Chilean organization — From 54 Scripts to an Engine of 11,474 Traceable Rules

The challenge

Processing customs files for seven countries depended on 54 standalone Python scripts, about 50,000 lines, one for each file model.

The solution

We designed a 16-stage engine with pandas and pandera. The 11,474 business rules live in PostgreSQL, each with the file and line it came from, and they cover 38 models. The engine is backed by 467 tests, strict typing with mypy, and continuous integration in Azure DevOps that audits traceability on every change.

Technologies used

• Python
• pandas
• PostgreSQL
• Azure DevOps

Results

• 11.474 Traceable rules — Each with its origin
• 467 Automated tests
• 38 File models covered