dr_dbt_init()

Write a small dbt-duckdb project with RAW inputs, staging casts, a core order model, customer revenue and uniqueness/not-null tests. Supply accepted R-ingested sources to use real immutable RAW relations. With no sources, the starter remains a runnable synthetic seed demonstration. Existing non-empty directories are never overwritten. No connection is opened.

Usage

dr_dbt_init(
  path,
  config,
  name = "dataraft_demo",
  executable = "dbt",
  sources = NULL
)

Arguments

path: Character scalar giving a new or empty directory.
config: A dataraft.lake::dr_lake_config() with a local DuckDB catalog and local storage.
name: Character scalar. A dbt project identifier containing letters, digits and underscores, starting with a letter.
executable: Character scalar giving a dbt executable name or path. It is passed to processx::run() without a shell.
sources: Optional named list accepted by dr_dbt_sources(). The canned order example requires an orders entry whose recorded schema includes order_id, customer_id and amount. Staging casts amount to SQL double, including when input amounts are character strings. Invalid numeric strings fail the dbt build. Other named sources also receive staging models. For arbitrary business models use dr_dbt_sources() with your own project instead of this order-specific starter.

Value

A dr_dbt_project() specification pointing to the written project and profile. Its source_config records the expected catalog. Source mode creates no seeds and never writes RAW tables. Configure layers = c("raw", "staging", "core", "marts") in dataraft.lake::dr_lake_config(). A named layer vector can map staging, core and marts to other schema names; the ingestion schema remains raw. Seed mode also works with the usual three-layer lake configuration, creating its dbt schemas on build.

Details

The generated profile targets the catalog in config through the lake attachment. Supported configurations use a local metadata catalog and local storage. For S3, PostgreSQL or dbt v2 catalog configuration, create and verify an appropriate profile yourself and use dr_dbt_project(). The starter uses the dbt-duckdb profile format; it does not install dbt or adapters.

Examples

\dontshow{if (requireNamespace("duckdb", quietly = TRUE) && requireNamespace("yaml", quietly = TRUE)) withAutoprint(\{ # examplesIf}
root <- tempfile("dataraft-example-")
config <- dataraft.lake::dr_lake_config(
  catalog = dataraft.lake::dr_registry_duckdb(file.path(root, "lake.duckdb")),
  storage = dataraft.lake::dr_storage_local(file.path(root, "data")),
  landing = file.path(root, "landing"), backend = "duckdb"
)
project <- dr_dbt_init(file.path(root, "dbt"), config)
project
unlink(root, recursive = TRUE)
\dontshow{\}) # examplesIf}