Check that override records reference real crossings and flag orphans,
duplicates, and conflicts. Optional step between lnk_override_load()
and lnk_override_apply() — recommended for production workflows where
overrides accumulate across field seasons.
Usage
lnk_override_validate(
conn,
overrides,
crossings,
col_id = "modelled_crossing_id",
verbose = TRUE
)Arguments
- conn
A DBI::DBIConnection object.
- overrides
Character. Schema-qualified override table.
- crossings
Character. Schema-qualified crossings table to validate against.
- col_id
Character. Join column (system-agnostic).
- verbose
Logical. Print a summary of findings.
Value
A list with:
- orphans
Override IDs not found in crossings (GPS error? wrong watershed? crossing removed from model?)
- duplicates
Crossing IDs that appear more than once in overrides (conflicting corrections — which one wins?)
- valid_count
Number of overrides that will apply cleanly.
- total_count
Total override records.
Details
Non-blocking: returns findings but does not error. The user decides whether orphans are acceptable (they often are — crossings get removed from models between versions).
Why validate? Override CSVs accumulate over years. Crossings get renumbered, GPS coordinates get corrected, models get rebuilt. Without validation, stale overrides silently fail to match and corrections are lost.
Examples
# --- Why validation matters ---
# You loaded 3 years of field reviews (1,200 overrides).
# The modelled crossings layer was rebuilt last month.
# How many overrides still point at valid crossings?
if (FALSE) { # \dontrun{
conn <- lnk_db_conn()
lnk_override_load(conn,
csv = c("data/overrides/2023_field.csv",
"data/overrides/2024_field.csv",
"data/overrides/2025_field.csv"),
to = "working.overrides_all")
result <- lnk_override_validate(conn,
overrides = "working.overrides_all",
crossings = "working.crossings")
# Override validation: working.overrides_all vs working.crossings
# Total overrides: 1,200
# Valid (matched): 1,147
# Orphans: 48 <-- crossings removed from model
# Duplicates: 5 <-- same crossing corrected twice
#
# The 48 orphans are expected — the model was rebuilt.
# The 5 duplicates need manual review: which correction wins?
# Inspect the orphans
result$orphans
# [1] 5042 5108 5203 ...
# Inspect the duplicates
result$duplicates
# [1] 1004 1007 ...
# If satisfied, apply
lnk_override_apply(conn, "working.crossings",
"working.overrides_all")
} # }