Apply overrides to a crossings table

Join loaded overrides onto a crossings table and update matching columns. Step two (or three) of the override pipeline: lnk_override_load() -> lnk_override_validate() -> apply.

Usage

lnk_override_apply(
  conn,
  crossings,
  overrides,
  col_id = "modelled_crossing_id",
  cols_update = NULL,
  cols_provenance = c("reviewer", "review_date", "source"),
  verbose = TRUE
)

Arguments

conn

A DBI::DBIConnection object.

crossings

Character. Schema-qualified crossings table to update (e.g., "working.crossings").

overrides

Character. Schema-qualified override table (output of lnk_override_load()).

col_id

Character. Join column — the crossing identifier shared by both tables. System-agnostic.

cols_update

Character vector. Columns to copy from overrides to crossings. NULL (default) auto-detects: all columns in overrides that also exist in crossings, excluding col_id and provenance columns.

cols_provenance

Character vector. Columns to exclude from auto-detection (they track who reviewed, not crossing attributes).

verbose

Logical. Report how many rows were updated.

Value

A list with n_updated (rows changed) and cols_updated (columns that were updated), invisibly.

Details

Auto-detect mode: when cols_update = NULL, the function finds columns that exist in both the overrides and crossings tables (excluding the join column and provenance columns) and updates those. This means if your override CSV has barrier_result_code and your crossings table has barrier_result_code, it just works — no configuration needed.

Explicit mode: set cols_update = c("barrier_result_code") when you want precision about exactly which columns change.

Idempotent: running twice produces the same result.

Examples

# --- The override pipeline: load, then apply ---
if (FALSE) { # \dontrun{
conn <- lnk_db_conn()

# Step 1: Load overrides
lnk_override_load(conn,
  csv = "data/overrides/modelled_xings_fixes.csv",
  to  = "working.overrides_modelled")

# Step 2: Apply — auto-detects which columns to update
result <- lnk_override_apply(conn,
  crossings = "working.crossings",
  overrides = "working.overrides_modelled")
# Updated 342 of 15,230 crossings (barrier_result_code)
#
# The verbose output tells you the magnitude of changes —
# essential for QA. 342 corrections from 3 years of field work.

# Step 3: Score the corrected crossings
lnk_score_severity(conn, "working.crossings")
# Severity scores now reflect field-verified barrier status,
# not just the raw modelled data.

# --- Explicit column selection ---
lnk_override_apply(conn,
  crossings   = "working.crossings",
  overrides   = "working.overrides_modelled",
  cols_update = c("barrier_result_code"))
# Only updates barrier_result_code, even if the override table
# has other columns that match.
} # }