girder / girder_worker_utils Goto Github PK

It would be nice if there were a recursive option for GirderUploadToFolder.
I was trying to use GirderUploadVolumePathToFolder from girder_worker, but I noticed that it is not recursive. It only uploads a single file or all of the contents of the top-level directory (no contents from subdirectories...).

There should be a way to deal with upload transforms that happen after long-running jobs, when it's possible that the user that started the job has ended their web session.

Easiest thing I can think of is to allow passing alternative credentials, like an API key, to the transform constructor.

Also, this behavior isn't documented and it should be. I would actually expect API token generation to be the default behavior. I'm inclined to call this a bug.

There are a number of job states to indicate downloading and uploading data that are used by girder_worker.run. We should consider implementing these in the transform classes.

Refs #13

This is the naming convention that ipython uses for example repr_html ...

Is there a plan to create a function similar to GirderFileId but for items ? A bit like in the docker plugin where an item is downloaded to the worker.

I have a use case where I need to download several items to a worker. Would you be interested if I propose in a PR:

• A function to download 1 item to a folder.
• A function to download several items to several folders or a folder with nested subfolders
  ?