> All CSVs contains a single consistent encoding > All records contain a single consistent encoding > All fields contain a single consistent encoding What? H

> All CSVs contains a single consistent encoding > All records contain a single consistent encoding > All fields contain a single consistent encoding What? How are you expected to handle these cases? Generally you have to guess the encoding of a plain text file like a CSV. I’m fairly sure the common case is the entire file will be a consistent encoding. If you were to guess per-record or per-field I suspect it’s more likely you’d guess some records/fields wrong than encountering a file with varying encoding. I’d be interested to see some real world examples of CSVs with varying encoding and how existing software handles that.