adding jupyterlab file paths for preview #925
Conversation
choldgraf
force-pushed
the
lab_nbgitpuller
branch
from
72a4a27
to
a98cfed
Compare
binderhub/main.py
Outdated
|
|
||
| # Check if we have a JupyterLab + file path, if so then use it for the filepath | ||
| urlpath = self.get_argument('urlpath', '').lstrip('/') | ||
| if 'lab/tree/' in urlpath: |
There was a problem hiding this comment.
| if 'lab/tree/' in urlpath: | |
| if urlpath.startswith("lab/tree"): |
would this be an improvement/less prone to matching accidentally?
There was a problem hiding this comment.
I was originally doing it this way to allow for workspaces in the url too, but upon looking it seems like that'd actually be a more complicated rule. What if instead we did something like:
if urlpath.startswith("lab") and `/tree/` in urlpath:
and then do the split on /tree/ and take whatever is to the right of the first instance of /tree/?
There was a problem hiding this comment.
this is now in the latest push, it seemed to work for me locally (both regular lab URLs and workspace URLs)
There was a problem hiding this comment.
What different kinds of URLs could we get here? I am not really familiar with what Jupyter Lab does/what workspaces are :-/
There was a problem hiding this comment.
See the link I sent above, that's what I was using to figure out what urls to expect
|
Merge? If you've tested it locally I'd be happy to press merge and see what happens. Investigating and learning about URL schemes and how they could go wrong seems like a high effort thing compared to the worst that could happen if we just deploy it. |
|
I'm +1 on merging, the only thing that can go wrong is that it doesn't work, which would be our current situation :-) |
closes #645
I think that this should add support for JupyterLab paths in the nbviewer preview.
It is pretty simple, basically just adds a few lines to:
urlpathis provided in the URLlab/tree/exists in theurlpathfilepathto whatever is on the right side oflab/tree/.I think this should work for at least vanilla jupyterlab paths. It might break in unpredicted ways if people start using jupyterlab but also other kinds of URL patterns, but we can beef up this function when we get to those situations.