hydra-crawler 2.8.4 → 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 -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 +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
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
import * as express from 'express';
|
|
2
|
-
import { ICommonsRequestWithStrictParams } from 'nodecommons-
|
|
3
|
-
import { CommonsRestApi } from 'nodecommons-
|
|
4
|
-
import { RestServer } from '../servers/rest.server';
|
|
5
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
+
import { ICommonsRequestWithStrictParams } from 'nodecommons-esm-express';
|
|
3
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
4
|
+
import { RestServer } from '../servers/rest.server.mjs';
|
|
5
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
6
6
|
export declare class AutocompleteApi extends CommonsRestApi<ICommonsRequestWithStrictParams, express.Response> {
|
|
7
7
|
private databaseService;
|
|
8
8
|
constructor(restServer: RestServer, databaseService: DatabaseService, path: string);
|
|
@@ -0,0 +1,14 @@
|
|
|
1
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
2
|
+
export class AutocompleteApi extends CommonsRestApi {
|
|
3
|
+
databaseService;
|
|
4
|
+
constructor(restServer, databaseService, path) {
|
|
5
|
+
super(restServer);
|
|
6
|
+
this.databaseService = databaseService;
|
|
7
|
+
super.getHandler(`${path}rest/autocomplete/domain/:term[idname]`, async (req, _res) => {
|
|
8
|
+
const domains = await this.databaseService.listDomainsByLike(req.strictParams.term);
|
|
9
|
+
return domains
|
|
10
|
+
.map((domain) => domain.domain);
|
|
11
|
+
});
|
|
12
|
+
}
|
|
13
|
+
}
|
|
14
|
+
//# sourceMappingURL=autocomplete.api.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"autocomplete.api.mjs","sourceRoot":"","sources":["../../src/apis/autocomplete.api.mts"],"names":[],"mappings":"AAKA,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAMtD,MAAM,OAAO,eAAgB,SAAQ,cAAiE;IAG3F;IAFV,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,wCAAwC,EAC/C,KAAK,EAAE,GAAoC,EAAE,IAAsB,EAAqB,EAAE;YACzF,MAAM,OAAO,GAAc,MAAM,IAAI,CAAC,eAAe,CAAC,iBAAiB,CAAC,GAAG,CAAC,YAAY,CAAC,IAAc,CAAC,CAAC;YAEzG,OAAO,OAAO;iBACX,GAAG,CAAC,CAAC,MAAe,EAAU,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;QACpD,CAAC,CACF,CAAC;IACH,CAAC;CACD"}
|
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
import * as express from 'express';
|
|
2
|
-
import { CommonsRestApi } from 'nodecommons-
|
|
3
|
-
import { ICommonsRequestWithStrictParams } from 'nodecommons-
|
|
4
|
-
import { RestServer } from '../servers/rest.server';
|
|
5
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
3
|
+
import { ICommonsRequestWithStrictParams } from 'nodecommons-esm-express';
|
|
4
|
+
import { RestServer } from '../servers/rest.server.mjs';
|
|
5
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
6
6
|
export declare class BugsApi extends CommonsRestApi<ICommonsRequestWithStrictParams, express.Response> {
|
|
7
7
|
private databaseService;
|
|
8
8
|
constructor(restServer: RestServer, databaseService: DatabaseService, path: string);
|
|
@@ -0,0 +1,18 @@
|
|
|
1
|
+
import { commonsTypeEncode } from 'tscommons-esm-core';
|
|
2
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
3
|
+
export class BugsApi extends CommonsRestApi {
|
|
4
|
+
databaseService;
|
|
5
|
+
constructor(restServer, databaseService, path) {
|
|
6
|
+
super(restServer);
|
|
7
|
+
this.databaseService = databaseService;
|
|
8
|
+
super.getHandler(`${path}rest/bugs/php`, async (_req, _res) => {
|
|
9
|
+
const result = await this.databaseService.listPhpErrors();
|
|
10
|
+
return commonsTypeEncode(result);
|
|
11
|
+
});
|
|
12
|
+
super.getHandler(`${path}rest/bugs/asp`, async (_req, _res) => {
|
|
13
|
+
const result = await this.databaseService.listAspErrors();
|
|
14
|
+
return commonsTypeEncode(result);
|
|
15
|
+
});
|
|
16
|
+
}
|
|
17
|
+
}
|
|
18
|
+
//# sourceMappingURL=bugs.api.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"bugs.api.mjs","sourceRoot":"","sources":["../../src/apis/bugs.api.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,iBAAiB,EAAY,MAAM,oBAAoB,CAAC;AAMjE,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAOtD,MAAM,OAAO,OAAQ,SAAQ,cAAiE;IAGnF;IAFV,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,eAAe,EACtB,KAAK,EAAE,IAAqB,EAAE,IAAsB,EAAqB,EAAE;YAC1E,MAAM,MAAM,GAA4B,MAAM,IAAI,CAAC,eAAe,CAAC,aAAa,EAAE,CAAC;YAEnF,OAAO,iBAAiB,CAAC,MAAM,CAAC,CAAC;QAClC,CAAC,CACF,CAAC;QAEF,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,eAAe,EACtB,KAAK,EAAE,IAAqB,EAAE,IAAsB,EAAqB,EAAE;YAC1E,MAAM,MAAM,GAA4B,MAAM,IAAI,CAAC,eAAe,CAAC,aAAa,EAAE,CAAC;YAEnF,OAAO,iBAAiB,CAAC,MAAM,CAAC,CAAC;QAClC,CAAC,CACF,CAAC;IACH,CAAC;CACD"}
|
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
import * as express from 'express';
|
|
2
|
-
import { CommonsRestApi } from 'nodecommons-
|
|
3
|
-
import { ICommonsRequestWithStrictParams } from 'nodecommons-
|
|
4
|
-
import { RestServer } from '../servers/rest.server';
|
|
5
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
3
|
+
import { ICommonsRequestWithStrictParams } from 'nodecommons-esm-express';
|
|
4
|
+
import { RestServer } from '../servers/rest.server.mjs';
|
|
5
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
6
6
|
export declare class CrawlApi extends CommonsRestApi<ICommonsRequestWithStrictParams, express.Response> {
|
|
7
7
|
private databaseService;
|
|
8
8
|
constructor(restServer: RestServer, databaseService: DatabaseService, path: string);
|
|
@@ -0,0 +1,17 @@
|
|
|
1
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
2
|
+
export class CrawlApi extends CommonsRestApi {
|
|
3
|
+
databaseService;
|
|
4
|
+
constructor(restServer, databaseService, path) {
|
|
5
|
+
super(restServer);
|
|
6
|
+
this.databaseService = databaseService;
|
|
7
|
+
super.getHandler(`${path}rest/crawl/queue`, async (_req, _res) => {
|
|
8
|
+
const tallies = await this.databaseService.listDomainQueuedTallies();
|
|
9
|
+
return Array.from(tallies.keys())
|
|
10
|
+
.map((domain) => ({
|
|
11
|
+
domain: domain,
|
|
12
|
+
queue: tallies.get(domain) || 0
|
|
13
|
+
}));
|
|
14
|
+
});
|
|
15
|
+
}
|
|
16
|
+
}
|
|
17
|
+
//# sourceMappingURL=crawl.api.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"crawl.api.mjs","sourceRoot":"","sources":["../../src/apis/crawl.api.mts"],"names":[],"mappings":"AAIA,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAOtD,MAAM,OAAO,QAAS,SAAQ,cAAiE;IAGpF;IAFV,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,kBAAkB,EACzB,KAAK,EAAE,IAAqB,EAAE,IAAsB,EAA2B,EAAE;YAChF,MAAM,OAAO,GAAwB,MAAM,IAAI,CAAC,eAAe,CAAC,uBAAuB,EAAE,CAAC;YAE1F,OAAO,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;iBAC9B,GAAG,CAAC,CAAC,MAAc,EAAgB,EAAE,CAAC,CAAC;gBACtC,MAAM,EAAE,MAAM;gBACd,KAAK,EAAE,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC;aAChC,CAAC,CAAC,CAAC;QACP,CAAC,CACF,CAAC;IAEH,CAAC;CACD"}
|
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
import * as express from 'express';
|
|
2
|
-
import { ICommonsRequestWithStrictParams } from 'nodecommons-
|
|
3
|
-
import { CommonsRestApi } from 'nodecommons-
|
|
4
|
-
import { RestServer } from '../servers/rest.server';
|
|
5
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
+
import { ICommonsRequestWithStrictParams } from 'nodecommons-esm-express';
|
|
3
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
4
|
+
import { RestServer } from '../servers/rest.server.mjs';
|
|
5
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
6
6
|
export declare class DomainsApi extends CommonsRestApi<ICommonsRequestWithStrictParams, express.Response> {
|
|
7
7
|
private databaseService;
|
|
8
8
|
constructor(restServer: RestServer, databaseService: DatabaseService, path: string);
|
|
@@ -1,26 +1,18 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
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 { commonsTreeStringSegmentArrayToTree, commonsTypeEncode } from 'tscommons-es-core';
|
|
11
|
-
import { CommonsRestApi, commonsRestApiBadRequest, commonsRestApiError } from 'nodecommons-es-rest';
|
|
1
|
+
import { commonsTreeStringSegmentArrayToTree, commonsTypeEncode } from 'tscommons-esm-core';
|
|
2
|
+
import { CommonsRestApi, commonsRestApiBadRequest, commonsRestApiError } from 'nodecommons-esm-rest';
|
|
12
3
|
export class DomainsApi extends CommonsRestApi {
|
|
4
|
+
databaseService;
|
|
13
5
|
constructor(restServer, databaseService, path) {
|
|
14
6
|
super(restServer);
|
|
15
7
|
this.databaseService = databaseService;
|
|
16
|
-
super.getHandler(`${path}rest/domains`, (_req, _res) =>
|
|
17
|
-
const domains =
|
|
8
|
+
super.getHandler(`${path}rest/domains`, async (_req, _res) => {
|
|
9
|
+
const domains = await this.databaseService.listDomains();
|
|
18
10
|
return commonsTypeEncode(domains);
|
|
19
|
-
})
|
|
20
|
-
super.getHandler(`${path}rest/domains/:domain[idname]/tree`, (req, _res) =>
|
|
11
|
+
});
|
|
12
|
+
super.getHandler(`${path}rest/domains/:domain[idname]/tree`, async (req, _res) => {
|
|
21
13
|
if (!/^[-a-z0-9]+(\.[-a-z0-9]+)*$/i.test(req.strictParams.domain))
|
|
22
14
|
return commonsRestApiBadRequest('Invalid domain');
|
|
23
|
-
const urls =
|
|
15
|
+
const urls = await this.databaseService.listDone200DomainUrls(req.params.domain);
|
|
24
16
|
const splits = [];
|
|
25
17
|
for (const url of urls) {
|
|
26
18
|
const regex = new RegExp(`^http(?:s?)://${req.params.domain}(?::[0-9]+)?(/.*)$`);
|
|
@@ -36,7 +28,7 @@ export class DomainsApi extends CommonsRestApi {
|
|
|
36
28
|
}
|
|
37
29
|
const tree = commonsTreeStringSegmentArrayToTree(splits);
|
|
38
30
|
return commonsTypeEncode(tree);
|
|
39
|
-
})
|
|
31
|
+
});
|
|
40
32
|
}
|
|
41
33
|
}
|
|
42
|
-
//# sourceMappingURL=domains.api.
|
|
34
|
+
//# sourceMappingURL=domains.api.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"domains.api.mjs","sourceRoot":"","sources":["../../src/apis/domains.api.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,mCAAmC,EAAE,iBAAiB,EAA0B,MAAM,oBAAoB,CAAC;AAKpH,OAAO,EAAE,cAAc,EAAE,wBAAwB,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAC;AAMrG,MAAM,OAAO,UAAW,SAAQ,cAAiE;IAGtF;IAFV,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,cAAc,EACrB,KAAK,EAAE,IAAqB,EAAE,IAAsB,EAAqB,EAAE;YAC1E,MAAM,OAAO,GAAc,MAAM,IAAI,CAAC,eAAe,CAAC,WAAW,EAAE,CAAC;YAEpE,OAAO,iBAAiB,CAAC,OAAO,CAAC,CAAC;QACnC,CAAC,CACF,CAAC;QAEF,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,mCAAmC,EAC1C,KAAK,EAAE,GAAoC,EAAE,IAAsB,EAAqB,EAAE;YACzF,IAAI,CAAC,8BAA8B,CAAC,IAAI,CAAC,GAAG,CAAC,YAAY,CAAC,MAAgB,CAAC;gBAAE,OAAO,wBAAwB,CAAC,gBAAgB,CAAC,CAAC;YAE/H,MAAM,IAAI,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,qBAAqB,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;YAE3F,MAAM,MAAM,GAAe,EAAE,CAAC;YAC9B,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBACxB,MAAM,KAAK,GAAW,IAAI,MAAM,CAAC,iBAAiB,GAAG,CAAC,MAAM,CAAC,MAAM,oBAAoB,CAAC,CAAC;gBACzF,MAAM,KAAK,GAAyB,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBAEpD,IAAI,KAAK,KAAK,IAAI;oBAAE,OAAO,mBAAmB,CAAC,uBAAuB,GAAG,EAAE,CAAC,CAAC;gBAE7E,IAAI,OAAO,GAAW,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC/B,IAAI,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;oBAAE,OAAO,GAAG,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;gBAEjE,MAAM,QAAQ,GAAa,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;gBAC9C,QAAQ,CAAC,GAAG,EAAE,CAAC,CAAC,kBAAkB;gBAElC,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;YACvB,CAAC;YAED,MAAM,IAAI,GAAiB,mCAAmC,CAAC,MAAM,CAAC,CAAC;YACvE,OAAO,iBAAiB,CAAC,IAAI,CAAC,CAAC;QAChC,CAAC,CACF,CAAC;IACH,CAAC;CACD"}
|
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
import * as express from 'express';
|
|
2
|
-
import { ICommonsRequestWithStrictParams } from 'nodecommons-
|
|
3
|
-
import { CommonsRestApi } from 'nodecommons-
|
|
4
|
-
import { RestServer } from '../servers/rest.server';
|
|
5
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
+
import { ICommonsRequestWithStrictParams } from 'nodecommons-esm-express';
|
|
3
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
4
|
+
import { RestServer } from '../servers/rest.server.mjs';
|
|
5
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
6
6
|
export declare class ImagesApi extends CommonsRestApi<ICommonsRequestWithStrictParams, express.Response> {
|
|
7
7
|
private databaseService;
|
|
8
8
|
constructor(restServer: RestServer, databaseService: DatabaseService, path: string);
|
|
@@ -0,0 +1,19 @@
|
|
|
1
|
+
import { commonsTypeEncode } from 'tscommons-esm-core';
|
|
2
|
+
import { toEComparator, ECOMPARATORS } from 'hydra-crawler-esm-assets';
|
|
3
|
+
import { CommonsRestApi, commonsRestApiBadRequest } from 'nodecommons-esm-rest';
|
|
4
|
+
export class ImagesApi extends CommonsRestApi {
|
|
5
|
+
databaseService;
|
|
6
|
+
constructor(restServer, databaseService, path) {
|
|
7
|
+
super(restServer);
|
|
8
|
+
this.databaseService = databaseService;
|
|
9
|
+
super.getHandler(`${path}rest/images/:comparator(${ECOMPARATORS.join('|')})/:size[int]`, async (req, _res) => {
|
|
10
|
+
const comparator = toEComparator(req.strictParams.comparator);
|
|
11
|
+
if (!comparator)
|
|
12
|
+
return commonsRestApiBadRequest('Invalid comparator');
|
|
13
|
+
const size = req.strictParams.size;
|
|
14
|
+
const matches = await this.databaseService.listImagesBySizeThreshold(size, comparator);
|
|
15
|
+
return commonsTypeEncode(matches);
|
|
16
|
+
});
|
|
17
|
+
}
|
|
18
|
+
}
|
|
19
|
+
//# sourceMappingURL=images.api.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"images.api.mjs","sourceRoot":"","sources":["../../src/apis/images.api.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,iBAAiB,EAAY,MAAM,oBAAoB,CAAC;AAGjE,OAAO,EAAe,aAAa,EAAE,YAAY,EAAE,MAAM,0BAA0B,CAAC;AAGpF,OAAO,EAAE,cAAc,EAAE,wBAAwB,EAAE,MAAM,sBAAsB,CAAC;AAMhF,MAAM,OAAO,SAAU,SAAQ,cAAiE;IAGrF;IAFV,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,2BAA2B,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,cAAc,EACtE,KAAK,EAAE,GAAoC,EAAE,IAAsB,EAAqB,EAAE;YACzF,MAAM,UAAU,GAA0B,aAAa,CAAC,GAAG,CAAC,YAAY,CAAC,UAAoB,CAAC,CAAC;YAC/F,IAAI,CAAC,UAAU;gBAAE,OAAO,wBAAwB,CAAC,oBAAoB,CAAC,CAAC;YAEvE,MAAM,IAAI,GAAW,GAAG,CAAC,YAAY,CAAC,IAAc,CAAC;YAErD,MAAM,OAAO,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,yBAAyB,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;YAEjG,OAAO,iBAAiB,CAAC,OAAO,CAAC,CAAC;QACnC,CAAC,CACF,CAAC;IACH,CAAC;CACD"}
|
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
import * as express from 'express';
|
|
2
|
-
import { CommonsRestApi } from 'nodecommons-
|
|
3
|
-
import { ICommonsRequestWithStrictParams } from 'nodecommons-
|
|
4
|
-
import { RestServer } from '../servers/rest.server';
|
|
5
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
3
|
+
import { ICommonsRequestWithStrictParams } from 'nodecommons-esm-express';
|
|
4
|
+
import { RestServer } from '../servers/rest.server.mjs';
|
|
5
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
6
6
|
export declare class StatisticsApi extends CommonsRestApi<ICommonsRequestWithStrictParams, express.Response> {
|
|
7
7
|
private databaseService;
|
|
8
8
|
private memcached;
|
|
@@ -0,0 +1,33 @@
|
|
|
1
|
+
import Cache from 'memcached-promisify';
|
|
2
|
+
import { commonsTypeHasPropertyEnum, commonsTypeHasPropertyNumber, commonsTypeIsTArray } from 'tscommons-esm-core';
|
|
3
|
+
import { isEStatus } from 'hydra-crawler-esm-assets';
|
|
4
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
5
|
+
function isTStatusTally(test) {
|
|
6
|
+
if (!commonsTypeHasPropertyEnum(test, 'status', isEStatus))
|
|
7
|
+
return false;
|
|
8
|
+
if (!commonsTypeHasPropertyNumber(test, 'tally'))
|
|
9
|
+
return false;
|
|
10
|
+
return true;
|
|
11
|
+
}
|
|
12
|
+
export class StatisticsApi extends CommonsRestApi {
|
|
13
|
+
databaseService;
|
|
14
|
+
memcached;
|
|
15
|
+
constructor(restServer, databaseService, path) {
|
|
16
|
+
super(restServer);
|
|
17
|
+
this.databaseService = databaseService;
|
|
18
|
+
this.memcached = new Cache({ keyPrefix: 'hydra', cacheHost: 'localhost:11211' });
|
|
19
|
+
super.getHandler(`${path}rest/statistics/status`, async (_req, _res) => {
|
|
20
|
+
const index = 'statistics/status';
|
|
21
|
+
const cached = await this.memcached.get(index);
|
|
22
|
+
if (cached && commonsTypeIsTArray(cached, isTStatusTally))
|
|
23
|
+
return cached;
|
|
24
|
+
const result = await this.databaseService.listStatusTallies();
|
|
25
|
+
const tallies = [];
|
|
26
|
+
for (const key of result.keys())
|
|
27
|
+
tallies.push({ status: key, tally: result.get(key) || 0 });
|
|
28
|
+
await this.memcached.set(index, tallies, 10);
|
|
29
|
+
return tallies;
|
|
30
|
+
});
|
|
31
|
+
}
|
|
32
|
+
}
|
|
33
|
+
//# sourceMappingURL=statistics.api.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"statistics.api.mjs","sourceRoot":"","sources":["../../src/apis/statistics.api.mts"],"names":[],"mappings":"AACA,OAAO,KAAK,MAAM,qBAAqB,CAAC;AAExC,OAAO,EAAE,0BAA0B,EAAE,4BAA4B,EAAE,mBAAmB,EAAE,MAAM,oBAAoB,CAAC;AAEnH,OAAO,EAAW,SAAS,EAAE,MAAM,0BAA0B,CAAC;AAE9D,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAWtD,SAAS,cAAc,CAAC,IAAa;IACpC,IAAI,CAAC,0BAA0B,CAAC,IAAI,EAAE,QAAQ,EAAE,SAAS,CAAC;QAAE,OAAO,KAAK,CAAC;IACzE,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,OAAO,CAAC;QAAE,OAAO,KAAK,CAAC;IAE/D,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAO,aAAc,SAAQ,cAAiE;IAKzF;IAJF,SAAS,CAAQ;IAEzB,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,IAAI,CAAC,SAAS,GAAG,IAAI,KAAK,CAAC,EAAE,SAAS,EAAE,OAAO,EAAE,SAAS,EAAE,iBAAiB,EAAE,CAAC,CAAC;QAEjF,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,wBAAwB,EAC/B,KAAK,EAAE,IAAqB,EAAE,IAAsB,EAA2B,EAAE;YAChF,MAAM,KAAK,GAAW,mBAAmB,CAAC;YAE1C,MAAM,MAAM,GAAY,MAAM,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;YACxD,IAAI,MAAM,IAAI,mBAAmB,CAAe,MAAM,EAAE,cAAc,CAAC;gBAAE,OAAO,MAAM,CAAC;YAEvF,MAAM,MAAM,GAAyB,MAAM,IAAI,CAAC,eAAe,CAAC,iBAAiB,EAAE,CAAC;YAEpF,MAAM,OAAO,GAAmB,EAAE,CAAC;YACnC,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,EAAE;gBAAE,OAAO,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,GAAG,EAAE,KAAK,EAAE,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;YAE5F,MAAM,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,KAAK,EAAE,OAAO,EAAE,EAAE,CAAC,CAAC;YAE7C,OAAO,OAAO,CAAC;QAChB,CAAC,CACF,CAAC;IACH,CAAC;CACD"}
|
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import * as express from 'express';
|
|
2
|
-
import { CommonsRestApi } from 'nodecommons-
|
|
3
|
-
import { ICommonsRequestWithStrictParams } from 'nodecommons-
|
|
4
|
-
import { RestServer } from '../servers/rest.server';
|
|
2
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
3
|
+
import { ICommonsRequestWithStrictParams } from 'nodecommons-esm-express';
|
|
4
|
+
import { RestServer } from '../servers/rest.server.mjs';
|
|
5
5
|
export declare class TestApi extends CommonsRestApi<ICommonsRequestWithStrictParams, express.Response> {
|
|
6
6
|
constructor(restServer: RestServer, path: string);
|
|
7
7
|
}
|
|
@@ -0,0 +1,10 @@
|
|
|
1
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
2
|
+
export class TestApi extends CommonsRestApi {
|
|
3
|
+
constructor(restServer, path) {
|
|
4
|
+
super(restServer);
|
|
5
|
+
super.getHandler(`${path}test/hello`,
|
|
6
|
+
// eslint-disable-next-line @typescript-eslint/require-await
|
|
7
|
+
async (_) => 'Hello world');
|
|
8
|
+
}
|
|
9
|
+
}
|
|
10
|
+
//# sourceMappingURL=test.api.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"test.api.mjs","sourceRoot":"","sources":["../../src/apis/test.api.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAKtD,MAAM,OAAO,OAAQ,SAAQ,cAAiE;IAC7F,YACE,UAAsB,EACtB,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAElB,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,YAAY;QACnB,4DAA4D;QAC5D,KAAK,EAAE,CAAU,EAAmB,EAAE,CAAC,aAAa,CACrD,CAAC;IACH,CAAC;CACD"}
|
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
import * as express from 'express';
|
|
2
|
-
import { CommonsRestApi } from 'nodecommons-
|
|
3
|
-
import { ICommonsRequestWithStrictParams } from 'nodecommons-
|
|
4
|
-
import { RestServer } from '../servers/rest.server';
|
|
5
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
3
|
+
import { ICommonsRequestWithStrictParams } from 'nodecommons-esm-express';
|
|
4
|
+
import { RestServer } from '../servers/rest.server.mjs';
|
|
5
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
6
6
|
export declare class UrlsApi extends CommonsRestApi<ICommonsRequestWithStrictParams, express.Response> {
|
|
7
7
|
private databaseService;
|
|
8
8
|
constructor(restServer: RestServer, databaseService: DatabaseService, path: string);
|
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
import { commonsTypeEncode } from 'tscommons-esm-core';
|
|
2
|
+
import { EDirection } from 'hydra-crawler-esm-assets';
|
|
3
|
+
import { CommonsRestApi } from 'nodecommons-esm-rest';
|
|
4
|
+
export class UrlsApi extends CommonsRestApi {
|
|
5
|
+
databaseService;
|
|
6
|
+
constructor(restServer, databaseService, path) {
|
|
7
|
+
super(restServer);
|
|
8
|
+
this.databaseService = databaseService;
|
|
9
|
+
// Can't use strict and ICommonsRequestWithStrictParams as the url could be anything, and idname won't accept that
|
|
10
|
+
super.getHandler(`${path}rest/urls/:url/links`, async (req, _res) => {
|
|
11
|
+
const outbound = await this.databaseService.listOutboundLinks(req.params.url);
|
|
12
|
+
const inbound = await this.databaseService.listInboundLinks(req.params.url);
|
|
13
|
+
const urlLinks = {
|
|
14
|
+
[EDirection.INBOUND]: inbound,
|
|
15
|
+
[EDirection.OUTBOUND]: outbound
|
|
16
|
+
};
|
|
17
|
+
return commonsTypeEncode(urlLinks);
|
|
18
|
+
});
|
|
19
|
+
}
|
|
20
|
+
}
|
|
21
|
+
//# sourceMappingURL=urls.api.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"urls.api.mjs","sourceRoot":"","sources":["../../src/apis/urls.api.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,iBAAiB,EAAY,MAAM,oBAAoB,CAAC;AAGjE,OAAO,EAAE,UAAU,EAAE,MAAM,0BAA0B,CAAC;AAEtD,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAOtD,MAAM,OAAO,OAAQ,SAAQ,cAAiE;IAGnF;IAFV,YACE,UAAsB,EACd,eAAgC,EACxC,IAAY;QAEb,KAAK,CAAC,UAAU,CAAC,CAAC;QAHT,oBAAe,GAAf,eAAe,CAAiB;QAKzC,kHAAkH;QAElH,KAAK,CAAC,UAAU,CACd,GAAG,IAAI,sBAAsB,EAC7B,KAAK,EAAE,GAAoB,EAAE,IAAsB,EAAqB,EAAE;YACzE,MAAM,QAAQ,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,iBAAiB,CAAC,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YACxF,MAAM,OAAO,GAAa,MAAM,IAAI,CAAC,eAAe,CAAC,gBAAgB,CAAC,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YAEtF,MAAM,QAAQ,GAAc;gBAC1B,CAAC,UAAU,CAAC,OAAO,CAAC,EAAE,OAAO;gBAC7B,CAAC,UAAU,CAAC,QAAQ,CAAC,EAAE,QAAQ;aAChC,CAAC;YAEF,OAAO,iBAAiB,CAAC,QAAQ,CAAC,CAAC;QACpC,CAAC,CACF,CAAC;IACH,CAAC;CACD"}
|
|
@@ -1,8 +1,8 @@
|
|
|
1
|
-
import { CommonsApp } from 'nodecommons-
|
|
2
|
-
import { DatabaseService } from '../services/database.service';
|
|
3
|
-
import { IMatch } from '../interfaces/imatch';
|
|
4
|
-
import { EList } from '../enums/elist';
|
|
5
|
-
import { IInternalHydraCommonListApp } 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 { EList } from '../enums/elist.mjs';
|
|
5
|
+
import { IInternalHydraCommonListApp } from './internal-hydra-common.app.mjs';
|
|
6
6
|
export declare class CleanupApp extends CommonsApp implements IInternalHydraCommonListApp {
|
|
7
7
|
private includeReset;
|
|
8
8
|
private includeLinks;
|
|
@@ -0,0 +1,129 @@
|
|
|
1
|
+
import { commonsTypeHasPropertyNumber } from 'tscommons-esm-core';
|
|
2
|
+
import { isIUrl } from 'hydra-crawler-esm-assets';
|
|
3
|
+
import { EStatus } from 'hydra-crawler-esm-assets';
|
|
4
|
+
import { commonsLogDoing } from 'nodecommons-esm-log';
|
|
5
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
6
|
+
import { Lists } from '../classes/lists.mjs';
|
|
7
|
+
import { Cleaner } from '../classes/cleaner.mjs';
|
|
8
|
+
export class CleanupApp extends CommonsApp {
|
|
9
|
+
includeReset;
|
|
10
|
+
includeLinks;
|
|
11
|
+
databaseService;
|
|
12
|
+
lists;
|
|
13
|
+
constructor(includeReset = false, includeLinks = false) {
|
|
14
|
+
super('hydra-crawler');
|
|
15
|
+
this.includeReset = includeReset;
|
|
16
|
+
this.includeLinks = includeLinks;
|
|
17
|
+
this.lists = new Lists();
|
|
18
|
+
}
|
|
19
|
+
getAppName() {
|
|
20
|
+
return 'Hydra - Cleanup';
|
|
21
|
+
}
|
|
22
|
+
setDatabaseService(databaseService) {
|
|
23
|
+
this.databaseService = databaseService;
|
|
24
|
+
}
|
|
25
|
+
addToList(list, entries) {
|
|
26
|
+
this.lists.add(list, entries);
|
|
27
|
+
}
|
|
28
|
+
async init() {
|
|
29
|
+
if (!this.databaseService)
|
|
30
|
+
throw new Error('Database service has not been set yet');
|
|
31
|
+
const log = commonsLogDoing('cleanup', 'Connecting to database');
|
|
32
|
+
await this.databaseService.init();
|
|
33
|
+
log.success();
|
|
34
|
+
await super.init();
|
|
35
|
+
}
|
|
36
|
+
async reset() {
|
|
37
|
+
if (!this.databaseService)
|
|
38
|
+
throw new Error('Database service has not been set');
|
|
39
|
+
const log = commonsLogDoing('cleanup', 'Reseting urls for DENY, FAILED, DEAD and DISALLOWED');
|
|
40
|
+
let tally = 0;
|
|
41
|
+
while (true) {
|
|
42
|
+
const result = this.databaseService.getUrls()
|
|
43
|
+
.find({ status: { $in: [
|
|
44
|
+
EStatus.DENY,
|
|
45
|
+
EStatus.FAILED,
|
|
46
|
+
EStatus.DEAD,
|
|
47
|
+
EStatus.DISALLOWED
|
|
48
|
+
] } })
|
|
49
|
+
.sort({ _id: 1 })
|
|
50
|
+
.skip(tally);
|
|
51
|
+
try {
|
|
52
|
+
while (true) {
|
|
53
|
+
tally++;
|
|
54
|
+
if ((tally % 100) === 0)
|
|
55
|
+
log.progress(tally);
|
|
56
|
+
const row = await result.next();
|
|
57
|
+
if (row === null)
|
|
58
|
+
break;
|
|
59
|
+
await this.databaseService.getLinks().deleteMany({ url: row.url });
|
|
60
|
+
}
|
|
61
|
+
break;
|
|
62
|
+
}
|
|
63
|
+
catch (err) {
|
|
64
|
+
if (!commonsTypeHasPropertyNumber(err, 'code') || err.code !== 43)
|
|
65
|
+
throw err;
|
|
66
|
+
}
|
|
67
|
+
}
|
|
68
|
+
log.result(tally);
|
|
69
|
+
}
|
|
70
|
+
async links() {
|
|
71
|
+
if (!this.databaseService)
|
|
72
|
+
throw new Error('Database service has not been set');
|
|
73
|
+
const orphans = [];
|
|
74
|
+
{ // scope
|
|
75
|
+
const log = commonsLogDoing('cleanup', 'Searching for link orphans');
|
|
76
|
+
let tally = 0;
|
|
77
|
+
const result = this.databaseService.getLinks().find({}, {});
|
|
78
|
+
let batch = [];
|
|
79
|
+
while (true) {
|
|
80
|
+
tally++;
|
|
81
|
+
if ((tally % 1000) === 0)
|
|
82
|
+
log.progress(`${tally}, ${orphans.length}`);
|
|
83
|
+
const link = await result.next();
|
|
84
|
+
if (link === null)
|
|
85
|
+
break;
|
|
86
|
+
if (!batch.includes(link.url))
|
|
87
|
+
batch.push(link.url);
|
|
88
|
+
if (!batch.includes(link.outgoing))
|
|
89
|
+
batch.push(link.outgoing);
|
|
90
|
+
if (batch.length < 10000)
|
|
91
|
+
continue;
|
|
92
|
+
const result2 = this.databaseService.getUrls().find({ url: { $in: batch } });
|
|
93
|
+
const matches = (await this.databaseService.listQueryResults(result2, isIUrl))
|
|
94
|
+
.map((u) => u.url);
|
|
95
|
+
for (const l of batch) {
|
|
96
|
+
if (!matches.includes(l))
|
|
97
|
+
orphans.push(l);
|
|
98
|
+
}
|
|
99
|
+
batch = [];
|
|
100
|
+
}
|
|
101
|
+
log.result(orphans.length);
|
|
102
|
+
}
|
|
103
|
+
{ // scope
|
|
104
|
+
const log = commonsLogDoing('cleanup', 'Removing orphan links');
|
|
105
|
+
let tally = 0;
|
|
106
|
+
for (const orphan of orphans) {
|
|
107
|
+
tally++;
|
|
108
|
+
if ((tally % 100) === 0)
|
|
109
|
+
log.progress(`${tally}`);
|
|
110
|
+
await this.databaseService.getLinks().deleteMany({ url: orphan });
|
|
111
|
+
await this.databaseService.getLinks().deleteMany({ outgoing: orphan });
|
|
112
|
+
}
|
|
113
|
+
log.success();
|
|
114
|
+
}
|
|
115
|
+
}
|
|
116
|
+
async run() {
|
|
117
|
+
if (!this.databaseService)
|
|
118
|
+
throw new Error('Database service has not been set');
|
|
119
|
+
if (this.includeReset)
|
|
120
|
+
await this.reset();
|
|
121
|
+
if (this.includeLinks)
|
|
122
|
+
await this.links();
|
|
123
|
+
const hardLimit = this.args.getFirstNumberOrUndefined('hard-limit');
|
|
124
|
+
const cleaner = new Cleaner(this.lists, this.databaseService);
|
|
125
|
+
await cleaner.purgeOrphanUrls(hardLimit);
|
|
126
|
+
await cleaner.purgeEmptyDomains();
|
|
127
|
+
}
|
|
128
|
+
}
|
|
129
|
+
//# sourceMappingURL=cleanup.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"cleanup.app.mjs","sourceRoot":"","sources":["../../src/apps/cleanup.app.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,4BAA4B,EAAE,MAAM,oBAAoB,CAAC;AAElE,OAAO,EAAQ,MAAM,EAAE,MAAM,0BAA0B,CAAC;AAExD,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAEnD,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAEjD,OAAO,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAC7C,OAAO,EAAE,OAAO,EAAE,MAAM,wBAAwB,CAAC;AAUjD,MAAM,OAAO,UAAW,SAAQ,UAAU;IAM/B;IACA;IANF,eAAe,CAA4B;IAE3C,KAAK,CAAQ;IAErB,YACU,eAAwB,KAAK,EAC7B,eAAwB,KAAK;QAEtC,KAAK,CAAC,eAAe,CAAC,CAAC;QAHd,iBAAY,GAAZ,YAAY,CAAiB;QAC7B,iBAAY,GAAZ,YAAY,CAAiB;QAItC,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;IAC1B,CAAC;IAEM,UAAU;QAChB,OAAO,iBAAiB,CAAC;IAC1B,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,wBAAwB,CAAC,CAAC;QACrF,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEO,KAAK,CAAC,KAAK;QAClB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,qDAAqD,CAAC,CAAC;QAElH,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,OAAO,IAAI,EAAE,CAAC;YACb,MAAM,MAAM,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBACpE,IAAI,CAAC,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE;wBACrB,OAAO,CAAC,IAAI;wBACZ,OAAO,CAAC,MAAM;wBACd,OAAO,CAAC,IAAI;wBACZ,OAAO,CAAC,UAAU;qBACnB,EAAE,EAAE,CAAC;iBACL,IAAI,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC;iBAChB,IAAI,CAAC,KAAK,CAAC,CAAC;YAEf,IAAI,CAAC;gBACJ,OAAO,IAAI,EAAE,CAAC;oBACb,KAAK,EAAE,CAAC;oBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;wBAAE,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;oBAE7C,MAAM,GAAG,GAAc,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;oBAC3C,IAAI,GAAG,KAAK,IAAI;wBAAE,MAAM;oBAExB,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC;gBACpE,CAAC;gBAED,MAAM;YACP,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACd,IAAI,CAAC,4BAA4B,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,GAAG,CAAC,IAAI,KAAK,EAAE;oBAAE,MAAM,GAAG,CAAC;YAC9E,CAAC;QACF,CAAC;QAED,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;IACnB,CAAC;IAEO,KAAK,CAAC,KAAK;QAClB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,MAAM,OAAO,GAAa,EAAE,CAAC;QAC7B,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,4BAA4B,CAAC,CAAC;YACzF,IAAI,KAAK,GAAW,CAAC,CAAC;YAEtB,MAAM,MAAM,GAA8B,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;YAE9F,IAAI,KAAK,GAAa,EAAE,CAAC;YACzB,OAAO,IAAI,EAAE,CAAC;gBACb,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC;oBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,KAAK,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC;gBAEtE,MAAM,IAAI,GAAe,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;gBAC7C,IAAI,IAAI,KAAK,IAAI;oBAAE,MAAM;gBAEzB,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC;oBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBACpD,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC;oBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBAE9D,IAAI,KAAK,CAAC,MAAM,GAAG,KAAK;oBAAE,SAAS;gBAEnC,MAAM,OAAO,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CAC3E,EAAE,GAAG,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,CACxB,CAAC;gBAEF,MAAM,OAAO,GAAa,CAAC,MAAM,IAAI,CAAC,eAAe,CAAC,gBAAgB,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;qBACrF,GAAG,CAAC,CAAC,CAAO,EAAU,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;gBAEnC,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;oBACvB,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC;wBAAE,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;gBAC3C,CAAC;gBAED,KAAK,GAAG,EAAE,CAAC;YACZ,CAAC;YACD,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;QAC5B,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,SAAS,EAAE,uBAAuB,CAAC,CAAC;YACpF,IAAI,KAAK,GAAW,CAAC,CAAC;YACtB,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;gBAC9B,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,EAAE,CAAC,CAAC;gBAClD,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,CAAC;gBAClE,MAAM,IAAI,CAAC,eAAe,CAAC,QAAQ,EAAE,CAAC,UAAU,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC;YACxE,CAAC;YACD,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;IACF,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,IAAI,IAAI,CAAC,YAAY;YAAE,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;QAC1C,IAAI,IAAI,CAAC,YAAY;YAAE,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC;QAE1C,MAAM,SAAS,GAAqB,IAAI,CAAC,IAAI,CAAC,yBAAyB,CAAC,YAAY,CAAC,CAAC;QAEtF,MAAM,OAAO,GAAY,IAAI,OAAO,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,eAAe,CAAC,CAAC;QACvE,MAAM,OAAO,CAAC,eAAe,CAAC,SAAS,CAAC,CAAC;QACzC,MAAM,OAAO,CAAC,iBAAiB,EAAE,CAAC;IACnC,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 CrossPopulateExportApp extends CommonsApp implements IInternalHydraCommonDbApp {
|
|
5
5
|
private filename;
|
|
6
6
|
private databaseService;
|
|
@@ -0,0 +1,61 @@
|
|
|
1
|
+
import * as fs from 'fs';
|
|
2
|
+
import { EStatus } from 'hydra-crawler-esm-assets';
|
|
3
|
+
import { commonsLogDoing } from 'nodecommons-esm-log';
|
|
4
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
5
|
+
// export the QUEUED, DONE and FAILED urls for importing into another instance
|
|
6
|
+
export class CrossPopulateExportApp extends CommonsApp {
|
|
7
|
+
filename;
|
|
8
|
+
databaseService;
|
|
9
|
+
constructor(filename) {
|
|
10
|
+
super('hydra-crawler');
|
|
11
|
+
this.filename = filename;
|
|
12
|
+
}
|
|
13
|
+
getAppName() {
|
|
14
|
+
return 'Hydra - Cross Populate Export';
|
|
15
|
+
}
|
|
16
|
+
setDatabaseService(databaseService) {
|
|
17
|
+
this.databaseService = databaseService;
|
|
18
|
+
}
|
|
19
|
+
async init() {
|
|
20
|
+
if (!this.databaseService)
|
|
21
|
+
throw new Error('Database service has not been set yet');
|
|
22
|
+
const log = commonsLogDoing('cross-populate-export', 'Connecting to database');
|
|
23
|
+
await this.databaseService.init();
|
|
24
|
+
log.success();
|
|
25
|
+
await super.init();
|
|
26
|
+
}
|
|
27
|
+
async run() {
|
|
28
|
+
if (!this.databaseService)
|
|
29
|
+
throw new Error('Database service has not been set');
|
|
30
|
+
const results = this.databaseService.getUrls().find({
|
|
31
|
+
status: { $in: [
|
|
32
|
+
EStatus.QUEUED,
|
|
33
|
+
EStatus.DONE,
|
|
34
|
+
EStatus.FAILED
|
|
35
|
+
] }
|
|
36
|
+
}, {});
|
|
37
|
+
// safer to do this directly rather than a call to listQueryResults
|
|
38
|
+
// ditto not using CommonsFile
|
|
39
|
+
const fd = fs.openSync(this.filename, 'w');
|
|
40
|
+
const log = commonsLogDoing('cross-populate-export', 'Exporting QUEUED, DONE and FAILED urls');
|
|
41
|
+
let tally = 0;
|
|
42
|
+
while (true) {
|
|
43
|
+
if (tally % 1000 === 0)
|
|
44
|
+
log.progress(tally);
|
|
45
|
+
const row = await results.next();
|
|
46
|
+
if (row === null)
|
|
47
|
+
break;
|
|
48
|
+
fs.writeSync(fd, JSON.stringify({
|
|
49
|
+
domain: row.domain,
|
|
50
|
+
url: row.url
|
|
51
|
+
}) + '\n');
|
|
52
|
+
if (tally % 1000 === 0)
|
|
53
|
+
fs.fdatasyncSync(fd);
|
|
54
|
+
tally++;
|
|
55
|
+
}
|
|
56
|
+
fs.fdatasyncSync(fd);
|
|
57
|
+
fs.closeSync(fd);
|
|
58
|
+
log.result(tally);
|
|
59
|
+
}
|
|
60
|
+
}
|
|
61
|
+
//# sourceMappingURL=cross-populate-export.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"cross-populate-export.app.mjs","sourceRoot":"","sources":["../../src/apps/cross-populate-export.app.mts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,MAAM,IAAI,CAAC;AAKzB,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAEnD,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAMjD,8EAA8E;AAE9E,MAAM,OAAO,sBAAuB,SAAQ,UAAU;IAI3C;IAHF,eAAe,CAA4B;IAEnD,YACU,QAAgB;QAEzB,KAAK,CAAC,eAAe,CAAC,CAAC;QAFd,aAAQ,GAAR,QAAQ,CAAQ;IAG1B,CAAC;IAEM,UAAU;QAChB,OAAO,+BAA+B,CAAC;IACxC,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,uBAAuB,EAAE,wBAAwB,CAAC,CAAC;QACnG,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,OAAO,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CAAO;YAClF,MAAM,EAAE,EAAE,GAAG,EAAE;oBACb,OAAO,CAAC,MAAM;oBACd,OAAO,CAAC,IAAI;oBACZ,OAAO,CAAC,MAAM;iBACf,EAAE;SACJ,EAAE,EAAE,CAAC,CAAC;QAEP,mEAAmE;QACnE,8BAA8B;QAE9B,MAAM,EAAE,GAAW,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;QAEnD,MAAM,GAAG,GAAuB,eAAe,CAAC,uBAAuB,EAAE,wCAAwC,CAAC,CAAC;QAEnH,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,OAAO,IAAI,EAAE,CAAC;YACb,IAAI,KAAK,GAAG,IAAI,KAAK,CAAC;gBAAE,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;YAE5C,MAAM,GAAG,GAAc,MAAM,OAAO,CAAC,IAAI,EAAE,CAAC;YAC5C,IAAI,GAAG,KAAK,IAAI;gBAAE,MAAM;YAExB,EAAE,CAAC,SAAS,CACV,EAAE,EACF,IAAI,CAAC,SAAS,CAAC;gBACb,MAAM,EAAE,GAAG,CAAC,MAAM;gBAClB,GAAG,EAAE,GAAG,CAAC,GAAG;aACb,CAAC,GAAG,IAAI,CACV,CAAC;YAEF,IAAI,KAAK,GAAG,IAAI,KAAK,CAAC;gBAAE,EAAE,CAAC,aAAa,CAAC,EAAE,CAAC,CAAC;YAC7C,KAAK,EAAE,CAAC;QACT,CAAC;QAED,EAAE,CAAC,aAAa,CAAC,EAAE,CAAC,CAAC;QACrB,EAAE,CAAC,SAAS,CAAC,EAAE,CAAC,CAAC;QAEjB,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;IACnB,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 CrossPopulateImportApp extends CommonsApp implements IInternalHydraCommonDbApp {
|
|
5
5
|
private filename;
|
|
6
6
|
private databaseService;
|