Labelled test cases for data-type detectors

A frozen set of valid and deliberately-corrupted example values for each data type, used to exercise the detectors in the R/dt-<family>.R files (e.g. R/dt-currency.R). Negatives are produced by the AutoType mutation strategies (see generate_autotype_test_cases) with a fixed seed, so the fixture is reproducible.

Usage

data_type_tests

Format

A named list, one element per data type (e.g. currency_code). Each element is a data frame with columns:

data_type

Character. The type name (matches the list element).

case

Character. The example value.

label

Character. "valid" for a real positive, or the mutation tier that produced a negative: "S1" (preserve structure), "S2" (preserve alphabet), "S3" (random), or "SP" (scramble structure).

Source

Positive examples seeded from the AutoType benchmark (Yan & He, SIGMOD 2018), cleaned and expanded; negatives generated by data-raw/build_test_cases.R.

See also

generate_autotype_test_cases, guess_data_type