FAQ

Why read everything as string first?
Pandas’ default type inference can corrupt data: 00742 becomes 742, and dates can be parsed in the wrong order. Reading as string and then applying our own detection keeps control and avoids silent errors.

Why not use dateutil.parser or dateparser?
They are too permissive (e.g. parsing "hello" as a date) and too slow for batch column analysis. We use strptime with explicit format strings for speed and reliability.

What if my file has no header?
The dialect detector tries to infer header presence. You can also pass pandas options via **pandas_kwargs (e.g. header=None and set column names yourself).

Can I use csvmedic with Excel (.xlsx)?
Install the optional dependency: pip install csvmedic[excel]. Support for .xlsx can be added in a future release.