I Built a Standalone .mdb to Parquet Exporter to Avoid Odbc Driver Hell

I’ve been stuck maintaining a legacy project that relies on massive .accdb files. I wasted two days trying to get the 64-bit ACE.OLEDB drivers to play nice with my Python environment without breaking other dependencies.

Access also kept segfaulting when I tried to export tables over 1GB to CSV, so I wrote a dedicated tool to handle the extraction.

The Tool:

Standalone: Doesn't require a local install of Office/Access. Streaming: Uses a stream reader so it doesn't OOM on large tables.

Parquet Support: Preserves data types better than CSV (and much smaller file size).

I threw in a basic SQL query window just to check data before dumping, but the main goal is just getting data out of Access and into a modern warehouse/dataframe as fast as possible.

It’s Windows-only for now (Access file locking is tricky on *nix), but let me know if it breaks on your specific schema.