Correct a source's ship code on one cruise only

A source can name the wrong ship for a single cruise while naming that ship correctly on every other cruise. A global rename (ship_renames.csv, by name and code) cannot fix that without breaking the cruises it already gets right. The CalCOFI bottle database and the picoplankton studyName call the July 2009 cruise 2009-07-14-C-31M4 (MCARTHUR, which left service in 2003), while every other dataset and the SWFSC reference call it 2009-07-33MH (MCARTHUR II), so the bottle casts were released on a derived cruise of their own and joined to nothing. Cruise 198502 is a real 31M4 cruise.

Usage

correct_cruise_ship(
  con,
  table_name,
  corrections_csv,
  dataset_key,
  cruise_ym_col = "cruise",
  ship_name_col = NULL
)

Arguments

con

DBI connection holding table_name.

table_name

Table with a ship_code column and cruise_ym_col.

corrections_csv

Path to the registry, with columns ship_code, cruise_ym, ship_code_new, dataset_key and evidence.

dataset_key

This ingest's dataset key. Only rows listing it, or listing no dataset, apply.

cruise_ym_col

Column holding the source's own cruise designation as YYYYMM (character or number; compared as an integer).

ship_name_col

Optional column to set to ship_name_new where the registry gives one (default NULL, so no name is changed).

Value

A tibble of the applied rows with n_updated.

Details

Each row of corrections_csv names a source ship code and a source cruise designation (cruise_ym, YYYYMM), plus the corrected code, and the rows of table_name holding both get the corrected ship_code. Run it before derive_cruise_key_on_casts(). A row that lists dataset_key (;-separated, or empty for every dataset) and matches nothing in that table is an error. A correction that no longer applies should be deleted from the registry, not left to pass silently.