Skip to contents

What each taxon_override.csv row matched, applied to and skipped

Source: R/taxa.R
report_taxon_overrides.Rd

Recomputes the override rule (see resolve_dataset_taxon()) from dataset_taxon alone — the dataset's vocabulary, its ds_source_json (what the source supplied) and the registry — so the same report is available in an ingest after the resolve and in release_database.qmd, where every dataset is present and only dataset_taxon is. One row per override row:

Usage

report_taxon_overrides(con, overrides, tbl = "dataset_taxon", verbose = TRUE)

Arguments

con

a DBI connection holding dataset_taxon

overrides

the override registry (metadata/taxon_override.csv)

tbl

crosswalk table name (default "dataset_taxon")

verbose

logical; message the per-dataset summary

Value

a data.frame (dataset_key, match_column, match_value, override_key, n_matched, n_applied, n_skipped, skipped_codes, source_json_known), one row per override row, invisibly when verbose = FALSE

Details

  • n_matched — vocabulary rows whose match_column equals match_value;

  • n_applied — of those, the rows the override keyed (a code match, or a row whose source supplied no id);

  • n_skipped — of those, the rows that kept the source's own id, with the first of their codes in skipped_codes. NA when source_json_known is FALSE: no row of that dataset carries ds_source_json, so the shard either predates calcofi4db 3.29.0 or its source supplied no ids at all, and the release cannot tell which — the ingest's own message can.

match_column is one of dataset_taxon's ds_taxa_code / ds_scientific_name / ds_common_name; anything else errors, as it does at ingest (the transitional aliases for the arms' own column names went with the arms in 4.0.0). A row that matches nothing is reported with zeros, not dropped — check_taxon_registries() is where a dataset nobody supplies fails.