hydra-crawler 2.8.4 → 3.0.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (467) 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 +135 -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 +81 -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 +78 -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 +264 -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.js → crawler.mjs} +216 -184
  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} +26 -28
  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.map +0 -1
  327. package/dist/classes/dns.js.map +0 -1
  328. package/dist/classes/expirer.js +0 -121
  329. package/dist/classes/expirer.js.map +0 -1
  330. package/dist/classes/expiry.js.map +0 -1
  331. package/dist/classes/lists.js.map +0 -1
  332. package/dist/classes/robot.js +0 -82
  333. package/dist/classes/robot.js.map +0 -1
  334. package/dist/classes/tracker.js +0 -120
  335. package/dist/classes/tracker.js.map +0 -1
  336. package/dist/cli.js.map +0 -1
  337. package/dist/enums/eavailable-strategy.js.map +0 -1
  338. package/dist/enums/elist.js.map +0 -1
  339. package/dist/enums/eserver.js.map +0 -1
  340. package/dist/enums/ex-powered-by.js.map +0 -1
  341. package/dist/helpers/matcher.js.map +0 -1
  342. package/dist/helpers/random.js.map +0 -1
  343. package/dist/helpers/utf-decoder.js.map +0 -1
  344. package/dist/interfaces/iexpiry.js.map +0 -1
  345. package/dist/interfaces/imatch.js.map +0 -1
  346. package/dist/interfaces/iparser-config.js.map +0 -1
  347. package/dist/interfaces/iparser.js +0 -2
  348. package/dist/interfaces/iparser.js.map +0 -1
  349. package/dist/interfaces/irequest-outcome.js +0 -2
  350. package/dist/interfaces/irequest-outcome.js.map +0 -1
  351. package/dist/interfaces/iserver.js +0 -2
  352. package/dist/interfaces/iserver.js.map +0 -1
  353. package/dist/parsers/accessibility-metrics.parser.d.ts +0 -11
  354. package/dist/parsers/accessibility-metrics.parser.js +0 -40
  355. package/dist/parsers/accessibility-metrics.parser.js.map +0 -1
  356. package/dist/parsers/asp-error.parser.d.ts +0 -12
  357. package/dist/parsers/asp-error.parser.js +0 -48
  358. package/dist/parsers/asp-error.parser.js.map +0 -1
  359. package/dist/parsers/bad-words.parser.js +0 -27
  360. package/dist/parsers/bad-words.parser.js.map +0 -1
  361. package/dist/parsers/complex-english.parser.d.ts +0 -15
  362. package/dist/parsers/complex-english.parser.js +0 -61
  363. package/dist/parsers/complex-english.parser.js.map +0 -1
  364. package/dist/parsers/data.parser.d.ts +0 -14
  365. package/dist/parsers/data.parser.js.map +0 -1
  366. package/dist/parsers/dictionary.parser.js +0 -73
  367. package/dist/parsers/dictionary.parser.js.map +0 -1
  368. package/dist/parsers/html.parser.d.ts +0 -13
  369. package/dist/parsers/html.parser.js.map +0 -1
  370. package/dist/parsers/hyperlinks.parser.d.ts +0 -20
  371. package/dist/parsers/hyperlinks.parser.js +0 -115
  372. package/dist/parsers/hyperlinks.parser.js.map +0 -1
  373. package/dist/parsers/image-tags.parser.d.ts +0 -19
  374. package/dist/parsers/image-tags.parser.js +0 -52
  375. package/dist/parsers/image-tags.parser.js.map +0 -1
  376. package/dist/parsers/interest.parser.js +0 -69
  377. package/dist/parsers/interest.parser.js.map +0 -1
  378. package/dist/parsers/jpeg.parser.d.ts +0 -11
  379. package/dist/parsers/jpeg.parser.js +0 -39
  380. package/dist/parsers/jpeg.parser.js.map +0 -1
  381. package/dist/parsers/llama-guard.parser.js +0 -65
  382. package/dist/parsers/llama-guard.parser.js.map +0 -1
  383. package/dist/parsers/offence.parser.js +0 -69
  384. package/dist/parsers/offence.parser.js.map +0 -1
  385. package/dist/parsers/ollama.parser.js +0 -51
  386. package/dist/parsers/ollama.parser.js.map +0 -1
  387. package/dist/parsers/paragraphs.parser.js +0 -49
  388. package/dist/parsers/paragraphs.parser.js.map +0 -1
  389. package/dist/parsers/parser.js +0 -57
  390. package/dist/parsers/parser.js.map +0 -1
  391. package/dist/parsers/php-error.parser.d.ts +0 -12
  392. package/dist/parsers/php-error.parser.js +0 -52
  393. package/dist/parsers/php-error.parser.js.map +0 -1
  394. package/dist/parsers/phrase.parser.js +0 -26
  395. package/dist/parsers/phrase.parser.js.map +0 -1
  396. package/dist/parsers/regex.parser.js +0 -43
  397. package/dist/parsers/regex.parser.js.map +0 -1
  398. package/dist/parsers/server.parser.d.ts +0 -11
  399. package/dist/parsers/server.parser.js +0 -67
  400. package/dist/parsers/server.parser.js.map +0 -1
  401. package/dist/parsers/spelling.parser.js +0 -27
  402. package/dist/parsers/spelling.parser.js.map +0 -1
  403. package/dist/parsers/string.parser.d.ts +0 -8
  404. package/dist/parsers/string.parser.js.map +0 -1
  405. package/dist/parsers/text.parser.js +0 -41
  406. package/dist/parsers/text.parser.js.map +0 -1
  407. package/dist/parsers/words.parser.js +0 -40
  408. package/dist/parsers/words.parser.js.map +0 -1
  409. package/dist/queries/complex-english.query.d.ts +0 -2
  410. package/dist/queries/complex-english.query.js.map +0 -1
  411. package/dist/queries/flash-content.query.d.ts +0 -2
  412. package/dist/queries/flash-content.query.js +0 -82
  413. package/dist/queries/flash-content.query.js.map +0 -1
  414. package/dist/queries/linking-to-domains.query.d.ts +0 -2
  415. package/dist/queries/linking-to-domains.query.js +0 -130
  416. package/dist/queries/linking-to-domains.query.js.map +0 -1
  417. package/dist/queries/llamaguard-unsafe-content.query.d.ts +0 -2
  418. package/dist/queries/llamaguard-unsafe-content.query.js +0 -92
  419. package/dist/queries/llamaguard-unsafe-content.query.js.map +0 -1
  420. package/dist/queries/readability-score.query.d.ts +0 -2
  421. package/dist/queries/readability-score.query.js.map +0 -1
  422. package/dist/servers/crawl.server.js +0 -198
  423. package/dist/servers/crawl.server.js.map +0 -1
  424. package/dist/servers/express.server.js +0 -13
  425. package/dist/servers/express.server.js.map +0 -1
  426. package/dist/servers/maintenance.server.js +0 -103
  427. package/dist/servers/maintenance.server.js.map +0 -1
  428. package/dist/servers/rest.server.js +0 -46
  429. package/dist/servers/rest.server.js.map +0 -1
  430. package/dist/servers/socket-io.server.js +0 -61
  431. package/dist/servers/socket-io.server.js.map +0 -1
  432. package/dist/services/database.service.js +0 -795
  433. package/dist/services/database.service.js.map +0 -1
  434. package/dist/services/ollama-rest.service.js +0 -46
  435. package/dist/services/ollama-rest.service.js.map +0 -1
  436. package/dist/services/rig-llama-guard.service.js.map +0 -1
  437. package/dist/services/rig-queue-length.service.js +0 -22
  438. package/dist/services/rig-queue-length.service.js.map +0 -1
  439. package/dist/services/rig-subjectivity-scale.service.js +0 -96
  440. package/dist/services/rig-subjectivity-scale.service.js.map +0 -1
  441. package/dist/services/rig-ticket-rest.service.js +0 -29
  442. package/dist/services/rig-ticket-rest.service.js.map +0 -1
  443. package/dist/services/rig-ticketed-promise.service.js.map +0 -1
  444. package/dist/types/tcrawl-config.js.map +0 -1
  445. package/dist/types/thydra-config.js.map +0 -1
  446. package/dist/types/tparser-ctor.d.ts +0 -7
  447. package/dist/types/tparser-ctor.js +0 -2
  448. package/dist/types/tparser-ctor.js.map +0 -1
  449. package/dist/types/tquery.d.ts +0 -7
  450. package/dist/types/tquery.js +0 -2
  451. package/dist/types/tquery.js.map +0 -1
  452. package/dist/types/tqueue-length.js +0 -2
  453. package/dist/types/tqueue-length.js.map +0 -1
  454. package/dist/types/trobots-config.js.map +0 -1
  455. /package/dist/classes/{dns.d.ts → dns.d.mts} +0 -0
  456. /package/dist/{cli.d.ts → cli.d.mts} +0 -0
  457. /package/dist/enums/{eavailable-strategy.d.ts → eavailable-strategy.d.mts} +0 -0
  458. /package/dist/enums/{elist.d.ts → elist.d.mts} +0 -0
  459. /package/dist/enums/{eserver.d.ts → eserver.d.mts} +0 -0
  460. /package/dist/enums/{ex-powered-by.d.ts → ex-powered-by.d.mts} +0 -0
  461. /package/dist/interfaces/{imatch.d.ts → imatch.d.mts} +0 -0
  462. /package/dist/interfaces/{iparser-config.d.ts → iparser-config.d.mts} +0 -0
  463. /package/dist/interfaces/{iserver.d.ts → iserver.d.mts} +0 -0
  464. /package/dist/types/{tcrawl-config.d.ts → tcrawl-config.d.mts} +0 -0
  465. /package/dist/types/{thydra-config.d.ts → thydra-config.d.mts} +0 -0
  466. /package/dist/types/{tqueue-length.d.ts → tqueue-length.d.mts} +0 -0
  467. /package/dist/types/{trobots-config.d.ts → trobots-config.d.mts} +0 -0
@@ -1,26 +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 { commonsTypeEncode } from 'tscommons-es-core';
11
- import { CommonsRestApi } from 'nodecommons-es-rest';
12
- export class BugsApi extends CommonsRestApi {
13
- constructor(restServer, databaseService, path) {
14
- super(restServer);
15
- this.databaseService = databaseService;
16
- super.getHandler(`${path}rest/bugs/php`, (_req, _res) => __awaiter(this, void 0, void 0, function* () {
17
- const result = yield this.databaseService.listPhpErrors();
18
- return commonsTypeEncode(result);
19
- }));
20
- super.getHandler(`${path}rest/bugs/asp`, (_req, _res) => __awaiter(this, void 0, void 0, function* () {
21
- const result = yield this.databaseService.listAspErrors();
22
- return commonsTypeEncode(result);
23
- }));
24
- }
25
- }
26
- //# sourceMappingURL=bugs.api.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"bugs.api.js","sourceRoot":"","sources":["../../src/apis/bugs.api.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,iBAAiB,EAAY,MAAM,mBAAmB,CAAC;AAMhE,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAOrD,MAAM,OAAO,OAAQ,SAAQ,cAAiE;IAC7F,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,eAAe,EACtB,CAAO,IAAqB,EAAE,IAAsB,EAAqB,EAAE;YAC1E,MAAM,MAAM,GAA4B,MAAM,IAAI,CAAC,eAAe,CAAC,aAAa,EAAE,CAAC;YAEnF,OAAO,iBAAiB,CAAC,MAAM,CAAC,CAAC;QAClC,CAAC,CAAA,CACF,CAAC;QAEF,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,eAAe,EACtB,CAAO,IAAqB,EAAE,IAAsB,EAAqB,EAAE;YAC1E,MAAM,MAAM,GAA4B,MAAM,IAAI,CAAC,eAAe,CAAC,aAAa,EAAE,CAAC;YAEnF,OAAO,iBAAiB,CAAC,MAAM,CAAC,CAAC;QAClC,CAAC,CAAA,CACF,CAAC;IACH,CAAC;CACD"}
@@ -1,25 +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 { CommonsRestApi } from 'nodecommons-es-rest';
11
- export class CrawlApi extends CommonsRestApi {
12
- constructor(restServer, databaseService, path) {
13
- super(restServer);
14
- this.databaseService = databaseService;
15
- super.getHandler(`${path}rest/crawl/queue`, (_req, _res) => __awaiter(this, void 0, void 0, function* () {
16
- const tallies = yield this.databaseService.listDomainQueuedTallies();
17
- return Array.from(tallies.keys())
18
- .map((domain) => ({
19
- domain: domain,
20
- queue: tallies.get(domain) || 0
21
- }));
22
- }));
23
- }
24
- }
25
- //# sourceMappingURL=crawl.api.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"crawl.api.js","sourceRoot":"","sources":["../../src/apis/crawl.api.ts"],"names":[],"mappings":";;;;;;;;;AAIA,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAOrD,MAAM,OAAO,QAAS,SAAQ,cAAiE;IAC9F,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,kBAAkB,EACzB,CAAO,IAAqB,EAAE,IAAsB,EAA2B,EAAE;YAChF,MAAM,OAAO,GAAwB,MAAM,IAAI,CAAC,eAAe,CAAC,uBAAuB,EAAE,CAAC;YAE1F,OAAO,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;iBAC9B,GAAG,CAAC,CAAC,MAAc,EAAgB,EAAE,CAAC,CAAC;gBACtC,MAAM,EAAE,MAAM;gBACd,KAAK,EAAE,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC;aAChC,CAAC,CAAC,CAAC;QACP,CAAC,CAAA,CACF,CAAC;IAEH,CAAC;CACD"}
@@ -1 +0,0 @@
1
- {"version":3,"file":"domains.api.js","sourceRoot":"","sources":["../../src/apis/domains.api.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,mCAAmC,EAAE,iBAAiB,EAA0B,MAAM,mBAAmB,CAAC;AAKnH,OAAO,EAAE,cAAc,EAAE,wBAAwB,EAAE,mBAAmB,EAAE,MAAM,qBAAqB,CAAC;AAMpG,MAAM,OAAO,UAAW,SAAQ,cAAiE;IAChG,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,cAAc,EACrB,CAAO,IAAqB,EAAE,IAAsB,EAAqB,EAAE;YAC1E,MAAM,OAAO,GAAc,MAAM,IAAI,CAAC,eAAe,CAAC,WAAW,EAAE,CAAC;YAEpE,OAAO,iBAAiB,CAAC,OAAO,CAAC,CAAC;QACnC,CAAC,CAAA,CACF,CAAC;QAEF,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,mCAAmC,EAC1C,CAAO,GAAoC,EAAE,IAAsB,EAAqB,EAAE;YACzF,IAAI,CAAC,8BAA8B,CAAC,IAAI,CAAC,GAAG,CAAC,YAAY,CAAC,MAAgB,CAAC;gBAAE,OAAO,wBAAwB,CAAC,gBAAgB,CAAC,CAAC;YAE/H,MAAM,IAAI,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,qBAAqB,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;YAE3F,MAAM,MAAM,GAAe,EAAE,CAAC;YAC9B,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE;gBACvB,MAAM,KAAK,GAAW,IAAI,MAAM,CAAC,iBAAiB,GAAG,CAAC,MAAM,CAAC,MAAM,oBAAoB,CAAC,CAAC;gBACzF,MAAM,KAAK,GAAyB,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBAEpD,IAAI,KAAK,KAAK,IAAI;oBAAE,OAAO,mBAAmB,CAAC,uBAAuB,GAAG,EAAE,CAAC,CAAC;gBAE7E,IAAI,OAAO,GAAW,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC/B,IAAI,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;oBAAE,OAAO,GAAG,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;gBAEjE,MAAM,QAAQ,GAAa,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBAC9C,QAAQ,CAAC,GAAG,EAAE,CAAC,CAAC,kBAAkB;gBAElC,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;aACtB;YAED,MAAM,IAAI,GAAiB,mCAAmC,CAAC,MAAM,CAAC,CAAC;YACvE,OAAO,iBAAiB,CAAC,IAAI,CAAC,CAAC;QAChC,CAAC,CAAA,CACF,CAAC;IACH,CAAC;CACD"}
@@ -1,27 +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 { commonsTypeEncode } from 'tscommons-es-core';
11
- import { toEComparator, ECOMPARATORS } from 'hydra-crawler-ts-assets';
12
- import { CommonsRestApi, commonsRestApiBadRequest } from 'nodecommons-es-rest';
13
- export class ImagesApi extends CommonsRestApi {
14
- constructor(restServer, databaseService, path) {
15
- super(restServer);
16
- this.databaseService = databaseService;
17
- super.getHandler(`${path}rest/images/:comparator(${ECOMPARATORS.join('|')})/:size[int]`, (req, _res) => __awaiter(this, void 0, void 0, function* () {
18
- const comparator = toEComparator(req.strictParams.comparator);
19
- if (!comparator)
20
- return commonsRestApiBadRequest('Invalid comparator');
21
- const size = req.strictParams.size;
22
- const matches = yield this.databaseService.listImagesBySizeThreshold(size, comparator);
23
- return commonsTypeEncode(matches);
24
- }));
25
- }
26
- }
27
- //# sourceMappingURL=images.api.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"images.api.js","sourceRoot":"","sources":["../../src/apis/images.api.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,iBAAiB,EAAY,MAAM,mBAAmB,CAAC;AAGhE,OAAO,EAAe,aAAa,EAAE,YAAY,EAAE,MAAM,yBAAyB,CAAC;AAGnF,OAAO,EAAE,cAAc,EAAE,wBAAwB,EAAE,MAAM,qBAAqB,CAAC;AAM/E,MAAM,OAAO,SAAU,SAAQ,cAAiE;IAC/F,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,2BAA2B,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,cAAc,EACtE,CAAO,GAAoC,EAAE,IAAsB,EAAqB,EAAE;YACzF,MAAM,UAAU,GAA0B,aAAa,CAAC,GAAG,CAAC,YAAY,CAAC,UAAoB,CAAC,CAAC;YAC/F,IAAI,CAAC,UAAU;gBAAE,OAAO,wBAAwB,CAAC,oBAAoB,CAAC,CAAC;YAEvE,MAAM,IAAI,GAAW,GAAG,CAAC,YAAY,CAAC,IAAc,CAAC;YAErD,MAAM,OAAO,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,yBAAyB,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;YAEjG,OAAO,iBAAiB,CAAC,OAAO,CAAC,CAAC;QACnC,CAAC,CAAA,CACF,CAAC;IACH,CAAC;CACD"}
@@ -1,41 +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 Cache from 'memcached-promisify';
11
- import { commonsTypeHasPropertyEnum, commonsTypeHasPropertyNumber, commonsTypeIsTArray } from 'tscommons-es-core';
12
- import { isEStatus } from 'hydra-crawler-ts-assets';
13
- import { CommonsRestApi } from 'nodecommons-es-rest';
14
- function isTStatusTally(test) {
15
- if (!commonsTypeHasPropertyEnum(test, 'status', isEStatus))
16
- return false;
17
- if (!commonsTypeHasPropertyNumber(test, 'tally'))
18
- return false;
19
- return true;
20
- }
21
- export class StatisticsApi extends CommonsRestApi {
22
- constructor(restServer, databaseService, path) {
23
- super(restServer);
24
- this.databaseService = databaseService;
25
- // eslint-disable-next-line @typescript-eslint/no-unsafe-assignment, @typescript-eslint/no-unsafe-call
26
- this.memcached = new Cache({ keyPrefix: 'hydra', cacheHost: 'localhost:11211' });
27
- super.getHandler(`${path}rest/statistics/status`, (_req, _res) => __awaiter(this, void 0, void 0, function* () {
28
- const index = 'statistics/status';
29
- const cached = yield this.memcached.get(index);
30
- if (cached && commonsTypeIsTArray(cached, isTStatusTally))
31
- return cached;
32
- const result = yield this.databaseService.listStatusTallies();
33
- const tallies = [];
34
- for (const key of result.keys())
35
- tallies.push({ status: key, tally: result.get(key) || 0 });
36
- yield this.memcached.set(index, tallies, 10);
37
- return tallies;
38
- }));
39
- }
40
- }
41
- //# sourceMappingURL=statistics.api.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"statistics.api.js","sourceRoot":"","sources":["../../src/apis/statistics.api.ts"],"names":[],"mappings":";;;;;;;;;AACA,OAAO,KAAK,MAAM,qBAAqB,CAAC;AAExC,OAAO,EAAE,0BAA0B,EAAE,4BAA4B,EAAE,mBAAmB,EAAE,MAAM,mBAAmB,CAAC;AAElH,OAAO,EAAW,SAAS,EAAE,MAAM,yBAAyB,CAAC;AAE7D,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAarD,SAAS,cAAc,CAAC,IAAa;IACpC,IAAI,CAAC,0BAA0B,CAAC,IAAI,EAAE,QAAQ,EAAE,SAAS,CAAC;QAAE,OAAO,KAAK,CAAC;IACzE,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,OAAO,CAAC;QAAE,OAAO,KAAK,CAAC;IAE/D,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAO,aAAc,SAAQ,cAAiE;IAGnG,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,sGAAsG;QACtG,IAAI,CAAC,SAAS,GAAG,IAAI,KAAK,CAAC,EAAE,SAAS,EAAE,OAAO,EAAE,SAAS,EAAE,iBAAiB,EAAE,CAAC,CAAC;QAEjF,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,wBAAwB,EAC/B,CAAO,IAAqB,EAAE,IAAsB,EAA2B,EAAE;YAChF,MAAM,KAAK,GAAW,mBAAmB,CAAC;YAE1C,MAAM,MAAM,GAAsB,MAAM,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;YAClE,IAAI,MAAM,IAAI,mBAAmB,CAAe,MAAM,EAAE,cAAc,CAAC;gBAAE,OAAO,MAAM,CAAC;YAEvF,MAAM,MAAM,GAAyB,MAAM,IAAI,CAAC,eAAe,CAAC,iBAAiB,EAAE,CAAC;YAEpF,MAAM,OAAO,GAAmB,EAAE,CAAC;YACnC,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,EAAE;gBAAE,OAAO,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,GAAG,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;YAE5F,MAAM,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAAC,CAAC;YAE7C,OAAO,OAAO,CAAC;QAChB,CAAC,CAAA,CACF,CAAC;IACH,CAAC;CACD"}
@@ -1,19 +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 { CommonsRestApi } from 'nodecommons-es-rest';
11
- export class TestApi extends CommonsRestApi {
12
- constructor(restServer, path) {
13
- super(restServer);
14
- super.getHandler(`${path}test/hello`,
15
- // eslint-disable-next-line @typescript-eslint/require-await
16
- (_) => __awaiter(this, void 0, void 0, function* () { return 'Hello world'; }));
17
- }
18
- }
19
- //# sourceMappingURL=test.api.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"test.api.js","sourceRoot":"","sources":["../../src/apis/test.api.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAKrD,MAAM,OAAO,OAAQ,SAAQ,cAAiE;IAC7F,YACE,UAAsB,EACtB,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAElB,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,YAAY;QACnB,4DAA4D;QAC5D,CAAO,CAAU,EAAmB,EAAE,gDAAC,OAAA,aAAa,CAAA,GAAA,CACrD,CAAC;IACH,CAAC;CACD"}
@@ -1,29 +0,0 @@
1
- var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
2
- function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
3
- return new (P || (P = Promise))(function (resolve, reject) {
4
- function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
5
- function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
6
- function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
7
- step((generator = generator.apply(thisArg, _arguments || [])).next());
8
- });
9
- };
10
- import { commonsTypeEncode } from 'tscommons-es-core';
11
- import { EDirection } from 'hydra-crawler-ts-assets';
12
- import { CommonsRestApi } from 'nodecommons-es-rest';
13
- export class UrlsApi extends CommonsRestApi {
14
- constructor(restServer, databaseService, path) {
15
- super(restServer);
16
- this.databaseService = databaseService;
17
- // Can't use strict and ICommonsRequestWithStrictParams as the url could be anything, and idname won't accept that
18
- super.getHandler(`${path}rest/urls/:url/links`, (req, _res) => __awaiter(this, void 0, void 0, function* () {
19
- const outbound = yield this.databaseService.listOutboundLinks(req.params.url);
20
- const inbound = yield this.databaseService.listInboundLinks(req.params.url);
21
- const urlLinks = {
22
- [EDirection.INBOUND]: inbound,
23
- [EDirection.OUTBOUND]: outbound
24
- };
25
- return commonsTypeEncode(urlLinks);
26
- }));
27
- }
28
- }
29
- //# sourceMappingURL=urls.api.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"urls.api.js","sourceRoot":"","sources":["../../src/apis/urls.api.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,iBAAiB,EAAY,MAAM,mBAAmB,CAAC;AAGhE,OAAO,EAAE,UAAU,EAAE,MAAM,yBAAyB,CAAC;AAErD,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAOrD,MAAM,OAAO,OAAQ,SAAQ,cAAiE;IAC7F,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,kHAAkH;QAElH,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,sBAAsB,EAC7B,CAAO,GAAoB,EAAE,IAAsB,EAAqB,EAAE;YACzE,MAAM,QAAQ,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,iBAAiB,CAAC,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YACxF,MAAM,OAAO,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,gBAAgB,CAAC,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YAEtF,MAAM,QAAQ,GAAc;gBAC1B,CAAC,UAAU,CAAC,OAAO,CAAC,EAAE,OAAO;gBAC7B,CAAC,UAAU,CAAC,QAAQ,CAAC,EAAE,QAAQ;aAChC,CAAC;YAEF,OAAO,iBAAiB,CAAC,QAAQ,CAAC,CAAC;QACpC,CAAC,CAAA,CACF,CAAC;IACH,CAAC;CACD"}
@@ -1,151 +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 { isIUrl } from 'hydra-crawler-ts-assets';
12
- import { EStatus } from 'hydra-crawler-ts-assets';
13
- import { commonsOutputDoing, commonsOutputPercent, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
14
- import { CommonsApp } from 'nodecommons-es-app';
15
- import { Lists } from '../classes/lists';
16
- import { Cleaner } from '../classes/cleaner';
17
- export class CleanupApp extends CommonsApp {
18
- constructor(includeReset = false, includeLinks = false) {
19
- super('hydra-crawler');
20
- this.includeReset = includeReset;
21
- this.includeLinks = includeLinks;
22
- this.lists = new Lists();
23
- }
24
- getAppName() {
25
- return 'Hydra - Cleanup';
26
- }
27
- setDatabaseService(databaseService) {
28
- this.databaseService = databaseService;
29
- }
30
- addToList(list, entries) {
31
- this.lists.add(list, entries);
32
- }
33
- init() {
34
- const _super = Object.create(null, {
35
- init: { get: () => super.init }
36
- });
37
- return __awaiter(this, void 0, void 0, function* () {
38
- if (!this.databaseService)
39
- throw new Error('Database service has not been set yet');
40
- commonsOutputDoing('Connecting to database');
41
- yield this.databaseService.init();
42
- commonsOutputSuccess();
43
- yield _super.init.call(this);
44
- });
45
- }
46
- reset() {
47
- return __awaiter(this, void 0, void 0, function* () {
48
- if (!this.databaseService)
49
- throw new Error('Database service has not been set');
50
- commonsOutputDoing('Enumerating urls for DENY, FAILED, DEAD and DISALLOWED');
51
- const urls = [];
52
- let tally = 0;
53
- while (true) {
54
- const result = this.databaseService.getUrls()
55
- .find({ status: { $in: [
56
- EStatus.DENY,
57
- EStatus.FAILED,
58
- EStatus.DEAD,
59
- EStatus.DISALLOWED
60
- ] } })
61
- .sort({ _id: 1 })
62
- .skip(tally);
63
- try {
64
- while (true) {
65
- tally++;
66
- if ((tally % 100) === 0)
67
- commonsOutputProgress(tally);
68
- const row = yield result.next();
69
- if (row === null)
70
- break;
71
- urls.push(row.url);
72
- }
73
- break;
74
- }
75
- catch (err) {
76
- if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
77
- throw err;
78
- }
79
- }
80
- commonsOutputResult(tally);
81
- commonsOutputDoing('Resetting links for DENY, FAILED, DEAD and DISALLOWED');
82
- let i = 0;
83
- for (const url of urls) {
84
- i++;
85
- if ((i % 100) === 0)
86
- commonsOutputPercent(i, urls.length);
87
- yield this.databaseService.getLinks().deleteMany({ url: url });
88
- }
89
- commonsOutputSuccess();
90
- });
91
- }
92
- links() {
93
- return __awaiter(this, void 0, void 0, function* () {
94
- if (!this.databaseService)
95
- throw new Error('Database service has not been set');
96
- commonsOutputDoing('Searching for link orphans');
97
- let tally = 0;
98
- const result = this.databaseService.getLinks().find({}, {});
99
- let batch = [];
100
- const orphans = [];
101
- while (true) {
102
- tally++;
103
- if ((tally % 1000) === 0)
104
- commonsOutputProgress(`${tally}, ${orphans.length}`);
105
- const link = yield result.next();
106
- if (link === null)
107
- break;
108
- if (!batch.includes(link.url))
109
- batch.push(link.url);
110
- if (!batch.includes(link.outgoing))
111
- batch.push(link.outgoing);
112
- if (batch.length < 10000)
113
- continue;
114
- const result2 = this.databaseService.getUrls().find({ url: { $in: batch } });
115
- const matches = (yield this.databaseService.listQueryResults(result2, isIUrl))
116
- .map((u) => u.url);
117
- for (const l of batch) {
118
- if (!matches.includes(l))
119
- orphans.push(l);
120
- }
121
- batch = [];
122
- }
123
- commonsOutputResult(orphans.length);
124
- commonsOutputDoing('Removing orphan links');
125
- tally = 0;
126
- for (const orphan of orphans) {
127
- tally++;
128
- if ((tally % 100) === 0)
129
- commonsOutputProgress(`${tally}`);
130
- yield this.databaseService.getLinks().deleteMany({ url: orphan });
131
- yield this.databaseService.getLinks().deleteMany({ outgoing: orphan });
132
- }
133
- commonsOutputSuccess();
134
- });
135
- }
136
- run() {
137
- return __awaiter(this, void 0, void 0, function* () {
138
- if (!this.databaseService)
139
- throw new Error('Database service has not been set');
140
- if (this.includeReset)
141
- yield this.reset();
142
- if (this.includeLinks)
143
- yield this.links();
144
- const hardLimit = this.getArgs().getNumberOrUndefined('hard-limit');
145
- const cleaner = new Cleaner(this.lists, this.databaseService);
146
- yield cleaner.purgeOrphanUrls(hardLimit);
147
- yield cleaner.purgeEmptyDomains();
148
- });
149
- }
150
- }
151
- //# sourceMappingURL=cleanup.app.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"cleanup.app.js","sourceRoot":"","sources":["../../src/apps/cleanup.app.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,4BAA4B,EAAE,MAAM,mBAAmB,CAAC;AAEjE,OAAO,EAAQ,MAAM,EAAE,MAAM,yBAAyB,CAAC;AAEvD,OAAO,EAAE,OAAO,EAAE,MAAM,yBAAyB,CAAC;AAElD,OAAO,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAChJ,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAEhD,OAAO,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAC;AACzC,OAAO,EAAE,OAAO,EAAE,MAAM,oBAAoB,CAAC;AAU7C,MAAM,OAAO,UAAW,SAAQ,UAAU;IAKzC,YACU,eAAwB,KAAK,EAC7B,eAAwB,KAAK;QAEtC,KAAK,CAAC,eAAe,CAAC,CAAC;QAHd,iBAAY,GAAZ,YAAY,CAAiB;QAC7B,iBAAY,GAAZ,YAAY,CAAiB;QAItC,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;IAC1B,CAAC;IAEM,UAAU;QAChB,OAAO,iBAAiB,CAAC;IAC1B,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;IAEa,KAAK;;YAClB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,kBAAkB,CAAC,wDAAwD,CAAC,CAAC;YAE7E,MAAM,IAAI,GAAa,EAAE,CAAC;YAE1B,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,OAAO,IAAI,EAAE;gBACZ,MAAM,MAAM,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;qBACpE,IAAI,CAAC,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE;4BACrB,OAAO,CAAC,IAAI;4BACZ,OAAO,CAAC,MAAM;4BACd,OAAO,CAAC,IAAI;4BACZ,OAAO,CAAC,UAAU;yBACnB,EAAE,EAAE,CAAC;qBACL,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,MAAM,CAAC,IAAI,EAAE,CAAC;wBAC3C,IAAI,GAAG,KAAK,IAAI;4BAAE,MAAM;wBAExB,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;qBACnB;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;iBAC7E;aACD;YAED,mBAAmB,CAAC,KAAK,CAAC,CAAC;YAE3B,kBAAkB,CAAC,uDAAuD,CAAC,CAAC;YAC5E,IAAI,CAAC,GAAW,CAAC,CAAC;YAClB,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE;gBACvB,CAAC,EAAE,CAAC;gBACJ,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,oBAAoB,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;gBAE1D,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC;aAC/D;YACD,oBAAoB,EAAE,CAAC;QACxB,CAAC;KAAA;IAEa,KAAK;;YAClB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,kBAAkB,CAAC,4BAA4B,CAAC,CAAC;YACjD,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,MAAM,MAAM,GAA8B,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;YAE9F,IAAI,KAAK,GAAa,EAAE,CAAC;YACzB,MAAM,OAAO,GAAa,EAAE,CAAC;YAC7B,OAAO,IAAI,EAAE;gBACZ,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,GAAG,KAAK,KAAK,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC;gBAE/E,MAAM,IAAI,GAAe,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;gBAC7C,IAAI,IAAI,KAAK,IAAI;oBAAE,MAAM;gBAEzB,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC;oBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBACpD,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC;oBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBAE9D,IAAI,KAAK,CAAC,MAAM,GAAG,KAAK;oBAAE,SAAS;gBAEnC,MAAM,OAAO,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CAC3E,EAAE,GAAG,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,CACxB,CAAC;gBAEF,MAAM,OAAO,GAAa,CAAC,MAAM,IAAI,CAAC,eAAe,CAAC,gBAAgB,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;qBACrF,GAAG,CAAC,CAAC,CAAO,EAAU,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;gBAEnC,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE;oBACtB,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC;wBAAE,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;iBAC1C;gBAED,KAAK,GAAG,EAAE,CAAC;aACX;YACD,mBAAmB,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;YAEpC,kBAAkB,CAAC,uBAAuB,CAAC,CAAC;YAC5C,KAAK,GAAG,CAAC,CAAC;YACV,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE;gBAC7B,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,GAAG,KAAK,EAAE,CAAC,CAAC;gBAC3D,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,CAAC;gBAClE,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC;aACvE;YACD,oBAAoB,EAAE,CAAC;QACxB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,IAAI,IAAI,CAAC,YAAY;gBAAE,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;YAC1C,IAAI,IAAI,CAAC,YAAY;gBAAE,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;YAE1C,MAAM,SAAS,GAAqB,IAAI,CAAC,OAAO,EAAE,CAAC,oBAAoB,CAAC,YAAY,CAAC,CAAC;YAEtF,MAAM,OAAO,GAAY,IAAI,OAAO,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,eAAe,CAAC,CAAC;YACvE,MAAM,OAAO,CAAC,eAAe,CAAC,SAAS,CAAC,CAAC;YACzC,MAAM,OAAO,CAAC,iBAAiB,EAAE,CAAC;QACnC,CAAC;KAAA;CACD"}
@@ -1,75 +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 * as fs from 'fs';
11
- import { EStatus } from 'hydra-crawler-ts-assets';
12
- import { commonsOutputDoing, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
13
- import { CommonsApp } from 'nodecommons-es-app';
14
- // export the QUEUED, DONE and FAILED urls for importing into another instance
15
- export class CrossPopulateExportApp extends CommonsApp {
16
- constructor(filename) {
17
- super('hydra-crawler');
18
- this.filename = filename;
19
- }
20
- getAppName() {
21
- return 'Hydra - Cross Populate Export';
22
- }
23
- setDatabaseService(databaseService) {
24
- this.databaseService = databaseService;
25
- }
26
- init() {
27
- const _super = Object.create(null, {
28
- init: { get: () => super.init }
29
- });
30
- return __awaiter(this, void 0, void 0, function* () {
31
- if (!this.databaseService)
32
- throw new Error('Database service has not been set yet');
33
- commonsOutputDoing('Connecting to database');
34
- yield this.databaseService.init();
35
- commonsOutputSuccess();
36
- yield _super.init.call(this);
37
- });
38
- }
39
- run() {
40
- return __awaiter(this, void 0, void 0, function* () {
41
- if (!this.databaseService)
42
- throw new Error('Database service has not been set');
43
- const results = this.databaseService.getUrls().find({
44
- status: { $in: [
45
- EStatus.QUEUED,
46
- EStatus.DONE,
47
- EStatus.FAILED
48
- ] }
49
- }, {});
50
- // safer to do this directly rather than a call to listQueryResults
51
- // ditto not using CommonsFile
52
- const fd = fs.openSync(this.filename, 'w');
53
- commonsOutputDoing('Exporting QUEUED, DONE and FAILED urls');
54
- let tally = 0;
55
- while (true) {
56
- if (tally % 1000 === 0)
57
- commonsOutputProgress(tally);
58
- const row = yield results.next();
59
- if (row === null)
60
- break;
61
- fs.writeSync(fd, JSON.stringify({
62
- domain: row.domain,
63
- url: row.url
64
- }) + '\n');
65
- if (tally % 1000 === 0)
66
- fs.fdatasyncSync(fd);
67
- tally++;
68
- }
69
- fs.fdatasyncSync(fd);
70
- fs.closeSync(fd);
71
- commonsOutputResult(tally);
72
- });
73
- }
74
- }
75
- //# sourceMappingURL=cross-populate-export.app.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"cross-populate-export.app.js","sourceRoot":"","sources":["../../src/apps/cross-populate-export.app.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,KAAK,EAAE,MAAM,IAAI,CAAC;AAKzB,OAAO,EAAE,OAAO,EAAE,MAAM,yBAAyB,CAAC;AAElD,OAAO,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAC1H,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAMhD,8EAA8E;AAE9E,MAAM,OAAO,sBAAuB,SAAQ,UAAU;IAGrD,YACU,QAAgB;QAEzB,KAAK,CAAC,eAAe,CAAC,CAAC;QAFd,aAAQ,GAAR,QAAQ,CAAQ;IAG1B,CAAC;IAEM,UAAU;QAChB,OAAO,+BAA+B,CAAC;IACxC,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,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,MAAM,OAAO,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CAAO;gBAClF,MAAM,EAAE,EAAE,GAAG,EAAE;wBACb,OAAO,CAAC,MAAM;wBACd,OAAO,CAAC,IAAI;wBACZ,OAAO,CAAC,MAAM;qBACf,EAAE;aACJ,EAAE,EAAE,CAAC,CAAC;YAEP,mEAAmE;YACnE,8BAA8B;YAE9B,MAAM,EAAE,GAAW,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;YAEnD,kBAAkB,CAAC,wCAAwC,CAAC,CAAC;YAE7D,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,OAAO,IAAI,EAAE;gBACZ,IAAI,KAAK,GAAG,IAAI,KAAK,CAAC;oBAAE,qBAAqB,CAAC,KAAK,CAAC,CAAC;gBAErD,MAAM,GAAG,GAAc,MAAM,OAAO,CAAC,IAAI,EAAE,CAAC;gBAC5C,IAAI,GAAG,KAAK,IAAI;oBAAE,MAAM;gBAExB,EAAE,CAAC,SAAS,CACV,EAAE,EACF,IAAI,CAAC,SAAS,CAAC;oBACb,MAAM,EAAE,GAAG,CAAC,MAAM;oBAClB,GAAG,EAAE,GAAG,CAAC,GAAG;iBACb,CAAC,GAAG,IAAI,CACV,CAAC;gBAEF,IAAI,KAAK,GAAG,IAAI,KAAK,CAAC;oBAAE,EAAE,CAAC,aAAa,CAAC,EAAE,CAAC,CAAC;gBAC7C,KAAK,EAAE,CAAC;aACR;YAED,EAAE,CAAC,aAAa,CAAC,EAAE,CAAC,CAAC;YACrB,EAAE,CAAC,SAAS,CAAC,EAAE,CAAC,CAAC;YAEjB,mBAAmB,CAAC,KAAK,CAAC,CAAC;QAC5B,CAAC;KAAA;CACD"}
@@ -1,100 +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
- var __asyncValues = (this && this.__asyncValues) || function (o) {
11
- if (!Symbol.asyncIterator) throw new TypeError("Symbol.asyncIterator is not defined.");
12
- var m = o[Symbol.asyncIterator], i;
13
- return m ? m.call(o) : (o = typeof __values === "function" ? __values(o) : o[Symbol.iterator](), i = {}, verb("next"), verb("throw"), verb("return"), i[Symbol.asyncIterator] = function () { return this; }, i);
14
- function verb(n) { i[n] = o[n] && function (v) { return new Promise(function (resolve, reject) { v = o[n](v), settle(resolve, reject, v.done, v.value); }); }; }
15
- function settle(resolve, reject, d, v) { Promise.resolve(v).then(function(v) { resolve({ value: v, done: d }); }, reject); }
16
- };
17
- import * as fs from 'fs';
18
- import * as readline from 'readline';
19
- import { commonsTypeHasPropertyString } from 'tscommons-es-core';
20
- import { commonsOutputDoing, commonsOutputError, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
21
- import { CommonsApp } from 'nodecommons-es-app';
22
- function isTRow(test) {
23
- if (!commonsTypeHasPropertyString(test, 'domain'))
24
- return false;
25
- if (!commonsTypeHasPropertyString(test, 'url'))
26
- return false;
27
- return true;
28
- }
29
- export class CrossPopulateImportApp extends CommonsApp {
30
- constructor(filename) {
31
- super('hydra-crawler');
32
- this.filename = filename;
33
- }
34
- getAppName() {
35
- return 'Hydra - Cross Populate Export';
36
- }
37
- setDatabaseService(databaseService) {
38
- this.databaseService = databaseService;
39
- }
40
- init() {
41
- const _super = Object.create(null, {
42
- init: { get: () => super.init }
43
- });
44
- return __awaiter(this, void 0, void 0, function* () {
45
- if (!this.databaseService)
46
- throw new Error('Database service has not been set yet');
47
- commonsOutputDoing('Connecting to database');
48
- yield this.databaseService.init();
49
- commonsOutputSuccess();
50
- yield _super.init.call(this);
51
- });
52
- }
53
- run() {
54
- var _a, e_1, _b, _c;
55
- return __awaiter(this, void 0, void 0, function* () {
56
- if (!this.databaseService)
57
- throw new Error('Database service has not been set');
58
- const fileStream = fs.createReadStream(this.filename);
59
- const rl = readline.createInterface({
60
- input: fileStream,
61
- crlfDelay: Infinity
62
- });
63
- commonsOutputDoing('Importing urls');
64
- let tally = 0;
65
- try {
66
- for (var _d = true, rl_1 = __asyncValues(rl), rl_1_1; rl_1_1 = yield rl_1.next(), _a = rl_1_1.done, !_a;) {
67
- _c = rl_1_1.value;
68
- _d = false;
69
- try {
70
- const line = _c;
71
- if (line.trim() === '')
72
- continue;
73
- if (tally % 1000 === 0)
74
- commonsOutputProgress(tally);
75
- const json = JSON.parse(line);
76
- if (!isTRow(json)) {
77
- commonsOutputError('Invalid row. Skipping');
78
- continue;
79
- }
80
- yield this.databaseService.queue(json.url, false);
81
- tally++;
82
- }
83
- finally {
84
- _d = true;
85
- }
86
- }
87
- }
88
- catch (e_1_1) { e_1 = { error: e_1_1 }; }
89
- finally {
90
- try {
91
- if (!_d && !_a && (_b = rl_1.return)) yield _b.call(rl_1);
92
- }
93
- finally { if (e_1) throw e_1.error; }
94
- }
95
- commonsOutputResult(tally);
96
- fileStream.close();
97
- });
98
- }
99
- }
100
- //# sourceMappingURL=cross-populate-import.app.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"cross-populate-import.app.js","sourceRoot":"","sources":["../../src/apps/cross-populate-import.app.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;AAAA,OAAO,KAAK,EAAE,MAAM,IAAI,CAAC;AACzB,OAAO,KAAK,QAAQ,MAAM,UAAU,CAAC;AAErC,OAAO,EAAE,4BAA4B,EAAE,MAAM,mBAAmB,CAAC;AAEjE,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAC9I,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAahD,SAAS,MAAM,CAAC,IAAa;IAC5B,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,QAAQ,CAAC;QAAE,OAAO,KAAK,CAAC;IAChE,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,KAAK,CAAC;QAAE,OAAO,KAAK,CAAC;IAE7D,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAO,sBAAuB,SAAQ,UAAU;IAGrD,YACU,QAAgB;QAEzB,KAAK,CAAC,eAAe,CAAC,CAAC;QAFd,aAAQ,GAAR,QAAQ,CAAQ;IAG1B,CAAC;IAEM,UAAU;QAChB,OAAO,+BAA+B,CAAC;IACxC,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,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,MAAM,UAAU,GAAkB,EAAE,CAAC,gBAAgB,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;YAErE,MAAM,EAAE,GAAuB,QAAQ,CAAC,eAAe,CAAC;gBACtD,KAAK,EAAE,UAAU;gBACjB,SAAS,EAAE,QAAQ;aACpB,CAAC,CAAC;YAEH,kBAAkB,CAAC,gBAAgB,CAAC,CAAC;YAErC,IAAI,KAAK,GAAW,CAAC,CAAC;;gBACtB,KAAyB,eAAA,OAAA,cAAA,EAAE,CAAA,QAAA;oBAAF,kBAAE;oBAAF,WAAE;;wBAAhB,MAAM,IAAI,KAAA,CAAA;wBACpB,IAAI,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE;4BAAE,SAAS;wBAEjC,IAAI,KAAK,GAAG,IAAI,KAAK,CAAC;4BAAE,qBAAqB,CAAC,KAAK,CAAC,CAAC;wBAErD,MAAM,IAAI,GAAY,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;wBACvC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE;4BAClB,kBAAkB,CAAC,uBAAuB,CAAC,CAAC;4BAC5C,SAAS;yBACT;wBAED,MAAM,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;wBAClD,KAAK,EAAE,CAAC;;;;;iBACR;;;;;;;;;YACD,mBAAmB,CAAC,KAAK,CAAC,CAAC;YAE3B,UAAU,CAAC,KAAK,EAAE,CAAC;;KACnB;CACD"}