Feedback after quick tests

[Pierlou]

Hi, thanks for this lib! It caught my interest because such cleaning methods could prove useful for an open data platform, such as data.gouv.fr. Quick feedback:

• you may want to add versions restrictions on install (for example I ran into an error using a quite old version of pandas, because of an unimplemented function back then)
• I didn't get the expected results testing the trimmer on quite ugly csv files (from this dataset and that one). We've found it tedious too to find good euristics to clean any csv. You can find many examples of badly structured csv files on the platform, if you want to make this lib even more agnostic
• as the input is a DataFrame and not directly the file-path, this lib could also be useful to clean excel (or other tabular) files, for instance from this dataset (files are really not suitable for any python analysis as it is)

For the record, we are maintaining an other csv-related lib: csv-detective, maybe there will be nice cross-overs in the future 🤞

[LucaCappelletti94]

Could you kindly provide a specific example of failure? Also, which version of Python are you using?

If you are using an old version of Python, you might have inadvertently installed an old version of this package, as it currently requires Python 3.9 exactly because the Pandas version it uses requires at least 3.9.

[Pierlou]

My env is in python 3.10 but I had pandas==1.5.3, which when simply using trimmer.trim() gave: AttributeError: 'DataFrame' object has no attribute 'map'. Updating pandas fix it as expected

[LucaCappelletti94]

Restricted pandas version requirements in commit 3560d252df84a5d9d6a1d0a7d8719647406ba57f