Closed t20100 closed 3 years ago
Does this fix also the issue when double-clicking on a file with +
in the name ?
No, only through H5Web()
.
BTW, only tested with jupyterlab v2.3.1, maybe v3 solves the issue?
BTW, only tested with jupyterlab v2.3.1, maybe v3 solves the issue?
No it doesn't.
I think we should try to find where the aggressive escaping of the URL takes place in the backend and try to circumvent it to tackle both contexts at once.
Right! But in the network inspector the URL looks to be already escaped (i.e., + replaced with space), so maybe the filename needs to be escaped in jupyterlab before being used.
But in the network inspector the URL looks to be already escaped (i.e., + replaced with space), so maybe the filename needs to be escaped in jupyterlab before being used.
That would be surprising as notebooks or other filetypes with +
in their names work fine. That makes me think that we are perhaps doing something wrong somewhere...
Indeed! It is supposed to work as any other file type, so the solution is maybe elsewhere...
This PR escapes the filename in
H5Web
so it is safely pass through the request.As it is, the file path displayed in h5web is also escaped so it is no really nice, but looks to work.
If you prefer to keep
_check_data
, I'll remove the 2nd commit.closes #45