Validate data.table Against ParamSet

Validates that a data.table is compatible with a given ParamSet. Checks column presence, types, bounds, and uniqueness within the domain and does type coercion if necessary.

Usage

assert_data_table_param_set(
  dt,
  param_set,
  require_uniqueness = TRUE,
  min_rows = 1L,
  presence = "all",
  allow_untyped = FALSE,
  allow_extra = TRUE,
  .param_set_name = "param_set",
  .dt_name = "dt"
)

Arguments

dt

(data.table)
data.table to validate.

param_set

(paradox::ParamSet)
ParamSet describing the space to validate against.

require_uniqueness

(logical(1))
Whether to require uniqueness within the space. Default is TRUE.

min_rows

(integer(1))
Minimum number of rows required in dt. Default is 1L.

presence

(character(1))
Whether all parameters must be present as columns in dt. Use "all" (default) to require all parameters, or "subset" to allow missing parameter columns and validate them as typed NA values.

allow_untyped

(logical(1))
Whether to allow p_uty in the ParamSet. Default is FALSE. p_uty columns are not checked for type compatibility. Their custom_check is applied when present.

allow_extra

(logical(1))
Whether columns outside the ParamSet are allowed. Default is TRUE.

.param_set_name

(character(1))
Name of the ParamSet for error messages. Default is "param_set".

.dt_name

(character(1))
Name of the data.table for error messages. Default is "dt".

Value

The validated and possibly coerced data.table (modified by reference).

Details

Extra columns are allowed and are passed through unchanged when allow_extra = TRUE. Missing parameter columns are rejected unless presence = "subset", in which case they are validated as typed NA values on an internal copy. When require_uniqueness is TRUE, tables with rows that are duplicated within the space of the ParamSet are rejected, even if they differ by other columns.