Add test for deposit of files with unicode names

[mih]

We know that we need directoryLabel mangeling. We do not know about files yet.

[mih]

Needs a solution to #232 first.

[christian-monch]

File names have certain restrictions in dataverse. The restrictions are described in issue #209. PR #211 adds dataverse-compatible encoding of file names.

With PR #240 all file names in a datalad dataset will be mapped on a unique dataverse-compatible name.

A remaining non-technical issue is that file names which contain Unicode characters will not be easily readable, which leads to the idea of using the Unidecode package to create dataverse-names that are "kind of readable", but not guaranteed to be unique, i.e. different datalad dataset file names might be mapped onto the same dataverse name.

It should also be mentioned that directory and file name restrictions due exclude a large number of ASCII characters from being used. These characters still need to be replaced. That can either be done "properly, but nasty" with PR #211 (which I would opt for), or "visually appealing, but faulty" by replacing all characters that are not allowed in dataverse-names with a "placeholder" character, for example with _.

[mih]

All referenced PRs are merged now.

test_dataset.py seems to be the most appropriate layer for such a test.

[mih]

Will not be ready in time. Reassigning milestone.

[pdurbin]

Hi! We just opened this related issue:

• Feature Request/Idea: Review of used regex IQSS/dataverse#10607

It's not prioritized or anything, but we do want to let you know that we hear you! ❤️