hydra-crawler 2.8.4 → 3.0.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/apis/{autocomplete.api.d.ts → autocomplete.api.d.mts} +4 -4
- package/dist/apis/autocomplete.api.mjs +14 -0
- package/dist/apis/autocomplete.api.mjs.map +1 -0
- package/dist/apis/{bugs.api.d.ts → bugs.api.d.mts} +4 -4
- package/dist/apis/bugs.api.mjs +18 -0
- package/dist/apis/bugs.api.mjs.map +1 -0
- package/dist/apis/{crawl.api.d.ts → crawl.api.d.mts} +4 -4
- package/dist/apis/crawl.api.mjs +17 -0
- package/dist/apis/crawl.api.mjs.map +1 -0
- package/dist/apis/{domains.api.d.ts → domains.api.d.mts} +4 -4
- package/dist/apis/{domains.api.js → domains.api.mjs} +10 -18
- package/dist/apis/domains.api.mjs.map +1 -0
- package/dist/apis/{images.api.d.ts → images.api.d.mts} +4 -4
- package/dist/apis/images.api.mjs +19 -0
- package/dist/apis/images.api.mjs.map +1 -0
- package/dist/apis/{statistics.api.d.ts → statistics.api.d.mts} +4 -4
- package/dist/apis/statistics.api.mjs +33 -0
- package/dist/apis/statistics.api.mjs.map +1 -0
- package/dist/apis/{test.api.d.ts → test.api.d.mts} +3 -3
- package/dist/apis/test.api.mjs +10 -0
- package/dist/apis/test.api.mjs.map +1 -0
- package/dist/apis/{urls.api.d.ts → urls.api.d.mts} +4 -4
- package/dist/apis/urls.api.mjs +21 -0
- package/dist/apis/urls.api.mjs.map +1 -0
- package/dist/apps/{cleanup.app.d.ts → cleanup.app.d.mts} +5 -5
- package/dist/apps/cleanup.app.mjs +129 -0
- package/dist/apps/cleanup.app.mjs.map +1 -0
- package/dist/apps/{cross-populate-export.app.d.ts → cross-populate-export.app.d.mts} +3 -3
- package/dist/apps/cross-populate-export.app.mjs +61 -0
- package/dist/apps/cross-populate-export.app.mjs.map +1 -0
- package/dist/apps/{cross-populate-import.app.d.ts → cross-populate-import.app.d.mts} +3 -3
- package/dist/apps/cross-populate-import.app.mjs +61 -0
- package/dist/apps/cross-populate-import.app.mjs.map +1 -0
- package/dist/apps/{denylist.app.d.ts → denylist.app.d.mts} +5 -5
- package/dist/apps/denylist.app.mjs +135 -0
- package/dist/apps/denylist.app.mjs.map +1 -0
- package/dist/apps/{expire.app.d.ts → expire.app.d.mts} +8 -7
- package/dist/apps/expire.app.mjs +51 -0
- package/dist/apps/expire.app.mjs.map +1 -0
- package/dist/apps/{export-domain-urls.d.ts → export-domain-urls.d.mts} +3 -3
- package/dist/apps/export-domain-urls.mjs +85 -0
- package/dist/apps/export-domain-urls.mjs.map +1 -0
- package/dist/apps/{extract-text.app.d.ts → extract-text.app.d.mts} +2 -2
- package/dist/apps/extract-text.app.mjs +43 -0
- package/dist/apps/extract-text.app.mjs.map +1 -0
- package/dist/apps/{hydra.app.d.ts → hydra.app.d.mts} +10 -10
- package/dist/apps/hydra.app.mjs +222 -0
- package/dist/apps/hydra.app.mjs.map +1 -0
- package/dist/apps/{import.app.d.ts → import.app.d.mts} +3 -3
- package/dist/apps/import.app.mjs +48 -0
- package/dist/apps/import.app.mjs.map +1 -0
- package/dist/apps/{internal-hydra-common.app.d.ts → internal-hydra-common.app.d.mts} +6 -6
- package/dist/apps/{internal-hydra-common.app.js → internal-hydra-common.app.mjs} +1 -1
- package/dist/apps/internal-hydra-common.app.mjs.map +1 -0
- package/dist/apps/{move-to-archive.app.d.ts → move-to-archive.app.d.mts} +2 -2
- package/dist/apps/move-to-archive.app.mjs +31 -0
- package/dist/apps/move-to-archive.app.mjs.map +1 -0
- package/dist/apps/{prune-archive.app.d.ts → prune-archive.app.d.mts} +2 -2
- package/dist/apps/prune-archive.app.mjs +40 -0
- package/dist/apps/prune-archive.app.mjs.map +1 -0
- package/dist/apps/{query.app.d.ts → query.app.d.mts} +6 -6
- package/dist/apps/query.app.mjs +69 -0
- package/dist/apps/query.app.mjs.map +1 -0
- package/dist/apps/{reattempt.app.d.ts → reattempt.app.d.mts} +6 -6
- package/dist/apps/reattempt.app.mjs +81 -0
- package/dist/apps/reattempt.app.mjs.map +1 -0
- package/dist/apps/{requeue-domain.app.d.ts → requeue-domain.app.d.mts} +3 -3
- package/dist/apps/requeue-domain.app.mjs +52 -0
- package/dist/apps/requeue-domain.app.mjs.map +1 -0
- package/dist/apps/{seed.app.d.ts → seed.app.d.mts} +5 -5
- package/dist/apps/seed.app.mjs +59 -0
- package/dist/apps/seed.app.mjs.map +1 -0
- package/dist/apps/{startup.app.d.ts → startup.app.d.mts} +3 -3
- package/dist/apps/startup.app.mjs +55 -0
- package/dist/apps/startup.app.mjs.map +1 -0
- package/dist/apps/{unarchive-urls.app.d.ts → unarchive-urls.app.d.mts} +5 -5
- package/dist/apps/unarchive-urls.app.mjs +78 -0
- package/dist/apps/unarchive-urls.app.mjs.map +1 -0
- package/dist/classes/{cleaner.d.ts → cleaner.d.mts} +2 -2
- package/dist/classes/cleaner.mjs +264 -0
- package/dist/classes/cleaner.mjs.map +1 -0
- package/dist/classes/{crawler.d.ts → crawler.d.mts} +10 -12
- package/dist/classes/{crawler.js → crawler.mjs} +216 -184
- package/dist/classes/crawler.mjs.map +1 -0
- package/dist/classes/{dns.js → dns.mjs} +4 -4
- package/dist/classes/dns.mjs.map +1 -0
- package/dist/classes/{expirer.d.ts → expirer.d.mts} +2 -2
- package/dist/classes/expirer.mjs +117 -0
- package/dist/classes/expirer.mjs.map +1 -0
- package/dist/classes/{expiry.d.ts → expiry.d.mts} +1 -1
- package/dist/classes/{expiry.js → expiry.mjs} +7 -9
- package/dist/classes/expiry.mjs.map +1 -0
- package/dist/classes/{lists.d.ts → lists.d.mts} +2 -2
- package/dist/classes/{lists.js → lists.mjs} +4 -4
- package/dist/classes/lists.mjs.map +1 -0
- package/dist/classes/{robot.d.ts → robot.d.mts} +3 -3
- package/dist/classes/robot.mjs +74 -0
- package/dist/classes/robot.mjs.map +1 -0
- package/dist/classes/{tracker.d.ts → tracker.d.mts} +3 -3
- package/dist/classes/tracker.mjs +101 -0
- package/dist/classes/tracker.mjs.map +1 -0
- package/dist/{cli.js → cli.mjs} +46 -58
- package/dist/cli.mjs.map +1 -0
- package/dist/enums/{eavailable-strategy.js → eavailable-strategy.mjs} +1 -1
- package/dist/enums/eavailable-strategy.mjs.map +1 -0
- package/dist/enums/{elist.js → elist.mjs} +1 -1
- package/dist/enums/elist.mjs.map +1 -0
- package/dist/enums/{eserver.js → eserver.mjs} +1 -1
- package/dist/enums/eserver.mjs.map +1 -0
- package/dist/enums/{ex-powered-by.js → ex-powered-by.mjs} +1 -1
- package/dist/enums/ex-powered-by.mjs.map +1 -0
- package/dist/helpers/{matcher.d.ts → matcher.d.mts} +1 -1
- package/dist/helpers/{matcher.js → matcher.mjs} +1 -1
- package/dist/helpers/matcher.mjs.map +1 -0
- package/dist/helpers/{random.d.ts → random.d.mts} +1 -1
- package/dist/helpers/{random.js → random.mjs} +1 -1
- package/dist/helpers/random.mjs.map +1 -0
- package/dist/helpers/{utf-decoder.d.ts → utf-decoder.d.mts} +0 -1
- package/dist/helpers/{utf-decoder.js → utf-decoder.mjs} +3 -3
- package/dist/helpers/utf-decoder.mjs.map +1 -0
- package/dist/interfaces/{iexpiry.d.ts → iexpiry.d.mts} +1 -1
- package/dist/interfaces/{iexpiry.js → iexpiry.mjs} +3 -3
- package/dist/interfaces/iexpiry.mjs.map +1 -0
- package/dist/interfaces/{imatch.js → imatch.mjs} +2 -2
- package/dist/interfaces/imatch.mjs.map +1 -0
- package/dist/interfaces/{iparser-config.js → iparser-config.mjs} +2 -2
- package/dist/interfaces/iparser-config.mjs.map +1 -0
- package/dist/interfaces/{iparser.d.ts → iparser.d.mts} +1 -1
- package/dist/interfaces/iparser.mjs +2 -0
- package/dist/interfaces/iparser.mjs.map +1 -0
- package/dist/interfaces/{irequest-outcome.d.ts → irequest-outcome.d.mts} +0 -2
- package/dist/interfaces/irequest-outcome.mjs +2 -0
- package/dist/interfaces/irequest-outcome.mjs.map +1 -0
- package/dist/interfaces/iserver.mjs +2 -0
- package/dist/interfaces/iserver.mjs.map +1 -0
- package/dist/parsers/accessibility-metrics.parser.d.mts +11 -0
- package/dist/parsers/accessibility-metrics.parser.mjs +30 -0
- package/dist/parsers/accessibility-metrics.parser.mjs.map +1 -0
- package/dist/parsers/asp-error.parser.d.mts +12 -0
- package/dist/parsers/asp-error.parser.mjs +38 -0
- package/dist/parsers/asp-error.parser.mjs.map +1 -0
- package/dist/parsers/{bad-words.parser.d.ts → bad-words.parser.d.mts} +6 -6
- package/dist/parsers/bad-words.parser.mjs +17 -0
- package/dist/parsers/bad-words.parser.mjs.map +1 -0
- package/dist/parsers/complex-english.parser.d.mts +15 -0
- package/dist/parsers/complex-english.parser.mjs +52 -0
- package/dist/parsers/complex-english.parser.mjs.map +1 -0
- package/dist/parsers/data.parser.d.mts +13 -0
- package/dist/parsers/{data.parser.js → data.parser.mjs} +8 -7
- package/dist/parsers/data.parser.mjs.map +1 -0
- package/dist/parsers/{dictionary.parser.d.ts → dictionary.parser.d.mts} +6 -6
- package/dist/parsers/dictionary.parser.mjs +63 -0
- package/dist/parsers/dictionary.parser.mjs.map +1 -0
- package/dist/parsers/html.parser.d.mts +13 -0
- package/dist/parsers/{html.parser.js → html.parser.mjs} +4 -3
- package/dist/parsers/html.parser.mjs.map +1 -0
- package/dist/parsers/hyperlinks.parser.d.mts +20 -0
- package/dist/parsers/hyperlinks.parser.mjs +104 -0
- package/dist/parsers/hyperlinks.parser.mjs.map +1 -0
- package/dist/parsers/image-tags.parser.d.mts +19 -0
- package/dist/parsers/image-tags.parser.mjs +42 -0
- package/dist/parsers/image-tags.parser.mjs.map +1 -0
- package/dist/parsers/{interest.parser.d.ts → interest.parser.d.mts} +7 -7
- package/dist/parsers/interest.parser.mjs +60 -0
- package/dist/parsers/interest.parser.mjs.map +1 -0
- package/dist/parsers/jpeg.parser.d.mts +11 -0
- package/dist/parsers/jpeg.parser.mjs +29 -0
- package/dist/parsers/jpeg.parser.mjs.map +1 -0
- package/dist/parsers/{llama-guard.parser.d.ts → llama-guard.parser.d.mts} +7 -7
- package/dist/parsers/llama-guard.parser.mjs +56 -0
- package/dist/parsers/llama-guard.parser.mjs.map +1 -0
- package/dist/parsers/{offence.parser.d.ts → offence.parser.d.mts} +7 -7
- package/dist/parsers/offence.parser.mjs +60 -0
- package/dist/parsers/offence.parser.mjs.map +1 -0
- package/dist/parsers/{ollama.parser.d.ts → ollama.parser.d.mts} +6 -6
- package/dist/parsers/ollama.parser.mjs +43 -0
- package/dist/parsers/ollama.parser.mjs.map +1 -0
- package/dist/parsers/{paragraphs.parser.d.ts → paragraphs.parser.d.mts} +5 -5
- package/dist/parsers/paragraphs.parser.mjs +38 -0
- package/dist/parsers/paragraphs.parser.mjs.map +1 -0
- package/dist/parsers/{parser.d.ts → parser.d.mts} +6 -6
- package/dist/parsers/parser.mjs +45 -0
- package/dist/parsers/parser.mjs.map +1 -0
- package/dist/parsers/php-error.parser.d.mts +12 -0
- package/dist/parsers/php-error.parser.mjs +42 -0
- package/dist/parsers/php-error.parser.mjs.map +1 -0
- package/dist/parsers/{phrase.parser.d.ts → phrase.parser.d.mts} +3 -3
- package/dist/parsers/phrase.parser.mjs +15 -0
- package/dist/parsers/phrase.parser.mjs.map +1 -0
- package/dist/parsers/{regex.parser.d.ts → regex.parser.d.mts} +3 -3
- package/dist/parsers/regex.parser.mjs +29 -0
- package/dist/parsers/regex.parser.mjs.map +1 -0
- package/dist/parsers/server.parser.d.mts +11 -0
- package/dist/parsers/server.parser.mjs +57 -0
- package/dist/parsers/server.parser.mjs.map +1 -0
- package/dist/parsers/{spelling.parser.d.ts → spelling.parser.d.mts} +6 -6
- package/dist/parsers/spelling.parser.mjs +17 -0
- package/dist/parsers/spelling.parser.mjs.map +1 -0
- package/dist/parsers/string.parser.d.mts +8 -0
- package/dist/parsers/{string.parser.js → string.parser.mjs} +5 -4
- package/dist/parsers/string.parser.mjs.map +1 -0
- package/dist/parsers/{text.parser.d.ts → text.parser.d.mts} +3 -3
- package/dist/parsers/text.parser.mjs +30 -0
- package/dist/parsers/text.parser.mjs.map +1 -0
- package/dist/parsers/{words.parser.d.ts → words.parser.d.mts} +3 -3
- package/dist/parsers/words.parser.mjs +29 -0
- package/dist/parsers/words.parser.mjs.map +1 -0
- package/dist/queries/complex-english.query.d.mts +2 -0
- package/dist/queries/{complex-english.query.js → complex-english.query.mjs} +26 -28
- package/dist/queries/complex-english.query.mjs.map +1 -0
- package/dist/queries/flash-content.query.d.mts +2 -0
- package/dist/queries/flash-content.query.mjs +80 -0
- package/dist/queries/flash-content.query.mjs.map +1 -0
- package/dist/queries/linking-to-domains.query.d.mts +2 -0
- package/dist/queries/linking-to-domains.query.mjs +128 -0
- package/dist/queries/linking-to-domains.query.mjs.map +1 -0
- package/dist/queries/llamaguard-unsafe-content.query.d.mts +2 -0
- package/dist/queries/llamaguard-unsafe-content.query.mjs +90 -0
- package/dist/queries/llamaguard-unsafe-content.query.mjs.map +1 -0
- package/dist/queries/readability-score.query.d.mts +2 -0
- package/dist/queries/{readability-score.query.js → readability-score.query.mjs} +11 -20
- package/dist/queries/readability-score.query.mjs.map +1 -0
- package/dist/servers/{crawl.server.d.ts → crawl.server.d.mts} +10 -10
- package/dist/servers/crawl.server.mjs +192 -0
- package/dist/servers/crawl.server.mjs.map +1 -0
- package/dist/servers/{express.server.d.ts → express.server.d.mts} +2 -3
- package/dist/servers/express.server.mjs +13 -0
- package/dist/servers/express.server.mjs.map +1 -0
- package/dist/servers/{maintenance.server.d.ts → maintenance.server.d.mts} +5 -5
- package/dist/servers/maintenance.server.mjs +97 -0
- package/dist/servers/maintenance.server.mjs.map +1 -0
- package/dist/servers/{rest.server.d.ts → rest.server.d.mts} +4 -4
- package/dist/servers/rest.server.mjs +62 -0
- package/dist/servers/rest.server.mjs.map +1 -0
- package/dist/servers/{socket-io.server.d.ts → socket-io.server.d.mts} +4 -4
- package/dist/servers/socket-io.server.mjs +22 -0
- package/dist/servers/socket-io.server.mjs.map +1 -0
- package/dist/services/{database.service.d.ts → database.service.d.mts} +12 -12
- package/dist/services/database.service.mjs +645 -0
- package/dist/services/database.service.mjs.map +1 -0
- package/dist/services/{ollama-rest.service.d.ts → ollama-rest.service.d.mts} +2 -2
- package/dist/services/ollama-rest.service.mjs +27 -0
- package/dist/services/ollama-rest.service.mjs.map +1 -0
- package/dist/services/{rig-llama-guard.service.d.ts → rig-llama-guard.service.d.mts} +2 -2
- package/dist/services/{rig-llama-guard.service.js → rig-llama-guard.service.mjs} +33 -43
- package/dist/services/rig-llama-guard.service.mjs.map +1 -0
- package/dist/services/{rig-queue-length.service.d.ts → rig-queue-length.service.d.mts} +1 -1
- package/dist/services/rig-queue-length.service.mjs +12 -0
- package/dist/services/rig-queue-length.service.mjs.map +1 -0
- package/dist/services/{rig-subjectivity-scale.service.d.ts → rig-subjectivity-scale.service.d.mts} +2 -2
- package/dist/services/rig-subjectivity-scale.service.mjs +82 -0
- package/dist/services/rig-subjectivity-scale.service.mjs.map +1 -0
- package/dist/services/{rig-ticket-rest.service.d.ts → rig-ticket-rest.service.d.mts} +3 -3
- package/dist/services/rig-ticket-rest.service.mjs +15 -0
- package/dist/services/rig-ticket-rest.service.mjs.map +1 -0
- package/dist/services/{rig-ticketed-promise.service.d.ts → rig-ticketed-promise.service.d.mts} +4 -6
- package/dist/services/{rig-ticketed-promise.service.js → rig-ticketed-promise.service.mjs} +7 -6
- package/dist/services/rig-ticketed-promise.service.mjs.map +1 -0
- package/dist/types/{tcrawl-config.js → tcrawl-config.mjs} +2 -2
- package/dist/types/tcrawl-config.mjs.map +1 -0
- package/dist/types/{thydra-config.js → thydra-config.mjs} +2 -2
- package/dist/types/thydra-config.mjs.map +1 -0
- package/dist/types/tparser-ctor.d.mts +7 -0
- package/dist/types/tparser-ctor.mjs +2 -0
- package/dist/types/tparser-ctor.mjs.map +1 -0
- package/dist/types/tquery.d.mts +7 -0
- package/dist/types/tquery.mjs +2 -0
- package/dist/types/tquery.mjs.map +1 -0
- package/dist/types/tqueue-length.mjs +2 -0
- package/dist/types/tqueue-length.mjs.map +1 -0
- package/dist/types/{trobots-config.js → trobots-config.mjs} +2 -2
- package/dist/types/trobots-config.mjs.map +1 -0
- package/package.json +37 -43
- package/dist/apis/autocomplete.api.js +0 -22
- package/dist/apis/autocomplete.api.js.map +0 -1
- package/dist/apis/bugs.api.js +0 -26
- package/dist/apis/bugs.api.js.map +0 -1
- package/dist/apis/crawl.api.js +0 -25
- package/dist/apis/crawl.api.js.map +0 -1
- package/dist/apis/domains.api.js.map +0 -1
- package/dist/apis/images.api.js +0 -27
- package/dist/apis/images.api.js.map +0 -1
- package/dist/apis/statistics.api.js +0 -41
- package/dist/apis/statistics.api.js.map +0 -1
- package/dist/apis/test.api.js +0 -19
- package/dist/apis/test.api.js.map +0 -1
- package/dist/apis/urls.api.js +0 -29
- package/dist/apis/urls.api.js.map +0 -1
- package/dist/apps/cleanup.app.js +0 -151
- package/dist/apps/cleanup.app.js.map +0 -1
- package/dist/apps/cross-populate-export.app.js +0 -75
- package/dist/apps/cross-populate-export.app.js.map +0 -1
- package/dist/apps/cross-populate-import.app.js +0 -100
- package/dist/apps/cross-populate-import.app.js.map +0 -1
- package/dist/apps/denylist.app.js +0 -132
- package/dist/apps/denylist.app.js.map +0 -1
- package/dist/apps/expire.app.js +0 -63
- package/dist/apps/expire.app.js.map +0 -1
- package/dist/apps/export-domain-urls.js +0 -99
- package/dist/apps/export-domain-urls.js.map +0 -1
- package/dist/apps/extract-text.app.js +0 -55
- package/dist/apps/extract-text.app.js.map +0 -1
- package/dist/apps/hydra.app.js +0 -218
- package/dist/apps/hydra.app.js.map +0 -1
- package/dist/apps/import.app.js +0 -57
- package/dist/apps/import.app.js.map +0 -1
- package/dist/apps/internal-hydra-common.app.js.map +0 -1
- package/dist/apps/move-to-archive.app.js +0 -46
- package/dist/apps/move-to-archive.app.js.map +0 -1
- package/dist/apps/prune-archive.app.js +0 -55
- package/dist/apps/prune-archive.app.js.map +0 -1
- package/dist/apps/query.app.js +0 -80
- package/dist/apps/query.app.js.map +0 -1
- package/dist/apps/reattempt.app.js +0 -83
- package/dist/apps/reattempt.app.js.map +0 -1
- package/dist/apps/requeue-domain.app.js +0 -64
- package/dist/apps/requeue-domain.app.js.map +0 -1
- package/dist/apps/seed.app.js +0 -69
- package/dist/apps/seed.app.js.map +0 -1
- package/dist/apps/startup.app.js +0 -64
- package/dist/apps/startup.app.js.map +0 -1
- package/dist/apps/unarchive-urls.app.js +0 -83
- package/dist/apps/unarchive-urls.app.js.map +0 -1
- package/dist/classes/cleaner.js +0 -266
- package/dist/classes/cleaner.js.map +0 -1
- package/dist/classes/crawler.js.map +0 -1
- package/dist/classes/dns.js.map +0 -1
- package/dist/classes/expirer.js +0 -121
- package/dist/classes/expirer.js.map +0 -1
- package/dist/classes/expiry.js.map +0 -1
- package/dist/classes/lists.js.map +0 -1
- package/dist/classes/robot.js +0 -82
- package/dist/classes/robot.js.map +0 -1
- package/dist/classes/tracker.js +0 -120
- package/dist/classes/tracker.js.map +0 -1
- package/dist/cli.js.map +0 -1
- package/dist/enums/eavailable-strategy.js.map +0 -1
- package/dist/enums/elist.js.map +0 -1
- package/dist/enums/eserver.js.map +0 -1
- package/dist/enums/ex-powered-by.js.map +0 -1
- package/dist/helpers/matcher.js.map +0 -1
- package/dist/helpers/random.js.map +0 -1
- package/dist/helpers/utf-decoder.js.map +0 -1
- package/dist/interfaces/iexpiry.js.map +0 -1
- package/dist/interfaces/imatch.js.map +0 -1
- package/dist/interfaces/iparser-config.js.map +0 -1
- package/dist/interfaces/iparser.js +0 -2
- package/dist/interfaces/iparser.js.map +0 -1
- package/dist/interfaces/irequest-outcome.js +0 -2
- package/dist/interfaces/irequest-outcome.js.map +0 -1
- package/dist/interfaces/iserver.js +0 -2
- package/dist/interfaces/iserver.js.map +0 -1
- package/dist/parsers/accessibility-metrics.parser.d.ts +0 -11
- package/dist/parsers/accessibility-metrics.parser.js +0 -40
- package/dist/parsers/accessibility-metrics.parser.js.map +0 -1
- package/dist/parsers/asp-error.parser.d.ts +0 -12
- package/dist/parsers/asp-error.parser.js +0 -48
- package/dist/parsers/asp-error.parser.js.map +0 -1
- package/dist/parsers/bad-words.parser.js +0 -27
- package/dist/parsers/bad-words.parser.js.map +0 -1
- package/dist/parsers/complex-english.parser.d.ts +0 -15
- package/dist/parsers/complex-english.parser.js +0 -61
- package/dist/parsers/complex-english.parser.js.map +0 -1
- package/dist/parsers/data.parser.d.ts +0 -14
- package/dist/parsers/data.parser.js.map +0 -1
- package/dist/parsers/dictionary.parser.js +0 -73
- package/dist/parsers/dictionary.parser.js.map +0 -1
- package/dist/parsers/html.parser.d.ts +0 -13
- package/dist/parsers/html.parser.js.map +0 -1
- package/dist/parsers/hyperlinks.parser.d.ts +0 -20
- package/dist/parsers/hyperlinks.parser.js +0 -115
- package/dist/parsers/hyperlinks.parser.js.map +0 -1
- package/dist/parsers/image-tags.parser.d.ts +0 -19
- package/dist/parsers/image-tags.parser.js +0 -52
- package/dist/parsers/image-tags.parser.js.map +0 -1
- package/dist/parsers/interest.parser.js +0 -69
- package/dist/parsers/interest.parser.js.map +0 -1
- package/dist/parsers/jpeg.parser.d.ts +0 -11
- package/dist/parsers/jpeg.parser.js +0 -39
- package/dist/parsers/jpeg.parser.js.map +0 -1
- package/dist/parsers/llama-guard.parser.js +0 -65
- package/dist/parsers/llama-guard.parser.js.map +0 -1
- package/dist/parsers/offence.parser.js +0 -69
- package/dist/parsers/offence.parser.js.map +0 -1
- package/dist/parsers/ollama.parser.js +0 -51
- package/dist/parsers/ollama.parser.js.map +0 -1
- package/dist/parsers/paragraphs.parser.js +0 -49
- package/dist/parsers/paragraphs.parser.js.map +0 -1
- package/dist/parsers/parser.js +0 -57
- package/dist/parsers/parser.js.map +0 -1
- package/dist/parsers/php-error.parser.d.ts +0 -12
- package/dist/parsers/php-error.parser.js +0 -52
- package/dist/parsers/php-error.parser.js.map +0 -1
- package/dist/parsers/phrase.parser.js +0 -26
- package/dist/parsers/phrase.parser.js.map +0 -1
- package/dist/parsers/regex.parser.js +0 -43
- package/dist/parsers/regex.parser.js.map +0 -1
- package/dist/parsers/server.parser.d.ts +0 -11
- package/dist/parsers/server.parser.js +0 -67
- package/dist/parsers/server.parser.js.map +0 -1
- package/dist/parsers/spelling.parser.js +0 -27
- package/dist/parsers/spelling.parser.js.map +0 -1
- package/dist/parsers/string.parser.d.ts +0 -8
- package/dist/parsers/string.parser.js.map +0 -1
- package/dist/parsers/text.parser.js +0 -41
- package/dist/parsers/text.parser.js.map +0 -1
- package/dist/parsers/words.parser.js +0 -40
- package/dist/parsers/words.parser.js.map +0 -1
- package/dist/queries/complex-english.query.d.ts +0 -2
- package/dist/queries/complex-english.query.js.map +0 -1
- package/dist/queries/flash-content.query.d.ts +0 -2
- package/dist/queries/flash-content.query.js +0 -82
- package/dist/queries/flash-content.query.js.map +0 -1
- package/dist/queries/linking-to-domains.query.d.ts +0 -2
- package/dist/queries/linking-to-domains.query.js +0 -130
- package/dist/queries/linking-to-domains.query.js.map +0 -1
- package/dist/queries/llamaguard-unsafe-content.query.d.ts +0 -2
- package/dist/queries/llamaguard-unsafe-content.query.js +0 -92
- package/dist/queries/llamaguard-unsafe-content.query.js.map +0 -1
- package/dist/queries/readability-score.query.d.ts +0 -2
- package/dist/queries/readability-score.query.js.map +0 -1
- package/dist/servers/crawl.server.js +0 -198
- package/dist/servers/crawl.server.js.map +0 -1
- package/dist/servers/express.server.js +0 -13
- package/dist/servers/express.server.js.map +0 -1
- package/dist/servers/maintenance.server.js +0 -103
- package/dist/servers/maintenance.server.js.map +0 -1
- package/dist/servers/rest.server.js +0 -46
- package/dist/servers/rest.server.js.map +0 -1
- package/dist/servers/socket-io.server.js +0 -61
- package/dist/servers/socket-io.server.js.map +0 -1
- package/dist/services/database.service.js +0 -795
- package/dist/services/database.service.js.map +0 -1
- package/dist/services/ollama-rest.service.js +0 -46
- package/dist/services/ollama-rest.service.js.map +0 -1
- package/dist/services/rig-llama-guard.service.js.map +0 -1
- package/dist/services/rig-queue-length.service.js +0 -22
- package/dist/services/rig-queue-length.service.js.map +0 -1
- package/dist/services/rig-subjectivity-scale.service.js +0 -96
- package/dist/services/rig-subjectivity-scale.service.js.map +0 -1
- package/dist/services/rig-ticket-rest.service.js +0 -29
- package/dist/services/rig-ticket-rest.service.js.map +0 -1
- package/dist/services/rig-ticketed-promise.service.js.map +0 -1
- package/dist/types/tcrawl-config.js.map +0 -1
- package/dist/types/thydra-config.js.map +0 -1
- package/dist/types/tparser-ctor.d.ts +0 -7
- package/dist/types/tparser-ctor.js +0 -2
- package/dist/types/tparser-ctor.js.map +0 -1
- package/dist/types/tquery.d.ts +0 -7
- package/dist/types/tquery.js +0 -2
- package/dist/types/tquery.js.map +0 -1
- package/dist/types/tqueue-length.js +0 -2
- package/dist/types/tqueue-length.js.map +0 -1
- package/dist/types/trobots-config.js.map +0 -1
- /package/dist/classes/{dns.d.ts → dns.d.mts} +0 -0
- /package/dist/{cli.d.ts → cli.d.mts} +0 -0
- /package/dist/enums/{eavailable-strategy.d.ts → eavailable-strategy.d.mts} +0 -0
- /package/dist/enums/{elist.d.ts → elist.d.mts} +0 -0
- /package/dist/enums/{eserver.d.ts → eserver.d.mts} +0 -0
- /package/dist/enums/{ex-powered-by.d.ts → ex-powered-by.d.mts} +0 -0
- /package/dist/interfaces/{imatch.d.ts → imatch.d.mts} +0 -0
- /package/dist/interfaces/{iparser-config.d.ts → iparser-config.d.mts} +0 -0
- /package/dist/interfaces/{iserver.d.ts → iserver.d.mts} +0 -0
- /package/dist/types/{tcrawl-config.d.ts → tcrawl-config.d.mts} +0 -0
- /package/dist/types/{thydra-config.d.ts → thydra-config.d.mts} +0 -0
- /package/dist/types/{tqueue-length.d.ts → tqueue-length.d.mts} +0 -0
- /package/dist/types/{trobots-config.d.ts → trobots-config.d.mts} +0 -0
|
@@ -0,0 +1,61 @@
|
|
|
1
|
+
import * as fs from 'fs';
|
|
2
|
+
import * as readline from 'readline';
|
|
3
|
+
import { commonsTypeHasPropertyString } from 'tscommons-esm-core';
|
|
4
|
+
import { commonsLogDoing, commonsLogError } from 'nodecommons-esm-log';
|
|
5
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
6
|
+
function isTRow(test) {
|
|
7
|
+
if (!commonsTypeHasPropertyString(test, 'domain'))
|
|
8
|
+
return false;
|
|
9
|
+
if (!commonsTypeHasPropertyString(test, 'url'))
|
|
10
|
+
return false;
|
|
11
|
+
return true;
|
|
12
|
+
}
|
|
13
|
+
export class CrossPopulateImportApp extends CommonsApp {
|
|
14
|
+
filename;
|
|
15
|
+
databaseService;
|
|
16
|
+
constructor(filename) {
|
|
17
|
+
super('hydra-crawler');
|
|
18
|
+
this.filename = filename;
|
|
19
|
+
}
|
|
20
|
+
getAppName() {
|
|
21
|
+
return 'Hydra - Cross Populate Export';
|
|
22
|
+
}
|
|
23
|
+
setDatabaseService(databaseService) {
|
|
24
|
+
this.databaseService = databaseService;
|
|
25
|
+
}
|
|
26
|
+
async init() {
|
|
27
|
+
if (!this.databaseService)
|
|
28
|
+
throw new Error('Database service has not been set yet');
|
|
29
|
+
const log = commonsLogDoing('cross-populate-import', 'Connecting to database');
|
|
30
|
+
await this.databaseService.init();
|
|
31
|
+
log.success();
|
|
32
|
+
await super.init();
|
|
33
|
+
}
|
|
34
|
+
async run() {
|
|
35
|
+
if (!this.databaseService)
|
|
36
|
+
throw new Error('Database service has not been set');
|
|
37
|
+
const fileStream = fs.createReadStream(this.filename);
|
|
38
|
+
const rl = readline.createInterface({
|
|
39
|
+
input: fileStream,
|
|
40
|
+
crlfDelay: Infinity
|
|
41
|
+
});
|
|
42
|
+
const log = commonsLogDoing('cross-populate-import', 'Importing urls');
|
|
43
|
+
let tally = 0;
|
|
44
|
+
for await (const line of rl) {
|
|
45
|
+
if (line.trim() === '')
|
|
46
|
+
continue;
|
|
47
|
+
if (tally % 1000 === 0)
|
|
48
|
+
log.progress(tally);
|
|
49
|
+
const json = JSON.parse(line);
|
|
50
|
+
if (!isTRow(json)) {
|
|
51
|
+
commonsLogError('Invalid row. Skipping');
|
|
52
|
+
continue;
|
|
53
|
+
}
|
|
54
|
+
await this.databaseService.queue(json.url, false);
|
|
55
|
+
tally++;
|
|
56
|
+
}
|
|
57
|
+
log.result(tally);
|
|
58
|
+
fileStream.close();
|
|
59
|
+
}
|
|
60
|
+
}
|
|
61
|
+
//# sourceMappingURL=cross-populate-import.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"cross-populate-import.app.mjs","sourceRoot":"","sources":["../../src/apps/cross-populate-import.app.mts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,MAAM,IAAI,CAAC;AACzB,OAAO,KAAK,QAAQ,MAAM,UAAU,CAAC;AAErC,OAAO,EAAE,4BAA4B,EAAE,MAAM,oBAAoB,CAAC;AAElE,OAAO,EAAsB,eAAe,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC3F,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAajD,SAAS,MAAM,CAAC,IAAa;IAC5B,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,QAAQ,CAAC;QAAE,OAAO,KAAK,CAAC;IAChE,IAAI,CAAC,4BAA4B,CAAC,IAAI,EAAE,KAAK,CAAC;QAAE,OAAO,KAAK,CAAC;IAE7D,OAAO,IAAI,CAAC;AACb,CAAC;AAED,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,UAAU,GAAkB,EAAE,CAAC,gBAAgB,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAErE,MAAM,EAAE,GAAuB,QAAQ,CAAC,eAAe,CAAC;YACtD,KAAK,EAAE,UAAU;YACjB,SAAS,EAAE,QAAQ;SACpB,CAAC,CAAC;QAEH,MAAM,GAAG,GAAuB,eAAe,CAAC,uBAAuB,EAAE,gBAAgB,CAAC,CAAC;QAE3F,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,EAAE,EAAE,CAAC;YAC7B,IAAI,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE;gBAAE,SAAS;YAEjC,IAAI,KAAK,GAAG,IAAI,KAAK,CAAC;gBAAE,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;YAE5C,MAAM,IAAI,GAAY,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;YACvC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC;gBACnB,eAAe,CAAC,uBAAuB,CAAC,CAAC;gBACzC,SAAS;YACV,CAAC;YAED,MAAM,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;YAClD,KAAK,EAAE,CAAC;QACT,CAAC;QACD,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QAElB,UAAU,CAAC,KAAK,EAAE,CAAC;IACpB,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 DenylistApp extends CommonsApp implements IInternalHydraCommonListApp {
|
|
7
7
|
private databaseService;
|
|
8
8
|
private lists;
|
|
@@ -0,0 +1,135 @@
|
|
|
1
|
+
import { commonsArrayChunk } from 'tscommons-esm-core';
|
|
2
|
+
import { EStatus } from 'hydra-crawler-esm-assets';
|
|
3
|
+
import { commonsLogDoing } from 'nodecommons-esm-log';
|
|
4
|
+
import { CommonsArgs } from 'nodecommons-esm-cli';
|
|
5
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
6
|
+
import { Lists } from '../classes/lists.mjs';
|
|
7
|
+
import { EList } from '../enums/elist.mjs';
|
|
8
|
+
export class DenylistApp extends CommonsApp {
|
|
9
|
+
databaseService;
|
|
10
|
+
lists;
|
|
11
|
+
constructor() {
|
|
12
|
+
super('hydra-crawler');
|
|
13
|
+
this.lists = new Lists();
|
|
14
|
+
}
|
|
15
|
+
getAppName() {
|
|
16
|
+
return 'Hydra - Denylist';
|
|
17
|
+
}
|
|
18
|
+
setDatabaseService(databaseService) {
|
|
19
|
+
this.databaseService = databaseService;
|
|
20
|
+
}
|
|
21
|
+
addToList(list, entries) {
|
|
22
|
+
this.lists.add(list, entries);
|
|
23
|
+
}
|
|
24
|
+
async init() {
|
|
25
|
+
if (!this.databaseService)
|
|
26
|
+
throw new Error('Database service has not been set yet');
|
|
27
|
+
const log = commonsLogDoing('denylist', 'Connecting to database');
|
|
28
|
+
await this.databaseService.init();
|
|
29
|
+
log.success();
|
|
30
|
+
await super.init();
|
|
31
|
+
}
|
|
32
|
+
async forward(limit) {
|
|
33
|
+
if (!this.databaseService)
|
|
34
|
+
throw new Error('Database service has not been set');
|
|
35
|
+
let tally = 0;
|
|
36
|
+
let found = 0;
|
|
37
|
+
const urls = [];
|
|
38
|
+
{ // scope
|
|
39
|
+
const log = commonsLogDoing('denylist', 'Searching for non-DENY urls on the denylist');
|
|
40
|
+
const result = this.databaseService.getUrls()
|
|
41
|
+
.find({ status: { $nin: [EStatus.DENY, EStatus.ARCHIVED] } }, {});
|
|
42
|
+
while (true) {
|
|
43
|
+
tally++;
|
|
44
|
+
if ((tally % 100) === 0)
|
|
45
|
+
log.progress(`${tally}, ${found}`);
|
|
46
|
+
if (limit !== undefined && tally > limit)
|
|
47
|
+
break;
|
|
48
|
+
const row = await result.next();
|
|
49
|
+
if (row === null)
|
|
50
|
+
break;
|
|
51
|
+
try {
|
|
52
|
+
if (!this.lists.match(EList.DENY, row.url))
|
|
53
|
+
continue;
|
|
54
|
+
}
|
|
55
|
+
catch (_e) {
|
|
56
|
+
// probably an invalid URL
|
|
57
|
+
// so mark it as denied
|
|
58
|
+
}
|
|
59
|
+
found++;
|
|
60
|
+
urls.push(row.url);
|
|
61
|
+
}
|
|
62
|
+
log.result(found);
|
|
63
|
+
}
|
|
64
|
+
{ // scope
|
|
65
|
+
const log = commonsLogDoing('denylist', 'Marking detected as DENY');
|
|
66
|
+
const batches = commonsArrayChunk(urls, 100);
|
|
67
|
+
tally = 0;
|
|
68
|
+
for (const batch of batches) {
|
|
69
|
+
await this.databaseService.getUrls().updateMany({ url: { $in: batch } }, { $set: {
|
|
70
|
+
status: EStatus.DENY
|
|
71
|
+
}, $unset: {
|
|
72
|
+
ttl: true
|
|
73
|
+
} });
|
|
74
|
+
tally += 100;
|
|
75
|
+
log.progress(tally);
|
|
76
|
+
}
|
|
77
|
+
log.success();
|
|
78
|
+
}
|
|
79
|
+
}
|
|
80
|
+
async reverse(limit) {
|
|
81
|
+
if (!this.databaseService)
|
|
82
|
+
throw new Error('Database service has not been set');
|
|
83
|
+
let tally = 0;
|
|
84
|
+
let found = 0;
|
|
85
|
+
const urls = [];
|
|
86
|
+
{ // scope
|
|
87
|
+
const log = commonsLogDoing('denylist', 'Searching for DENY urls not on the denylist');
|
|
88
|
+
const result = this.databaseService.getUrls()
|
|
89
|
+
.find({ status: EStatus.DENY }, {});
|
|
90
|
+
while (true) {
|
|
91
|
+
tally++;
|
|
92
|
+
if ((tally % 100) === 0)
|
|
93
|
+
log.progress(`${tally}, ${found}`);
|
|
94
|
+
if (limit !== undefined && tally > limit)
|
|
95
|
+
break;
|
|
96
|
+
const row = await result.next();
|
|
97
|
+
if (row === null)
|
|
98
|
+
break;
|
|
99
|
+
try {
|
|
100
|
+
if (this.lists.match(EList.DENY, row.url))
|
|
101
|
+
continue;
|
|
102
|
+
}
|
|
103
|
+
catch (_e) {
|
|
104
|
+
// probably an invalid URL
|
|
105
|
+
// so don't re-enable it
|
|
106
|
+
continue;
|
|
107
|
+
}
|
|
108
|
+
found++;
|
|
109
|
+
urls.push(row.url);
|
|
110
|
+
}
|
|
111
|
+
log.result(found);
|
|
112
|
+
}
|
|
113
|
+
{ // scope
|
|
114
|
+
const log = commonsLogDoing('denylist', 'Marking detected as QUEUED');
|
|
115
|
+
const batches = commonsArrayChunk(urls, 100);
|
|
116
|
+
tally = 0;
|
|
117
|
+
for (const batch of batches) {
|
|
118
|
+
await this.databaseService.getUrls().updateMany({ url: { $in: batch } }, { $set: {
|
|
119
|
+
status: EStatus.QUEUED
|
|
120
|
+
}, $unset: {
|
|
121
|
+
ttl: true
|
|
122
|
+
} });
|
|
123
|
+
tally += 100;
|
|
124
|
+
log.progress(tally);
|
|
125
|
+
}
|
|
126
|
+
log.success();
|
|
127
|
+
}
|
|
128
|
+
}
|
|
129
|
+
async run() {
|
|
130
|
+
const args = new CommonsArgs();
|
|
131
|
+
await this.forward(args.getFirstNumberOrUndefined('limit'));
|
|
132
|
+
await this.reverse(args.getFirstNumberOrUndefined('limit'));
|
|
133
|
+
}
|
|
134
|
+
}
|
|
135
|
+
//# sourceMappingURL=denylist.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"denylist.app.mjs","sourceRoot":"","sources":["../../src/apps/denylist.app.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,iBAAiB,EAAE,MAAM,oBAAoB,CAAC;AAEvD,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAGnD,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,WAAW,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAEjD,OAAO,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAM7C,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAC;AAI3C,MAAM,OAAO,WAAY,SAAQ,UAAU;IAClC,eAAe,CAA4B;IAE3C,KAAK,CAAQ;IAErB;QACC,KAAK,CAAC,eAAe,CAAC,CAAC;QAEvB,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;IAC1B,CAAC;IAEM,UAAU;QAChB,OAAO,kBAAkB,CAAC;IAC3B,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,UAAU,EAAE,wBAAwB,CAAC,CAAC;QACtF,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,OAAO,CAAC,KAAc;QACnC,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,IAAI,KAAK,GAAW,CAAC,CAAC;QAEtB,MAAM,IAAI,GAAa,EAAE,CAAC;QAC1B,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,UAAU,EAAE,6CAA6C,CAAC,CAAC;YAE3G,MAAM,MAAM,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBACpE,IAAI,CACH,EAAE,MAAM,EAAE,EAAE,IAAI,EAAE,CAAE,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC,QAAQ,CAAE,EAAE,EAAE,EACxD,EAAE,CACH,CAAC;YAEJ,OAAO,IAAI,EAAE,CAAC;gBACb,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,KAAK,KAAK,EAAE,CAAC,CAAC;gBAC5D,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,GAAG,KAAK;oBAAE,MAAM;gBAEhD,MAAM,GAAG,GAAc,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;gBAC3C,IAAI,GAAG,KAAK,IAAI;oBAAE,MAAM;gBAExB,IAAI,CAAC;oBACJ,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,EAAE,GAAG,CAAC,GAAG,CAAC;wBAAE,SAAS;gBACtD,CAAC;gBAAC,OAAO,EAAE,EAAE,CAAC;oBACb,0BAA0B;oBAC1B,uBAAuB;gBACxB,CAAC;gBAED,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACpB,CAAC;YAED,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACnB,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,UAAU,EAAE,0BAA0B,CAAC,CAAC;YAExF,MAAM,OAAO,GAAe,iBAAiB,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;YAEzD,KAAK,GAAG,CAAC,CAAC;YACV,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;gBAC7B,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,UAAU,CAC7C,EAAE,GAAG,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,EACvB,EAAE,IAAI,EAAE;wBACN,MAAM,EAAE,OAAO,CAAC,IAAI;qBACrB,EAAE,MAAM,EAAE;wBACT,GAAG,EAAE,IAAI;qBACV,EAAE,CACJ,CAAC;gBAEF,KAAK,IAAI,GAAG,CAAC;gBACb,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;YACrB,CAAC;YACD,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;IACF,CAAC;IAEO,KAAK,CAAC,OAAO,CAAC,KAAc;QACnC,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,IAAI,KAAK,GAAW,CAAC,CAAC;QAEtB,MAAM,IAAI,GAAa,EAAE,CAAC;QAC1B,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,UAAU,EAAE,6CAA6C,CAAC,CAAC;YAE3G,MAAM,MAAM,GAA6B,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;iBACpE,IAAI,CACH,EAAE,MAAM,EAAE,OAAO,CAAC,IAAI,EAAE,EACxB,EAAE,CACH,CAAC;YAEJ,OAAO,IAAI,EAAE,CAAC;gBACb,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;oBAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,KAAK,KAAK,EAAE,CAAC,CAAC;gBAC5D,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,GAAG,KAAK;oBAAE,MAAM;gBAEhD,MAAM,GAAG,GAAc,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;gBAC3C,IAAI,GAAG,KAAK,IAAI;oBAAE,MAAM;gBAExB,IAAI,CAAC;oBACJ,IAAI,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,EAAE,GAAG,CAAC,GAAG,CAAC;wBAAE,SAAS;gBACrD,CAAC;gBAAC,OAAO,EAAE,EAAE,CAAC;oBACb,0BAA0B;oBAC1B,wBAAwB;oBACxB,SAAS;gBACV,CAAC;gBAED,KAAK,EAAE,CAAC;gBACR,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACpB,CAAC;YAED,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACnB,CAAC;QAED,CAAC,CAAC,QAAQ;YACT,MAAM,GAAG,GAAuB,eAAe,CAAC,UAAU,EAAE,4BAA4B,CAAC,CAAC;YAE1F,MAAM,OAAO,GAAe,iBAAiB,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;YAEzD,KAAK,GAAG,CAAC,CAAC;YACV,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;gBAC7B,MAAM,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE,CAAC,UAAU,CAC7C,EAAE,GAAG,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,EACvB,EAAE,IAAI,EAAE;wBACN,MAAM,EAAE,OAAO,CAAC,MAAM;qBACvB,EAAE,MAAM,EAAE;wBACT,GAAG,EAAE,IAAI;qBACV,EAAE,CACJ,CAAC;gBAEF,KAAK,IAAI,GAAG,CAAC;gBACb,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;YACrB,CAAC;YACD,GAAG,CAAC,OAAO,EAAE,CAAC;QACf,CAAC;IACF,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,MAAM,IAAI,GAAgB,IAAI,WAAW,EAAE,CAAC;QAE5C,MAAM,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,yBAAyB,CAAC,OAAO,CAAC,CAAC,CAAC;QAC5D,MAAM,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,yBAAyB,CAAC,OAAO,CAAC,CAAC,CAAC;IAC7D,CAAC;CACD"}
|
|
@@ -1,15 +1,16 @@
|
|
|
1
|
-
import {
|
|
2
|
-
import {
|
|
3
|
-
import {
|
|
4
|
-
import {
|
|
5
|
-
import {
|
|
6
|
-
import {
|
|
1
|
+
import { CommonsFixedDate } from 'tscommons-esm-core';
|
|
2
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
3
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
4
|
+
import { IMatch } from '../interfaces/imatch.mjs';
|
|
5
|
+
import { IExpiry } from '../interfaces/iexpiry.mjs';
|
|
6
|
+
import { EList } from '../enums/elist.mjs';
|
|
7
|
+
import { IInternalHydraCommonListApp, IInternalHydraCommonExpiryApp } from './internal-hydra-common.app.mjs';
|
|
7
8
|
export declare class ExpireApp extends CommonsApp implements IInternalHydraCommonListApp, IInternalHydraCommonExpiryApp {
|
|
8
9
|
private threshold;
|
|
9
10
|
private databaseService;
|
|
10
11
|
private lists;
|
|
11
12
|
private expiry;
|
|
12
|
-
constructor(threshold:
|
|
13
|
+
constructor(threshold: CommonsFixedDate | undefined);
|
|
13
14
|
getAppName(): string;
|
|
14
15
|
setDatabaseService(databaseService: DatabaseService): void;
|
|
15
16
|
addToList(list: EList, entries: IMatch[]): void;
|
|
@@ -0,0 +1,51 @@
|
|
|
1
|
+
import { commonsLogDoing } from 'nodecommons-esm-log';
|
|
2
|
+
import { CommonsArgs } from 'nodecommons-esm-cli';
|
|
3
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
4
|
+
import { Lists } from '../classes/lists.mjs';
|
|
5
|
+
import { Expirer } from '../classes/expirer.mjs';
|
|
6
|
+
import { Expiry } from '../classes/expiry.mjs';
|
|
7
|
+
export class ExpireApp extends CommonsApp {
|
|
8
|
+
threshold;
|
|
9
|
+
databaseService;
|
|
10
|
+
lists;
|
|
11
|
+
expiry;
|
|
12
|
+
constructor(threshold) {
|
|
13
|
+
super('hydra-crawler');
|
|
14
|
+
this.threshold = threshold;
|
|
15
|
+
this.lists = new Lists();
|
|
16
|
+
this.expiry = new Expiry();
|
|
17
|
+
}
|
|
18
|
+
getAppName() {
|
|
19
|
+
return 'Hydra - Expire';
|
|
20
|
+
}
|
|
21
|
+
setDatabaseService(databaseService) {
|
|
22
|
+
this.databaseService = databaseService;
|
|
23
|
+
}
|
|
24
|
+
addToList(list, entries) {
|
|
25
|
+
this.lists.add(list, entries);
|
|
26
|
+
}
|
|
27
|
+
addToExpiry(expiries) {
|
|
28
|
+
this.expiry.add(expiries);
|
|
29
|
+
}
|
|
30
|
+
async init() {
|
|
31
|
+
if (!this.databaseService)
|
|
32
|
+
throw new Error('Database service has not been set yet');
|
|
33
|
+
const log = commonsLogDoing('expire', 'Connecting to database');
|
|
34
|
+
await this.databaseService.init();
|
|
35
|
+
log.success();
|
|
36
|
+
await super.init();
|
|
37
|
+
}
|
|
38
|
+
async run() {
|
|
39
|
+
if (!this.databaseService)
|
|
40
|
+
throw new Error('Database service has not been set');
|
|
41
|
+
const args = new CommonsArgs();
|
|
42
|
+
const expirer = new Expirer(this.expiry, this.databaseService);
|
|
43
|
+
if (this.threshold !== undefined) {
|
|
44
|
+
await expirer.expireFixed(this.threshold.UTCDate);
|
|
45
|
+
}
|
|
46
|
+
else {
|
|
47
|
+
await expirer.expire(args.getFirstNumberOrUndefined('limit'));
|
|
48
|
+
}
|
|
49
|
+
}
|
|
50
|
+
}
|
|
51
|
+
//# sourceMappingURL=expire.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"expire.app.mjs","sourceRoot":"","sources":["../../src/apps/expire.app.mts"],"names":[],"mappings":"AAEA,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,WAAW,EAAE,MAAM,qBAAqB,CAAC;AAClD,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAEjD,OAAO,EAAE,KAAK,EAAE,MAAM,sBAAsB,CAAC;AAC7C,OAAO,EAAE,OAAO,EAAE,MAAM,wBAAwB,CAAC;AACjD,OAAO,EAAE,MAAM,EAAE,MAAM,uBAAuB,CAAC;AAW/C,MAAM,OAAO,SAAU,SAAQ,UAAU;IAM9B;IALF,eAAe,CAA4B;IAC3C,KAAK,CAAQ;IACb,MAAM,CAAS;IAEvB,YACU,SAAqC;QAE9C,KAAK,CAAC,eAAe,CAAC,CAAC;QAFd,cAAS,GAAT,SAAS,CAA4B;QAI9C,IAAI,CAAC,KAAK,GAAG,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,CAAC,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;IAC5B,CAAC;IAEM,UAAU;QAChB,OAAO,gBAAgB,CAAC;IACzB,CAAC;IAEM,kBAAkB,CACvB,eAAgC;QAEjC,IAAI,CAAC,eAAe,GAAG,eAAe,CAAC;IACxC,CAAC;IAEM,SAAS,CACd,IAAW,EACX,OAAiB;QAElB,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/B,CAAC;IAEM,WAAW,CAChB,QAAmB;QAEpB,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;IAC3B,CAAC;IAEe,KAAK,CAAC,IAAI;QACzB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC;QAEpF,MAAM,GAAG,GAAuB,eAAe,CAAC,QAAQ,EAAE,wBAAwB,CAAC,CAAC;QACpF,MAAM,IAAI,CAAC,eAAe,CAAC,IAAI,EAAE,CAAC;QAClC,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,KAAK,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,IAAI,CAAC,IAAI,CAAC,eAAe;YAAE,MAAM,IAAI,KAAK,CAAC,mCAAmC,CAAC,CAAC;QAEhF,MAAM,IAAI,GAAgB,IAAI,WAAW,EAAE,CAAC;QAE5C,MAAM,OAAO,GAAY,IAAI,OAAO,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,eAAe,CAAC,CAAC;QAExE,IAAI,IAAI,CAAC,SAAS,KAAK,SAAS,EAAE,CAAC;YAClC,MAAM,OAAO,CAAC,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC;QACnD,CAAC;aAAM,CAAC;YACP,MAAM,OAAO,CAAC,MAAM,CAClB,IAAI,CAAC,yBAAyB,CAAC,OAAO,CAAC,CACxC,CAAC;QACH,CAAC;IACF,CAAC;CACD"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { CommonsApp } from 'nodecommons-
|
|
2
|
-
import { DatabaseService } from '../services/database.service';
|
|
3
|
-
import { IInternalHydraCommonDbApp } from './internal-hydra-common.app';
|
|
1
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
2
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
3
|
+
import { IInternalHydraCommonDbApp } from './internal-hydra-common.app.mjs';
|
|
4
4
|
export declare class ExportDomainUrlsApp extends CommonsApp implements IInternalHydraCommonDbApp {
|
|
5
5
|
private domain;
|
|
6
6
|
private databaseService;
|
|
@@ -0,0 +1,85 @@
|
|
|
1
|
+
import { EStatus } from 'hydra-crawler-esm-assets';
|
|
2
|
+
import { commonsLogDoing } from 'nodecommons-esm-log';
|
|
3
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
4
|
+
// export the QUEUED, DONE and FAILED urls for importing into another instance
|
|
5
|
+
export class ExportDomainUrlsApp extends CommonsApp {
|
|
6
|
+
domain;
|
|
7
|
+
databaseService;
|
|
8
|
+
constructor(domain) {
|
|
9
|
+
super('hydra-crawler');
|
|
10
|
+
this.domain = domain;
|
|
11
|
+
}
|
|
12
|
+
getAppName() {
|
|
13
|
+
return 'Hydra - Export Domain Urls';
|
|
14
|
+
}
|
|
15
|
+
setDatabaseService(databaseService) {
|
|
16
|
+
this.databaseService = databaseService;
|
|
17
|
+
}
|
|
18
|
+
async init() {
|
|
19
|
+
if (!this.databaseService)
|
|
20
|
+
throw new Error('Database service has not been set yet');
|
|
21
|
+
const log = commonsLogDoing('export-domain-urls', 'Connecting to database');
|
|
22
|
+
await this.databaseService.init();
|
|
23
|
+
log.success();
|
|
24
|
+
await super.init();
|
|
25
|
+
}
|
|
26
|
+
async run() {
|
|
27
|
+
if (!this.databaseService)
|
|
28
|
+
throw new Error('Database service has not been set');
|
|
29
|
+
const validOnly = this.args.hasAttribute('valid-only');
|
|
30
|
+
const pagesOnly = this.args.hasAttribute('pages-only');
|
|
31
|
+
const mergeHttp = this.args.hasAttribute('merge-http');
|
|
32
|
+
const matchPipeline = [
|
|
33
|
+
{ $match: { domain: this.domain } }
|
|
34
|
+
];
|
|
35
|
+
if (validOnly) {
|
|
36
|
+
matchPipeline.push({ $match: {
|
|
37
|
+
status: { $in: [EStatus.DONE, EStatus.QUEUED, EStatus.ACTIVE] },
|
|
38
|
+
statusCode: { $in: [200, 204] }
|
|
39
|
+
} });
|
|
40
|
+
}
|
|
41
|
+
if (pagesOnly) {
|
|
42
|
+
matchPipeline.push(...[
|
|
43
|
+
{ $match: {
|
|
44
|
+
url: { $not: /[?]/ }
|
|
45
|
+
} },
|
|
46
|
+
{ $match: {
|
|
47
|
+
url: { $not: /(gif|jpeg|jpg|png|webp|avif)$/i }
|
|
48
|
+
} }
|
|
49
|
+
]);
|
|
50
|
+
}
|
|
51
|
+
const cursor = this.databaseService.getUrls()
|
|
52
|
+
.aggregate([
|
|
53
|
+
...matchPipeline,
|
|
54
|
+
{ $project: { _id: false, url: true } }
|
|
55
|
+
]);
|
|
56
|
+
// safer to do this directly rather than a call to listQueryResults
|
|
57
|
+
// ditto not using CommonsFile
|
|
58
|
+
const log = commonsLogDoing('export-domain-urls', `Exporting URLs for domain ${this.domain}`);
|
|
59
|
+
const urls = [];
|
|
60
|
+
while (true) {
|
|
61
|
+
if (urls.length % 1000 === 0)
|
|
62
|
+
log.progress(urls.length);
|
|
63
|
+
const row = await cursor.next();
|
|
64
|
+
if (row === null)
|
|
65
|
+
break;
|
|
66
|
+
let url = row.url
|
|
67
|
+
.trim()
|
|
68
|
+
.replace(/[?].*$/, '')
|
|
69
|
+
.replace(/\/index\.(htm|html)$/i, '/')
|
|
70
|
+
.replace(/\/$/, '')
|
|
71
|
+
.trim();
|
|
72
|
+
if (mergeHttp) {
|
|
73
|
+
url = url.replace(/^http(s?):/, '');
|
|
74
|
+
}
|
|
75
|
+
if (!urls.includes(url))
|
|
76
|
+
urls.push(url);
|
|
77
|
+
}
|
|
78
|
+
log.result(urls.length);
|
|
79
|
+
urls.sort();
|
|
80
|
+
for (const url of urls) {
|
|
81
|
+
console.log(url);
|
|
82
|
+
}
|
|
83
|
+
}
|
|
84
|
+
}
|
|
85
|
+
//# sourceMappingURL=export-domain-urls.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"export-domain-urls.mjs","sourceRoot":"","sources":["../../src/apps/export-domain-urls.mts"],"names":[],"mappings":"AAEA,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,mBAAoB,SAAQ,UAAU;IAIxC;IAHF,eAAe,CAA4B;IAEnD,YACU,MAAc;QAEvB,KAAK,CAAC,eAAe,CAAC,CAAC;QAFd,WAAM,GAAN,MAAM,CAAQ;IAGxB,CAAC;IAEM,UAAU;QAChB,OAAO,4BAA4B,CAAC;IACrC,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,oBAAoB,EAAE,wBAAwB,CAAC,CAAC;QAChG,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,SAAS,GAAY,IAAI,CAAC,IAAI,CAAC,YAAY,CAAC,YAAY,CAAC,CAAC;QAChE,MAAM,SAAS,GAAY,IAAI,CAAC,IAAI,CAAC,YAAY,CAAC,YAAY,CAAC,CAAC;QAChE,MAAM,SAAS,GAAY,IAAI,CAAC,IAAI,CAAC,YAAY,CAAC,YAAY,CAAC,CAAC;QAEhE,MAAM,aAAa,GAA4B;YAC7C,EAAE,MAAM,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,EAAE;SACpC,CAAC;QACF,IAAI,SAAS,EAAE,CAAC;YACf,aAAa,CAAC,IAAI,CAChB,EAAE,MAAM,EAAE;oBACR,MAAM,EAAE,EAAE,GAAG,EAAE,CAAE,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC,MAAM,EAAE,OAAO,CAAC,MAAM,CAAE,EAAE;oBACjE,UAAU,EAAE,EAAE,GAAG,EAAE,CAAE,GAAG,EAAE,GAAG,CAAE,EAAE;iBAClC,EAAE,CACJ,CAAC;QACH,CAAC;QACD,IAAI,SAAS,EAAE,CAAC;YACf,aAAa,CAAC,IAAI,CAAC,GAAG;gBACpB,EAAE,MAAM,EAAE;wBACR,GAAG,EAAE,EAAE,IAAI,EAAE,KAAK,EAAE;qBACrB,EAAE;gBACH,EAAE,MAAM,EAAE;wBACR,GAAG,EAAE,EAAE,IAAI,EAAE,gCAAgC,EAAE;qBAChD,EAAE;aACJ,CAAC,CAAC;QACJ,CAAC;QAED,MAAM,MAAM,GAA+C,IAAI,CAAC,eAAe,CAAC,OAAO,EAAE;aACtF,SAAS,CAAC;YACT,GAAG,aAAa;YAChB,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,EAAE;SACxC,CAAC,CAAC;QAEL,mEAAmE;QACnE,8BAA8B;QAE9B,MAAM,GAAG,GAAuB,eAAe,CAAC,oBAAoB,EAAE,6BAA6B,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;QAElH,MAAM,IAAI,GAAa,EAAE,CAAC;QAC1B,OAAO,IAAI,EAAE,CAAC;YACb,IAAI,IAAI,CAAC,MAAM,GAAG,IAAI,KAAK,CAAC;gBAAE,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YAExD,MAAM,GAAG,GAAyB,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;YACtD,IAAI,GAAG,KAAK,IAAI;gBAAE,MAAM;YAExB,IAAI,GAAG,GAAW,GAAG,CAAC,GAAG;iBACtB,IAAI,EAAE;iBACN,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC;iBACrB,OAAO,CAAC,uBAAuB,EAAE,GAAG,CAAC;iBACrC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC;iBAClB,IAAI,EAAE,CAAC;YAEV,IAAI,SAAS,EAAE,CAAC;gBACf,GAAG,GAAG,GAAG,CAAC,OAAO,CAAC,YAAY,EAAE,EAAE,CAAC,CAAC;YACrC,CAAC;YAED,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;gBAAE,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACzC,CAAC;QACD,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAExB,IAAI,CAAC,IAAI,EAAE,CAAC;QACZ,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;YACxB,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QAClB,CAAC;IACF,CAAC;CACD"}
|
|
@@ -1,5 +1,5 @@
|
|
|
1
|
-
import { CommonsApp } from 'nodecommons-
|
|
2
|
-
import { IInternalHydraCommonApp } from './internal-hydra-common.app';
|
|
1
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
2
|
+
import { IInternalHydraCommonApp } from './internal-hydra-common.app.mjs';
|
|
3
3
|
export declare class ExtractTextApp extends CommonsApp implements IInternalHydraCommonApp {
|
|
4
4
|
private url;
|
|
5
5
|
constructor(url: string);
|
|
@@ -0,0 +1,43 @@
|
|
|
1
|
+
import * as http from 'http';
|
|
2
|
+
import * as https from 'https';
|
|
3
|
+
import { commonsLogDoing } from 'nodecommons-esm-log';
|
|
4
|
+
import { CommonsApp } from 'nodecommons-esm-app';
|
|
5
|
+
import { Crawler } from '../classes/crawler.mjs';
|
|
6
|
+
import { TextParser } from '../parsers/text.parser.mjs';
|
|
7
|
+
// extract the text from the HTML of a URL
|
|
8
|
+
class TextParserInstance extends TextParser {
|
|
9
|
+
// eslint-disable-next-line @typescript-eslint/require-await
|
|
10
|
+
async parseText(_database, text) {
|
|
11
|
+
console.log(text);
|
|
12
|
+
}
|
|
13
|
+
}
|
|
14
|
+
export class ExtractTextApp extends CommonsApp {
|
|
15
|
+
url;
|
|
16
|
+
constructor(url) {
|
|
17
|
+
super('hydra-crawler');
|
|
18
|
+
this.url = url;
|
|
19
|
+
}
|
|
20
|
+
getAppName() {
|
|
21
|
+
return 'Hydra - Extract Text';
|
|
22
|
+
}
|
|
23
|
+
async run() {
|
|
24
|
+
const log = commonsLogDoing('extract-text', `Downloading from URL at ${this.url}`);
|
|
25
|
+
const whatwg = new URL(this.url);
|
|
26
|
+
let handler;
|
|
27
|
+
switch (whatwg.protocol) {
|
|
28
|
+
case 'http:':
|
|
29
|
+
handler = http;
|
|
30
|
+
break;
|
|
31
|
+
case 'https:':
|
|
32
|
+
handler = https;
|
|
33
|
+
break;
|
|
34
|
+
default:
|
|
35
|
+
throw new Error(`unable to handle protocol ${whatwg.protocol}`);
|
|
36
|
+
}
|
|
37
|
+
const outcome = await Crawler.request(handler, this.url, 10000, 4096000);
|
|
38
|
+
log.success();
|
|
39
|
+
const parser = new TextParserInstance(outcome);
|
|
40
|
+
await parser.parse(undefined);
|
|
41
|
+
}
|
|
42
|
+
}
|
|
43
|
+
//# sourceMappingURL=extract-text.app.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"extract-text.app.mjs","sourceRoot":"","sources":["../../src/apps/extract-text.app.mts"],"names":[],"mappings":"AAAA,OAAO,KAAK,IAAI,MAAM,MAAM,CAAC;AAC7B,OAAO,KAAK,KAAK,MAAM,OAAO,CAAC;AAE/B,OAAO,EAAsB,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAEjD,OAAO,EAAE,OAAO,EAAE,MAAM,wBAAwB,CAAC;AAIjD,OAAO,EAAE,UAAU,EAAE,MAAM,4BAA4B,CAAC;AAOxD,0CAA0C;AAE1C,MAAM,kBAAmB,SAAQ,UAAyB;IACzD,4DAA4D;IAClD,KAAK,CAAC,SAAS,CACvB,SAA0B,EAC1B,IAAY;QAEb,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACnB,CAAC;CACD;AAED,MAAM,OAAO,cAAe,SAAQ,UAAU;IAEnC;IADV,YACU,GAAW;QAEpB,KAAK,CAAC,eAAe,CAAC,CAAC;QAFd,QAAG,GAAH,GAAG,CAAQ;IAGrB,CAAC;IAEM,UAAU;QAChB,OAAO,sBAAsB,CAAC;IAC/B,CAAC;IAEe,KAAK,CAAC,GAAG;QACxB,MAAM,GAAG,GAAuB,eAAe,CAAC,cAAc,EAAE,2BAA2B,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC;QAEvG,MAAM,MAAM,GAAQ,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAEtC,IAAI,OAA2C,CAAC;QAChD,QAAQ,MAAM,CAAC,QAAQ,EAAE,CAAC;YACzB,KAAK,OAAO;gBAAE,OAAO,GAAG,IAAI,CAAC;gBAAC,MAAM;YACpC,KAAK,QAAQ;gBAAE,OAAO,GAAG,KAAK,CAAC;gBAAC,MAAM;YACtC;gBACC,MAAM,IAAI,KAAK,CAAC,6BAA6B,MAAM,CAAC,QAAQ,EAAE,CAAC,CAAC;QAClE,CAAC;QAED,MAAM,OAAO,GAAoB,MAAM,OAAO,CAAC,OAAO,CACpD,OAAO,EACP,IAAI,CAAC,GAAG,EACR,KAAK,EACL,OAAO,CACR,CAAC;QACF,GAAG,CAAC,OAAO,EAAE,CAAC;QAEd,MAAM,MAAM,GAAuB,IAAI,kBAAkB,CACvD,OAAO,CACR,CAAC;QACF,MAAM,MAAM,CAAC,KAAK,CAAC,SAAuC,CAAC,CAAC;IAC7D,CAAC;CACD"}
|
|
@@ -1,13 +1,13 @@
|
|
|
1
|
-
import { TCommonsScheduleTime } from 'tscommons-
|
|
2
|
-
import { ICommonsExpressConfig } from 'nodecommons-
|
|
3
|
-
import { CommonsStrictExpressServer } from 'nodecommons-
|
|
4
|
-
import { CommonsSocketIoApp } from 'nodecommons-
|
|
5
|
-
import { SocketIoServer } from '../servers/socket-io.server';
|
|
6
|
-
import { DatabaseService } from '../services/database.service';
|
|
7
|
-
import { IMatch } from '../interfaces/imatch';
|
|
8
|
-
import { IExpiry } from '../interfaces/iexpiry';
|
|
9
|
-
import { EList } from '../enums/elist';
|
|
10
|
-
import { IInternalHydraCommonListApp, IInternalHydraCommonExpiryApp, IInternalHydraCommonMaintenanceApp } from './internal-hydra-common.app';
|
|
1
|
+
import { TCommonsScheduleTime } from 'tscommons-esm-async';
|
|
2
|
+
import { ICommonsExpressConfig } from 'nodecommons-esm-express';
|
|
3
|
+
import { CommonsStrictExpressServer } from 'nodecommons-esm-express';
|
|
4
|
+
import { CommonsSocketIoApp } from 'nodecommons-esm-app-socket-io';
|
|
5
|
+
import { SocketIoServer } from '../servers/socket-io.server.mjs';
|
|
6
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
7
|
+
import { IMatch } from '../interfaces/imatch.mjs';
|
|
8
|
+
import { IExpiry } from '../interfaces/iexpiry.mjs';
|
|
9
|
+
import { EList } from '../enums/elist.mjs';
|
|
10
|
+
import { IInternalHydraCommonListApp, IInternalHydraCommonExpiryApp, IInternalHydraCommonMaintenanceApp } from './internal-hydra-common.app.mjs';
|
|
11
11
|
export declare class HydraApp extends CommonsSocketIoApp<SocketIoServer> implements IInternalHydraCommonListApp, IInternalHydraCommonExpiryApp, IInternalHydraCommonMaintenanceApp {
|
|
12
12
|
private databaseService;
|
|
13
13
|
private lists;
|