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.
- package/dist/apis/{autocomplete.api.d.ts → autocomplete.api.d.mts} +4 -4
- package/dist/apis/autocomplete.api.mjs +14 -0
- package/dist/apis/autocomplete.api.mjs.map +1 -0
- package/dist/apis/{bugs.api.d.ts → bugs.api.d.mts} +4 -4
- package/dist/apis/bugs.api.mjs +18 -0
- package/dist/apis/bugs.api.mjs.map +1 -0
- package/dist/apis/{crawl.api.d.ts → crawl.api.d.mts} +4 -4
- package/dist/apis/crawl.api.mjs +17 -0
- package/dist/apis/crawl.api.mjs.map +1 -0
- package/dist/apis/{domains.api.d.ts → domains.api.d.mts} +4 -4
- package/dist/apis/{domains.api.js → domains.api.mjs} +10 -18
- package/dist/apis/domains.api.mjs.map +1 -0
- package/dist/apis/{images.api.d.ts → images.api.d.mts} +4 -4
- package/dist/apis/images.api.mjs +19 -0
- package/dist/apis/images.api.mjs.map +1 -0
- package/dist/apis/{statistics.api.d.ts → statistics.api.d.mts} +4 -4
- package/dist/apis/statistics.api.mjs +33 -0
- package/dist/apis/statistics.api.mjs.map +1 -0
- package/dist/apis/{test.api.d.ts → test.api.d.mts} +3 -3
- package/dist/apis/test.api.mjs +10 -0
- package/dist/apis/test.api.mjs.map +1 -0
- package/dist/apis/{urls.api.d.ts → urls.api.d.mts} +4 -4
- package/dist/apis/urls.api.mjs +21 -0
- package/dist/apis/urls.api.mjs.map +1 -0
- package/dist/apps/{cleanup.app.d.ts → cleanup.app.d.mts} +5 -5
- package/dist/apps/cleanup.app.mjs +129 -0
- package/dist/apps/cleanup.app.mjs.map +1 -0
- package/dist/apps/{cross-populate-export.app.d.ts → cross-populate-export.app.d.mts} +3 -3
- package/dist/apps/cross-populate-export.app.mjs +61 -0
- package/dist/apps/cross-populate-export.app.mjs.map +1 -0
- package/dist/apps/{cross-populate-import.app.d.ts → cross-populate-import.app.d.mts} +3 -3
- package/dist/apps/cross-populate-import.app.mjs +61 -0
- package/dist/apps/cross-populate-import.app.mjs.map +1 -0
- package/dist/apps/{denylist.app.d.ts → denylist.app.d.mts} +5 -5
- package/dist/apps/denylist.app.mjs +135 -0
- package/dist/apps/denylist.app.mjs.map +1 -0
- package/dist/apps/{expire.app.d.ts → expire.app.d.mts} +8 -7
- package/dist/apps/expire.app.mjs +51 -0
- package/dist/apps/expire.app.mjs.map +1 -0
- package/dist/apps/{export-domain-urls.d.ts → export-domain-urls.d.mts} +3 -3
- package/dist/apps/export-domain-urls.mjs +85 -0
- package/dist/apps/export-domain-urls.mjs.map +1 -0
- package/dist/apps/{extract-text.app.d.ts → extract-text.app.d.mts} +2 -2
- package/dist/apps/extract-text.app.mjs +43 -0
- package/dist/apps/extract-text.app.mjs.map +1 -0
- package/dist/apps/{hydra.app.d.ts → hydra.app.d.mts} +10 -10
- package/dist/apps/hydra.app.mjs +222 -0
- package/dist/apps/hydra.app.mjs.map +1 -0
- package/dist/apps/{import.app.d.ts → import.app.d.mts} +3 -3
- package/dist/apps/import.app.mjs +48 -0
- package/dist/apps/import.app.mjs.map +1 -0
- package/dist/apps/{internal-hydra-common.app.d.ts → internal-hydra-common.app.d.mts} +6 -6
- package/dist/apps/{internal-hydra-common.app.js → internal-hydra-common.app.mjs} +1 -1
- package/dist/apps/internal-hydra-common.app.mjs.map +1 -0
- package/dist/apps/{move-to-archive.app.d.ts → move-to-archive.app.d.mts} +2 -2
- package/dist/apps/move-to-archive.app.mjs +31 -0
- package/dist/apps/move-to-archive.app.mjs.map +1 -0
- package/dist/apps/{prune-archive.app.d.ts → prune-archive.app.d.mts} +2 -2
- package/dist/apps/prune-archive.app.mjs +40 -0
- package/dist/apps/prune-archive.app.mjs.map +1 -0
- package/dist/apps/{query.app.d.ts → query.app.d.mts} +6 -6
- package/dist/apps/query.app.mjs +69 -0
- package/dist/apps/query.app.mjs.map +1 -0
- package/dist/apps/{reattempt.app.d.ts → reattempt.app.d.mts} +6 -6
- package/dist/apps/reattempt.app.mjs +81 -0
- package/dist/apps/reattempt.app.mjs.map +1 -0
- package/dist/apps/{requeue-domain.app.d.ts → requeue-domain.app.d.mts} +3 -3
- package/dist/apps/requeue-domain.app.mjs +52 -0
- package/dist/apps/requeue-domain.app.mjs.map +1 -0
- package/dist/apps/{seed.app.d.ts → seed.app.d.mts} +5 -5
- package/dist/apps/seed.app.mjs +59 -0
- package/dist/apps/seed.app.mjs.map +1 -0
- package/dist/apps/{startup.app.d.ts → startup.app.d.mts} +3 -3
- package/dist/apps/startup.app.mjs +55 -0
- package/dist/apps/startup.app.mjs.map +1 -0
- package/dist/apps/{unarchive-urls.app.d.ts → unarchive-urls.app.d.mts} +5 -5
- package/dist/apps/unarchive-urls.app.mjs +78 -0
- package/dist/apps/unarchive-urls.app.mjs.map +1 -0
- package/dist/classes/{cleaner.d.ts → cleaner.d.mts} +2 -2
- package/dist/classes/cleaner.mjs +264 -0
- package/dist/classes/cleaner.mjs.map +1 -0
- package/dist/classes/{crawler.d.ts → crawler.d.mts} +10 -12
- package/dist/classes/{crawler.js → crawler.mjs} +216 -184
- package/dist/classes/crawler.mjs.map +1 -0
- package/dist/classes/{dns.js → dns.mjs} +4 -4
- package/dist/classes/dns.mjs.map +1 -0
- package/dist/classes/{expirer.d.ts → expirer.d.mts} +2 -2
- package/dist/classes/expirer.mjs +117 -0
- package/dist/classes/expirer.mjs.map +1 -0
- package/dist/classes/{expiry.d.ts → expiry.d.mts} +1 -1
- package/dist/classes/{expiry.js → expiry.mjs} +7 -9
- package/dist/classes/expiry.mjs.map +1 -0
- package/dist/classes/{lists.d.ts → lists.d.mts} +2 -2
- package/dist/classes/{lists.js → lists.mjs} +4 -4
- package/dist/classes/lists.mjs.map +1 -0
- package/dist/classes/{robot.d.ts → robot.d.mts} +3 -3
- package/dist/classes/robot.mjs +74 -0
- package/dist/classes/robot.mjs.map +1 -0
- package/dist/classes/{tracker.d.ts → tracker.d.mts} +3 -3
- package/dist/classes/tracker.mjs +101 -0
- package/dist/classes/tracker.mjs.map +1 -0
- package/dist/{cli.js → cli.mjs} +46 -58
- package/dist/cli.mjs.map +1 -0
- package/dist/enums/{eavailable-strategy.js → eavailable-strategy.mjs} +1 -1
- package/dist/enums/eavailable-strategy.mjs.map +1 -0
- package/dist/enums/{elist.js → elist.mjs} +1 -1
- package/dist/enums/elist.mjs.map +1 -0
- package/dist/enums/{eserver.js → eserver.mjs} +1 -1
- package/dist/enums/eserver.mjs.map +1 -0
- package/dist/enums/{ex-powered-by.js → ex-powered-by.mjs} +1 -1
- package/dist/enums/ex-powered-by.mjs.map +1 -0
- package/dist/helpers/{matcher.d.ts → matcher.d.mts} +1 -1
- package/dist/helpers/{matcher.js → matcher.mjs} +1 -1
- package/dist/helpers/matcher.mjs.map +1 -0
- package/dist/helpers/{random.d.ts → random.d.mts} +1 -1
- package/dist/helpers/{random.js → random.mjs} +1 -1
- package/dist/helpers/random.mjs.map +1 -0
- package/dist/helpers/{utf-decoder.d.ts → utf-decoder.d.mts} +0 -1
- package/dist/helpers/{utf-decoder.js → utf-decoder.mjs} +3 -3
- package/dist/helpers/utf-decoder.mjs.map +1 -0
- package/dist/interfaces/{iexpiry.d.ts → iexpiry.d.mts} +1 -1
- package/dist/interfaces/{iexpiry.js → iexpiry.mjs} +3 -3
- package/dist/interfaces/iexpiry.mjs.map +1 -0
- package/dist/interfaces/{imatch.js → imatch.mjs} +2 -2
- package/dist/interfaces/imatch.mjs.map +1 -0
- package/dist/interfaces/{iparser-config.js → iparser-config.mjs} +2 -2
- package/dist/interfaces/iparser-config.mjs.map +1 -0
- package/dist/interfaces/{iparser.d.ts → iparser.d.mts} +1 -1
- package/dist/interfaces/iparser.mjs +2 -0
- package/dist/interfaces/iparser.mjs.map +1 -0
- package/dist/interfaces/{irequest-outcome.d.ts → irequest-outcome.d.mts} +0 -2
- package/dist/interfaces/irequest-outcome.mjs +2 -0
- package/dist/interfaces/irequest-outcome.mjs.map +1 -0
- package/dist/interfaces/iserver.mjs +2 -0
- package/dist/interfaces/iserver.mjs.map +1 -0
- package/dist/parsers/accessibility-metrics.parser.d.mts +11 -0
- package/dist/parsers/accessibility-metrics.parser.mjs +30 -0
- package/dist/parsers/accessibility-metrics.parser.mjs.map +1 -0
- package/dist/parsers/asp-error.parser.d.mts +12 -0
- package/dist/parsers/asp-error.parser.mjs +38 -0
- package/dist/parsers/asp-error.parser.mjs.map +1 -0
- package/dist/parsers/{bad-words.parser.d.ts → bad-words.parser.d.mts} +6 -6
- package/dist/parsers/bad-words.parser.mjs +17 -0
- package/dist/parsers/bad-words.parser.mjs.map +1 -0
- package/dist/parsers/complex-english.parser.d.mts +15 -0
- package/dist/parsers/complex-english.parser.mjs +52 -0
- package/dist/parsers/complex-english.parser.mjs.map +1 -0
- package/dist/parsers/data.parser.d.mts +13 -0
- package/dist/parsers/{data.parser.js → data.parser.mjs} +8 -7
- package/dist/parsers/data.parser.mjs.map +1 -0
- package/dist/parsers/{dictionary.parser.d.ts → dictionary.parser.d.mts} +6 -6
- package/dist/parsers/dictionary.parser.mjs +63 -0
- package/dist/parsers/dictionary.parser.mjs.map +1 -0
- package/dist/parsers/html.parser.d.mts +13 -0
- package/dist/parsers/{html.parser.js → html.parser.mjs} +4 -3
- package/dist/parsers/html.parser.mjs.map +1 -0
- package/dist/parsers/hyperlinks.parser.d.mts +20 -0
- package/dist/parsers/hyperlinks.parser.mjs +104 -0
- package/dist/parsers/hyperlinks.parser.mjs.map +1 -0
- package/dist/parsers/image-tags.parser.d.mts +19 -0
- package/dist/parsers/image-tags.parser.mjs +42 -0
- package/dist/parsers/image-tags.parser.mjs.map +1 -0
- package/dist/parsers/{interest.parser.d.ts → interest.parser.d.mts} +7 -7
- package/dist/parsers/interest.parser.mjs +60 -0
- package/dist/parsers/interest.parser.mjs.map +1 -0
- package/dist/parsers/jpeg.parser.d.mts +11 -0
- package/dist/parsers/jpeg.parser.mjs +29 -0
- package/dist/parsers/jpeg.parser.mjs.map +1 -0
- package/dist/parsers/{llama-guard.parser.d.ts → llama-guard.parser.d.mts} +7 -7
- package/dist/parsers/llama-guard.parser.mjs +56 -0
- package/dist/parsers/llama-guard.parser.mjs.map +1 -0
- package/dist/parsers/{offence.parser.d.ts → offence.parser.d.mts} +7 -7
- package/dist/parsers/offence.parser.mjs +60 -0
- package/dist/parsers/offence.parser.mjs.map +1 -0
- package/dist/parsers/{ollama.parser.d.ts → ollama.parser.d.mts} +6 -6
- package/dist/parsers/ollama.parser.mjs +43 -0
- package/dist/parsers/ollama.parser.mjs.map +1 -0
- package/dist/parsers/{paragraphs.parser.d.ts → paragraphs.parser.d.mts} +5 -5
- package/dist/parsers/paragraphs.parser.mjs +38 -0
- package/dist/parsers/paragraphs.parser.mjs.map +1 -0
- package/dist/parsers/{parser.d.ts → parser.d.mts} +6 -6
- package/dist/parsers/parser.mjs +45 -0
- package/dist/parsers/parser.mjs.map +1 -0
- package/dist/parsers/php-error.parser.d.mts +12 -0
- package/dist/parsers/php-error.parser.mjs +42 -0
- package/dist/parsers/php-error.parser.mjs.map +1 -0
- package/dist/parsers/{phrase.parser.d.ts → phrase.parser.d.mts} +3 -3
- package/dist/parsers/phrase.parser.mjs +15 -0
- package/dist/parsers/phrase.parser.mjs.map +1 -0
- package/dist/parsers/{regex.parser.d.ts → regex.parser.d.mts} +3 -3
- package/dist/parsers/regex.parser.mjs +29 -0
- package/dist/parsers/regex.parser.mjs.map +1 -0
- package/dist/parsers/server.parser.d.mts +11 -0
- package/dist/parsers/server.parser.mjs +57 -0
- package/dist/parsers/server.parser.mjs.map +1 -0
- package/dist/parsers/{spelling.parser.d.ts → spelling.parser.d.mts} +6 -6
- package/dist/parsers/spelling.parser.mjs +17 -0
- package/dist/parsers/spelling.parser.mjs.map +1 -0
- package/dist/parsers/string.parser.d.mts +8 -0
- package/dist/parsers/{string.parser.js → string.parser.mjs} +5 -4
- package/dist/parsers/string.parser.mjs.map +1 -0
- package/dist/parsers/{text.parser.d.ts → text.parser.d.mts} +3 -3
- package/dist/parsers/text.parser.mjs +30 -0
- package/dist/parsers/text.parser.mjs.map +1 -0
- package/dist/parsers/{words.parser.d.ts → words.parser.d.mts} +3 -3
- package/dist/parsers/words.parser.mjs +29 -0
- package/dist/parsers/words.parser.mjs.map +1 -0
- package/dist/queries/complex-english.query.d.mts +2 -0
- package/dist/queries/{complex-english.query.js → complex-english.query.mjs} +26 -28
- package/dist/queries/complex-english.query.mjs.map +1 -0
- package/dist/queries/flash-content.query.d.mts +2 -0
- package/dist/queries/flash-content.query.mjs +80 -0
- package/dist/queries/flash-content.query.mjs.map +1 -0
- package/dist/queries/linking-to-domains.query.d.mts +2 -0
- package/dist/queries/linking-to-domains.query.mjs +128 -0
- package/dist/queries/linking-to-domains.query.mjs.map +1 -0
- package/dist/queries/llamaguard-unsafe-content.query.d.mts +2 -0
- package/dist/queries/llamaguard-unsafe-content.query.mjs +90 -0
- package/dist/queries/llamaguard-unsafe-content.query.mjs.map +1 -0
- package/dist/queries/readability-score.query.d.mts +2 -0
- package/dist/queries/{readability-score.query.js → readability-score.query.mjs} +11 -20
- package/dist/queries/readability-score.query.mjs.map +1 -0
- package/dist/servers/{crawl.server.d.ts → crawl.server.d.mts} +10 -10
- package/dist/servers/crawl.server.mjs +192 -0
- package/dist/servers/crawl.server.mjs.map +1 -0
- package/dist/servers/{express.server.d.ts → express.server.d.mts} +2 -3
- package/dist/servers/express.server.mjs +13 -0
- package/dist/servers/express.server.mjs.map +1 -0
- package/dist/servers/{maintenance.server.d.ts → maintenance.server.d.mts} +5 -5
- package/dist/servers/maintenance.server.mjs +97 -0
- package/dist/servers/maintenance.server.mjs.map +1 -0
- package/dist/servers/{rest.server.d.ts → rest.server.d.mts} +4 -4
- package/dist/servers/rest.server.mjs +62 -0
- package/dist/servers/rest.server.mjs.map +1 -0
- package/dist/servers/{socket-io.server.d.ts → socket-io.server.d.mts} +4 -4
- package/dist/servers/socket-io.server.mjs +22 -0
- package/dist/servers/socket-io.server.mjs.map +1 -0
- package/dist/services/{database.service.d.ts → database.service.d.mts} +12 -12
- package/dist/services/database.service.mjs +645 -0
- package/dist/services/database.service.mjs.map +1 -0
- package/dist/services/{ollama-rest.service.d.ts → ollama-rest.service.d.mts} +2 -2
- package/dist/services/ollama-rest.service.mjs +27 -0
- package/dist/services/ollama-rest.service.mjs.map +1 -0
- package/dist/services/{rig-llama-guard.service.d.ts → rig-llama-guard.service.d.mts} +2 -2
- package/dist/services/{rig-llama-guard.service.js → rig-llama-guard.service.mjs} +33 -43
- package/dist/services/rig-llama-guard.service.mjs.map +1 -0
- package/dist/services/{rig-queue-length.service.d.ts → rig-queue-length.service.d.mts} +1 -1
- package/dist/services/rig-queue-length.service.mjs +12 -0
- package/dist/services/rig-queue-length.service.mjs.map +1 -0
- package/dist/services/{rig-subjectivity-scale.service.d.ts → rig-subjectivity-scale.service.d.mts} +2 -2
- package/dist/services/rig-subjectivity-scale.service.mjs +82 -0
- package/dist/services/rig-subjectivity-scale.service.mjs.map +1 -0
- package/dist/services/{rig-ticket-rest.service.d.ts → rig-ticket-rest.service.d.mts} +3 -3
- package/dist/services/rig-ticket-rest.service.mjs +15 -0
- package/dist/services/rig-ticket-rest.service.mjs.map +1 -0
- package/dist/services/{rig-ticketed-promise.service.d.ts → rig-ticketed-promise.service.d.mts} +4 -6
- package/dist/services/{rig-ticketed-promise.service.js → rig-ticketed-promise.service.mjs} +7 -6
- package/dist/services/rig-ticketed-promise.service.mjs.map +1 -0
- package/dist/types/{tcrawl-config.js → tcrawl-config.mjs} +2 -2
- package/dist/types/tcrawl-config.mjs.map +1 -0
- package/dist/types/{thydra-config.js → thydra-config.mjs} +2 -2
- package/dist/types/thydra-config.mjs.map +1 -0
- package/dist/types/tparser-ctor.d.mts +7 -0
- package/dist/types/tparser-ctor.mjs +2 -0
- package/dist/types/tparser-ctor.mjs.map +1 -0
- package/dist/types/tquery.d.mts +7 -0
- package/dist/types/tquery.mjs +2 -0
- package/dist/types/tquery.mjs.map +1 -0
- package/dist/types/tqueue-length.mjs +2 -0
- package/dist/types/tqueue-length.mjs.map +1 -0
- package/dist/types/{trobots-config.js → trobots-config.mjs} +2 -2
- package/dist/types/trobots-config.mjs.map +1 -0
- package/package.json +37 -43
- package/dist/apis/autocomplete.api.js +0 -22
- package/dist/apis/autocomplete.api.js.map +0 -1
- package/dist/apis/bugs.api.js +0 -26
- package/dist/apis/bugs.api.js.map +0 -1
- package/dist/apis/crawl.api.js +0 -25
- package/dist/apis/crawl.api.js.map +0 -1
- package/dist/apis/domains.api.js.map +0 -1
- package/dist/apis/images.api.js +0 -27
- package/dist/apis/images.api.js.map +0 -1
- package/dist/apis/statistics.api.js +0 -41
- package/dist/apis/statistics.api.js.map +0 -1
- package/dist/apis/test.api.js +0 -19
- package/dist/apis/test.api.js.map +0 -1
- package/dist/apis/urls.api.js +0 -29
- package/dist/apis/urls.api.js.map +0 -1
- package/dist/apps/cleanup.app.js +0 -151
- package/dist/apps/cleanup.app.js.map +0 -1
- package/dist/apps/cross-populate-export.app.js +0 -75
- package/dist/apps/cross-populate-export.app.js.map +0 -1
- package/dist/apps/cross-populate-import.app.js +0 -100
- package/dist/apps/cross-populate-import.app.js.map +0 -1
- package/dist/apps/denylist.app.js +0 -132
- package/dist/apps/denylist.app.js.map +0 -1
- package/dist/apps/expire.app.js +0 -63
- package/dist/apps/expire.app.js.map +0 -1
- package/dist/apps/export-domain-urls.js +0 -99
- package/dist/apps/export-domain-urls.js.map +0 -1
- package/dist/apps/extract-text.app.js +0 -55
- package/dist/apps/extract-text.app.js.map +0 -1
- package/dist/apps/hydra.app.js +0 -218
- package/dist/apps/hydra.app.js.map +0 -1
- package/dist/apps/import.app.js +0 -57
- package/dist/apps/import.app.js.map +0 -1
- package/dist/apps/internal-hydra-common.app.js.map +0 -1
- package/dist/apps/move-to-archive.app.js +0 -46
- package/dist/apps/move-to-archive.app.js.map +0 -1
- package/dist/apps/prune-archive.app.js +0 -55
- package/dist/apps/prune-archive.app.js.map +0 -1
- package/dist/apps/query.app.js +0 -80
- package/dist/apps/query.app.js.map +0 -1
- package/dist/apps/reattempt.app.js +0 -83
- package/dist/apps/reattempt.app.js.map +0 -1
- package/dist/apps/requeue-domain.app.js +0 -64
- package/dist/apps/requeue-domain.app.js.map +0 -1
- package/dist/apps/seed.app.js +0 -69
- package/dist/apps/seed.app.js.map +0 -1
- package/dist/apps/startup.app.js +0 -64
- package/dist/apps/startup.app.js.map +0 -1
- package/dist/apps/unarchive-urls.app.js +0 -83
- package/dist/apps/unarchive-urls.app.js.map +0 -1
- package/dist/classes/cleaner.js +0 -266
- package/dist/classes/cleaner.js.map +0 -1
- package/dist/classes/crawler.js.map +0 -1
- package/dist/classes/dns.js.map +0 -1
- package/dist/classes/expirer.js +0 -121
- package/dist/classes/expirer.js.map +0 -1
- package/dist/classes/expiry.js.map +0 -1
- package/dist/classes/lists.js.map +0 -1
- package/dist/classes/robot.js +0 -82
- package/dist/classes/robot.js.map +0 -1
- package/dist/classes/tracker.js +0 -120
- package/dist/classes/tracker.js.map +0 -1
- package/dist/cli.js.map +0 -1
- package/dist/enums/eavailable-strategy.js.map +0 -1
- package/dist/enums/elist.js.map +0 -1
- package/dist/enums/eserver.js.map +0 -1
- package/dist/enums/ex-powered-by.js.map +0 -1
- package/dist/helpers/matcher.js.map +0 -1
- package/dist/helpers/random.js.map +0 -1
- package/dist/helpers/utf-decoder.js.map +0 -1
- package/dist/interfaces/iexpiry.js.map +0 -1
- package/dist/interfaces/imatch.js.map +0 -1
- package/dist/interfaces/iparser-config.js.map +0 -1
- package/dist/interfaces/iparser.js +0 -2
- package/dist/interfaces/iparser.js.map +0 -1
- package/dist/interfaces/irequest-outcome.js +0 -2
- package/dist/interfaces/irequest-outcome.js.map +0 -1
- package/dist/interfaces/iserver.js +0 -2
- package/dist/interfaces/iserver.js.map +0 -1
- package/dist/parsers/accessibility-metrics.parser.d.ts +0 -11
- package/dist/parsers/accessibility-metrics.parser.js +0 -40
- package/dist/parsers/accessibility-metrics.parser.js.map +0 -1
- package/dist/parsers/asp-error.parser.d.ts +0 -12
- package/dist/parsers/asp-error.parser.js +0 -48
- package/dist/parsers/asp-error.parser.js.map +0 -1
- package/dist/parsers/bad-words.parser.js +0 -27
- package/dist/parsers/bad-words.parser.js.map +0 -1
- package/dist/parsers/complex-english.parser.d.ts +0 -15
- package/dist/parsers/complex-english.parser.js +0 -61
- package/dist/parsers/complex-english.parser.js.map +0 -1
- package/dist/parsers/data.parser.d.ts +0 -14
- package/dist/parsers/data.parser.js.map +0 -1
- package/dist/parsers/dictionary.parser.js +0 -73
- package/dist/parsers/dictionary.parser.js.map +0 -1
- package/dist/parsers/html.parser.d.ts +0 -13
- package/dist/parsers/html.parser.js.map +0 -1
- package/dist/parsers/hyperlinks.parser.d.ts +0 -20
- package/dist/parsers/hyperlinks.parser.js +0 -115
- package/dist/parsers/hyperlinks.parser.js.map +0 -1
- package/dist/parsers/image-tags.parser.d.ts +0 -19
- package/dist/parsers/image-tags.parser.js +0 -52
- package/dist/parsers/image-tags.parser.js.map +0 -1
- package/dist/parsers/interest.parser.js +0 -69
- package/dist/parsers/interest.parser.js.map +0 -1
- package/dist/parsers/jpeg.parser.d.ts +0 -11
- package/dist/parsers/jpeg.parser.js +0 -39
- package/dist/parsers/jpeg.parser.js.map +0 -1
- package/dist/parsers/llama-guard.parser.js +0 -65
- package/dist/parsers/llama-guard.parser.js.map +0 -1
- package/dist/parsers/offence.parser.js +0 -69
- package/dist/parsers/offence.parser.js.map +0 -1
- package/dist/parsers/ollama.parser.js +0 -51
- package/dist/parsers/ollama.parser.js.map +0 -1
- package/dist/parsers/paragraphs.parser.js +0 -49
- package/dist/parsers/paragraphs.parser.js.map +0 -1
- package/dist/parsers/parser.js +0 -57
- package/dist/parsers/parser.js.map +0 -1
- package/dist/parsers/php-error.parser.d.ts +0 -12
- package/dist/parsers/php-error.parser.js +0 -52
- package/dist/parsers/php-error.parser.js.map +0 -1
- package/dist/parsers/phrase.parser.js +0 -26
- package/dist/parsers/phrase.parser.js.map +0 -1
- package/dist/parsers/regex.parser.js +0 -43
- package/dist/parsers/regex.parser.js.map +0 -1
- package/dist/parsers/server.parser.d.ts +0 -11
- package/dist/parsers/server.parser.js +0 -67
- package/dist/parsers/server.parser.js.map +0 -1
- package/dist/parsers/spelling.parser.js +0 -27
- package/dist/parsers/spelling.parser.js.map +0 -1
- package/dist/parsers/string.parser.d.ts +0 -8
- package/dist/parsers/string.parser.js.map +0 -1
- package/dist/parsers/text.parser.js +0 -41
- package/dist/parsers/text.parser.js.map +0 -1
- package/dist/parsers/words.parser.js +0 -40
- package/dist/parsers/words.parser.js.map +0 -1
- package/dist/queries/complex-english.query.d.ts +0 -2
- package/dist/queries/complex-english.query.js.map +0 -1
- package/dist/queries/flash-content.query.d.ts +0 -2
- package/dist/queries/flash-content.query.js +0 -82
- package/dist/queries/flash-content.query.js.map +0 -1
- package/dist/queries/linking-to-domains.query.d.ts +0 -2
- package/dist/queries/linking-to-domains.query.js +0 -130
- package/dist/queries/linking-to-domains.query.js.map +0 -1
- package/dist/queries/llamaguard-unsafe-content.query.d.ts +0 -2
- package/dist/queries/llamaguard-unsafe-content.query.js +0 -92
- package/dist/queries/llamaguard-unsafe-content.query.js.map +0 -1
- package/dist/queries/readability-score.query.d.ts +0 -2
- package/dist/queries/readability-score.query.js.map +0 -1
- package/dist/servers/crawl.server.js +0 -198
- package/dist/servers/crawl.server.js.map +0 -1
- package/dist/servers/express.server.js +0 -13
- package/dist/servers/express.server.js.map +0 -1
- package/dist/servers/maintenance.server.js +0 -103
- package/dist/servers/maintenance.server.js.map +0 -1
- package/dist/servers/rest.server.js +0 -46
- package/dist/servers/rest.server.js.map +0 -1
- package/dist/servers/socket-io.server.js +0 -61
- package/dist/servers/socket-io.server.js.map +0 -1
- package/dist/services/database.service.js +0 -795
- package/dist/services/database.service.js.map +0 -1
- package/dist/services/ollama-rest.service.js +0 -46
- package/dist/services/ollama-rest.service.js.map +0 -1
- package/dist/services/rig-llama-guard.service.js.map +0 -1
- package/dist/services/rig-queue-length.service.js +0 -22
- package/dist/services/rig-queue-length.service.js.map +0 -1
- package/dist/services/rig-subjectivity-scale.service.js +0 -96
- package/dist/services/rig-subjectivity-scale.service.js.map +0 -1
- package/dist/services/rig-ticket-rest.service.js +0 -29
- package/dist/services/rig-ticket-rest.service.js.map +0 -1
- package/dist/services/rig-ticketed-promise.service.js.map +0 -1
- package/dist/types/tcrawl-config.js.map +0 -1
- package/dist/types/thydra-config.js.map +0 -1
- package/dist/types/tparser-ctor.d.ts +0 -7
- package/dist/types/tparser-ctor.js +0 -2
- package/dist/types/tparser-ctor.js.map +0 -1
- package/dist/types/tquery.d.ts +0 -7
- package/dist/types/tquery.js +0 -2
- package/dist/types/tquery.js.map +0 -1
- package/dist/types/tqueue-length.js +0 -2
- package/dist/types/tqueue-length.js.map +0 -1
- package/dist/types/trobots-config.js.map +0 -1
- /package/dist/classes/{dns.d.ts → dns.d.mts} +0 -0
- /package/dist/{cli.d.ts → cli.d.mts} +0 -0
- /package/dist/enums/{eavailable-strategy.d.ts → eavailable-strategy.d.mts} +0 -0
- /package/dist/enums/{elist.d.ts → elist.d.mts} +0 -0
- /package/dist/enums/{eserver.d.ts → eserver.d.mts} +0 -0
- /package/dist/enums/{ex-powered-by.d.ts → ex-powered-by.d.mts} +0 -0
- /package/dist/interfaces/{imatch.d.ts → imatch.d.mts} +0 -0
- /package/dist/interfaces/{iparser-config.d.ts → iparser-config.d.mts} +0 -0
- /package/dist/interfaces/{iserver.d.ts → iserver.d.mts} +0 -0
- /package/dist/types/{tcrawl-config.d.ts → tcrawl-config.d.mts} +0 -0
- /package/dist/types/{thydra-config.d.ts → thydra-config.d.mts} +0 -0
- /package/dist/types/{tqueue-length.d.ts → tqueue-length.d.mts} +0 -0
- /package/dist/types/{trobots-config.d.ts → trobots-config.d.mts} +0 -0
|
@@ -0,0 +1,222 @@
|
|
|
1
|
+
import { commonsStringSplitString, commonsTypeIsTPropertyObjectT } from 'tscommons-esm-core';
|
|
2
|
+
import { commonsLogDebug, commonsLogDoing, commonsLogInfo } from 'nodecommons-esm-log';
|
|
3
|
+
import { commonsExpressBuildDefaultServer, isICommonsExpressConfig } from 'nodecommons-esm-express';
|
|
4
|
+
import { CommonsSocketIoApp } from 'nodecommons-esm-app-socket-io';
|
|
5
|
+
import { AutocompleteApi } from '../apis/autocomplete.api.mjs';
|
|
6
|
+
import { BugsApi } from '../apis/bugs.api.mjs';
|
|
7
|
+
import { CrawlApi } from '../apis/crawl.api.mjs';
|
|
8
|
+
import { DomainsApi } from '../apis/domains.api.mjs';
|
|
9
|
+
import { ImagesApi } from '../apis/images.api.mjs';
|
|
10
|
+
import { StatisticsApi } from '../apis/statistics.api.mjs';
|
|
11
|
+
import { TestApi } from '../apis/test.api.mjs';
|
|
12
|
+
import { UrlsApi } from '../apis/urls.api.mjs';
|
|
13
|
+
import { Lists } from '../classes/lists.mjs';
|
|
14
|
+
import { Expiry } from '../classes/expiry.mjs';
|
|
15
|
+
import { Tracker } from '../classes/tracker.mjs';
|
|
16
|
+
import { Crawler } from '../classes/crawler.mjs';
|
|
17
|
+
import { SocketIoServer } from '../servers/socket-io.server.mjs';
|
|
18
|
+
import { CrawlServer } from '../servers/crawl.server.mjs';
|
|
19
|
+
import { MaintenanceServer } from '../servers/maintenance.server.mjs';
|
|
20
|
+
import { ServerParser } from '../parsers/server.parser.mjs';
|
|
21
|
+
import { HyperlinksParser } from '../parsers/hyperlinks.parser.mjs';
|
|
22
|
+
import { ImageTagsParser } from '../parsers/image-tags.parser.mjs';
|
|
23
|
+
import { JpegParser } from '../parsers/jpeg.parser.mjs';
|
|
24
|
+
import { SpellingParser } from '../parsers/spelling.parser.mjs';
|
|
25
|
+
import { BadWordsParser } from '../parsers/bad-words.parser.mjs';
|
|
26
|
+
import { PhpErrorParser } from '../parsers/php-error.parser.mjs';
|
|
27
|
+
import { AspErrorParser } from '../parsers/asp-error.parser.mjs';
|
|
28
|
+
import { ComplexEnglishParser } from '../parsers/complex-english.parser.mjs';
|
|
29
|
+
import { AccessibilityMetricsParser } from '../parsers/accessibility-metrics.parser.mjs';
|
|
30
|
+
// import { OffenceParser } from '../parsers/offence.parser.mjs';
|
|
31
|
+
// import { InterestParser } from '../parsers/interest.parser.mjs';
|
|
32
|
+
import { LlamaGuardParser } from '../parsers/llama-guard.parser.mjs';
|
|
33
|
+
import { isIParserConfig } from '../interfaces/iparser-config.mjs';
|
|
34
|
+
import { isTHydraConfig } from '../types/thydra-config.mjs';
|
|
35
|
+
import { isTCrawlConfig } from '../types/tcrawl-config.mjs';
|
|
36
|
+
import { isTRobotsConfig } from '../types/trobots-config.mjs';
|
|
37
|
+
export class HydraApp extends CommonsSocketIoApp {
|
|
38
|
+
databaseService;
|
|
39
|
+
lists;
|
|
40
|
+
expiry;
|
|
41
|
+
hydraConfig;
|
|
42
|
+
crawlConfig;
|
|
43
|
+
parsersConfig;
|
|
44
|
+
robotsConfig;
|
|
45
|
+
crawlServer;
|
|
46
|
+
maintenanceTimes = [];
|
|
47
|
+
maintenanceServer;
|
|
48
|
+
tracker;
|
|
49
|
+
constructor() {
|
|
50
|
+
super('hydra-crawler');
|
|
51
|
+
this.lists = new Lists();
|
|
52
|
+
this.expiry = new Expiry();
|
|
53
|
+
const hydraConfig = this.getConfigArea('hydra');
|
|
54
|
+
if (!isTHydraConfig(hydraConfig))
|
|
55
|
+
throw new Error('Hydra config is not valid');
|
|
56
|
+
this.hydraConfig = hydraConfig;
|
|
57
|
+
const crawlConfig = this.getConfigArea('crawl');
|
|
58
|
+
if (!isTCrawlConfig(crawlConfig))
|
|
59
|
+
throw new Error('Crawl config is not valid');
|
|
60
|
+
this.crawlConfig = crawlConfig;
|
|
61
|
+
const parsersConfig = this.getConfigArea('parsers');
|
|
62
|
+
if (!commonsTypeIsTPropertyObjectT(parsersConfig, isIParserConfig))
|
|
63
|
+
throw new Error('Parsers config is not valid');
|
|
64
|
+
this.parsersConfig = parsersConfig;
|
|
65
|
+
const robotsConfig = this.getConfigArea('robots');
|
|
66
|
+
if (!isTRobotsConfig(robotsConfig))
|
|
67
|
+
throw new Error('Robots config is not valid');
|
|
68
|
+
this.robotsConfig = robotsConfig;
|
|
69
|
+
}
|
|
70
|
+
buildHttpServer() {
|
|
71
|
+
const expressConfig = this.getConfigArea('express');
|
|
72
|
+
if (!isICommonsExpressConfig(expressConfig))
|
|
73
|
+
throw new Error('Invalid express config');
|
|
74
|
+
return commonsExpressBuildDefaultServer(this.httpConfig.port, expressConfig.bodyParserLimit);
|
|
75
|
+
}
|
|
76
|
+
buildSocketIoServer(expressServer, expressConfig) {
|
|
77
|
+
return new SocketIoServer(expressServer, expressConfig);
|
|
78
|
+
}
|
|
79
|
+
getAppName() {
|
|
80
|
+
return 'Hydra';
|
|
81
|
+
}
|
|
82
|
+
setDatabaseService(databaseService) {
|
|
83
|
+
this.databaseService = databaseService;
|
|
84
|
+
}
|
|
85
|
+
addToList(list, entries) {
|
|
86
|
+
this.lists.add(list, entries);
|
|
87
|
+
}
|
|
88
|
+
addToExpiry(expiries) {
|
|
89
|
+
this.expiry.add(expiries);
|
|
90
|
+
}
|
|
91
|
+
setMaintenanceSchedule(times) {
|
|
92
|
+
this.maintenanceTimes = times;
|
|
93
|
+
}
|
|
94
|
+
async init() {
|
|
95
|
+
if (!this.databaseService)
|
|
96
|
+
throw new Error('Database service has not been set yet');
|
|
97
|
+
this.install((restServer, path) => {
|
|
98
|
+
if (!this.databaseService)
|
|
99
|
+
throw new Error('Database service has not been defined yet');
|
|
100
|
+
{ // scope
|
|
101
|
+
const log = commonsLogDoing('hydra', 'Installing auto-complete API');
|
|
102
|
+
new AutocompleteApi(restServer, this.databaseService, path);
|
|
103
|
+
log.success();
|
|
104
|
+
}
|
|
105
|
+
{ // scope
|
|
106
|
+
const log = commonsLogDoing('hydra', 'Installing bugs API');
|
|
107
|
+
new BugsApi(restServer, this.databaseService, path);
|
|
108
|
+
log.success();
|
|
109
|
+
}
|
|
110
|
+
{ // scope
|
|
111
|
+
const log = commonsLogDoing('hydra', 'Installing crawl API');
|
|
112
|
+
new CrawlApi(restServer, this.databaseService, path);
|
|
113
|
+
log.success();
|
|
114
|
+
}
|
|
115
|
+
{ // scope
|
|
116
|
+
const log = commonsLogDoing('hydra', 'Installing domains API');
|
|
117
|
+
new DomainsApi(restServer, this.databaseService, path);
|
|
118
|
+
log.success();
|
|
119
|
+
}
|
|
120
|
+
{ // scope
|
|
121
|
+
const log = commonsLogDoing('hydra', 'Installing images API');
|
|
122
|
+
new ImagesApi(restServer, this.databaseService, path);
|
|
123
|
+
log.success();
|
|
124
|
+
}
|
|
125
|
+
{ // scope
|
|
126
|
+
const log = commonsLogDoing('hydra', 'Installing statistics API');
|
|
127
|
+
new StatisticsApi(restServer, this.databaseService, path);
|
|
128
|
+
log.success();
|
|
129
|
+
}
|
|
130
|
+
{ // scope
|
|
131
|
+
const log = commonsLogDoing('hydra', 'Installing test API');
|
|
132
|
+
new TestApi(restServer, path);
|
|
133
|
+
log.success();
|
|
134
|
+
}
|
|
135
|
+
{ // scope
|
|
136
|
+
const log = commonsLogDoing('hydra', 'Installing urls API');
|
|
137
|
+
new UrlsApi(restServer, this.databaseService, path);
|
|
138
|
+
log.success();
|
|
139
|
+
}
|
|
140
|
+
});
|
|
141
|
+
await super.init();
|
|
142
|
+
const socketIoServer = this.socketIoServer;
|
|
143
|
+
if (!socketIoServer)
|
|
144
|
+
throw new Error('SocketIo server has not been set yet');
|
|
145
|
+
this.tracker = new Tracker(this.databaseService, socketIoServer);
|
|
146
|
+
this.crawlServer = new CrawlServer(this.databaseService, this.hydraConfig, this.crawlConfig, this.parsersConfig, this.robotsConfig, this.lists, this.tracker);
|
|
147
|
+
{ // scope
|
|
148
|
+
const log = commonsLogDoing('hydra', 'Connecting to database');
|
|
149
|
+
await this.databaseService.init();
|
|
150
|
+
log.success();
|
|
151
|
+
}
|
|
152
|
+
for (const parser of [
|
|
153
|
+
ServerParser,
|
|
154
|
+
HyperlinksParser,
|
|
155
|
+
ImageTagsParser,
|
|
156
|
+
JpegParser,
|
|
157
|
+
SpellingParser,
|
|
158
|
+
BadWordsParser,
|
|
159
|
+
PhpErrorParser,
|
|
160
|
+
AspErrorParser,
|
|
161
|
+
ComplexEnglishParser,
|
|
162
|
+
AccessibilityMetricsParser,
|
|
163
|
+
// OffenceParser,
|
|
164
|
+
// InterestParser,
|
|
165
|
+
LlamaGuardParser
|
|
166
|
+
]) {
|
|
167
|
+
commonsLogDebug(`Installing parser ${parser.name}`);
|
|
168
|
+
this.crawlServer.addParser(parser);
|
|
169
|
+
await this.databaseService.initParser(parser);
|
|
170
|
+
}
|
|
171
|
+
const maintenanceHardLimit = this.args.getFirstNumberOrUndefined('maintenance-hard-limit');
|
|
172
|
+
if (maintenanceHardLimit)
|
|
173
|
+
commonsLogDebug(`Setting maintenance hard limit to ${maintenanceHardLimit}`);
|
|
174
|
+
this.maintenanceServer = new MaintenanceServer(this.maintenanceTimes, this.expiry, this.lists, this.databaseService, this.crawlServer, maintenanceHardLimit // to prevent running out of heap on very large collections
|
|
175
|
+
);
|
|
176
|
+
}
|
|
177
|
+
listening() {
|
|
178
|
+
void (async () => {
|
|
179
|
+
if (!this.crawlServer)
|
|
180
|
+
throw new Error('No crawl server instantiated! This should not be possible');
|
|
181
|
+
await this.crawlServer.start();
|
|
182
|
+
this.abort();
|
|
183
|
+
})();
|
|
184
|
+
}
|
|
185
|
+
async run() {
|
|
186
|
+
const singleUrl = this.args.getFirstValueOrUndefined('single-url');
|
|
187
|
+
if (singleUrl) {
|
|
188
|
+
if (!this.databaseService)
|
|
189
|
+
throw new Error('Database service has not been set yet');
|
|
190
|
+
if (!this.crawlServer)
|
|
191
|
+
throw new Error('Crawl server has not been set yet');
|
|
192
|
+
const whatwg = new URL(singleUrl);
|
|
193
|
+
const crawler = new Crawler(whatwg.hostname, this.databaseService, this.crawlConfig, this.parsersConfig, this.robotsConfig, this.crawlServer.listParsers(), this.lists);
|
|
194
|
+
commonsLogInfo(`Running single crawl of: ${singleUrl}`);
|
|
195
|
+
const outcome = await crawler.fetch(singleUrl, false);
|
|
196
|
+
commonsLogInfo(`HTTP status code outcome was ${outcome}`);
|
|
197
|
+
return;
|
|
198
|
+
}
|
|
199
|
+
const domainRestrictTo = this.args.getFirstValueOrUndefined('domain-restrict-to');
|
|
200
|
+
if (domainRestrictTo) {
|
|
201
|
+
const domains = commonsStringSplitString(domainRestrictTo, ',', true, true);
|
|
202
|
+
commonsLogDebug(`Restricting crawl domains to: ${domains.join(',')}`);
|
|
203
|
+
if (!this.crawlServer)
|
|
204
|
+
throw new Error('CrawlServer has not been defined! This should not be possible');
|
|
205
|
+
this.crawlServer.restrictTo = domains;
|
|
206
|
+
}
|
|
207
|
+
if (!this.tracker)
|
|
208
|
+
throw new Error('Tracker has not been defined! This should not be possible');
|
|
209
|
+
if (!this.maintenanceServer)
|
|
210
|
+
throw new Error('Maintenance server has not been defined! This should not be possible');
|
|
211
|
+
this.maintenanceServer.start();
|
|
212
|
+
void this.tracker.start();
|
|
213
|
+
await super.run();
|
|
214
|
+
}
|
|
215
|
+
async shutdown() {
|
|
216
|
+
if (!this.crawlServer)
|
|
217
|
+
throw new Error('No crawl server instantiated! This should not be possible');
|
|
218
|
+
await this.crawlServer.shutdown();
|
|
219
|
+
await super.shutdown();
|
|
220
|
+
}
|
|
221
|
+
}
|
|
222
|
+
//# sourceMappingURL=hydra.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"hydra.app.mjs","sourceRoot":"","sources":["../../src/apps/hydra.app.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,wBAAwB,EAAE,6BAA6B,EAAmB,MAAM,oBAAoB,CAAC;AAG9G,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,cAAc,EAAsB,MAAM,qBAAqB,CAAC;AAC3G,OAAO,EAAE,gCAAgC,EAA0D,uBAAuB,EAAE,MAAM,yBAAyB,CAAC;AAG5J,OAAO,EAAE,kBAAkB,EAAE,MAAM,+BAA+B,CAAC;AAEnE,OAAO,EAAE,eAAe,EAAE,MAAM,8BAA8B,CAAC;AAC/D,OAAO,EAAE,OAAO,EAAE,MAAM,sBAAsB,CAAC;AAC/C,OAAO,EAAE,QAAQ,EAAE,MAAM,uBAAuB,CAAC;AACjD,OAAO,EAAE,UAAU,EAAE,MAAM,yBAAyB,CAAC;AACrD,OAAO,EAAE,SAAS,EAAE,MAAM,wBAAwB,CAAC;AACnD,OAAO,EAAE,aAAa,EAAE,MAAM,4BAA4B,CAAC;AAC3D,OAAO,EAAE,OAAO,EAAE,MAAM,sBAAsB,CAAC;AAC/C,OAAO,EAAE,OAAO,EAAE,MAAM,sBAAsB,CAAC;AAE/C,OAAO,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAC7C,OAAO,EAAE,MAAM,EAAE,MAAM,uBAAuB,CAAC;AAC/C,OAAO,EAAE,OAAO,EAAE,MAAM,wBAAwB,CAAC;AACjD,OAAO,EAAE,OAAO,EAAE,MAAM,wBAAwB,CAAC;AAEjD,OAAO,EAAE,cAAc,EAAE,MAAM,iCAAiC,CAAC;AACjE,OAAO,EAAE,WAAW,EAAE,MAAM,6BAA6B,CAAC;AAC1D,OAAO,EAAE,iBAAiB,EAAE,MAAM,mCAAmC,CAAC;AAItE,OAAO,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAC;AAC5D,OAAO,EAAE,gBAAgB,EAAE,MAAM,kCAAkC,CAAC;AACpE,OAAO,EAAE,eAAe,EAAE,MAAM,kCAAkC,CAAC;AACnE,OAAO,EAAE,UAAU,EAAE,MAAM,4BAA4B,CAAC;AACxD,OAAO,EAAE,cAAc,EAAE,MAAM,gCAAgC,CAAC;AAChE,OAAO,EAAE,cAAc,EAAE,MAAM,iCAAiC,CAAC;AACjE,OAAO,EAAE,cAAc,EAAE,MAAM,iCAAiC,CAAC;AACjE,OAAO,EAAE,cAAc,EAAE,MAAM,iCAAiC,CAAC;AACjE,OAAO,EAAE,oBAAoB,EAAE,MAAM,uCAAuC,CAAC;AAC7E,OAAO,EAAE,0BAA0B,EAAE,MAAM,6CAA6C,CAAC;AACzF,iEAAiE;AACjE,mEAAmE;AACnE,OAAO,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAC;AAErE,OAAO,EAAiB,eAAe,EAAE,MAAM,kCAAkC,CAAC;AAIlF,OAAO,EAAgB,cAAc,EAAE,MAAM,4BAA4B,CAAC;AAC1E,OAAO,EAAgB,cAAc,EAAE,MAAM,4BAA4B,CAAC;AAC1E,OAAO,EAAiB,eAAe,EAAE,MAAM,6BAA6B,CAAC;AAU7E,MAAM,OAAO,QAAS,SAAQ,kBAAkC;IACvD,eAAe,CAA4B;IAE3C,KAAK,CAAQ;IACb,MAAM,CAAS;IAEf,WAAW,CAAe;IAC1B,WAAW,CAAe;IAC1B,aAAa,CAAiC;IAC9C,YAAY,CAAgB;IAE5B,WAAW,CAAwB;IAEnC,gBAAgB,GAA2B,EAAE,CAAC;IAC9C,iBAAiB,CAA8B;IAE/C,OAAO,CAAoB;IAEnC;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;QAEvB,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,CAAC,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;QAE3B,MAAM,WAAW,GAAY,IAAI,CAAC,aAAa,CAAC,OAAO,CAAC,CAAC;QACzD,IAAI,CAAC,cAAc,CAAC,WAAW,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,2BAA2B,CAAC,CAAC;QAC/E,IAAI,CAAC,WAAW,GAAG,WAAW,CAAC;QAE/B,MAAM,WAAW,GAAY,IAAI,CAAC,aAAa,CAAC,OAAO,CAAC,CAAC;QACzD,IAAI,CAAC,cAAc,CAAC,WAAW,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,2BAA2B,CAAC,CAAC;QAC/E,IAAI,CAAC,WAAW,GAAG,WAAW,CAAC;QAE/B,MAAM,aAAa,GAAY,IAAI,CAAC,aAAa,CAAC,SAAS,CAAC,CAAC;QAC7D,IAAI,CAAC,6BAA6B,CAAgB,aAAa,EAAE,eAAe,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,6BAA6B,CAAC,CAAC;QAClI,IAAI,CAAC,aAAa,GAAG,aAAa,CAAC;QAEnC,MAAM,YAAY,GAAY,IAAI,CAAC,aAAa,CAAC,QAAQ,CAAC,CAAC;QAC3D,IAAI,CAAC,eAAe,CAAC,YAAY,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC;QAClF,IAAI,CAAC,YAAY,GAAG,YAAY,CAAC;IAClC,CAAC;IAEkB,eAAe;QACjC,MAAM,aAAa,GAAY,IAAI,CAAC,aAAa,CAAC,SAAS,CAAC,CAAC;QAC7D,IAAI,CAAC,uBAAuB,CAAC,aAAa,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,wBAAwB,CAAC,CAAC;QAEvF,OAAO,gCAAgC,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,aAAa,CAAC,eAAe,CAAC,CAAC;IAC9F,CAAC;IAES,mBAAmB,CAC3B,aAAyC,EACzC,aAAoC;QAErC,OAAO,IAAI,cAAc,CACvB,aAAa,EACb,aAAa,CACd,CAAC;IACH,CAAC;IAEM,UAAU;QAChB,OAAO,OAAO,CAAC;IAChB,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEM,WAAW,CAChB,QAAmB;QAEpB,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;IAC3B,CAAC;IAEM,sBAAsB,CAC3B,KAA6B;QAE9B,IAAI,CAAC,gBAAgB,GAAG,KAAK,CAAC;IAC/B,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,IAAI,CAAC,OAAO,CACV,CAAC,UAAgF,EAAE,IAAY,EAAQ,EAAE;YACxG,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,2CAA2C,CAAC,CAAC;YAExF,CAAC,CAAC,QAAQ;gBACT,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,8BAA8B,CAAC,CAAC;gBACzF,IAAI,eAAe,CACjB,UAAU,EACV,IAAI,CAAC,eAAe,EACpB,IAAI,CACL,CAAC;gBACF,GAAG,CAAC,OAAO,EAAE,CAAC;YACf,CAAC;YAED,CAAC,CAAC,QAAQ;gBACT,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,qBAAqB,CAAC,CAAC;gBAChF,IAAI,OAAO,CACT,UAAU,EACV,IAAI,CAAC,eAAe,EACpB,IAAI,CACL,CAAC;gBACF,GAAG,CAAC,OAAO,EAAE,CAAC;YACf,CAAC;YAED,CAAC,CAAC,QAAQ;gBACT,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,sBAAsB,CAAC,CAAC;gBACjF,IAAI,QAAQ,CACV,UAAU,EACV,IAAI,CAAC,eAAe,EACpB,IAAI,CACL,CAAC;gBACF,GAAG,CAAC,OAAO,EAAE,CAAC;YACf,CAAC;YAED,CAAC,CAAC,QAAQ;gBACT,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,wBAAwB,CAAC,CAAC;gBACnF,IAAI,UAAU,CACZ,UAAU,EACV,IAAI,CAAC,eAAe,EACpB,IAAI,CACL,CAAC;gBACF,GAAG,CAAC,OAAO,EAAE,CAAC;YACf,CAAC;YAED,CAAC,CAAC,QAAQ;gBACT,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,uBAAuB,CAAC,CAAC;gBAClF,IAAI,SAAS,CACX,UAAU,EACV,IAAI,CAAC,eAAe,EACpB,IAAI,CACL,CAAC;gBACF,GAAG,CAAC,OAAO,EAAE,CAAC;YACf,CAAC;YAED,CAAC,CAAC,QAAQ;gBACT,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,2BAA2B,CAAC,CAAC;gBACtF,IAAI,aAAa,CACf,UAAU,EACV,IAAI,CAAC,eAAe,EACpB,IAAI,CACL,CAAC;gBACF,GAAG,CAAC,OAAO,EAAE,CAAC;YACf,CAAC;YAED,CAAC,CAAC,QAAQ;gBACT,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,qBAAqB,CAAC,CAAC;gBAChF,IAAI,OAAO,CACT,UAAU,EACV,IAAI,CACL,CAAC;gBACF,GAAG,CAAC,OAAO,EAAE,CAAC;YACf,CAAC;YAED,CAAC,CAAC,QAAQ;gBACT,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,qBAAqB,CAAC,CAAC;gBAChF,IAAI,OAAO,CACT,UAAU,EACV,IAAI,CAAC,eAAe,EACpB,IAAI,CACL,CAAC;gBACF,GAAG,CAAC,OAAO,EAAE,CAAC;YACf,CAAC;QACF,CAAC,CACF,CAAC;QAEF,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;QAEnB,MAAM,cAAc,GAA6B,IAAI,CAAC,cAAc,CAAC;QACrE,IAAI,CAAC,cAAc;YAAE,MAAM,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC;QAE7E,IAAI,CAAC,OAAO,GAAG,IAAI,OAAO,CACxB,IAAI,CAAC,eAAe,EACpB,cAAc,CACf,CAAC;QAEF,IAAI,CAAC,WAAW,GAAG,IAAI,WAAW,CAChC,IAAI,CAAC,eAAe,EACpB,IAAI,CAAC,WAAW,EAChB,IAAI,CAAC,WAAW,EAChB,IAAI,CAAC,aAAa,EAClB,IAAI,CAAC,YAAY,EACjB,IAAI,CAAC,KAAK,EACV,IAAI,CAAC,OAAO,CACb,CAAC;QAEF,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,wBAAwB,CAAC,CAAC;YACnF,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;QAED,KAAK,MAAM,MAAM,IAAI;YACnB,YAAY;YACZ,gBAAgB;YAChB,eAAe;YACf,UAAU;YACV,cAAc;YACd,cAAc;YACd,cAAc;YACd,cAAc;YACd,oBAAoB;YACpB,0BAA0B;YAC1B,iBAAiB;YACjB,kBAAkB;YAClB,gBAAgB;SACjB,EAAE,CAAC;YACH,eAAe,CAAC,qBAAqB,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC;YACpD,IAAI,CAAC,WAAW,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;YACnC,MAAM,IAAI,CAAC,eAAe,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;QAC/C,CAAC;QAED,MAAM,oBAAoB,GAAqB,IAAI,CAAC,IAAI,CAAC,yBAAyB,CAAC,wBAAwB,CAAC,CAAC;QAC7G,IAAI,oBAAoB;YAAE,eAAe,CAAC,qCAAqC,oBAAoB,EAAE,CAAC,CAAC;QAEvG,IAAI,CAAC,iBAAiB,GAAG,IAAI,iBAAiB,CAC5C,IAAI,CAAC,gBAAgB,EACrB,IAAI,CAAC,MAAM,EACX,IAAI,CAAC,KAAK,EACV,IAAI,CAAC,eAAe,EACpB,IAAI,CAAC,WAAW,EAChB,oBAAoB,CAAC,2DAA2D;SACjF,CAAC;IACH,CAAC;IAEkB,SAAS;QAC3B,KAAK,CAAC,KAAK,IAAmB,EAAE;YAC/B,IAAI,CAAC,IAAI,CAAC,WAAW;gBAAE,MAAM,IAAI,KAAK,CAAC,2DAA2D,CAAC,CAAC;YAEpG,MAAM,IAAI,CAAC,WAAW,CAAC,KAAK,EAAE,CAAC;YAE/B,IAAI,CAAC,KAAK,EAAE,CAAC;QACd,CAAC,CAAC,EAAE,CAAC;IACN,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,MAAM,SAAS,GAAqB,IAAI,CAAC,IAAI,CAAC,wBAAwB,CAAC,YAAY,CAAC,CAAC;QACrF,IAAI,SAAS,EAAE,CAAC;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YACpF,IAAI,CAAC,IAAI,CAAC,WAAW;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAE5E,MAAM,MAAM,GAAQ,IAAI,GAAG,CAAC,SAAS,CAAC,CAAC;YAEvC,MAAM,OAAO,GAAY,IAAI,OAAO,CAClC,MAAM,CAAC,QAAQ,EACf,IAAI,CAAC,eAAe,EACpB,IAAI,CAAC,WAAW,EAChB,IAAI,CAAC,aAAa,EAClB,IAAI,CAAC,YAAY,EACjB,IAAI,CAAC,WAAW,CAAC,WAAW,EAAE,EAC9B,IAAI,CAAC,KAAK,CACX,CAAC;YAEF,cAAc,CAAC,4BAA4B,SAAS,EAAE,CAAC,CAAC;YACxD,MAAM,OAAO,GAAW,MAAM,OAAO,CAAC,KAAK,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC;YAC9D,cAAc,CAAC,gCAAgC,OAAO,EAAE,CAAC,CAAC;YAE1D,OAAO;QACR,CAAC;QAED,MAAM,gBAAgB,GAAqB,IAAI,CAAC,IAAI,CAAC,wBAAwB,CAAC,oBAAoB,CAAC,CAAC;QACpG,IAAI,gBAAgB,EAAE,CAAC;YACtB,MAAM,OAAO,GAAa,wBAAwB,CAAC,gBAAgB,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;YACtF,eAAe,CAAC,iCAAiC,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YAEtE,IAAI,CAAC,IAAI,CAAC,WAAW;gBAAE,MAAM,IAAI,KAAK,CAAC,+DAA+D,CAAC,CAAC;YACxG,IAAI,CAAC,WAAW,CAAC,UAAU,GAAG,OAAO,CAAC;QACvC,CAAC;QAED,IAAI,CAAC,IAAI,CAAC,OAAO;YAAE,MAAM,IAAI,KAAK,CAAC,2DAA2D,CAAC,CAAC;QAChG,IAAI,CAAC,IAAI,CAAC,iBAAiB;YAAE,MAAM,IAAI,KAAK,CAAC,sEAAsE,CAAC,CAAC;QAErH,IAAI,CAAC,iBAAiB,CAAC,KAAK,EAAE,CAAC;QAC/B,KAAK,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,CAAC;QAE1B,MAAM,KAAK,CAAC,GAAG,EAAE,CAAC;IACnB,CAAC;IAEkB,KAAK,CAAC,QAAQ;QAChC,IAAI,CAAC,IAAI,CAAC,WAAW;YAAE,MAAM,IAAI,KAAK,CAAC,2DAA2D,CAAC,CAAC;QAEpG,MAAM,IAAI,CAAC,WAAW,CAAC,QAAQ,EAAE,CAAC;QAElC,MAAM,KAAK,CAAC,QAAQ,EAAE,CAAC;IACxB,CAAC;CACD"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { CommonsApp } from 'nodecommons-
|
|
2
|
-
import { DatabaseService } from '../services/database.service';
|
|
3
|
-
import { IInternalHydraCommonDbApp } from './internal-hydra-common.app';
|
|
1
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
2
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
3
|
+
import { IInternalHydraCommonDbApp } from './internal-hydra-common.app.mjs';
|
|
4
4
|
export declare class ImportApp extends CommonsApp implements IInternalHydraCommonDbApp {
|
|
5
5
|
private databaseService;
|
|
6
6
|
constructor();
|
|
@@ -0,0 +1,48 @@
|
|
|
1
|
+
import { commonsInputReadAllStdIn } from 'nodecommons-esm-cli';
|
|
2
|
+
import { commonsLogDoing } from 'nodecommons-esm-log';
|
|
3
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
4
|
+
export class ImportApp extends CommonsApp {
|
|
5
|
+
databaseService;
|
|
6
|
+
constructor() {
|
|
7
|
+
super('hydra-crawler');
|
|
8
|
+
}
|
|
9
|
+
getAppName() {
|
|
10
|
+
return 'Hydra - Import';
|
|
11
|
+
}
|
|
12
|
+
setDatabaseService(databaseService) {
|
|
13
|
+
this.databaseService = databaseService;
|
|
14
|
+
}
|
|
15
|
+
async init() {
|
|
16
|
+
if (!this.databaseService)
|
|
17
|
+
throw new Error('Database service has not been set yet');
|
|
18
|
+
const log = commonsLogDoing('import', 'Connecting to database');
|
|
19
|
+
await this.databaseService.init();
|
|
20
|
+
log.success();
|
|
21
|
+
await super.init();
|
|
22
|
+
}
|
|
23
|
+
async run() {
|
|
24
|
+
if (!this.databaseService)
|
|
25
|
+
throw new Error('Database service has not been set');
|
|
26
|
+
let urls = [];
|
|
27
|
+
{ // scope
|
|
28
|
+
const log = commonsLogDoing('import', 'Loading URLs from stdin');
|
|
29
|
+
urls = await commonsInputReadAllStdIn();
|
|
30
|
+
log.result(urls.length);
|
|
31
|
+
}
|
|
32
|
+
{ // scope
|
|
33
|
+
const log = commonsLogDoing('import', 'Seeding into queue');
|
|
34
|
+
let tally = 0;
|
|
35
|
+
let done = 0;
|
|
36
|
+
for (const url of urls) {
|
|
37
|
+
if ((++tally % 100) === 0)
|
|
38
|
+
log.progress(`${tally}, imported ${done}`);
|
|
39
|
+
const outcome = await this.databaseService.queue(url);
|
|
40
|
+
if (outcome) {
|
|
41
|
+
done++;
|
|
42
|
+
}
|
|
43
|
+
}
|
|
44
|
+
log.result(done);
|
|
45
|
+
}
|
|
46
|
+
}
|
|
47
|
+
}
|
|
48
|
+
//# sourceMappingURL=import.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"import.app.mjs","sourceRoot":"","sources":["../../src/apps/import.app.mts"],"names":[],"mappings":"AAAA,OAAO,EAAE,wBAAwB,EAAE,MAAM,qBAAqB,CAAC;AAC/D,OAAO,EAAE,eAAe,EAAsB,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAMjD,MAAM,OAAO,SAAU,SAAQ,UAAU;IAChC,eAAe,CAA4B;IAEnD;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;IACxB,CAAC;IAEM,UAAU;QAChB,OAAO,gBAAgB,CAAC;IACzB,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,QAAQ,EAAE,wBAAwB,CAAC,CAAC;QACpF,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,IAAI,IAAI,GAAa,EAAE,CAAC;QACxB,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,QAAQ,EAAE,yBAAyB,CAAC,CAAC;YACrF,IAAI,GAAG,MAAM,wBAAwB,EAAE,CAAC;YACxC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACzB,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,QAAQ,EAAE,oBAAoB,CAAC,CAAC;YAChF,IAAI,KAAK,GAAW,CAAC,CAAC;YAAC,IAAI,IAAI,GAAW,CAAC,CAAC;YAC5C,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBACxB,IAAI,CAAC,EAAE,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,cAAc,IAAI,EAAE,CAAC,CAAC;gBAEtE,MAAM,OAAO,GAAY,MAAM,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBAC/D,IAAI,OAAO,EAAE,CAAC;oBACb,IAAI,EAAE,CAAC;gBACR,CAAC;YACF,CAAC;YACD,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QAClB,CAAC;IACF,CAAC;CACD"}
|
|
@@ -1,9 +1,9 @@
|
|
|
1
|
-
import { TCommonsScheduleTime } from 'tscommons-
|
|
2
|
-
import { CommonsConfigFile } from 'nodecommons-
|
|
3
|
-
import { DatabaseService } from '../services/database.service';
|
|
4
|
-
import { IMatch } from '../interfaces/imatch';
|
|
5
|
-
import { IExpiry } from '../interfaces/iexpiry';
|
|
6
|
-
import { EList } from '../enums/elist';
|
|
1
|
+
import { TCommonsScheduleTime } from 'tscommons-esm-async';
|
|
2
|
+
import { CommonsConfigFile } from 'nodecommons-esm-config';
|
|
3
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
4
|
+
import { IMatch } from '../interfaces/imatch.mjs';
|
|
5
|
+
import { IExpiry } from '../interfaces/iexpiry.mjs';
|
|
6
|
+
import { EList } from '../enums/elist.mjs';
|
|
7
7
|
export interface IInternalHydraCommonApp {
|
|
8
8
|
loadConfigFile(file: string): CommonsConfigFile;
|
|
9
9
|
loadRawJsonConfigFile(file: string): unknown;
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"internal-hydra-common.app.mjs","sourceRoot":"","sources":["../../src/apps/internal-hydra-common.app.mts"],"names":[],"mappings":"AAyBA,MAAM,UAAU,2BAA2B,CAAC,IAA6B;IACxE,OAAO,oBAAoB,IAAI,IAAI,CAAC;AACrC,CAAC;AAQD,MAAM,UAAU,6BAA6B,CAAC,IAA6B;IAC1E,IAAI,CAAC,2BAA2B,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IACrD,OAAO,WAAW,IAAI,IAAI,CAAC;AAC5B,CAAC;AAKD,MAAM,UAAU,+BAA+B,CAAC,IAA6B;IAC5E,IAAI,CAAC,2BAA2B,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IACrD,OAAO,aAAa,IAAI,IAAI,CAAC;AAC9B,CAAC;AAKD,MAAM,UAAU,oCAAoC,CAAC,IAA6B;IACjF,IAAI,CAAC,2BAA2B,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IACrD,OAAO,wBAAwB,IAAI,IAAI,CAAC;AACzC,CAAC"}
|
|
@@ -1,5 +1,5 @@
|
|
|
1
|
-
import { CommonsApp } from 'nodecommons-
|
|
2
|
-
import { DatabaseService } from '../services/database.service';
|
|
1
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
2
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
3
3
|
export declare class MoveToArchiveApp extends CommonsApp {
|
|
4
4
|
private databaseService;
|
|
5
5
|
constructor();
|
|
@@ -0,0 +1,31 @@
|
|
|
1
|
+
import { commonsLogDoing, commonsLogInfo } from 'nodecommons-esm-log';
|
|
2
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
3
|
+
export class MoveToArchiveApp extends CommonsApp {
|
|
4
|
+
databaseService;
|
|
5
|
+
constructor() {
|
|
6
|
+
super('hydra-crawler');
|
|
7
|
+
}
|
|
8
|
+
getAppName() {
|
|
9
|
+
return 'Hydra - Move to archive';
|
|
10
|
+
}
|
|
11
|
+
setDatabaseService(databaseService) {
|
|
12
|
+
this.databaseService = databaseService;
|
|
13
|
+
}
|
|
14
|
+
async init() {
|
|
15
|
+
if (!this.databaseService)
|
|
16
|
+
throw new Error('Database service has not been set yet');
|
|
17
|
+
const log = commonsLogDoing('move-to-archive', 'Connecting to database');
|
|
18
|
+
await this.databaseService.init();
|
|
19
|
+
log.success();
|
|
20
|
+
await super.init();
|
|
21
|
+
}
|
|
22
|
+
async run() {
|
|
23
|
+
const batchSize = this.args.getFirstNumber('batch-size');
|
|
24
|
+
const limit = this.args.getFirstNumberOrUndefined('limit');
|
|
25
|
+
if (!this.databaseService)
|
|
26
|
+
throw new Error('Database service has not been set');
|
|
27
|
+
commonsLogInfo('Moving ARCHIVED urls into archives collection.');
|
|
28
|
+
await this.databaseService.moveArchivedUrlsToArchive(batchSize, limit);
|
|
29
|
+
}
|
|
30
|
+
}
|
|
31
|
+
//# sourceMappingURL=move-to-archive.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"move-to-archive.app.mjs","sourceRoot":"","sources":["../../src/apps/move-to-archive.app.mts"],"names":[],"mappings":"AAAA,OAAO,EAAsB,eAAe,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAC;AAC1F,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAIjD,MAAM,OAAO,gBAAiB,SAAQ,UAAU;IACvC,eAAe,CAA4B;IAEnD;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;IACxB,CAAC;IAEM,UAAU;QAChB,OAAO,yBAAyB,CAAC;IAClC,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,iBAAiB,EAAE,wBAAwB,CAAC,CAAC;QAC7F,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,MAAM,SAAS,GAAW,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,YAAY,CAAC,CAAC;QACjE,MAAM,KAAK,GAAqB,IAAI,CAAC,IAAI,CAAC,yBAAyB,CAAC,OAAO,CAAC,CAAC;QAE7E,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,cAAc,CAAC,gDAAgD,CAAC,CAAC;QACjE,MAAM,IAAI,CAAC,eAAe,CAAC,yBAAyB,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC;IACxE,CAAC;CACD"}
|
|
@@ -1,5 +1,5 @@
|
|
|
1
|
-
import { CommonsApp } from 'nodecommons-
|
|
2
|
-
import { DatabaseService } from '../services/database.service';
|
|
1
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
2
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
3
3
|
export declare class PruneArchiveApp extends CommonsApp {
|
|
4
4
|
private databaseService;
|
|
5
5
|
constructor();
|
|
@@ -0,0 +1,40 @@
|
|
|
1
|
+
import { commonsLogAlert, commonsLogDoing, commonsLogInfo } from 'nodecommons-esm-log';
|
|
2
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
3
|
+
export class PruneArchiveApp extends CommonsApp {
|
|
4
|
+
databaseService;
|
|
5
|
+
constructor() {
|
|
6
|
+
super('hydra-crawler');
|
|
7
|
+
}
|
|
8
|
+
getAppName() {
|
|
9
|
+
return 'Hydra - Prune archive';
|
|
10
|
+
}
|
|
11
|
+
setDatabaseService(databaseService) {
|
|
12
|
+
this.databaseService = databaseService;
|
|
13
|
+
}
|
|
14
|
+
async init() {
|
|
15
|
+
if (!this.databaseService)
|
|
16
|
+
throw new Error('Database service has not been set yet');
|
|
17
|
+
const log = commonsLogDoing('prune-archive', 'Connecting to database');
|
|
18
|
+
await this.databaseService.init();
|
|
19
|
+
log.success();
|
|
20
|
+
await super.init();
|
|
21
|
+
}
|
|
22
|
+
async run() {
|
|
23
|
+
if (!this.databaseService)
|
|
24
|
+
throw new Error('Database service has not been set');
|
|
25
|
+
commonsLogInfo('Searching for duplicate entries in archiveds table.');
|
|
26
|
+
const duplicates = await this.databaseService.listDuplicateArchivedUrls();
|
|
27
|
+
if (this.args.hasAttribute('purge')) {
|
|
28
|
+
const log = commonsLogDoing('prune-archive', 'Purging duplicates');
|
|
29
|
+
for (const duplicate of duplicates) {
|
|
30
|
+
log.progress(duplicate.substring(0, 90));
|
|
31
|
+
await this.databaseService.purgeArchiveDuplicates(duplicate);
|
|
32
|
+
}
|
|
33
|
+
log.success();
|
|
34
|
+
}
|
|
35
|
+
else {
|
|
36
|
+
commonsLogAlert('Specify --purge flag to perform the purge');
|
|
37
|
+
}
|
|
38
|
+
}
|
|
39
|
+
}
|
|
40
|
+
//# sourceMappingURL=prune-archive.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"prune-archive.app.mjs","sourceRoot":"","sources":["../../src/apps/prune-archive.app.mts"],"names":[],"mappings":"AAAA,OAAO,EAAE,eAAe,EAAE,eAAe,EAAE,cAAc,EAAsB,MAAM,qBAAqB,CAAC;AAC3G,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAIjD,MAAM,OAAO,eAAgB,SAAQ,UAAU;IACtC,eAAe,CAA4B;IAEnD;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;IACxB,CAAC;IAEM,UAAU;QAChB,OAAO,uBAAuB,CAAC;IAChC,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,eAAe,EAAE,wBAAwB,CAAC,CAAC;QAC3F,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,cAAc,CAAC,qDAAqD,CAAC,CAAC;QACtE,MAAM,UAAU,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,yBAAyB,EAAE,CAAC;QAEpF,IAAI,IAAI,CAAC,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC,EAAE,CAAC;YACrC,MAAM,GAAG,GAAuB,eAAe,CAAC,eAAe,EAAE,oBAAoB,CAAC,CAAC;YAEvF,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;gBACpC,GAAG,CAAC,QAAQ,CAAC,SAAS,CAAC,SAAS,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;gBAEzC,MAAM,IAAI,CAAC,eAAe,CAAC,sBAAsB,CAAC,SAAS,CAAC,CAAC;YAC9D,CAAC;YAED,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;aAAM,CAAC;YACP,eAAe,CAAC,2CAA2C,CAAC,CAAC;QAC9D,CAAC;IACF,CAAC;CACD"}
|
|
@@ -1,9 +1,9 @@
|
|
|
1
|
-
import { CommonsApp } from 'nodecommons-
|
|
2
|
-
import { DatabaseService } from '../services/database.service';
|
|
3
|
-
import { IMatch } from '../interfaces/imatch';
|
|
4
|
-
import { IExpiry } from '../interfaces/iexpiry';
|
|
5
|
-
import { EList } from '../enums/elist';
|
|
6
|
-
import { IInternalHydraCommonDbApp } from './internal-hydra-common.app';
|
|
1
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
2
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
3
|
+
import { IMatch } from '../interfaces/imatch.mjs';
|
|
4
|
+
import { IExpiry } from '../interfaces/iexpiry.mjs';
|
|
5
|
+
import { EList } from '../enums/elist.mjs';
|
|
6
|
+
import { IInternalHydraCommonDbApp } from './internal-hydra-common.app.mjs';
|
|
7
7
|
export declare class QueryApp extends CommonsApp implements IInternalHydraCommonDbApp {
|
|
8
8
|
private queryName;
|
|
9
9
|
private databaseService;
|
|
@@ -0,0 +1,69 @@
|
|
|
1
|
+
import { commonsTypeIsTPropertyObjectT } from 'tscommons-esm-core';
|
|
2
|
+
import { commonsLogDie, commonsLogDoing } from 'nodecommons-esm-log';
|
|
3
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
4
|
+
import { Lists } from '../classes/lists.mjs';
|
|
5
|
+
import { Expiry } from '../classes/expiry.mjs';
|
|
6
|
+
import { isIParserConfig } from '../interfaces/iparser-config.mjs';
|
|
7
|
+
import { QUERY as QUERY_COMPLEX_ENGLISH } from '../queries/complex-english.query.mjs';
|
|
8
|
+
import { QUERY as QUERY_FLASH_CONTENT } from '../queries/flash-content.query.mjs';
|
|
9
|
+
import { QUERY as QUERY_LINKING_TO_DOMAINS } from '../queries/linking-to-domains.query.mjs';
|
|
10
|
+
import { QUERY as QUERY_READABILITY_SCORE } from '../queries/readability-score.query.mjs';
|
|
11
|
+
import { QUERY as QUERY_LLAMAGUARD_UNSAFE_CONTENT } from '../queries/llamaguard-unsafe-content.query.mjs';
|
|
12
|
+
export class QueryApp extends CommonsApp {
|
|
13
|
+
queryName;
|
|
14
|
+
databaseService;
|
|
15
|
+
lists;
|
|
16
|
+
expiry;
|
|
17
|
+
parsersConfig;
|
|
18
|
+
constructor(queryName) {
|
|
19
|
+
super('hydra-crawler');
|
|
20
|
+
this.queryName = queryName;
|
|
21
|
+
if (/[^-a-z0-9]/i.test(queryName))
|
|
22
|
+
throw new Error('Invalid query name');
|
|
23
|
+
this.lists = new Lists();
|
|
24
|
+
this.expiry = new Expiry();
|
|
25
|
+
const parsersConfig = this.getConfigArea('parsers');
|
|
26
|
+
if (!commonsTypeIsTPropertyObjectT(parsersConfig, isIParserConfig))
|
|
27
|
+
throw new Error('Parsers config is not valid');
|
|
28
|
+
this.parsersConfig = parsersConfig;
|
|
29
|
+
}
|
|
30
|
+
getAppName() {
|
|
31
|
+
return 'Hydra - Query';
|
|
32
|
+
}
|
|
33
|
+
setDatabaseService(databaseService) {
|
|
34
|
+
this.databaseService = databaseService;
|
|
35
|
+
}
|
|
36
|
+
addToList(list, entries) {
|
|
37
|
+
this.lists.add(list, entries);
|
|
38
|
+
}
|
|
39
|
+
addToExpiry(expiries) {
|
|
40
|
+
this.expiry.add(expiries);
|
|
41
|
+
}
|
|
42
|
+
async init() {
|
|
43
|
+
if (!this.databaseService)
|
|
44
|
+
throw new Error('Database service has not been set yet');
|
|
45
|
+
const log = commonsLogDoing('query', 'Connecting to database');
|
|
46
|
+
await this.databaseService.init();
|
|
47
|
+
log.success();
|
|
48
|
+
await super.init();
|
|
49
|
+
}
|
|
50
|
+
async run() {
|
|
51
|
+
if (!this.databaseService)
|
|
52
|
+
throw new Error('Database service has not been set');
|
|
53
|
+
let query;
|
|
54
|
+
if (this.queryName === 'complex-english')
|
|
55
|
+
query = QUERY_COMPLEX_ENGLISH;
|
|
56
|
+
if (this.queryName === 'flash-content')
|
|
57
|
+
query = QUERY_FLASH_CONTENT;
|
|
58
|
+
if (this.queryName === 'linking-to-domains')
|
|
59
|
+
query = QUERY_LINKING_TO_DOMAINS;
|
|
60
|
+
if (this.queryName === 'readability-score')
|
|
61
|
+
query = QUERY_READABILITY_SCORE;
|
|
62
|
+
if (this.queryName === 'llamaguard-unsafe-content')
|
|
63
|
+
query = QUERY_LLAMAGUARD_UNSAFE_CONTENT;
|
|
64
|
+
if (!query)
|
|
65
|
+
commonsLogDie('No such available query');
|
|
66
|
+
await query(this.args, this.databaseService, this.lists, this.expiry, this.parsersConfig);
|
|
67
|
+
}
|
|
68
|
+
}
|
|
69
|
+
//# sourceMappingURL=query.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"query.app.mjs","sourceRoot":"","sources":["../../src/apps/query.app.mts"],"names":[],"mappings":"AAAA,OAAO,EAAE,6BAA6B,EAAmB,MAAM,oBAAoB,CAAC;AAEpF,OAAO,EAAE,aAAa,EAAE,eAAe,EAAsB,MAAM,qBAAqB,CAAC;AACzF,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAEjD,OAAO,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAC7C,OAAO,EAAE,MAAM,EAAE,MAAM,uBAAuB,CAAC;AAI/C,OAAO,EAAiB,eAAe,EAAE,MAAM,kCAAkC,CAAC;AAQlF,OAAO,EAAE,KAAK,IAAI,qBAAqB,EAAE,MAAM,sCAAsC,CAAC;AACtF,OAAO,EAAE,KAAK,IAAI,mBAAmB,EAAE,MAAM,oCAAoC,CAAC;AAClF,OAAO,EAAE,KAAK,IAAI,wBAAwB,EAAE,MAAM,yCAAyC,CAAC;AAC5F,OAAO,EAAE,KAAK,IAAI,uBAAuB,EAAE,MAAM,wCAAwC,CAAC;AAC1F,OAAO,EAAE,KAAK,IAAI,+BAA+B,EAAE,MAAM,gDAAgD,CAAC;AAI1G,MAAM,OAAO,QAAS,SAAQ,UAAU;IAS7B;IARF,eAAe,CAA4B;IAE3C,KAAK,CAAQ;IACb,MAAM,CAAS;IAEf,aAAa,CAAiC;IAEtD,YACU,SAAiB;QAE1B,KAAK,CAAC,eAAe,CAAC,CAAC;QAFd,cAAS,GAAT,SAAS,CAAQ;QAI1B,IAAI,aAAa,CAAC,IAAI,CAAC,SAAS,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,oBAAoB,CAAC,CAAC;QAEzE,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,CAAC,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;QAE3B,MAAM,aAAa,GAAY,IAAI,CAAC,aAAa,CAAC,SAAS,CAAC,CAAC;QAC7D,IAAI,CAAC,6BAA6B,CAAgB,aAAa,EAAE,eAAe,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,6BAA6B,CAAC,CAAC;QAClI,IAAI,CAAC,aAAa,GAAG,aAAa,CAAC;IACpC,CAAC;IAEM,UAAU;QAChB,OAAO,eAAe,CAAC;IACxB,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEM,WAAW,CAChB,QAAmB;QAEpB,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;IAC3B,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,OAAO,EAAE,wBAAwB,CAAC,CAAC;QACnF,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,IAAI,KAAuB,CAAC;QAE5B,IAAI,IAAI,CAAC,SAAS,KAAK,iBAAiB;YAAE,KAAK,GAAG,qBAAqB,CAAC;QACxE,IAAI,IAAI,CAAC,SAAS,KAAK,eAAe;YAAE,KAAK,GAAG,mBAAmB,CAAC;QACpE,IAAI,IAAI,CAAC,SAAS,KAAK,oBAAoB;YAAE,KAAK,GAAG,wBAAwB,CAAC;QAC9E,IAAI,IAAI,CAAC,SAAS,KAAK,mBAAmB;YAAE,KAAK,GAAG,uBAAuB,CAAC;QAC5E,IAAI,IAAI,CAAC,SAAS,KAAK,2BAA2B;YAAE,KAAK,GAAG,+BAA+B,CAAC;QAE5F,IAAI,CAAC,KAAK;YAAE,aAAa,CAAC,yBAAyB,CAAC,CAAC;QAErD,MAAM,KAAK,CACT,IAAI,CAAC,IAAI,EACT,IAAI,CAAC,eAAe,EACpB,IAAI,CAAC,KAAK,EACV,IAAI,CAAC,MAAM,EACX,IAAI,CAAC,aAAa,CACnB,CAAC;IACH,CAAC;CACD"}
|
|
@@ -1,9 +1,9 @@
|
|
|
1
|
-
import { EStatus } from 'hydra-crawler-
|
|
2
|
-
import { CommonsApp } from 'nodecommons-
|
|
3
|
-
import { DatabaseService } from '../services/database.service';
|
|
4
|
-
import { IMatch } from '../interfaces/imatch';
|
|
5
|
-
import { EList } from '../enums/elist';
|
|
6
|
-
import { IInternalHydraCommonListApp } from './internal-hydra-common.app';
|
|
1
|
+
import { EStatus } from 'hydra-crawler-esm-assets';
|
|
2
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
3
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
4
|
+
import { IMatch } from '../interfaces/imatch.mjs';
|
|
5
|
+
import { EList } from '../enums/elist.mjs';
|
|
6
|
+
import { IInternalHydraCommonListApp } from './internal-hydra-common.app.mjs';
|
|
7
7
|
export declare class ReattemptApp extends CommonsApp implements IInternalHydraCommonListApp {
|
|
8
8
|
private include;
|
|
9
9
|
private databaseService;
|
|
@@ -0,0 +1,81 @@
|
|
|
1
|
+
import { EStatus } from 'hydra-crawler-esm-assets';
|
|
2
|
+
import { commonsLogDoing } from 'nodecommons-esm-log';
|
|
3
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
4
|
+
import { Lists } from '../classes/lists.mjs';
|
|
5
|
+
import { EList } from '../enums/elist.mjs';
|
|
6
|
+
export class ReattemptApp extends CommonsApp {
|
|
7
|
+
include;
|
|
8
|
+
databaseService;
|
|
9
|
+
lists;
|
|
10
|
+
constructor(include) {
|
|
11
|
+
super('hydra-crawler');
|
|
12
|
+
this.include = include;
|
|
13
|
+
this.lists = new Lists();
|
|
14
|
+
}
|
|
15
|
+
getAppName() {
|
|
16
|
+
return 'Hydra - Reattempt';
|
|
17
|
+
}
|
|
18
|
+
setDatabaseService(databaseService) {
|
|
19
|
+
this.databaseService = databaseService;
|
|
20
|
+
}
|
|
21
|
+
addToList(list, entries) {
|
|
22
|
+
this.lists.add(list, entries);
|
|
23
|
+
}
|
|
24
|
+
async init() {
|
|
25
|
+
if (!this.databaseService)
|
|
26
|
+
throw new Error('Database service has not been set yet');
|
|
27
|
+
const log = commonsLogDoing('reattempt', 'Connecting to database');
|
|
28
|
+
await this.databaseService.init();
|
|
29
|
+
log.success();
|
|
30
|
+
await super.init();
|
|
31
|
+
}
|
|
32
|
+
async run() {
|
|
33
|
+
if (!this.databaseService)
|
|
34
|
+
throw new Error('Database service has not been set');
|
|
35
|
+
const urls = [];
|
|
36
|
+
{ // scope
|
|
37
|
+
let tally = 0;
|
|
38
|
+
const log = commonsLogDoing('reattempt', `Searching for allowlisted ${this.include.join(' and ')}`);
|
|
39
|
+
const results = this.databaseService.getUrls()
|
|
40
|
+
.find({
|
|
41
|
+
status: { $in: this.include }
|
|
42
|
+
}, {});
|
|
43
|
+
tally = 0;
|
|
44
|
+
while (true) {
|
|
45
|
+
if ((tally % 100) === 0)
|
|
46
|
+
log.progress(tally);
|
|
47
|
+
const row = await results.next();
|
|
48
|
+
if (row === null)
|
|
49
|
+
break;
|
|
50
|
+
try {
|
|
51
|
+
if (!this.lists.match(EList.ALLOW, row.url))
|
|
52
|
+
continue;
|
|
53
|
+
}
|
|
54
|
+
catch (_e) {
|
|
55
|
+
// probably an invalid URL
|
|
56
|
+
// so ignore
|
|
57
|
+
continue;
|
|
58
|
+
}
|
|
59
|
+
urls.push(row.url);
|
|
60
|
+
tally++;
|
|
61
|
+
}
|
|
62
|
+
log.result(tally);
|
|
63
|
+
}
|
|
64
|
+
{ // scope
|
|
65
|
+
const log = commonsLogDoing('reattempt', 'Requeuing matches');
|
|
66
|
+
let tally = 0;
|
|
67
|
+
for (const url of urls) {
|
|
68
|
+
tally++;
|
|
69
|
+
if ((tally % 10) === 0)
|
|
70
|
+
log.progress(tally);
|
|
71
|
+
await this.databaseService.getUrls()
|
|
72
|
+
.updateOne({ url: url }, {
|
|
73
|
+
$set: { status: EStatus.QUEUED },
|
|
74
|
+
$unset: { ttl: 1 }
|
|
75
|
+
});
|
|
76
|
+
}
|
|
77
|
+
log.result(tally);
|
|
78
|
+
}
|
|
79
|
+
}
|
|
80
|
+
}
|
|
81
|
+
//# sourceMappingURL=reattempt.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"reattempt.app.mjs","sourceRoot":"","sources":["../../src/apps/reattempt.app.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAGnD,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAEjD,OAAO,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAM7C,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAC;AAI3C,MAAM,OAAO,YAAa,SAAQ,UAAU;IAMjC;IALF,eAAe,CAA4B;IAE3C,KAAK,CAAQ;IAErB,YACU,OAAkB;QAE3B,KAAK,CAAC,eAAe,CAAC,CAAC;QAFd,YAAO,GAAP,OAAO,CAAW;QAI3B,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;IAC1B,CAAC;IAEM,UAAU;QAChB,OAAO,mBAAmB,CAAC;IAC5B,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,WAAW,EAAE,wBAAwB,CAAC,CAAC;QACvF,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,MAAM,IAAI,GAAa,EAAE,CAAC;QAC1B,CAAC,CAAC,QAAQ;YACT,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,MAAM,GAAG,GAAuB,eAAe,CAAC,WAAW,EAAE,6BAA6B,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;YAExH,MAAM,OAAO,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBACrE,IAAI,CACH;gBACE,MAAM,EAAE,EAAE,GAAG,EAAE,IAAI,CAAC,OAAO,EAAE;aAC9B,EACD,EAAE,CACH,CAAC;YAEJ,KAAK,GAAG,CAAC,CAAC;YAEV,OAAO,IAAI,EAAE,CAAC;gBACb,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;gBAE7C,MAAM,GAAG,GAAc,MAAM,OAAO,CAAC,IAAI,EAAE,CAAC;gBAC5C,IAAI,GAAG,KAAK,IAAI;oBAAE,MAAM;gBAExB,IAAI,CAAC;oBACJ,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,GAAG,CAAC;wBAAE,SAAS;gBACvD,CAAC;gBAAC,OAAO,EAAE,EAAE,CAAC;oBACb,0BAA0B;oBAC1B,YAAY;oBACZ,SAAS;gBACV,CAAC;gBAED,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;gBAEnB,KAAK,EAAE,CAAC;YACT,CAAC;YACD,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACnB,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,WAAW,EAAE,mBAAmB,CAAC,CAAC;YAClF,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBACxB,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,KAAK,CAAC;oBAAE,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;gBAE5C,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;qBACjC,SAAS,CAAC,EAAE,GAAG,EAAE,GAAG,EAAE,EAAE;oBACvB,IAAI,EAAE,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE;oBAChC,MAAM,EAAE,EAAE,GAAG,EAAE,CAAC,EAAE;iBACnB,CAAC,CAAC;YACN,CAAC;YACD,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACnB,CAAC;IACF,CAAC;CACD"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { CommonsApp } from 'nodecommons-
|
|
2
|
-
import { DatabaseService } from '../services/database.service';
|
|
3
|
-
import { IInternalHydraCommonDbApp } from './internal-hydra-common.app';
|
|
1
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
2
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
3
|
+
import { IInternalHydraCommonDbApp } from './internal-hydra-common.app.mjs';
|
|
4
4
|
export declare class RequeueDomainApp extends CommonsApp implements IInternalHydraCommonDbApp {
|
|
5
5
|
private domain;
|
|
6
6
|
private done200Only;
|