Getting started¶

From a bc2adls export to running staging tables in four steps.

1. Export from Business Central¶

Set up bc2adls to export your 4PS Construct tables to a Unity Catalog volume. Alongside the CSV deltas it writes CDM json files (deltas.manifest.cdm.json plus one <Table>-<id>.cdm.json per table) — those are the generator's input.

2. Download the CDM metadata¶

Install the generator (or run it from a checkout — see the generator CLI):

uv tool install dbt-4ps-generator

Authenticate via the Databricks SDK's default chain (DATABRICKS_HOST / DATABRICKS_TOKEN, or a ~/.databrickscfg profile), then:

dbt-4ps-generator download \
  --volume-path /Volumes/<catalog>/<schema>/<volume> \
  --output-directory ./cdm

No Business Central at hand?

The repo ships a 3-table example export in _cdm/ so you can try everything without an environment. See CDM metadata.

3. Generate the models¶

Copy the example project as your starting point, then generate your models into it — this replaces the sample models with your full table set:

cp -r example_dbt_project my_4ps_dbt
dbt-4ps-generator generate \
  --manifest ./cdm/deltas.manifest.cdm.json \
  --output-directory my_4ps_dbt/models/staging/4ps

This writes one stg_4ps__<table>.sql per entity plus a _4ps__models.yaml with descriptions and primary-key uniqueness tests.

4. Configure and run dbt¶

Point the project at your environment and run it — see the example dbt project for the vars and environment variables:

cd my_4ps_dbt
export DBT_DATABRICKS_HOST=<workspace>.azuredatabricks.net
export DBT_DATABRICKS_HTTP_PATH=/sql/1.0/warehouses/<warehouse-id>
export DBT_DATABRICKS_CATALOG=<catalog>
export DBT_DATABRICKS_TOKEN=<token>

uv sync
uv run dbt deps --profiles-dir .
uv run dbt run --profiles-dir .
uv run dbt test --profiles-dir .

Each dbt run triggers an incremental refresh: new deltas are ingested, already-processed files are skipped.