The number of test units with each row status.
Comparison.status_counts: dict[str, int]
Where n_passed and n_failed say how many rows differ, status_counts says how they differ. The statuses are described in the Test Units and Row Statuses section of compare(). The counts are computed when the comparison is made, so reading them doesn’t touch the tables again.
The counts are the row statuses alone and aren’t affected by the schema check. If the schema check fails, there can be "match" rows even though no test units pass.
In this example, station S3 appears twice in the source table.
import pointblank as pb
import polars as pl
readings_v1 = pl.DataFrame(
{"station": ["S1", "S2", "S3", "S3", "S4"], "temp": [12.1, 14.0, 9.8, 9.9, 11.2]}
)
readings_v2 = pl.DataFrame(
{"station": ["S1", "S2", "S3", "S5"], "temp": [12.1, 14.3, 9.8, 8.0]}
)
comparison = pb.compare(readings_v1, readings_v2, keys="station")
comparison.status_counts{'match': 1, 'changed': 1, 'missing': 1, 'extra': 1, 'dup_key': 2}Each status has a different cause. S2 changed, S4 is missing from the target, and S5 is extra. The duplicated S3 key counts as two "dup_key" units, since it occurs twice in the source. Those rows can’t be paired up, so they’re not compared. To see the rows behind any of these counts, pass the status to rows(), as in comparison.rows("dup_key").