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
@@ -0,0 +1,20 @@
1
+ import * as cheerio from 'cheerio';
2
+ import { TPropertyObject } from 'tscommons-esm-core';
3
+ import { DatabaseService } from '../services/database.service.mjs';
4
+ import { IParserConfig } from '../interfaces/iparser-config.mjs';
5
+ import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
6
+ import { HtmlParser } from './html.parser.mjs';
7
+ import { IDataConfig } from './data.parser.mjs';
8
+ export interface IHyperlinksConfig extends IDataConfig {
9
+ sources: string[];
10
+ }
11
+ export declare function isIHyperlinksConfig(test: unknown): test is IHyperlinksConfig;
12
+ export declare class HyperlinksParser extends HtmlParser<IHyperlinksConfig> {
13
+ private url?;
14
+ static parseTagLinks(dom: cheerio.CheerioAPI, tag: string, attribute: string): string[];
15
+ static getBaseHref(dom: cheerio.CheerioAPI, url: string): string | undefined;
16
+ private hyperlinksConfig;
17
+ constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>);
18
+ parse(database: DatabaseService): Promise<void>;
19
+ links(): Promise<string[]>;
20
+ }
@@ -0,0 +1,104 @@
1
+ import { URL } from 'url';
2
+ import { commonsArrayUnique, commonsTypeHasPropertyStringArray } from 'tscommons-esm-core';
3
+ import { HtmlParser } from './html.parser.mjs';
4
+ import { isIDataConfig } from './data.parser.mjs';
5
+ export function isIHyperlinksConfig(test) {
6
+ if (!isIDataConfig(test))
7
+ return false;
8
+ if (!commonsTypeHasPropertyStringArray(test, 'sources'))
9
+ return false;
10
+ return true;
11
+ }
12
+ export class HyperlinksParser extends HtmlParser {
13
+ url;
14
+ static parseTagLinks(dom, tag, attribute) {
15
+ const links = [];
16
+ dom(`${tag}[${attribute}]`).each((_index, element) => {
17
+ const link = dom(element).attr(attribute);
18
+ if (link !== undefined)
19
+ links.push(link);
20
+ return true;
21
+ });
22
+ return commonsArrayUnique(links);
23
+ }
24
+ static getBaseHref(dom, url) {
25
+ try {
26
+ const base = dom('base[href]');
27
+ if (base.length === 0)
28
+ return undefined;
29
+ const href = dom(base[0]).attr('href');
30
+ if (!href)
31
+ return undefined;
32
+ const whatwg = new URL(href, url);
33
+ return whatwg.toString();
34
+ }
35
+ catch (_ex) { /* do nothing */ }
36
+ return undefined;
37
+ }
38
+ hyperlinksConfig;
39
+ constructor(url, outcome, config) {
40
+ super(outcome, config, 'hyperlinks');
41
+ this.url = url;
42
+ this.hyperlinksConfig = this.getConfig(isIHyperlinksConfig);
43
+ }
44
+ async parse(database) {
45
+ if (!this.dom || !this.url || !this.hyperlinksConfig)
46
+ return; // invalid parse
47
+ let title;
48
+ try {
49
+ title = this.dom('title').text();
50
+ }
51
+ catch (_ex) { /* do nothing */ }
52
+ if (title !== undefined && title !== null && title !== '') {
53
+ await database.setData(this.url, 'title', title);
54
+ }
55
+ const links = {};
56
+ for (const source of this.hyperlinksConfig.sources) {
57
+ const pattern = source.match(/^([a-z]+)\[(src|href|action)\]$/);
58
+ if (pattern)
59
+ links[pattern[1]] = HyperlinksParser.parseTagLinks(this.dom, pattern[1], pattern[2]);
60
+ }
61
+ await database.setData(this.url, 'links', links);
62
+ }
63
+ // eslint-disable-next-line @typescript-eslint/require-await
64
+ async links() {
65
+ if (!this.dom || !this.url || !this.hyperlinksConfig)
66
+ return []; // invalid parse
67
+ let links = [];
68
+ for (const source of this.hyperlinksConfig.sources) {
69
+ const pattern = source.match(/^([a-z]+)\[(src|href|action)\]$/);
70
+ if (pattern) {
71
+ if (pattern[1] === 'form')
72
+ continue;
73
+ links = [...links, ...HyperlinksParser.parseTagLinks(this.dom, pattern[1], pattern[2])];
74
+ }
75
+ }
76
+ const unique = [...new Set(links)];
77
+ let baseHref = HyperlinksParser.getBaseHref(this.dom, this.url);
78
+ if (baseHref === undefined)
79
+ baseHref = this.url;
80
+ const results = [];
81
+ for (let l of unique) {
82
+ const anchor = l.split('#');
83
+ l = anchor[0];
84
+ if (l === '')
85
+ continue;
86
+ try {
87
+ const u = new URL(l, baseHref);
88
+ if (u.toString().indexOf('#') !== -1) {
89
+ process.exit(1);
90
+ }
91
+ if (!u.protocol.match(/^http(s?):$/))
92
+ continue;
93
+ if (u.toString() === this.url || u.toString() === baseHref) {
94
+ // skip self-links, or presumed self-links
95
+ continue;
96
+ }
97
+ results.push(u.toString());
98
+ }
99
+ catch (_ex) { /* do nothing */ }
100
+ }
101
+ return results;
102
+ }
103
+ }
104
+ //# sourceMappingURL=hyperlinks.parser.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"hyperlinks.parser.mjs","sourceRoot":"","sources":["../../src/parsers/hyperlinks.parser.mts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,KAAK,CAAC;AAI1B,OAAO,EAAE,kBAAkB,EAAE,iCAAiC,EAAmB,MAAM,oBAAoB,CAAC;AAO5G,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAC/C,OAAO,EAAe,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAK/D,MAAM,UAAU,mBAAmB,CAAC,IAAa;IAChD,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IAEvC,IAAI,CAAC,iCAAiC,CAAC,IAAI,EAAE,SAAS,CAAC;QAAE,OAAO,KAAK,CAAC;IAEtE,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAO,gBAAiB,SAAQ,UAA6B;IAmCxD;IAjCH,MAAM,CAAC,aAAa,CACzB,GAAuB,EACvB,GAAW,EACX,SAAiB;QAElB,MAAM,KAAK,GAAa,EAAE,CAAC;QAE3B,GAAG,CAAC,GAAG,GAAG,IAAI,SAAS,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAY,EAAW,EAAE;YAC1E,MAAM,IAAI,GAAqB,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;YAC5D,IAAI,IAAI,KAAK,SAAS;gBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACzC,OAAO,IAAI,CAAC;QACb,CAAC,CAAC,CAAC;QAEH,OAAO,kBAAkB,CAAC,KAAK,CAAC,CAAC;IAClC,CAAC;IAEM,MAAM,CAAC,WAAW,CAAC,GAAuB,EAAE,GAAW;QAC7D,IAAI,CAAC;YACJ,MAAM,IAAI,GAAyB,GAAG,CAAC,YAAY,CAAC,CAAC;YACrD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;gBAAE,OAAO,SAAS,CAAC;YAExC,MAAM,IAAI,GAAqB,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YACzD,IAAI,CAAC,IAAI;gBAAE,OAAO,SAAS,CAAC;YAE5B,MAAM,MAAM,GAAQ,IAAI,GAAG,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;YACvC,OAAO,MAAM,CAAC,QAAQ,EAAE,CAAC;QAC1B,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC,CAAC,gBAAgB,CAAC,CAAC;QAClC,OAAO,SAAS,CAAC;IAClB,CAAC;IAEO,gBAAgB,CAA8B;IAEtD,YACU,GAAY,EACpB,OAAyB,EACzB,MAAuC;QAExC,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,YAAY,CAAC,CAAC;QAJ5B,QAAG,GAAH,GAAG,CAAS;QAMrB,IAAI,CAAC,gBAAgB,GAAG,IAAI,CAAC,SAAS,CAAC,mBAAmB,CAAC,CAAC;IAC7D,CAAC;IAEe,KAAK,CAAC,KAAK,CAAC,QAAyB;QACpD,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,gBAAgB;YAAE,OAAO,CAAC,gBAAgB;QAE9E,IAAI,KAAuB,CAAC;QAC5B,IAAI,CAAC;YACJ,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,CAAC;QAClC,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC,CAAC,gBAAgB,CAAC,CAAC;QAElC,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,KAAK,EAAE,EAAE,CAAC;YAC3D,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC;QAClD,CAAC;QAED,MAAM,KAAK,GAAkC,EAAE,CAAC;QAChD,KAAK,MAAM,MAAM,IAAI,IAAI,CAAC,gBAAgB,CAAC,OAAO,EAAE,CAAC;YACpD,MAAM,OAAO,GAA0B,MAAM,CAAC,KAAK,CAAC,iCAAiC,CAAC,CAAC;YACvF,IAAI,OAAO;gBAAE,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,GAAG,gBAAgB,CAAC,aAAa,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;QACnG,CAAC;QAED,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC;IAClD,CAAC;IAED,4DAA4D;IAC5C,KAAK,CAAC,KAAK;QAC1B,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,gBAAgB;YAAE,OAAO,EAAE,CAAC,CAAC,gBAAgB;QAEjF,IAAI,KAAK,GAAa,EAAE,CAAC;QACzB,KAAK,MAAM,MAAM,IAAI,IAAI,CAAC,gBAAgB,CAAC,OAAO,EAAE,CAAC;YACpD,MAAM,OAAO,GAA0B,MAAM,CAAC,KAAK,CAAC,iCAAiC,CAAC,CAAC;YACvF,IAAI,OAAO,EAAE,CAAC;gBACb,IAAI,OAAO,CAAC,CAAC,CAAC,KAAK,MAAM;oBAAE,SAAS;gBACpC,KAAK,GAAG,CAAC,GAAG,KAAK,EAAE,GAAG,gBAAgB,CAAC,aAAa,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YACzF,CAAC;QACF,CAAC;QAED,MAAM,MAAM,GAAa,CAAC,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC;QAE7C,IAAI,QAAQ,GAAqB,gBAAgB,CAAC,WAAW,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC;QAClF,IAAI,QAAQ,KAAK,SAAS;YAAE,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC;QAEhD,MAAM,OAAO,GAAa,EAAE,CAAC;QAC7B,KAAK,IAAI,CAAC,IAAI,MAAM,EAAE,CAAC;YACtB,MAAM,MAAM,GAAa,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YACtC,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;YACd,IAAI,CAAC,KAAK,EAAE;gBAAE,SAAS;YAEvB,IAAI,CAAC;gBACJ,MAAM,CAAC,GAAQ,IAAI,GAAG,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC;gBACpC,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;oBACtC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;gBACjB,CAAC;gBACD,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,aAAa,CAAC;oBAAE,SAAS;gBAC/C,IAAI,CAAC,CAAC,QAAQ,EAAE,KAAK,IAAI,CAAC,GAAG,IAAI,CAAC,CAAC,QAAQ,EAAE,KAAK,QAAQ,EAAE,CAAC;oBAC5D,0CAA0C;oBAC1C,SAAS;gBACV,CAAC;gBAED,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAC;YAC5B,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC,CAAC,gBAAgB,CAAC,CAAC;QACnC,CAAC;QAED,OAAO,OAAO,CAAC;IAChB,CAAC;CACD"}
@@ -0,0 +1,19 @@
1
+ import * as cheerio from 'cheerio';
2
+ import { TPropertyObject } from 'tscommons-esm-core';
3
+ import { DatabaseService } from '../services/database.service.mjs';
4
+ import { IParserConfig } from '../interfaces/iparser-config.mjs';
5
+ import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
6
+ import { HtmlParser } from './html.parser.mjs';
7
+ import { IDataConfig } from './data.parser.mjs';
8
+ type TImageTag = {
9
+ src: string;
10
+ alt?: string;
11
+ title?: string;
12
+ };
13
+ export declare class ImageTagsParser extends HtmlParser<IDataConfig> {
14
+ private url?;
15
+ static parseImageTags(dom: cheerio.CheerioAPI): TImageTag[];
16
+ constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>);
17
+ parse(database: DatabaseService): Promise<void>;
18
+ }
19
+ export {};
@@ -0,0 +1,42 @@
1
+ import { commonsLogDebug } from 'nodecommons-esm-log';
2
+ import { HtmlParser } from './html.parser.mjs';
3
+ export class ImageTagsParser extends HtmlParser {
4
+ url;
5
+ static parseImageTags(dom) {
6
+ const images = [];
7
+ dom('img').each((_index, element) => {
8
+ const src = dom(element).attr('src');
9
+ if (src === null || src === undefined)
10
+ return true;
11
+ const alt = dom(element).attr('alt');
12
+ const title = dom(element).attr('title');
13
+ const image = {
14
+ src: src
15
+ };
16
+ if (alt !== undefined && alt !== null)
17
+ image.alt = alt;
18
+ if (title !== undefined && title !== null)
19
+ image.title = title;
20
+ images.push(image);
21
+ return true;
22
+ });
23
+ return images;
24
+ }
25
+ constructor(url, outcome, config) {
26
+ super(outcome, config, 'imageTags');
27
+ this.url = url;
28
+ }
29
+ async parse(database) {
30
+ if (!this.dom || !this.url)
31
+ return; // invalid parse
32
+ const images = ImageTagsParser.parseImageTags(this.dom);
33
+ if (images.length > 0) {
34
+ commonsLogDebug(`Detected images for ${this.url}`);
35
+ await database.setData(this.url, 'images', images);
36
+ }
37
+ else {
38
+ await database.unsetData(this.url, 'images');
39
+ }
40
+ }
41
+ }
42
+ //# sourceMappingURL=image-tags.parser.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"image-tags.parser.mjs","sourceRoot":"","sources":["../../src/parsers/image-tags.parser.mts"],"names":[],"mappings":"AAIA,OAAO,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAOtD,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAS/C,MAAM,OAAO,eAAgB,SAAQ,UAAuB;IA4BjD;IA3BH,MAAM,CAAC,cAAc,CAC1B,GAAuB;QAExB,MAAM,MAAM,GAAgB,EAAE,CAAC;QAE/B,GAAG,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAY,EAAW,EAAE;YACzD,MAAM,GAAG,GAA0B,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YAC5D,IAAI,GAAG,KAAK,IAAI,IAAI,GAAG,KAAK,SAAS;gBAAE,OAAO,IAAI,CAAC;YAEnD,MAAM,GAAG,GAA0B,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YAC5D,MAAM,KAAK,GAA0B,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;YAEhE,MAAM,KAAK,GAAc;gBACvB,GAAG,EAAE,GAAG;aACT,CAAC;YACF,IAAI,GAAG,KAAK,SAAS,IAAI,GAAG,KAAK,IAAI;gBAAE,KAAK,CAAC,GAAG,GAAG,GAAG,CAAC;YACvD,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,IAAI;gBAAE,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC;YAE/D,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YAEnB,OAAO,IAAI,CAAC;QACb,CAAC,CAAC,CAAC;QAEH,OAAO,MAAM,CAAC;IACf,CAAC;IAED,YACU,GAAY,EACpB,OAAyB,EACzB,MAAuC;QAExC,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,WAAW,CAAC,CAAC;QAJ3B,QAAG,GAAH,GAAG,CAAS;IAKtB,CAAC;IAEe,KAAK,CAAC,KAAK,CAAC,QAAyB;QACpD,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG;YAAE,OAAO,CAAC,gBAAgB;QAEpD,MAAM,MAAM,GAAgB,eAAe,CAAC,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACrE,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACvB,eAAe,CAAC,uBAAuB,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC;YACnD,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,EAAE,MAAM,CAAC,CAAC;QACpD,CAAC;aAAM,CAAC;YACP,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC;QAC9C,CAAC;IACF,CAAC;CACD"}
@@ -1,9 +1,9 @@
1
- import { TKeyObject } from 'tscommons-es-core';
2
- import { DatabaseService } from '../services/database.service';
3
- import { IParserConfig } from '../interfaces/iparser-config';
4
- import { IRequestOutcome } from '../interfaces/irequest-outcome';
5
- import { IOllamaConfig, OllamaParser } from './ollama.parser';
6
- import { IDataConfig } from './data.parser';
1
+ import { TPropertyObject } from 'tscommons-esm-core';
2
+ import { DatabaseService } from '../services/database.service.mjs';
3
+ import { IParserConfig } from '../interfaces/iparser-config.mjs';
4
+ import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
5
+ import { IOllamaConfig, OllamaParser } from './ollama.parser.mjs';
6
+ import { IDataConfig } from './data.parser.mjs';
7
7
  export interface IInterestConfig extends IDataConfig {
8
8
  model: string;
9
9
  permittedDomains: string[];
@@ -13,6 +13,6 @@ export declare class InterestParser extends OllamaParser<IOllamaConfig> {
13
13
  private url?;
14
14
  private model;
15
15
  private permittedDomains;
16
- constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TKeyObject<IParserConfig>);
16
+ constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>);
17
17
  protected parseText(database: DatabaseService, text: string): Promise<void>;
18
18
  }
@@ -0,0 +1,60 @@
1
+ import { commonsTypeHasPropertyString, commonsTypeHasPropertyStringArray, commonsTypeHasPropertyT } from 'tscommons-esm-core';
2
+ import { commonsLogDebug, commonsLogError } from 'nodecommons-esm-log';
3
+ import { RigSubjectivityScaleService } from '../services/rig-subjectivity-scale.service.mjs';
4
+ import { OllamaParser } from './ollama.parser.mjs';
5
+ import { isIDataConfig } from './data.parser.mjs';
6
+ export function isIInterestConfig(test) {
7
+ if (!isIDataConfig(test))
8
+ return false;
9
+ if (!commonsTypeHasPropertyString(test, 'model'))
10
+ return false;
11
+ if (!commonsTypeHasPropertyStringArray(test, 'permittedDomains'))
12
+ return false;
13
+ return true;
14
+ }
15
+ export class InterestParser extends OllamaParser {
16
+ url;
17
+ model;
18
+ permittedDomains = [];
19
+ constructor(url, outcome, config) {
20
+ super(outcome, config, 'interest');
21
+ this.url = url;
22
+ if (!config)
23
+ return;
24
+ if (!commonsTypeHasPropertyT(config, 'interest', isIInterestConfig)) {
25
+ commonsLogError('Invalid config for InterestParser');
26
+ }
27
+ this.model = config['interest'].model;
28
+ this.permittedDomains = config['interest'].permittedDomains;
29
+ }
30
+ async parseText(database, text) {
31
+ if (!this.url)
32
+ return;
33
+ if (!this.ollamaUrl || !this.ollamaKey || !this.model)
34
+ return;
35
+ const hostname = new URL(this.url).hostname;
36
+ if (!this.permittedDomains.includes(hostname))
37
+ return;
38
+ if (text.length === 0) {
39
+ await database.unsetData(this.url, 'interest');
40
+ return;
41
+ }
42
+ const rigService = new RigSubjectivityScaleService(this.ollamaUrl, this.ollamaKey, this.model);
43
+ await this.debugGetLength();
44
+ // run this outside of await so that it doesn't block other things
45
+ void (async () => {
46
+ if (!this.url)
47
+ return;
48
+ const score = await rigService.generateInterestScore(text, this.ollamaKeepAlive);
49
+ if (score === undefined)
50
+ return;
51
+ if (score.score < 2)
52
+ delete score.reason;
53
+ commonsLogDebug(`Interest score is: ${score.score.toString(10)}`);
54
+ if (score.reason)
55
+ commonsLogDebug(`Reason for score is: ${score.reason}`);
56
+ await database.setData(this.url, 'interest', score);
57
+ })();
58
+ }
59
+ }
60
+ //# sourceMappingURL=interest.parser.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"interest.parser.mjs","sourceRoot":"","sources":["../../src/parsers/interest.parser.mts"],"names":[],"mappings":"AAAA,OAAO,EAAmB,4BAA4B,EAAE,iCAAiC,EAAE,uBAAuB,EAAE,MAAM,oBAAoB,CAAC;AAE/I,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAGvE,OAAO,EAAE,2BAA2B,EAAuB,MAAM,gDAAgD,CAAC;AAKlH,OAAO,EAAiB,YAAY,EAAE,MAAM,qBAAqB,CAAC;AAClE,OAAO,EAAe,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAM/D,MAAM,UAAU,iBAAiB,CAAC,IAAa;IAC9C,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IAEvC,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,OAAO,CAAC;QAAE,OAAO,KAAK,CAAC;IAC/D,IAAI,CAAC,iCAAiC,CAAC,IAAI,EAAE,kBAAkB,CAAC;QAAE,OAAO,KAAK,CAAC;IAE/E,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAO,cAAe,SAAQ,YAA2B;IAKpD;IAJF,KAAK,CAAmB;IACxB,gBAAgB,GAAa,EAAE,CAAC;IAExC,YACU,GAAY,EACpB,OAAyB,EACzB,MAAuC;QAExC,KAAK,CACH,OAAO,EACP,MAAM,EACN,UAAU,CACX,CAAC;QARO,QAAG,GAAH,GAAG,CAAS;QAUrB,IAAI,CAAC,MAAM;YAAE,OAAO;QAEpB,IAAI,CAAC,uBAAuB,CAAkB,MAAM,EAAE,UAAU,EAAE,iBAAiB,CAAC,EAAE,CAAC;YACtF,eAAe,CAAC,mCAAmC,CAAC,CAAC;QACtD,CAAC;QACD,IAAI,CAAC,KAAK,GAAI,MAAM,CAAC,UAAU,CAAqB,CAAC,KAAK,CAAC;QAC3D,IAAI,CAAC,gBAAgB,GAAI,MAAM,CAAC,UAAU,CAAqB,CAAC,gBAAgB,CAAC;IAClF,CAAC;IAES,KAAK,CAAC,SAAS,CAAC,QAAyB,EAAE,IAAY;QAChE,IAAI,CAAC,IAAI,CAAC,GAAG;YAAE,OAAO;QACtB,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,KAAK;YAAE,OAAO;QAE9D,MAAM,QAAQ,GAAW,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC;QACpD,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,QAAQ,CAAC;YAAE,OAAO;QAEtD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACvB,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,UAAU,CAAC,CAAC;YAC/C,OAAO;QACR,CAAC;QAED,MAAM,UAAU,GAAgC,IAAI,2BAA2B,CAC7E,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,KAAK,CACX,CAAC;QAEF,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;QAE5B,kEAAkE;QAClE,KAAK,CAAC,KAAK,IAAmB,EAAE;YAC/B,IAAI,CAAC,IAAI,CAAC,GAAG;gBAAE,OAAO;YAEtB,MAAM,KAAK,GAAkC,MAAM,UAAU,CAAC,qBAAqB,CACjF,IAAI,EACJ,IAAI,CAAC,eAAe,CACrB,CAAC;YACF,IAAI,KAAK,KAAK,SAAS;gBAAE,OAAO;YAEhC,IAAI,KAAK,CAAC,KAAK,GAAG,CAAC;gBAAE,OAAO,KAAK,CAAC,MAAM,CAAC;YAEzC,eAAe,CAAC,sBAAsB,KAAK,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;YAClE,IAAI,KAAK,CAAC,MAAM;gBAAE,eAAe,CAAC,wBAAwB,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC;YAE1E,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,UAAU,EAAE,KAAK,CAAC,CAAC;QACrD,CAAC,CAAC,EAAE,CAAC;IACN,CAAC;CACD"}
@@ -0,0 +1,11 @@
1
+ import { TPropertyObject } from 'tscommons-esm-core';
2
+ import { DatabaseService } from '../services/database.service.mjs';
3
+ import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
4
+ import { IParserConfig } from '../interfaces/iparser-config.mjs';
5
+ import { DataParser, IDataConfig } from './data.parser.mjs';
6
+ export declare class JpegParser extends DataParser<IDataConfig> {
7
+ private url?;
8
+ constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>);
9
+ supports(contentType: string, isAllow: boolean): boolean;
10
+ parse(database: DatabaseService): Promise<void>;
11
+ }
@@ -0,0 +1,29 @@
1
+ import * as jpeg from 'jpeg-js';
2
+ import { DataParser } from './data.parser.mjs';
3
+ export class JpegParser extends DataParser {
4
+ url;
5
+ constructor(url, outcome, config) {
6
+ super(outcome, config, 'jpeg');
7
+ this.url = url;
8
+ }
9
+ supports(contentType, isAllow) {
10
+ if (!isAllow)
11
+ return false;
12
+ return contentType.match(/^image\/(jpeg|jpg)$/) ? true : false;
13
+ }
14
+ async parse(database) {
15
+ if (!this.outcome || !this.url)
16
+ return;
17
+ if (!this.outcome.data)
18
+ return;
19
+ if (this.outcome.exceeded)
20
+ return; // don't try and parse vast JPEGs than haven't been downloaded properly, regardless of what the IDataConfig.allowExceeded is set to
21
+ try {
22
+ const rawImageData = jpeg.decode(this.outcome.data);
23
+ await database.setData(this.url, 'width', rawImageData.width);
24
+ await database.setData(this.url, 'height', rawImageData.height);
25
+ }
26
+ catch (_ex) { /* do nothing */ }
27
+ }
28
+ }
29
+ //# sourceMappingURL=jpeg.parser.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"jpeg.parser.mjs","sourceRoot":"","sources":["../../src/parsers/jpeg.parser.mts"],"names":[],"mappings":"AAAA,OAAO,KAAK,IAAI,MAAM,SAAS,CAAC;AAShC,OAAO,EAAE,UAAU,EAAe,MAAM,mBAAmB,CAAC;AAE5D,MAAM,OAAO,UAAW,SAAQ,UAAuB;IAE5C;IADV,YACU,GAAY,EACpB,OAAyB,EACzB,MAAuC;QAExC,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC;QAJtB,QAAG,GAAH,GAAG,CAAS;IAKtB,CAAC;IAEM,QAAQ,CAAC,WAAmB,EAAE,OAAgB;QACpD,IAAI,CAAC,OAAO;YAAE,OAAO,KAAK,CAAC;QAC3B,OAAO,WAAW,CAAC,KAAK,CAAC,qBAAqB,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC;IAChE,CAAC;IAEe,KAAK,CAAC,KAAK,CAAC,QAAyB;QACpD,IAAI,CAAC,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,CAAC,GAAG;YAAE,OAAO;QAEvC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI;YAAE,OAAO;QAC/B,IAAI,IAAI,CAAC,OAAO,CAAC,QAAQ;YAAE,OAAO,CAAC,mIAAmI;QAEtK,IAAI,CAAC;YACJ,MAAM,YAAY,GAAmB,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;YAEpE,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,EAAE,YAAY,CAAC,KAAK,CAAC,CAAC;YAC9D,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,EAAE,YAAY,CAAC,MAAM,CAAC,CAAC;QACjE,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC,CAAC,gBAAgB,CAAC,CAAC;IACnC,CAAC;CACD"}
@@ -1,9 +1,9 @@
1
- import { TKeyObject } from 'tscommons-es-core';
2
- import { DatabaseService } from '../services/database.service';
3
- import { IParserConfig } from '../interfaces/iparser-config';
4
- import { IRequestOutcome } from '../interfaces/irequest-outcome';
5
- import { IOllamaConfig, OllamaParser } from './ollama.parser';
6
- import { IDataConfig } from './data.parser';
1
+ import { TPropertyObject } from 'tscommons-esm-core';
2
+ import { DatabaseService } from '../services/database.service.mjs';
3
+ import { IParserConfig } from '../interfaces/iparser-config.mjs';
4
+ import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
5
+ import { IOllamaConfig, OllamaParser } from './ollama.parser.mjs';
6
+ import { IDataConfig } from './data.parser.mjs';
7
7
  export interface ILlamaGuardConfig extends IDataConfig {
8
8
  model: string;
9
9
  permittedDomains: string[];
@@ -13,6 +13,6 @@ export declare class LlamaGuardParser extends OllamaParser<IOllamaConfig> {
13
13
  private url?;
14
14
  private model;
15
15
  private permittedDomains;
16
- constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TKeyObject<IParserConfig>);
16
+ constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>);
17
17
  protected parseText(database: DatabaseService, text: string): Promise<void>;
18
18
  }
@@ -0,0 +1,56 @@
1
+ import { commonsTypeHasPropertyString, commonsTypeHasPropertyStringArray, commonsTypeHasPropertyT } from 'tscommons-esm-core';
2
+ import { commonsLogDebug, commonsLogError } from 'nodecommons-esm-log';
3
+ import { ELlamaGuardClassification, RigLlamaGuardService } from '../services/rig-llama-guard.service.mjs';
4
+ import { OllamaParser } from './ollama.parser.mjs';
5
+ import { isIDataConfig } from './data.parser.mjs';
6
+ export function isILlamaGuardConfig(test) {
7
+ if (!isIDataConfig(test))
8
+ return false;
9
+ if (!commonsTypeHasPropertyString(test, 'model'))
10
+ return false;
11
+ if (!commonsTypeHasPropertyStringArray(test, 'permittedDomains'))
12
+ return false;
13
+ return true;
14
+ }
15
+ export class LlamaGuardParser extends OllamaParser {
16
+ url;
17
+ model;
18
+ permittedDomains = [];
19
+ constructor(url, outcome, config) {
20
+ super(outcome, config, 'llamaguard');
21
+ this.url = url;
22
+ if (!config)
23
+ return;
24
+ if (!commonsTypeHasPropertyT(config, 'llamaguard', isILlamaGuardConfig)) {
25
+ commonsLogError('Invalid config for LlamaGuardParser');
26
+ }
27
+ this.model = config['llamaguard'].model;
28
+ this.permittedDomains = config['llamaguard'].permittedDomains;
29
+ }
30
+ async parseText(database, text) {
31
+ if (!this.url)
32
+ return;
33
+ if (!this.ollamaUrl || !this.ollamaKey || !this.model)
34
+ return;
35
+ const hostname = new URL(this.url).hostname;
36
+ if (!this.permittedDomains.includes(hostname))
37
+ return;
38
+ if (text.length === 0) {
39
+ await database.unsetData(this.url, 'llamaguard');
40
+ return;
41
+ }
42
+ const rigService = new RigLlamaGuardService(this.ollamaUrl, this.ollamaKey, this.model);
43
+ await this.debugGetLength();
44
+ // run this outside of await so that it doesn't block other things
45
+ void (async () => {
46
+ if (!this.url)
47
+ return;
48
+ const result = await rigService.generate(text, this.ollamaKeepAlive);
49
+ if (result === undefined)
50
+ return;
51
+ commonsLogDebug(`LlamaGuard result is: ${result.classification.toString()}${result.classification === ELlamaGuardClassification.UNSAFE ? ` (${result.categories.join(',')})` : ''}`);
52
+ await database.setData(this.url, 'llamaguard', result);
53
+ })();
54
+ }
55
+ }
56
+ //# sourceMappingURL=llama-guard.parser.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"llama-guard.parser.mjs","sourceRoot":"","sources":["../../src/parsers/llama-guard.parser.mts"],"names":[],"mappings":"AAAA,OAAO,EAAmB,4BAA4B,EAAE,iCAAiC,EAAE,uBAAuB,EAAE,MAAM,oBAAoB,CAAC;AAE/I,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAGvE,OAAO,EAAE,yBAAyB,EAAE,oBAAoB,EAAqB,MAAM,yCAAyC,CAAC;AAK7H,OAAO,EAAiB,YAAY,EAAE,MAAM,qBAAqB,CAAC;AAClE,OAAO,EAAe,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAM/D,MAAM,UAAU,mBAAmB,CAAC,IAAa;IAChD,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IAEvC,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,OAAO,CAAC;QAAE,OAAO,KAAK,CAAC;IAC/D,IAAI,CAAC,iCAAiC,CAAC,IAAI,EAAE,kBAAkB,CAAC;QAAE,OAAO,KAAK,CAAC;IAE/E,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAO,gBAAiB,SAAQ,YAA2B;IAKtD;IAJF,KAAK,CAAmB;IACxB,gBAAgB,GAAa,EAAE,CAAC;IAExC,YACU,GAAY,EACpB,OAAyB,EACzB,MAAuC;QAExC,KAAK,CACH,OAAO,EACP,MAAM,EACN,YAAY,CACb,CAAC;QARO,QAAG,GAAH,GAAG,CAAS;QAUrB,IAAI,CAAC,MAAM;YAAE,OAAO;QAEpB,IAAI,CAAC,uBAAuB,CAAoB,MAAM,EAAE,YAAY,EAAE,mBAAmB,CAAC,EAAE,CAAC;YAC5F,eAAe,CAAC,qCAAqC,CAAC,CAAC;QACxD,CAAC;QACD,IAAI,CAAC,KAAK,GAAI,MAAM,CAAC,YAAY,CAAuB,CAAC,KAAK,CAAC;QAC/D,IAAI,CAAC,gBAAgB,GAAI,MAAM,CAAC,YAAY,CAAuB,CAAC,gBAAgB,CAAC;IACtF,CAAC;IAES,KAAK,CAAC,SAAS,CAAC,QAAyB,EAAE,IAAY;QAChE,IAAI,CAAC,IAAI,CAAC,GAAG;YAAE,OAAO;QACtB,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,KAAK;YAAE,OAAO;QAE9D,MAAM,QAAQ,GAAW,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC;QACpD,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,QAAQ,CAAC;YAAE,OAAO;QAEtD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACvB,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,YAAY,CAAC,CAAC;YACjD,OAAO;QACR,CAAC;QAED,MAAM,UAAU,GAAyB,IAAI,oBAAoB,CAC/D,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,KAAK,CACX,CAAC;QAEF,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;QAE5B,kEAAkE;QAClE,KAAK,CAAC,KAAK,IAAmB,EAAE;YAC/B,IAAI,CAAC,IAAI,CAAC,GAAG;gBAAE,OAAO;YAEtB,MAAM,MAAM,GAAgC,MAAM,UAAU,CAAC,QAAQ,CACnE,IAAI,EACJ,IAAI,CAAC,eAAe,CACrB,CAAC;YACF,IAAI,MAAM,KAAK,SAAS;gBAAE,OAAO;YAEjC,eAAe,CAAC,yBAAyB,MAAM,CAAC,cAAc,CAAC,QAAQ,EAAE,GAAG,MAAM,CAAC,cAAc,KAAK,yBAAyB,CAAC,MAAM,CAAC,CAAC,CAAC,KAAK,MAAM,CAAC,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;YAErL,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,YAAY,EAAE,MAAM,CAAC,CAAC;QACxD,CAAC,CAAC,EAAE,CAAC;IACN,CAAC;CACD"}
@@ -1,9 +1,9 @@
1
- import { TKeyObject } from 'tscommons-es-core';
2
- import { DatabaseService } from '../services/database.service';
3
- import { IParserConfig } from '../interfaces/iparser-config';
4
- import { IRequestOutcome } from '../interfaces/irequest-outcome';
5
- import { IOllamaConfig, OllamaParser } from './ollama.parser';
6
- import { IDataConfig } from './data.parser';
1
+ import { TPropertyObject } from 'tscommons-esm-core';
2
+ import { DatabaseService } from '../services/database.service.mjs';
3
+ import { IParserConfig } from '../interfaces/iparser-config.mjs';
4
+ import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
5
+ import { IOllamaConfig, OllamaParser } from './ollama.parser.mjs';
6
+ import { IDataConfig } from './data.parser.mjs';
7
7
  export interface IOffenceConfig extends IDataConfig {
8
8
  model: string;
9
9
  permittedDomains: string[];
@@ -13,6 +13,6 @@ export declare class OffenceParser extends OllamaParser<IOllamaConfig> {
13
13
  private url?;
14
14
  private model;
15
15
  private permittedDomains;
16
- constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TKeyObject<IParserConfig>);
16
+ constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>);
17
17
  protected parseText(database: DatabaseService, text: string): Promise<void>;
18
18
  }
@@ -0,0 +1,60 @@
1
+ import { commonsTypeHasPropertyString, commonsTypeHasPropertyStringArray, commonsTypeHasPropertyT } from 'tscommons-esm-core';
2
+ import { commonsLogDebug, commonsLogError } from 'nodecommons-esm-log';
3
+ import { RigSubjectivityScaleService } from '../services/rig-subjectivity-scale.service.mjs';
4
+ import { OllamaParser } from './ollama.parser.mjs';
5
+ import { isIDataConfig } from './data.parser.mjs';
6
+ export function isIOffenceConfig(test) {
7
+ if (!isIDataConfig(test))
8
+ return false;
9
+ if (!commonsTypeHasPropertyString(test, 'model'))
10
+ return false;
11
+ if (!commonsTypeHasPropertyStringArray(test, 'permittedDomains'))
12
+ return false;
13
+ return true;
14
+ }
15
+ export class OffenceParser extends OllamaParser {
16
+ url;
17
+ model;
18
+ permittedDomains = [];
19
+ constructor(url, outcome, config) {
20
+ super(outcome, config, 'offence');
21
+ this.url = url;
22
+ if (!config)
23
+ return;
24
+ if (!commonsTypeHasPropertyT(config, 'offence', isIOffenceConfig)) {
25
+ commonsLogError('Invalid config for OffenceParser');
26
+ }
27
+ this.model = config['offence'].model;
28
+ this.permittedDomains = config['offence'].permittedDomains;
29
+ }
30
+ async parseText(database, text) {
31
+ if (!this.url)
32
+ return;
33
+ if (!this.ollamaUrl || !this.ollamaKey || !this.model)
34
+ return;
35
+ const hostname = new URL(this.url).hostname;
36
+ if (!this.permittedDomains.includes(hostname))
37
+ return;
38
+ if (text.length === 0) {
39
+ await database.unsetData(this.url, 'offence');
40
+ return;
41
+ }
42
+ const rigService = new RigSubjectivityScaleService(this.ollamaUrl, this.ollamaKey, this.model);
43
+ await this.debugGetLength();
44
+ // run this outside of await so that it doesn't block other things
45
+ void (async () => {
46
+ if (!this.url)
47
+ return;
48
+ const score = await rigService.generateOffenceScore(text, this.ollamaKeepAlive);
49
+ if (score === undefined)
50
+ return;
51
+ if (score.score < 1)
52
+ delete score.reason;
53
+ commonsLogDebug(`Offence score is: ${score.score.toString(10)}`);
54
+ if (score.reason)
55
+ commonsLogDebug(`Reason for score is: ${score.reason}`);
56
+ await database.setData(this.url, 'offence', score);
57
+ })();
58
+ }
59
+ }
60
+ //# sourceMappingURL=offence.parser.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"offence.parser.mjs","sourceRoot":"","sources":["../../src/parsers/offence.parser.mts"],"names":[],"mappings":"AAAA,OAAO,EAAmB,4BAA4B,EAAE,iCAAiC,EAAE,uBAAuB,EAAE,MAAM,oBAAoB,CAAC;AAE/I,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAGvE,OAAO,EAAE,2BAA2B,EAAuB,MAAM,gDAAgD,CAAC;AAKlH,OAAO,EAAiB,YAAY,EAAE,MAAM,qBAAqB,CAAC;AAClE,OAAO,EAAe,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAM/D,MAAM,UAAU,gBAAgB,CAAC,IAAa;IAC7C,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IAEvC,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,OAAO,CAAC;QAAE,OAAO,KAAK,CAAC;IAC/D,IAAI,CAAC,iCAAiC,CAAC,IAAI,EAAE,kBAAkB,CAAC;QAAE,OAAO,KAAK,CAAC;IAE/E,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAO,aAAc,SAAQ,YAA2B;IAKnD;IAJF,KAAK,CAAmB;IACxB,gBAAgB,GAAa,EAAE,CAAC;IAExC,YACU,GAAY,EACpB,OAAyB,EACzB,MAAuC;QAExC,KAAK,CACH,OAAO,EACP,MAAM,EACN,SAAS,CACV,CAAC;QARO,QAAG,GAAH,GAAG,CAAS;QAUrB,IAAI,CAAC,MAAM;YAAE,OAAO;QAEpB,IAAI,CAAC,uBAAuB,CAAiB,MAAM,EAAE,SAAS,EAAE,gBAAgB,CAAC,EAAE,CAAC;YACnF,eAAe,CAAC,kCAAkC,CAAC,CAAC;QACrD,CAAC;QACD,IAAI,CAAC,KAAK,GAAI,MAAM,CAAC,SAAS,CAAoB,CAAC,KAAK,CAAC;QACzD,IAAI,CAAC,gBAAgB,GAAI,MAAM,CAAC,SAAS,CAAoB,CAAC,gBAAgB,CAAC;IAChF,CAAC;IAES,KAAK,CAAC,SAAS,CAAC,QAAyB,EAAE,IAAY;QAChE,IAAI,CAAC,IAAI,CAAC,GAAG;YAAE,OAAO;QACtB,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,KAAK;YAAE,OAAO;QAE9D,MAAM,QAAQ,GAAW,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC;QACpD,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,QAAQ,CAAC;YAAE,OAAO;QAEtD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACvB,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,SAAS,CAAC,CAAC;YAC9C,OAAO;QACR,CAAC;QAED,MAAM,UAAU,GAAgC,IAAI,2BAA2B,CAC7E,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,KAAK,CACX,CAAC;QAEF,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;QAE5B,kEAAkE;QAClE,KAAK,CAAC,KAAK,IAAmB,EAAE;YAC/B,IAAI,CAAC,IAAI,CAAC,GAAG;gBAAE,OAAO;YAEtB,MAAM,KAAK,GAAkC,MAAM,UAAU,CAAC,oBAAoB,CAChF,IAAI,EACJ,IAAI,CAAC,eAAe,CACrB,CAAC;YACF,IAAI,KAAK,KAAK,SAAS;gBAAE,OAAO;YAEhC,IAAI,KAAK,CAAC,KAAK,GAAG,CAAC;gBAAE,OAAO,KAAK,CAAC,MAAM,CAAC;YAEzC,eAAe,CAAC,qBAAqB,KAAK,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;YACjE,IAAI,KAAK,CAAC,MAAM;gBAAE,eAAe,CAAC,wBAAwB,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC;YAE1E,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,SAAS,EAAE,KAAK,CAAC,CAAC;QACpD,CAAC,CAAC,EAAE,CAAC;IACN,CAAC;CACD"}
@@ -1,8 +1,8 @@
1
- import { CommonsFixedDuration, TKeyObject } from 'tscommons-es-core';
2
- import { IParserConfig } from '../interfaces/iparser-config';
3
- import { IRequestOutcome } from '../interfaces/irequest-outcome';
4
- import { TextParser } from './text.parser';
5
- import { IDataConfig } from './data.parser';
1
+ import { CommonsFixedDuration, TPropertyObject } from 'tscommons-esm-core';
2
+ import { IParserConfig } from '../interfaces/iparser-config.mjs';
3
+ import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
4
+ import { TextParser } from './text.parser.mjs';
5
+ import { IDataConfig } from './data.parser.mjs';
6
6
  export interface IOllamaConfig extends IDataConfig {
7
7
  url: string;
8
8
  key: string;
@@ -13,6 +13,6 @@ export declare abstract class OllamaParser<T extends IOllamaConfig> extends Text
13
13
  protected readonly ollamaUrl: string | undefined;
14
14
  protected readonly ollamaKey: string | undefined;
15
15
  protected readonly ollamaKeepAlive: CommonsFixedDuration | undefined;
16
- constructor(outcome?: IRequestOutcome, config?: TKeyObject<IParserConfig>, configKey?: string);
16
+ constructor(outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>, configKey?: string);
17
17
  debugGetLength(): Promise<void>;
18
18
  }
@@ -0,0 +1,43 @@
1
+ import { CommonsFixedDuration, commonsTypeHasPropertyNumberOrUndefined, commonsTypeHasPropertyString, commonsTypeHasPropertyT } from 'tscommons-esm-core';
2
+ import { commonsLogDebug, commonsLogError } from 'nodecommons-esm-log';
3
+ import { RigQueueLengthService } from '../services/rig-queue-length.service.mjs';
4
+ import { TextParser } from './text.parser.mjs';
5
+ import { isIDataConfig } from './data.parser.mjs';
6
+ export function isIOllamaConfig(test) {
7
+ if (!isIDataConfig(test))
8
+ return false;
9
+ if (!commonsTypeHasPropertyString(test, 'url'))
10
+ return false;
11
+ if (!commonsTypeHasPropertyString(test, 'key'))
12
+ return false;
13
+ if (!commonsTypeHasPropertyNumberOrUndefined(test, 'keepAlive'))
14
+ return false;
15
+ return true;
16
+ }
17
+ export class OllamaParser extends TextParser {
18
+ ollamaUrl;
19
+ ollamaKey;
20
+ ollamaKeepAlive;
21
+ constructor(outcome, config, configKey) {
22
+ super(outcome, config, configKey);
23
+ if (!config)
24
+ return;
25
+ if (!commonsTypeHasPropertyT(config, 'ollama', isIOllamaConfig)) {
26
+ commonsLogError('Invalid config for OllamaParser');
27
+ }
28
+ const ollamaConfig = config['ollama'];
29
+ this.ollamaUrl = ollamaConfig.url;
30
+ this.ollamaKey = ollamaConfig.key;
31
+ if (ollamaConfig.keepAlive) {
32
+ this.ollamaKeepAlive = CommonsFixedDuration.fromMillis(ollamaConfig.keepAlive);
33
+ }
34
+ }
35
+ async debugGetLength() {
36
+ if (!this.ollamaUrl || !this.ollamaKey)
37
+ return;
38
+ const rigService = new RigQueueLengthService(this.ollamaUrl, this.ollamaKey);
39
+ const length = await rigService.getLength();
40
+ commonsLogDebug(`Rig queue length is: ${length.active} active, ${length.queue} queued: ${length.line} total line length.`);
41
+ }
42
+ }
43
+ //# sourceMappingURL=ollama.parser.mjs.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"ollama.parser.mjs","sourceRoot":"","sources":["../../src/parsers/ollama.parser.mts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,uCAAuC,EAAE,4BAA4B,EAAE,uBAAuB,EAAmB,MAAM,oBAAoB,CAAC;AAE3K,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAEvE,OAAO,EAAE,qBAAqB,EAAE,MAAM,0CAA0C,CAAC;AAOjF,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAC/C,OAAO,EAAe,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAO/D,MAAM,UAAU,eAAe,CAAC,IAAa;IAC5C,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IAEvC,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,KAAK,CAAC;QAAE,OAAO,KAAK,CAAC;IAC7D,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,KAAK,CAAC;QAAE,OAAO,KAAK,CAAC;IAC7D,IAAI,CAAC,uCAAuC,CAAC,IAAI,EAAE,WAAW,CAAC;QAAE,OAAO,KAAK,CAAC;IAE9E,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAgB,YAAsC,SAAQ,UAAa;IAC7D,SAAS,CAAmB;IAC5B,SAAS,CAAmB;IAC5B,eAAe,CAAiC;IAEnE,YACE,OAAyB,EACzB,MAAuC,EACvC,SAAkB;QAEnB,KAAK,CACH,OAAO,EACP,MAAM,EACN,SAAS,CACV,CAAC;QAEF,IAAI,CAAC,MAAM;YAAE,OAAO;QAEpB,IAAI,CAAC,uBAAuB,CAAgB,MAAM,EAAE,QAAQ,EAAE,eAAe,CAAC,EAAE,CAAC;YAChF,eAAe,CAAC,iCAAiC,CAAC,CAAC;QACpD,CAAC;QACD,MAAM,YAAY,GAAkB,MAAM,CAAC,QAAQ,CAAkB,CAAC;QAEtE,IAAI,CAAC,SAAS,GAAG,YAAY,CAAC,GAAG,CAAC;QAClC,IAAI,CAAC,SAAS,GAAG,YAAY,CAAC,GAAG,CAAC;QAElC,IAAI,YAAY,CAAC,SAAS,EAAE,CAAC;YAC5B,IAAI,CAAC,eAAe,GAAG,oBAAoB,CAAC,UAAU,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC;QAChF,CAAC;IACF,CAAC;IAEM,KAAK,CAAC,cAAc;QAC1B,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,SAAS;YAAE,OAAO;QAE/C,MAAM,UAAU,GAA0B,IAAI,qBAAqB,CACjE,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,SAAS,CACf,CAAC;QAEF,MAAM,MAAM,GAAiB,MAAM,UAAU,CAAC,SAAS,EAAE,CAAC;QAC1D,eAAe,CAAC,wBAAwB,MAAM,CAAC,MAAM,YAAY,MAAM,CAAC,KAAK,YAAY,MAAM,CAAC,IAAI,qBAAqB,CAAC,CAAC;IAC5H,CAAC;CACD"}
@@ -1,13 +1,13 @@
1
- /// <reference types="cheerio" />
2
- import { DatabaseService } from '../services/database.service';
3
- import { HtmlParser } from './html.parser';
4
- import { IDataConfig } from './data.parser';
1
+ import * as cheerio from 'cheerio';
2
+ import { DatabaseService } from '../services/database.service.mjs';
3
+ import { HtmlParser } from './html.parser.mjs';
4
+ import { IDataConfig } from './data.parser.mjs';
5
5
  export interface IParagraphsConfig extends IDataConfig {
6
6
  paragraphWordsThreshold?: number;
7
7
  }
8
8
  export declare function isIParagraphsConfig(test: unknown): test is IParagraphsConfig;
9
9
  export declare abstract class ParagraphsParser<T extends IParagraphsConfig> extends HtmlParser<T> {
10
- static parseParagraphs(dom: cheerio.Root, paragraphsConfig: IParagraphsConfig): string[];
10
+ static parseParagraphs(dom: cheerio.CheerioAPI, paragraphsConfig: IParagraphsConfig): string[];
11
11
  protected abstract parseParagraphs(database: DatabaseService, paragraphs: string[]): Promise<void>;
12
12
  parse(database: DatabaseService): Promise<void>;
13
13
  }