dr_dbt_contract()

Create an ordinary YAML-ready list containing explicit column types, not-null tests and a single-column unique-key test. The default SQL types target dbt-duckdb. Other adapters can supply explicit types overrides. This does not install dbt packages, modify a project or execute validation.

Usage

dr_dbt_contract(
  contract,
  name,
  enforced = TRUE,
  types = NULL,
  unsupported = c("error", "report")
)

Arguments

contract: A dataraft.core::dr_contract() specification.
name: dbt model identifier.
enforced: Whether dbt should enforce the exact model schema.
types: Optional named character vector overriding SQL types by column. For example, c(amount = "decimal(18,2)"). Nested R list columns require an explicit SQL type because their shape cannot be inferred from list.
unsupported: How to handle checks that cannot be translated: "error" (default) or "report" with a warning and metadata.

Value

A regular list with version = 2 and one models entry, suitable for yaml::write_yaml() or composition into dbt model properties.

Details

Schema enforcement is supported for SQL models with compatible dbt materializations. It enforces the exact exported names and types. Required fields and keys use dbt data tests, whose execution is the responsibility of dbt build or dbt test; no database constraint guarantee is implied.

R functions, formulas, pointblank rules and composite-key checks are not translated. By default they cause an error. With unsupported = "report", a warning names them and config.meta.dataraft.untranslated records them in the exported model. Missing SQL types always cause an error. R lifecycle policies such as empty-data, freshness and additional-column handling are recorded separately under r_policies; they still require R validation. The export is deliberately a bounded schema bridge, not a replacement for the original contract or its quality engines.

Examples

schema <- dataraft.core::dr_contract(columns = c(id = "integer", amount = "numeric"), key = "id")
properties <- dr_dbt_contract(schema, "orders")
properties$models[[1]]$columns