pyfabric-dev 0.4.0

Local development framework for Microsoft Fabric notebooks: source-of-truth Python, generated .Notebook artifacts, mocked Fabric APIs, parallel local testing.

pyfabric-dev

A development framework for Microsoft Fabric notebooks. Write plain Python modules in src/, generate Fabric .Notebook artifacts from them, and run + test everything locally with mocked Fabric APIs.

Status: v0.4.0. APIs are stabilizing but pre-1.0; minor versions may break.

Disclaimer: not affiliated with or endorsed by Microsoft.

Why

Fabric notebooks have two pain points:

1. They can't import Python modules. Code must be inlined into notebook cells, which fights every Python convention.
2. They're nearly impossible to test locally. mssparkutils and notebookutils only exist in the Fabric runtime.

pyfabric-dev solves both:

• A source-of-truth pattern: you write modules under src/; the generator compiles them into .Notebook/ artifacts (committed to git, ready for Fabric Git integration).
• A local execution layer: mocks mssparkutils / notebookutils, sets up PySpark + Delta with a per-checkout Derby metastore, and runs your notebooks and pipelines off-cloud.
• A parallel test runner that mirrors a Fabric pipeline's batching, so the same suite that passes locally is the one that runs in CI.

Install

Requires Python 3.13 or 3.14 and a JDK 17+ on $PATH (for PySpark).

pip install pyfabric-dev

CLIs

After installing, four console scripts are on your PATH:

Command	Purpose
pyfabric-generate	Compile src/ modules into Fabric .Notebook/ artifacts
pyfabric-run-notebook	Execute one notebook locally (resolves %run deps)
pyfabric-run-pipeline	Execute a pipeline-content.json DAG locally
pyfabric-test	Parallel pytest runner with isolated per-worker metastores

All four accept --project-root <dir> (default: CWD).

Five-minute walkthrough

git clone https://github.com/CashHero/pyfabric-dev.git
cd pyfabric-dev/examples/minimal-medallion

pyfabric-generate    --project-root .
pyfabric-test        --project-root . --config config/test_batches.json
pyfabric-run-pipeline --project-root . --dry-run sales_etl

examples/minimal-medallion/ is a working bronze/silver/gold pipeline over toy sales data. See its README.md for details.

Documentation

• Getting started
• Source-of-truth pattern — how src/ → .Notebook/ works
• Local execution — how the runner mocks Fabric APIs
• Testing — fixtures, parallelism, batching
• Hooks reference — customizing the runners for your project

Used in production by

• CashHero — the internal pipeline this framework was extracted from.

Contributing

See CONTRIBUTING.md. PRs welcome.

License

Apache 2.0. See NOTICE for attribution.