The ingest asserts its own taxon crosswalk (taxon plan D6)

Call it after resolve_dataset_taxon() and before append_obs(). Three findings, each a row of the returned report:

Usage

check_dataset_taxon(
  con,
  dataset_key,
  allow = character(),
  halt = TRUE,
  codes = NULL,
  verbose = TRUE
)

Arguments

con

a DBI connection holding dataset_taxon (and taxon)

dataset_key

the dataset whose crosswalk is checked

allow

character vector of taxon_keys accepted as-is: dataset-local keys of non-taxonomic classes, or a worms: key for an Aves taxon with no TSN

halt

logical; stop() on any finding (default TRUE)

codes

optional character vector of the ds_taxa_codes the observations reference (e.g. DISTINCT species_code of the source observation table); every one must be in the vocabulary

verbose

logical; message the summary

Value

a data.frame with one row per finding (check, ds_taxon_key, ds_taxa_code, taxon_key, detail); zero rows when clean. Invisible when verbose = FALSE.

Details

• missing_code — a code the observations reference (codes) that is not in this dataset's dataset_taxon. Farallon's MEGU (the pre-split Mew Gull code, present in the observations and absent from the species list) is the case that motivated it: an obs projection joining on the code would drop or NULL those rows with no error anywhere.

• unresolved — a dataset_taxon row with no authority taxon_key (worms: / itis:), unless its dataset-local key is in allow — the ingest's own declaration of a genuinely non-taxonomic class (zooscan "nauplii", phyto "undefined code"), one key at a time, with a comment.

• aves_not_itis — a taxon whose class is Aves that did not key itis: (no accepted TSN resolved; see taxon_key_of()). Checked here because this is where it is cheap to fix — a TSN in taxon_override.csv. An ingest that accepts the worms: key lists that key in allow. Needs taxon in con (build_taxon_reference()) for the class; skipped without it.

release_database.qmd's check_taxon_ids() stays as the backstop.