Build a labelled test-case table for one data type — generate_autotype_test

Assembles the pure inputs used by the bundled data_type_tests object: the real positive examples plus deduplicated negatives from each mutation tier, labelled by origin. Unlike build_autotype_results(), this stores no detector predictions – it is the fixture a detector is later scored against.

Usage

generate_autotype_test_cases(
  P,
  data_type,
  tiers = c("S1", "S2", "S3", "SP"),
  n_per_positive = 20,
  p = 0.5,
  seed = 514
)

Arguments

P: A character vector of positive examples.
data_type: A single string naming the type (e.g. "currency_code").
tiers: Mutation tiers to include.
n_per_positive: Number of mutants to attempt per positive per tier.
p: Per-character mutation probability.
seed: Integer seed; set for reproducible fixtures.

Value

A data frame with columns data_type, case, and label, where label is "valid" for real positives or the tier name ("S1", "S2", "S3", "SP") for negatives. Negatives coinciding with a real positive are dropped.