- Fixed bug where encoding of file was not correctly detected if non-ASCII characters occurred after the first 2KB. The tool now inspects the first 2MB of content (as originally intended) and also upgrades ASCII to `latin-1` since ASCII is a complete subset of `latin-1` and using `latin-1` increases the chance of a successful import. [25](https://github.com/simonw/datasette-upload-csvs/issues/25)