@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 +1,56 @@
|
|
|
1
|
+
import { buildAr5ivHtmlUrl, buildArxivAbsUrl, } from "../../sources/arxiv/client";
|
|
2
|
+
import { isLikelyErrorHtml, isArxivAbsHtml, } from "./shared";
|
|
3
|
+
const UA = process.env.UA || "drylab-reveng-api/1.0";
|
|
4
|
+
async function fetchArxivHtmlIfAvailable(id, UA) {
|
|
5
|
+
const headers = {
|
|
6
|
+
Accept: "text/html,application/xhtml+xml",
|
|
7
|
+
"User-Agent": UA || "drylab-reveng-api/1.0",
|
|
8
|
+
};
|
|
9
|
+
const absRes = await fetch(`https://arxiv.org/abs/${id}`, { headers });
|
|
10
|
+
if (!absRes.ok)
|
|
11
|
+
return null;
|
|
12
|
+
const absHtml = await absRes.text();
|
|
13
|
+
const escaped = id.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
|
|
14
|
+
const m = absHtml.match(new RegExp(`(?:https://arxiv\\.org)?/html/${escaped}v\\d+`, "i"));
|
|
15
|
+
if (!m)
|
|
16
|
+
return null;
|
|
17
|
+
const htmlUrl = m[0].startsWith("/") ? `https://arxiv.org${m[0]}` : m[0];
|
|
18
|
+
const htmlRes = await fetch(htmlUrl, { headers });
|
|
19
|
+
if (!htmlRes.ok)
|
|
20
|
+
return null;
|
|
21
|
+
const html = await htmlRes.text();
|
|
22
|
+
return html && html.trim() ? { url: htmlUrl, html } : null;
|
|
23
|
+
}
|
|
24
|
+
export const arxivStrategy = {
|
|
25
|
+
name: "arxiv",
|
|
26
|
+
matches: (ids) => Boolean(ids.arxivId),
|
|
27
|
+
getFullText: async (ids, ctx) => {
|
|
28
|
+
const arxivId = ids.arxivId;
|
|
29
|
+
const ar5ivUrl = buildAr5ivHtmlUrl(arxivId);
|
|
30
|
+
const { text: ar5ivHtml, finalUrl } = await ctx.fetchText(ar5ivUrl, "text/html,application/xhtml+xml", 3);
|
|
31
|
+
if (ar5ivHtml &&
|
|
32
|
+
!isLikelyErrorHtml(ar5ivHtml) &&
|
|
33
|
+
!isArxivAbsHtml(ar5ivHtml) &&
|
|
34
|
+
!(finalUrl && /^https?:\/\/arxiv\.org\//i.test(finalUrl))) {
|
|
35
|
+
return { type: "html", source: "ar5iv", content: ar5ivHtml };
|
|
36
|
+
}
|
|
37
|
+
const exp = await fetchArxivHtmlIfAvailable(arxivId, UA);
|
|
38
|
+
if (exp)
|
|
39
|
+
return {
|
|
40
|
+
type: "html",
|
|
41
|
+
source: "arxiv-html",
|
|
42
|
+
content: exp.html,
|
|
43
|
+
url: exp.url,
|
|
44
|
+
};
|
|
45
|
+
const jinaUrl = `https://r.jina.ai/http://ar5iv.org/html/${arxivId.replace(/^arxiv:/i, "")}`;
|
|
46
|
+
const { text: jina } = await ctx.fetchText(jinaUrl, "text/plain, text/html;q=0.9", 2);
|
|
47
|
+
if (jina && jina.trim().length > 0)
|
|
48
|
+
return { type: "html", source: "ar5iv-proxy", content: jina };
|
|
49
|
+
const absUrl = buildArxivAbsUrl(arxivId);
|
|
50
|
+
const { text: absHtml } = await ctx.fetchText(absUrl, "text/html,application/xhtml+xml", 2);
|
|
51
|
+
if (absHtml && absHtml.trim().length > 0)
|
|
52
|
+
return { type: "html", source: "arxiv-abs", content: absHtml };
|
|
53
|
+
return { type: "html", source: "ar5iv", url: ar5ivUrl };
|
|
54
|
+
},
|
|
55
|
+
};
|
|
56
|
+
//# sourceMappingURL=arxiv.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"arxiv.js","sourceRoot":"","sources":["../../../src/works/strategies/arxiv.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,iBAAiB,EACjB,gBAAgB,GACjB,MAAM,4BAA4B,CAAC;AAEpC,OAAO,EACL,iBAAiB,EACjB,cAAc,GACf,MAAM,UAAU,CAAC;AAElB,MAAM,EAAE,GAAG,OAAO,CAAC,GAAG,CAAC,EAAE,IAAI,uBAAuB,CAAC;AAErD,KAAK,UAAU,yBAAyB,CAAC,EAAU,EAAE,EAAW;IAC9D,MAAM,OAAO,GAAG;QACd,MAAM,EAAE,iCAAiC;QACzC,YAAY,EAAE,EAAE,IAAI,uBAAuB;KAClB,CAAC;IAC5B,MAAM,MAAM,GAAG,MAAM,KAAK,CAAC,yBAAyB,EAAE,EAAE,EAAE,EAAE,OAAO,EAAS,CAAC,CAAC;IAC9E,IAAI,CAAC,MAAM,CAAC,EAAE;QAAE,OAAO,IAAI,CAAC;IAC5B,MAAM,OAAO,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;IAEpC,MAAM,OAAO,GAAG,EAAE,CAAC,OAAO,CAAC,qBAAqB,EAAE,MAAM,CAAC,CAAC;IAC1D,MAAM,CAAC,GAAG,OAAO,CAAC,KAAK,CACrB,IAAI,MAAM,CAAC,iCAAiC,OAAO,OAAO,EAAE,GAAG,CAAC,CACjE,CAAC;IACF,IAAI,CAAC,CAAC;QAAE,OAAO,IAAI,CAAC;IAEpB,MAAM,OAAO,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,oBAAoB,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACzE,MAAM,OAAO,GAAG,MAAM,KAAK,CAAC,OAAO,EAAE,EAAE,OAAO,EAAS,CAAC,CAAC;IACzD,IAAI,CAAC,OAAO,CAAC,EAAE;QAAE,OAAO,IAAI,CAAC;IAE7B,MAAM,IAAI,GAAG,MAAM,OAAO,CAAC,IAAI,EAAE,CAAC;IAClC,OAAO,IAAI,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;AAC7D,CAAC;AAED,MAAM,CAAC,MAAM,aAAa,GAAqB;IAC7C,IAAI,EAAE,OAAO;IACb,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC;IACtC,WAAW,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,EAAE,EAAE;QAC9B,MAAM,OAAO,GAAG,GAAG,CAAC,OAAQ,CAAC;QAC7B,MAAM,QAAQ,GAAG,iBAAiB,CAAC,OAAO,CAAC,CAAC;QAE5C,MAAM,EAAE,IAAI,EAAE,SAAS,EAAE,QAAQ,EAAE,GAAG,MAAM,GAAG,CAAC,SAAS,CACvD,QAAQ,EACR,iCAAiC,EACjC,CAAC,CACF,CAAC;QACF,IACE,SAAS;YACT,CAAC,iBAAiB,CAAC,SAAS,CAAC;YAC7B,CAAC,cAAc,CAAC,SAAS,CAAC;YAC1B,CAAC,CAAC,QAAQ,IAAI,2BAA2B,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,EACzD,CAAC;YACD,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,OAAO,EAAE,SAAS,EAAE,CAAC;QAC/D,CAAC;QAED,MAAM,GAAG,GAAG,MAAM,yBAAyB,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;QACzD,IAAI,GAAG;YACL,OAAO;gBACL,IAAI,EAAE,MAAM;gBACZ,MAAM,EAAE,YAAY;gBACpB,OAAO,EAAE,GAAG,CAAC,IAAI;gBACjB,GAAG,EAAE,GAAG,CAAC,GAAG;aACb,CAAC;QAEJ,MAAM,OAAO,GAAG,2CAA2C,OAAO,CAAC,OAAO,CACxE,UAAU,EACV,EAAE,CACH,EAAE,CAAC;QACJ,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,MAAM,GAAG,CAAC,SAAS,CACxC,OAAO,EACP,6BAA6B,EAC7B,CAAC,CACF,CAAC;QACF,IAAI,IAAI,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;YAChC,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC;QAEhE,MAAM,MAAM,GAAG,gBAAgB,CAAC,OAAO,CAAC,CAAC;QACzC,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,GAAG,MAAM,GAAG,CAAC,SAAS,CAC3C,MAAM,EACN,iCAAiC,EACjC,CAAC,CACF,CAAC;QACF,IAAI,OAAO,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;YACtC,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,WAAW,EAAE,OAAO,EAAE,OAAO,EAAE,CAAC;QAEjE,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,GAAG,EAAE,QAAQ,EAAE,CAAC;IAC1D,CAAC;CACF,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"biorxiv.d.ts","sourceRoot":"","sources":["../../../src/works/strategies/biorxiv.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAWjD,eAAO,MAAM,eAAe,EAAE,gBA6E7B,CAAC"}
|
|
@@ -0,0 +1,63 @@
|
|
|
1
|
+
import { fetchBiorxivSourceXmlViaApi, buildBiorxivLatestSourceXmlUrlViaApi, fetchBiorxivFullHtmlViaProxy, } from "../../sources/biorxiv/client";
|
|
2
|
+
import { parseVersionSuffix, extractVersionSuffix, firstFiniteVersion, labelFromVersion, buildBiorxivVersionedId, } from "./shared";
|
|
3
|
+
const UA = process.env.UA || "drylab-reveng-api/1.0";
|
|
4
|
+
export const biorxivStrategy = {
|
|
5
|
+
name: "biorxiv-id",
|
|
6
|
+
matches: (ids) => Boolean(ids.biorxivId),
|
|
7
|
+
getFullText: async (ids) => {
|
|
8
|
+
const biorxivId = ids.biorxivId;
|
|
9
|
+
const idForApi = biorxivId.startsWith("10.1101/")
|
|
10
|
+
? biorxivId
|
|
11
|
+
: `10.1101/${biorxivId.replace(/v\d+$/i, "")}`;
|
|
12
|
+
const preferVersion = parseVersionSuffix(biorxivId);
|
|
13
|
+
const fetchOpts = { UA, preferVersion };
|
|
14
|
+
const hit = await fetchBiorxivSourceXmlViaApi(idForApi, fetchOpts);
|
|
15
|
+
if (hit) {
|
|
16
|
+
const resolvedVersion = firstFiniteVersion(hit.version, extractVersionSuffix(hit.url), preferVersion, parseVersionSuffix(biorxivId));
|
|
17
|
+
const versionLabel = labelFromVersion(resolvedVersion);
|
|
18
|
+
const versionedIdForHtml = buildBiorxivVersionedId(idForApi, resolvedVersion) ??
|
|
19
|
+
(resolvedVersion
|
|
20
|
+
? buildBiorxivVersionedId(biorxivId, resolvedVersion)
|
|
21
|
+
: null);
|
|
22
|
+
const hasBody = /<(?:body|article-body|sec)\b/i.test(hit.xml);
|
|
23
|
+
if (!hasBody) {
|
|
24
|
+
const htmlHit = versionedIdForHtml
|
|
25
|
+
? await fetchBiorxivFullHtmlViaProxy(versionedIdForHtml, UA)
|
|
26
|
+
: null;
|
|
27
|
+
if (htmlHit) {
|
|
28
|
+
return {
|
|
29
|
+
type: "html",
|
|
30
|
+
source: "biorxiv",
|
|
31
|
+
content: htmlHit.html,
|
|
32
|
+
url: htmlHit.url,
|
|
33
|
+
versionLabel,
|
|
34
|
+
};
|
|
35
|
+
}
|
|
36
|
+
}
|
|
37
|
+
return {
|
|
38
|
+
type: "jats-xml",
|
|
39
|
+
source: "biorxiv",
|
|
40
|
+
content: hit.xml,
|
|
41
|
+
url: hit.url,
|
|
42
|
+
versionLabel,
|
|
43
|
+
};
|
|
44
|
+
}
|
|
45
|
+
const latestRecord = await buildBiorxivLatestSourceXmlUrlViaApi(idForApi, fetchOpts);
|
|
46
|
+
if (latestRecord?.url) {
|
|
47
|
+
const resolvedVersion = firstFiniteVersion(latestRecord.version, extractVersionSuffix(latestRecord.url), preferVersion, parseVersionSuffix(biorxivId));
|
|
48
|
+
return {
|
|
49
|
+
type: "jats-xml",
|
|
50
|
+
source: "biorxiv",
|
|
51
|
+
url: latestRecord.url,
|
|
52
|
+
versionLabel: labelFromVersion(resolvedVersion),
|
|
53
|
+
};
|
|
54
|
+
}
|
|
55
|
+
return {
|
|
56
|
+
type: "jats-xml",
|
|
57
|
+
source: "biorxiv",
|
|
58
|
+
url: `https://www.biorxiv.org/content/${idForApi}.source.xml`,
|
|
59
|
+
versionLabel: labelFromVersion(preferVersion),
|
|
60
|
+
};
|
|
61
|
+
},
|
|
62
|
+
};
|
|
63
|
+
//# sourceMappingURL=biorxiv.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"biorxiv.js","sourceRoot":"","sources":["../../../src/works/strategies/biorxiv.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,2BAA2B,EAC3B,oCAAoC,EACpC,4BAA4B,GAC7B,MAAM,8BAA8B,CAAC;AAEtC,OAAO,EACL,kBAAkB,EAClB,oBAAoB,EACpB,kBAAkB,EAClB,gBAAgB,EAChB,uBAAuB,GACxB,MAAM,UAAU,CAAC;AAElB,MAAM,EAAE,GAAG,OAAO,CAAC,GAAG,CAAC,EAAE,IAAI,uBAAuB,CAAC;AAErD,MAAM,CAAC,MAAM,eAAe,GAAqB;IAC/C,IAAI,EAAE,YAAY;IAClB,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC;IACxC,WAAW,EAAE,KAAK,EAAE,GAAG,EAAE,EAAE;QACzB,MAAM,SAAS,GAAG,GAAG,CAAC,SAAU,CAAC;QACjC,MAAM,QAAQ,GAAG,SAAS,CAAC,UAAU,CAAC,UAAU,CAAC;YAC/C,CAAC,CAAC,SAAS;YACX,CAAC,CAAC,WAAW,SAAS,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,EAAE,CAAC;QACjD,MAAM,aAAa,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAC;QACpD,MAAM,SAAS,GAAG,EAAE,EAAE,EAAE,aAAa,EAAW,CAAC;QAEjD,MAAM,GAAG,GAAG,MAAM,2BAA2B,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC;QACnE,IAAI,GAAG,EAAE,CAAC;YACR,MAAM,eAAe,GAAG,kBAAkB,CACxC,GAAG,CAAC,OAAO,EACX,oBAAoB,CAAC,GAAG,CAAC,GAAG,CAAC,EAC7B,aAAa,EACb,kBAAkB,CAAC,SAAS,CAAC,CAC9B,CAAC;YACF,MAAM,YAAY,GAAG,gBAAgB,CAAC,eAAe,CAAC,CAAC;YACvD,MAAM,kBAAkB,GACtB,uBAAuB,CAAC,QAAQ,EAAE,eAAe,CAAC;gBAClD,CAAC,eAAe;oBACd,CAAC,CAAC,uBAAuB,CAAC,SAAS,EAAE,eAAe,CAAC;oBACrD,CAAC,CAAC,IAAI,CAAC,CAAC;YAEZ,MAAM,OAAO,GAAG,+BAA+B,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YAC9D,IAAI,CAAC,OAAO,EAAE,CAAC;gBACb,MAAM,OAAO,GAAG,kBAAkB;oBAChC,CAAC,CAAC,MAAM,4BAA4B,CAAC,kBAAkB,EAAE,EAAE,CAAC;oBAC5D,CAAC,CAAC,IAAI,CAAC;gBACT,IAAI,OAAO,EAAE,CAAC;oBACZ,OAAO;wBACL,IAAI,EAAE,MAAM;wBACZ,MAAM,EAAE,SAAS;wBACjB,OAAO,EAAE,OAAO,CAAC,IAAI;wBACrB,GAAG,EAAE,OAAO,CAAC,GAAG;wBAChB,YAAY;qBACb,CAAC;gBACJ,CAAC;YACH,CAAC;YAED,OAAO;gBACL,IAAI,EAAE,UAAU;gBAChB,MAAM,EAAE,SAAS;gBACjB,OAAO,EAAE,GAAG,CAAC,GAAG;gBAChB,GAAG,EAAE,GAAG,CAAC,GAAG;gBACZ,YAAY;aACb,CAAC;QACJ,CAAC;QAED,MAAM,YAAY,GAAG,MAAM,oCAAoC,CAC7D,QAAQ,EACR,SAAS,CACV,CAAC;QACF,IAAI,YAAY,EAAE,GAAG,EAAE,CAAC;YACtB,MAAM,eAAe,GAAG,kBAAkB,CACxC,YAAY,CAAC,OAAO,EACpB,oBAAoB,CAAC,YAAY,CAAC,GAAG,CAAC,EACtC,aAAa,EACb,kBAAkB,CAAC,SAAS,CAAC,CAC9B,CAAC;YACF,OAAO;gBACL,IAAI,EAAE,UAAU;gBAChB,MAAM,EAAE,SAAS;gBACjB,GAAG,EAAE,YAAY,CAAC,GAAG;gBACrB,YAAY,EAAE,gBAAgB,CAAC,eAAe,CAAC;aAChD,CAAC;QACJ,CAAC;QAED,OAAO;YACL,IAAI,EAAE,UAAU;YAChB,MAAM,EAAE,SAAS;YACjB,GAAG,EAAE,mCAAmC,QAAQ,aAAa;YAC7D,YAAY,EAAE,gBAAgB,CAAC,aAAa,CAAC;SAC9C,CAAC;IACJ,CAAC;CACF,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"europepmc.d.ts","sourceRoot":"","sources":["../../../src/works/strategies/europepmc.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAGV,gBAAgB,EACjB,MAAM,UAAU,CAAC;AAElB,eAAO,MAAM,iBAAiB,EAAE,gBAU/B,CAAC"}
|
|
@@ -0,0 +1,15 @@
|
|
|
1
|
+
import { fetchEpmcJatsByPmcid } from "../../sources/europepmc/client";
|
|
2
|
+
export const europePmcStrategy = {
|
|
3
|
+
name: "europepmc-pmcid",
|
|
4
|
+
matches: (ids) => Boolean(ids.pmcid),
|
|
5
|
+
getFullText: async (ids) => {
|
|
6
|
+
try {
|
|
7
|
+
const xml = await fetchEpmcJatsByPmcid(ids.pmcid);
|
|
8
|
+
if (xml)
|
|
9
|
+
return { type: "jats-xml", source: "europepmc", content: xml };
|
|
10
|
+
}
|
|
11
|
+
catch { }
|
|
12
|
+
return null;
|
|
13
|
+
},
|
|
14
|
+
};
|
|
15
|
+
//# sourceMappingURL=europepmc.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"europepmc.js","sourceRoot":"","sources":["../../../src/works/strategies/europepmc.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,gCAAgC,CAAC;AAOtE,MAAM,CAAC,MAAM,iBAAiB,GAAqB;IACjD,IAAI,EAAE,iBAAiB;IACvB,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC;IACpC,WAAW,EAAE,KAAK,EAAE,GAAG,EAAE,EAAE;QACzB,IAAI,CAAC;YACH,MAAM,GAAG,GAAG,MAAM,oBAAoB,CAAC,GAAG,CAAC,KAAM,CAAC,CAAC;YACnD,IAAI,GAAG;gBAAE,OAAO,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,WAAW,EAAE,OAAO,EAAE,GAAG,EAAE,CAAC;QAC1E,CAAC;QAAC,MAAM,CAAC,CAAA,CAAC;QACV,OAAO,IAAI,CAAC;IACd,CAAC;CACF,CAAC"}
|
|
@@ -0,0 +1,12 @@
|
|
|
1
|
+
export { europePmcStrategy } from "./europepmc";
|
|
2
|
+
export { plosStrategy } from "./plos";
|
|
3
|
+
export { biorxivStrategy } from "./biorxiv";
|
|
4
|
+
export { medrxivStrategy } from "./medrxiv";
|
|
5
|
+
export { ten1101FallbackStrategy } from "./ten1101";
|
|
6
|
+
export { arxivStrategy } from "./arxiv";
|
|
7
|
+
export { landingUrlStrategy } from "./landing-url";
|
|
8
|
+
export { fetchTextWithRetries } from "./shared";
|
|
9
|
+
import type { FullTextStrategy } from "../types";
|
|
10
|
+
export declare const strategiesInOrder: FullTextStrategy[];
|
|
11
|
+
export type { FullTextStrategy, StrategyContext, FetchTextFn, } from "../types";
|
|
12
|
+
//# sourceMappingURL=index.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/works/strategies/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,MAAM,aAAa,CAAC;AAChD,OAAO,EAAE,YAAY,EAAE,MAAM,QAAQ,CAAC;AACtC,OAAO,EAAE,eAAe,EAAE,MAAM,WAAW,CAAC;AAC5C,OAAO,EAAE,eAAe,EAAE,MAAM,WAAW,CAAC;AAC5C,OAAO,EAAE,uBAAuB,EAAE,MAAM,WAAW,CAAC;AACpD,OAAO,EAAE,aAAa,EAAE,MAAM,SAAS,CAAC;AACxC,OAAO,EAAE,kBAAkB,EAAE,MAAM,eAAe,CAAC;AAEnD,OAAO,EAAE,oBAAoB,EAAE,MAAM,UAAU,CAAC;AAEhD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAWjD,eAAO,MAAM,iBAAiB,EAAE,gBAAgB,EAQ/C,CAAC;AAEF,YAAY,EACV,gBAAgB,EAChB,eAAe,EACf,WAAW,GACZ,MAAM,UAAU,CAAC"}
|
|
@@ -0,0 +1,19 @@
|
|
|
1
|
+
export { europePmcStrategy } from "./europepmc";
|
|
2
|
+
export { plosStrategy } from "./plos";
|
|
3
|
+
export { biorxivStrategy } from "./biorxiv";
|
|
4
|
+
export { medrxivStrategy } from "./medrxiv";
|
|
5
|
+
export { ten1101FallbackStrategy } from "./ten1101";
|
|
6
|
+
export { arxivStrategy } from "./arxiv";
|
|
7
|
+
export { landingUrlStrategy } from "./landing-url";
|
|
8
|
+
export { fetchTextWithRetries } from "./shared";
|
|
9
|
+
import { europePmcStrategy, plosStrategy, biorxivStrategy, medrxivStrategy, ten1101FallbackStrategy, arxivStrategy, landingUrlStrategy, } from "./index";
|
|
10
|
+
export const strategiesInOrder = [
|
|
11
|
+
europePmcStrategy,
|
|
12
|
+
plosStrategy,
|
|
13
|
+
biorxivStrategy,
|
|
14
|
+
medrxivStrategy,
|
|
15
|
+
ten1101FallbackStrategy,
|
|
16
|
+
arxivStrategy,
|
|
17
|
+
landingUrlStrategy,
|
|
18
|
+
];
|
|
19
|
+
//# sourceMappingURL=index.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../src/works/strategies/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,MAAM,aAAa,CAAC;AAChD,OAAO,EAAE,YAAY,EAAE,MAAM,QAAQ,CAAC;AACtC,OAAO,EAAE,eAAe,EAAE,MAAM,WAAW,CAAC;AAC5C,OAAO,EAAE,eAAe,EAAE,MAAM,WAAW,CAAC;AAC5C,OAAO,EAAE,uBAAuB,EAAE,MAAM,WAAW,CAAC;AACpD,OAAO,EAAE,aAAa,EAAE,MAAM,SAAS,CAAC;AACxC,OAAO,EAAE,kBAAkB,EAAE,MAAM,eAAe,CAAC;AAEnD,OAAO,EAAE,oBAAoB,EAAE,MAAM,UAAU,CAAC;AAGhD,OAAO,EACL,iBAAiB,EACjB,YAAY,EACZ,eAAe,EACf,eAAe,EACf,uBAAuB,EACvB,aAAa,EACb,kBAAkB,GACnB,MAAM,SAAS,CAAC;AAEjB,MAAM,CAAC,MAAM,iBAAiB,GAAuB;IACnD,iBAAiB;IACjB,YAAY;IACZ,eAAe;IACf,eAAe;IACf,uBAAuB;IACvB,aAAa;IACb,kBAAkB;CACnB,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"landing-url.d.ts","sourceRoot":"","sources":["../../../src/works/strategies/landing-url.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAEjD,eAAO,MAAM,kBAAkB,EAAE,gBAQhC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"landing-url.js","sourceRoot":"","sources":["../../../src/works/strategies/landing-url.ts"],"names":[],"mappings":"AAEA,MAAM,CAAC,MAAM,kBAAkB,GAAqB;IAClD,IAAI,EAAE,aAAa;IACnB,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,UAAU,CAAC;IACzC,WAAW,EAAE,KAAK,EAAE,GAAG,EAAE,EAAE,CAAC,CAAC;QAC3B,IAAI,EAAE,MAAM;QACZ,MAAM,EAAE,WAAW;QACnB,GAAG,EAAE,GAAG,CAAC,UAAW;KACrB,CAAC;CACH,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"medrxiv.d.ts","sourceRoot":"","sources":["../../../src/works/strategies/medrxiv.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAUjD,eAAO,MAAM,eAAe,EAAE,gBAuD7B,CAAC"}
|
|
@@ -0,0 +1,47 @@
|
|
|
1
|
+
import { fetchMedrxivSourceXmlViaApi, buildMedrxivLatestSourceXmlUrlViaApi, } from "../../sources/medrxiv/client";
|
|
2
|
+
import { parseVersionSuffix, extractVersionSuffix, firstFiniteVersion, labelFromVersion, } from "./shared";
|
|
3
|
+
const UA = process.env.UA || "drylab-reveng-api/1.0";
|
|
4
|
+
export const medrxivStrategy = {
|
|
5
|
+
name: "medrxiv-id",
|
|
6
|
+
matches: (ids) => Boolean(ids.medrxivId),
|
|
7
|
+
getFullText: async (ids) => {
|
|
8
|
+
const medrxivId = ids.medrxivId;
|
|
9
|
+
const core = medrxivId.replace(/^10\.1101\//i, "").replace(/v\d+$/i, "");
|
|
10
|
+
const doiForApi = `10.1101/${core}`;
|
|
11
|
+
const preferVersion = parseVersionSuffix(medrxivId);
|
|
12
|
+
const hit = await fetchMedrxivSourceXmlViaApi(doiForApi, {
|
|
13
|
+
UA,
|
|
14
|
+
preferVersion,
|
|
15
|
+
});
|
|
16
|
+
if (hit) {
|
|
17
|
+
const resolvedVersion = firstFiniteVersion(hit.version, extractVersionSuffix(hit.url), preferVersion, parseVersionSuffix(medrxivId));
|
|
18
|
+
return {
|
|
19
|
+
type: "jats-xml",
|
|
20
|
+
source: "medrxiv",
|
|
21
|
+
content: hit.xml,
|
|
22
|
+
url: hit.url,
|
|
23
|
+
versionLabel: labelFromVersion(resolvedVersion),
|
|
24
|
+
};
|
|
25
|
+
}
|
|
26
|
+
const latestRecord = await buildMedrxivLatestSourceXmlUrlViaApi(doiForApi, {
|
|
27
|
+
UA,
|
|
28
|
+
preferVersion,
|
|
29
|
+
});
|
|
30
|
+
if (latestRecord?.url) {
|
|
31
|
+
const resolvedVersion = firstFiniteVersion(latestRecord.version, extractVersionSuffix(latestRecord.url), preferVersion, parseVersionSuffix(medrxivId));
|
|
32
|
+
return {
|
|
33
|
+
type: "jats-xml",
|
|
34
|
+
source: "medrxiv",
|
|
35
|
+
url: latestRecord.url,
|
|
36
|
+
versionLabel: labelFromVersion(resolvedVersion),
|
|
37
|
+
};
|
|
38
|
+
}
|
|
39
|
+
return {
|
|
40
|
+
type: "jats-xml",
|
|
41
|
+
source: "medrxiv",
|
|
42
|
+
url: `https://www.medrxiv.org/content/10.1101/${core}.source.xml`,
|
|
43
|
+
versionLabel: labelFromVersion(preferVersion),
|
|
44
|
+
};
|
|
45
|
+
},
|
|
46
|
+
};
|
|
47
|
+
//# sourceMappingURL=medrxiv.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"medrxiv.js","sourceRoot":"","sources":["../../../src/works/strategies/medrxiv.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,2BAA2B,EAC3B,oCAAoC,GACrC,MAAM,8BAA8B,CAAC;AAEtC,OAAO,EACL,kBAAkB,EAClB,oBAAoB,EACpB,kBAAkB,EAClB,gBAAgB,GACjB,MAAM,UAAU,CAAC;AAElB,MAAM,EAAE,GAAG,OAAO,CAAC,GAAG,CAAC,EAAE,IAAI,uBAAuB,CAAC;AAErD,MAAM,CAAC,MAAM,eAAe,GAAqB;IAC/C,IAAI,EAAE,YAAY;IAClB,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,SAAS,CAAC;IACxC,WAAW,EAAE,KAAK,EAAE,GAAG,EAAE,EAAE;QACzB,MAAM,SAAS,GAAG,GAAG,CAAC,SAAU,CAAC;QACjC,MAAM,IAAI,GAAG,SAAS,CAAC,OAAO,CAAC,cAAc,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QACzE,MAAM,SAAS,GAAG,WAAW,IAAI,EAAE,CAAC;QACpC,MAAM,aAAa,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAC;QAEpD,MAAM,GAAG,GAAG,MAAM,2BAA2B,CAAC,SAAS,EAAE;YACvD,EAAE;YACF,aAAa;SACd,CAAC,CAAC;QACH,IAAI,GAAG,EAAE,CAAC;YACR,MAAM,eAAe,GAAG,kBAAkB,CACxC,GAAG,CAAC,OAAO,EACX,oBAAoB,CAAC,GAAG,CAAC,GAAG,CAAC,EAC7B,aAAa,EACb,kBAAkB,CAAC,SAAS,CAAC,CAC9B,CAAC;YACF,OAAO;gBACL,IAAI,EAAE,UAAU;gBAChB,MAAM,EAAE,SAAS;gBACjB,OAAO,EAAE,GAAG,CAAC,GAAG;gBAChB,GAAG,EAAE,GAAG,CAAC,GAAG;gBACZ,YAAY,EAAE,gBAAgB,CAAC,eAAe,CAAC;aAChD,CAAC;QACJ,CAAC;QAED,MAAM,YAAY,GAAG,MAAM,oCAAoC,CAAC,SAAS,EAAE;YACzE,EAAE;YACF,aAAa;SACd,CAAC,CAAC;QACH,IAAI,YAAY,EAAE,GAAG,EAAE,CAAC;YACtB,MAAM,eAAe,GAAG,kBAAkB,CACxC,YAAY,CAAC,OAAO,EACpB,oBAAoB,CAAC,YAAY,CAAC,GAAG,CAAC,EACtC,aAAa,EACb,kBAAkB,CAAC,SAAS,CAAC,CAC9B,CAAC;YACF,OAAO;gBACL,IAAI,EAAE,UAAU;gBAChB,MAAM,EAAE,SAAS;gBACjB,GAAG,EAAE,YAAY,CAAC,GAAG;gBACrB,YAAY,EAAE,gBAAgB,CAAC,eAAe,CAAC;aAChD,CAAC;QACJ,CAAC;QAED,OAAO;YACL,IAAI,EAAE,UAAU;YAChB,MAAM,EAAE,SAAS;YACjB,GAAG,EAAE,2CAA2C,IAAI,aAAa;YACjE,YAAY,EAAE,gBAAgB,CAAC,aAAa,CAAC;SAC9C,CAAC;IACJ,CAAC;CACF,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"plos.d.ts","sourceRoot":"","sources":["../../../src/works/strategies/plos.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAEjD,eAAO,MAAM,YAAY,EAAE,gBAU1B,CAAC"}
|
|
@@ -0,0 +1,15 @@
|
|
|
1
|
+
import { fetchPlosJatsXml } from "../../sources/plos/client";
|
|
2
|
+
export const plosStrategy = {
|
|
3
|
+
name: "plos-doi",
|
|
4
|
+
matches: (ids) => Boolean(ids.doi && /^10\.1371\//i.test(ids.doi)),
|
|
5
|
+
getFullText: async (ids) => {
|
|
6
|
+
try {
|
|
7
|
+
const xml = await fetchPlosJatsXml(ids.doi);
|
|
8
|
+
if (xml)
|
|
9
|
+
return { type: "jats-xml", source: "plos", content: xml };
|
|
10
|
+
}
|
|
11
|
+
catch { }
|
|
12
|
+
return null;
|
|
13
|
+
},
|
|
14
|
+
};
|
|
15
|
+
//# sourceMappingURL=plos.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"plos.js","sourceRoot":"","sources":["../../../src/works/strategies/plos.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,MAAM,2BAA2B,CAAC;AAG7D,MAAM,CAAC,MAAM,YAAY,GAAqB;IAC5C,IAAI,EAAE,UAAU;IAChB,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,GAAG,IAAI,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;IAClE,WAAW,EAAE,KAAK,EAAE,GAAG,EAAE,EAAE;QACzB,IAAI,CAAC;YACH,MAAM,GAAG,GAAG,MAAM,gBAAgB,CAAC,GAAG,CAAC,GAAI,CAAC,CAAC;YAC7C,IAAI,GAAG;gBAAE,OAAO,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,GAAG,EAAE,CAAC;QACrE,CAAC;QAAC,MAAM,CAAC,CAAA,CAAC;QACV,OAAO,IAAI,CAAC;IACd,CAAC;CACF,CAAC"}
|
|
@@ -0,0 +1,11 @@
|
|
|
1
|
+
import type { FetchTextFn } from "../types";
|
|
2
|
+
export declare const cloneJson: <T>(value: T) => T;
|
|
3
|
+
export declare function parseVersionSuffix(value?: string | null): number | null;
|
|
4
|
+
export declare function labelFromVersion(version?: number | null): string | undefined;
|
|
5
|
+
export declare const extractVersionSuffix: (value?: string | null) => number | null;
|
|
6
|
+
export declare const firstFiniteVersion: (...candidates: Array<number | null | undefined>) => number | null;
|
|
7
|
+
export declare const buildBiorxivVersionedId: (idOrDoi: string, version?: number | null) => string | null;
|
|
8
|
+
export declare const isLikelyErrorHtml: (html: string) => boolean;
|
|
9
|
+
export declare const isArxivAbsHtml: (html: string) => boolean;
|
|
10
|
+
export declare const fetchTextWithRetries: FetchTextFn;
|
|
11
|
+
//# sourceMappingURL=shared.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"shared.d.ts","sourceRoot":"","sources":["../../../src/works/strategies/shared.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,WAAW,EAAmB,MAAM,UAAU,CAAC;AAI7D,eAAO,MAAM,SAAS,GAAI,CAAC,EAAE,OAAO,CAAC,KAAG,CAGD,CAAC;AAExC,wBAAgB,kBAAkB,CAAC,KAAK,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,MAAM,GAAG,IAAI,CAMvE;AAED,wBAAgB,gBAAgB,CAAC,OAAO,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,MAAM,GAAG,SAAS,CAG5E;AAED,eAAO,MAAM,oBAAoB,GAAI,QAAQ,MAAM,GAAG,IAAI,KAAG,MAAM,GAAG,IAQrE,CAAC;AAEF,eAAO,MAAM,kBAAkB,GAC7B,GAAG,YAAY,KAAK,CAAC,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC,KAC9C,MAAM,GAAG,IAQX,CAAC;AAEF,eAAO,MAAM,uBAAuB,GAClC,SAAS,MAAM,EACf,UAAU,MAAM,GAAG,IAAI,KACtB,MAAM,GAAG,IASX,CAAC;AAEF,eAAO,MAAM,iBAAiB,GAAI,MAAM,MAAM,KAAG,OAOhD,CAAC;AAEF,eAAO,MAAM,cAAc,GAAI,MAAM,MAAM,KAAG,OAGN,CAAC;AAEzC,eAAO,MAAM,oBAAoB,EAAE,WAqClC,CAAC"}
|
|
@@ -0,0 +1,97 @@
|
|
|
1
|
+
const UA = process.env.UA || "drylab-reveng-api/1.0";
|
|
2
|
+
export const cloneJson = (value) => value === null || value === undefined
|
|
3
|
+
? value
|
|
4
|
+
: JSON.parse(JSON.stringify(value));
|
|
5
|
+
export function parseVersionSuffix(value) {
|
|
6
|
+
if (!value)
|
|
7
|
+
return null;
|
|
8
|
+
const m = value.trim().match(/v(\d+)$/i);
|
|
9
|
+
if (!m)
|
|
10
|
+
return null;
|
|
11
|
+
const n = parseInt(m[1], 10);
|
|
12
|
+
return Number.isFinite(n) ? n : null;
|
|
13
|
+
}
|
|
14
|
+
export function labelFromVersion(version) {
|
|
15
|
+
if (!version || !Number.isFinite(version))
|
|
16
|
+
return undefined;
|
|
17
|
+
return `v${version}`;
|
|
18
|
+
}
|
|
19
|
+
export const extractVersionSuffix = (value) => {
|
|
20
|
+
if (!value)
|
|
21
|
+
return null;
|
|
22
|
+
const match = String(value)
|
|
23
|
+
.trim()
|
|
24
|
+
.match(/v(\d+)(?:[^0-9]|$)/i);
|
|
25
|
+
if (!match || !match[1])
|
|
26
|
+
return null;
|
|
27
|
+
const n = parseInt(match[1], 10);
|
|
28
|
+
return Number.isFinite(n) ? n : null;
|
|
29
|
+
};
|
|
30
|
+
export const firstFiniteVersion = (...candidates) => {
|
|
31
|
+
for (const c of candidates) {
|
|
32
|
+
if (c !== null && c !== undefined) {
|
|
33
|
+
const n = Number(c);
|
|
34
|
+
if (Number.isFinite(n) && n > 0)
|
|
35
|
+
return n;
|
|
36
|
+
}
|
|
37
|
+
}
|
|
38
|
+
return null;
|
|
39
|
+
};
|
|
40
|
+
export const buildBiorxivVersionedId = (idOrDoi, version) => {
|
|
41
|
+
if (!version || !Number.isFinite(version))
|
|
42
|
+
return null;
|
|
43
|
+
const trimmed = String(idOrDoi || "").trim();
|
|
44
|
+
if (!trimmed)
|
|
45
|
+
return null;
|
|
46
|
+
const hasPrefix = /^10\.1101\//i.test(trimmed);
|
|
47
|
+
const base = trimmed.replace(/^10\.1101\//i, "").replace(/v\d+$/i, "");
|
|
48
|
+
if (!base)
|
|
49
|
+
return null;
|
|
50
|
+
const suffix = `v${version}`;
|
|
51
|
+
return `${hasPrefix ? "10.1101/" : ""}${base}${suffix}`;
|
|
52
|
+
};
|
|
53
|
+
export const isLikelyErrorHtml = (html) => {
|
|
54
|
+
const s = html.slice(0, 4096).toLowerCase();
|
|
55
|
+
return (s.includes("<title>502") ||
|
|
56
|
+
s.includes("<h1>502") ||
|
|
57
|
+
(s.includes("error") && s.includes("ar5iv")));
|
|
58
|
+
};
|
|
59
|
+
export const isArxivAbsHtml = (html) => /<link\s+rel=["']canonical["']\s+href=["']https:\/\/arxiv\.org\/abs\//i.test(html) || /id=["']abs-outer["']/.test(html);
|
|
60
|
+
export const fetchTextWithRetries = async (url, accept = "text/html,application/xhtml+xml", tries = 2) => {
|
|
61
|
+
const headers = {
|
|
62
|
+
Accept: accept,
|
|
63
|
+
"User-Agent": UA || "drylab-reveng-api/1.0",
|
|
64
|
+
};
|
|
65
|
+
let lastStatus;
|
|
66
|
+
let failureReason;
|
|
67
|
+
for (let i = 0; i < Math.max(1, tries); i++) {
|
|
68
|
+
try {
|
|
69
|
+
const res = await fetch(url, { headers });
|
|
70
|
+
lastStatus = res.status;
|
|
71
|
+
if (res.ok) {
|
|
72
|
+
const text = await res.text();
|
|
73
|
+
if (text && text.length > 0)
|
|
74
|
+
return { text, finalUrl: res.url };
|
|
75
|
+
failureReason = "empty_response";
|
|
76
|
+
}
|
|
77
|
+
else if (res.status === 404) {
|
|
78
|
+
failureReason = "not_found";
|
|
79
|
+
}
|
|
80
|
+
else if (res.status === 429) {
|
|
81
|
+
failureReason = "rate_limited";
|
|
82
|
+
}
|
|
83
|
+
else if (res.status >= 500) {
|
|
84
|
+
failureReason = "server_error";
|
|
85
|
+
}
|
|
86
|
+
else {
|
|
87
|
+
failureReason = "server_error";
|
|
88
|
+
}
|
|
89
|
+
}
|
|
90
|
+
catch {
|
|
91
|
+
failureReason = "network_error";
|
|
92
|
+
}
|
|
93
|
+
await new Promise((r) => setTimeout(r, 300 * (i + 1)));
|
|
94
|
+
}
|
|
95
|
+
return { text: null, failureReason, lastStatus };
|
|
96
|
+
};
|
|
97
|
+
//# sourceMappingURL=shared.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"shared.js","sourceRoot":"","sources":["../../../src/works/strategies/shared.ts"],"names":[],"mappings":"AAEA,MAAM,EAAE,GAAG,OAAO,CAAC,GAAG,CAAC,EAAE,IAAI,uBAAuB,CAAC;AAErD,MAAM,CAAC,MAAM,SAAS,GAAG,CAAI,KAAQ,EAAK,EAAE,CAC1C,KAAK,KAAK,IAAI,IAAI,KAAK,KAAK,SAAS;IACnC,CAAC,CAAC,KAAK;IACP,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC;AAExC,MAAM,UAAU,kBAAkB,CAAC,KAAqB;IACtD,IAAI,CAAC,KAAK;QAAE,OAAO,IAAI,CAAC;IACxB,MAAM,CAAC,GAAG,KAAK,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;IACzC,IAAI,CAAC,CAAC;QAAE,OAAO,IAAI,CAAC;IACpB,MAAM,CAAC,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAE,EAAE,EAAE,CAAC,CAAC;IAC9B,OAAO,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;AACvC,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,OAAuB;IACtD,IAAI,CAAC,OAAO,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,OAAO,CAAC;QAAE,OAAO,SAAS,CAAC;IAC5D,OAAO,IAAI,OAAO,EAAE,CAAC;AACvB,CAAC;AAED,MAAM,CAAC,MAAM,oBAAoB,GAAG,CAAC,KAAqB,EAAiB,EAAE;IAC3E,IAAI,CAAC,KAAK;QAAE,OAAO,IAAI,CAAC;IACxB,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;SACxB,IAAI,EAAE;SACN,KAAK,CAAC,qBAAqB,CAAC,CAAC;IAChC,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;QAAE,OAAO,IAAI,CAAC;IACrC,MAAM,CAAC,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;IACjC,OAAO,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;AACvC,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAChC,GAAG,UAA4C,EAChC,EAAE;IACjB,KAAK,MAAM,CAAC,IAAI,UAAU,EAAE,CAAC;QAC3B,IAAI,CAAC,KAAK,IAAI,IAAI,CAAC,KAAK,SAAS,EAAE,CAAC;YAClC,MAAM,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;YACpB,IAAI,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC;gBAAE,OAAO,CAAC,CAAC;QAC5C,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,uBAAuB,GAAG,CACrC,OAAe,EACf,OAAuB,EACR,EAAE;IACjB,IAAI,CAAC,OAAO,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,OAAO,CAAC;QAAE,OAAO,IAAI,CAAC;IACvD,MAAM,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;IAC7C,IAAI,CAAC,OAAO;QAAE,OAAO,IAAI,CAAC;IAC1B,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC/C,MAAM,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,cAAc,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;IACvE,IAAI,CAAC,IAAI;QAAE,OAAO,IAAI,CAAC;IACvB,MAAM,MAAM,GAAG,IAAI,OAAO,EAAE,CAAC;IAC7B,OAAO,GAAG,SAAS,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,EAAE,GAAG,IAAI,GAAG,MAAM,EAAE,CAAC;AAC1D,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,iBAAiB,GAAG,CAAC,IAAY,EAAW,EAAE;IACzD,MAAM,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,WAAW,EAAE,CAAC;IAC5C,OAAO,CACL,CAAC,CAAC,QAAQ,CAAC,YAAY,CAAC;QACxB,CAAC,CAAC,QAAQ,CAAC,SAAS,CAAC;QACrB,CAAC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,CAC7C,CAAC;AACJ,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,cAAc,GAAG,CAAC,IAAY,EAAW,EAAE,CACtD,uEAAuE,CAAC,IAAI,CAC1E,IAAI,CACL,IAAI,sBAAsB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAEzC,MAAM,CAAC,MAAM,oBAAoB,GAAgB,KAAK,EACpD,GAAG,EACH,MAAM,GAAG,iCAAiC,EAC1C,KAAK,GAAG,CAAC,EACT,EAAE;IACF,MAAM,OAAO,GAAG;QACd,MAAM,EAAE,MAAM;QACd,YAAY,EAAE,EAAE,IAAI,uBAAuB;KAClB,CAAC;IAE5B,IAAI,UAA8B,CAAC;IACnC,IAAI,aAA+C,CAAC;IAEpD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;QAC5C,IAAI,CAAC;YACH,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE,EAAE,OAAO,EAAS,CAAC,CAAC;YACjD,UAAU,GAAG,GAAG,CAAC,MAAM,CAAC;YACxB,IAAI,GAAG,CAAC,EAAE,EAAE,CAAC;gBACX,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,IAAI,EAAE,CAAC;gBAC9B,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC;oBACzB,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAG,GAAW,CAAC,GAAG,EAAE,CAAC;gBAC9C,aAAa,GAAG,gBAAgB,CAAC;YACnC,CAAC;iBAAM,IAAI,GAAG,CAAC,MAAM,KAAK,GAAG,EAAE,CAAC;gBAC9B,aAAa,GAAG,WAAW,CAAC;YAC9B,CAAC;iBAAM,IAAI,GAAG,CAAC,MAAM,KAAK,GAAG,EAAE,CAAC;gBAC9B,aAAa,GAAG,cAAc,CAAC;YACjC,CAAC;iBAAM,IAAI,GAAG,CAAC,MAAM,IAAI,GAAG,EAAE,CAAC;gBAC7B,aAAa,GAAG,cAAc,CAAC;YACjC,CAAC;iBAAM,CAAC;gBACN,aAAa,GAAG,cAAc,CAAC;YACjC,CAAC;QACH,CAAC;QAAC,MAAM,CAAC;YACP,aAAa,GAAG,eAAe,CAAC;QAClC,CAAC;QACD,MAAM,IAAI,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,UAAU,CAAC,CAAC,EAAE,GAAG,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IACzD,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,aAAa,EAAE,UAAU,EAAE,CAAC;AACnD,CAAC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"ten1101.d.ts","sourceRoot":"","sources":["../../../src/works/strategies/ten1101.ts"],"names":[],"mappings":"AAQA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,UAAU,CAAC;AAUjD,eAAO,MAAM,uBAAuB,EAAE,gBAkGrC,CAAC"}
|
|
@@ -0,0 +1,84 @@
|
|
|
1
|
+
import { fetchBiorxivSourceXmlViaApi, buildBiorxivLatestSourceXmlUrlViaApi, } from "../../sources/biorxiv/client";
|
|
2
|
+
import { fetchMedrxivSourceXmlViaApi, buildMedrxivLatestSourceXmlUrlViaApi, } from "../../sources/medrxiv/client";
|
|
3
|
+
import { parseVersionSuffix, extractVersionSuffix, firstFiniteVersion, labelFromVersion, } from "./shared";
|
|
4
|
+
const UA = process.env.UA || "drylab-reveng-api/1.0";
|
|
5
|
+
export const ten1101FallbackStrategy = {
|
|
6
|
+
name: "generic-10.1101",
|
|
7
|
+
matches: (ids) => Boolean(!ids.biorxivId &&
|
|
8
|
+
!ids.medrxivId &&
|
|
9
|
+
ids.doi &&
|
|
10
|
+
/^10\.1101\//i.test(ids.doi)),
|
|
11
|
+
getFullText: async (ids) => {
|
|
12
|
+
const doi = ids.doi;
|
|
13
|
+
const core = doi.replace(/^10\.1101\//i, "").replace(/v\d+$/i, "");
|
|
14
|
+
const doiForApi = `10.1101/${core}`;
|
|
15
|
+
const preferVersion = parseVersionSuffix(doi);
|
|
16
|
+
try {
|
|
17
|
+
const hitBx = await fetchBiorxivSourceXmlViaApi(doiForApi, {
|
|
18
|
+
UA,
|
|
19
|
+
preferVersion,
|
|
20
|
+
});
|
|
21
|
+
if (hitBx) {
|
|
22
|
+
const resolvedVersion = firstFiniteVersion(hitBx.version, extractVersionSuffix(hitBx.url), preferVersion);
|
|
23
|
+
return {
|
|
24
|
+
type: "jats-xml",
|
|
25
|
+
source: "biorxiv",
|
|
26
|
+
content: hitBx.xml,
|
|
27
|
+
url: hitBx.url,
|
|
28
|
+
versionLabel: labelFromVersion(resolvedVersion),
|
|
29
|
+
};
|
|
30
|
+
}
|
|
31
|
+
const latestBx = await buildBiorxivLatestSourceXmlUrlViaApi(doiForApi, {
|
|
32
|
+
UA,
|
|
33
|
+
preferVersion,
|
|
34
|
+
});
|
|
35
|
+
if (latestBx?.url) {
|
|
36
|
+
const resolvedVersion = firstFiniteVersion(latestBx.version, extractVersionSuffix(latestBx.url), preferVersion);
|
|
37
|
+
return {
|
|
38
|
+
type: "jats-xml",
|
|
39
|
+
source: "biorxiv",
|
|
40
|
+
url: latestBx.url,
|
|
41
|
+
versionLabel: labelFromVersion(resolvedVersion),
|
|
42
|
+
};
|
|
43
|
+
}
|
|
44
|
+
}
|
|
45
|
+
catch { }
|
|
46
|
+
try {
|
|
47
|
+
const hitMx = await fetchMedrxivSourceXmlViaApi(doiForApi, {
|
|
48
|
+
UA,
|
|
49
|
+
preferVersion,
|
|
50
|
+
});
|
|
51
|
+
if (hitMx) {
|
|
52
|
+
const resolvedVersion = firstFiniteVersion(hitMx.version, extractVersionSuffix(hitMx.url), preferVersion);
|
|
53
|
+
return {
|
|
54
|
+
type: "jats-xml",
|
|
55
|
+
source: "medrxiv",
|
|
56
|
+
content: hitMx.xml,
|
|
57
|
+
url: hitMx.url,
|
|
58
|
+
versionLabel: labelFromVersion(resolvedVersion),
|
|
59
|
+
};
|
|
60
|
+
}
|
|
61
|
+
const latestMx = await buildMedrxivLatestSourceXmlUrlViaApi(doiForApi, {
|
|
62
|
+
UA,
|
|
63
|
+
preferVersion,
|
|
64
|
+
});
|
|
65
|
+
if (latestMx?.url) {
|
|
66
|
+
const resolvedVersion = firstFiniteVersion(latestMx.version, extractVersionSuffix(latestMx.url), preferVersion);
|
|
67
|
+
return {
|
|
68
|
+
type: "jats-xml",
|
|
69
|
+
source: "medrxiv",
|
|
70
|
+
url: latestMx.url,
|
|
71
|
+
versionLabel: labelFromVersion(resolvedVersion),
|
|
72
|
+
};
|
|
73
|
+
}
|
|
74
|
+
}
|
|
75
|
+
catch { }
|
|
76
|
+
return {
|
|
77
|
+
type: "jats-xml",
|
|
78
|
+
source: "preprint",
|
|
79
|
+
url: `https://www.biorxiv.org/content/${doiForApi}.source.xml`,
|
|
80
|
+
versionLabel: labelFromVersion(preferVersion),
|
|
81
|
+
};
|
|
82
|
+
},
|
|
83
|
+
};
|
|
84
|
+
//# sourceMappingURL=ten1101.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"ten1101.js","sourceRoot":"","sources":["../../../src/works/strategies/ten1101.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,2BAA2B,EAC3B,oCAAoC,GACrC,MAAM,8BAA8B,CAAC;AACtC,OAAO,EACL,2BAA2B,EAC3B,oCAAoC,GACrC,MAAM,8BAA8B,CAAC;AAEtC,OAAO,EACL,kBAAkB,EAClB,oBAAoB,EACpB,kBAAkB,EAClB,gBAAgB,GACjB,MAAM,UAAU,CAAC;AAElB,MAAM,EAAE,GAAG,OAAO,CAAC,GAAG,CAAC,EAAE,IAAI,uBAAuB,CAAC;AAErD,MAAM,CAAC,MAAM,uBAAuB,GAAqB;IACvD,IAAI,EAAE,iBAAiB;IACvB,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE,CACf,OAAO,CACL,CAAC,GAAG,CAAC,SAAS;QACZ,CAAC,GAAG,CAAC,SAAS;QACd,GAAG,CAAC,GAAG;QACP,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAC/B;IACH,WAAW,EAAE,KAAK,EAAE,GAAG,EAAE,EAAE;QACzB,MAAM,GAAG,GAAG,GAAG,CAAC,GAAI,CAAC;QACrB,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,cAAc,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QACnE,MAAM,SAAS,GAAG,WAAW,IAAI,EAAE,CAAC;QACpC,MAAM,aAAa,GAAG,kBAAkB,CAAC,GAAG,CAAC,CAAC;QAE9C,IAAI,CAAC;YACH,MAAM,KAAK,GAAG,MAAM,2BAA2B,CAAC,SAAS,EAAE;gBACzD,EAAE;gBACF,aAAa;aACd,CAAC,CAAC;YACH,IAAI,KAAK,EAAE,CAAC;gBACV,MAAM,eAAe,GAAG,kBAAkB,CACxC,KAAK,CAAC,OAAO,EACb,oBAAoB,CAAC,KAAK,CAAC,GAAG,CAAC,EAC/B,aAAa,CACd,CAAC;gBACF,OAAO;oBACL,IAAI,EAAE,UAAU;oBAChB,MAAM,EAAE,SAAS;oBACjB,OAAO,EAAE,KAAK,CAAC,GAAG;oBAClB,GAAG,EAAE,KAAK,CAAC,GAAG;oBACd,YAAY,EAAE,gBAAgB,CAAC,eAAe,CAAC;iBAChD,CAAC;YACJ,CAAC;YACD,MAAM,QAAQ,GAAG,MAAM,oCAAoC,CAAC,SAAS,EAAE;gBACrE,EAAE;gBACF,aAAa;aACd,CAAC,CAAC;YACH,IAAI,QAAQ,EAAE,GAAG,EAAE,CAAC;gBAClB,MAAM,eAAe,GAAG,kBAAkB,CACxC,QAAQ,CAAC,OAAO,EAChB,oBAAoB,CAAC,QAAQ,CAAC,GAAG,CAAC,EAClC,aAAa,CACd,CAAC;gBACF,OAAO;oBACL,IAAI,EAAE,UAAU;oBAChB,MAAM,EAAE,SAAS;oBACjB,GAAG,EAAE,QAAQ,CAAC,GAAG;oBACjB,YAAY,EAAE,gBAAgB,CAAC,eAAe,CAAC;iBAChD,CAAC;YACJ,CAAC;QACH,CAAC;QAAC,MAAM,CAAC,CAAA,CAAC;QAEV,IAAI,CAAC;YACH,MAAM,KAAK,GAAG,MAAM,2BAA2B,CAAC,SAAS,EAAE;gBACzD,EAAE;gBACF,aAAa;aACd,CAAC,CAAC;YACH,IAAI,KAAK,EAAE,CAAC;gBACV,MAAM,eAAe,GAAG,kBAAkB,CACxC,KAAK,CAAC,OAAO,EACb,oBAAoB,CAAC,KAAK,CAAC,GAAG,CAAC,EAC/B,aAAa,CACd,CAAC;gBACF,OAAO;oBACL,IAAI,EAAE,UAAU;oBAChB,MAAM,EAAE,SAAS;oBACjB,OAAO,EAAE,KAAK,CAAC,GAAG;oBAClB,GAAG,EAAE,KAAK,CAAC,GAAG;oBACd,YAAY,EAAE,gBAAgB,CAAC,eAAe,CAAC;iBAChD,CAAC;YACJ,CAAC;YACD,MAAM,QAAQ,GAAG,MAAM,oCAAoC,CAAC,SAAS,EAAE;gBACrE,EAAE;gBACF,aAAa;aACd,CAAC,CAAC;YACH,IAAI,QAAQ,EAAE,GAAG,EAAE,CAAC;gBAClB,MAAM,eAAe,GAAG,kBAAkB,CACxC,QAAQ,CAAC,OAAO,EAChB,oBAAoB,CAAC,QAAQ,CAAC,GAAG,CAAC,EAClC,aAAa,CACd,CAAC;gBACF,OAAO;oBACL,IAAI,EAAE,UAAU;oBAChB,MAAM,EAAE,SAAS;oBACjB,GAAG,EAAE,QAAQ,CAAC,GAAG;oBACjB,YAAY,EAAE,gBAAgB,CAAC,eAAe,CAAC;iBAChD,CAAC;YACJ,CAAC;QACH,CAAC;QAAC,MAAM,CAAC,CAAA,CAAC;QAEV,OAAO;YACL,IAAI,EAAE,UAAU;YAChB,MAAM,EAAE,UAAU;YAClB,GAAG,EAAE,mCAAmC,SAAS,aAAa;YAC9D,YAAY,EAAE,gBAAgB,CAAC,aAAa,CAAC;SAC9C,CAAC;IACJ,CAAC;CACF,CAAC"}
|
|
@@ -0,0 +1,7 @@
|
|
|
1
|
+
import type { FullTextResult, CompleteIds, StrategyAttempt, LoggerLike } from "../types";
|
|
2
|
+
export declare function getBestFullText(ids: CompleteIds, logger?: LoggerLike): Promise<FullTextResult | null>;
|
|
3
|
+
export declare function getBestFullTextWithTrace(ids: CompleteIds, logger?: LoggerLike): Promise<{
|
|
4
|
+
result: FullTextResult | null;
|
|
5
|
+
trace: StrategyAttempt[];
|
|
6
|
+
}>;
|
|
7
|
+
//# sourceMappingURL=acquire-fulltext.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"acquire-fulltext.d.ts","sourceRoot":"","sources":["../../../src/works/text/acquire-fulltext.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EACV,cAAc,EACd,WAAW,EACX,eAAe,EACf,UAAU,EACX,MAAM,UAAU,CAAC;AA6ElB,wBAAsB,eAAe,CACnC,GAAG,EAAE,WAAW,EAChB,MAAM,CAAC,EAAE,UAAU,GAClB,OAAO,CAAC,cAAc,GAAG,IAAI,CAAC,CAGhC;AAED,wBAAsB,wBAAwB,CAC5C,GAAG,EAAE,WAAW,EAChB,MAAM,CAAC,EAAE,UAAU,GAClB,OAAO,CAAC;IAAE,MAAM,EAAE,cAAc,GAAG,IAAI,CAAC;IAAC,KAAK,EAAE,eAAe,EAAE,CAAA;CAAE,CAAC,CAEtE"}
|