dbt-polnor 0.1.0

dbt adapter for Polnor — run dbt models on a Polnor SQL warehouse (DuckDB reads + Spark Iceberg writes).

dbt-polnor

dbt adapter for Polnor — run your dbt models on a Polnor SQL warehouse with Iceberg tables, the same way you'd run them on Databricks or Snowflake.

Install

pip install dbt-polnor

This installs dbt-core>=1.7, polnor>=1.0, and the adapter glue.

Configure

In ~/.dbt/profiles.yml:

my_project:
  target: dev
  outputs:
    dev:
      type: polnor
      host: https://api.polnor.net
      token: "{{ env_var('POLNOR_TOKEN') }}"
      warehouse_id: wh-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
      workspace_id: ws-xxxxxxxx                # optional
      database: polnor                          # Spark catalog name (= Lakekeeper warehouse)
      schema: demo                              # default namespace for unqualified models
      threads: 4

Or interactively :

dbt init my_project
# select polnor as the database

Use

Anywhere your dbt project does :

dbt run
dbt test
dbt build
dbt docs generate

Models with materialized='table' get CREATE TABLE ... USING iceberg through the Spark sidecar; materialized='view' becomes a SQL view via Spark. Incremental models work via Spark's MERGE INTO.

What's supported (V0.1)

Feature	Status
materialized='table' (Iceberg)	✅
materialized='view'	✅
materialized='incremental'	🟡 append + merge (no delete+insert yet)
Sources + ref()	✅
Tests (generic + singular)	✅
Snapshots	🟡 V2
Seeds	✅
Macros	✅
dbt docs generate	✅
Python models (Spark)	⏳ V2

Limitations (V1)

• No transaction rollback — Polnor's engine is auto-commit per statement.
• No transient / temporary tables — Iceberg doesn't support them.
• dbt seed uses INSERT batches (1k rows / chunk) — for ≫100k rows prefer copying CSV to S3 then CREATE TABLE FROM directly.

Links

• Polnor: https://polnor.net
• Core SDK: https://pypi.org/project/polnor/
• SQLAlchemy dialect: https://pypi.org/project/polnor-sqlalchemy/
• Source / issues: https://github.com/polnor/polnor

License

Apache-2.0.