (true) {
73
- const urls = this.databaseService.getUrls()
74
- .find({
75
- $and: [
76
- { status: { $in: statuses } },
77
- { $or: [
78
- { orphan: null },
79
- { orphan: false }
80
- ] }
81
- ]
82
- }, {})
83
- .sort({ _id: 1 })
84
- .skip(tally);
85
- try {
86
- while (true) {
87
- tally++;
88
- if ((tally % 100) === 0)
89
- commonsOutputProgress(`${tally} urls, ${found} orphans`);
90
- if (hardLimit && tally >= hardLimit)
91
- break;
92
- const row = yield urls.next();
93
- if (row === null)
94
- break;
95
- if (this.lists.match(EList.ALLOW, row.url))
96
- continue;
97
- const incoming = yield this.databaseService.getLinks().findOne({ outgoing: row.url }, { limit: 1 });
98
- if (incoming)
99
- continue;
100
- // eslint-disable-next-line @typescript-eslint/no-unsafe-assignment
101
- yield this.databaseService.getUrls().updateOne({ _id: row['_id'] }, { $set: { orphan: true } });
102
- found++;
103
- }
104
- break;
105
- }
106
- catch (err) {
107
- if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
108
- throw err;
109
- commonsOutputDebug('Code 43 encountered. Continuing');
110
- }
111
- }
112
- commonsOutputResult(`${tally} urls, ${found} orphans`);
113
- });
114
- }
115
- detectStatusCodeOrphans(gt, lt, hardLimit) {
116
- return __awaiter(this, void 0, void 0, function* () {
117
- let tally = 0;
118
- let found = 0;
119
- commonsOutputDoing(`Detecting DONE status code between ${gt}<${lt} orphan urls`);
120
- while (true) {
121
- const urls = this.databaseService.getUrls()
122
- .find({
123
- $and: [
124
- { status: EStatus.DONE },
125
- { statusCode: { $gt: gt } },
126
- { statusCode: { $lt: lt } },
127
- { $or: [
128
- { orphan: null },
129
- { orphan: false }
130
- ] }
131
- ]
132
- }, {})
133
- .sort({ _id: 1 })
134
- .skip(tally);
135
- try {
136
- while (true) {
137
- tally++;
138
- if ((tally % 100) === 0)
139
- commonsOutputProgress(`${tally} urls, ${found} orphans`);
140
- if (hardLimit && tally >= hardLimit)
141
- break;
142
- const row = yield urls.next();
143
- if (row === null)
144
- break;
145
- const incoming = yield this.databaseService.getLinks().findOne({ outgoing: row.url }, { limit: 1 });
146
- if (incoming)
147
- continue;
148
- // eslint-disable-next-line @typescript-eslint/no-unsafe-assignment
149
- yield this.databaseService.getUrls().updateOne({ _id: row['_id'] }, { $set: { orphan: true } });
150
- found++;
151
- }
152
- break;
153
- }
154
- catch (err) {
155
- if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
156
- throw err;
157
- commonsOutputDebug('Code 43 encountered. Continuing');
158
- }
159
- }
160
- commonsOutputResult(`${tally} urls, ${found} orphans`);
161
- });
162
- }
163
- purgeOrphanUrls(hardLimit) {
164
- return __awaiter(this, void 0, void 0, function* () {
165
- yield this.detectStatusOrphans([
166
- EStatus.DENY,
167
- EStatus.FAILED,
168
- EStatus.DEAD,
169
- EStatus.DISALLOWED
170
- ], hardLimit);
171
- yield this.detectNonAllowlistOrphans([
172
- EStatus.QUEUED,
173
- EStatus.DONE
174
- ], hardLimit);
175
- yield this.detectStatusCodeOrphans(300, 310, hardLimit);
176
- yield this.detectStatusCodeOrphans(400, 500, hardLimit);
177
- let tally = 0;
178
- commonsOutputDoing('Removing orphan outgoing links');
179
- while (true) {
180
- const urls2 = this.databaseService.getUrls()
181
- .find({ orphan: true }, {})
182
- .sort({ _id: 1 })
183
- .skip(tally);
184
- try {
185
- while (true) {
186
- tally++;
187
- if ((tally % 100) === 0)
188
- commonsOutputProgress(tally);
189
- const row = yield urls2.next();
190
- if (row === null)
191
- break;
192
- yield this.databaseService.getLinks().deleteMany({ url: row.url });
193
- }
194
- break;
195
- }
196
- catch (err) {
197
- if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
198
- throw err;
199
- commonsOutputDebug('Code 43 encountered. Continuing');
200
- }
201
- }
202
- commonsOutputResult(tally);
203
- commonsOutputDoing('Archiving orphans');
204
- yield this.databaseService.getUrls().updateMany({ orphan: true }, {
205
- $set: { status: EStatus.ARCHIVED },
206
- $unset: {
207
- orphan: true,
208
- headers: true,
209
- server: true,
210
- hash: true,
211
- hashSet: true,
212
- ttl: true,
213
- reason: true,
214
- links: true,
215
- title: true
216
- }
217
- });
218
- commonsOutputSuccess();
219
- });
220
- }
221
- purgeEmptyDomains() {
222
- return __awaiter(this, void 0, void 0, function* () {
223
- let tally = 0;
224
- let found = 0;
225
- commonsOutputDoing('Enumerating domains');
226
- const results = this.databaseService.getDomains().find({}, {});
227
- const domains = [];
228
- try {
229
- tally = 0;
230
- while (true) {
231
- tally++;
232
- if ((tally % 100) === 0)
233
- commonsOutputProgress(tally);
234
- const row = yield results.next();
235
- if (row === null)
236
- break;
237
- domains.push(row.domain);
238
- }
239
- }
240
- catch (err) {
241
- if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
242
- throw err;
243
- }
244
- commonsOutputResult(tally);
245
- tally = 0;
246
- found = 0;
247
- commonsOutputDoing('Detecting and removing empty domains');
248
- for (const domain of domains) {
249
- tally++;
250
- if ((tally % 10) === 0)
251
- commonsOutputProgress(`${tally} domains, ${found} removed`);
252
- const urls = yield this.databaseService.getUrls()
253
- .findOne({
254
- domain: domain,
255
- status: { $ne: EStatus.ARCHIVED }
256
- }, { limit: 1 });
257
- if (urls)
258
- continue;
259
- yield this.databaseService.getDomains().deleteOne({ domain: domain });
260
- found++;
261
- }
262
- commonsOutputResult(`${tally} domains, ${found} removed`);
263
- });
264
- }
265
- }
266
- //# sourceMappingURL=cleaner.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"cleaner.js","sourceRoot":"","sources":["../../src/classes/cleaner.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,4BAA4B,EAAE,MAAM,mBAAmB,CAAC;AAGjE,OAAO,EAAE,OAAO,EAAE,MAAM,yBAAyB,CAAC;AAGlD,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAM9I,OAAO,EAAE,KAAK,EAAE,MAAM,gBAAgB,CAAC;AAEvC,MAAM,OAAO,OAAO;IACnB,YACU,KAAY,EACZ,eAAgC;QADhC,UAAK,GAAL,KAAK,CAAO;QACZ,oBAAe,GAAf,eAAe,CAAiB;IACvC,CAAC;IAEU,mBAAmB,CAC/B,QAAmB,EACnB,SAAkB;;YAEnB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,kBAAkB,CAAC,aAAa,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;YAEnE,OAAO,IAAI,EAAE;gBACZ,MAAM,IAAI,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;qBAClE,IAAI,CACH;oBACE,IAAI,EAAE;wBACJ,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE,QAAQ,EAAE,EAAE;wBAC7B,EAAE,GAAG,EAAE;gCACL,EAAE,MAAM,EAAE,IAAI,EAAE;gCAChB,EAAE,MAAM,EAAE,KAAK,EAAE;6BAClB,EAAE;qBACJ;iBACF,EACD,EAAE,CACH;qBACA,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;qBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;gBAEf,IAAI;oBACH,OAAO,IAAI,EAAE;wBACZ,KAAK,EAAE,CAAC;wBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;4BAAE,qBAAqB,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;wBAElF,IAAI,SAAS,IAAI,KAAK,IAAI,SAAS;4BAAE,MAAM;wBAE3C,MAAM,GAAG,GAAc,MAAM,IAAI,CAAC,IAAI,EAAE,CAAC;wBACzC,IAAI,GAAG,KAAK,IAAI;4BAAE,MAAM;wBAExB,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;wBACF,IAAI,QAAQ;4BAAE,SAAS;wBAEvB,mEAAmE;wBACnE,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,SAAS,CAAC,EAAE,GAAG,EAAE,GAAG,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,CAAC;wBAChG,KAAK,EAAE,CAAC;qBACR;oBAED,MAAM;iBACN;gBAAC,OAAO,GAAG,EAAE;oBACb,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;wBAAE,MAAM,GAAG,CAAC;oBAC7E,kBAAkB,CAAC,iCAAiC,CAAC,CAAC;iBACtD;aACD;YACD,mBAAmB,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;QACxD,CAAC;KAAA;IAEa,yBAAyB,CACrC,QAAmB,EACnB,SAAkB;;YAEnB,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,kBAAkB,CAAC,2BAA2B,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;YAEjF,OAAO,IAAI,EAAE;gBACZ,MAAM,IAAI,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;qBAClE,IAAI,CACH;oBACE,IAAI,EAAE;wBACJ,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE,QAAQ,EAAE,EAAE;wBAC7B,EAAE,GAAG,EAAE;gCACL,EAAE,MAAM,EAAE,IAAI,EAAE;gCAChB,EAAE,MAAM,EAAE,KAAK,EAAE;6BAClB,EAAE;qBACJ;iBACF,EACD,EAAE,CACH;qBACA,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;qBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;gBAEf,IAAI;oBACH,OAAO,IAAI,EAAE;wBACZ,KAAK,EAAE,CAAC;wBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;4BAAE,qBAAqB,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;wBAElF,IAAI,SAAS,IAAI,KAAK,IAAI,SAAS;4BAAE,MAAM;wBAE3C,MAAM,GAAG,GAAc,MAAM,IAAI,CAAC,IAAI,EAAE,CAAC;wBACzC,IAAI,GAAG,KAAK,IAAI;4BAAE,MAAM;wBAExB,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,GAAG,CAAC;4BAAE,SAAS;wBAErD,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;wBACF,IAAI,QAAQ;4BAAE,SAAS;wBAEvB,mEAAmE;wBACnE,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,SAAS,CAAC,EAAE,GAAG,EAAE,GAAG,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,CAAC;wBAChG,KAAK,EAAE,CAAC;qBACR;oBAED,MAAM;iBACN;gBAAC,OAAO,GAAG,EAAE;oBACb,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;wBAAE,MAAM,GAAG,CAAC;oBAC7E,kBAAkB,CAAC,iCAAiC,CAAC,CAAC;iBACtD;aACD;YACD,mBAAmB,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;QACxD,CAAC;KAAA;IAEa,uBAAuB,CACnC,EAAU,EACV,EAAU,EACV,SAAkB;;YAEnB,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,kBAAkB,CAAC,sCAAsC,EAAE,IAAI,EAAE,cAAc,CAAC,CAAC;YAEjF,OAAO,IAAI,EAAE;gBACZ,MAAM,IAAI,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;qBAClE,IAAI,CACH;oBACE,IAAI,EAAE;wBACJ,EAAE,MAAM,EAAE,OAAO,CAAC,IAAI,EAAE;wBACxB,EAAE,UAAU,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,EAAE;wBAC3B,EAAE,UAAU,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,EAAE;wBAC3B,EAAE,GAAG,EAAE;gCACL,EAAE,MAAM,EAAE,IAAI,EAAE;gCAChB,EAAE,MAAM,EAAE,KAAK,EAAE;6BAClB,EAAE;qBACJ;iBACF,EACD,EAAE,CACH;qBACA,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;qBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;gBAEf,IAAI;oBACH,OAAO,IAAI,EAAE;wBACZ,KAAK,EAAE,CAAC;wBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;4BAAE,qBAAqB,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;wBAElF,IAAI,SAAS,IAAI,KAAK,IAAI,SAAS;4BAAE,MAAM;wBAE3C,MAAM,GAAG,GAAc,MAAM,IAAI,CAAC,IAAI,EAAE,CAAC;wBACzC,IAAI,GAAG,KAAK,IAAI;4BAAE,MAAM;wBAExB,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;wBACF,IAAI,QAAQ;4BAAE,SAAS;wBAEvB,mEAAmE;wBACnE,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,SAAS,CAAC,EAAE,GAAG,EAAE,GAAG,CAAC,KAAK,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,CAAC;wBAChG,KAAK,EAAE,CAAC;qBACR;oBAED,MAAM;iBACN;gBAAC,OAAO,GAAG,EAAE;oBACb,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;wBAAE,MAAM,GAAG,CAAC;oBAC7E,kBAAkB,CAAC,iCAAiC,CAAC,CAAC;iBACtD;aACD;YAED,mBAAmB,CAAC,GAAG,KAAK,UAAU,KAAK,UAAU,CAAC,CAAC;QACxD,CAAC;KAAA;IAEY,eAAe,CAAC,SAAkB;;YAC9C,MAAM,IAAI,CAAC,mBAAmB,CAAC;gBAC7B,OAAO,CAAC,IAAI;gBACZ,OAAO,CAAC,MAAM;gBACd,OAAO,CAAC,IAAI;gBACZ,OAAO,CAAC,UAAU;aACnB,EAAE,SAAS,CAAC,CAAC;YAEd,MAAM,IAAI,CAAC,yBAAyB,CAAC;gBACnC,OAAO,CAAC,MAAM;gBACd,OAAO,CAAC,IAAI;aACb,EAAE,SAAS,CAAC,CAAC;YAEd,MAAM,IAAI,CAAC,uBAAuB,CAAC,GAAG,EAAE,GAAG,EAAE,SAAS,CAAC,CAAC;YACxD,MAAM,IAAI,CAAC,uBAAuB,CAAC,GAAG,EAAE,GAAG,EAAE,SAAS,CAAC,CAAC;YAExD,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,kBAAkB,CAAC,gCAAgC,CAAC,CAAC;YACrD,OAAO,IAAI,EAAE;gBACZ,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;oBACH,OAAO,IAAI,EAAE;wBACZ,KAAK,EAAE,CAAC;wBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;4BAAE,qBAAqB,CAAC,KAAK,CAAC,CAAC;wBAEtD,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;qBACnE;oBAED,MAAM;iBACN;gBAAC,OAAO,GAAG,EAAE;oBACb,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;wBAAE,MAAM,GAAG,CAAC;oBAC7E,kBAAkB,CAAC,iCAAiC,CAAC,CAAC;iBACtD;aACD;YACD,mBAAmB,CAAC,KAAK,CAAC,CAAC;YAE3B,kBAAkB,CAAC,mBAAmB,CAAC,CAAC;YACxC,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,oBAAoB,EAAE,CAAC;QACxB,CAAC;KAAA;IAEY,iBAAiB;;YAC7B,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,kBAAkB,CAAC,qBAAqB,CAAC,CAAC;YAE1C,MAAM,OAAO,GAAgC,IAAI,CAAC,eAAe,CAAC,UAAU,EAAE,CAAC,IAAI,CAAU,EAAE,EAAE,EAAE,CAAC,CAAC;YAErG,MAAM,OAAO,GAAa,EAAE,CAAC;YAC7B,IAAI;gBACH,KAAK,GAAG,CAAC,CAAC;gBAEV,OAAO,IAAI,EAAE;oBACZ,KAAK,EAAE,CAAC;oBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;wBAAE,qBAAqB,CAAC,KAAK,CAAC,CAAC;oBAEtD,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;iBACzB;aACD;YAAC,OAAO,GAAG,EAAE;gBACb,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;oBAAE,MAAM,GAAG,CAAC;aAC7E;YAED,mBAAmB,CAAC,KAAK,CAAC,CAAC;YAE3B,KAAK,GAAG,CAAC,CAAC;YAAC,KAAK,GAAG,CAAC,CAAC;YACrB,kBAAkB,CAAC,sCAAsC,CAAC,CAAC;YAC3D,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE;gBAC7B,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,GAAG,KAAK,aAAa,KAAK,UAAU,CAAC,CAAC;gBAEpF,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;aACR;YACD,mBAAmB,CAAC,GAAG,KAAK,aAAa,KAAK,UAAU,CAAC,CAAC;QAC3D,CAAC;KAAA;CAED"}