`isid`¶

isid is a read-only key-uniqueness quality gate inspired by Stata's isid, SAS key-validation workflows, and SPSS duplicate-ID validation. It makes a composite-key invariant explicit and scriptable without changing the active dataset.

Syntax¶

isid varlist [, missok]

missok is the only supported option. At least one key variable is required.

Semantics¶

Each complete key combination must occur at most once.
Null key values compare equal for duplicate grouping, so repeated null-containing combinations fail.
Without missok, any row with a null in any key variable fails even when that incomplete key is otherwise unique.
With missok, incomplete keys are allowed only when their complete combinations remain unique.
Empty datasets pass with zero rows and zero unique groups.
Unknown variables are rejected before scanning. The command is read-only and preserves labels, panel metadata, named tables, and the active lazy plan.
Eager, DuckDB-lazy, and Polars-lazy execution use aggregate scans.

Output¶

isid passed
Key variables: patient_id visit
Rows checked: 3
Unique groups: 3
Rows with missing keys: 0
Missing keys allowed: no

JSON output uses result_type: "IsidResult" with variables, total_rows, unique_groups, missing_key_rows, and missok fields.

Failures use deterministic command-level errors, for example:

isid failed: 2 rows have missing key values (use , missok to permit them); 3 rows are in 1 duplicate key groups

Examples¶

use visits.parquet, lazy engine=polars
isid patient_id visit
isid patient_id visit, missok

The first form requires every key component to be present; the second allows incomplete keys only when those combinations are unique.

Invalid forms¶

isid
isid patient_id if active == true
isid patient_id, report
isid patient_id, missok(true)
isid missing_column

These forms fail with deterministic parse or execution errors and do not mutate active state.

isid¶