tiozin 1.4.0

Tiozin, your friendly ETL framework

Tiozin

---

ETL shouldn't require 80 files, 50 YAMLs, and a PhD in complexity.

Tiozin brings it back to basics: Transform. Input. Output. Nothing more, nothing less.

A lightweight Python framework that makes data jobs declarative, testable, and actually enjoyable to write.

Quick Start

pip install tiozin

Option 1: Define a job in YAML

kind: Job
name: kinglear_word_count_job

org: tiozin
region: latam
domain: literature
product: shakespeare
model: kinglear
layer: refined

runner:
  kind: SparkRunner

inputs:
  - kind: SparkFileInput
    name: load_poems
    path: s3://{{org}}-{{domain}}-raw/{{product}}/{{model}}/date={{ DAY[-1] }}

transforms:
  - kind: SparkWordCountTransform
    name: word_count

outputs:
  - kind: SparkFileOutput
    name: save_word_counts
    path: s3://{{org}}-{{domain}}-{{layer}}/{{product}}/{{model}}/date={{ today }}

Run it:

tiozin run examples/jobs/shakespeare/kinglear_word_count_job.yaml

Option 2: Use Python directly

from tiozin import TiozinApp

app = TiozinApp()
app.run("examples/jobs/shakespeare/kinglear_word_count_job.yaml")

Done. No ceremony, no boilerplate.

Documentation

• Installation
• Jobs
• Runners
• Transforms, Inputs & Outputs
• Registries
• Plugins
• Testing

Philosophy

Your uncle's advice: Keep it simple, readable, and testable.

• Declarative – Define what, not how
• Pluggable – Swap runners, registries, plugins as needed
• Metadata – Built-in metadata integration
• Observable – Logs that help
• Testable – Mock anything, validate everything

No magic. No surprises. Just clean data pipelines.

Contributing

Contributions are welcome! See CONTRIBUTING.md for guidelines.

License

This project is licensed under the Mozilla Public License 2.0.