python-airflow-to-orchestra

Python: Airflow → Orchestra Conversion

Overview

Airflow's PythonOperator runs a callable inline in the worker — the code lives right there in the DAG file, not in some separate repo the DAG checks out at runtime. Orchestra's Python integration (Execute Script) has two modes: source: INLINE runs code pasted directly into the task's parameters.code, and source: GIT runs a file checked out from a Git repo via parameters.command. Default to INLINE — it's the direct match for what Airflow already does (code inline in the DAG), needs no Git repo or connection wiring, and skips a whole conversion step. Only reach for GIT when the callable itself checks out and runs a script that already lives in a separate repo (rare for PythonOperator — more of a BashOperator pattern).

Parameter Mapping

Orchestra YAML Structure

version: v1
name: <pipeline-name>
pipeline:
  <stage-uuid>:
    tasks:
      <task-uuid>:
        integration: PYTHON
        integration_job: PYTHON_EXECUTE_SCRIPT
        name: <task_id value from Airflow>
        connection: null                          # usually null for INLINE — no repo/creds needed unless the code itself needs a specific connection's secrets
        parameters:
          source: INLINE                           # default — code already lives in the DAG, no Git repo involved
          code: |
            <callable body, copied verbatim>
          build_command: 'pip install pandas'      # optional — only for non-stdlib imports
          python_version: '3.12'
        depends_on: []
        condition: null
        tags: []