I realize that GitHub integration (for the purpose of version control, not module import) is not currently supported by Datalore. That said, I’ve got a creative workaround that I think I can get make work if only I knew the location of the “Notebook” file(s) used to store all the sheets in which interactive development can occur. Any ideas where that is on the Docker container’s file system?
Same question is here.
Hi,
Unfortunately, notebooks are stored in the database in an internal format (not .ipynb).
Thank you Mikhail for the answer.
A follow up question: Are all notebook output data stored in the same database? Like pictures created from arrays(not read from a file), large dataframes, generated CSVs that were not saved into a file and etc.?
Files created by notebooks and attached data are stored in /opt/data/web-dav/0
folder. The path to the specific notebook is /opt/data/web-dav/0/prefix/userId/notebookId
, where prefix
is 2 first symbols of userId. userId
and notebookId
are what you can see in a URL when opening notebook, for example for URL https://datalore.jetbrains.com/notebook/DM3u6DOxMaeVm6AXnXeFhX/lbmSQeHcMSP6pqe75TTmrF/
the useId
is DM3u6DOxMaeVm6AXnXeFhX
, and the notebookId
is lbmSQeHcMSP6pqe75TTmrF
.
The notebooks are split into three parts: input (the content of input cells), output (the content of output cells), and blobs (the big pieces of output, like plots). Inputs and outputs are stored in the database. While blobs are stored on the disk on the /opt/data/blob
path. But, unfortunately, it is stored under internal ids, not userId
or notebookId
.
That makes it more clear about the storage solutions Datalore is using. Thank you.