hydra-crawler 2.8.3 → 3.0.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/apis/{autocomplete.api.d.ts → autocomplete.api.d.mts} +4 -4
- package/dist/apis/autocomplete.api.mjs +14 -0
- package/dist/apis/autocomplete.api.mjs.map +1 -0
- package/dist/apis/{bugs.api.d.ts → bugs.api.d.mts} +4 -4
- package/dist/apis/bugs.api.mjs +18 -0
- package/dist/apis/bugs.api.mjs.map +1 -0
- package/dist/apis/{crawl.api.d.ts → crawl.api.d.mts} +4 -4
- package/dist/apis/crawl.api.mjs +17 -0
- package/dist/apis/crawl.api.mjs.map +1 -0
- package/dist/apis/{domains.api.d.ts → domains.api.d.mts} +4 -4
- package/dist/apis/{domains.api.js → domains.api.mjs} +10 -18
- package/dist/apis/domains.api.mjs.map +1 -0
- package/dist/apis/{images.api.d.ts → images.api.d.mts} +4 -4
- package/dist/apis/images.api.mjs +19 -0
- package/dist/apis/images.api.mjs.map +1 -0
- package/dist/apis/{statistics.api.d.ts → statistics.api.d.mts} +4 -4
- package/dist/apis/statistics.api.mjs +33 -0
- package/dist/apis/statistics.api.mjs.map +1 -0
- package/dist/apis/{test.api.d.ts → test.api.d.mts} +3 -3
- package/dist/apis/test.api.mjs +10 -0
- package/dist/apis/test.api.mjs.map +1 -0
- package/dist/apis/{urls.api.d.ts → urls.api.d.mts} +4 -4
- package/dist/apis/urls.api.mjs +21 -0
- package/dist/apis/urls.api.mjs.map +1 -0
- package/dist/apps/{cleanup.app.d.ts → cleanup.app.d.mts} +5 -5
- package/dist/apps/cleanup.app.mjs +129 -0
- package/dist/apps/cleanup.app.mjs.map +1 -0
- package/dist/apps/{cross-populate-export.app.d.ts → cross-populate-export.app.d.mts} +3 -3
- package/dist/apps/cross-populate-export.app.mjs +61 -0
- package/dist/apps/cross-populate-export.app.mjs.map +1 -0
- package/dist/apps/{cross-populate-import.app.d.ts → cross-populate-import.app.d.mts} +3 -3
- package/dist/apps/cross-populate-import.app.mjs +61 -0
- package/dist/apps/cross-populate-import.app.mjs.map +1 -0
- package/dist/apps/{denylist.app.d.ts → denylist.app.d.mts} +5 -5
- package/dist/apps/denylist.app.mjs +122 -0
- package/dist/apps/denylist.app.mjs.map +1 -0
- package/dist/apps/{expire.app.d.ts → expire.app.d.mts} +8 -7
- package/dist/apps/expire.app.mjs +51 -0
- package/dist/apps/expire.app.mjs.map +1 -0
- package/dist/apps/{export-domain-urls.d.ts → export-domain-urls.d.mts} +3 -3
- package/dist/apps/export-domain-urls.mjs +85 -0
- package/dist/apps/export-domain-urls.mjs.map +1 -0
- package/dist/apps/{extract-text.app.d.ts → extract-text.app.d.mts} +2 -2
- package/dist/apps/extract-text.app.mjs +43 -0
- package/dist/apps/extract-text.app.mjs.map +1 -0
- package/dist/apps/{hydra.app.d.ts → hydra.app.d.mts} +10 -10
- package/dist/apps/hydra.app.mjs +222 -0
- package/dist/apps/hydra.app.mjs.map +1 -0
- package/dist/apps/{import.app.d.ts → import.app.d.mts} +3 -3
- package/dist/apps/import.app.mjs +48 -0
- package/dist/apps/import.app.mjs.map +1 -0
- package/dist/apps/{internal-hydra-common.app.d.ts → internal-hydra-common.app.d.mts} +6 -6
- package/dist/apps/{internal-hydra-common.app.js → internal-hydra-common.app.mjs} +1 -1
- package/dist/apps/internal-hydra-common.app.mjs.map +1 -0
- package/dist/apps/{move-to-archive.app.d.ts → move-to-archive.app.d.mts} +2 -2
- package/dist/apps/move-to-archive.app.mjs +31 -0
- package/dist/apps/move-to-archive.app.mjs.map +1 -0
- package/dist/apps/{prune-archive.app.d.ts → prune-archive.app.d.mts} +2 -2
- package/dist/apps/prune-archive.app.mjs +40 -0
- package/dist/apps/prune-archive.app.mjs.map +1 -0
- package/dist/apps/{query.app.d.ts → query.app.d.mts} +6 -6
- package/dist/apps/query.app.mjs +69 -0
- package/dist/apps/query.app.mjs.map +1 -0
- package/dist/apps/{reattempt.app.d.ts → reattempt.app.d.mts} +6 -6
- package/dist/apps/reattempt.app.mjs +74 -0
- package/dist/apps/reattempt.app.mjs.map +1 -0
- package/dist/apps/{requeue-domain.app.d.ts → requeue-domain.app.d.mts} +3 -3
- package/dist/apps/requeue-domain.app.mjs +52 -0
- package/dist/apps/requeue-domain.app.mjs.map +1 -0
- package/dist/apps/{seed.app.d.ts → seed.app.d.mts} +5 -5
- package/dist/apps/seed.app.mjs +59 -0
- package/dist/apps/seed.app.mjs.map +1 -0
- package/dist/apps/{startup.app.d.ts → startup.app.d.mts} +3 -3
- package/dist/apps/startup.app.mjs +55 -0
- package/dist/apps/startup.app.mjs.map +1 -0
- package/dist/apps/{unarchive-urls.app.d.ts → unarchive-urls.app.d.mts} +5 -5
- package/dist/apps/unarchive-urls.app.mjs +71 -0
- package/dist/apps/unarchive-urls.app.mjs.map +1 -0
- package/dist/classes/{cleaner.d.ts → cleaner.d.mts} +2 -2
- package/dist/classes/cleaner.mjs +257 -0
- package/dist/classes/cleaner.mjs.map +1 -0
- package/dist/classes/{crawler.d.ts → crawler.d.mts} +10 -12
- package/dist/classes/crawler.mjs +418 -0
- package/dist/classes/crawler.mjs.map +1 -0
- package/dist/classes/{dns.js → dns.mjs} +4 -4
- package/dist/classes/dns.mjs.map +1 -0
- package/dist/classes/{expirer.d.ts → expirer.d.mts} +2 -2
- package/dist/classes/expirer.mjs +117 -0
- package/dist/classes/expirer.mjs.map +1 -0
- package/dist/classes/{expiry.d.ts → expiry.d.mts} +1 -1
- package/dist/classes/{expiry.js → expiry.mjs} +7 -9
- package/dist/classes/expiry.mjs.map +1 -0
- package/dist/classes/{lists.d.ts → lists.d.mts} +2 -2
- package/dist/classes/{lists.js → lists.mjs} +4 -4
- package/dist/classes/lists.mjs.map +1 -0
- package/dist/classes/{robot.d.ts → robot.d.mts} +3 -3
- package/dist/classes/robot.mjs +74 -0
- package/dist/classes/robot.mjs.map +1 -0
- package/dist/classes/{tracker.d.ts → tracker.d.mts} +3 -3
- package/dist/classes/tracker.mjs +101 -0
- package/dist/classes/tracker.mjs.map +1 -0
- package/dist/{cli.js → cli.mjs} +46 -58
- package/dist/cli.mjs.map +1 -0
- package/dist/enums/{eavailable-strategy.js → eavailable-strategy.mjs} +1 -1
- package/dist/enums/eavailable-strategy.mjs.map +1 -0
- package/dist/enums/{elist.js → elist.mjs} +1 -1
- package/dist/enums/elist.mjs.map +1 -0
- package/dist/enums/{eserver.js → eserver.mjs} +1 -1
- package/dist/enums/eserver.mjs.map +1 -0
- package/dist/enums/{ex-powered-by.js → ex-powered-by.mjs} +1 -1
- package/dist/enums/ex-powered-by.mjs.map +1 -0
- package/dist/helpers/{matcher.d.ts → matcher.d.mts} +1 -1
- package/dist/helpers/{matcher.js → matcher.mjs} +1 -1
- package/dist/helpers/matcher.mjs.map +1 -0
- package/dist/helpers/{random.d.ts → random.d.mts} +1 -1
- package/dist/helpers/{random.js → random.mjs} +1 -1
- package/dist/helpers/random.mjs.map +1 -0
- package/dist/helpers/{utf-decoder.d.ts → utf-decoder.d.mts} +0 -1
- package/dist/helpers/{utf-decoder.js → utf-decoder.mjs} +3 -3
- package/dist/helpers/utf-decoder.mjs.map +1 -0
- package/dist/interfaces/{iexpiry.d.ts → iexpiry.d.mts} +1 -1
- package/dist/interfaces/{iexpiry.js → iexpiry.mjs} +3 -3
- package/dist/interfaces/iexpiry.mjs.map +1 -0
- package/dist/interfaces/{imatch.js → imatch.mjs} +2 -2
- package/dist/interfaces/imatch.mjs.map +1 -0
- package/dist/interfaces/{iparser-config.js → iparser-config.mjs} +2 -2
- package/dist/interfaces/iparser-config.mjs.map +1 -0
- package/dist/interfaces/{iparser.d.ts → iparser.d.mts} +1 -1
- package/dist/interfaces/iparser.mjs +2 -0
- package/dist/interfaces/iparser.mjs.map +1 -0
- package/dist/interfaces/{irequest-outcome.d.ts → irequest-outcome.d.mts} +0 -2
- package/dist/interfaces/irequest-outcome.mjs +2 -0
- package/dist/interfaces/irequest-outcome.mjs.map +1 -0
- package/dist/interfaces/iserver.mjs +2 -0
- package/dist/interfaces/iserver.mjs.map +1 -0
- package/dist/parsers/accessibility-metrics.parser.d.mts +11 -0
- package/dist/parsers/accessibility-metrics.parser.mjs +30 -0
- package/dist/parsers/accessibility-metrics.parser.mjs.map +1 -0
- package/dist/parsers/asp-error.parser.d.mts +12 -0
- package/dist/parsers/asp-error.parser.mjs +38 -0
- package/dist/parsers/asp-error.parser.mjs.map +1 -0
- package/dist/parsers/{bad-words.parser.d.ts → bad-words.parser.d.mts} +6 -6
- package/dist/parsers/bad-words.parser.mjs +17 -0
- package/dist/parsers/bad-words.parser.mjs.map +1 -0
- package/dist/parsers/complex-english.parser.d.mts +15 -0
- package/dist/parsers/complex-english.parser.mjs +52 -0
- package/dist/parsers/complex-english.parser.mjs.map +1 -0
- package/dist/parsers/data.parser.d.mts +13 -0
- package/dist/parsers/{data.parser.js → data.parser.mjs} +8 -7
- package/dist/parsers/data.parser.mjs.map +1 -0
- package/dist/parsers/{dictionary.parser.d.ts → dictionary.parser.d.mts} +6 -6
- package/dist/parsers/dictionary.parser.mjs +63 -0
- package/dist/parsers/dictionary.parser.mjs.map +1 -0
- package/dist/parsers/html.parser.d.mts +13 -0
- package/dist/parsers/{html.parser.js → html.parser.mjs} +4 -3
- package/dist/parsers/html.parser.mjs.map +1 -0
- package/dist/parsers/hyperlinks.parser.d.mts +20 -0
- package/dist/parsers/hyperlinks.parser.mjs +104 -0
- package/dist/parsers/hyperlinks.parser.mjs.map +1 -0
- package/dist/parsers/image-tags.parser.d.mts +19 -0
- package/dist/parsers/image-tags.parser.mjs +42 -0
- package/dist/parsers/image-tags.parser.mjs.map +1 -0
- package/dist/parsers/{interest.parser.d.ts → interest.parser.d.mts} +7 -7
- package/dist/parsers/interest.parser.mjs +60 -0
- package/dist/parsers/interest.parser.mjs.map +1 -0
- package/dist/parsers/jpeg.parser.d.mts +11 -0
- package/dist/parsers/jpeg.parser.mjs +29 -0
- package/dist/parsers/jpeg.parser.mjs.map +1 -0
- package/dist/parsers/{llama-guard.parser.d.ts → llama-guard.parser.d.mts} +7 -7
- package/dist/parsers/llama-guard.parser.mjs +56 -0
- package/dist/parsers/llama-guard.parser.mjs.map +1 -0
- package/dist/parsers/{offence.parser.d.ts → offence.parser.d.mts} +7 -7
- package/dist/parsers/offence.parser.mjs +60 -0
- package/dist/parsers/offence.parser.mjs.map +1 -0
- package/dist/parsers/{ollama.parser.d.ts → ollama.parser.d.mts} +6 -6
- package/dist/parsers/ollama.parser.mjs +43 -0
- package/dist/parsers/ollama.parser.mjs.map +1 -0
- package/dist/parsers/{paragraphs.parser.d.ts → paragraphs.parser.d.mts} +5 -5
- package/dist/parsers/paragraphs.parser.mjs +38 -0
- package/dist/parsers/paragraphs.parser.mjs.map +1 -0
- package/dist/parsers/{parser.d.ts → parser.d.mts} +6 -6
- package/dist/parsers/parser.mjs +45 -0
- package/dist/parsers/parser.mjs.map +1 -0
- package/dist/parsers/php-error.parser.d.mts +12 -0
- package/dist/parsers/php-error.parser.mjs +42 -0
- package/dist/parsers/php-error.parser.mjs.map +1 -0
- package/dist/parsers/{phrase.parser.d.ts → phrase.parser.d.mts} +3 -3
- package/dist/parsers/phrase.parser.mjs +15 -0
- package/dist/parsers/phrase.parser.mjs.map +1 -0
- package/dist/parsers/{regex.parser.d.ts → regex.parser.d.mts} +3 -3
- package/dist/parsers/regex.parser.mjs +29 -0
- package/dist/parsers/regex.parser.mjs.map +1 -0
- package/dist/parsers/server.parser.d.mts +11 -0
- package/dist/parsers/server.parser.mjs +57 -0
- package/dist/parsers/server.parser.mjs.map +1 -0
- package/dist/parsers/{spelling.parser.d.ts → spelling.parser.d.mts} +6 -6
- package/dist/parsers/spelling.parser.mjs +17 -0
- package/dist/parsers/spelling.parser.mjs.map +1 -0
- package/dist/parsers/string.parser.d.mts +8 -0
- package/dist/parsers/{string.parser.js → string.parser.mjs} +5 -4
- package/dist/parsers/string.parser.mjs.map +1 -0
- package/dist/parsers/{text.parser.d.ts → text.parser.d.mts} +3 -3
- package/dist/parsers/text.parser.mjs +30 -0
- package/dist/parsers/text.parser.mjs.map +1 -0
- package/dist/parsers/{words.parser.d.ts → words.parser.d.mts} +3 -3
- package/dist/parsers/words.parser.mjs +29 -0
- package/dist/parsers/words.parser.mjs.map +1 -0
- package/dist/queries/complex-english.query.d.mts +2 -0
- package/dist/queries/{complex-english.query.js → complex-english.query.mjs} +18 -27
- package/dist/queries/complex-english.query.mjs.map +1 -0
- package/dist/queries/flash-content.query.d.mts +2 -0
- package/dist/queries/flash-content.query.mjs +80 -0
- package/dist/queries/flash-content.query.mjs.map +1 -0
- package/dist/queries/linking-to-domains.query.d.mts +2 -0
- package/dist/queries/linking-to-domains.query.mjs +128 -0
- package/dist/queries/linking-to-domains.query.mjs.map +1 -0
- package/dist/queries/llamaguard-unsafe-content.query.d.mts +2 -0
- package/dist/queries/llamaguard-unsafe-content.query.mjs +90 -0
- package/dist/queries/llamaguard-unsafe-content.query.mjs.map +1 -0
- package/dist/queries/readability-score.query.d.mts +2 -0
- package/dist/queries/{readability-score.query.js → readability-score.query.mjs} +11 -20
- package/dist/queries/readability-score.query.mjs.map +1 -0
- package/dist/servers/{crawl.server.d.ts → crawl.server.d.mts} +10 -10
- package/dist/servers/crawl.server.mjs +192 -0
- package/dist/servers/crawl.server.mjs.map +1 -0
- package/dist/servers/{express.server.d.ts → express.server.d.mts} +2 -3
- package/dist/servers/express.server.mjs +13 -0
- package/dist/servers/express.server.mjs.map +1 -0
- package/dist/servers/{maintenance.server.d.ts → maintenance.server.d.mts} +5 -5
- package/dist/servers/maintenance.server.mjs +97 -0
- package/dist/servers/maintenance.server.mjs.map +1 -0
- package/dist/servers/{rest.server.d.ts → rest.server.d.mts} +4 -4
- package/dist/servers/rest.server.mjs +62 -0
- package/dist/servers/rest.server.mjs.map +1 -0
- package/dist/servers/{socket-io.server.d.ts → socket-io.server.d.mts} +4 -4
- package/dist/servers/socket-io.server.mjs +22 -0
- package/dist/servers/socket-io.server.mjs.map +1 -0
- package/dist/services/{database.service.d.ts → database.service.d.mts} +12 -12
- package/dist/services/database.service.mjs +645 -0
- package/dist/services/database.service.mjs.map +1 -0
- package/dist/services/{ollama-rest.service.d.ts → ollama-rest.service.d.mts} +2 -2
- package/dist/services/ollama-rest.service.mjs +27 -0
- package/dist/services/ollama-rest.service.mjs.map +1 -0
- package/dist/services/{rig-llama-guard.service.d.ts → rig-llama-guard.service.d.mts} +2 -2
- package/dist/services/{rig-llama-guard.service.js → rig-llama-guard.service.mjs} +33 -43
- package/dist/services/rig-llama-guard.service.mjs.map +1 -0
- package/dist/services/{rig-queue-length.service.d.ts → rig-queue-length.service.d.mts} +1 -1
- package/dist/services/rig-queue-length.service.mjs +12 -0
- package/dist/services/rig-queue-length.service.mjs.map +1 -0
- package/dist/services/{rig-subjectivity-scale.service.d.ts → rig-subjectivity-scale.service.d.mts} +2 -2
- package/dist/services/rig-subjectivity-scale.service.mjs +82 -0
- package/dist/services/rig-subjectivity-scale.service.mjs.map +1 -0
- package/dist/services/{rig-ticket-rest.service.d.ts → rig-ticket-rest.service.d.mts} +3 -3
- package/dist/services/rig-ticket-rest.service.mjs +15 -0
- package/dist/services/rig-ticket-rest.service.mjs.map +1 -0
- package/dist/services/{rig-ticketed-promise.service.d.ts → rig-ticketed-promise.service.d.mts} +4 -6
- package/dist/services/{rig-ticketed-promise.service.js → rig-ticketed-promise.service.mjs} +7 -6
- package/dist/services/rig-ticketed-promise.service.mjs.map +1 -0
- package/dist/types/{tcrawl-config.js → tcrawl-config.mjs} +2 -2
- package/dist/types/tcrawl-config.mjs.map +1 -0
- package/dist/types/{thydra-config.js → thydra-config.mjs} +2 -2
- package/dist/types/thydra-config.mjs.map +1 -0
- package/dist/types/tparser-ctor.d.mts +7 -0
- package/dist/types/tparser-ctor.mjs +2 -0
- package/dist/types/tparser-ctor.mjs.map +1 -0
- package/dist/types/tquery.d.mts +7 -0
- package/dist/types/tquery.mjs +2 -0
- package/dist/types/tquery.mjs.map +1 -0
- package/dist/types/tqueue-length.mjs +2 -0
- package/dist/types/tqueue-length.mjs.map +1 -0
- package/dist/types/{trobots-config.js → trobots-config.mjs} +2 -2
- package/dist/types/trobots-config.mjs.map +1 -0
- package/package.json +37 -43
- package/dist/apis/autocomplete.api.js +0 -22
- package/dist/apis/autocomplete.api.js.map +0 -1
- package/dist/apis/bugs.api.js +0 -26
- package/dist/apis/bugs.api.js.map +0 -1
- package/dist/apis/crawl.api.js +0 -25
- package/dist/apis/crawl.api.js.map +0 -1
- package/dist/apis/domains.api.js.map +0 -1
- package/dist/apis/images.api.js +0 -27
- package/dist/apis/images.api.js.map +0 -1
- package/dist/apis/statistics.api.js +0 -41
- package/dist/apis/statistics.api.js.map +0 -1
- package/dist/apis/test.api.js +0 -19
- package/dist/apis/test.api.js.map +0 -1
- package/dist/apis/urls.api.js +0 -29
- package/dist/apis/urls.api.js.map +0 -1
- package/dist/apps/cleanup.app.js +0 -141
- package/dist/apps/cleanup.app.js.map +0 -1
- package/dist/apps/cross-populate-export.app.js +0 -75
- package/dist/apps/cross-populate-export.app.js.map +0 -1
- package/dist/apps/cross-populate-import.app.js +0 -100
- package/dist/apps/cross-populate-import.app.js.map +0 -1
- package/dist/apps/denylist.app.js +0 -132
- package/dist/apps/denylist.app.js.map +0 -1
- package/dist/apps/expire.app.js +0 -63
- package/dist/apps/expire.app.js.map +0 -1
- package/dist/apps/export-domain-urls.js +0 -99
- package/dist/apps/export-domain-urls.js.map +0 -1
- package/dist/apps/extract-text.app.js +0 -55
- package/dist/apps/extract-text.app.js.map +0 -1
- package/dist/apps/hydra.app.js +0 -218
- package/dist/apps/hydra.app.js.map +0 -1
- package/dist/apps/import.app.js +0 -57
- package/dist/apps/import.app.js.map +0 -1
- package/dist/apps/internal-hydra-common.app.js.map +0 -1
- package/dist/apps/move-to-archive.app.js +0 -46
- package/dist/apps/move-to-archive.app.js.map +0 -1
- package/dist/apps/prune-archive.app.js +0 -55
- package/dist/apps/prune-archive.app.js.map +0 -1
- package/dist/apps/query.app.js +0 -80
- package/dist/apps/query.app.js.map +0 -1
- package/dist/apps/reattempt.app.js +0 -83
- package/dist/apps/reattempt.app.js.map +0 -1
- package/dist/apps/requeue-domain.app.js +0 -64
- package/dist/apps/requeue-domain.app.js.map +0 -1
- package/dist/apps/seed.app.js +0 -69
- package/dist/apps/seed.app.js.map +0 -1
- package/dist/apps/startup.app.js +0 -64
- package/dist/apps/startup.app.js.map +0 -1
- package/dist/apps/unarchive-urls.app.js +0 -83
- package/dist/apps/unarchive-urls.app.js.map +0 -1
- package/dist/classes/cleaner.js +0 -266
- package/dist/classes/cleaner.js.map +0 -1
- package/dist/classes/crawler.js +0 -422
- package/dist/classes/crawler.js.map +0 -1
- package/dist/classes/dns.js.map +0 -1
- package/dist/classes/expirer.js +0 -121
- package/dist/classes/expirer.js.map +0 -1
- package/dist/classes/expiry.js.map +0 -1
- package/dist/classes/lists.js.map +0 -1
- package/dist/classes/robot.js +0 -82
- package/dist/classes/robot.js.map +0 -1
- package/dist/classes/tracker.js +0 -120
- package/dist/classes/tracker.js.map +0 -1
- package/dist/cli.js.map +0 -1
- package/dist/enums/eavailable-strategy.js.map +0 -1
- package/dist/enums/elist.js.map +0 -1
- package/dist/enums/eserver.js.map +0 -1
- package/dist/enums/ex-powered-by.js.map +0 -1
- package/dist/helpers/matcher.js.map +0 -1
- package/dist/helpers/random.js.map +0 -1
- package/dist/helpers/utf-decoder.js.map +0 -1
- package/dist/interfaces/iexpiry.js.map +0 -1
- package/dist/interfaces/imatch.js.map +0 -1
- package/dist/interfaces/iparser-config.js.map +0 -1
- package/dist/interfaces/iparser.js +0 -2
- package/dist/interfaces/iparser.js.map +0 -1
- package/dist/interfaces/irequest-outcome.js +0 -2
- package/dist/interfaces/irequest-outcome.js.map +0 -1
- package/dist/interfaces/iserver.js +0 -2
- package/dist/interfaces/iserver.js.map +0 -1
- package/dist/parsers/accessibility-metrics.parser.d.ts +0 -11
- package/dist/parsers/accessibility-metrics.parser.js +0 -40
- package/dist/parsers/accessibility-metrics.parser.js.map +0 -1
- package/dist/parsers/asp-error.parser.d.ts +0 -12
- package/dist/parsers/asp-error.parser.js +0 -48
- package/dist/parsers/asp-error.parser.js.map +0 -1
- package/dist/parsers/bad-words.parser.js +0 -27
- package/dist/parsers/bad-words.parser.js.map +0 -1
- package/dist/parsers/complex-english.parser.d.ts +0 -15
- package/dist/parsers/complex-english.parser.js +0 -61
- package/dist/parsers/complex-english.parser.js.map +0 -1
- package/dist/parsers/data.parser.d.ts +0 -14
- package/dist/parsers/data.parser.js.map +0 -1
- package/dist/parsers/dictionary.parser.js +0 -73
- package/dist/parsers/dictionary.parser.js.map +0 -1
- package/dist/parsers/html.parser.d.ts +0 -13
- package/dist/parsers/html.parser.js.map +0 -1
- package/dist/parsers/hyperlinks.parser.d.ts +0 -20
- package/dist/parsers/hyperlinks.parser.js +0 -115
- package/dist/parsers/hyperlinks.parser.js.map +0 -1
- package/dist/parsers/image-tags.parser.d.ts +0 -19
- package/dist/parsers/image-tags.parser.js +0 -52
- package/dist/parsers/image-tags.parser.js.map +0 -1
- package/dist/parsers/interest.parser.js +0 -69
- package/dist/parsers/interest.parser.js.map +0 -1
- package/dist/parsers/jpeg.parser.d.ts +0 -11
- package/dist/parsers/jpeg.parser.js +0 -39
- package/dist/parsers/jpeg.parser.js.map +0 -1
- package/dist/parsers/llama-guard.parser.js +0 -65
- package/dist/parsers/llama-guard.parser.js.map +0 -1
- package/dist/parsers/offence.parser.js +0 -69
- package/dist/parsers/offence.parser.js.map +0 -1
- package/dist/parsers/ollama.parser.js +0 -51
- package/dist/parsers/ollama.parser.js.map +0 -1
- package/dist/parsers/paragraphs.parser.js +0 -49
- package/dist/parsers/paragraphs.parser.js.map +0 -1
- package/dist/parsers/parser.js +0 -57
- package/dist/parsers/parser.js.map +0 -1
- package/dist/parsers/php-error.parser.d.ts +0 -12
- package/dist/parsers/php-error.parser.js +0 -52
- package/dist/parsers/php-error.parser.js.map +0 -1
- package/dist/parsers/phrase.parser.js +0 -26
- package/dist/parsers/phrase.parser.js.map +0 -1
- package/dist/parsers/regex.parser.js +0 -43
- package/dist/parsers/regex.parser.js.map +0 -1
- package/dist/parsers/server.parser.d.ts +0 -11
- package/dist/parsers/server.parser.js +0 -67
- package/dist/parsers/server.parser.js.map +0 -1
- package/dist/parsers/spelling.parser.js +0 -27
- package/dist/parsers/spelling.parser.js.map +0 -1
- package/dist/parsers/string.parser.d.ts +0 -8
- package/dist/parsers/string.parser.js.map +0 -1
- package/dist/parsers/text.parser.js +0 -41
- package/dist/parsers/text.parser.js.map +0 -1
- package/dist/parsers/words.parser.js +0 -40
- package/dist/parsers/words.parser.js.map +0 -1
- package/dist/queries/complex-english.query.d.ts +0 -2
- package/dist/queries/complex-english.query.js.map +0 -1
- package/dist/queries/flash-content.query.d.ts +0 -2
- package/dist/queries/flash-content.query.js +0 -82
- package/dist/queries/flash-content.query.js.map +0 -1
- package/dist/queries/linking-to-domains.query.d.ts +0 -2
- package/dist/queries/linking-to-domains.query.js +0 -130
- package/dist/queries/linking-to-domains.query.js.map +0 -1
- package/dist/queries/llamaguard-unsafe-content.query.d.ts +0 -2
- package/dist/queries/llamaguard-unsafe-content.query.js +0 -92
- package/dist/queries/llamaguard-unsafe-content.query.js.map +0 -1
- package/dist/queries/readability-score.query.d.ts +0 -2
- package/dist/queries/readability-score.query.js.map +0 -1
- package/dist/servers/crawl.server.js +0 -198
- package/dist/servers/crawl.server.js.map +0 -1
- package/dist/servers/express.server.js +0 -13
- package/dist/servers/express.server.js.map +0 -1
- package/dist/servers/maintenance.server.js +0 -103
- package/dist/servers/maintenance.server.js.map +0 -1
- package/dist/servers/rest.server.js +0 -46
- package/dist/servers/rest.server.js.map +0 -1
- package/dist/servers/socket-io.server.js +0 -61
- package/dist/servers/socket-io.server.js.map +0 -1
- package/dist/services/database.service.js +0 -795
- package/dist/services/database.service.js.map +0 -1
- package/dist/services/ollama-rest.service.js +0 -46
- package/dist/services/ollama-rest.service.js.map +0 -1
- package/dist/services/rig-llama-guard.service.js.map +0 -1
- package/dist/services/rig-queue-length.service.js +0 -22
- package/dist/services/rig-queue-length.service.js.map +0 -1
- package/dist/services/rig-subjectivity-scale.service.js +0 -96
- package/dist/services/rig-subjectivity-scale.service.js.map +0 -1
- package/dist/services/rig-ticket-rest.service.js +0 -29
- package/dist/services/rig-ticket-rest.service.js.map +0 -1
- package/dist/services/rig-ticketed-promise.service.js.map +0 -1
- package/dist/types/tcrawl-config.js.map +0 -1
- package/dist/types/thydra-config.js.map +0 -1
- package/dist/types/tparser-ctor.d.ts +0 -7
- package/dist/types/tparser-ctor.js +0 -2
- package/dist/types/tparser-ctor.js.map +0 -1
- package/dist/types/tquery.d.ts +0 -7
- package/dist/types/tquery.js +0 -2
- package/dist/types/tquery.js.map +0 -1
- package/dist/types/tqueue-length.js +0 -2
- package/dist/types/tqueue-length.js.map +0 -1
- package/dist/types/trobots-config.js.map +0 -1
- /package/dist/classes/{dns.d.ts → dns.d.mts} +0 -0
- /package/dist/{cli.d.ts → cli.d.mts} +0 -0
- /package/dist/enums/{eavailable-strategy.d.ts → eavailable-strategy.d.mts} +0 -0
- /package/dist/enums/{elist.d.ts → elist.d.mts} +0 -0
- /package/dist/enums/{eserver.d.ts → eserver.d.mts} +0 -0
- /package/dist/enums/{ex-powered-by.d.ts → ex-powered-by.d.mts} +0 -0
- /package/dist/interfaces/{imatch.d.ts → imatch.d.mts} +0 -0
- /package/dist/interfaces/{iparser-config.d.ts → iparser-config.d.mts} +0 -0
- /package/dist/interfaces/{iserver.d.ts → iserver.d.mts} +0 -0
- /package/dist/types/{tcrawl-config.d.ts → tcrawl-config.d.mts} +0 -0
- /package/dist/types/{thydra-config.d.ts → thydra-config.d.mts} +0 -0
- /package/dist/types/{tqueue-length.d.ts → tqueue-length.d.mts} +0 -0
- /package/dist/types/{trobots-config.d.ts → trobots-config.d.mts} +0 -0
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"php-error.parser.mjs","sourceRoot":"","sources":["../../src/parsers/php-error.parser.mts"],"names":[],"mappings":"AAGA,OAAO,EAAiB,mBAAmB,EAAE,MAAM,0BAA0B,CAAC;AAO9E,OAAO,EAAE,YAAY,EAAE,MAAM,qBAAqB,CAAC;AAGnD,MAAM,SAAS,GAAW,iGAAiG,CAAC;AAE5H,MAAM,OAAO,cAAe,SAAQ,YAAyB;IAElD;IADV,YACU,GAAY,EACpB,OAAyB,EACzB,MAAuC;QAExC,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,UAAU,CAAC,CAAC;QAJ1B,QAAG,GAAH,GAAG,CAAS;IAKtB,CAAC;IAEe,KAAK,CAAC,KAAK,CAAC,QAAyB;QACpD,IAAI,CAAC,IAAI,CAAC,UAAU,IAAI,CAAC,IAAI,CAAC,GAAG;YAAE,OAAO;QAE1C,MAAM,OAAO,GAAW,IAAI,MAAM,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC;QAEnD,MAAM,MAAM,GAAgB,EAAE,CAAC;QAC/B,OAAO,IAAI,EAAE,CAAC;YACb,MAAM,MAAM,GAAyB,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;YACnE,IAAI,MAAM,KAAK,IAAI;gBAAE,MAAM;YAE3B,MAAM,IAAI,GAA4B,mBAAmB,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;YACrE,IAAI,IAAI,KAAK,SAAS;gBAAE,SAAS;YAEjC,MAAM,OAAO,GAAW,MAAM,CAAC,CAAC,CAAC,CAAC;YAClC,MAAM,IAAI,GAAW,MAAM,CAAC,CAAC,CAAC,CAAC;YAC/B,MAAM,IAAI,GAAW,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;YAE7C,MAAM,KAAK,GAAc;gBACvB,IAAI,EAAE,IAAI;gBACV,OAAO,EAAE,OAAO;gBAChB,IAAI,EAAE,IAAI;gBACV,IAAI,EAAE,IAAI;aACX,CAAC;YACF,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACpB,CAAC;QAED,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC;YAAE,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,WAAW,EAAE,MAAM,CAAC,CAAC;;YACxE,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,WAAW,CAAC,CAAC;IACtD,CAAC;IAEM,QAAQ,CAAC,YAAoB,EAAE,OAAgB;QACrD,OAAO,OAAO,CAAC,CAAC,yBAAyB;IAC1C,CAAC;CACD"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
-
import { RegexParser } from './regex.parser';
|
|
3
|
-
import { IDictionaryConfig } from './dictionary.parser';
|
|
1
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
2
|
+
import { RegexParser } from './regex.parser.mjs';
|
|
3
|
+
import { IDictionaryConfig } from './dictionary.parser.mjs';
|
|
4
4
|
export declare abstract class PhraseParser<T extends IDictionaryConfig> extends RegexParser<T> {
|
|
5
5
|
setPhrases(phrases: string[], caseInsensitive: boolean): void;
|
|
6
6
|
protected abstract parsePhrases(database: DatabaseService, phrases: string[]): Promise<void>;
|
|
@@ -0,0 +1,15 @@
|
|
|
1
|
+
import { RegexParser } from './regex.parser.mjs';
|
|
2
|
+
export class PhraseParser extends RegexParser {
|
|
3
|
+
setPhrases(phrases, caseInsensitive) {
|
|
4
|
+
const map = new Map();
|
|
5
|
+
for (const phrase of phrases) {
|
|
6
|
+
const regex = new RegExp(`(^|[^0-9A-Za-z])${phrase}($|[^0-9A-Za-z])`, `g${caseInsensitive ? 'i' : ''}`);
|
|
7
|
+
map.set(phrase, regex);
|
|
8
|
+
}
|
|
9
|
+
this.setRegExs(map);
|
|
10
|
+
}
|
|
11
|
+
async parseRegExs(database, matches) {
|
|
12
|
+
await this.parsePhrases(database, Array.from(matches.keys()));
|
|
13
|
+
}
|
|
14
|
+
}
|
|
15
|
+
//# sourceMappingURL=phrase.parser.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"phrase.parser.mjs","sourceRoot":"","sources":["../../src/parsers/phrase.parser.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AAGjD,MAAM,OAAgB,YAA0C,SAAQ,WAAc;IAC9E,UAAU,CACf,OAAiB,EACjB,eAAwB;QAEzB,MAAM,GAAG,GAAwB,IAAI,GAAG,EAAkB,CAAC;QAC3D,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;YAC9B,MAAM,KAAK,GAAW,IAAI,MAAM,CAAC,mBAAmB,MAAM,kBAAkB,EAAE,IAAI,eAAe,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;YAChH,GAAG,CAAC,GAAG,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC;QACxB,CAAC;QAED,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;IACrB,CAAC;IAIS,KAAK,CAAC,WAAW,CAAC,QAAyB,EAAE,OAAuC;QAC7F,MAAM,IAAI,CAAC,YAAY,CAAC,QAAQ,EAAE,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;IAC/D,CAAC;CACD"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
-
import { TextParser } from './text.parser';
|
|
3
|
-
import { IDataConfig } from './data.parser';
|
|
1
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
2
|
+
import { TextParser } from './text.parser.mjs';
|
|
3
|
+
import { IDataConfig } from './data.parser.mjs';
|
|
4
4
|
export declare abstract class RegexParser<T extends IDataConfig> extends TextParser<T> {
|
|
5
5
|
private regexs;
|
|
6
6
|
protected setRegExs(regexs: Map<string, RegExp>): void;
|
|
@@ -0,0 +1,29 @@
|
|
|
1
|
+
import { TextParser } from './text.parser.mjs';
|
|
2
|
+
export class RegexParser extends TextParser {
|
|
3
|
+
regexs = new Map();
|
|
4
|
+
setRegExs(regexs) {
|
|
5
|
+
for (const key of regexs.keys()) {
|
|
6
|
+
this.regexs.set(key, regexs.get(key));
|
|
7
|
+
}
|
|
8
|
+
}
|
|
9
|
+
getRegExs() {
|
|
10
|
+
return this.regexs;
|
|
11
|
+
}
|
|
12
|
+
async parseText(database, text) {
|
|
13
|
+
const matches = new Map();
|
|
14
|
+
for (const key of this.regexs.keys()) {
|
|
15
|
+
const regex = this.regexs.get(key);
|
|
16
|
+
const results = [];
|
|
17
|
+
while (true) {
|
|
18
|
+
const result = regex.exec(text);
|
|
19
|
+
if (!result || result === null)
|
|
20
|
+
break;
|
|
21
|
+
results.push(result);
|
|
22
|
+
}
|
|
23
|
+
if (results.length > 0)
|
|
24
|
+
matches.set(key, results);
|
|
25
|
+
}
|
|
26
|
+
await this.parseRegExs(database, matches);
|
|
27
|
+
}
|
|
28
|
+
}
|
|
29
|
+
//# sourceMappingURL=regex.parser.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"regex.parser.mjs","sourceRoot":"","sources":["../../src/parsers/regex.parser.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAG/C,MAAM,OAAgB,WAAmC,SAAQ,UAAa;IACrE,MAAM,GAAwB,IAAI,GAAG,EAAkB,CAAC;IAEtD,SAAS,CAAC,MAA2B;QAC9C,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC;YACjC,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,GAAG,CAAC,GAAG,CAAE,CAAC,CAAC;QACxC,CAAC;IACF,CAAC;IAEM,SAAS;QACf,OAAO,IAAI,CAAC,MAAM,CAAC;IACpB,CAAC;IAIS,KAAK,CAAC,SAAS,CAAC,QAAyB,EAAE,IAAY;QAChE,MAAM,OAAO,GAAmC,IAAI,GAAG,EAA6B,CAAC;QACrF,KAAK,MAAM,GAAG,IAAI,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC;YACtC,MAAM,KAAK,GAAW,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,GAAG,CAAE,CAAC;YAE5C,MAAM,OAAO,GAAsB,EAAE,CAAC;YACtC,OAAO,IAAI,EAAE,CAAC;gBACb,MAAM,MAAM,GAAyB,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBACtD,IAAI,CAAC,MAAM,IAAI,MAAM,KAAK,IAAI;oBAAE,MAAM;gBAEtC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YACtB,CAAC;YAED,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC;gBAAE,OAAO,CAAC,GAAG,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;QACnD,CAAC;QAED,MAAM,IAAI,CAAC,WAAW,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;IAC3C,CAAC;CACD"}
|
|
@@ -0,0 +1,11 @@
|
|
|
1
|
+
import { TPropertyObject } from 'tscommons-esm-core';
|
|
2
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
3
|
+
import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
|
|
4
|
+
import { IParserConfig } from '../interfaces/iparser-config.mjs';
|
|
5
|
+
import { Parser } from './parser.mjs';
|
|
6
|
+
export declare class ServerParser extends Parser<IParserConfig> {
|
|
7
|
+
private url?;
|
|
8
|
+
constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>);
|
|
9
|
+
supports(_contentType: string): boolean;
|
|
10
|
+
parse(database: DatabaseService): Promise<void>;
|
|
11
|
+
}
|
|
@@ -0,0 +1,57 @@
|
|
|
1
|
+
import { EServer } from '../enums/eserver.mjs';
|
|
2
|
+
import { EXPoweredBy } from '../enums/ex-powered-by.mjs';
|
|
3
|
+
import { Parser } from './parser.mjs';
|
|
4
|
+
export class ServerParser extends Parser {
|
|
5
|
+
url;
|
|
6
|
+
constructor(url, outcome, config) {
|
|
7
|
+
super(outcome, config, 'server');
|
|
8
|
+
this.url = url;
|
|
9
|
+
}
|
|
10
|
+
supports(_contentType) {
|
|
11
|
+
return true; // everything
|
|
12
|
+
}
|
|
13
|
+
async parse(database) {
|
|
14
|
+
if (!this.outcome || !this.url)
|
|
15
|
+
return;
|
|
16
|
+
const server = this.outcome.headers['server'] || undefined;
|
|
17
|
+
const xPoweredBy = this.outcome.headers['x-powered-by'] || undefined;
|
|
18
|
+
if (server === undefined && xPoweredBy === undefined) {
|
|
19
|
+
await database.unsetData(this.url, 'server');
|
|
20
|
+
return;
|
|
21
|
+
}
|
|
22
|
+
const data = {};
|
|
23
|
+
if (server) {
|
|
24
|
+
if (/^Apache($|\/)/.test(server))
|
|
25
|
+
data.server = EServer.APACHE;
|
|
26
|
+
if (/^Microsoft-IIS($|\/)/.test(server))
|
|
27
|
+
data.server = EServer.IIS;
|
|
28
|
+
if (/^nginx($|\/)/.test(server))
|
|
29
|
+
data.server = EServer.NGINX;
|
|
30
|
+
if (/^cloudflare($|\/)/.test(server))
|
|
31
|
+
data.server = EServer.CLOUDFLARE;
|
|
32
|
+
if (/^AmazonS3($|\/)/.test(server))
|
|
33
|
+
data.server = EServer.AMAZONS3;
|
|
34
|
+
if (/^Jetty($|\/)/.test(server))
|
|
35
|
+
data.server = EServer.JETTY;
|
|
36
|
+
}
|
|
37
|
+
if (xPoweredBy) {
|
|
38
|
+
if (/^PHP($|\/)/.test(xPoweredBy))
|
|
39
|
+
data.xPoweredBy = EXPoweredBy.PHP;
|
|
40
|
+
if (/^ASP.NET($|\/)/.test(xPoweredBy))
|
|
41
|
+
data.xPoweredBy = EXPoweredBy.ASPNET;
|
|
42
|
+
if (/^Express($|\/)/.test(xPoweredBy))
|
|
43
|
+
data.xPoweredBy = EXPoweredBy.EXPRESS;
|
|
44
|
+
if (/^Zope($|\/)/.test(xPoweredBy))
|
|
45
|
+
data.xPoweredBy = EXPoweredBy.ZOPE;
|
|
46
|
+
}
|
|
47
|
+
if (data.server === undefined && data.xPoweredBy === undefined) {
|
|
48
|
+
await database.unsetData(this.url, 'server');
|
|
49
|
+
return;
|
|
50
|
+
}
|
|
51
|
+
try {
|
|
52
|
+
await database.setData(this.url, 'server', data);
|
|
53
|
+
}
|
|
54
|
+
catch (_ex) { /* do nothing */ }
|
|
55
|
+
}
|
|
56
|
+
}
|
|
57
|
+
//# sourceMappingURL=server.parser.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"server.parser.mjs","sourceRoot":"","sources":["../../src/parsers/server.parser.mts"],"names":[],"mappings":"AAQA,OAAO,EAAE,OAAO,EAAE,MAAM,sBAAsB,CAAC;AAC/C,OAAO,EAAE,WAAW,EAAE,MAAM,4BAA4B,CAAC;AAEzD,OAAO,EAAE,MAAM,EAAE,MAAM,cAAc,CAAC;AAEtC,MAAM,OAAO,YAAa,SAAQ,MAAqB;IAE5C;IADV,YACU,GAAY,EACpB,OAAyB,EACzB,MAAuC;QAExC,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,QAAQ,CAAC,CAAC;QAJxB,QAAG,GAAH,GAAG,CAAS;IAKtB,CAAC;IAEM,QAAQ,CAAC,YAAoB;QACnC,OAAO,IAAI,CAAC,CAAC,aAAa;IAC3B,CAAC;IAEe,KAAK,CAAC,KAAK,CAAC,QAAyB;QACpD,IAAI,CAAC,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,CAAC,GAAG;YAAE,OAAO;QAEvC,MAAM,MAAM,GAAsB,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,QAAQ,CAAY,IAAI,SAAS,CAAC;QACzF,MAAM,UAAU,GAAsB,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,cAAc,CAAY,IAAI,SAAS,CAAC;QAEnG,IAAI,MAAM,KAAK,SAAS,IAAI,UAAU,KAAK,SAAS,EAAE,CAAC;YACtD,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC;YAC7C,OAAO;QACR,CAAC;QAED,MAAM,IAAI,GAAY,EAAE,CAAC;QAEzB,IAAI,MAAM,EAAE,CAAC;YACZ,IAAI,eAAe,CAAC,IAAI,CAAC,MAAM,CAAC;gBAAE,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;YAC/D,IAAI,sBAAsB,CAAC,IAAI,CAAC,MAAM,CAAC;gBAAE,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC;YACnE,IAAI,cAAc,CAAC,IAAI,CAAC,MAAM,CAAC;gBAAE,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,KAAK,CAAC;YAC7D,IAAI,mBAAmB,CAAC,IAAI,CAAC,MAAM,CAAC;gBAAE,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,UAAU,CAAC;YACvE,IAAI,iBAAiB,CAAC,IAAI,CAAC,MAAM,CAAC;gBAAE,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,QAAQ,CAAC;YACnE,IAAI,cAAc,CAAC,IAAI,CAAC,MAAM,CAAC;gBAAE,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,KAAK,CAAC;QAC9D,CAAC;QAED,IAAI,UAAU,EAAE,CAAC;YAChB,IAAI,YAAY,CAAC,IAAI,CAAC,UAAU,CAAC;gBAAE,IAAI,CAAC,UAAU,GAAG,WAAW,CAAC,GAAG,CAAC;YACrE,IAAI,gBAAgB,CAAC,IAAI,CAAC,UAAU,CAAC;gBAAE,IAAI,CAAC,UAAU,GAAG,WAAW,CAAC,MAAM,CAAC;YAC5E,IAAI,gBAAgB,CAAC,IAAI,CAAC,UAAU,CAAC;gBAAE,IAAI,CAAC,UAAU,GAAG,WAAW,CAAC,OAAO,CAAC;YAC7E,IAAI,aAAa,CAAC,IAAI,CAAC,UAAU,CAAC;gBAAE,IAAI,CAAC,UAAU,GAAG,WAAW,CAAC,IAAI,CAAC;QACxE,CAAC;QAED,IAAI,IAAI,CAAC,MAAM,KAAK,SAAS,IAAI,IAAI,CAAC,UAAU,KAAK,SAAS,EAAE,CAAC;YAChE,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC;YAC7C,OAAO;QACR,CAAC;QAED,IAAI,CAAC;YACJ,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,EAAE,IAAI,CAAC,CAAC;QAClD,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC,CAAC,gBAAgB,CAAC,CAAC;IACnC,CAAC;CACD"}
|
|
@@ -1,10 +1,10 @@
|
|
|
1
|
-
import {
|
|
2
|
-
import { DatabaseService } from '../services/database.service';
|
|
3
|
-
import { IParserConfig } from '../interfaces/iparser-config';
|
|
4
|
-
import { IRequestOutcome } from '../interfaces/irequest-outcome';
|
|
5
|
-
import { DictionaryParser, IDictionaryConfig } from './dictionary.parser';
|
|
1
|
+
import { TPropertyObject } from 'tscommons-esm-core';
|
|
2
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
3
|
+
import { IParserConfig } from '../interfaces/iparser-config.mjs';
|
|
4
|
+
import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
|
|
5
|
+
import { DictionaryParser, IDictionaryConfig } from './dictionary.parser.mjs';
|
|
6
6
|
export declare class SpellingParser extends DictionaryParser<IDictionaryConfig> {
|
|
7
7
|
private url?;
|
|
8
|
-
constructor(url?: string | undefined, outcome?: IRequestOutcome, config?:
|
|
8
|
+
constructor(url?: string | undefined, outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>);
|
|
9
9
|
protected parseMatches(database: DatabaseService, matches: string[], _nonMatches: string[]): Promise<void>;
|
|
10
10
|
}
|
|
@@ -0,0 +1,17 @@
|
|
|
1
|
+
import { DictionaryParser } from './dictionary.parser.mjs';
|
|
2
|
+
export class SpellingParser extends DictionaryParser {
|
|
3
|
+
url;
|
|
4
|
+
constructor(url, outcome, config) {
|
|
5
|
+
super(outcome, config, 'spelling');
|
|
6
|
+
this.url = url;
|
|
7
|
+
}
|
|
8
|
+
async parseMatches(database, matches, _nonMatches) {
|
|
9
|
+
if (!this.url)
|
|
10
|
+
return;
|
|
11
|
+
if (matches.length > 0)
|
|
12
|
+
await database.setData(this.url, 'spelling', matches);
|
|
13
|
+
else
|
|
14
|
+
await database.unsetData(this.url, 'spelling');
|
|
15
|
+
}
|
|
16
|
+
}
|
|
17
|
+
//# sourceMappingURL=spelling.parser.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"spelling.parser.mjs","sourceRoot":"","sources":["../../src/parsers/spelling.parser.mts"],"names":[],"mappings":"AAOA,OAAO,EAAE,gBAAgB,EAAqB,MAAM,yBAAyB,CAAC;AAE9E,MAAM,OAAO,cAAe,SAAQ,gBAAmC;IAE5D;IADV,YACU,GAAY,EACpB,OAAyB,EACzB,MAAuC;QAExC,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,UAAU,CAAC,CAAC;QAJ1B,QAAG,GAAH,GAAG,CAAS;IAKtB,CAAC;IAES,KAAK,CAAC,YAAY,CAAC,QAAyB,EAAE,OAAiB,EAAE,WAAqB;QAC/F,IAAI,CAAC,IAAI,CAAC,GAAG;YAAE,OAAO;QAEtB,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC;YAAE,MAAM,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,UAAU,EAAE,OAAO,CAAC,CAAC;;YACzE,MAAM,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,EAAE,UAAU,CAAC,CAAC;IACrD,CAAC;CACD"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
import { TPropertyObject } from 'tscommons-esm-core';
|
|
2
|
+
import { IRequestOutcome } from '../interfaces/irequest-outcome.mjs';
|
|
3
|
+
import { IParserConfig } from '../interfaces/iparser-config.mjs';
|
|
4
|
+
import { DataParser, IDataConfig } from './data.parser.mjs';
|
|
5
|
+
export declare abstract class StringParser<T extends IDataConfig> extends DataParser<T> {
|
|
6
|
+
protected stringData: string | undefined;
|
|
7
|
+
constructor(outcome?: IRequestOutcome, config?: TPropertyObject<IParserConfig>, configKey?: string);
|
|
8
|
+
}
|
|
@@ -1,6 +1,7 @@
|
|
|
1
|
-
import { UtfDecoder } from '../helpers/utf-decoder';
|
|
2
|
-
import { DataParser } from './data.parser';
|
|
1
|
+
import { UtfDecoder } from '../helpers/utf-decoder.mjs';
|
|
2
|
+
import { DataParser } from './data.parser.mjs';
|
|
3
3
|
export class StringParser extends DataParser {
|
|
4
|
+
stringData;
|
|
4
5
|
constructor(outcome, config, configKey) {
|
|
5
6
|
super(outcome, config, configKey);
|
|
6
7
|
if (!this.data)
|
|
@@ -8,9 +9,9 @@ export class StringParser extends DataParser {
|
|
|
8
9
|
try {
|
|
9
10
|
this.stringData = UtfDecoder.fromBuffer(this.data);
|
|
10
11
|
}
|
|
11
|
-
catch (
|
|
12
|
+
catch (_ex) {
|
|
12
13
|
// ignore
|
|
13
14
|
}
|
|
14
15
|
}
|
|
15
16
|
}
|
|
16
|
-
//# sourceMappingURL=string.parser.
|
|
17
|
+
//# sourceMappingURL=string.parser.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"string.parser.mjs","sourceRoot":"","sources":["../../src/parsers/string.parser.mts"],"names":[],"mappings":"AAEA,OAAO,EAAE,UAAU,EAAE,MAAM,4BAA4B,CAAC;AAKxD,OAAO,EAAE,UAAU,EAAe,MAAM,mBAAmB,CAAC;AAE5D,MAAM,OAAgB,YAAoC,SAAQ,UAAa;IACpE,UAAU,CAAmB;IAEvC,YACE,OAAyB,EACzB,MAAuC,EACvC,SAAkB;QAEnB,KAAK,CAAC,OAAO,EAAE,MAAM,EAAE,SAAS,CAAC,CAAC;QAElC,IAAI,CAAC,IAAI,CAAC,IAAI;YAAE,OAAO;QAEvB,IAAI,CAAC;YACJ,IAAI,CAAC,UAAU,GAAG,UAAU,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACpD,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACd,SAAS;QACV,CAAC;IACF,CAAC;CACD"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
-
import { HtmlParser } from './html.parser';
|
|
3
|
-
import { IDataConfig } from './data.parser';
|
|
1
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
2
|
+
import { HtmlParser } from './html.parser.mjs';
|
|
3
|
+
import { IDataConfig } from './data.parser.mjs';
|
|
4
4
|
export declare abstract class TextParser<T extends IDataConfig> extends HtmlParser<T> {
|
|
5
5
|
static attemptExtract(html: string): string;
|
|
6
6
|
protected abstract parseText(database: DatabaseService, text: string): Promise<void>;
|
|
@@ -0,0 +1,30 @@
|
|
|
1
|
+
import extractor from 'unfluff';
|
|
2
|
+
import { HtmlParser } from './html.parser.mjs';
|
|
3
|
+
export class TextParser extends HtmlParser {
|
|
4
|
+
static attemptExtract(html) {
|
|
5
|
+
try {
|
|
6
|
+
// eslint-disable-next-line @typescript-eslint/no-unsafe-assignment, @typescript-eslint/no-unsafe-call
|
|
7
|
+
const extracted = extractor(html);
|
|
8
|
+
if (!extracted)
|
|
9
|
+
throw new Error('Unable to run unfluff');
|
|
10
|
+
return extracted.text
|
|
11
|
+
.split(/\n+/)
|
|
12
|
+
.map((s) => s.trim())
|
|
13
|
+
.filter((s) => s !== '')
|
|
14
|
+
.join('\n');
|
|
15
|
+
}
|
|
16
|
+
catch (_ex) {
|
|
17
|
+
return '';
|
|
18
|
+
}
|
|
19
|
+
}
|
|
20
|
+
async parse(database) {
|
|
21
|
+
if (!this.dom) {
|
|
22
|
+
await this.parseText(database, '');
|
|
23
|
+
return;
|
|
24
|
+
}
|
|
25
|
+
const html = this.dom.html();
|
|
26
|
+
const extracted = TextParser.attemptExtract(html);
|
|
27
|
+
await this.parseText(database, extracted);
|
|
28
|
+
}
|
|
29
|
+
}
|
|
30
|
+
//# sourceMappingURL=text.parser.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"text.parser.mjs","sourceRoot":"","sources":["../../src/parsers/text.parser.mts"],"names":[],"mappings":"AAAA,OAAO,SAAS,MAAM,SAAS,CAAC;AAIhC,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAG/C,MAAM,OAAgB,UAAkC,SAAQ,UAAa;IACrE,MAAM,CAAC,cAAc,CAAC,IAAY;QACxC,IAAI,CAAC;YACJ,sGAAsG;YACtG,MAAM,SAAS,GAAqB,SAAS,CAAC,IAAI,CAAC,CAAC;YACpD,IAAI,CAAC,SAAS;gBAAE,MAAM,IAAI,KAAK,CAAC,uBAAuB,CAAC,CAAC;YAEzD,OAAO,SAAS,CAAC,IAAI;iBAClB,KAAK,CAAC,KAAK,CAAC;iBACZ,GAAG,CAAC,CAAC,CAAS,EAAU,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;iBACpC,MAAM,CAAC,CAAC,CAAS,EAAW,EAAE,CAAC,CAAC,KAAK,EAAE,CAAC;iBACxC,IAAI,CAAC,IAAI,CAAC,CAAC;QACf,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACd,OAAO,EAAE,CAAC;QACX,CAAC;IACF,CAAC;IAIe,KAAK,CAAC,KAAK,CAAC,QAAyB;QACpD,IAAI,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC;YACf,MAAM,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;YACnC,OAAO;QACR,CAAC;QAED,MAAM,IAAI,GAAW,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC;QACrC,MAAM,SAAS,GAAW,UAAU,CAAC,cAAc,CAAC,IAAI,CAAC,CAAC;QAE1D,MAAM,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC;IAC3C,CAAC;CACD"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { DatabaseService } from '../services/database.service';
|
|
2
|
-
import { TextParser } from './text.parser';
|
|
3
|
-
import { IDataConfig } from './data.parser';
|
|
1
|
+
import { DatabaseService } from '../services/database.service.mjs';
|
|
2
|
+
import { TextParser } from './text.parser.mjs';
|
|
3
|
+
import { IDataConfig } from './data.parser.mjs';
|
|
4
4
|
export interface IWordsConfig extends IDataConfig {
|
|
5
5
|
allowHyphenatedWords?: boolean;
|
|
6
6
|
}
|
|
@@ -0,0 +1,29 @@
|
|
|
1
|
+
import { commonsTypeHasPropertyBooleanOrUndefined } from 'tscommons-esm-core';
|
|
2
|
+
import { TextParser } from './text.parser.mjs';
|
|
3
|
+
import { isIDataConfig } from './data.parser.mjs';
|
|
4
|
+
const WORD_WITH_HYPHEN = '(?<![a-z0-9])[a-z0-9](?:[-\']?[a-z0-9])*(?![a-z0-9])';
|
|
5
|
+
const WORD_WITHOUT_HYPHEN = '(?<![a-z0-9])[a-z0-9](?:[\']?[a-z0-9])*(?![a-z0-9])';
|
|
6
|
+
export function isIWordsConfig(test) {
|
|
7
|
+
if (!isIDataConfig(test))
|
|
8
|
+
return false;
|
|
9
|
+
if (!commonsTypeHasPropertyBooleanOrUndefined(test, 'allowHyphenatedWords'))
|
|
10
|
+
return false;
|
|
11
|
+
return true;
|
|
12
|
+
}
|
|
13
|
+
export class WordsParser extends TextParser {
|
|
14
|
+
async parseText(database, text) {
|
|
15
|
+
const wordsConfig = this.getConfig(isIWordsConfig);
|
|
16
|
+
if (!wordsConfig)
|
|
17
|
+
return;
|
|
18
|
+
const pattern = wordsConfig.allowHyphenatedWords ? new RegExp(WORD_WITH_HYPHEN, 'ig') : new RegExp(WORD_WITHOUT_HYPHEN, 'ig');
|
|
19
|
+
const words = [];
|
|
20
|
+
while (true) {
|
|
21
|
+
const result = pattern.exec(text);
|
|
22
|
+
if (result === null)
|
|
23
|
+
break;
|
|
24
|
+
words.push(result[0]);
|
|
25
|
+
}
|
|
26
|
+
await this.parseWords(database, words);
|
|
27
|
+
}
|
|
28
|
+
}
|
|
29
|
+
//# sourceMappingURL=words.parser.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"words.parser.mjs","sourceRoot":"","sources":["../../src/parsers/words.parser.mts"],"names":[],"mappings":"AAAA,OAAO,EAAE,wCAAwC,EAAE,MAAM,oBAAoB,CAAC;AAI9E,OAAO,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AAC/C,OAAO,EAAe,aAAa,EAAE,MAAM,mBAAmB,CAAC;AAE/D,MAAM,gBAAgB,GAAW,sDAAsD,CAAC;AACxF,MAAM,mBAAmB,GAAW,qDAAqD,CAAC;AAK1F,MAAM,UAAU,cAAc,CAAC,IAAa;IAC3C,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAC;IAEvC,IAAI,CAAC,wCAAwC,CAAC,IAAI,EAAE,sBAAsB,CAAC;QAAE,OAAO,KAAK,CAAC;IAE1F,OAAO,IAAI,CAAC;AACb,CAAC;AAED,MAAM,OAAgB,WAAoC,SAAQ,UAAa;IAGpE,KAAK,CAAC,SAAS,CAAC,QAAyB,EAAE,IAAY;QAChE,MAAM,WAAW,GAA2B,IAAI,CAAC,SAAS,CAAC,cAAc,CAAC,CAAC;QAC3E,IAAI,CAAC,WAAW;YAAE,OAAO;QAEzB,MAAM,OAAO,GAAW,WAAW,CAAC,oBAAoB,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,gBAAgB,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,mBAAmB,EAAE,IAAI,CAAC,CAAC;QAEtI,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,OAAO,IAAI,EAAE,CAAC;YACb,MAAM,MAAM,GAAyB,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACxD,IAAI,MAAM,KAAK,IAAI;gBAAE,MAAM;YAE3B,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;QACvB,CAAC;QAED,MAAM,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;IACxC,CAAC;CACD"}
|
|
@@ -1,23 +1,14 @@
|
|
|
1
|
-
var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
|
|
2
|
-
function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
|
|
3
|
-
return new (P || (P = Promise))(function (resolve, reject) {
|
|
4
|
-
function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
|
|
5
|
-
function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
|
|
6
|
-
function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
|
|
7
|
-
step((generator = generator.apply(thisArg, _arguments || [])).next());
|
|
8
|
-
});
|
|
9
|
-
};
|
|
10
1
|
import * as cheerio from 'cheerio';
|
|
11
|
-
import { commonsArrayUnique } from 'tscommons-
|
|
12
|
-
import { EStatus } from 'hydra-crawler-
|
|
13
|
-
import {
|
|
14
|
-
import { commonsHttpReadUrlAsBuffer } from 'nodecommons-
|
|
15
|
-
import { ComplexEnglishParser } from '../parsers/complex-english.parser';
|
|
16
|
-
import { EList } from '../enums/elist';
|
|
17
|
-
export const QUERY = (args, databaseService, lists, _expiry, parsersConfig) =>
|
|
18
|
-
const domain = args.
|
|
2
|
+
import { commonsArrayUnique } from 'tscommons-esm-core';
|
|
3
|
+
import { EStatus } from 'hydra-crawler-esm-assets';
|
|
4
|
+
import { commonsLogDoing, commonsLogError } from 'nodecommons-esm-log';
|
|
5
|
+
import { commonsHttpReadUrlAsBuffer } from 'nodecommons-esm-http';
|
|
6
|
+
import { ComplexEnglishParser } from '../parsers/complex-english.parser.mjs';
|
|
7
|
+
import { EList } from '../enums/elist.mjs';
|
|
8
|
+
export const QUERY = async (args, databaseService, lists, _expiry, parsersConfig) => {
|
|
9
|
+
const domain = args.getFirstValue('domain');
|
|
19
10
|
if (!domain) {
|
|
20
|
-
|
|
11
|
+
commonsLogError('No domain specified');
|
|
21
12
|
return;
|
|
22
13
|
}
|
|
23
14
|
const parser = new ComplexEnglishParser(undefined, undefined, parsersConfig);
|
|
@@ -25,7 +16,7 @@ export const QUERY = (args, databaseService, lists, _expiry, parsersConfig) => _
|
|
|
25
16
|
const keyDictionary = parser.getDictionary();
|
|
26
17
|
if (!keyDictionary)
|
|
27
18
|
throw new Error('No dictionary available');
|
|
28
|
-
|
|
19
|
+
const log = commonsLogDoing('complex-english-parser', `Searching for complex english detections for domain ${domain}`);
|
|
29
20
|
const result = databaseService.getUrls().find({
|
|
30
21
|
domain: domain,
|
|
31
22
|
status: EStatus.DONE,
|
|
@@ -37,8 +28,8 @@ export const QUERY = (args, databaseService, lists, _expiry, parsersConfig) => _
|
|
|
37
28
|
while (true) {
|
|
38
29
|
tally++;
|
|
39
30
|
if ((tally % 100) === 0)
|
|
40
|
-
|
|
41
|
-
const row =
|
|
31
|
+
log.progress(`${tally}`);
|
|
32
|
+
const row = await result.next();
|
|
42
33
|
if (row === null)
|
|
43
34
|
break;
|
|
44
35
|
const typecast = row;
|
|
@@ -51,18 +42,18 @@ export const QUERY = (args, databaseService, lists, _expiry, parsersConfig) => _
|
|
|
51
42
|
complexEnglish: typecast.complexEnglish
|
|
52
43
|
});
|
|
53
44
|
}
|
|
54
|
-
|
|
45
|
+
log.result(tally);
|
|
55
46
|
for (const match of matches) {
|
|
56
47
|
console.log('----------------------------------------------------');
|
|
57
48
|
console.log(match.url);
|
|
58
|
-
const data =
|
|
49
|
+
const data = await commonsHttpReadUrlAsBuffer(match.url);
|
|
59
50
|
if (!data) {
|
|
60
|
-
|
|
51
|
+
commonsLogError('Unable to read URL. Skipping');
|
|
61
52
|
continue;
|
|
62
53
|
}
|
|
63
54
|
const dom = cheerio.load(data);
|
|
64
55
|
if (!dom) {
|
|
65
|
-
|
|
56
|
+
commonsLogError('Unable to parse HTML. Skipping');
|
|
66
57
|
continue;
|
|
67
58
|
}
|
|
68
59
|
const nodes = dom('*')
|
|
@@ -100,5 +91,5 @@ export const QUERY = (args, databaseService, lists, _expiry, parsersConfig) => _
|
|
|
100
91
|
}
|
|
101
92
|
}
|
|
102
93
|
}
|
|
103
|
-
}
|
|
104
|
-
//# sourceMappingURL=complex-english.query.
|
|
94
|
+
};
|
|
95
|
+
//# sourceMappingURL=complex-english.query.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"complex-english.query.mjs","sourceRoot":"","sources":["../../src/queries/complex-english.query.mts"],"names":[],"mappings":"AAAA,OAAO,KAAK,OAAO,MAAM,SAAS,CAAC;AAGnC,OAAO,EAAE,kBAAkB,EAAmB,MAAM,oBAAoB,CAAC;AAEzE,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAInD,OAAO,EAAE,eAAe,EAAE,eAAe,EAAsB,MAAM,qBAAqB,CAAC;AAC3F,OAAO,EAAE,0BAA0B,EAAE,MAAM,sBAAsB,CAAC;AAOlE,OAAO,EAAE,oBAAoB,EAAE,MAAM,uCAAuC,CAAC;AAM7E,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAC;AAU3C,MAAM,CAAC,MAAM,KAAK,GAAW,KAAK,EAChC,IAAiB,EACjB,eAAgC,EAChC,KAAY,EACZ,OAAe,EACf,aAA6C,EAC9B,EAAE;IAClB,MAAM,MAAM,GAAqB,IAAI,CAAC,aAAa,CAAC,QAAQ,CAAC,CAAC;IAC9D,IAAI,CAAC,MAAM,EAAE,CAAC;QACb,eAAe,CAAC,qBAAqB,CAAC,CAAC;QACvC,OAAO;IACR,CAAC;IAED,MAAM,MAAM,GAAyB,IAAI,oBAAoB,CAC3D,SAAS,EACT,SAAS,EACT,aAAa,CACd,CAAC;IACF,MAAM,MAAM,GAAwB,MAAM,CAAC,SAAS,EAAE,CAAC;IAEvD,MAAM,aAAa,GAAwC,MAAM,CAAC,aAAa,EAAE,CAAC;IAClF,IAAI,CAAC,aAAa;QAAE,MAAM,IAAI,KAAK,CAAC,yBAAyB,CAAC,CAAC;IAE/D,MAAM,GAAG,GAAuB,eAAe,CAAC,wBAAwB,EAAE,uDAAuD,MAAM,EAAE,CAAC,CAAC;IAE3I,MAAM,MAAM,GAA6B,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CACrE;QACE,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,OAAO,CAAC,IAAI;QACpB,UAAU,EAAE,GAAG;QACf,cAAc,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE;KAClC,EACD,EAAE,CACH,CAAC;IAEF,MAAM,OAAO,GAAoB,EAAE,CAAC;IACpC,IAAI,KAAK,GAAW,CAAC,CAAC;IACtB,OAAO,IAAI,EAAE,CAAC;QACb,KAAK,EAAE,CAAC;QACR,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC;YAAE,GAAG,CAAC,QAAQ,CAAC,GAAG,KAAK,EAAE,CAAC,CAAC;QAElD,MAAM,GAAG,GAAc,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QAC3C,IAAI,GAAG,KAAK,IAAI;YAAE,MAAM;QAExB,MAAM,QAAQ,GAAkB,GAA+B,CAAC;QAEhE,IAAI,QAAQ,CAAC,OAAO,KAAK,SAAS,IAAI,QAAQ,CAAC,OAAO,CAAC,cAAc,CAAC,KAAK,SAAS,EAAE,CAAC;YACtF,IAAI,CAAC,MAAM,CAAC,QAAQ,CAClB,QAAQ,CAAC,OAAO,CAAC,cAAc,CAAC,EAChC,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,GAAG,CAAC,CAClC;gBAAE,SAAS;QACb,CAAC;QAED,OAAO,CAAC,IAAI,CAAC;YACX,GAAG,EAAE,GAAG,CAAC,GAAG;YACZ,cAAc,EAAE,QAAQ,CAAC,cAAc;SACxC,CAAC,CAAC;IACJ,CAAC;IAED,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;IAElB,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC7B,OAAO,CAAC,GAAG,CAAC,sDAAsD,CAAC,CAAC;QACpE,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAEvB,MAAM,IAAI,GAAqB,MAAM,0BAA0B,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAC3E,IAAI,CAAC,IAAI,EAAE,CAAC;YACX,eAAe,CAAC,8BAA8B,CAAC,CAAC;YAChD,SAAS;QACV,CAAC;QAED,MAAM,GAAG,GAAiC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC7D,IAAI,CAAC,GAAG,EAAE,CAAC;YACV,eAAe,CAAC,gCAAgC,CAAC,CAAC;YAClD,SAAS;QACV,CAAC;QAED,MAAM,KAAK,GAAyB,GAAG,CAAC,GAAG,CAAC;aACzC,QAAQ,EAAE;YACX,sEAAsE;aACrE,MAAM,CAAC,CAAC,MAAc,EAAE,OAAgB,EAAW,EAAE,CAAE,OAAe,CAAC,QAAQ,KAAK,CAAC,CAAC,CAAC;QAE1F,MAAM,WAAW,GAAa,EAAE,CAAC;QAEjC,GAAG,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwC,EAAQ,EAAE;YAClF,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;QACvC,CAAC,CAAC,CAAC;QAEH,MAAM,KAAK,GAAa,WAAW;aAChC,IAAI,CAAC,IAAI,CAAC;aACV,OAAO,CAAC,YAAY,EAAE,IAAI,CAAC;aAC3B,KAAK,CAAC,IAAI,CAAC;aACX,GAAG,CAAC,CAAC,CAAS,EAAU,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;aACpC,MAAM,CAAC,CAAC,CAAS,EAAW,EAAE,CAAC,CAAC,KAAK,EAAE,CAAC,CAAC;QAE5C,MAAM,MAAM,GAAa,kBAAkB,CAAC,KAAK,CAAC,CAAC;QAEnD,KAAK,MAAM,IAAI,IAAI,MAAM,EAAE,CAAC;YAC3B,IAAI,OAAO,GAAW,IAAI,CAAC;YAE3B,KAAK,MAAM,OAAO,IAAI,KAAK,CAAC,cAAc,EAAE,CAAC;gBAC5C,MAAM,KAAK,GAAqB,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;gBACpD,IAAI,CAAC,KAAK;oBAAE,SAAS;gBAErB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;oBAAE,SAAS;gBAEhC,MAAM,WAAW,GAAa,aAAa,CAAC,OAAO,CAAC,CAAC;gBAErD,MAAM,KAAK,GAAa,CAAE,KAAK,OAAO,EAAE,CAAE,CAAC;gBAC3C,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC;oBAAE,KAAK,CAAC,IAAI,CAAC,KAAK,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;gBAErE,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,KAAK,EAAE,MAAM,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;YAC9D,CAAC;YAED,IAAI,OAAO,KAAK,IAAI,EAAE,CAAC;gBACtB,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;gBACrB,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YACjB,CAAC;QACF,CAAC;IACF,CAAC;AACF,CAAC,CAAC"}
|
|
@@ -0,0 +1,80 @@
|
|
|
1
|
+
import { ECommonsCsvColumnType } from 'tscommons-esm-format';
|
|
2
|
+
import { isIUrl } from 'hydra-crawler-esm-assets';
|
|
3
|
+
import { isTLink } from 'hydra-crawler-esm-assets';
|
|
4
|
+
import { EStatus } from 'hydra-crawler-esm-assets';
|
|
5
|
+
import { commonsLogDoing } from 'nodecommons-esm-log';
|
|
6
|
+
import { CommonsCsv } from 'nodecommons-esm-file';
|
|
7
|
+
export const QUERY = async (args, databaseService, _lists, _expiry, _parsersConfig) => {
|
|
8
|
+
const filename = args.getFirstValue('filename');
|
|
9
|
+
let urls;
|
|
10
|
+
{ // scope
|
|
11
|
+
const log = commonsLogDoing('flash-content-query', 'Searching for FLV and SWF URLs');
|
|
12
|
+
const results = databaseService.getUrls().find({
|
|
13
|
+
url: /\.(flv|swf)$/i,
|
|
14
|
+
status: { $in: [
|
|
15
|
+
EStatus.ACTIVE,
|
|
16
|
+
EStatus.DENY,
|
|
17
|
+
EStatus.DISALLOWED,
|
|
18
|
+
EStatus.DONE,
|
|
19
|
+
EStatus.FAILED,
|
|
20
|
+
EStatus.QUEUED
|
|
21
|
+
] }
|
|
22
|
+
}, {});
|
|
23
|
+
urls = await databaseService.listQueryResults(results, isIUrl);
|
|
24
|
+
log.result(urls.length);
|
|
25
|
+
}
|
|
26
|
+
const urlLinks = new Map();
|
|
27
|
+
{ // scope
|
|
28
|
+
const log = commonsLogDoing('flash-content-query', 'Searching for outgoing links to URLs');
|
|
29
|
+
let tally = 0;
|
|
30
|
+
for (const url of urls) {
|
|
31
|
+
const results3 = databaseService.getLinks().find({
|
|
32
|
+
outgoing: url.url
|
|
33
|
+
}, {});
|
|
34
|
+
const links = await databaseService.listQueryResults(results3, isTLink);
|
|
35
|
+
if (links.length === 0)
|
|
36
|
+
continue;
|
|
37
|
+
tally += links.length;
|
|
38
|
+
log.progress(tally);
|
|
39
|
+
urlLinks.set(url, links);
|
|
40
|
+
}
|
|
41
|
+
log.result(tally);
|
|
42
|
+
}
|
|
43
|
+
const rows = [];
|
|
44
|
+
{ // scope
|
|
45
|
+
const log = commonsLogDoing('flash-content-query', 'Building CSV array');
|
|
46
|
+
for (const url of urls) {
|
|
47
|
+
for (const link of (urlLinks.get(url) || [])) {
|
|
48
|
+
rows.push({
|
|
49
|
+
src: link.url,
|
|
50
|
+
dest: url.url
|
|
51
|
+
});
|
|
52
|
+
}
|
|
53
|
+
}
|
|
54
|
+
rows
|
|
55
|
+
.sort((a, b) => {
|
|
56
|
+
if (a.src < b.src)
|
|
57
|
+
return -1;
|
|
58
|
+
if (a.src > b.src)
|
|
59
|
+
return 1;
|
|
60
|
+
if (a.dest < b.dest)
|
|
61
|
+
return -1;
|
|
62
|
+
if (a.dest > b.dest)
|
|
63
|
+
return 1;
|
|
64
|
+
return 0;
|
|
65
|
+
});
|
|
66
|
+
log.success();
|
|
67
|
+
}
|
|
68
|
+
const csv = new CommonsCsv([
|
|
69
|
+
{
|
|
70
|
+
name: 'src',
|
|
71
|
+
type: ECommonsCsvColumnType.STRING
|
|
72
|
+
},
|
|
73
|
+
{
|
|
74
|
+
name: 'dest',
|
|
75
|
+
type: ECommonsCsvColumnType.STRING
|
|
76
|
+
}
|
|
77
|
+
]);
|
|
78
|
+
csv.save(rows, filename, true);
|
|
79
|
+
};
|
|
80
|
+
//# sourceMappingURL=flash-content.query.mjs.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"flash-content.query.mjs","sourceRoot":"","sources":["../../src/queries/flash-content.query.mts"],"names":[],"mappings":"AAGA,OAAO,EAAE,qBAAqB,EAAE,MAAM,sBAAsB,CAAC;AAE7D,OAAO,EAAQ,MAAM,EAAE,MAAM,0BAA0B,CAAC;AACxD,OAAO,EAAS,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAC1D,OAAO,EAAE,OAAO,EAAE,MAAM,0BAA0B,CAAC;AAGnD,OAAO,EAAE,eAAe,EAAsB,MAAM,qBAAqB,CAAC;AAC1E,OAAO,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAC;AAgBlD,MAAM,CAAC,MAAM,KAAK,GAAW,KAAK,EAChC,IAAiB,EACjB,eAAgC,EAChC,MAAa,EACb,OAAe,EACf,cAA8C,EAC/B,EAAE;IAClB,MAAM,QAAQ,GAAW,IAAI,CAAC,aAAa,CAAC,UAAU,CAAC,CAAC;IAExD,IAAI,IAAY,CAAC;IACjB,CAAC,CAAC,QAAQ;QACT,MAAM,GAAG,GAAuB,eAAe,CAAC,qBAAqB,EAAE,gCAAgC,CAAC,CAAC;QACzG,MAAM,OAAO,GAA6B,eAAe,CAAC,OAAO,EAAE,CAAC,IAAI,CACtE;YACE,GAAG,EAAE,eAAe;YACpB,MAAM,EAAE,EAAE,GAAG,EAAE;oBACb,OAAO,CAAC,MAAM;oBACd,OAAO,CAAC,IAAI;oBACZ,OAAO,CAAC,UAAU;oBAClB,OAAO,CAAC,IAAI;oBACZ,OAAO,CAAC,MAAM;oBACd,OAAO,CAAC,MAAM;iBACf,EAAE;SACJ,EACD,EAAE,CACH,CAAC;QACF,IAAI,GAAG,MAAM,eAAe,CAAC,gBAAgB,CAC3C,OAAO,EACP,MAAM,CACP,CAAC;QACF,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IACzB,CAAC;IAED,MAAM,QAAQ,GAAuB,IAAI,GAAG,EAAiB,CAAC;IAC9D,CAAC,CAAC,QAAQ;QACT,MAAM,GAAG,GAAuB,eAAe,CAAC,qBAAqB,EAAE,sCAAsC,CAAC,CAAC;QAC/G,IAAI,KAAK,GAAW,CAAC,CAAC;QACtB,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;YACxB,MAAM,QAAQ,GAA8B,eAAe,CAAC,QAAQ,EAAE,CAAC,IAAI,CACzE;gBACE,QAAQ,EAAE,GAAG,CAAC,GAAG;aAClB,EACD,EAAE,CACH,CAAC;YACF,MAAM,KAAK,GAAY,MAAM,eAAe,CAAC,gBAAgB,CAC3D,QAAQ,EACR,OAAO,CACR,CAAC;YAEF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;gBAAE,SAAS;YAEjC,KAAK,IAAI,KAAK,CAAC,MAAM,CAAC;YACtB,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;YAEpB,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;QAC1B,CAAC;QACD,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;IACnB,CAAC;IAED,MAAM,IAAI,GAAW,EAAE,CAAC;IACxB,CAAC,CAAC,QAAQ;QACT,MAAM,GAAG,GAAuB,eAAe,CAAC,qBAAqB,EAAE,oBAAoB,CAAC,CAAC;QAE7F,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;YACxB,KAAK,MAAM,IAAI,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;gBAC9C,IAAI,CAAC,IAAI,CAAC;oBACR,GAAG,EAAE,IAAI,CAAC,GAAG;oBACb,IAAI,EAAE,GAAG,CAAC,GAAG;iBACd,CAAC,CAAC;YACJ,CAAC;QACF,CAAC;QACD,IAAI;aACD,IAAI,CAAC,CAAC,CAAO,EAAE,CAAO,EAAU,EAAE;YAClC,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,GAAG;gBAAE,OAAO,CAAC,CAAC,CAAC;YAC7B,IAAI,CAAC,CAAC,GAAG,GAAG,CAAC,CAAC,GAAG;gBAAE,OAAO,CAAC,CAAC;YAC5B,IAAI,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI;gBAAE,OAAO,CAAC,CAAC,CAAC;YAC/B,IAAI,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI;gBAAE,OAAO,CAAC,CAAC;YAC9B,OAAO,CAAC,CAAC;QACV,CAAC,CAAC,CAAC;QAEL,GAAG,CAAC,OAAO,EAAE,CAAC;IACf,CAAC;IAED,MAAM,GAAG,GAAe,IAAI,UAAU,CAAC;QACrC;YACE,IAAI,EAAE,KAAK;YACX,IAAI,EAAE,qBAAqB,CAAC,MAAM;SACnC;QACD;YACE,IAAI,EAAE,MAAM;YACZ,IAAI,EAAE,qBAAqB,CAAC,MAAM;SACnC;KACF,CAAC,CAAC;IACH,GAAG,CAAC,IAAI,CACN,IAAI,EACJ,QAAQ,EACR,IAAI,CACL,CAAC;AACH,CAAC,CAAC"}
|