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.
- 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 +122 -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 +74 -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 +71 -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 +257 -0
- package/dist/classes/cleaner.mjs.map +1 -0
- package/dist/classes/{crawler.d.ts → crawler.d.mts} +10 -12
- package/dist/classes/crawler.mjs +418 -0
- 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} +18 -27
- 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 -141
- 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 +0 -422
- 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
package/dist/apps/import.app.js
DELETED
|
@@ -1,57 +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 { commonsInputReadAllStdIn, commonsOutputDoing, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
|
|
11
|
-
import { CommonsApp } from 'nodecommons-es-app';
|
|
12
|
-
export class ImportApp extends CommonsApp {
|
|
13
|
-
constructor() {
|
|
14
|
-
super('hydra-crawler');
|
|
15
|
-
}
|
|
16
|
-
getAppName() {
|
|
17
|
-
return 'Hydra - Import';
|
|
18
|
-
}
|
|
19
|
-
setDatabaseService(databaseService) {
|
|
20
|
-
this.databaseService = databaseService;
|
|
21
|
-
}
|
|
22
|
-
init() {
|
|
23
|
-
const _super = Object.create(null, {
|
|
24
|
-
init: { get: () => super.init }
|
|
25
|
-
});
|
|
26
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
27
|
-
if (!this.databaseService)
|
|
28
|
-
throw new Error('Database service has not been set yet');
|
|
29
|
-
commonsOutputDoing('Connecting to database');
|
|
30
|
-
yield this.databaseService.init();
|
|
31
|
-
commonsOutputSuccess();
|
|
32
|
-
yield _super.init.call(this);
|
|
33
|
-
});
|
|
34
|
-
}
|
|
35
|
-
run() {
|
|
36
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
37
|
-
if (!this.databaseService)
|
|
38
|
-
throw new Error('Database service has not been set');
|
|
39
|
-
commonsOutputDoing('Loading URLs from stdin');
|
|
40
|
-
const urls = yield commonsInputReadAllStdIn();
|
|
41
|
-
commonsOutputResult(urls.length);
|
|
42
|
-
commonsOutputDoing('Seeding into queue');
|
|
43
|
-
let tally = 0;
|
|
44
|
-
let done = 0;
|
|
45
|
-
for (const url of urls) {
|
|
46
|
-
if ((++tally % 100) === 0)
|
|
47
|
-
commonsOutputProgress(`${tally}, imported ${done}`);
|
|
48
|
-
const outcome = yield this.databaseService.queue(url);
|
|
49
|
-
if (outcome) {
|
|
50
|
-
done++;
|
|
51
|
-
}
|
|
52
|
-
}
|
|
53
|
-
commonsOutputResult(done);
|
|
54
|
-
});
|
|
55
|
-
}
|
|
56
|
-
}
|
|
57
|
-
//# sourceMappingURL=import.app.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"import.app.js","sourceRoot":"","sources":["../../src/apps/import.app.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAE,wBAAwB,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AACpJ,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAMhD,MAAM,OAAO,SAAU,SAAQ,UAAU;IAGxC;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;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,kBAAkB,CAAC,yBAAyB,CAAC,CAAC;YAC9C,MAAM,IAAI,GAAa,MAAM,wBAAwB,EAAE,CAAC;YACxD,mBAAmB,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YAEjC,kBAAkB,CAAC,oBAAoB,CAAC,CAAC;YACzC,IAAI,KAAK,GAAW,CAAC,CAAC;YAAC,IAAI,IAAI,GAAW,CAAC,CAAC;YAC5C,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE;gBACvB,IAAI,CAAC,EAAE,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,GAAG,KAAK,cAAc,IAAI,EAAE,CAAC,CAAC;gBAE/E,MAAM,OAAO,GAAY,MAAM,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBAC/D,IAAI,OAAO,EAAE;oBACZ,IAAI,EAAE,CAAC;iBACP;aACD;YACD,mBAAmB,CAAC,IAAI,CAAC,CAAC;QAC3B,CAAC;KAAA;CACD"}
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"internal-hydra-common.app.js","sourceRoot":"","sources":["../../src/apps/internal-hydra-common.app.ts"],"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"}
|
|
@@ -1,46 +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 { commonsOutputDoing, commonsOutputInfo, commonsOutputSuccess } from 'nodecommons-es-cli';
|
|
11
|
-
import { CommonsApp } from 'nodecommons-es-app';
|
|
12
|
-
export class MoveToArchiveApp extends CommonsApp {
|
|
13
|
-
constructor() {
|
|
14
|
-
super('hydra-crawler');
|
|
15
|
-
}
|
|
16
|
-
getAppName() {
|
|
17
|
-
return 'Hydra - Move to archive';
|
|
18
|
-
}
|
|
19
|
-
setDatabaseService(databaseService) {
|
|
20
|
-
this.databaseService = databaseService;
|
|
21
|
-
}
|
|
22
|
-
init() {
|
|
23
|
-
const _super = Object.create(null, {
|
|
24
|
-
init: { get: () => super.init }
|
|
25
|
-
});
|
|
26
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
27
|
-
if (!this.databaseService)
|
|
28
|
-
throw new Error('Database service has not been set yet');
|
|
29
|
-
commonsOutputDoing('Connecting to database');
|
|
30
|
-
yield this.databaseService.init();
|
|
31
|
-
commonsOutputSuccess();
|
|
32
|
-
yield _super.init.call(this);
|
|
33
|
-
});
|
|
34
|
-
}
|
|
35
|
-
run() {
|
|
36
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
37
|
-
const batchSize = this.getArgs().getNumber('batch-size');
|
|
38
|
-
const limit = this.getArgs().getNumberOrUndefined('limit');
|
|
39
|
-
if (!this.databaseService)
|
|
40
|
-
throw new Error('Database service has not been set');
|
|
41
|
-
commonsOutputInfo('Moving ARCHIVED urls into archives collection.');
|
|
42
|
-
yield this.databaseService.moveArchivedUrlsToArchive(batchSize, limit);
|
|
43
|
-
});
|
|
44
|
-
}
|
|
45
|
-
}
|
|
46
|
-
//# sourceMappingURL=move-to-archive.app.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"move-to-archive.app.js","sourceRoot":"","sources":["../../src/apps/move-to-archive.app.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAE,kBAAkB,EAAE,iBAAiB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AACjG,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAIhD,MAAM,OAAO,gBAAiB,SAAQ,UAAU;IAG/C;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;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEY,GAAG;;YACf,MAAM,SAAS,GAAW,IAAI,CAAC,OAAO,EAAE,CAAC,SAAS,CAAC,YAAY,CAAC,CAAC;YACjE,MAAM,KAAK,GAAqB,IAAI,CAAC,OAAO,EAAE,CAAC,oBAAoB,CAAC,OAAO,CAAC,CAAC;YAE7E,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,iBAAiB,CAAC,gDAAgD,CAAC,CAAC;YACpE,MAAM,IAAI,CAAC,eAAe,CAAC,yBAAyB,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC;QACxE,CAAC;KAAA;CACD"}
|
|
@@ -1,55 +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 { commonsOutputAlert, commonsOutputDoing, commonsOutputInfo, commonsOutputProgress, commonsOutputSuccess } from 'nodecommons-es-cli';
|
|
11
|
-
import { CommonsApp } from 'nodecommons-es-app';
|
|
12
|
-
export class PruneArchiveApp extends CommonsApp {
|
|
13
|
-
constructor() {
|
|
14
|
-
super('hydra-crawler');
|
|
15
|
-
}
|
|
16
|
-
getAppName() {
|
|
17
|
-
return 'Hydra - Prune archive';
|
|
18
|
-
}
|
|
19
|
-
setDatabaseService(databaseService) {
|
|
20
|
-
this.databaseService = databaseService;
|
|
21
|
-
}
|
|
22
|
-
init() {
|
|
23
|
-
const _super = Object.create(null, {
|
|
24
|
-
init: { get: () => super.init }
|
|
25
|
-
});
|
|
26
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
27
|
-
if (!this.databaseService)
|
|
28
|
-
throw new Error('Database service has not been set yet');
|
|
29
|
-
commonsOutputDoing('Connecting to database');
|
|
30
|
-
yield this.databaseService.init();
|
|
31
|
-
commonsOutputSuccess();
|
|
32
|
-
yield _super.init.call(this);
|
|
33
|
-
});
|
|
34
|
-
}
|
|
35
|
-
run() {
|
|
36
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
37
|
-
if (!this.databaseService)
|
|
38
|
-
throw new Error('Database service has not been set');
|
|
39
|
-
commonsOutputInfo('Searching for duplicate entries in archiveds table.');
|
|
40
|
-
const duplicates = yield this.databaseService.listDuplicateArchivedUrls();
|
|
41
|
-
if (this.getArgs().hasAttribute('purge')) {
|
|
42
|
-
commonsOutputDoing('Purging duplicates');
|
|
43
|
-
for (const duplicate of duplicates) {
|
|
44
|
-
commonsOutputProgress(duplicate.substring(0, 90));
|
|
45
|
-
yield this.databaseService.purgeArchiveDuplicates(duplicate);
|
|
46
|
-
}
|
|
47
|
-
commonsOutputSuccess();
|
|
48
|
-
}
|
|
49
|
-
else {
|
|
50
|
-
commonsOutputAlert('Specify --purge flag to perform the purge');
|
|
51
|
-
}
|
|
52
|
-
});
|
|
53
|
-
}
|
|
54
|
-
}
|
|
55
|
-
//# sourceMappingURL=prune-archive.app.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"prune-archive.app.js","sourceRoot":"","sources":["../../src/apps/prune-archive.app.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,iBAAiB,EAAE,qBAAqB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAC5I,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAIhD,MAAM,OAAO,eAAgB,SAAQ,UAAU;IAG9C;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;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,iBAAiB,CAAC,qDAAqD,CAAC,CAAC;YACzE,MAAM,UAAU,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,yBAAyB,EAAE,CAAC;YAEpF,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC,YAAY,CAAC,OAAO,CAAC,EAAE;gBACzC,kBAAkB,CAAC,oBAAoB,CAAC,CAAC;gBAEzC,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE;oBACnC,qBAAqB,CAAC,SAAS,CAAC,SAAS,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;oBAElD,MAAM,IAAI,CAAC,eAAe,CAAC,sBAAsB,CAAC,SAAS,CAAC,CAAC;iBAC7D;gBAED,oBAAoB,EAAE,CAAC;aACvB;iBAAM;gBACN,kBAAkB,CAAC,2CAA2C,CAAC,CAAC;aAChE;QACF,CAAC;KAAA;CACD"}
|
package/dist/apps/query.app.js
DELETED
|
@@ -1,80 +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 { commonsTypeIsTKeyObject } from 'tscommons-es-core';
|
|
11
|
-
import { commonsOutputDie, commonsOutputDoing, commonsOutputSuccess } from 'nodecommons-es-cli';
|
|
12
|
-
import { CommonsApp } from 'nodecommons-es-app';
|
|
13
|
-
import { Lists } from '../classes/lists';
|
|
14
|
-
import { Expiry } from '../classes/expiry';
|
|
15
|
-
import { isIParserConfig } from '../interfaces/iparser-config';
|
|
16
|
-
import { QUERY as QUERY_COMPLEX_ENGLISH } from '../queries/complex-english.query';
|
|
17
|
-
import { QUERY as QUERY_FLASH_CONTENT } from '../queries/flash-content.query';
|
|
18
|
-
import { QUERY as QUERY_LINKING_TO_DOMAINS } from '../queries/linking-to-domains.query';
|
|
19
|
-
import { QUERY as QUERY_READABILITY_SCORE } from '../queries/readability-score.query';
|
|
20
|
-
import { QUERY as QUERY_LLAMAGUARD_UNSAFE_CONTENT } from '../queries/llamaguard-unsafe-content.query';
|
|
21
|
-
export class QueryApp extends CommonsApp {
|
|
22
|
-
constructor(queryName) {
|
|
23
|
-
super('hydra-crawler');
|
|
24
|
-
this.queryName = queryName;
|
|
25
|
-
if (/[^-a-z0-9]/i.test(queryName))
|
|
26
|
-
throw new Error('Invalid query name');
|
|
27
|
-
this.lists = new Lists();
|
|
28
|
-
this.expiry = new Expiry();
|
|
29
|
-
const parsersConfig = this.getConfigArea('parsers');
|
|
30
|
-
if (!commonsTypeIsTKeyObject(parsersConfig, isIParserConfig))
|
|
31
|
-
throw new Error('Parsers config is not valid');
|
|
32
|
-
this.parsersConfig = parsersConfig;
|
|
33
|
-
}
|
|
34
|
-
getAppName() {
|
|
35
|
-
return 'Hydra - Query';
|
|
36
|
-
}
|
|
37
|
-
setDatabaseService(databaseService) {
|
|
38
|
-
this.databaseService = databaseService;
|
|
39
|
-
}
|
|
40
|
-
addToList(list, entries) {
|
|
41
|
-
this.lists.add(list, entries);
|
|
42
|
-
}
|
|
43
|
-
addToExpiry(expiries) {
|
|
44
|
-
this.expiry.add(expiries);
|
|
45
|
-
}
|
|
46
|
-
init() {
|
|
47
|
-
const _super = Object.create(null, {
|
|
48
|
-
init: { get: () => super.init }
|
|
49
|
-
});
|
|
50
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
51
|
-
if (!this.databaseService)
|
|
52
|
-
throw new Error('Database service has not been set yet');
|
|
53
|
-
commonsOutputDoing('Connecting to database');
|
|
54
|
-
yield this.databaseService.init();
|
|
55
|
-
commonsOutputSuccess();
|
|
56
|
-
yield _super.init.call(this);
|
|
57
|
-
});
|
|
58
|
-
}
|
|
59
|
-
run() {
|
|
60
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
61
|
-
if (!this.databaseService)
|
|
62
|
-
throw new Error('Database service has not been set');
|
|
63
|
-
let query;
|
|
64
|
-
if (this.queryName === 'complex-english')
|
|
65
|
-
query = QUERY_COMPLEX_ENGLISH;
|
|
66
|
-
if (this.queryName === 'flash-content')
|
|
67
|
-
query = QUERY_FLASH_CONTENT;
|
|
68
|
-
if (this.queryName === 'linking-to-domains')
|
|
69
|
-
query = QUERY_LINKING_TO_DOMAINS;
|
|
70
|
-
if (this.queryName === 'readability-score')
|
|
71
|
-
query = QUERY_READABILITY_SCORE;
|
|
72
|
-
if (this.queryName === 'llamaguard-unsafe-content')
|
|
73
|
-
query = QUERY_LLAMAGUARD_UNSAFE_CONTENT;
|
|
74
|
-
if (!query)
|
|
75
|
-
commonsOutputDie('No such available query');
|
|
76
|
-
yield query(this.getArgs(), this.databaseService, this.lists, this.expiry, this.parsersConfig);
|
|
77
|
-
});
|
|
78
|
-
}
|
|
79
|
-
}
|
|
80
|
-
//# sourceMappingURL=query.app.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"query.app.js","sourceRoot":"","sources":["../../src/apps/query.app.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAE,uBAAuB,EAAc,MAAM,mBAAmB,CAAC;AAExE,OAAO,EAAE,gBAAgB,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAChG,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAEhD,OAAO,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAC;AACzC,OAAO,EAAE,MAAM,EAAE,MAAM,mBAAmB,CAAC;AAI3C,OAAO,EAAiB,eAAe,EAAE,MAAM,8BAA8B,CAAC;AAQ9E,OAAO,EAAE,KAAK,IAAI,qBAAqB,EAAE,MAAM,kCAAkC,CAAC;AAClF,OAAO,EAAE,KAAK,IAAI,mBAAmB,EAAE,MAAM,gCAAgC,CAAC;AAC9E,OAAO,EAAE,KAAK,IAAI,wBAAwB,EAAE,MAAM,qCAAqC,CAAC;AACxF,OAAO,EAAE,KAAK,IAAI,uBAAuB,EAAE,MAAM,oCAAoC,CAAC;AACtF,OAAO,EAAE,KAAK,IAAI,+BAA+B,EAAE,MAAM,4CAA4C,CAAC;AAItG,MAAM,OAAO,QAAS,SAAQ,UAAU;IAQvC,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,uBAAuB,CAAgB,aAAa,EAAE,eAAe,CAAC;YAAE,MAAM,IAAI,KAAK,CAAC,6BAA6B,CAAC,CAAC;QAC5H,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;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,IAAI,KAAuB,CAAC;YAE5B,IAAI,IAAI,CAAC,SAAS,KAAK,iBAAiB;gBAAE,KAAK,GAAG,qBAAqB,CAAC;YACxE,IAAI,IAAI,CAAC,SAAS,KAAK,eAAe;gBAAE,KAAK,GAAG,mBAAmB,CAAC;YACpE,IAAI,IAAI,CAAC,SAAS,KAAK,oBAAoB;gBAAE,KAAK,GAAG,wBAAwB,CAAC;YAC9E,IAAI,IAAI,CAAC,SAAS,KAAK,mBAAmB;gBAAE,KAAK,GAAG,uBAAuB,CAAC;YAC5E,IAAI,IAAI,CAAC,SAAS,KAAK,2BAA2B;gBAAE,KAAK,GAAG,+BAA+B,CAAC;YAE5F,IAAI,CAAC,KAAK;gBAAE,gBAAgB,CAAC,yBAAyB,CAAC,CAAC;YAExD,MAAM,KAAK,CACT,IAAI,CAAC,OAAO,EAAE,EACd,IAAI,CAAC,eAAe,EACpB,IAAI,CAAC,KAAK,EACV,IAAI,CAAC,MAAM,EACX,IAAI,CAAC,aAAa,CACnB,CAAC;QACH,CAAC;KAAA;CACD"}
|
|
@@ -1,83 +0,0 @@
|
|
|
1
|
-
var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
|
|
2
|
-
function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
|
|
3
|
-
return new (P || (P = Promise))(function (resolve, reject) {
|
|
4
|
-
function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
|
|
5
|
-
function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
|
|
6
|
-
function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
|
|
7
|
-
step((generator = generator.apply(thisArg, _arguments || [])).next());
|
|
8
|
-
});
|
|
9
|
-
};
|
|
10
|
-
import { EStatus } from 'hydra-crawler-ts-assets';
|
|
11
|
-
import { commonsOutputDoing, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
|
|
12
|
-
import { CommonsApp } from 'nodecommons-es-app';
|
|
13
|
-
import { Lists } from '../classes/lists';
|
|
14
|
-
import { EList } from '../enums/elist';
|
|
15
|
-
export class ReattemptApp extends CommonsApp {
|
|
16
|
-
constructor(include) {
|
|
17
|
-
super('hydra-crawler');
|
|
18
|
-
this.include = include;
|
|
19
|
-
this.lists = new Lists();
|
|
20
|
-
}
|
|
21
|
-
getAppName() {
|
|
22
|
-
return 'Hydra - Reattempt';
|
|
23
|
-
}
|
|
24
|
-
setDatabaseService(databaseService) {
|
|
25
|
-
this.databaseService = databaseService;
|
|
26
|
-
}
|
|
27
|
-
addToList(list, entries) {
|
|
28
|
-
this.lists.add(list, entries);
|
|
29
|
-
}
|
|
30
|
-
init() {
|
|
31
|
-
const _super = Object.create(null, {
|
|
32
|
-
init: { get: () => super.init }
|
|
33
|
-
});
|
|
34
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
35
|
-
if (!this.databaseService)
|
|
36
|
-
throw new Error('Database service has not been set yet');
|
|
37
|
-
commonsOutputDoing('Connecting to database');
|
|
38
|
-
yield this.databaseService.init();
|
|
39
|
-
commonsOutputSuccess();
|
|
40
|
-
yield _super.init.call(this);
|
|
41
|
-
});
|
|
42
|
-
}
|
|
43
|
-
run() {
|
|
44
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
45
|
-
if (!this.databaseService)
|
|
46
|
-
throw new Error('Database service has not been set');
|
|
47
|
-
let tally = 0;
|
|
48
|
-
commonsOutputDoing(`Searching for allowlisted ${this.include.join(' and ')}`);
|
|
49
|
-
const results = this.databaseService.getUrls()
|
|
50
|
-
.find({
|
|
51
|
-
status: { $in: this.include }
|
|
52
|
-
}, {});
|
|
53
|
-
const urls = [];
|
|
54
|
-
tally = 0;
|
|
55
|
-
while (true) {
|
|
56
|
-
if ((tally % 100) === 0)
|
|
57
|
-
commonsOutputProgress(tally);
|
|
58
|
-
const row = yield results.next();
|
|
59
|
-
if (row === null)
|
|
60
|
-
break;
|
|
61
|
-
if (!this.lists.match(EList.ALLOW, row.url))
|
|
62
|
-
continue;
|
|
63
|
-
urls.push(row.url);
|
|
64
|
-
tally++;
|
|
65
|
-
}
|
|
66
|
-
commonsOutputResult(tally);
|
|
67
|
-
commonsOutputDoing('Requeuing matches');
|
|
68
|
-
tally = 0;
|
|
69
|
-
for (const url of urls) {
|
|
70
|
-
tally++;
|
|
71
|
-
if ((tally % 10) === 0)
|
|
72
|
-
commonsOutputProgress(tally);
|
|
73
|
-
yield this.databaseService.getUrls()
|
|
74
|
-
.updateOne({ url: url }, {
|
|
75
|
-
$set: { status: EStatus.QUEUED },
|
|
76
|
-
$unset: { ttl: 1 }
|
|
77
|
-
});
|
|
78
|
-
}
|
|
79
|
-
commonsOutputResult(tally);
|
|
80
|
-
});
|
|
81
|
-
}
|
|
82
|
-
}
|
|
83
|
-
//# sourceMappingURL=reattempt.app.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"reattempt.app.js","sourceRoot":"","sources":["../../src/apps/reattempt.app.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,OAAO,EAAE,MAAM,yBAAyB,CAAC;AAGlD,OAAO,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAC1H,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAEhD,OAAO,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAC;AAMzC,OAAO,EAAE,KAAK,EAAE,MAAM,gBAAgB,CAAC;AAIvC,MAAM,OAAO,YAAa,SAAQ,UAAU;IAK3C,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;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,kBAAkB,CAAC,6BAA6B,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;YAE9E,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,MAAM,IAAI,GAAa,EAAE,CAAC;YAC1B,KAAK,GAAG,CAAC,CAAC;YAEV,OAAO,IAAI,EAAE;gBACZ,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,KAAK,CAAC,CAAC;gBAEtD,MAAM,GAAG,GAAc,MAAM,OAAO,CAAC,IAAI,EAAE,CAAC;gBAC5C,IAAI,GAAG,KAAK,IAAI;oBAAE,MAAM;gBAExB,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,GAAG,CAAC;oBAAE,SAAS;gBAEtD,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;gBAEnB,KAAK,EAAE,CAAC;aACR;YACD,mBAAmB,CAAC,KAAK,CAAC,CAAC;YAE3B,kBAAkB,CAAC,mBAAmB,CAAC,CAAC;YACxC,KAAK,GAAG,CAAC,CAAC;YACV,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE;gBACvB,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,EAAE,CAAC,KAAK,CAAC;oBAAE,qBAAqB,CAAC,KAAK,CAAC,CAAC;gBAErD,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;aACL;YACD,mBAAmB,CAAC,KAAK,CAAC,CAAC;QAC5B,CAAC;KAAA;CACD"}
|
|
@@ -1,64 +0,0 @@
|
|
|
1
|
-
var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
|
|
2
|
-
function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
|
|
3
|
-
return new (P || (P = Promise))(function (resolve, reject) {
|
|
4
|
-
function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
|
|
5
|
-
function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
|
|
6
|
-
function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
|
|
7
|
-
step((generator = generator.apply(thisArg, _arguments || [])).next());
|
|
8
|
-
});
|
|
9
|
-
};
|
|
10
|
-
import { EStatus } from 'hydra-crawler-ts-assets';
|
|
11
|
-
import { commonsOutputDoing, commonsOutputSuccess } from 'nodecommons-es-cli';
|
|
12
|
-
import { CommonsApp } from 'nodecommons-es-app';
|
|
13
|
-
export class RequeueDomainApp extends CommonsApp {
|
|
14
|
-
constructor(domain, done200Only) {
|
|
15
|
-
super('hydra-crawler');
|
|
16
|
-
this.domain = domain;
|
|
17
|
-
this.done200Only = done200Only;
|
|
18
|
-
}
|
|
19
|
-
getAppName() {
|
|
20
|
-
return 'Hydra - Requeue domain';
|
|
21
|
-
}
|
|
22
|
-
setDatabaseService(databaseService) {
|
|
23
|
-
this.databaseService = databaseService;
|
|
24
|
-
}
|
|
25
|
-
init() {
|
|
26
|
-
const _super = Object.create(null, {
|
|
27
|
-
init: { get: () => super.init }
|
|
28
|
-
});
|
|
29
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
30
|
-
if (!this.databaseService)
|
|
31
|
-
throw new Error('Database service has not been set yet');
|
|
32
|
-
commonsOutputDoing('Connecting to database');
|
|
33
|
-
yield this.databaseService.init();
|
|
34
|
-
commonsOutputSuccess();
|
|
35
|
-
yield _super.init.call(this);
|
|
36
|
-
});
|
|
37
|
-
}
|
|
38
|
-
run() {
|
|
39
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
40
|
-
if (!this.databaseService)
|
|
41
|
-
throw new Error('Database service has not been set');
|
|
42
|
-
const conditions = {
|
|
43
|
-
domain: this.domain,
|
|
44
|
-
status: { $in: [] }
|
|
45
|
-
};
|
|
46
|
-
if (this.done200Only) {
|
|
47
|
-
conditions.status.$in = [EStatus.DONE];
|
|
48
|
-
conditions.statusCode = 200;
|
|
49
|
-
commonsOutputDoing(`Marking all 200 DONE as QUEUED for ${this.domain}`);
|
|
50
|
-
}
|
|
51
|
-
else {
|
|
52
|
-
conditions.status.$in = [
|
|
53
|
-
EStatus.DONE,
|
|
54
|
-
EStatus.FAILED
|
|
55
|
-
];
|
|
56
|
-
commonsOutputDoing(`Marking all DONE and FAILED as QUEUED for ${this.domain}`);
|
|
57
|
-
}
|
|
58
|
-
yield this.databaseService.getUrls()
|
|
59
|
-
.updateMany(conditions, { $set: { status: EStatus.QUEUED } });
|
|
60
|
-
commonsOutputSuccess();
|
|
61
|
-
});
|
|
62
|
-
}
|
|
63
|
-
}
|
|
64
|
-
//# sourceMappingURL=requeue-domain.app.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"requeue-domain.app.js","sourceRoot":"","sources":["../../src/apps/requeue-domain.app.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAE,OAAO,EAAE,MAAM,yBAAyB,CAAC;AAElD,OAAO,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAC9E,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAMhD,MAAM,OAAO,gBAAiB,SAAQ,UAAU;IAG/C,YACU,MAAc,EACd,WAAoB;QAE7B,KAAK,CAAC,eAAe,CAAC,CAAC;QAHd,WAAM,GAAN,MAAM,CAAQ;QACd,gBAAW,GAAX,WAAW,CAAS;IAG9B,CAAC;IAEM,UAAU;QAChB,OAAO,wBAAwB,CAAC;IACjC,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,MAAM,UAAU,GAMZ;gBACF,MAAM,EAAE,IAAI,CAAC,MAAM;gBACnB,MAAM,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE;aACpB,CAAC;YAEF,IAAI,IAAI,CAAC,WAAW,EAAE;gBACrB,UAAU,CAAC,MAAM,CAAC,GAAG,GAAG,CAAE,OAAO,CAAC,IAAI,CAAE,CAAC;gBACzC,UAAU,CAAC,UAAU,GAAG,GAAG,CAAC;gBAE5B,kBAAkB,CAAC,sCAAsC,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;aACxE;iBAAM;gBACN,UAAU,CAAC,MAAM,CAAC,GAAG,GAAG;oBACtB,OAAO,CAAC,IAAI;oBACZ,OAAO,CAAC,MAAM;iBACf,CAAC;gBAEF,kBAAkB,CAAC,6CAA6C,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;aAC/E;YAED,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBACjC,UAAU,CACT,UAAU,EACR,EAAE,IAAI,EAAE,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,EAAE,CACvC,CAAC;YAEJ,oBAAoB,EAAE,CAAC;QACxB,CAAC;KAAA;CACD"}
|
package/dist/apps/seed.app.js
DELETED
|
@@ -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 { commonsOutputDoing, commonsOutputProgress, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
|
|
11
|
-
import { CommonsApp } from 'nodecommons-es-app';
|
|
12
|
-
import { Lists } from '../classes/lists';
|
|
13
|
-
import { EList } from '../enums/elist';
|
|
14
|
-
export class SeedApp extends CommonsApp {
|
|
15
|
-
constructor() {
|
|
16
|
-
super('hydra-crawler');
|
|
17
|
-
this.lists = new Lists();
|
|
18
|
-
}
|
|
19
|
-
getAppName() {
|
|
20
|
-
return 'Hydra - Seed';
|
|
21
|
-
}
|
|
22
|
-
setDatabaseService(databaseService) {
|
|
23
|
-
this.databaseService = databaseService;
|
|
24
|
-
}
|
|
25
|
-
addToList(list, entries) {
|
|
26
|
-
this.lists.add(list, entries);
|
|
27
|
-
}
|
|
28
|
-
init() {
|
|
29
|
-
const _super = Object.create(null, {
|
|
30
|
-
init: { get: () => super.init }
|
|
31
|
-
});
|
|
32
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
33
|
-
if (!this.databaseService)
|
|
34
|
-
throw new Error('Database service has not been set yet');
|
|
35
|
-
commonsOutputDoing('Connecting to database');
|
|
36
|
-
yield this.databaseService.init();
|
|
37
|
-
commonsOutputSuccess();
|
|
38
|
-
yield _super.init.call(this);
|
|
39
|
-
});
|
|
40
|
-
}
|
|
41
|
-
run() {
|
|
42
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
43
|
-
if (!this.databaseService)
|
|
44
|
-
throw new Error('Database service has not been set');
|
|
45
|
-
commonsOutputDoing('Building allowlist hostname seeds');
|
|
46
|
-
const hostnames = this.lists.listHostnames(EList.ALLOW);
|
|
47
|
-
const unique = [];
|
|
48
|
-
for (const hostname of hostnames) {
|
|
49
|
-
if (!unique.includes(hostname))
|
|
50
|
-
unique.push(`http://${hostname}/`);
|
|
51
|
-
if (!unique.includes(`www.${hostname}`))
|
|
52
|
-
unique.push(`http://www.${hostname}/`);
|
|
53
|
-
commonsOutputProgress(unique.length);
|
|
54
|
-
}
|
|
55
|
-
commonsOutputResult(unique.length);
|
|
56
|
-
commonsOutputDoing('Seeding into queue');
|
|
57
|
-
let tally = 0;
|
|
58
|
-
for (const url of unique) {
|
|
59
|
-
const outcome = yield this.databaseService.queue(url);
|
|
60
|
-
if (outcome) {
|
|
61
|
-
tally++;
|
|
62
|
-
commonsOutputProgress(tally);
|
|
63
|
-
}
|
|
64
|
-
}
|
|
65
|
-
commonsOutputResult(tally);
|
|
66
|
-
});
|
|
67
|
-
}
|
|
68
|
-
}
|
|
69
|
-
//# sourceMappingURL=seed.app.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"seed.app.js","sourceRoot":"","sources":["../../src/apps/seed.app.ts"],"names":[],"mappings":";;;;;;;;;AAAA,OAAO,EAAE,kBAAkB,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AAC1H,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAEhD,OAAO,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAC;AAMzC,OAAO,EAAE,KAAK,EAAE,MAAM,gBAAgB,CAAC;AAIvC,MAAM,OAAO,OAAQ,SAAQ,UAAU;IAKtC;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;QAEvB,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;IAC1B,CAAC;IAEM,UAAU;QAChB,OAAO,cAAc,CAAC;IACvB,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,kBAAkB,CAAC,mCAAmC,CAAC,CAAC;YACxD,MAAM,SAAS,GAAa,IAAI,CAAC,KAAK,CAAC,aAAa,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;YAElE,MAAM,MAAM,GAAa,EAAE,CAAC;YAC5B,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE;gBACjC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC;oBAAE,MAAM,CAAC,IAAI,CAAC,UAAU,QAAQ,GAAG,CAAC,CAAC;gBACnE,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,OAAO,QAAQ,EAAE,CAAC;oBAAE,MAAM,CAAC,IAAI,CAAC,cAAc,QAAQ,GAAG,CAAC,CAAC;gBAChF,qBAAqB,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;aACrC;YAED,mBAAmB,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;YAEnC,kBAAkB,CAAC,oBAAoB,CAAC,CAAC;YACzC,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,KAAK,MAAM,GAAG,IAAI,MAAM,EAAE;gBACzB,MAAM,OAAO,GAAY,MAAM,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBAC/D,IAAI,OAAO,EAAE;oBACZ,KAAK,EAAE,CAAC;oBACR,qBAAqB,CAAC,KAAK,CAAC,CAAC;iBAC7B;aACD;YACD,mBAAmB,CAAC,KAAK,CAAC,CAAC;QAC5B,CAAC;KAAA;CACD"}
|
package/dist/apps/startup.app.js
DELETED
|
@@ -1,64 +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 { commonsOutputDoing, commonsOutputResult, commonsOutputSuccess } from 'nodecommons-es-cli';
|
|
11
|
-
import { CommonsApp } from 'nodecommons-es-app';
|
|
12
|
-
// load the tables into Mongo's memory
|
|
13
|
-
export class StartupApp extends CommonsApp {
|
|
14
|
-
constructor() {
|
|
15
|
-
super('hydra-crawler');
|
|
16
|
-
}
|
|
17
|
-
getAppName() {
|
|
18
|
-
return 'Hydra - Startup';
|
|
19
|
-
}
|
|
20
|
-
setDatabaseService(databaseService) {
|
|
21
|
-
this.databaseService = databaseService;
|
|
22
|
-
}
|
|
23
|
-
init() {
|
|
24
|
-
const _super = Object.create(null, {
|
|
25
|
-
init: { get: () => super.init }
|
|
26
|
-
});
|
|
27
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
28
|
-
if (!this.databaseService)
|
|
29
|
-
throw new Error('Database service has not been set yet');
|
|
30
|
-
commonsOutputDoing('Connecting to database');
|
|
31
|
-
yield this.databaseService.init();
|
|
32
|
-
commonsOutputSuccess();
|
|
33
|
-
yield _super.init.call(this);
|
|
34
|
-
});
|
|
35
|
-
}
|
|
36
|
-
run() {
|
|
37
|
-
return __awaiter(this, void 0, void 0, function* () {
|
|
38
|
-
if (!this.databaseService)
|
|
39
|
-
throw new Error('Database service has not been set');
|
|
40
|
-
const dbo = this.databaseService.getRawDatabase();
|
|
41
|
-
for (const collection of 'domains,urls,links'.split(',')) {
|
|
42
|
-
commonsOutputDoing(`Loading ${collection} collection into memory (simple enumeration)`);
|
|
43
|
-
const tally = yield dbo.collection(collection).find({}).count();
|
|
44
|
-
commonsOutputResult(tally);
|
|
45
|
-
}
|
|
46
|
-
commonsOutputDoing('Loading domains collection into memory (regex enumeration)');
|
|
47
|
-
yield this.databaseService.getDomains().find({
|
|
48
|
-
domain: /[0-9]temp[0-9]/
|
|
49
|
-
}).count();
|
|
50
|
-
commonsOutputSuccess();
|
|
51
|
-
commonsOutputDoing('Loading urls collection into memory (regex enumeration)');
|
|
52
|
-
yield this.databaseService.getUrls().find({
|
|
53
|
-
url: /[0-9]temp[0-9]/
|
|
54
|
-
}).count();
|
|
55
|
-
commonsOutputSuccess();
|
|
56
|
-
commonsOutputDoing('Loading links collection into memory (regex enumeration)');
|
|
57
|
-
yield this.databaseService.getLinks().find({
|
|
58
|
-
outgoing: /[0-9]temp[0-9]/
|
|
59
|
-
}).count();
|
|
60
|
-
commonsOutputSuccess();
|
|
61
|
-
});
|
|
62
|
-
}
|
|
63
|
-
}
|
|
64
|
-
//# sourceMappingURL=startup.app.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"startup.app.js","sourceRoot":"","sources":["../../src/apps/startup.app.ts"],"names":[],"mappings":";;;;;;;;;AAEA,OAAO,EAAE,kBAAkB,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AACnG,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;AAMhD,sCAAsC;AAEtC,MAAM,OAAO,UAAW,SAAQ,UAAU;IAGzC;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;IACxB,CAAC;IAEM,UAAU;QAChB,OAAO,iBAAiB,CAAC;IAC1B,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEY,IAAI;;;;;YAChB,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;YAEpF,kBAAkB,CAAC,wBAAwB,CAAC,CAAC;YAC7C,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;YAClC,oBAAoB,EAAE,CAAC;YAEvB,MAAM,OAAM,IAAI,WAAE,CAAC;QACpB,CAAC;KAAA;IAEY,GAAG;;YACf,IAAI,CAAC,IAAI,CAAC,eAAe;gBAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;YAEhF,MAAM,GAAG,GAAe,IAAI,CAAC,eAAe,CAAC,cAAc,EAAE,CAAC;YAE9D,KAAK,MAAM,UAAU,IAAI,oBAAoB,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE;gBACzD,kBAAkB,CAAC,WAAW,UAAU,8CAA8C,CAAC,CAAC;gBACxF,MAAM,KAAK,GAAW,MAAM,GAAG,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,KAAK,EAAE,CAAC;gBACxE,mBAAmB,CAAC,KAAK,CAAC,CAAC;aAC3B;YAED,kBAAkB,CAAC,4DAA4D,CAAC,CAAC;YACjF,MAAM,IAAI,CAAC,eAAe,CAAC,UAAU,EAAE,CAAC,IAAI,CAAC;gBAC3C,MAAM,EAAE,gBAAgB;aACzB,CAAC,CAAC,KAAK,EAAE,CAAC;YACX,oBAAoB,EAAE,CAAC;YAEvB,kBAAkB,CAAC,yDAAyD,CAAC,CAAC;YAC9E,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CAAC;gBACxC,GAAG,EAAE,gBAAgB;aACtB,CAAC,CAAC,KAAK,EAAE,CAAC;YACX,oBAAoB,EAAE,CAAC;YAEvB,kBAAkB,CAAC,0DAA0D,CAAC,CAAC;YAC/E,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;gBACzC,QAAQ,EAAE,gBAAgB;aAC3B,CAAC,CAAC,KAAK,EAAE,CAAC;YACX,oBAAoB,EAAE,CAAC;QACxB,CAAC;KAAA;CACD"}
|