hydra-crawler 2.8.3 → 3.0.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (468) hide show
  1. package/dist/apis/{autocomplete.api.d.ts → autocomplete.api.d.mts} +4 -4
  2. package/dist/apis/autocomplete.api.mjs +14 -0
  3. package/dist/apis/autocomplete.api.mjs.map +1 -0
  4. package/dist/apis/{bugs.api.d.ts → bugs.api.d.mts} +4 -4
  5. package/dist/apis/bugs.api.mjs +18 -0
  6. package/dist/apis/bugs.api.mjs.map +1 -0
  7. package/dist/apis/{crawl.api.d.ts → crawl.api.d.mts} +4 -4
  8. package/dist/apis/crawl.api.mjs +17 -0
  9. package/dist/apis/crawl.api.mjs.map +1 -0
  10. package/dist/apis/{domains.api.d.ts → domains.api.d.mts} +4 -4
  11. package/dist/apis/{domains.api.js → domains.api.mjs} +10 -18
  12. package/dist/apis/domains.api.mjs.map +1 -0
  13. package/dist/apis/{images.api.d.ts → images.api.d.mts} +4 -4
  14. package/dist/apis/images.api.mjs +19 -0
  15. package/dist/apis/images.api.mjs.map +1 -0
  16. package/dist/apis/{statistics.api.d.ts → statistics.api.d.mts} +4 -4
  17. package/dist/apis/statistics.api.mjs +33 -0
  18. package/dist/apis/statistics.api.mjs.map +1 -0
  19. package/dist/apis/{test.api.d.ts → test.api.d.mts} +3 -3
  20. package/dist/apis/test.api.mjs +10 -0
  21. package/dist/apis/test.api.mjs.map +1 -0
  22. package/dist/apis/{urls.api.d.ts → urls.api.d.mts} +4 -4
  23. package/dist/apis/urls.api.mjs +21 -0
  24. package/dist/apis/urls.api.mjs.map +1 -0
  25. package/dist/apps/{cleanup.app.d.ts → cleanup.app.d.mts} +5 -5
  26. package/dist/apps/cleanup.app.mjs +129 -0
  27. package/dist/apps/cleanup.app.mjs.map +1 -0
  28. package/dist/apps/{cross-populate-export.app.d.ts → cross-populate-export.app.d.mts} +3 -3
  29. package/dist/apps/cross-populate-export.app.mjs +61 -0
  30. package/dist/apps/cross-populate-export.app.mjs.map +1 -0
  31. package/dist/apps/{cross-populate-import.app.d.ts → cross-populate-import.app.d.mts} +3 -3
  32. package/dist/apps/cross-populate-import.app.mjs +61 -0
  33. package/dist/apps/cross-populate-import.app.mjs.map +1 -0
  34. package/dist/apps/{denylist.app.d.ts → denylist.app.d.mts} +5 -5
  35. package/dist/apps/denylist.app.mjs +122 -0
  36. package/dist/apps/denylist.app.mjs.map +1 -0
  37. package/dist/apps/{expire.app.d.ts → expire.app.d.mts} +8 -7
  38. package/dist/apps/expire.app.mjs +51 -0
  39. package/dist/apps/expire.app.mjs.map +1 -0
  40. package/dist/apps/{export-domain-urls.d.ts → export-domain-urls.d.mts} +3 -3
  41. package/dist/apps/export-domain-urls.mjs +85 -0
  42. package/dist/apps/export-domain-urls.mjs.map +1 -0
  43. package/dist/apps/{extract-text.app.d.ts → extract-text.app.d.mts} +2 -2
  44. package/dist/apps/extract-text.app.mjs +43 -0
  45. package/dist/apps/extract-text.app.mjs.map +1 -0
  46. package/dist/apps/{hydra.app.d.ts → hydra.app.d.mts} +10 -10
  47. package/dist/apps/hydra.app.mjs +222 -0
  48. package/dist/apps/hydra.app.mjs.map +1 -0
  49. package/dist/apps/{import.app.d.ts → import.app.d.mts} +3 -3
  50. package/dist/apps/import.app.mjs +48 -0
  51. package/dist/apps/import.app.mjs.map +1 -0
  52. package/dist/apps/{internal-hydra-common.app.d.ts → internal-hydra-common.app.d.mts} +6 -6
  53. package/dist/apps/{internal-hydra-common.app.js → internal-hydra-common.app.mjs} +1 -1
  54. package/dist/apps/internal-hydra-common.app.mjs.map +1 -0
  55. package/dist/apps/{move-to-archive.app.d.ts → move-to-archive.app.d.mts} +2 -2
  56. package/dist/apps/move-to-archive.app.mjs +31 -0
  57. package/dist/apps/move-to-archive.app.mjs.map +1 -0
  58. package/dist/apps/{prune-archive.app.d.ts → prune-archive.app.d.mts} +2 -2
  59. package/dist/apps/prune-archive.app.mjs +40 -0
  60. package/dist/apps/prune-archive.app.mjs.map +1 -0
  61. package/dist/apps/{query.app.d.ts → query.app.d.mts} +6 -6
  62. package/dist/apps/query.app.mjs +69 -0
  63. package/dist/apps/query.app.mjs.map +1 -0
  64. package/dist/apps/{reattempt.app.d.ts → reattempt.app.d.mts} +6 -6
  65. package/dist/apps/reattempt.app.mjs +74 -0
  66. package/dist/apps/reattempt.app.mjs.map +1 -0
  67. package/dist/apps/{requeue-domain.app.d.ts → requeue-domain.app.d.mts} +3 -3
  68. package/dist/apps/requeue-domain.app.mjs +52 -0
  69. package/dist/apps/requeue-domain.app.mjs.map +1 -0
  70. package/dist/apps/{seed.app.d.ts → seed.app.d.mts} +5 -5
  71. package/dist/apps/seed.app.mjs +59 -0
  72. package/dist/apps/seed.app.mjs.map +1 -0
  73. package/dist/apps/{startup.app.d.ts → startup.app.d.mts} +3 -3
  74. package/dist/apps/startup.app.mjs +55 -0
  75. package/dist/apps/startup.app.mjs.map +1 -0
  76. package/dist/apps/{unarchive-urls.app.d.ts → unarchive-urls.app.d.mts} +5 -5
  77. package/dist/apps/unarchive-urls.app.mjs +71 -0
  78. package/dist/apps/unarchive-urls.app.mjs.map +1 -0
  79. package/dist/classes/{cleaner.d.ts → cleaner.d.mts} +2 -2
  80. package/dist/classes/cleaner.mjs +257 -0
  81. package/dist/classes/cleaner.mjs.map +1 -0
  82. package/dist/classes/{crawler.d.ts → crawler.d.mts} +10 -12
  83. package/dist/classes/crawler.mjs +418 -0
  84. package/dist/classes/crawler.mjs.map +1 -0
  85. package/dist/classes/{dns.js → dns.mjs} +4 -4
  86. package/dist/classes/dns.mjs.map +1 -0
  87. package/dist/classes/{expirer.d.ts → expirer.d.mts} +2 -2
  88. package/dist/classes/expirer.mjs +117 -0
  89. package/dist/classes/expirer.mjs.map +1 -0
  90. package/dist/classes/{expiry.d.ts → expiry.d.mts} +1 -1
  91. package/dist/classes/{expiry.js → expiry.mjs} +7 -9
  92. package/dist/classes/expiry.mjs.map +1 -0
  93. package/dist/classes/{lists.d.ts → lists.d.mts} +2 -2
  94. package/dist/classes/{lists.js → lists.mjs} +4 -4
  95. package/dist/classes/lists.mjs.map +1 -0
  96. package/dist/classes/{robot.d.ts → robot.d.mts} +3 -3
  97. package/dist/classes/robot.mjs +74 -0
  98. package/dist/classes/robot.mjs.map +1 -0
  99. package/dist/classes/{tracker.d.ts → tracker.d.mts} +3 -3
  100. package/dist/classes/tracker.mjs +101 -0
  101. package/dist/classes/tracker.mjs.map +1 -0
  102. package/dist/{cli.js → cli.mjs} +46 -58
  103. package/dist/cli.mjs.map +1 -0
  104. package/dist/enums/{eavailable-strategy.js → eavailable-strategy.mjs} +1 -1
  105. package/dist/enums/eavailable-strategy.mjs.map +1 -0
  106. package/dist/enums/{elist.js → elist.mjs} +1 -1
  107. package/dist/enums/elist.mjs.map +1 -0
  108. package/dist/enums/{eserver.js → eserver.mjs} +1 -1
  109. package/dist/enums/eserver.mjs.map +1 -0
  110. package/dist/enums/{ex-powered-by.js → ex-powered-by.mjs} +1 -1
  111. package/dist/enums/ex-powered-by.mjs.map +1 -0
  112. package/dist/helpers/{matcher.d.ts → matcher.d.mts} +1 -1
  113. package/dist/helpers/{matcher.js → matcher.mjs} +1 -1
  114. package/dist/helpers/matcher.mjs.map +1 -0
  115. package/dist/helpers/{random.d.ts → random.d.mts} +1 -1
  116. package/dist/helpers/{random.js → random.mjs} +1 -1
  117. package/dist/helpers/random.mjs.map +1 -0
  118. package/dist/helpers/{utf-decoder.d.ts → utf-decoder.d.mts} +0 -1
  119. package/dist/helpers/{utf-decoder.js → utf-decoder.mjs} +3 -3
  120. package/dist/helpers/utf-decoder.mjs.map +1 -0
  121. package/dist/interfaces/{iexpiry.d.ts → iexpiry.d.mts} +1 -1
  122. package/dist/interfaces/{iexpiry.js → iexpiry.mjs} +3 -3
  123. package/dist/interfaces/iexpiry.mjs.map +1 -0
  124. package/dist/interfaces/{imatch.js → imatch.mjs} +2 -2
  125. package/dist/interfaces/imatch.mjs.map +1 -0
  126. package/dist/interfaces/{iparser-config.js → iparser-config.mjs} +2 -2
  127. package/dist/interfaces/iparser-config.mjs.map +1 -0
  128. package/dist/interfaces/{iparser.d.ts → iparser.d.mts} +1 -1
  129. package/dist/interfaces/iparser.mjs +2 -0
  130. package/dist/interfaces/iparser.mjs.map +1 -0
  131. package/dist/interfaces/{irequest-outcome.d.ts → irequest-outcome.d.mts} +0 -2
  132. package/dist/interfaces/irequest-outcome.mjs +2 -0
  133. package/dist/interfaces/irequest-outcome.mjs.map +1 -0
  134. package/dist/interfaces/iserver.mjs +2 -0
  135. package/dist/interfaces/iserver.mjs.map +1 -0
  136. package/dist/parsers/accessibility-metrics.parser.d.mts +11 -0
  137. package/dist/parsers/accessibility-metrics.parser.mjs +30 -0
  138. package/dist/parsers/accessibility-metrics.parser.mjs.map +1 -0
  139. package/dist/parsers/asp-error.parser.d.mts +12 -0
  140. package/dist/parsers/asp-error.parser.mjs +38 -0
  141. package/dist/parsers/asp-error.parser.mjs.map +1 -0
  142. package/dist/parsers/{bad-words.parser.d.ts → bad-words.parser.d.mts} +6 -6
  143. package/dist/parsers/bad-words.parser.mjs +17 -0
  144. package/dist/parsers/bad-words.parser.mjs.map +1 -0
  145. package/dist/parsers/complex-english.parser.d.mts +15 -0
  146. package/dist/parsers/complex-english.parser.mjs +52 -0
  147. package/dist/parsers/complex-english.parser.mjs.map +1 -0
  148. package/dist/parsers/data.parser.d.mts +13 -0
  149. package/dist/parsers/{data.parser.js → data.parser.mjs} +8 -7
  150. package/dist/parsers/data.parser.mjs.map +1 -0
  151. package/dist/parsers/{dictionary.parser.d.ts → dictionary.parser.d.mts} +6 -6
  152. package/dist/parsers/dictionary.parser.mjs +63 -0
  153. package/dist/parsers/dictionary.parser.mjs.map +1 -0
  154. package/dist/parsers/html.parser.d.mts +13 -0
  155. package/dist/parsers/{html.parser.js → html.parser.mjs} +4 -3
  156. package/dist/parsers/html.parser.mjs.map +1 -0
  157. package/dist/parsers/hyperlinks.parser.d.mts +20 -0
  158. package/dist/parsers/hyperlinks.parser.mjs +104 -0
  159. package/dist/parsers/hyperlinks.parser.mjs.map +1 -0
  160. package/dist/parsers/image-tags.parser.d.mts +19 -0
  161. package/dist/parsers/image-tags.parser.mjs +42 -0
  162. package/dist/parsers/image-tags.parser.mjs.map +1 -0
  163. package/dist/parsers/{interest.parser.d.ts → interest.parser.d.mts} +7 -7
  164. package/dist/parsers/interest.parser.mjs +60 -0
  165. package/dist/parsers/interest.parser.mjs.map +1 -0
  166. package/dist/parsers/jpeg.parser.d.mts +11 -0
  167. package/dist/parsers/jpeg.parser.mjs +29 -0
  168. package/dist/parsers/jpeg.parser.mjs.map +1 -0
  169. package/dist/parsers/{llama-guard.parser.d.ts → llama-guard.parser.d.mts} +7 -7
  170. package/dist/parsers/llama-guard.parser.mjs +56 -0
  171. package/dist/parsers/llama-guard.parser.mjs.map +1 -0
  172. package/dist/parsers/{offence.parser.d.ts → offence.parser.d.mts} +7 -7
  173. package/dist/parsers/offence.parser.mjs +60 -0
  174. package/dist/parsers/offence.parser.mjs.map +1 -0
  175. package/dist/parsers/{ollama.parser.d.ts → ollama.parser.d.mts} +6 -6
  176. package/dist/parsers/ollama.parser.mjs +43 -0
  177. package/dist/parsers/ollama.parser.mjs.map +1 -0
  178. package/dist/parsers/{paragraphs.parser.d.ts → paragraphs.parser.d.mts} +5 -5
  179. package/dist/parsers/paragraphs.parser.mjs +38 -0
  180. package/dist/parsers/paragraphs.parser.mjs.map +1 -0
  181. package/dist/parsers/{parser.d.ts → parser.d.mts} +6 -6
  182. package/dist/parsers/parser.mjs +45 -0
  183. package/dist/parsers/parser.mjs.map +1 -0
  184. package/dist/parsers/php-error.parser.d.mts +12 -0
  185. package/dist/parsers/php-error.parser.mjs +42 -0
  186. package/dist/parsers/php-error.parser.mjs.map +1 -0
  187. package/dist/parsers/{phrase.parser.d.ts → phrase.parser.d.mts} +3 -3
  188. package/dist/parsers/phrase.parser.mjs +15 -0
  189. package/dist/parsers/phrase.parser.mjs.map +1 -0
  190. package/dist/parsers/{regex.parser.d.ts → regex.parser.d.mts} +3 -3
  191. package/dist/parsers/regex.parser.mjs +29 -0
  192. package/dist/parsers/regex.parser.mjs.map +1 -0
  193. package/dist/parsers/server.parser.d.mts +11 -0
  194. package/dist/parsers/server.parser.mjs +57 -0
  195. package/dist/parsers/server.parser.mjs.map +1 -0
  196. package/dist/parsers/{spelling.parser.d.ts → spelling.parser.d.mts} +6 -6
  197. package/dist/parsers/spelling.parser.mjs +17 -0
  198. package/dist/parsers/spelling.parser.mjs.map +1 -0
  199. package/dist/parsers/string.parser.d.mts +8 -0
  200. package/dist/parsers/{string.parser.js → string.parser.mjs} +5 -4
  201. package/dist/parsers/string.parser.mjs.map +1 -0
  202. package/dist/parsers/{text.parser.d.ts → text.parser.d.mts} +3 -3
  203. package/dist/parsers/text.parser.mjs +30 -0
  204. package/dist/parsers/text.parser.mjs.map +1 -0
  205. package/dist/parsers/{words.parser.d.ts → words.parser.d.mts} +3 -3
  206. package/dist/parsers/words.parser.mjs +29 -0
  207. package/dist/parsers/words.parser.mjs.map +1 -0
  208. package/dist/queries/complex-english.query.d.mts +2 -0
  209. package/dist/queries/{complex-english.query.js → complex-english.query.mjs} +18 -27
  210. package/dist/queries/complex-english.query.mjs.map +1 -0
  211. package/dist/queries/flash-content.query.d.mts +2 -0
  212. package/dist/queries/flash-content.query.mjs +80 -0
  213. package/dist/queries/flash-content.query.mjs.map +1 -0
  214. package/dist/queries/linking-to-domains.query.d.mts +2 -0
  215. package/dist/queries/linking-to-domains.query.mjs +128 -0
  216. package/dist/queries/linking-to-domains.query.mjs.map +1 -0
  217. package/dist/queries/llamaguard-unsafe-content.query.d.mts +2 -0
  218. package/dist/queries/llamaguard-unsafe-content.query.mjs +90 -0
  219. package/dist/queries/llamaguard-unsafe-content.query.mjs.map +1 -0
  220. package/dist/queries/readability-score.query.d.mts +2 -0
  221. package/dist/queries/{readability-score.query.js → readability-score.query.mjs} +11 -20
  222. package/dist/queries/readability-score.query.mjs.map +1 -0
  223. package/dist/servers/{crawl.server.d.ts → crawl.server.d.mts} +10 -10
  224. package/dist/servers/crawl.server.mjs +192 -0
  225. package/dist/servers/crawl.server.mjs.map +1 -0
  226. package/dist/servers/{express.server.d.ts → express.server.d.mts} +2 -3
  227. package/dist/servers/express.server.mjs +13 -0
  228. package/dist/servers/express.server.mjs.map +1 -0
  229. package/dist/servers/{maintenance.server.d.ts → maintenance.server.d.mts} +5 -5
  230. package/dist/servers/maintenance.server.mjs +97 -0
  231. package/dist/servers/maintenance.server.mjs.map +1 -0
  232. package/dist/servers/{rest.server.d.ts → rest.server.d.mts} +4 -4
  233. package/dist/servers/rest.server.mjs +62 -0
  234. package/dist/servers/rest.server.mjs.map +1 -0
  235. package/dist/servers/{socket-io.server.d.ts → socket-io.server.d.mts} +4 -4
  236. package/dist/servers/socket-io.server.mjs +22 -0
  237. package/dist/servers/socket-io.server.mjs.map +1 -0
  238. package/dist/services/{database.service.d.ts → database.service.d.mts} +12 -12
  239. package/dist/services/database.service.mjs +645 -0
  240. package/dist/services/database.service.mjs.map +1 -0
  241. package/dist/services/{ollama-rest.service.d.ts → ollama-rest.service.d.mts} +2 -2
  242. package/dist/services/ollama-rest.service.mjs +27 -0
  243. package/dist/services/ollama-rest.service.mjs.map +1 -0
  244. package/dist/services/{rig-llama-guard.service.d.ts → rig-llama-guard.service.d.mts} +2 -2
  245. package/dist/services/{rig-llama-guard.service.js → rig-llama-guard.service.mjs} +33 -43
  246. package/dist/services/rig-llama-guard.service.mjs.map +1 -0
  247. package/dist/services/{rig-queue-length.service.d.ts → rig-queue-length.service.d.mts} +1 -1
  248. package/dist/services/rig-queue-length.service.mjs +12 -0
  249. package/dist/services/rig-queue-length.service.mjs.map +1 -0
  250. package/dist/services/{rig-subjectivity-scale.service.d.ts → rig-subjectivity-scale.service.d.mts} +2 -2
  251. package/dist/services/rig-subjectivity-scale.service.mjs +82 -0
  252. package/dist/services/rig-subjectivity-scale.service.mjs.map +1 -0
  253. package/dist/services/{rig-ticket-rest.service.d.ts → rig-ticket-rest.service.d.mts} +3 -3
  254. package/dist/services/rig-ticket-rest.service.mjs +15 -0
  255. package/dist/services/rig-ticket-rest.service.mjs.map +1 -0
  256. package/dist/services/{rig-ticketed-promise.service.d.ts → rig-ticketed-promise.service.d.mts} +4 -6
  257. package/dist/services/{rig-ticketed-promise.service.js → rig-ticketed-promise.service.mjs} +7 -6
  258. package/dist/services/rig-ticketed-promise.service.mjs.map +1 -0
  259. package/dist/types/{tcrawl-config.js → tcrawl-config.mjs} +2 -2
  260. package/dist/types/tcrawl-config.mjs.map +1 -0
  261. package/dist/types/{thydra-config.js → thydra-config.mjs} +2 -2
  262. package/dist/types/thydra-config.mjs.map +1 -0
  263. package/dist/types/tparser-ctor.d.mts +7 -0
  264. package/dist/types/tparser-ctor.mjs +2 -0
  265. package/dist/types/tparser-ctor.mjs.map +1 -0
  266. package/dist/types/tquery.d.mts +7 -0
  267. package/dist/types/tquery.mjs +2 -0
  268. package/dist/types/tquery.mjs.map +1 -0
  269. package/dist/types/tqueue-length.mjs +2 -0
  270. package/dist/types/tqueue-length.mjs.map +1 -0
  271. package/dist/types/{trobots-config.js → trobots-config.mjs} +2 -2
  272. package/dist/types/trobots-config.mjs.map +1 -0
  273. package/package.json +37 -43
  274. package/dist/apis/autocomplete.api.js +0 -22
  275. package/dist/apis/autocomplete.api.js.map +0 -1
  276. package/dist/apis/bugs.api.js +0 -26
  277. package/dist/apis/bugs.api.js.map +0 -1
  278. package/dist/apis/crawl.api.js +0 -25
  279. package/dist/apis/crawl.api.js.map +0 -1
  280. package/dist/apis/domains.api.js.map +0 -1
  281. package/dist/apis/images.api.js +0 -27
  282. package/dist/apis/images.api.js.map +0 -1
  283. package/dist/apis/statistics.api.js +0 -41
  284. package/dist/apis/statistics.api.js.map +0 -1
  285. package/dist/apis/test.api.js +0 -19
  286. package/dist/apis/test.api.js.map +0 -1
  287. package/dist/apis/urls.api.js +0 -29
  288. package/dist/apis/urls.api.js.map +0 -1
  289. package/dist/apps/cleanup.app.js +0 -141
  290. package/dist/apps/cleanup.app.js.map +0 -1
  291. package/dist/apps/cross-populate-export.app.js +0 -75
  292. package/dist/apps/cross-populate-export.app.js.map +0 -1
  293. package/dist/apps/cross-populate-import.app.js +0 -100
  294. package/dist/apps/cross-populate-import.app.js.map +0 -1
  295. package/dist/apps/denylist.app.js +0 -132
  296. package/dist/apps/denylist.app.js.map +0 -1
  297. package/dist/apps/expire.app.js +0 -63
  298. package/dist/apps/expire.app.js.map +0 -1
  299. package/dist/apps/export-domain-urls.js +0 -99
  300. package/dist/apps/export-domain-urls.js.map +0 -1
  301. package/dist/apps/extract-text.app.js +0 -55
  302. package/dist/apps/extract-text.app.js.map +0 -1
  303. package/dist/apps/hydra.app.js +0 -218
  304. package/dist/apps/hydra.app.js.map +0 -1
  305. package/dist/apps/import.app.js +0 -57
  306. package/dist/apps/import.app.js.map +0 -1
  307. package/dist/apps/internal-hydra-common.app.js.map +0 -1
  308. package/dist/apps/move-to-archive.app.js +0 -46
  309. package/dist/apps/move-to-archive.app.js.map +0 -1
  310. package/dist/apps/prune-archive.app.js +0 -55
  311. package/dist/apps/prune-archive.app.js.map +0 -1
  312. package/dist/apps/query.app.js +0 -80
  313. package/dist/apps/query.app.js.map +0 -1
  314. package/dist/apps/reattempt.app.js +0 -83
  315. package/dist/apps/reattempt.app.js.map +0 -1
  316. package/dist/apps/requeue-domain.app.js +0 -64
  317. package/dist/apps/requeue-domain.app.js.map +0 -1
  318. package/dist/apps/seed.app.js +0 -69
  319. package/dist/apps/seed.app.js.map +0 -1
  320. package/dist/apps/startup.app.js +0 -64
  321. package/dist/apps/startup.app.js.map +0 -1
  322. package/dist/apps/unarchive-urls.app.js +0 -83
  323. package/dist/apps/unarchive-urls.app.js.map +0 -1
  324. package/dist/classes/cleaner.js +0 -266
  325. package/dist/classes/cleaner.js.map +0 -1
  326. package/dist/classes/crawler.js +0 -422
  327. package/dist/classes/crawler.js.map +0 -1
  328. package/dist/classes/dns.js.map +0 -1
  329. package/dist/classes/expirer.js +0 -121
  330. package/dist/classes/expirer.js.map +0 -1
  331. package/dist/classes/expiry.js.map +0 -1
  332. package/dist/classes/lists.js.map +0 -1
  333. package/dist/classes/robot.js +0 -82
  334. package/dist/classes/robot.js.map +0 -1
  335. package/dist/classes/tracker.js +0 -120
  336. package/dist/classes/tracker.js.map +0 -1
  337. package/dist/cli.js.map +0 -1
  338. package/dist/enums/eavailable-strategy.js.map +0 -1
  339. package/dist/enums/elist.js.map +0 -1
  340. package/dist/enums/eserver.js.map +0 -1
  341. package/dist/enums/ex-powered-by.js.map +0 -1
  342. package/dist/helpers/matcher.js.map +0 -1
  343. package/dist/helpers/random.js.map +0 -1
  344. package/dist/helpers/utf-decoder.js.map +0 -1
  345. package/dist/interfaces/iexpiry.js.map +0 -1
  346. package/dist/interfaces/imatch.js.map +0 -1
  347. package/dist/interfaces/iparser-config.js.map +0 -1
  348. package/dist/interfaces/iparser.js +0 -2
  349. package/dist/interfaces/iparser.js.map +0 -1
  350. package/dist/interfaces/irequest-outcome.js +0 -2
  351. package/dist/interfaces/irequest-outcome.js.map +0 -1
  352. package/dist/interfaces/iserver.js +0 -2
  353. package/dist/interfaces/iserver.js.map +0 -1
  354. package/dist/parsers/accessibility-metrics.parser.d.ts +0 -11
  355. package/dist/parsers/accessibility-metrics.parser.js +0 -40
  356. package/dist/parsers/accessibility-metrics.parser.js.map +0 -1
  357. package/dist/parsers/asp-error.parser.d.ts +0 -12
  358. package/dist/parsers/asp-error.parser.js +0 -48
  359. package/dist/parsers/asp-error.parser.js.map +0 -1
  360. package/dist/parsers/bad-words.parser.js +0 -27
  361. package/dist/parsers/bad-words.parser.js.map +0 -1
  362. package/dist/parsers/complex-english.parser.d.ts +0 -15
  363. package/dist/parsers/complex-english.parser.js +0 -61
  364. package/dist/parsers/complex-english.parser.js.map +0 -1
  365. package/dist/parsers/data.parser.d.ts +0 -14
  366. package/dist/parsers/data.parser.js.map +0 -1
  367. package/dist/parsers/dictionary.parser.js +0 -73
  368. package/dist/parsers/dictionary.parser.js.map +0 -1
  369. package/dist/parsers/html.parser.d.ts +0 -13
  370. package/dist/parsers/html.parser.js.map +0 -1
  371. package/dist/parsers/hyperlinks.parser.d.ts +0 -20
  372. package/dist/parsers/hyperlinks.parser.js +0 -115
  373. package/dist/parsers/hyperlinks.parser.js.map +0 -1
  374. package/dist/parsers/image-tags.parser.d.ts +0 -19
  375. package/dist/parsers/image-tags.parser.js +0 -52
  376. package/dist/parsers/image-tags.parser.js.map +0 -1
  377. package/dist/parsers/interest.parser.js +0 -69
  378. package/dist/parsers/interest.parser.js.map +0 -1
  379. package/dist/parsers/jpeg.parser.d.ts +0 -11
  380. package/dist/parsers/jpeg.parser.js +0 -39
  381. package/dist/parsers/jpeg.parser.js.map +0 -1
  382. package/dist/parsers/llama-guard.parser.js +0 -65
  383. package/dist/parsers/llama-guard.parser.js.map +0 -1
  384. package/dist/parsers/offence.parser.js +0 -69
  385. package/dist/parsers/offence.parser.js.map +0 -1
  386. package/dist/parsers/ollama.parser.js +0 -51
  387. package/dist/parsers/ollama.parser.js.map +0 -1
  388. package/dist/parsers/paragraphs.parser.js +0 -49
  389. package/dist/parsers/paragraphs.parser.js.map +0 -1
  390. package/dist/parsers/parser.js +0 -57
  391. package/dist/parsers/parser.js.map +0 -1
  392. package/dist/parsers/php-error.parser.d.ts +0 -12
  393. package/dist/parsers/php-error.parser.js +0 -52
  394. package/dist/parsers/php-error.parser.js.map +0 -1
  395. package/dist/parsers/phrase.parser.js +0 -26
  396. package/dist/parsers/phrase.parser.js.map +0 -1
  397. package/dist/parsers/regex.parser.js +0 -43
  398. package/dist/parsers/regex.parser.js.map +0 -1
  399. package/dist/parsers/server.parser.d.ts +0 -11
  400. package/dist/parsers/server.parser.js +0 -67
  401. package/dist/parsers/server.parser.js.map +0 -1
  402. package/dist/parsers/spelling.parser.js +0 -27
  403. package/dist/parsers/spelling.parser.js.map +0 -1
  404. package/dist/parsers/string.parser.d.ts +0 -8
  405. package/dist/parsers/string.parser.js.map +0 -1
  406. package/dist/parsers/text.parser.js +0 -41
  407. package/dist/parsers/text.parser.js.map +0 -1
  408. package/dist/parsers/words.parser.js +0 -40
  409. package/dist/parsers/words.parser.js.map +0 -1
  410. package/dist/queries/complex-english.query.d.ts +0 -2
  411. package/dist/queries/complex-english.query.js.map +0 -1
  412. package/dist/queries/flash-content.query.d.ts +0 -2
  413. package/dist/queries/flash-content.query.js +0 -82
  414. package/dist/queries/flash-content.query.js.map +0 -1
  415. package/dist/queries/linking-to-domains.query.d.ts +0 -2
  416. package/dist/queries/linking-to-domains.query.js +0 -130
  417. package/dist/queries/linking-to-domains.query.js.map +0 -1
  418. package/dist/queries/llamaguard-unsafe-content.query.d.ts +0 -2
  419. package/dist/queries/llamaguard-unsafe-content.query.js +0 -92
  420. package/dist/queries/llamaguard-unsafe-content.query.js.map +0 -1
  421. package/dist/queries/readability-score.query.d.ts +0 -2
  422. package/dist/queries/readability-score.query.js.map +0 -1
  423. package/dist/servers/crawl.server.js +0 -198
  424. package/dist/servers/crawl.server.js.map +0 -1
  425. package/dist/servers/express.server.js +0 -13
  426. package/dist/servers/express.server.js.map +0 -1
  427. package/dist/servers/maintenance.server.js +0 -103
  428. package/dist/servers/maintenance.server.js.map +0 -1
  429. package/dist/servers/rest.server.js +0 -46
  430. package/dist/servers/rest.server.js.map +0 -1
  431. package/dist/servers/socket-io.server.js +0 -61
  432. package/dist/servers/socket-io.server.js.map +0 -1
  433. package/dist/services/database.service.js +0 -795
  434. package/dist/services/database.service.js.map +0 -1
  435. package/dist/services/ollama-rest.service.js +0 -46
  436. package/dist/services/ollama-rest.service.js.map +0 -1
  437. package/dist/services/rig-llama-guard.service.js.map +0 -1
  438. package/dist/services/rig-queue-length.service.js +0 -22
  439. package/dist/services/rig-queue-length.service.js.map +0 -1
  440. package/dist/services/rig-subjectivity-scale.service.js +0 -96
  441. package/dist/services/rig-subjectivity-scale.service.js.map +0 -1
  442. package/dist/services/rig-ticket-rest.service.js +0 -29
  443. package/dist/services/rig-ticket-rest.service.js.map +0 -1
  444. package/dist/services/rig-ticketed-promise.service.js.map +0 -1
  445. package/dist/types/tcrawl-config.js.map +0 -1
  446. package/dist/types/thydra-config.js.map +0 -1
  447. package/dist/types/tparser-ctor.d.ts +0 -7
  448. package/dist/types/tparser-ctor.js +0 -2
  449. package/dist/types/tparser-ctor.js.map +0 -1
  450. package/dist/types/tquery.d.ts +0 -7
  451. package/dist/types/tquery.js +0 -2
  452. package/dist/types/tquery.js.map +0 -1
  453. package/dist/types/tqueue-length.js +0 -2
  454. package/dist/types/tqueue-length.js.map +0 -1
  455. package/dist/types/trobots-config.js.map +0 -1
  456. /package/dist/classes/{dns.d.ts → dns.d.mts} +0 -0
  457. /package/dist/{cli.d.ts → cli.d.mts} +0 -0
  458. /package/dist/enums/{eavailable-strategy.d.ts → eavailable-strategy.d.mts} +0 -0
  459. /package/dist/enums/{elist.d.ts → elist.d.mts} +0 -0
  460. /package/dist/enums/{eserver.d.ts → eserver.d.mts} +0 -0
  461. /package/dist/enums/{ex-powered-by.d.ts → ex-powered-by.d.mts} +0 -0
  462. /package/dist/interfaces/{imatch.d.ts → imatch.d.mts} +0 -0
  463. /package/dist/interfaces/{iparser-config.d.ts → iparser-config.d.mts} +0 -0
  464. /package/dist/interfaces/{iserver.d.ts → iserver.d.mts} +0 -0
  465. /package/dist/types/{tcrawl-config.d.ts → tcrawl-config.d.mts} +0 -0
  466. /package/dist/types/{thydra-config.d.ts → thydra-config.d.mts} +0 -0
  467. /package/dist/types/{tqueue-length.d.ts → tqueue-length.d.mts} +0 -0
  468. /package/dist/types/{trobots-config.d.ts → trobots-config.d.mts} +0 -0
@@ -1,73 +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 { commonsTypeHasPropertyBooleanOrUndefined, commonsTypeHasPropertyString, commonsTypeHasPropertyT, commonsTypeIsStringArray } from 'tscommons-es-core';
11
- import { commonsOutputError } from 'nodecommons-es-cli';
12
- import { commonsFileCwdRelativeOrAbsolutePath, commonsFileReadJsonFile } from 'nodecommons-es-file';
13
- import { WordsParser, isIWordsConfig } from './words.parser';
14
- export function isIDictionaryConfig(test) {
15
- if (!isIWordsConfig(test))
16
- return false;
17
- if (!commonsTypeHasPropertyString(test, 'dictionary'))
18
- return false;
19
- if (!commonsTypeHasPropertyBooleanOrUndefined(test, 'caseSensitive'))
20
- return false;
21
- return true;
22
- }
23
- const CACHE = new Map();
24
- function loadDictionary(config) {
25
- const filename = commonsFileCwdRelativeOrAbsolutePath(`config/${config.dictionary}`);
26
- const json = commonsFileReadJsonFile(filename);
27
- if (json === undefined)
28
- throw new Error('Unable to read dictionary file');
29
- if (!commonsTypeIsStringArray(json))
30
- throw new Error('Dictionary file is not a JSON array');
31
- return json;
32
- }
33
- export class DictionaryParser extends WordsParser {
34
- constructor(outcome, config, configKey) {
35
- super(outcome, config, configKey);
36
- this.caseSensitive = false;
37
- this.dictionary = [];
38
- if (!config || !configKey)
39
- return;
40
- this.cacheId = configKey;
41
- if (!commonsTypeHasPropertyT(config, configKey, isIDictionaryConfig)) {
42
- commonsOutputError(`Invalid config for DictionaryParser (${configKey})`);
43
- }
44
- const dictionaryConfig = config[configKey];
45
- this.caseSensitive = dictionaryConfig.caseSensitive || false;
46
- if (!CACHE.has(this.cacheId)) {
47
- const loaded = loadDictionary(dictionaryConfig);
48
- CACHE.set(this.cacheId, loaded);
49
- }
50
- }
51
- getDictionary() {
52
- if (!this.cacheId)
53
- return [];
54
- return CACHE.get(this.cacheId) || [];
55
- }
56
- parseWords(database, words) {
57
- return __awaiter(this, void 0, void 0, function* () {
58
- this.dictionary = this.getDictionary()
59
- .map((word) => this.caseSensitive ? word : word.toLowerCase());
60
- const matches = [];
61
- const nonMatches = [];
62
- for (const word of words) {
63
- const compare = this.caseSensitive ? word : word.toLowerCase();
64
- if (this.dictionary.includes(compare))
65
- matches.push(word);
66
- else
67
- nonMatches.push(word);
68
- }
69
- yield this.parseMatches(database, matches, nonMatches);
70
- });
71
- }
72
- }
73
- //# sourceMappingURL=dictionary.parser.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"dictionary.parser.js","sourceRoot":"","sources":["../../src/parsers/dictionary.parser.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAE,wCAAwC,EAAE,4BAA4B,EAAE,uBAAuB,EAAE,wBAAwB,EAAc,MAAM,mBAAmB,CAAC;AAE1K,OAAO,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AACxD,OAAO,EAAE,oCAAoC,EAAE,uBAAuB,EAAE,MAAM,qBAAqB,CAAC;AAOpG,OAAO,EAAE,WAAW,EAAgB,cAAc,EAAE,MAAM,gBAAgB,CAAC;AAM3E,MAAM,UAAU,mBAAmB,CAAC,IAAa;IAChD,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IAExC,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,YAAY,CAAC;QAAE,OAAO,KAAK,CAAC;IACpE,IAAI,CAAC,wCAAwC,CAAC,IAAI,EAAE,eAAe,CAAC;QAAE,OAAO,KAAK,CAAC;IAEnF,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,KAAK,GAA0B,IAAI,GAAG,EAAoB,CAAC;AAEjE,SAAS,cAAc,CAAC,MAAyB;IAChD,MAAM,QAAQ,GAAW,oCAAoC,CAAC,UAAU,MAAM,CAAC,UAAU,EAAE,CAAC,CAAC;IAC7F,MAAM,IAAI,GAAsB,uBAAuB,CAAC,QAAQ,CAAC,CAAC;IAClE,IAAI,IAAI,KAAK,SAAS;QAAE,MAAM,IAAI,KAAK,CAAC,gCAAgC,CAAC,CAAC;IAE1E,IAAI,CAAC,wBAAwB,CAAC,IAAI,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,qCAAqC,CAAC,CAAC;IAE5F,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAgB,gBAA8C,SAAQ,WAAc;IAKzF,YACE,OAAyB,EACzB,MAAkC,EAClC,SAAkB;QAEnB,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,SAAS,CAAC,CAAC;QATzB,kBAAa,GAAY,KAAK,CAAC;QACjC,eAAU,GAAa,EAAE,CAAC;QAUjC,IAAI,CAAC,MAAM,IAAI,CAAC,SAAS;YAAE,OAAO;QAElC,IAAI,CAAC,OAAO,GAAG,SAAS,CAAC;QAEzB,IAAI,CAAC,uBAAuB,CAAoB,MAAM,EAAE,SAAS,EAAE,mBAAmB,CAAC,EAAE;YACxF,kBAAkB,CAAC,wCAAwC,SAAS,GAAG,CAAC,CAAC;SACzE;QACD,MAAM,gBAAgB,GAAsB,MAAM,CAAC,SAAS,CAAsB,CAAC;QAEnF,IAAI,CAAC,aAAa,GAAG,gBAAgB,CAAC,aAAa,IAAI,KAAK,CAAC;QAE7D,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE;YAC7B,MAAM,MAAM,GAAa,cAAc,CAAC,gBAAgB,CAAC,CAAC;YAC1D,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;SAChC;IACF,CAAC;IAES,aAAa;QACtB,IAAI,CAAC,IAAI,CAAC,OAAO;YAAE,OAAO,EAAE,CAAC;QAE7B,OAAO,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;IACtC,CAAC;IAIe,UAAU,CAAC,QAAyB,EAAE,KAAe;;YACpE,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,aAAa,EAAE;iBACnC,GAAG,CAAC,CAAC,IAAY,EAAU,EAAE,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,CAAC;YAEjF,MAAM,OAAO,GAAa,EAAE,CAAC;YAC7B,MAAM,UAAU,GAAa,EAAE,CAAC;YAEhC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE;gBACzB,MAAM,OAAO,GAAW,IAAI,CAAC,aAAa,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC;gBACvE,IAAI,IAAI,CAAC,UAAU,CAAC,QAAQ,CAAC,OAAO,CAAC;oBAAE,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;;oBACrD,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;aAC3B;YAED,MAAM,IAAI,CAAC,YAAY,CAAC,QAAQ,EAAE,OAAO,EAAE,UAAU,CAAC,CAAC;QACxD,CAAC;KAAA;CACD"}
@@ -1,13 +0,0 @@
1
- /// <reference types="cheerio" />
2
- import { TKeyObject } from 'tscommons-es-core';
3
- import { IRequestOutcome } from '../interfaces/irequest-outcome';
4
- import { IParserConfig } from '../interfaces/iparser-config';
5
- import { StringParser } from './string.parser';
6
- import { IDataConfig } from './data.parser';
7
- export declare abstract class HtmlParser<T extends IDataConfig> extends StringParser<T> {
8
- static collapseWhitespace(raw: string): string;
9
- static loadDom(html: string): cheerio.Root | undefined;
10
- protected dom: cheerio.Root | undefined;
11
- constructor(outcome?: IRequestOutcome, config?: TKeyObject<IParserConfig>, configKey?: string);
12
- supports(contentType: string, isAllow: boolean): boolean;
13
- }
@@ -1 +0,0 @@
1
- {"version":3,"file":"html.parser.js","sourceRoot":"","sources":["../../src/parsers/html.parser.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,OAAO,MAAM,SAAS,CAAC;AAOnC,OAAO,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAG/C,MAAM,OAAgB,UAAkC,SAAQ,YAAe;IACvE,MAAM,CAAC,kBAAkB,CAAC,GAAW;QAC3C,OAAO,GAAG,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;IACjC,CAAC;IAEM,MAAM,CAAC,OAAO,CAAC,IAAY;QACjC,IAAI;YACH,OAAO,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;SAC1B;QAAC,OAAO,EAAE,EAAE;YACZ,SAAS;YACT,OAAO,SAAS,CAAC;SACjB;IACF,CAAC;IAID,YACE,OAAyB,EACzB,MAAkC,EAClC,SAAkB;QAEnB,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,SAAS,CAAC,CAAC;QAElC,IAAI,IAAI,CAAC,UAAU,IAAI,IAAI,CAAC,UAAU,CAAC,MAAM,EAAE;YAC9C,MAAM,SAAS,GAAW,UAAU,CAAC,kBAAkB,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;YAEzE,IAAI,CAAC,GAAG,GAAG,UAAU,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC;SACzC;IACF,CAAC;IAEM,QAAQ,CAAC,WAAmB,EAAE,OAAgB;QACpD,IAAI,CAAC,OAAO;YAAE,OAAO,KAAK,CAAC;QAC3B,OAAO,4BAA4B,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;IACvD,CAAC;CACD"}
@@ -1,20 +0,0 @@
1
- /// <reference types="cheerio" />
2
- import { TKeyObject } from 'tscommons-es-core';
3
- import { DatabaseService } from '../services/database.service';
4
- import { IParserConfig } from '../interfaces/iparser-config';
5
- import { IRequestOutcome } from '../interfaces/irequest-outcome';
6
- import { HtmlParser } from './html.parser';
7
- import { IDataConfig } from './data.parser';
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.Root, tag: string, attribute: string): string[];
15
- static getBaseHref(dom: cheerio.Root, url: string): string | undefined;
16
- private hyperlinksConfig;
17
- constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TKeyObject<IParserConfig>);
18
- parse(database: DatabaseService): Promise<void>;
19
- links(): Promise<string[]>;
20
- }
@@ -1,115 +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 { URL } from 'url';
11
- import { commonsArrayUnique, commonsTypeHasPropertyStringArray } from 'tscommons-es-core';
12
- import { HtmlParser } from './html.parser';
13
- import { isIDataConfig } from './data.parser';
14
- export function isIHyperlinksConfig(test) {
15
- if (!isIDataConfig(test))
16
- return false;
17
- if (!commonsTypeHasPropertyStringArray(test, 'sources'))
18
- return false;
19
- return true;
20
- }
21
- export class HyperlinksParser extends HtmlParser {
22
- static parseTagLinks(dom, tag, attribute) {
23
- const links = [];
24
- dom(`${tag}[${attribute}]`).each((_index, element) => {
25
- const link = dom(element).attr(attribute);
26
- if (link !== undefined)
27
- links.push(link);
28
- return true;
29
- });
30
- return commonsArrayUnique(links);
31
- }
32
- static getBaseHref(dom, url) {
33
- try {
34
- const base = dom('base[href]');
35
- if (base.length === 0)
36
- return undefined;
37
- const href = dom(base[0]).attr('href');
38
- if (!href)
39
- return undefined;
40
- const whatwg = new URL(href, url);
41
- return whatwg.toString();
42
- }
43
- catch (ex) { /* do nothing */ }
44
- return undefined;
45
- }
46
- constructor(url, outcome, config) {
47
- super(outcome, config, 'hyperlinks');
48
- this.url = url;
49
- this.hyperlinksConfig = this.getConfig(isIHyperlinksConfig);
50
- }
51
- parse(database) {
52
- return __awaiter(this, void 0, void 0, function* () {
53
- if (!this.dom || !this.url || !this.hyperlinksConfig)
54
- return; // invalid parse
55
- let title;
56
- try {
57
- title = this.dom('title').text();
58
- }
59
- catch (ex) { /* do nothing */ }
60
- if (title !== undefined && title !== null && title !== '') {
61
- yield database.setData(this.url, 'title', title);
62
- }
63
- const links = {};
64
- for (const source of this.hyperlinksConfig.sources) {
65
- const pattern = source.match(/^([a-z]+)\[(src|href|action)\]$/);
66
- if (pattern)
67
- links[pattern[1]] = HyperlinksParser.parseTagLinks(this.dom, pattern[1], pattern[2]);
68
- }
69
- yield database.setData(this.url, 'links', links);
70
- });
71
- }
72
- // eslint-disable-next-line @typescript-eslint/require-await
73
- links() {
74
- return __awaiter(this, void 0, void 0, function* () {
75
- if (!this.dom || !this.url || !this.hyperlinksConfig)
76
- return []; // invalid parse
77
- let links = [];
78
- for (const source of this.hyperlinksConfig.sources) {
79
- const pattern = source.match(/^([a-z]+)\[(src|href|action)\]$/);
80
- if (pattern) {
81
- if (pattern[1] === 'form')
82
- continue;
83
- links = [...links, ...HyperlinksParser.parseTagLinks(this.dom, pattern[1], pattern[2])];
84
- }
85
- }
86
- const unique = [...new Set(links)];
87
- let baseHref = HyperlinksParser.getBaseHref(this.dom, this.url);
88
- if (baseHref === undefined)
89
- baseHref = this.url;
90
- const results = [];
91
- for (let l of unique) {
92
- const anchor = l.split('#');
93
- l = anchor[0];
94
- if (l === '')
95
- continue;
96
- try {
97
- const u = new URL(l, baseHref);
98
- if (u.toString().indexOf('#') !== -1) {
99
- process.exit(1);
100
- }
101
- if (!u.protocol.match(/^http(s?):$/))
102
- continue;
103
- if (u.toString() === this.url || u.toString() === baseHref) {
104
- // skip self-links, or presumed self-links
105
- continue;
106
- }
107
- results.push(u.toString());
108
- }
109
- catch (ex) { /* do nothing */ }
110
- }
111
- return results;
112
- });
113
- }
114
- }
115
- //# sourceMappingURL=hyperlinks.parser.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"hyperlinks.parser.js","sourceRoot":"","sources":["../../src/parsers/hyperlinks.parser.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,KAAK,CAAC;AAE1B,OAAO,EAAE,kBAAkB,EAAE,iCAAiC,EAAc,MAAM,mBAAmB,CAAC;AAOtG,OAAO,EAAE,UAAU,EAAE,MAAM,eAAe,CAAC;AAC3C,OAAO,EAAe,aAAa,EAAE,MAAM,eAAe,CAAC;AAK3D,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;IAE3D,MAAM,CAAC,aAAa,CACzB,GAAiB,EACjB,GAAW,EACX,SAAiB;QAElB,MAAM,KAAK,GAAa,EAAE,CAAC;QAE3B,GAAG,CAAC,GAAG,GAAG,IAAI,SAAS,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwB,EAAW,EAAE;YACtF,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,GAAiB,EAAE,GAAW;QACvD,IAAI;YACH,MAAM,IAAI,GAAoB,GAAG,CAAC,YAAY,CAAC,CAAC;YAChD,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;SACzB;QAAC,OAAO,EAAE,EAAE,EAAE,gBAAgB,EAAE;QACjC,OAAO,SAAS,CAAC;IAClB,CAAC;IAID,YACU,GAAY,EACpB,OAAyB,EACzB,MAAkC;QAEnC,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;IAEY,KAAK,CAAC,QAAyB;;YAC3C,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,gBAAgB;gBAAE,OAAO,CAAC,gBAAgB;YAE9E,IAAI,KAAuB,CAAC;YAC5B,IAAI;gBACH,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,CAAC;aACjC;YAAC,OAAO,EAAE,EAAE,EAAE,gBAAgB,EAAE;YAEjC,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,KAAK,EAAE,EAAE;gBAC1D,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC;aACjD;YAED,MAAM,KAAK,GAAkC,EAAE,CAAC;YAChD,KAAK,MAAM,MAAM,IAAI,IAAI,CAAC,gBAAgB,CAAC,OAAO,EAAE;gBACnD,MAAM,OAAO,GAA0B,MAAM,CAAC,KAAK,CAAC,iCAAiC,CAAC,CAAC;gBACvF,IAAI,OAAO;oBAAE,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;aAClG;YAED,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC;QAClD,CAAC;KAAA;IAED,4DAA4D;IAC/C,KAAK;;YACjB,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,gBAAgB;gBAAE,OAAO,EAAE,CAAC,CAAC,gBAAgB;YAEjF,IAAI,KAAK,GAAa,EAAE,CAAC;YACzB,KAAK,MAAM,MAAM,IAAI,IAAI,CAAC,gBAAgB,CAAC,OAAO,EAAE;gBACnD,MAAM,OAAO,GAA0B,MAAM,CAAC,KAAK,CAAC,iCAAiC,CAAC,CAAC;gBACvF,IAAI,OAAO,EAAE;oBACZ,IAAI,OAAO,CAAC,CAAC,CAAC,KAAK,MAAM;wBAAE,SAAS;oBACpC,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;iBACxF;aACD;YAED,MAAM,MAAM,GAAa,CAAC,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC;YAE7C,IAAI,QAAQ,GAAqB,gBAAgB,CAAC,WAAW,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC;YAClF,IAAI,QAAQ,KAAK,SAAS;gBAAE,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC;YAEhD,MAAM,OAAO,GAAa,EAAE,CAAC;YAC7B,KAAK,IAAI,CAAC,IAAI,MAAM,EAAE;gBACrB,MAAM,MAAM,GAAa,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBACtC,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;gBACd,IAAI,CAAC,KAAK,EAAE;oBAAE,SAAS;gBAEvB,IAAI;oBACH,MAAM,CAAC,GAAQ,IAAI,GAAG,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC;oBACpC,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE;wBACrC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;qBAChB;oBACD,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,aAAa,CAAC;wBAAE,SAAS;oBAC/C,IAAI,CAAC,CAAC,QAAQ,EAAE,KAAK,IAAI,CAAC,GAAG,IAAI,CAAC,CAAC,QAAQ,EAAE,KAAK,QAAQ,EAAE;wBAC3D,0CAA0C;wBAC1C,SAAS;qBACT;oBAED,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAC;iBAC3B;gBAAC,OAAO,EAAE,EAAE,EAAE,gBAAgB,EAAE;aACjC;YAED,OAAO,OAAO,CAAC;QAChB,CAAC;KAAA;CACD"}
@@ -1,19 +0,0 @@
1
- /// <reference types="cheerio" />
2
- import { TKeyObject } from 'tscommons-es-core';
3
- import { DatabaseService } from '../services/database.service';
4
- import { IParserConfig } from '../interfaces/iparser-config';
5
- import { IRequestOutcome } from '../interfaces/irequest-outcome';
6
- import { HtmlParser } from './html.parser';
7
- import { IDataConfig } from './data.parser';
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.Root): TImageTag[];
16
- constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TKeyObject<IParserConfig>);
17
- parse(database: DatabaseService): Promise<void>;
18
- }
19
- export {};
@@ -1,52 +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 { commonsOutputDebug } from 'nodecommons-es-cli';
11
- import { HtmlParser } from './html.parser';
12
- export class ImageTagsParser extends HtmlParser {
13
- static parseImageTags(dom) {
14
- const images = [];
15
- dom('img').each((_index, element) => {
16
- const src = dom(element).attr('src');
17
- if (src === null || src === undefined)
18
- return true;
19
- const alt = dom(element).attr('alt');
20
- const title = dom(element).attr('title');
21
- const image = {
22
- src: src
23
- };
24
- if (alt !== undefined && alt !== null)
25
- image.alt = alt;
26
- if (title !== undefined && title !== null)
27
- image.title = title;
28
- images.push(image);
29
- return true;
30
- });
31
- return images;
32
- }
33
- constructor(url, outcome, config) {
34
- super(outcome, config, 'imageTags');
35
- this.url = url;
36
- }
37
- parse(database) {
38
- return __awaiter(this, void 0, void 0, function* () {
39
- if (!this.dom || !this.url)
40
- return; // invalid parse
41
- const images = ImageTagsParser.parseImageTags(this.dom);
42
- if (images.length > 0) {
43
- commonsOutputDebug(`Detected images for ${this.url}`);
44
- yield database.setData(this.url, 'images', images);
45
- }
46
- else {
47
- yield database.unsetData(this.url, 'images');
48
- }
49
- });
50
- }
51
- }
52
- //# sourceMappingURL=image-tags.parser.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"image-tags.parser.js","sourceRoot":"","sources":["../../src/parsers/image-tags.parser.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AAOxD,OAAO,EAAE,UAAU,EAAE,MAAM,eAAe,CAAC;AAS3C,MAAM,OAAO,eAAgB,SAAQ,UAAuB;IACpD,MAAM,CAAC,cAAc,CAC1B,GAAiB;QAElB,MAAM,MAAM,GAAgB,EAAE,CAAC;QAE/B,GAAG,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwB,EAAW,EAAE;YACrE,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,MAAkC;QAEnC,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,WAAW,CAAC,CAAC;QAJ3B,QAAG,GAAH,GAAG,CAAS;IAKtB,CAAC;IAEY,KAAK,CAAC,QAAyB;;YAC3C,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG;gBAAE,OAAO,CAAC,gBAAgB;YAEpD,MAAM,MAAM,GAAgB,eAAe,CAAC,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACrE,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE;gBACtB,kBAAkB,CAAC,uBAAuB,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC;gBACtD,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,EAAE,MAAM,CAAC,CAAC;aACnD;iBAAM;gBACN,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC;aAC7C;QACF,CAAC;KAAA;CACD"}
@@ -1,69 +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 { commonsTypeHasPropertyString, commonsTypeHasPropertyStringArray, commonsTypeHasPropertyT } from 'tscommons-es-core';
11
- import { commonsOutputDebug, commonsOutputError } from 'nodecommons-es-cli';
12
- import { RigSubjectivityScaleService } from '../services/rig-subjectivity-scale.service';
13
- import { OllamaParser } from './ollama.parser';
14
- import { isIDataConfig } from './data.parser';
15
- export function isIInterestConfig(test) {
16
- if (!isIDataConfig(test))
17
- return false;
18
- if (!commonsTypeHasPropertyString(test, 'model'))
19
- return false;
20
- if (!commonsTypeHasPropertyStringArray(test, 'permittedDomains'))
21
- return false;
22
- return true;
23
- }
24
- export class InterestParser extends OllamaParser {
25
- constructor(url, outcome, config) {
26
- super(outcome, config, 'interest');
27
- this.url = url;
28
- this.permittedDomains = [];
29
- if (!config)
30
- return;
31
- if (!commonsTypeHasPropertyT(config, 'interest', isIInterestConfig)) {
32
- commonsOutputError('Invalid config for InterestParser');
33
- }
34
- this.model = config['interest'].model;
35
- this.permittedDomains = config['interest'].permittedDomains;
36
- }
37
- parseText(database, text) {
38
- return __awaiter(this, void 0, void 0, function* () {
39
- if (!this.url)
40
- return;
41
- if (!this.ollamaUrl || !this.ollamaKey || !this.model)
42
- return;
43
- const hostname = new URL(this.url).hostname;
44
- if (!this.permittedDomains.includes(hostname))
45
- return;
46
- if (text.length === 0) {
47
- yield database.unsetData(this.url, 'interest');
48
- return;
49
- }
50
- const rigService = new RigSubjectivityScaleService(this.ollamaUrl, this.ollamaKey, this.model);
51
- yield this.debugGetLength();
52
- // run this outside of await so that it doesn't block other things
53
- void (() => __awaiter(this, void 0, void 0, function* () {
54
- if (!this.url)
55
- return;
56
- const score = yield rigService.generateInterestScore(text, this.ollamaKeepAlive);
57
- if (score === undefined)
58
- return;
59
- if (score.score < 2)
60
- delete score.reason;
61
- commonsOutputDebug(`Interest score is: ${score.score.toString(10)}`);
62
- if (score.reason)
63
- commonsOutputDebug(`Reason for score is: ${score.reason}`);
64
- yield database.setData(this.url, 'interest', score);
65
- }))();
66
- });
67
- }
68
- }
69
- //# sourceMappingURL=interest.parser.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"interest.parser.js","sourceRoot":"","sources":["../../src/parsers/interest.parser.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAc,4BAA4B,EAAE,iCAAiC,EAAE,uBAAuB,EAAE,MAAM,mBAAmB,CAAC;AAEzI,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AAG5E,OAAO,EAAE,2BAA2B,EAAuB,MAAM,4CAA4C,CAAC;AAK9G,OAAO,EAAiB,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC9D,OAAO,EAAe,aAAa,EAAE,MAAM,eAAe,CAAC;AAM3D,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;IAI9D,YACU,GAAY,EACpB,OAAyB,EACzB,MAAkC;QAEnC,KAAK,CACH,OAAO,EACP,MAAM,EACN,UAAU,CACX,CAAC;QARO,QAAG,GAAH,GAAG,CAAS;QAHd,qBAAgB,GAAa,EAAE,CAAC;QAavC,IAAI,CAAC,MAAM;YAAE,OAAO;QAEpB,IAAI,CAAC,uBAAuB,CAAkB,MAAM,EAAE,UAAU,EAAE,iBAAiB,CAAC,EAAE;YACrF,kBAAkB,CAAC,mCAAmC,CAAC,CAAC;SACxD;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;IAEe,SAAS,CAAC,QAAyB,EAAE,IAAY;;YAChE,IAAI,CAAC,IAAI,CAAC,GAAG;gBAAE,OAAO;YACtB,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,KAAK;gBAAE,OAAO;YAE9D,MAAM,QAAQ,GAAW,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC;YACpD,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,QAAQ,CAAC;gBAAE,OAAO;YAEtD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE;gBACtB,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,UAAU,CAAC,CAAC;gBAC/C,OAAO;aACP;YAED,MAAM,UAAU,GAAgC,IAAI,2BAA2B,CAC7E,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,KAAK,CACX,CAAC;YAEF,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;YAE5B,kEAAkE;YAClE,KAAK,CAAC,GAAwB,EAAE;gBAC/B,IAAI,CAAC,IAAI,CAAC,GAAG;oBAAE,OAAO;gBAEtB,MAAM,KAAK,GAAkC,MAAM,UAAU,CAAC,qBAAqB,CACjF,IAAI,EACJ,IAAI,CAAC,eAAe,CACrB,CAAC;gBACF,IAAI,KAAK,KAAK,SAAS;oBAAE,OAAO;gBAEhC,IAAI,KAAK,CAAC,KAAK,GAAG,CAAC;oBAAE,OAAO,KAAK,CAAC,MAAM,CAAC;gBAEzC,kBAAkB,CAAC,sBAAsB,KAAK,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;gBACrE,IAAI,KAAK,CAAC,MAAM;oBAAE,kBAAkB,CAAC,wBAAwB,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC;gBAE7E,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,UAAU,EAAE,KAAK,CAAC,CAAC;YACrD,CAAC,CAAA,CAAC,EAAE,CAAC;QACN,CAAC;KAAA;CACD"}
@@ -1,11 +0,0 @@
1
- import { TKeyObject } from 'tscommons-es-core';
2
- import { DatabaseService } from '../services/database.service';
3
- import { IRequestOutcome } from '../interfaces/irequest-outcome';
4
- import { IParserConfig } from '../interfaces/iparser-config';
5
- import { DataParser, IDataConfig } from './data.parser';
6
- export declare class JpegParser extends DataParser<IDataConfig> {
7
- private url?;
8
- constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TKeyObject<IParserConfig>);
9
- supports(contentType: string, isAllow: boolean): boolean;
10
- parse(database: DatabaseService): Promise<void>;
11
- }
@@ -1,39 +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 jpeg from 'jpeg-js';
11
- import { DataParser } from './data.parser';
12
- export class JpegParser extends DataParser {
13
- constructor(url, outcome, config) {
14
- super(outcome, config, 'jpeg');
15
- this.url = url;
16
- }
17
- supports(contentType, isAllow) {
18
- if (!isAllow)
19
- return false;
20
- return contentType.match(/^image\/(jpeg|jpg)$/) ? true : false;
21
- }
22
- parse(database) {
23
- return __awaiter(this, void 0, void 0, function* () {
24
- if (!this.outcome || !this.url)
25
- return;
26
- if (!this.outcome.data)
27
- return;
28
- if (this.outcome.exceeded)
29
- return; // don't try and parse vast JPEGs than haven't been downloaded properly, regardless of what the IDataConfig.allowExceeded is set to
30
- try {
31
- const rawImageData = jpeg.decode(this.outcome.data);
32
- yield database.setData(this.url, 'width', rawImageData.width);
33
- yield database.setData(this.url, 'height', rawImageData.height);
34
- }
35
- catch (ex) { /* do nothing */ }
36
- });
37
- }
38
- }
39
- //# sourceMappingURL=jpeg.parser.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"jpeg.parser.js","sourceRoot":"","sources":["../../src/parsers/jpeg.parser.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,KAAK,IAAI,MAAM,SAAS,CAAC;AAShC,OAAO,EAAE,UAAU,EAAe,MAAM,eAAe,CAAC;AAExD,MAAM,OAAO,UAAW,SAAQ,UAAuB;IACtD,YACU,GAAY,EACpB,OAAyB,EACzB,MAAkC;QAEnC,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;IAEY,KAAK,CAAC,QAAyB;;YAC3C,IAAI,CAAC,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,CAAC,GAAG;gBAAE,OAAO;YAEvC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI;gBAAE,OAAO;YAC/B,IAAI,IAAI,CAAC,OAAO,CAAC,QAAQ;gBAAE,OAAO,CAAC,mIAAmI;YAEtK,IAAI;gBACH,MAAM,YAAY,GAAmB,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;gBAEpE,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,OAAO,EAAE,YAAY,CAAC,KAAK,CAAC,CAAC;gBAC9D,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,EAAE,YAAY,CAAC,MAAM,CAAC,CAAC;aAChE;YAAC,OAAO,EAAE,EAAE,EAAE,gBAAgB,EAAE;QAClC,CAAC;KAAA;CACD"}
@@ -1,65 +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 { commonsTypeHasPropertyString, commonsTypeHasPropertyStringArray, commonsTypeHasPropertyT } from 'tscommons-es-core';
11
- import { commonsOutputDebug, commonsOutputError } from 'nodecommons-es-cli';
12
- import { ELlamaGuardClassification, RigLlamaGuardService } from '../services/rig-llama-guard.service';
13
- import { OllamaParser } from './ollama.parser';
14
- import { isIDataConfig } from './data.parser';
15
- export function isILlamaGuardConfig(test) {
16
- if (!isIDataConfig(test))
17
- return false;
18
- if (!commonsTypeHasPropertyString(test, 'model'))
19
- return false;
20
- if (!commonsTypeHasPropertyStringArray(test, 'permittedDomains'))
21
- return false;
22
- return true;
23
- }
24
- export class LlamaGuardParser extends OllamaParser {
25
- constructor(url, outcome, config) {
26
- super(outcome, config, 'llamaguard');
27
- this.url = url;
28
- this.permittedDomains = [];
29
- if (!config)
30
- return;
31
- if (!commonsTypeHasPropertyT(config, 'llamaguard', isILlamaGuardConfig)) {
32
- commonsOutputError('Invalid config for LlamaGuardParser');
33
- }
34
- this.model = config['llamaguard'].model;
35
- this.permittedDomains = config['llamaguard'].permittedDomains;
36
- }
37
- parseText(database, text) {
38
- return __awaiter(this, void 0, void 0, function* () {
39
- if (!this.url)
40
- return;
41
- if (!this.ollamaUrl || !this.ollamaKey || !this.model)
42
- return;
43
- const hostname = new URL(this.url).hostname;
44
- if (!this.permittedDomains.includes(hostname))
45
- return;
46
- if (text.length === 0) {
47
- yield database.unsetData(this.url, 'llamaguard');
48
- return;
49
- }
50
- const rigService = new RigLlamaGuardService(this.ollamaUrl, this.ollamaKey, this.model);
51
- yield this.debugGetLength();
52
- // run this outside of await so that it doesn't block other things
53
- void (() => __awaiter(this, void 0, void 0, function* () {
54
- if (!this.url)
55
- return;
56
- const result = yield rigService.generate(text, this.ollamaKeepAlive);
57
- if (result === undefined)
58
- return;
59
- commonsOutputDebug(`LlamaGuard result is: ${result.classification.toString()}${result.classification === ELlamaGuardClassification.UNSAFE ? ` (${result.categories.join(',')})` : ''}`);
60
- yield database.setData(this.url, 'llamaguard', result);
61
- }))();
62
- });
63
- }
64
- }
65
- //# sourceMappingURL=llama-guard.parser.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"llama-guard.parser.js","sourceRoot":"","sources":["../../src/parsers/llama-guard.parser.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAc,4BAA4B,EAAE,iCAAiC,EAAE,uBAAuB,EAAE,MAAM,mBAAmB,CAAC;AAEzI,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AAG5E,OAAO,EAAE,yBAAyB,EAAE,oBAAoB,EAAqB,MAAM,qCAAqC,CAAC;AAKzH,OAAO,EAAiB,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC9D,OAAO,EAAe,aAAa,EAAE,MAAM,eAAe,CAAC;AAM3D,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;IAIhE,YACU,GAAY,EACpB,OAAyB,EACzB,MAAkC;QAEnC,KAAK,CACH,OAAO,EACP,MAAM,EACN,YAAY,CACb,CAAC;QARO,QAAG,GAAH,GAAG,CAAS;QAHd,qBAAgB,GAAa,EAAE,CAAC;QAavC,IAAI,CAAC,MAAM;YAAE,OAAO;QAEpB,IAAI,CAAC,uBAAuB,CAAoB,MAAM,EAAE,YAAY,EAAE,mBAAmB,CAAC,EAAE;YAC3F,kBAAkB,CAAC,qCAAqC,CAAC,CAAC;SAC1D;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;IAEe,SAAS,CAAC,QAAyB,EAAE,IAAY;;YAChE,IAAI,CAAC,IAAI,CAAC,GAAG;gBAAE,OAAO;YACtB,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,KAAK;gBAAE,OAAO;YAE9D,MAAM,QAAQ,GAAW,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC;YACpD,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,QAAQ,CAAC;gBAAE,OAAO;YAEtD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE;gBACtB,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,YAAY,CAAC,CAAC;gBACjD,OAAO;aACP;YAED,MAAM,UAAU,GAAyB,IAAI,oBAAoB,CAC/D,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,KAAK,CACX,CAAC;YAEF,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;YAE5B,kEAAkE;YAClE,KAAK,CAAC,GAAwB,EAAE;gBAC/B,IAAI,CAAC,IAAI,CAAC,GAAG;oBAAE,OAAO;gBAEtB,MAAM,MAAM,GAAgC,MAAM,UAAU,CAAC,QAAQ,CACnE,IAAI,EACJ,IAAI,CAAC,eAAe,CACrB,CAAC;gBACF,IAAI,MAAM,KAAK,SAAS;oBAAE,OAAO;gBAEjC,kBAAkB,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;gBAExL,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,YAAY,EAAE,MAAM,CAAC,CAAC;YACxD,CAAC,CAAA,CAAC,EAAE,CAAC;QACN,CAAC;KAAA;CACD"}
@@ -1,69 +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 { commonsTypeHasPropertyString, commonsTypeHasPropertyStringArray, commonsTypeHasPropertyT } from 'tscommons-es-core';
11
- import { commonsOutputDebug, commonsOutputError } from 'nodecommons-es-cli';
12
- import { RigSubjectivityScaleService } from '../services/rig-subjectivity-scale.service';
13
- import { OllamaParser } from './ollama.parser';
14
- import { isIDataConfig } from './data.parser';
15
- export function isIOffenceConfig(test) {
16
- if (!isIDataConfig(test))
17
- return false;
18
- if (!commonsTypeHasPropertyString(test, 'model'))
19
- return false;
20
- if (!commonsTypeHasPropertyStringArray(test, 'permittedDomains'))
21
- return false;
22
- return true;
23
- }
24
- export class OffenceParser extends OllamaParser {
25
- constructor(url, outcome, config) {
26
- super(outcome, config, 'offence');
27
- this.url = url;
28
- this.permittedDomains = [];
29
- if (!config)
30
- return;
31
- if (!commonsTypeHasPropertyT(config, 'offence', isIOffenceConfig)) {
32
- commonsOutputError('Invalid config for OffenceParser');
33
- }
34
- this.model = config['offence'].model;
35
- this.permittedDomains = config['offence'].permittedDomains;
36
- }
37
- parseText(database, text) {
38
- return __awaiter(this, void 0, void 0, function* () {
39
- if (!this.url)
40
- return;
41
- if (!this.ollamaUrl || !this.ollamaKey || !this.model)
42
- return;
43
- const hostname = new URL(this.url).hostname;
44
- if (!this.permittedDomains.includes(hostname))
45
- return;
46
- if (text.length === 0) {
47
- yield database.unsetData(this.url, 'offence');
48
- return;
49
- }
50
- const rigService = new RigSubjectivityScaleService(this.ollamaUrl, this.ollamaKey, this.model);
51
- yield this.debugGetLength();
52
- // run this outside of await so that it doesn't block other things
53
- void (() => __awaiter(this, void 0, void 0, function* () {
54
- if (!this.url)
55
- return;
56
- const score = yield rigService.generateOffenceScore(text, this.ollamaKeepAlive);
57
- if (score === undefined)
58
- return;
59
- if (score.score < 1)
60
- delete score.reason;
61
- commonsOutputDebug(`Offence score is: ${score.score.toString(10)}`);
62
- if (score.reason)
63
- commonsOutputDebug(`Reason for score is: ${score.reason}`);
64
- yield database.setData(this.url, 'offence', score);
65
- }))();
66
- });
67
- }
68
- }
69
- //# sourceMappingURL=offence.parser.js.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"offence.parser.js","sourceRoot":"","sources":["../../src/parsers/offence.parser.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAc,4BAA4B,EAAE,iCAAiC,EAAE,uBAAuB,EAAE,MAAM,mBAAmB,CAAC;AAEzI,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAC;AAG5E,OAAO,EAAE,2BAA2B,EAAuB,MAAM,4CAA4C,CAAC;AAK9G,OAAO,EAAiB,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC9D,OAAO,EAAe,aAAa,EAAE,MAAM,eAAe,CAAC;AAM3D,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;IAI7D,YACU,GAAY,EACpB,OAAyB,EACzB,MAAkC;QAEnC,KAAK,CACH,OAAO,EACP,MAAM,EACN,SAAS,CACV,CAAC;QARO,QAAG,GAAH,GAAG,CAAS;QAHd,qBAAgB,GAAa,EAAE,CAAC;QAavC,IAAI,CAAC,MAAM;YAAE,OAAO;QAEpB,IAAI,CAAC,uBAAuB,CAAiB,MAAM,EAAE,SAAS,EAAE,gBAAgB,CAAC,EAAE;YAClF,kBAAkB,CAAC,kCAAkC,CAAC,CAAC;SACvD;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;IAEe,SAAS,CAAC,QAAyB,EAAE,IAAY;;YAChE,IAAI,CAAC,IAAI,CAAC,GAAG;gBAAE,OAAO;YACtB,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,SAAS,IAAI,CAAC,IAAI,CAAC,KAAK;gBAAE,OAAO;YAE9D,MAAM,QAAQ,GAAW,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC;YACpD,IAAI,CAAC,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,QAAQ,CAAC;gBAAE,OAAO;YAEtD,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE;gBACtB,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,SAAS,CAAC,CAAC;gBAC9C,OAAO;aACP;YAED,MAAM,UAAU,GAAgC,IAAI,2BAA2B,CAC7E,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,SAAS,EACd,IAAI,CAAC,KAAK,CACX,CAAC;YAEF,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;YAE5B,kEAAkE;YAClE,KAAK,CAAC,GAAwB,EAAE;gBAC/B,IAAI,CAAC,IAAI,CAAC,GAAG;oBAAE,OAAO;gBAEtB,MAAM,KAAK,GAAkC,MAAM,UAAU,CAAC,oBAAoB,CAChF,IAAI,EACJ,IAAI,CAAC,eAAe,CACrB,CAAC;gBACF,IAAI,KAAK,KAAK,SAAS;oBAAE,OAAO;gBAEhC,IAAI,KAAK,CAAC,KAAK,GAAG,CAAC;oBAAE,OAAO,KAAK,CAAC,MAAM,CAAC;gBAEzC,kBAAkB,CAAC,qBAAqB,KAAK,CAAC,KAAK,CAAC,QAAQ,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC;gBACpE,IAAI,KAAK,CAAC,MAAM;oBAAE,kBAAkB,CAAC,wBAAwB,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC;gBAE7E,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,SAAS,EAAE,KAAK,CAAC,CAAC;YACpD,CAAC,CAAA,CAAC,EAAE,CAAC;QACN,CAAC;KAAA;CACD"}