Skip to contents

Report where row evidence disagrees.

Source: R/row_evidence.R
evidence_conflicts.Rd

Find the (row_key, field) pairs carrying more than one distinct value, i.e. the places where two producers, or two runs of one producer, made different claims about the same row. Reports rather than aborts, so a caller can decide which disagreements matter; nrow() == 0 is the clean result.

A repeated identical claim is not a conflict, so re-recording the same evidence is harmless.

Usage

evidence_conflicts(evidence)

Arguments

evidence

Evidence table, validated against row_evidence_schema().

Value

A tibble with one row per conflicting (row_key, field), ordered by both:

  • row_key, field: The row and evidence field in dispute.

  • n_values: Number of distinct values claimed. NA counts as a value of its own, since "unknown" and "0.5" are a disagreement.

  • values: The claimed values, " | "-separated, with NA rendered as "NA".

  • source_ids: The producers involved, comma-separated.

Examples

rows <- tibble::tibble(area_code = 724L, year = 2020L)
key <- c("area_code", "year")
recorded_at <- as.POSIXct("2026-01-01", tz = "UTC")
agreeing <- combine_row_evidence(
  row_evidence(rows, "A", key, list(flag = "E"), recorded_at = recorded_at),
  row_evidence(rows, "B", key, list(flag = "E"), recorded_at = recorded_at)
)
evidence_conflicts(agreeing)
#> # A tibble: 0 × 5
#> # ℹ 5 variables: row_key <chr>, field <chr>, n_values <int>, values <chr>,
#> #   source_ids <chr>

disagreeing <- combine_row_evidence(
  agreeing,
  row_evidence(rows, "C", key, list(flag = "A"), recorded_at = recorded_at)
)
evidence_conflicts(disagreeing)
#> # A tibble: 1 × 5
#>   row_key         field n_values values source_ids
#>   <chr>           <chr>    <int> <chr>  <chr>     
#> 1 "724\u001f2020" flag         2 A | E  A, B, C