@mantra-ai/core 0.1.0
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/ai/google/client.d.ts +67 -0
- package/dist/ai/google/client.d.ts.map +1 -0
- package/dist/ai/google/client.js +169 -0
- package/dist/ai/google/client.js.map +1 -0
- package/dist/ai/google/generate.d.ts +10 -0
- package/dist/ai/google/generate.d.ts.map +1 -0
- package/dist/ai/google/generate.js +137 -0
- package/dist/ai/google/generate.js.map +1 -0
- package/dist/ai/google/index.d.ts +4 -0
- package/dist/ai/google/index.d.ts.map +1 -0
- package/dist/ai/google/index.js +4 -0
- package/dist/ai/google/index.js.map +1 -0
- package/dist/ai/google/types.d.ts +88 -0
- package/dist/ai/google/types.d.ts.map +1 -0
- package/dist/ai/google/types.js +55 -0
- package/dist/ai/google/types.js.map +1 -0
- package/dist/ai/index.d.ts +3 -0
- package/dist/ai/index.d.ts.map +1 -0
- package/dist/ai/index.js +3 -0
- package/dist/ai/index.js.map +1 -0
- package/dist/ai/openai/client.d.ts +22 -0
- package/dist/ai/openai/client.d.ts.map +1 -0
- package/dist/ai/openai/client.js +49 -0
- package/dist/ai/openai/client.js.map +1 -0
- package/dist/ai/openai/generate.d.ts +14 -0
- package/dist/ai/openai/generate.d.ts.map +1 -0
- package/dist/ai/openai/generate.js +178 -0
- package/dist/ai/openai/generate.js.map +1 -0
- package/dist/ai/openai/index.d.ts +4 -0
- package/dist/ai/openai/index.d.ts.map +1 -0
- package/dist/ai/openai/index.js +4 -0
- package/dist/ai/openai/index.js.map +1 -0
- package/dist/ai/openai/types.d.ts +86 -0
- package/dist/ai/openai/types.d.ts.map +1 -0
- package/dist/ai/openai/types.js +56 -0
- package/dist/ai/openai/types.js.map +1 -0
- package/dist/ai/prompts/index.d.ts +1 -0
- package/dist/ai/prompts/index.d.ts.map +1 -0
- package/dist/ai/prompts/index.js +2 -0
- package/dist/ai/prompts/index.js.map +1 -0
- package/dist/errors/index.d.ts +3 -0
- package/dist/errors/index.d.ts.map +1 -0
- package/dist/errors/index.js +4 -0
- package/dist/errors/index.js.map +1 -0
- package/dist/errors/schemas.d.ts +304 -0
- package/dist/errors/schemas.d.ts.map +1 -0
- package/dist/errors/schemas.js +57 -0
- package/dist/errors/schemas.js.map +1 -0
- package/dist/errors/types.d.ts +30 -0
- package/dist/errors/types.d.ts.map +1 -0
- package/dist/errors/types.js +33 -0
- package/dist/errors/types.js.map +1 -0
- package/dist/index.d.ts +21 -0
- package/dist/index.d.ts.map +1 -0
- package/dist/index.js +21 -0
- package/dist/index.js.map +1 -0
- package/dist/normalization/jats/index.d.ts +4 -0
- package/dist/normalization/jats/index.d.ts.map +1 -0
- package/dist/normalization/jats/index.js +3 -0
- package/dist/normalization/jats/index.js.map +1 -0
- package/dist/normalization/jats/normalize.d.ts +7 -0
- package/dist/normalization/jats/normalize.d.ts.map +1 -0
- package/dist/normalization/jats/normalize.js +213 -0
- package/dist/normalization/jats/normalize.js.map +1 -0
- package/dist/normalization/jats/utils/build/finalize.d.ts +3 -0
- package/dist/normalization/jats/utils/build/finalize.d.ts.map +1 -0
- package/dist/normalization/jats/utils/build/finalize.js +462 -0
- package/dist/normalization/jats/utils/build/finalize.js.map +1 -0
- package/dist/normalization/jats/utils/build/flatten.d.ts +20 -0
- package/dist/normalization/jats/utils/build/flatten.d.ts.map +1 -0
- package/dist/normalization/jats/utils/build/flatten.js +502 -0
- package/dist/normalization/jats/utils/build/flatten.js.map +1 -0
- package/dist/normalization/jats/utils/build/meta.d.ts +10 -0
- package/dist/normalization/jats/utils/build/meta.d.ts.map +1 -0
- package/dist/normalization/jats/utils/build/meta.js +32 -0
- package/dist/normalization/jats/utils/build/meta.js.map +1 -0
- package/dist/normalization/jats/utils/build/version.d.ts +3 -0
- package/dist/normalization/jats/utils/build/version.d.ts.map +1 -0
- package/dist/normalization/jats/utils/build/version.js +11 -0
- package/dist/normalization/jats/utils/build/version.js.map +1 -0
- package/dist/normalization/jats/utils/category.d.ts +11 -0
- package/dist/normalization/jats/utils/category.d.ts.map +1 -0
- package/dist/normalization/jats/utils/category.js +431 -0
- package/dist/normalization/jats/utils/category.js.map +1 -0
- package/dist/normalization/jats/utils/collectors/abstracts.d.ts +3 -0
- package/dist/normalization/jats/utils/collectors/abstracts.d.ts.map +1 -0
- package/dist/normalization/jats/utils/collectors/abstracts.js +168 -0
- package/dist/normalization/jats/utils/collectors/abstracts.js.map +1 -0
- package/dist/normalization/jats/utils/collectors/back.d.ts +35 -0
- package/dist/normalization/jats/utils/collectors/back.d.ts.map +1 -0
- package/dist/normalization/jats/utils/collectors/back.js +801 -0
- package/dist/normalization/jats/utils/collectors/back.js.map +1 -0
- package/dist/normalization/jats/utils/collectors/contributors.d.ts +4 -0
- package/dist/normalization/jats/utils/collectors/contributors.d.ts.map +1 -0
- package/dist/normalization/jats/utils/collectors/contributors.js +77 -0
- package/dist/normalization/jats/utils/collectors/contributors.js.map +1 -0
- package/dist/normalization/jats/utils/collectors/keywords.d.ts +2 -0
- package/dist/normalization/jats/utils/collectors/keywords.d.ts.map +1 -0
- package/dist/normalization/jats/utils/collectors/keywords.js +14 -0
- package/dist/normalization/jats/utils/collectors/keywords.js.map +1 -0
- package/dist/normalization/jats/utils/collectors/meta.d.ts +6 -0
- package/dist/normalization/jats/utils/collectors/meta.d.ts.map +1 -0
- package/dist/normalization/jats/utils/collectors/meta.js +103 -0
- package/dist/normalization/jats/utils/collectors/meta.js.map +1 -0
- package/dist/normalization/jats/utils/collectors/sections.d.ts +7 -0
- package/dist/normalization/jats/utils/collectors/sections.d.ts.map +1 -0
- package/dist/normalization/jats/utils/collectors/sections.js +484 -0
- package/dist/normalization/jats/utils/collectors/sections.js.map +1 -0
- package/dist/normalization/jats/utils/licenses.d.ts +5 -0
- package/dist/normalization/jats/utils/licenses.d.ts.map +1 -0
- package/dist/normalization/jats/utils/licenses.js +64 -0
- package/dist/normalization/jats/utils/licenses.js.map +1 -0
- package/dist/normalization/jats/utils/po/nodes.d.ts +6 -0
- package/dist/normalization/jats/utils/po/nodes.d.ts.map +1 -0
- package/dist/normalization/jats/utils/po/nodes.js +60 -0
- package/dist/normalization/jats/utils/po/nodes.js.map +1 -0
- package/dist/normalization/jats/utils/po/query.d.ts +7 -0
- package/dist/normalization/jats/utils/po/query.d.ts.map +1 -0
- package/dist/normalization/jats/utils/po/query.js +67 -0
- package/dist/normalization/jats/utils/po/query.js.map +1 -0
- package/dist/normalization/jats/utils/po/serialize.d.ts +4 -0
- package/dist/normalization/jats/utils/po/serialize.d.ts.map +1 -0
- package/dist/normalization/jats/utils/po/serialize.js +329 -0
- package/dist/normalization/jats/utils/po/serialize.js.map +1 -0
- package/dist/normalization/jats/utils/po/text.d.ts +7 -0
- package/dist/normalization/jats/utils/po/text.d.ts.map +1 -0
- package/dist/normalization/jats/utils/po/text.js +114 -0
- package/dist/normalization/jats/utils/po/text.js.map +1 -0
- package/dist/normalization/jats/utils/references.d.ts +26 -0
- package/dist/normalization/jats/utils/references.d.ts.map +1 -0
- package/dist/normalization/jats/utils/references.js +371 -0
- package/dist/normalization/jats/utils/references.js.map +1 -0
- package/dist/normalization/jats/utils/strings.d.ts +8 -0
- package/dist/normalization/jats/utils/strings.d.ts.map +1 -0
- package/dist/normalization/jats/utils/strings.js +197 -0
- package/dist/normalization/jats/utils/strings.js.map +1 -0
- package/dist/normalization/jats/utils/types.d.ts +233 -0
- package/dist/normalization/jats/utils/types.d.ts.map +1 -0
- package/dist/normalization/jats/utils/types.js +2 -0
- package/dist/normalization/jats/utils/types.js.map +1 -0
- package/dist/normalization/jats/utils/xml.d.ts +5 -0
- package/dist/normalization/jats/utils/xml.d.ts.map +1 -0
- package/dist/normalization/jats/utils/xml.js +69 -0
- package/dist/normalization/jats/utils/xml.js.map +1 -0
- package/dist/normalization/normalized-doc-schema.d.ts +1094 -0
- package/dist/normalization/normalized-doc-schema.d.ts.map +1 -0
- package/dist/normalization/normalized-doc-schema.js +410 -0
- package/dist/normalization/normalized-doc-schema.js.map +1 -0
- package/dist/normalization/pdf/index.d.ts +4 -0
- package/dist/normalization/pdf/index.d.ts.map +1 -0
- package/dist/normalization/pdf/index.js +3 -0
- package/dist/normalization/pdf/index.js.map +1 -0
- package/dist/normalization/pdf/normalize.d.ts +31 -0
- package/dist/normalization/pdf/normalize.d.ts.map +1 -0
- package/dist/normalization/pdf/normalize.js +321 -0
- package/dist/normalization/pdf/normalize.js.map +1 -0
- package/dist/normalization/pdf/prompt.d.ts +3 -0
- package/dist/normalization/pdf/prompt.d.ts.map +1 -0
- package/dist/normalization/pdf/prompt.js +118 -0
- package/dist/normalization/pdf/prompt.js.map +1 -0
- package/dist/sources/arxiv/client.d.ts +4 -0
- package/dist/sources/arxiv/client.d.ts.map +1 -0
- package/dist/sources/arxiv/client.js +13 -0
- package/dist/sources/arxiv/client.js.map +1 -0
- package/dist/sources/biorxiv/client.d.ts +21 -0
- package/dist/sources/biorxiv/client.d.ts.map +1 -0
- package/dist/sources/biorxiv/client.js +173 -0
- package/dist/sources/biorxiv/client.js.map +1 -0
- package/dist/sources/crossref/client.d.ts +3 -0
- package/dist/sources/crossref/client.d.ts.map +1 -0
- package/dist/sources/crossref/client.js +24 -0
- package/dist/sources/crossref/client.js.map +1 -0
- package/dist/sources/europepmc/client.d.ts +3 -0
- package/dist/sources/europepmc/client.d.ts.map +1 -0
- package/dist/sources/europepmc/client.js +29 -0
- package/dist/sources/europepmc/client.js.map +1 -0
- package/dist/sources/medrxiv/browser.d.ts +16 -0
- package/dist/sources/medrxiv/browser.d.ts.map +1 -0
- package/dist/sources/medrxiv/browser.js +210 -0
- package/dist/sources/medrxiv/browser.js.map +1 -0
- package/dist/sources/medrxiv/client.d.ts +34 -0
- package/dist/sources/medrxiv/client.d.ts.map +1 -0
- package/dist/sources/medrxiv/client.js +673 -0
- package/dist/sources/medrxiv/client.js.map +1 -0
- package/dist/sources/medrxiv/shared.d.ts +7 -0
- package/dist/sources/medrxiv/shared.d.ts.map +1 -0
- package/dist/sources/medrxiv/shared.js +18 -0
- package/dist/sources/medrxiv/shared.js.map +1 -0
- package/dist/sources/plos/client.d.ts +13 -0
- package/dist/sources/plos/client.d.ts.map +1 -0
- package/dist/sources/plos/client.js +147 -0
- package/dist/sources/plos/client.js.map +1 -0
- package/dist/sources/preprint-discovery.d.ts +55 -0
- package/dist/sources/preprint-discovery.d.ts.map +1 -0
- package/dist/sources/preprint-discovery.js +115 -0
- package/dist/sources/preprint-discovery.js.map +1 -0
- package/dist/types/expand.d.ts +5 -0
- package/dist/types/expand.d.ts.map +1 -0
- package/dist/types/expand.js +20 -0
- package/dist/types/expand.js.map +1 -0
- package/dist/types/methods-types.d.ts +37 -0
- package/dist/types/methods-types.d.ts.map +1 -0
- package/dist/types/methods-types.js +2 -0
- package/dist/types/methods-types.js.map +1 -0
- package/dist/types/multi-input-types.d.ts +57 -0
- package/dist/types/multi-input-types.d.ts.map +1 -0
- package/dist/types/multi-input-types.js +2 -0
- package/dist/types/multi-input-types.js.map +1 -0
- package/dist/types/paper/types.d.ts +41 -0
- package/dist/types/paper/types.d.ts.map +1 -0
- package/dist/types/paper/types.js +2 -0
- package/dist/types/paper/types.js.map +1 -0
- package/dist/types/results-types.d.ts +122 -0
- package/dist/types/results-types.d.ts.map +1 -0
- package/dist/types/results-types.js +17 -0
- package/dist/types/results-types.js.map +1 -0
- package/dist/types/supp-types.d.ts +6 -0
- package/dist/types/supp-types.d.ts.map +1 -0
- package/dist/types/supp-types.js +2 -0
- package/dist/types/supp-types.js.map +1 -0
- package/dist/types/version.d.ts +1828 -0
- package/dist/types/version.d.ts.map +1 -0
- package/dist/types/version.js +311 -0
- package/dist/types/version.js.map +1 -0
- package/dist/types/work.d.ts +4455 -0
- package/dist/types/work.d.ts.map +1 -0
- package/dist/types/work.js +330 -0
- package/dist/types/work.js.map +1 -0
- package/dist/works/adapters/crossref.d.ts +28 -0
- package/dist/works/adapters/crossref.d.ts.map +1 -0
- package/dist/works/adapters/crossref.js +43 -0
- package/dist/works/adapters/crossref.js.map +1 -0
- package/dist/works/adapters/europepmc.d.ts +14 -0
- package/dist/works/adapters/europepmc.d.ts.map +1 -0
- package/dist/works/adapters/europepmc.js +46 -0
- package/dist/works/adapters/europepmc.js.map +1 -0
- package/dist/works/adapters/openalex.d.ts +5 -0
- package/dist/works/adapters/openalex.d.ts.map +1 -0
- package/dist/works/adapters/openalex.js +75 -0
- package/dist/works/adapters/openalex.js.map +1 -0
- package/dist/works/errors.d.ts +23 -0
- package/dist/works/errors.d.ts.map +1 -0
- package/dist/works/errors.js +37 -0
- package/dist/works/errors.js.map +1 -0
- package/dist/works/id/detect-identifier.d.ts +15 -0
- package/dist/works/id/detect-identifier.d.ts.map +1 -0
- package/dist/works/id/detect-identifier.js +50 -0
- package/dist/works/id/detect-identifier.js.map +1 -0
- package/dist/works/id/normalize-external-id.d.ts +3 -0
- package/dist/works/id/normalize-external-id.d.ts.map +1 -0
- package/dist/works/id/normalize-external-id.js +44 -0
- package/dist/works/id/normalize-external-id.js.map +1 -0
- package/dist/works/id/normalize-ids.d.ts +66 -0
- package/dist/works/id/normalize-ids.d.ts.map +1 -0
- package/dist/works/id/normalize-ids.js +112 -0
- package/dist/works/id/normalize-ids.js.map +1 -0
- package/dist/works/id/normalize-internals.d.ts +7 -0
- package/dist/works/id/normalize-internals.d.ts.map +1 -0
- package/dist/works/id/normalize-internals.js +65 -0
- package/dist/works/id/normalize-internals.js.map +1 -0
- package/dist/works/id/resolve.d.ts +31 -0
- package/dist/works/id/resolve.d.ts.map +1 -0
- package/dist/works/id/resolve.js +123 -0
- package/dist/works/id/resolve.js.map +1 -0
- package/dist/works/id/resolveIds/assign.d.ts +4 -0
- package/dist/works/id/resolveIds/assign.d.ts.map +1 -0
- package/dist/works/id/resolveIds/assign.js +15 -0
- package/dist/works/id/resolveIds/assign.js.map +1 -0
- package/dist/works/id/resolveIds/flags.d.ts +11 -0
- package/dist/works/id/resolveIds/flags.d.ts.map +1 -0
- package/dist/works/id/resolveIds/flags.js +27 -0
- package/dist/works/id/resolveIds/flags.js.map +1 -0
- package/dist/works/id/resolveIds/idctx.d.ts +4 -0
- package/dist/works/id/resolveIds/idctx.d.ts.map +1 -0
- package/dist/works/id/resolveIds/idctx.js +25 -0
- package/dist/works/id/resolveIds/idctx.js.map +1 -0
- package/dist/works/id/resolveIds/index.d.ts +13 -0
- package/dist/works/id/resolveIds/index.d.ts.map +1 -0
- package/dist/works/id/resolveIds/index.js +498 -0
- package/dist/works/id/resolveIds/index.js.map +1 -0
- package/dist/works/id/resolveIds/versioning.d.ts +27 -0
- package/dist/works/id/resolveIds/versioning.d.ts.map +1 -0
- package/dist/works/id/resolveIds/versioning.js +156 -0
- package/dist/works/id/resolveIds/versioning.js.map +1 -0
- package/dist/works/id/resolveIds/workWhere.d.ts +3 -0
- package/dist/works/id/resolveIds/workWhere.d.ts.map +1 -0
- package/dist/works/id/resolveIds/workWhere.js +35 -0
- package/dist/works/id/resolveIds/workWhere.js.map +1 -0
- package/dist/works/id/types.d.ts +6 -0
- package/dist/works/id/types.d.ts.map +1 -0
- package/dist/works/id/types.js +2 -0
- package/dist/works/id/types.js.map +1 -0
- package/dist/works/pdf-fallback/candidates.d.ts +12 -0
- package/dist/works/pdf-fallback/candidates.d.ts.map +1 -0
- package/dist/works/pdf-fallback/candidates.js +51 -0
- package/dist/works/pdf-fallback/candidates.js.map +1 -0
- package/dist/works/pdf-fallback/fetch.d.ts +21 -0
- package/dist/works/pdf-fallback/fetch.d.ts.map +1 -0
- package/dist/works/pdf-fallback/fetch.js +89 -0
- package/dist/works/pdf-fallback/fetch.js.map +1 -0
- package/dist/works/pdf-fallback/index.d.ts +28 -0
- package/dist/works/pdf-fallback/index.d.ts.map +1 -0
- package/dist/works/pdf-fallback/index.js +35 -0
- package/dist/works/pdf-fallback/index.js.map +1 -0
- package/dist/works/plan.d.ts +8 -0
- package/dist/works/plan.d.ts.map +1 -0
- package/dist/works/plan.js +62 -0
- package/dist/works/plan.js.map +1 -0
- package/dist/works/strategies/arxiv.d.ts +3 -0
- package/dist/works/strategies/arxiv.d.ts.map +1 -0
- package/dist/works/strategies/arxiv.js +56 -0
- package/dist/works/strategies/arxiv.js.map +1 -0
- package/dist/works/strategies/biorxiv.d.ts +3 -0
- package/dist/works/strategies/biorxiv.d.ts.map +1 -0
- package/dist/works/strategies/biorxiv.js +63 -0
- package/dist/works/strategies/biorxiv.js.map +1 -0
- package/dist/works/strategies/europepmc.d.ts +3 -0
- package/dist/works/strategies/europepmc.d.ts.map +1 -0
- package/dist/works/strategies/europepmc.js +15 -0
- package/dist/works/strategies/europepmc.js.map +1 -0
- package/dist/works/strategies/index.d.ts +12 -0
- package/dist/works/strategies/index.d.ts.map +1 -0
- package/dist/works/strategies/index.js +19 -0
- package/dist/works/strategies/index.js.map +1 -0
- package/dist/works/strategies/landing-url.d.ts +3 -0
- package/dist/works/strategies/landing-url.d.ts.map +1 -0
- package/dist/works/strategies/landing-url.js +10 -0
- package/dist/works/strategies/landing-url.js.map +1 -0
- package/dist/works/strategies/medrxiv.d.ts +3 -0
- package/dist/works/strategies/medrxiv.d.ts.map +1 -0
- package/dist/works/strategies/medrxiv.js +47 -0
- package/dist/works/strategies/medrxiv.js.map +1 -0
- package/dist/works/strategies/plos.d.ts +3 -0
- package/dist/works/strategies/plos.d.ts.map +1 -0
- package/dist/works/strategies/plos.js +15 -0
- package/dist/works/strategies/plos.js.map +1 -0
- package/dist/works/strategies/shared.d.ts +11 -0
- package/dist/works/strategies/shared.d.ts.map +1 -0
- package/dist/works/strategies/shared.js +97 -0
- package/dist/works/strategies/shared.js.map +1 -0
- package/dist/works/strategies/ten1101.d.ts +3 -0
- package/dist/works/strategies/ten1101.d.ts.map +1 -0
- package/dist/works/strategies/ten1101.js +84 -0
- package/dist/works/strategies/ten1101.js.map +1 -0
- package/dist/works/text/acquire-fulltext.d.ts +7 -0
- package/dist/works/text/acquire-fulltext.d.ts.map +1 -0
- package/dist/works/text/acquire-fulltext.js +62 -0
- package/dist/works/text/acquire-fulltext.js.map +1 -0
- package/dist/works/text/normalize.d.ts +40 -0
- package/dist/works/text/normalize.d.ts.map +1 -0
- package/dist/works/text/normalize.js +188 -0
- package/dist/works/text/normalize.js.map +1 -0
- package/dist/works/types.d.ts +215 -0
- package/dist/works/types.d.ts.map +1 -0
- package/dist/works/types.js +6 -0
- package/dist/works/types.js.map +1 -0
- package/dist/works/util/debug.d.ts +7 -0
- package/dist/works/util/debug.d.ts.map +1 -0
- package/dist/works/util/debug.js +9 -0
- package/dist/works/util/debug.js.map +1 -0
- package/dist/works/util/license.d.ts +9 -0
- package/dist/works/util/license.d.ts.map +1 -0
- package/dist/works/util/license.js +39 -0
- package/dist/works/util/license.js.map +1 -0
- package/dist/works/util/normalize.d.ts +2 -0
- package/dist/works/util/normalize.d.ts.map +1 -0
- package/dist/works/util/normalize.js +76 -0
- package/dist/works/util/normalize.js.map +1 -0
- package/dist/works/util/parse.d.ts +8 -0
- package/dist/works/util/parse.d.ts.map +1 -0
- package/dist/works/util/parse.js +32 -0
- package/dist/works/util/parse.js.map +1 -0
- package/dist/works/util/source.d.ts +10 -0
- package/dist/works/util/source.d.ts.map +1 -0
- package/dist/works/util/source.js +48 -0
- package/dist/works/util/source.js.map +1 -0
- package/dist/works/util/version-label.d.ts +2 -0
- package/dist/works/util/version-label.d.ts.map +1 -0
- package/dist/works/util/version-label.js +8 -0
- package/dist/works/util/version-label.js.map +1 -0
- package/dist/works/util/work-id.d.ts +2 -0
- package/dist/works/util/work-id.d.ts.map +1 -0
- package/dist/works/util/work-id.js +27 -0
- package/dist/works/util/work-id.js.map +1 -0
- package/package.json +208 -0
|
@@ -0,0 +1,27 @@
|
|
|
1
|
+
export type VersionLabel = "published" | `v${number}`;
|
|
2
|
+
export type VersionStage = "journal" | "preprint" | "unknown";
|
|
3
|
+
export type VersionInference = {
|
|
4
|
+
label: VersionLabel;
|
|
5
|
+
stage: VersionStage;
|
|
6
|
+
confidence: number;
|
|
7
|
+
evidence: string[];
|
|
8
|
+
};
|
|
9
|
+
type MinimalIds = {
|
|
10
|
+
doi?: string | null;
|
|
11
|
+
arxivId?: string | null;
|
|
12
|
+
biorxivId?: string | null;
|
|
13
|
+
medrxivId?: string | null;
|
|
14
|
+
landingUrl?: string | null;
|
|
15
|
+
journal?: string | null;
|
|
16
|
+
openalexHostVenue?: string | null;
|
|
17
|
+
};
|
|
18
|
+
type ExternalHints = {
|
|
19
|
+
crossrefType?: string | null;
|
|
20
|
+
crossrefSubtype?: string | null;
|
|
21
|
+
openalexVenueName?: string | null;
|
|
22
|
+
openalexIsPreprintVenue?: boolean | null;
|
|
23
|
+
latestKnownPreprintVersion?: number | null;
|
|
24
|
+
};
|
|
25
|
+
export declare function inferVersionLabel(ids: MinimalIds, hints?: ExternalHints): Promise<VersionInference>;
|
|
26
|
+
export {};
|
|
27
|
+
//# sourceMappingURL=versioning.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"versioning.d.ts","sourceRoot":"","sources":["../../../../src/works/id/resolveIds/versioning.ts"],"names":[],"mappings":"AAEA,MAAM,MAAM,YAAY,GAAG,WAAW,GAAG,IAAI,MAAM,EAAE,CAAC;AACtD,MAAM,MAAM,YAAY,GAAG,SAAS,GAAG,UAAU,GAAG,SAAS,CAAC;AAE9D,MAAM,MAAM,gBAAgB,GAAG;IAC7B,KAAK,EAAE,YAAY,CAAC;IACpB,KAAK,EAAE,YAAY,CAAC;IACpB,UAAU,EAAE,MAAM,CAAC;IACnB,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB,CAAC;AAEF,KAAK,UAAU,GAAG;IAChB,GAAG,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACpB,OAAO,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,SAAS,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IAC1B,SAAS,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IAC1B,UAAU,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IAC3B,OAAO,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,iBAAiB,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CACnC,CAAC;AAEF,KAAK,aAAa,GAAG;IAEnB,YAAY,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IAC7B,eAAe,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IAChC,iBAAiB,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IAClC,uBAAuB,CAAC,EAAE,OAAO,GAAG,IAAI,CAAC;IACzC,0BAA0B,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CAC5C,CAAC;AAsDF,wBAAsB,iBAAiB,CACrC,GAAG,EAAE,UAAU,EACf,KAAK,GAAE,aAAkB,GACxB,OAAO,CAAC,gBAAgB,CAAC,CAwH3B"}
|
|
@@ -0,0 +1,156 @@
|
|
|
1
|
+
import { fetchLatestMedrxivVersionNumber } from "../../../sources/medrxiv/client";
|
|
2
|
+
const PREPRINT_HOSTS = [
|
|
3
|
+
"arxiv.org",
|
|
4
|
+
"www.arxiv.org",
|
|
5
|
+
"biorxiv.org",
|
|
6
|
+
"www.biorxiv.org",
|
|
7
|
+
"medrxiv.org",
|
|
8
|
+
"www.medrxiv.org",
|
|
9
|
+
"osf.io",
|
|
10
|
+
"www.osf.io",
|
|
11
|
+
"preprints.org",
|
|
12
|
+
"www.preprints.org",
|
|
13
|
+
];
|
|
14
|
+
const PREPRINT_DOI_PREFIXES = [
|
|
15
|
+
/^10\.1101\//i, // bioRxiv / medRxiv
|
|
16
|
+
/^10\.48550\/(arxiv(\.|:)?)/i, // DataCite-minted arXiv DOI
|
|
17
|
+
/^10\.31219\/osf\.io\//i, // OSF preprints (SocArXiv, etc.)
|
|
18
|
+
/^10\.31234\/osf\.io\//i, // OSF PsyArXiv
|
|
19
|
+
/^10\.7287\/peerj\.preprints/i, // PeerJ Preprints (legacy)
|
|
20
|
+
];
|
|
21
|
+
function isPreprintDoi(doi) {
|
|
22
|
+
if (!doi)
|
|
23
|
+
return false;
|
|
24
|
+
return PREPRINT_DOI_PREFIXES.some((rx) => rx.test(doi));
|
|
25
|
+
}
|
|
26
|
+
function hostFromUrl(u) {
|
|
27
|
+
try {
|
|
28
|
+
if (!u)
|
|
29
|
+
return null;
|
|
30
|
+
return new URL(u).host.toLowerCase();
|
|
31
|
+
}
|
|
32
|
+
catch {
|
|
33
|
+
return null;
|
|
34
|
+
}
|
|
35
|
+
}
|
|
36
|
+
function parseExplicitPreprintV(ids) {
|
|
37
|
+
const rxV = /v(\d+)$/i;
|
|
38
|
+
for (const maybe of [ids.biorxivId, ids.medrxivId]) {
|
|
39
|
+
if (maybe) {
|
|
40
|
+
const m = maybe.match(rxV);
|
|
41
|
+
if (m)
|
|
42
|
+
return Number(m[1]);
|
|
43
|
+
}
|
|
44
|
+
}
|
|
45
|
+
return null;
|
|
46
|
+
}
|
|
47
|
+
function looksLikePreprintVenue(name) {
|
|
48
|
+
if (!name)
|
|
49
|
+
return false;
|
|
50
|
+
const n = name.toLowerCase();
|
|
51
|
+
return /(arxiv|biorxiv|medrxiv|preprint|osf)/i.test(n);
|
|
52
|
+
}
|
|
53
|
+
export async function inferVersionLabel(ids, hints = {}) {
|
|
54
|
+
// Augment hints with latest medRxiv version iff:
|
|
55
|
+
// - a medrxivId is present, and
|
|
56
|
+
// - no explicit vN is embedded in ids.{biorxivId|medrxivId}
|
|
57
|
+
const augmented = { ...hints };
|
|
58
|
+
const explicitV0 = parseExplicitPreprintV(ids);
|
|
59
|
+
if (ids.medrxivId && !explicitV0) {
|
|
60
|
+
try {
|
|
61
|
+
const idOrDoi = ids.doi && /^10\.1101\//i.test(ids.doi) ? ids.doi : ids.medrxivId;
|
|
62
|
+
const latest = await fetchLatestMedrxivVersionNumber(idOrDoi);
|
|
63
|
+
if (latest && latest > 0) {
|
|
64
|
+
augmented.latestKnownPreprintVersion = latest;
|
|
65
|
+
}
|
|
66
|
+
}
|
|
67
|
+
catch {
|
|
68
|
+
// ignore network errors; fall back to pure logic below
|
|
69
|
+
}
|
|
70
|
+
}
|
|
71
|
+
const evidence = [];
|
|
72
|
+
let preprintV = null;
|
|
73
|
+
let preprintSignal = false;
|
|
74
|
+
let publishedSignal = false;
|
|
75
|
+
// --- 1) Explicit preprint versions
|
|
76
|
+
preprintV = parseExplicitPreprintV(ids);
|
|
77
|
+
if (preprintV) {
|
|
78
|
+
preprintSignal = true;
|
|
79
|
+
evidence.push(`explicit preprint version v${preprintV} found in repository ID/DOI`);
|
|
80
|
+
}
|
|
81
|
+
// --- 2) DOI classification
|
|
82
|
+
if (ids.doi) {
|
|
83
|
+
if (isPreprintDoi(ids.doi)) {
|
|
84
|
+
preprintSignal = true;
|
|
85
|
+
evidence.push(`DOI classified as preprint: ${ids.doi}`);
|
|
86
|
+
if (!preprintV) {
|
|
87
|
+
const m = ids.doi.match(/v(\d+)$/i);
|
|
88
|
+
if (m) {
|
|
89
|
+
preprintV = Number(m[1]);
|
|
90
|
+
evidence.push(`extracted v${preprintV} from preprint DOI`);
|
|
91
|
+
}
|
|
92
|
+
}
|
|
93
|
+
}
|
|
94
|
+
else {
|
|
95
|
+
publishedSignal = true;
|
|
96
|
+
evidence.push(`non-preprint DOI present (journal candidate): ${ids.doi}`);
|
|
97
|
+
}
|
|
98
|
+
}
|
|
99
|
+
// --- 3) Landing URL host heuristics
|
|
100
|
+
const host = hostFromUrl(ids.landingUrl);
|
|
101
|
+
if (host && PREPRINT_HOSTS.includes(host)) {
|
|
102
|
+
preprintSignal = true;
|
|
103
|
+
evidence.push(`landing URL host is a known preprint host: ${host}`);
|
|
104
|
+
}
|
|
105
|
+
// --- 4) Venue/name/type hints
|
|
106
|
+
if (augmented.crossrefType === "posted-content" &&
|
|
107
|
+
augmented.crossrefSubtype === "preprint") {
|
|
108
|
+
preprintSignal = true;
|
|
109
|
+
evidence.push(`Crossref type=subtype indicates preprint (posted-content/preprint)`);
|
|
110
|
+
}
|
|
111
|
+
if (augmented.openalexIsPreprintVenue) {
|
|
112
|
+
preprintSignal = true;
|
|
113
|
+
evidence.push(`OpenAlex marks venue as preprint`);
|
|
114
|
+
}
|
|
115
|
+
else if (looksLikePreprintVenue(augmented.openalexVenueName ?? ids.openalexHostVenue)) {
|
|
116
|
+
preprintSignal = true;
|
|
117
|
+
evidence.push(`Venue name looks like preprint: ${augmented.openalexVenueName ?? ids.openalexHostVenue}`);
|
|
118
|
+
}
|
|
119
|
+
// --- 4b) use repository-provided latest version if no explicit v found
|
|
120
|
+
if (preprintSignal &&
|
|
121
|
+
!preprintV &&
|
|
122
|
+
augmented.latestKnownPreprintVersion &&
|
|
123
|
+
augmented.latestKnownPreprintVersion > 0) {
|
|
124
|
+
preprintV = augmented.latestKnownPreprintVersion;
|
|
125
|
+
evidence.push(`using repository latest version v${preprintV} from medRxiv API`);
|
|
126
|
+
}
|
|
127
|
+
// --- 5) Final precedence
|
|
128
|
+
let label;
|
|
129
|
+
let stage;
|
|
130
|
+
if (publishedSignal) {
|
|
131
|
+
label = "published";
|
|
132
|
+
stage = "journal";
|
|
133
|
+
evidence.push("published signal takes precedence over preprint");
|
|
134
|
+
}
|
|
135
|
+
else if (preprintSignal) {
|
|
136
|
+
const v = preprintV ?? 1;
|
|
137
|
+
label = `v${v}`;
|
|
138
|
+
stage = "preprint";
|
|
139
|
+
}
|
|
140
|
+
else {
|
|
141
|
+
label = "v1";
|
|
142
|
+
stage = "unknown";
|
|
143
|
+
evidence.push("no strong signals; defaulting to v1");
|
|
144
|
+
}
|
|
145
|
+
// --- 6) Confidence
|
|
146
|
+
let confidence = 0;
|
|
147
|
+
if (publishedSignal)
|
|
148
|
+
confidence += 0.4;
|
|
149
|
+
if (preprintSignal)
|
|
150
|
+
confidence += 0.4;
|
|
151
|
+
if (preprintV)
|
|
152
|
+
confidence += 0.2;
|
|
153
|
+
confidence = Math.min(1, Math.max(0.4, confidence));
|
|
154
|
+
return { label, stage, confidence, evidence };
|
|
155
|
+
}
|
|
156
|
+
//# sourceMappingURL=versioning.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"versioning.js","sourceRoot":"","sources":["../../../../src/works/id/resolveIds/versioning.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,+BAA+B,EAAE,MAAM,iCAAiC,CAAC;AA+BlF,MAAM,cAAc,GAAG;IACrB,WAAW;IACX,eAAe;IACf,aAAa;IACb,iBAAiB;IACjB,aAAa;IACb,iBAAiB;IACjB,QAAQ;IACR,YAAY;IACZ,eAAe;IACf,mBAAmB;CACpB,CAAC;AAEF,MAAM,qBAAqB,GAAG;IAC5B,cAAc,EAAE,oBAAoB;IACpC,6BAA6B,EAAE,4BAA4B;IAC3D,wBAAwB,EAAE,iCAAiC;IAC3D,wBAAwB,EAAE,eAAe;IACzC,8BAA8B,EAAE,2BAA2B;CAC5D,CAAC;AAEF,SAAS,aAAa,CAAC,GAAmB;IACxC,IAAI,CAAC,GAAG;QAAE,OAAO,KAAK,CAAC;IACvB,OAAO,qBAAqB,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;AAC1D,CAAC;AAED,SAAS,WAAW,CAAC,CAAiB;IACpC,IAAI,CAAC;QACH,IAAI,CAAC,CAAC;YAAE,OAAO,IAAI,CAAC;QACpB,OAAO,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC;IACvC,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,IAAI,CAAC;IACd,CAAC;AACH,CAAC;AAED,SAAS,sBAAsB,CAAC,GAAe;IAC7C,MAAM,GAAG,GAAG,UAAU,CAAC;IACvB,KAAK,MAAM,KAAK,IAAI,CAAC,GAAG,CAAC,SAAS,EAAE,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC;QACnD,IAAI,KAAK,EAAE,CAAC;YACV,MAAM,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YAC3B,IAAI,CAAC;gBAAE,OAAO,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC7B,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED,SAAS,sBAAsB,CAAC,IAAoB;IAClD,IAAI,CAAC,IAAI;QAAE,OAAO,KAAK,CAAC;IACxB,MAAM,CAAC,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;IAC7B,OAAO,uCAAuC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;AACzD,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,iBAAiB,CACrC,GAAe,EACf,QAAuB,EAAE;IAEzB,iDAAiD;IACjD,iCAAiC;IACjC,6DAA6D;IAC7D,MAAM,SAAS,GAAkB,EAAE,GAAG,KAAK,EAAE,CAAC;IAC9C,MAAM,UAAU,GAAG,sBAAsB,CAAC,GAAG,CAAC,CAAC;IAC/C,IAAI,GAAG,CAAC,SAAS,IAAI,CAAC,UAAU,EAAE,CAAC;QACjC,IAAI,CAAC;YACH,MAAM,OAAO,GACX,GAAG,CAAC,GAAG,IAAI,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC;YACpE,MAAM,MAAM,GAAG,MAAM,+BAA+B,CAAC,OAAO,CAAC,CAAC;YAC9D,IAAI,MAAM,IAAI,MAAM,GAAG,CAAC,EAAE,CAAC;gBACzB,SAAS,CAAC,0BAA0B,GAAG,MAAM,CAAC;YAChD,CAAC;QACH,CAAC;QAAC,MAAM,CAAC;YACP,uDAAuD;QACzD,CAAC;IACH,CAAC;IAED,MAAM,QAAQ,GAAa,EAAE,CAAC;IAC9B,IAAI,SAAS,GAAkB,IAAI,CAAC;IACpC,IAAI,cAAc,GAAG,KAAK,CAAC;IAC3B,IAAI,eAAe,GAAG,KAAK,CAAC;IAE5B,oCAAoC;IACpC,SAAS,GAAG,sBAAsB,CAAC,GAAG,CAAC,CAAC;IACxC,IAAI,SAAS,EAAE,CAAC;QACd,cAAc,GAAG,IAAI,CAAC;QACtB,QAAQ,CAAC,IAAI,CACX,8BAA8B,SAAS,6BAA6B,CACrE,CAAC;IACJ,CAAC;IAED,4BAA4B;IAC5B,IAAI,GAAG,CAAC,GAAG,EAAE,CAAC;QACZ,IAAI,aAAa,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;YAC3B,cAAc,GAAG,IAAI,CAAC;YACtB,QAAQ,CAAC,IAAI,CAAC,+BAA+B,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC;YACxD,IAAI,CAAC,SAAS,EAAE,CAAC;gBACf,MAAM,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;gBACpC,IAAI,CAAC,EAAE,CAAC;oBACN,SAAS,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;oBACzB,QAAQ,CAAC,IAAI,CAAC,cAAc,SAAS,oBAAoB,CAAC,CAAC;gBAC7D,CAAC;YACH,CAAC;QACH,CAAC;aAAM,CAAC;YACN,eAAe,GAAG,IAAI,CAAC;YACvB,QAAQ,CAAC,IAAI,CAAC,iDAAiD,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC;QAC5E,CAAC;IACH,CAAC;IAED,qCAAqC;IACrC,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC;IACzC,IAAI,IAAI,IAAI,cAAc,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;QAC1C,cAAc,GAAG,IAAI,CAAC;QACtB,QAAQ,CAAC,IAAI,CAAC,8CAA8C,IAAI,EAAE,CAAC,CAAC;IACtE,CAAC;IAED,+BAA+B;IAC/B,IACE,SAAS,CAAC,YAAY,KAAK,gBAAgB;QAC3C,SAAS,CAAC,eAAe,KAAK,UAAU,EACxC,CAAC;QACD,cAAc,GAAG,IAAI,CAAC;QACtB,QAAQ,CAAC,IAAI,CACX,oEAAoE,CACrE,CAAC;IACJ,CAAC;IACD,IAAI,SAAS,CAAC,uBAAuB,EAAE,CAAC;QACtC,cAAc,GAAG,IAAI,CAAC;QACtB,QAAQ,CAAC,IAAI,CAAC,kCAAkC,CAAC,CAAC;IACpD,CAAC;SAAM,IACL,sBAAsB,CAAC,SAAS,CAAC,iBAAiB,IAAI,GAAG,CAAC,iBAAiB,CAAC,EAC5E,CAAC;QACD,cAAc,GAAG,IAAI,CAAC;QACtB,QAAQ,CAAC,IAAI,CACX,mCACE,SAAS,CAAC,iBAAiB,IAAI,GAAG,CAAC,iBACrC,EAAE,CACH,CAAC;IACJ,CAAC;IAED,wEAAwE;IACxE,IACE,cAAc;QACd,CAAC,SAAS;QACV,SAAS,CAAC,0BAA0B;QACpC,SAAS,CAAC,0BAA0B,GAAG,CAAC,EACxC,CAAC;QACD,SAAS,GAAG,SAAS,CAAC,0BAA0B,CAAC;QACjD,QAAQ,CAAC,IAAI,CACX,oCAAoC,SAAS,mBAAmB,CACjE,CAAC;IACJ,CAAC;IAED,0BAA0B;IAC1B,IAAI,KAAmB,CAAC;IACxB,IAAI,KAAmB,CAAC;IACxB,IAAI,eAAe,EAAE,CAAC;QACpB,KAAK,GAAG,WAAW,CAAC;QACpB,KAAK,GAAG,SAAS,CAAC;QAClB,QAAQ,CAAC,IAAI,CAAC,iDAAiD,CAAC,CAAC;IACnE,CAAC;SAAM,IAAI,cAAc,EAAE,CAAC;QAC1B,MAAM,CAAC,GAAG,SAAS,IAAI,CAAC,CAAC;QACzB,KAAK,GAAG,IAAI,CAAC,EAAE,CAAC;QAChB,KAAK,GAAG,UAAU,CAAC;IACrB,CAAC;SAAM,CAAC;QACN,KAAK,GAAG,IAAI,CAAC;QACb,KAAK,GAAG,SAAS,CAAC;QAClB,QAAQ,CAAC,IAAI,CAAC,qCAAqC,CAAC,CAAC;IACvD,CAAC;IAED,oBAAoB;IACpB,IAAI,UAAU,GAAG,CAAC,CAAC;IACnB,IAAI,eAAe;QAAE,UAAU,IAAI,GAAG,CAAC;IACvC,IAAI,cAAc;QAAE,UAAU,IAAI,GAAG,CAAC;IACtC,IAAI,SAAS;QAAE,UAAU,IAAI,GAAG,CAAC;IACjC,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,UAAU,CAAC,CAAC,CAAC;IAEpD,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,QAAQ,EAAE,CAAC;AAChD,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"workWhere.d.ts","sourceRoot":"","sources":["../../../../src/works/id/resolveIds/workWhere.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAQ/C,wBAAgB,0BAA0B,CAAC,GAAG,EAAE,WAAW,4BAoB1D"}
|
|
@@ -0,0 +1,35 @@
|
|
|
1
|
+
const stripVersionSuffix = (value) => {
|
|
2
|
+
if (typeof value !== "string")
|
|
3
|
+
return null;
|
|
4
|
+
const t = value.trim();
|
|
5
|
+
return t ? t.replace(/v\d+$/i, "") : null;
|
|
6
|
+
};
|
|
7
|
+
export function buildWorkWhereOrConditions(ids) {
|
|
8
|
+
const orConds = [];
|
|
9
|
+
if (ids.doi)
|
|
10
|
+
orConds.push({ doi: ids.doi });
|
|
11
|
+
if (ids.pmid)
|
|
12
|
+
orConds.push({ pmid: ids.pmid });
|
|
13
|
+
if (ids.pmcid)
|
|
14
|
+
orConds.push({ pmcid: ids.pmcid });
|
|
15
|
+
if (ids.arxivId)
|
|
16
|
+
orConds.push({ arxivId: ids.arxivId });
|
|
17
|
+
if (ids.biorxivId) {
|
|
18
|
+
const base = stripVersionSuffix(ids.biorxivId);
|
|
19
|
+
if (base)
|
|
20
|
+
orConds.push({ biorxivId: base });
|
|
21
|
+
if (base && base !== ids.biorxivId)
|
|
22
|
+
orConds.push({ biorxivId: ids.biorxivId });
|
|
23
|
+
}
|
|
24
|
+
if (ids.medrxivId) {
|
|
25
|
+
const base = stripVersionSuffix(ids.medrxivId);
|
|
26
|
+
if (base)
|
|
27
|
+
orConds.push({ medrxivId: base });
|
|
28
|
+
if (base && base !== ids.medrxivId)
|
|
29
|
+
orConds.push({ medrxivId: ids.medrxivId });
|
|
30
|
+
}
|
|
31
|
+
if (ids.openalexId)
|
|
32
|
+
orConds.push({ openalexId: ids.openalexId });
|
|
33
|
+
return orConds;
|
|
34
|
+
}
|
|
35
|
+
//# sourceMappingURL=workWhere.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"workWhere.js","sourceRoot":"","sources":["../../../../src/works/id/resolveIds/workWhere.ts"],"names":[],"mappings":"AAEA,MAAM,kBAAkB,GAAG,CAAC,KAAqB,EAAiB,EAAE;IAClE,IAAI,OAAO,KAAK,KAAK,QAAQ;QAAE,OAAO,IAAI,CAAC;IAC3C,MAAM,CAAC,GAAG,KAAK,CAAC,IAAI,EAAE,CAAC;IACvB,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;AAC5C,CAAC,CAAC;AAEF,MAAM,UAAU,0BAA0B,CAAC,GAAgB;IACzD,MAAM,OAAO,GAAkC,EAAE,CAAC;IAClD,IAAI,GAAG,CAAC,GAAG;QAAE,OAAO,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC;IAC5C,IAAI,GAAG,CAAC,IAAI;QAAE,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC;IAC/C,IAAI,GAAG,CAAC,KAAK;QAAE,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,GAAG,CAAC,KAAK,EAAE,CAAC,CAAC;IAClD,IAAI,GAAG,CAAC,OAAO;QAAE,OAAO,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;IACxD,IAAI,GAAG,CAAC,SAAS,EAAE,CAAC;QAClB,MAAM,IAAI,GAAG,kBAAkB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QAC/C,IAAI,IAAI;YAAE,OAAO,CAAC,IAAI,CAAC,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;QAC5C,IAAI,IAAI,IAAI,IAAI,KAAK,GAAG,CAAC,SAAS;YAChC,OAAO,CAAC,IAAI,CAAC,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,CAAC,CAAC;IAC/C,CAAC;IACD,IAAI,GAAG,CAAC,SAAS,EAAE,CAAC;QAClB,MAAM,IAAI,GAAG,kBAAkB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QAC/C,IAAI,IAAI;YAAE,OAAO,CAAC,IAAI,CAAC,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;QAC5C,IAAI,IAAI,IAAI,IAAI,KAAK,GAAG,CAAC,SAAS;YAChC,OAAO,CAAC,IAAI,CAAC,EAAE,SAAS,EAAE,GAAG,CAAC,SAAS,EAAE,CAAC,CAAC;IAC/C,CAAC;IACD,IAAI,GAAG,CAAC,UAAU;QAAE,OAAO,CAAC,IAAI,CAAC,EAAE,UAAU,EAAE,GAAG,CAAC,UAAU,EAAE,CAAC,CAAC;IACjE,OAAO,OAAO,CAAC;AACjB,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../../src/works/id/types.ts"],"names":[],"mappings":"AAAA,MAAM,MAAM,cAAc,GAAG,KAAK,GAAG,MAAM,GAAG,OAAO,GAAG,OAAO,GAAG,SAAS,GAAG,SAAS,CAAC;AACxF,MAAM,MAAM,kBAAkB,GAAG;IAAE,IAAI,EAAE,cAAc,CAAC;IAAC,KAAK,EAAE,MAAM,CAAA;CAAE,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"types.js","sourceRoot":"","sources":["../../../src/works/id/types.ts"],"names":[],"mappings":""}
|
|
@@ -0,0 +1,12 @@
|
|
|
1
|
+
import type { CompleteIds } from "../types";
|
|
2
|
+
export interface PdfCandidate {
|
|
3
|
+
url: string;
|
|
4
|
+
sourceKind: "preprint" | "oa-publisher" | "resolved-metadata";
|
|
5
|
+
sourceName: string;
|
|
6
|
+
}
|
|
7
|
+
/**
|
|
8
|
+
* Build an ordered list of candidate PDF URLs from resolved identifiers.
|
|
9
|
+
* Order: preprint servers (known URL patterns) → bestPdfUrl from OpenAlex → DOI-derived.
|
|
10
|
+
*/
|
|
11
|
+
export declare function buildPdfCandidates(ids: CompleteIds): PdfCandidate[];
|
|
12
|
+
//# sourceMappingURL=candidates.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"candidates.d.ts","sourceRoot":"","sources":["../../../src/works/pdf-fallback/candidates.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,UAAU,CAAC;AAE5C,MAAM,WAAW,YAAY;IAC3B,GAAG,EAAE,MAAM,CAAC;IACZ,UAAU,EAAE,UAAU,GAAG,cAAc,GAAG,mBAAmB,CAAC;IAC9D,UAAU,EAAE,MAAM,CAAC;CACpB;AAED;;;GAGG;AACH,wBAAgB,kBAAkB,CAAC,GAAG,EAAE,WAAW,GAAG,YAAY,EAAE,CAkDnE"}
|
|
@@ -0,0 +1,51 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Build an ordered list of candidate PDF URLs from resolved identifiers.
|
|
3
|
+
* Order: preprint servers (known URL patterns) → bestPdfUrl from OpenAlex → DOI-derived.
|
|
4
|
+
*/
|
|
5
|
+
export function buildPdfCandidates(ids) {
|
|
6
|
+
const candidates = [];
|
|
7
|
+
// biorxiv: /content/{doi}v{N}.full.pdf
|
|
8
|
+
if (ids.biorxivId) {
|
|
9
|
+
const bare = ids.biorxivId.replace(/v\d+$/i, "");
|
|
10
|
+
const doi = bare.startsWith("10.1101/") ? bare : `10.1101/${bare}`;
|
|
11
|
+
const version = ids.version ?? "1";
|
|
12
|
+
candidates.push({
|
|
13
|
+
url: `https://www.biorxiv.org/content/${doi}v${version}.full.pdf`,
|
|
14
|
+
sourceKind: "preprint",
|
|
15
|
+
sourceName: "biorxiv",
|
|
16
|
+
});
|
|
17
|
+
}
|
|
18
|
+
// medrxiv: same pattern as biorxiv
|
|
19
|
+
if (ids.medrxivId) {
|
|
20
|
+
const bare = ids.medrxivId.replace(/v\d+$/i, "");
|
|
21
|
+
const doi = bare.startsWith("10.1101/") ? bare : `10.1101/${bare}`;
|
|
22
|
+
const version = ids.version ?? "1";
|
|
23
|
+
candidates.push({
|
|
24
|
+
url: `https://www.medrxiv.org/content/${doi}v${version}.full.pdf`,
|
|
25
|
+
sourceKind: "preprint",
|
|
26
|
+
sourceName: "medrxiv",
|
|
27
|
+
});
|
|
28
|
+
}
|
|
29
|
+
// arxiv: /pdf/{id}
|
|
30
|
+
if (ids.arxivId) {
|
|
31
|
+
const bare = ids.arxivId.replace(/^arxiv:/i, "");
|
|
32
|
+
candidates.push({
|
|
33
|
+
url: `https://arxiv.org/pdf/${bare}`,
|
|
34
|
+
sourceKind: "preprint",
|
|
35
|
+
sourceName: "arxiv",
|
|
36
|
+
});
|
|
37
|
+
}
|
|
38
|
+
// bestPdfUrl from OpenAlex / resolveIds
|
|
39
|
+
if (ids.bestPdfUrl) {
|
|
40
|
+
const alreadyAdded = candidates.some((c) => c.url === ids.bestPdfUrl);
|
|
41
|
+
if (!alreadyAdded) {
|
|
42
|
+
candidates.push({
|
|
43
|
+
url: ids.bestPdfUrl,
|
|
44
|
+
sourceKind: "oa-publisher",
|
|
45
|
+
sourceName: "openalex-best-pdf",
|
|
46
|
+
});
|
|
47
|
+
}
|
|
48
|
+
}
|
|
49
|
+
return candidates;
|
|
50
|
+
}
|
|
51
|
+
//# sourceMappingURL=candidates.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"candidates.js","sourceRoot":"","sources":["../../../src/works/pdf-fallback/candidates.ts"],"names":[],"mappings":"AAQA;;;GAGG;AACH,MAAM,UAAU,kBAAkB,CAAC,GAAgB;IACjD,MAAM,UAAU,GAAmB,EAAE,CAAC;IAEtC,uCAAuC;IACvC,IAAI,GAAG,CAAC,SAAS,EAAE,CAAC;QAClB,MAAM,IAAI,GAAG,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QACjD,MAAM,GAAG,GAAG,IAAI,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,WAAW,IAAI,EAAE,CAAC;QACnE,MAAM,OAAO,GAAG,GAAG,CAAC,OAAO,IAAI,GAAG,CAAC;QACnC,UAAU,CAAC,IAAI,CAAC;YACd,GAAG,EAAE,mCAAmC,GAAG,IAAI,OAAO,WAAW;YACjE,UAAU,EAAE,UAAU;YACtB,UAAU,EAAE,SAAS;SACtB,CAAC,CAAC;IACL,CAAC;IAED,mCAAmC;IACnC,IAAI,GAAG,CAAC,SAAS,EAAE,CAAC;QAClB,MAAM,IAAI,GAAG,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QACjD,MAAM,GAAG,GAAG,IAAI,CAAC,UAAU,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,WAAW,IAAI,EAAE,CAAC;QACnE,MAAM,OAAO,GAAG,GAAG,CAAC,OAAO,IAAI,GAAG,CAAC;QACnC,UAAU,CAAC,IAAI,CAAC;YACd,GAAG,EAAE,mCAAmC,GAAG,IAAI,OAAO,WAAW;YACjE,UAAU,EAAE,UAAU;YACtB,UAAU,EAAE,SAAS;SACtB,CAAC,CAAC;IACL,CAAC;IAED,mBAAmB;IACnB,IAAI,GAAG,CAAC,OAAO,EAAE,CAAC;QAChB,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;QACjD,UAAU,CAAC,IAAI,CAAC;YACd,GAAG,EAAE,yBAAyB,IAAI,EAAE;YACpC,UAAU,EAAE,UAAU;YACtB,UAAU,EAAE,OAAO;SACpB,CAAC,CAAC;IACL,CAAC;IAED,wCAAwC;IACxC,IAAI,GAAG,CAAC,UAAU,EAAE,CAAC;QACnB,MAAM,YAAY,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,GAAG,KAAK,GAAG,CAAC,UAAU,CAAC,CAAC;QACtE,IAAI,CAAC,YAAY,EAAE,CAAC;YAClB,UAAU,CAAC,IAAI,CAAC;gBACd,GAAG,EAAE,GAAG,CAAC,UAAU;gBACnB,UAAU,EAAE,cAAc;gBAC1B,UAAU,EAAE,mBAAmB;aAChC,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,OAAO,UAAU,CAAC;AACpB,CAAC"}
|
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
import type { PdfCandidate } from "./candidates";
|
|
2
|
+
import type { LoggerLike } from "../types";
|
|
3
|
+
export interface PdfFetchSuccess {
|
|
4
|
+
ok: true;
|
|
5
|
+
buffer: Buffer;
|
|
6
|
+
candidate: PdfCandidate;
|
|
7
|
+
finalUrl: string;
|
|
8
|
+
contentType: string;
|
|
9
|
+
}
|
|
10
|
+
export interface PdfFetchFailure {
|
|
11
|
+
ok: false;
|
|
12
|
+
candidate: PdfCandidate;
|
|
13
|
+
reason: string;
|
|
14
|
+
}
|
|
15
|
+
export type PdfFetchResult = PdfFetchSuccess | PdfFetchFailure;
|
|
16
|
+
/**
|
|
17
|
+
* Attempt to fetch a PDF from a candidate URL.
|
|
18
|
+
* Validates: HTTP status, content-type, magic bytes, size.
|
|
19
|
+
*/
|
|
20
|
+
export declare function fetchPdfCandidate(candidate: PdfCandidate, logger?: LoggerLike): Promise<PdfFetchResult>;
|
|
21
|
+
//# sourceMappingURL=fetch.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"fetch.d.ts","sourceRoot":"","sources":["../../../src/works/pdf-fallback/fetch.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,cAAc,CAAC;AACjD,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,UAAU,CAAC;AAM3C,MAAM,WAAW,eAAe;IAC9B,EAAE,EAAE,IAAI,CAAC;IACT,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,EAAE,YAAY,CAAC;IACxB,QAAQ,EAAE,MAAM,CAAC;IACjB,WAAW,EAAE,MAAM,CAAC;CACrB;AAED,MAAM,WAAW,eAAe;IAC9B,EAAE,EAAE,KAAK,CAAC;IACV,SAAS,EAAE,YAAY,CAAC;IACxB,MAAM,EAAE,MAAM,CAAC;CAChB;AAED,MAAM,MAAM,cAAc,GAAG,eAAe,GAAG,eAAe,CAAC;AAE/D;;;GAGG;AACH,wBAAsB,iBAAiB,CACrC,SAAS,EAAE,YAAY,EACvB,MAAM,CAAC,EAAE,UAAU,GAClB,OAAO,CAAC,cAAc,CAAC,CAgGzB"}
|
|
@@ -0,0 +1,89 @@
|
|
|
1
|
+
const PDF_MAGIC = Buffer.from("%PDF");
|
|
2
|
+
const MAX_PDF_SIZE = 50 * 1024 * 1024; // 50 MB
|
|
3
|
+
const FETCH_TIMEOUT_MS = 30_000;
|
|
4
|
+
/**
|
|
5
|
+
* Attempt to fetch a PDF from a candidate URL.
|
|
6
|
+
* Validates: HTTP status, content-type, magic bytes, size.
|
|
7
|
+
*/
|
|
8
|
+
export async function fetchPdfCandidate(candidate, logger) {
|
|
9
|
+
try {
|
|
10
|
+
const controller = new AbortController();
|
|
11
|
+
const timeout = setTimeout(() => controller.abort(), FETCH_TIMEOUT_MS);
|
|
12
|
+
const response = await fetch(candidate.url, {
|
|
13
|
+
method: "GET",
|
|
14
|
+
signal: controller.signal,
|
|
15
|
+
redirect: "follow",
|
|
16
|
+
headers: {
|
|
17
|
+
Accept: "application/pdf",
|
|
18
|
+
"User-Agent": "DryLab-RevEng-API/1.0 (PDF-fallback)",
|
|
19
|
+
},
|
|
20
|
+
});
|
|
21
|
+
clearTimeout(timeout);
|
|
22
|
+
if (!response.ok) {
|
|
23
|
+
return {
|
|
24
|
+
ok: false,
|
|
25
|
+
candidate,
|
|
26
|
+
reason: `HTTP ${response.status} ${response.statusText}`,
|
|
27
|
+
};
|
|
28
|
+
}
|
|
29
|
+
const contentType = response.headers.get("content-type") ?? "";
|
|
30
|
+
const isPdfContentType = contentType.includes("application/pdf") ||
|
|
31
|
+
contentType.includes("application/x-pdf");
|
|
32
|
+
const arrayBuffer = await response.arrayBuffer();
|
|
33
|
+
const buffer = Buffer.from(arrayBuffer);
|
|
34
|
+
if (buffer.length > MAX_PDF_SIZE) {
|
|
35
|
+
return {
|
|
36
|
+
ok: false,
|
|
37
|
+
candidate,
|
|
38
|
+
reason: `PDF too large: ${buffer.length} bytes (max ${MAX_PDF_SIZE})`,
|
|
39
|
+
};
|
|
40
|
+
}
|
|
41
|
+
if (buffer.length < 4) {
|
|
42
|
+
return {
|
|
43
|
+
ok: false,
|
|
44
|
+
candidate,
|
|
45
|
+
reason: `Response too small: ${buffer.length} bytes`,
|
|
46
|
+
};
|
|
47
|
+
}
|
|
48
|
+
// Validate magic bytes
|
|
49
|
+
const hasMagic = buffer.subarray(0, 4).equals(PDF_MAGIC);
|
|
50
|
+
if (!hasMagic) {
|
|
51
|
+
if (!isPdfContentType) {
|
|
52
|
+
return {
|
|
53
|
+
ok: false,
|
|
54
|
+
candidate,
|
|
55
|
+
reason: `Not a PDF: content-type=${contentType}, no PDF magic bytes`,
|
|
56
|
+
};
|
|
57
|
+
}
|
|
58
|
+
// Has PDF content-type but no magic bytes — might be HTML login page
|
|
59
|
+
const preview = buffer.subarray(0, 200).toString("utf8");
|
|
60
|
+
if (preview.includes("<html") || preview.includes("<!DOCTYPE")) {
|
|
61
|
+
return {
|
|
62
|
+
ok: false,
|
|
63
|
+
candidate,
|
|
64
|
+
reason: "Response is HTML, not PDF (likely login/paywall page)",
|
|
65
|
+
};
|
|
66
|
+
}
|
|
67
|
+
}
|
|
68
|
+
const finalUrl = response.url || candidate.url;
|
|
69
|
+
logger?.info?.({
|
|
70
|
+
source: candidate.sourceName,
|
|
71
|
+
url: finalUrl,
|
|
72
|
+
size: buffer.length,
|
|
73
|
+
}, "pdf.fallback.fetch.success");
|
|
74
|
+
return {
|
|
75
|
+
ok: true,
|
|
76
|
+
buffer,
|
|
77
|
+
candidate,
|
|
78
|
+
finalUrl,
|
|
79
|
+
contentType,
|
|
80
|
+
};
|
|
81
|
+
}
|
|
82
|
+
catch (err) {
|
|
83
|
+
const reason = err?.name === "AbortError"
|
|
84
|
+
? `Timeout after ${FETCH_TIMEOUT_MS}ms`
|
|
85
|
+
: err?.message || String(err);
|
|
86
|
+
return { ok: false, candidate, reason };
|
|
87
|
+
}
|
|
88
|
+
}
|
|
89
|
+
//# sourceMappingURL=fetch.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"fetch.js","sourceRoot":"","sources":["../../../src/works/pdf-fallback/fetch.ts"],"names":[],"mappings":"AAGA,MAAM,SAAS,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AACtC,MAAM,YAAY,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,QAAQ;AAC/C,MAAM,gBAAgB,GAAG,MAAM,CAAC;AAkBhC;;;GAGG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CACrC,SAAuB,EACvB,MAAmB;IAEnB,IAAI,CAAC;QACH,MAAM,UAAU,GAAG,IAAI,eAAe,EAAE,CAAC;QACzC,MAAM,OAAO,GAAG,UAAU,CAAC,GAAG,EAAE,CAAC,UAAU,CAAC,KAAK,EAAE,EAAE,gBAAgB,CAAC,CAAC;QAEvE,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,SAAS,CAAC,GAAG,EAAE;YAC1C,MAAM,EAAE,KAAK;YACb,MAAM,EAAE,UAAU,CAAC,MAAM;YACzB,QAAQ,EAAE,QAAQ;YAClB,OAAO,EAAE;gBACP,MAAM,EAAE,iBAAiB;gBACzB,YAAY,EAAE,sCAAsC;aACrD;SACF,CAAC,CAAC;QAEH,YAAY,CAAC,OAAO,CAAC,CAAC;QAEtB,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;YACjB,OAAO;gBACL,EAAE,EAAE,KAAK;gBACT,SAAS;gBACT,MAAM,EAAE,QAAQ,QAAQ,CAAC,MAAM,IAAI,QAAQ,CAAC,UAAU,EAAE;aACzD,CAAC;QACJ,CAAC;QAED,MAAM,WAAW,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,IAAI,EAAE,CAAC;QAC/D,MAAM,gBAAgB,GACpB,WAAW,CAAC,QAAQ,CAAC,iBAAiB,CAAC;YACvC,WAAW,CAAC,QAAQ,CAAC,mBAAmB,CAAC,CAAC;QAE5C,MAAM,WAAW,GAAG,MAAM,QAAQ,CAAC,WAAW,EAAE,CAAC;QACjD,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;QAExC,IAAI,MAAM,CAAC,MAAM,GAAG,YAAY,EAAE,CAAC;YACjC,OAAO;gBACL,EAAE,EAAE,KAAK;gBACT,SAAS;gBACT,MAAM,EAAE,kBAAkB,MAAM,CAAC,MAAM,eAAe,YAAY,GAAG;aACtE,CAAC;QACJ,CAAC;QAED,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACtB,OAAO;gBACL,EAAE,EAAE,KAAK;gBACT,SAAS;gBACT,MAAM,EAAE,uBAAuB,MAAM,CAAC,MAAM,QAAQ;aACrD,CAAC;QACJ,CAAC;QAED,uBAAuB;QACvB,MAAM,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC;QACzD,IAAI,CAAC,QAAQ,EAAE,CAAC;YACd,IAAI,CAAC,gBAAgB,EAAE,CAAC;gBACtB,OAAO;oBACL,EAAE,EAAE,KAAK;oBACT,SAAS;oBACT,MAAM,EAAE,2BAA2B,WAAW,sBAAsB;iBACrE,CAAC;YACJ,CAAC;YACD,qEAAqE;YACrE,MAAM,OAAO,GAAG,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YACzD,IAAI,OAAO,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;gBAC/D,OAAO;oBACL,EAAE,EAAE,KAAK;oBACT,SAAS;oBACT,MAAM,EAAE,uDAAuD;iBAChE,CAAC;YACJ,CAAC;QACH,CAAC;QAED,MAAM,QAAQ,GAAG,QAAQ,CAAC,GAAG,IAAI,SAAS,CAAC,GAAG,CAAC;QAE/C,MAAM,EAAE,IAAI,EAAE,CACZ;YACE,MAAM,EAAE,SAAS,CAAC,UAAU;YAC5B,GAAG,EAAE,QAAQ;YACb,IAAI,EAAE,MAAM,CAAC,MAAM;SACpB,EACD,4BAA4B,CAC7B,CAAC;QAEF,OAAO;YACL,EAAE,EAAE,IAAI;YACR,MAAM;YACN,SAAS;YACT,QAAQ;YACR,WAAW;SACZ,CAAC;IACJ,CAAC;IAAC,OAAO,GAAQ,EAAE,CAAC;QAClB,MAAM,MAAM,GACV,GAAG,EAAE,IAAI,KAAK,YAAY;YACxB,CAAC,CAAC,iBAAiB,gBAAgB,IAAI;YACvC,CAAC,CAAC,GAAG,EAAE,OAAO,IAAI,MAAM,CAAC,GAAG,CAAC,CAAC;QAElC,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,SAAS,EAAE,MAAM,EAAE,CAAC;IAC1C,CAAC;AACH,CAAC"}
|
|
@@ -0,0 +1,28 @@
|
|
|
1
|
+
import type { CompleteIds, LoggerLike } from "../types";
|
|
2
|
+
import { type PdfCandidate } from "./candidates";
|
|
3
|
+
export type { PdfCandidate } from "./candidates";
|
|
4
|
+
export type { PdfFetchSuccess, PdfFetchFailure, PdfFetchResult } from "./fetch";
|
|
5
|
+
export interface PdfFallbackAttempt {
|
|
6
|
+
candidate: PdfCandidate;
|
|
7
|
+
success: boolean;
|
|
8
|
+
reason?: string;
|
|
9
|
+
}
|
|
10
|
+
export interface PdfFallbackSuccess {
|
|
11
|
+
ok: true;
|
|
12
|
+
buffer: Buffer;
|
|
13
|
+
candidate: PdfCandidate;
|
|
14
|
+
finalUrl: string;
|
|
15
|
+
attempts: PdfFallbackAttempt[];
|
|
16
|
+
}
|
|
17
|
+
export interface PdfFallbackExhausted {
|
|
18
|
+
ok: false;
|
|
19
|
+
attempts: PdfFallbackAttempt[];
|
|
20
|
+
}
|
|
21
|
+
export type PdfFallbackResult = PdfFallbackSuccess | PdfFallbackExhausted;
|
|
22
|
+
/**
|
|
23
|
+
* Try to fetch a PDF for the given identifiers by working through
|
|
24
|
+
* an ordered list of candidate URLs. Returns the first successful
|
|
25
|
+
* fetch or a structured failure with all attempts.
|
|
26
|
+
*/
|
|
27
|
+
export declare function resolvePdfFallback(ids: CompleteIds, logger?: LoggerLike): Promise<PdfFallbackResult>;
|
|
28
|
+
//# sourceMappingURL=index.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/works/pdf-fallback/index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,UAAU,CAAC;AACxD,OAAO,EAAsB,KAAK,YAAY,EAAE,MAAM,cAAc,CAAC;AAGrE,YAAY,EAAE,YAAY,EAAE,MAAM,cAAc,CAAC;AACjD,YAAY,EAAE,eAAe,EAAE,eAAe,EAAE,cAAc,EAAE,MAAM,SAAS,CAAC;AAEhF,MAAM,WAAW,kBAAkB;IACjC,SAAS,EAAE,YAAY,CAAC;IACxB,OAAO,EAAE,OAAO,CAAC;IACjB,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,WAAW,kBAAkB;IACjC,EAAE,EAAE,IAAI,CAAC;IACT,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,EAAE,YAAY,CAAC;IACxB,QAAQ,EAAE,MAAM,CAAC;IACjB,QAAQ,EAAE,kBAAkB,EAAE,CAAC;CAChC;AAED,MAAM,WAAW,oBAAoB;IACnC,EAAE,EAAE,KAAK,CAAC;IACV,QAAQ,EAAE,kBAAkB,EAAE,CAAC;CAChC;AAED,MAAM,MAAM,iBAAiB,GAAG,kBAAkB,GAAG,oBAAoB,CAAC;AAE1E;;;;GAIG;AACH,wBAAsB,kBAAkB,CACtC,GAAG,EAAE,WAAW,EAChB,MAAM,CAAC,EAAE,UAAU,GAClB,OAAO,CAAC,iBAAiB,CAAC,CAkC5B"}
|
|
@@ -0,0 +1,35 @@
|
|
|
1
|
+
import { buildPdfCandidates } from "./candidates";
|
|
2
|
+
import { fetchPdfCandidate } from "./fetch";
|
|
3
|
+
/**
|
|
4
|
+
* Try to fetch a PDF for the given identifiers by working through
|
|
5
|
+
* an ordered list of candidate URLs. Returns the first successful
|
|
6
|
+
* fetch or a structured failure with all attempts.
|
|
7
|
+
*/
|
|
8
|
+
export async function resolvePdfFallback(ids, logger) {
|
|
9
|
+
const candidates = buildPdfCandidates(ids);
|
|
10
|
+
const attempts = [];
|
|
11
|
+
if (candidates.length === 0) {
|
|
12
|
+
return { ok: false, attempts };
|
|
13
|
+
}
|
|
14
|
+
for (const candidate of candidates) {
|
|
15
|
+
const result = await fetchPdfCandidate(candidate, logger);
|
|
16
|
+
if (result.ok) {
|
|
17
|
+
attempts.push({ candidate, success: true });
|
|
18
|
+
return {
|
|
19
|
+
ok: true,
|
|
20
|
+
buffer: result.buffer,
|
|
21
|
+
candidate: result.candidate,
|
|
22
|
+
finalUrl: result.finalUrl,
|
|
23
|
+
attempts,
|
|
24
|
+
};
|
|
25
|
+
}
|
|
26
|
+
attempts.push({ candidate, success: false, reason: result.reason });
|
|
27
|
+
logger?.warn?.({
|
|
28
|
+
source: candidate.sourceName,
|
|
29
|
+
url: candidate.url,
|
|
30
|
+
reason: result.reason,
|
|
31
|
+
}, "pdf.fallback.candidate.failed");
|
|
32
|
+
}
|
|
33
|
+
return { ok: false, attempts };
|
|
34
|
+
}
|
|
35
|
+
//# sourceMappingURL=index.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../src/works/pdf-fallback/index.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,kBAAkB,EAAqB,MAAM,cAAc,CAAC;AACrE,OAAO,EAAE,iBAAiB,EAAuB,MAAM,SAAS,CAAC;AA0BjE;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACtC,GAAgB,EAChB,MAAmB;IAEnB,MAAM,UAAU,GAAG,kBAAkB,CAAC,GAAG,CAAC,CAAC;IAC3C,MAAM,QAAQ,GAAyB,EAAE,CAAC;IAE1C,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC5B,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC;IACjC,CAAC;IAED,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACnC,MAAM,MAAM,GAAG,MAAM,iBAAiB,CAAC,SAAS,EAAE,MAAM,CAAC,CAAC;QAE1D,IAAI,MAAM,CAAC,EAAE,EAAE,CAAC;YACd,QAAQ,CAAC,IAAI,CAAC,EAAE,SAAS,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC;YAC5C,OAAO;gBACL,EAAE,EAAE,IAAI;gBACR,MAAM,EAAE,MAAM,CAAC,MAAM;gBACrB,SAAS,EAAE,MAAM,CAAC,SAAS;gBAC3B,QAAQ,EAAE,MAAM,CAAC,QAAQ;gBACzB,QAAQ;aACT,CAAC;QACJ,CAAC;QAED,QAAQ,CAAC,IAAI,CAAC,EAAE,SAAS,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC;QACpE,MAAM,EAAE,IAAI,EAAE,CACZ;YACE,MAAM,EAAE,SAAS,CAAC,UAAU;YAC5B,GAAG,EAAE,SAAS,CAAC,GAAG;YAClB,MAAM,EAAE,MAAM,CAAC,MAAM;SACtB,EACD,+BAA+B,CAChC,CAAC;IACJ,CAAC;IAED,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,QAAQ,EAAE,CAAC;AACjC,CAAC"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
import type { Step } from "./types";
|
|
2
|
+
import type { Coordinator } from "./types";
|
|
3
|
+
export type PlanOptions = {
|
|
4
|
+
/** When true, use PDF normalization instead of JATS fetch+normalize. */
|
|
5
|
+
hasPdf?: boolean;
|
|
6
|
+
};
|
|
7
|
+
export declare function planSteps(decision: Coordinator, chosenVersionLabel: string, cid?: string, opts?: PlanOptions): Step[];
|
|
8
|
+
//# sourceMappingURL=plan.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"plan.d.ts","sourceRoot":"","sources":["../../src/works/plan.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,SAAS,CAAC;AACpC,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,SAAS,CAAC;AAE3C,MAAM,MAAM,WAAW,GAAG;IACxB,wEAAwE;IACxE,MAAM,CAAC,EAAE,OAAO,CAAC;CAClB,CAAC;AAEF,wBAAgB,SAAS,CACvB,QAAQ,EAAE,WAAW,EACrB,kBAAkB,EAAE,MAAM,EAC1B,GAAG,CAAC,EAAE,MAAM,EACZ,IAAI,CAAC,EAAE,WAAW,GACjB,IAAI,EAAE,CAmER"}
|
|
@@ -0,0 +1,62 @@
|
|
|
1
|
+
export function planSteps(decision, chosenVersionLabel, cid, opts) {
|
|
2
|
+
// Conflict decisions must be handled before planning — return no steps
|
|
3
|
+
if (decision.conflict) {
|
|
4
|
+
return [];
|
|
5
|
+
}
|
|
6
|
+
if (decision.workExists &&
|
|
7
|
+
decision.versionExists &&
|
|
8
|
+
decision.matched.work_id &&
|
|
9
|
+
!opts?.hasPdf) {
|
|
10
|
+
return [
|
|
11
|
+
{
|
|
12
|
+
kind: "ReturnExisting",
|
|
13
|
+
work_id: decision.matched.work_id,
|
|
14
|
+
versionId: decision.matched.versionId ?? null,
|
|
15
|
+
},
|
|
16
|
+
];
|
|
17
|
+
}
|
|
18
|
+
const steps = [];
|
|
19
|
+
if (opts?.hasPdf) {
|
|
20
|
+
// PDF upload path: skip FetchFullText, use NormalizePdf instead
|
|
21
|
+
if (!decision.actions.ensureWork && decision.workExists && decision.matched.work_id) {
|
|
22
|
+
steps.push({ kind: "LoadMatchedWork", work_id: decision.matched.work_id });
|
|
23
|
+
}
|
|
24
|
+
steps.push({ kind: "NormalizePdf", versionLabel: chosenVersionLabel });
|
|
25
|
+
if (decision.actions.ensureWork) {
|
|
26
|
+
steps.push({ kind: "EnsureWork" });
|
|
27
|
+
}
|
|
28
|
+
// PersistRawBundle will gracefully skip when rawXml is null (PDF has no XML)
|
|
29
|
+
steps.push({ kind: "PersistRawBundle", versionLabel: chosenVersionLabel });
|
|
30
|
+
steps.push({ kind: "EnsureVersion", versionLabel: chosenVersionLabel });
|
|
31
|
+
steps.push({ kind: "FinalizeFullText", versionLabel: chosenVersionLabel });
|
|
32
|
+
}
|
|
33
|
+
else if (decision.actions.fetchFullText) {
|
|
34
|
+
// Standard JATS path
|
|
35
|
+
if (!decision.actions.ensureWork && decision.workExists && decision.matched.work_id) {
|
|
36
|
+
steps.push({ kind: "LoadMatchedWork", work_id: decision.matched.work_id });
|
|
37
|
+
}
|
|
38
|
+
steps.push({ kind: "FetchFullText" });
|
|
39
|
+
steps.push({ kind: "NormalizeFullText", versionLabel: chosenVersionLabel });
|
|
40
|
+
if (decision.actions.ensureWork) {
|
|
41
|
+
steps.push({ kind: "EnsureWork" });
|
|
42
|
+
}
|
|
43
|
+
steps.push({ kind: "PersistRawBundle", versionLabel: chosenVersionLabel });
|
|
44
|
+
steps.push({ kind: "EnsureVersion", versionLabel: chosenVersionLabel });
|
|
45
|
+
steps.push({ kind: "FinalizeFullText", versionLabel: chosenVersionLabel });
|
|
46
|
+
}
|
|
47
|
+
else if (decision.actions.ensureWork) {
|
|
48
|
+
steps.push({ kind: "EnsureWork" });
|
|
49
|
+
}
|
|
50
|
+
else if (decision.workExists && decision.matched.work_id) {
|
|
51
|
+
// Existing work matched but no fetch/ensureWork needed (metadata-only re-ingest)
|
|
52
|
+
return [
|
|
53
|
+
{
|
|
54
|
+
kind: "ReturnExisting",
|
|
55
|
+
work_id: decision.matched.work_id,
|
|
56
|
+
versionId: decision.matched.versionId ?? null,
|
|
57
|
+
},
|
|
58
|
+
];
|
|
59
|
+
}
|
|
60
|
+
return steps;
|
|
61
|
+
}
|
|
62
|
+
//# sourceMappingURL=plan.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"plan.js","sourceRoot":"","sources":["../../src/works/plan.ts"],"names":[],"mappings":"AAQA,MAAM,UAAU,SAAS,CACvB,QAAqB,EACrB,kBAA0B,EAC1B,GAAY,EACZ,IAAkB;IAElB,uEAAuE;IACvE,IAAI,QAAQ,CAAC,QAAQ,EAAE,CAAC;QACtB,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,IACE,QAAQ,CAAC,UAAU;QACnB,QAAQ,CAAC,aAAa;QACtB,QAAQ,CAAC,OAAO,CAAC,OAAO;QACxB,CAAC,IAAI,EAAE,MAAM,EACb,CAAC;QACD,OAAO;YACL;gBACE,IAAI,EAAE,gBAAgB;gBACtB,OAAO,EAAE,QAAQ,CAAC,OAAO,CAAC,OAAO;gBACjC,SAAS,EAAE,QAAQ,CAAC,OAAO,CAAC,SAAS,IAAI,IAAI;aAC9C;SACF,CAAC;IACJ,CAAC;IAED,MAAM,KAAK,GAAW,EAAE,CAAC;IAEzB,IAAI,IAAI,EAAE,MAAM,EAAE,CAAC;QACjB,gEAAgE;QAChE,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,UAAU,IAAI,QAAQ,CAAC,UAAU,IAAI,QAAQ,CAAC,OAAO,CAAC,OAAO,EAAE,CAAC;YACpF,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,iBAAiB,EAAE,OAAO,EAAE,QAAQ,CAAC,OAAO,CAAC,OAAO,EAAE,CAAC,CAAC;QAC7E,CAAC;QAED,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,cAAc,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC,CAAC;QAEvE,IAAI,QAAQ,CAAC,OAAO,CAAC,UAAU,EAAE,CAAC;YAChC,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,YAAY,EAAE,CAAC,CAAC;QACrC,CAAC;QACD,6EAA6E;QAC7E,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,kBAAkB,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC,CAAC;QAC3E,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,eAAe,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC,CAAC;QACxE,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,kBAAkB,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC,CAAC;IAC7E,CAAC;SAAM,IAAI,QAAQ,CAAC,OAAO,CAAC,aAAa,EAAE,CAAC;QAC1C,qBAAqB;QACrB,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,UAAU,IAAI,QAAQ,CAAC,UAAU,IAAI,QAAQ,CAAC,OAAO,CAAC,OAAO,EAAE,CAAC;YACpF,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,iBAAiB,EAAE,OAAO,EAAE,QAAQ,CAAC,OAAO,CAAC,OAAO,EAAE,CAAC,CAAC;QAC7E,CAAC;QAED,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,eAAe,EAAE,CAAC,CAAC;QACtC,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,mBAAmB,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC,CAAC;QAE5E,IAAI,QAAQ,CAAC,OAAO,CAAC,UAAU,EAAE,CAAC;YAChC,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,YAAY,EAAE,CAAC,CAAC;QACrC,CAAC;QACD,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,kBAAkB,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC,CAAC;QAC3E,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,eAAe,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC,CAAC;QACxE,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,kBAAkB,EAAE,YAAY,EAAE,kBAAkB,EAAE,CAAC,CAAC;IAC7E,CAAC;SAAM,IAAI,QAAQ,CAAC,OAAO,CAAC,UAAU,EAAE,CAAC;QACvC,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,YAAY,EAAE,CAAC,CAAC;IACrC,CAAC;SAAM,IAAI,QAAQ,CAAC,UAAU,IAAI,QAAQ,CAAC,OAAO,CAAC,OAAO,EAAE,CAAC;QAC3D,iFAAiF;QACjF,OAAO;YACL;gBACE,IAAI,EAAE,gBAAgB;gBACtB,OAAO,EAAE,QAAQ,CAAC,OAAO,CAAC,OAAO;gBACjC,SAAS,EAAE,QAAQ,CAAC,OAAO,CAAC,SAAS,IAAI,IAAI;aAC9C;SACF,CAAC;IACJ,CAAC;IAED,OAAO,KAAK,CAAC;AACf,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"arxiv.d.ts","sourceRoot":"","sources":["../../../src/works/strategies/arxiv.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AA+BjD,eAAO,MAAM,aAAa,EAAE,gBAqD3B,CAAC"}
|