Build one dataset's EML 2.2 document from the catalog record

Plan § D-8: eml/{dataset_key}.xml in every release, generated from the record and the descriptive sidecar so the DwC-A, the EDI package, ERDDAP's globals and the page's JSON-LD all read one document. The mapping, field by field:

Usage

build_eml(
  record,
  sidecar = NULL,
  meta = NULL,
  coverage = NULL,
  release = NULL,
  gear = NULL
)

Arguments

record

one dataset record (an element of datasets.json's datasets[])

sidecar

the dataset's dataset_meta.yml as a list (the narrative fields — methods_md, study_extent, sampling_description, quality_control_md, maintenance, associated_parties — live only there)

meta

the release metadata.json (path or parsed list), for columns{}

coverage

the release coverage.json (path or parsed list), for taxa[]

release

the release block of datasets.json (version, date, citation); defaults to the record's own release element when it carries one

gear

the tibble from read_gear_registry(), or NULL

Value

A named list ready for EML::write_eml(), carrying an "eml_notes" attribute (which fallbacks were used) that check_eml() reads.

Details

EML	from
packageId / system	{dataset_key}.{release version} / calcofi.io
dataset/alternateIdentifier	attribution.doi (as a DOI URL) and links.page
dataset/shortName · title	dataset_name_short · dataset_name
dataset/creator	attribution.creators[], else pi_names with the provider organization, else the provider organization alone
dataset/pubDate	the release release_date
dataset/language	eng
dataset/abstract	description_md, rendered to paragraphs
dataset/keywordSet	keywords — the GCMD terms under their thesaurus, plus the category and the observed variable names
dataset/intellectualRights · licensed	attribution.license / license_name / license_url (metadata/license.csv)
dataset/distribution	links.page
dataset/coverage	geographic from coverage.bbox + coverage.spatial; temporal from coverage.year_min/max; taxonomic from coverage.json's taxa[] for this dataset (WoRMS / ITIS ids)
dataset/maintenance	the sidecar's maintenance
dataset/contact	attribution.contact, else a creator email, else eml_contact_address()
dataset/methods	the sidecar's methods_md, quality_control_md, study_extent, sampling_description, with metadata/gear.csv's dwc_samplingProtocol sentences for the dataset's tow_types
dataset/project	dataset_name + attribution.funding (else acknowledgement), personnel from the creators
dataset/dataTable[]	the record's tables[], attributeList from metadata.json's columns{} (label, definition, units, storage type), physical from the record's objects[]
additionalMetadata	the release citation, the dataset citation and the record's own provenance

Absent optional fields are omitted; a missing required field is a check_eml() finding, never a placeholder.

See also

check_eml(), write_eml_files()