Usage example

[vadimkantorov]

Hi!

I created a sample file on my Google Drive hello.txt and made it accessible for Anyone with the link: https://drive.google.com/file/d/1I_bokg0Fejy7dVxNkZcJD4gCagawLIVY/view?usp=drive_link

E.g. I have an fsspec-using program catfsspec.py:

import sys, fsspec
with fsspec.open(sys.argv[1], 'rt') as f:
    print(f.read())

# python catfsspec.py file://hi.txt
# planet

Does installation of PyDrive2 auto-register a protocol with fsspec automatically (like sshfs does: https://pypi.org/project/sshfs/)? Is it possible to access my sample Google Drive file

Ideally, I think being able to do the following would make a lot of sense:

# should not be needed for accessing anon-acessible files: export FSSPEC_GDRIVE_SECRETS=/path/to/client_secrets.json

python catfsspec.py gdrivefs://1I_bokg0Fejy7dVxNkZcJD4gCagawLIVY
# world

[martindurant]

Note that neither this repo nor pydrive2 are particularly complete, although the latter may have progressed since I checked on it last. The auth and API with gdrive is not particularly well documented and hard to test against.

---

If you have a public HTTP link, you don't need gdrivefs at all.

Does installation of PyDrive2 auto-register a protocol with fsspec automatically

Yes, in fact fsspec already knows about gdrivefs.

should not be needed for accessing anon-acessible files

I think you need credentials to poke the GDrive API at all, even to get the URL of public files.

[vadimkantorov]

If you have a public HTTP link, you don't need gdrivefs at all.

It can be tricky to access these (these needed at some point the cookies etc) sometimes, so I thought that gdrivefs would be more robust

Yes, in fact fsspec already knows about gdrivefs.

What is the protocol? Are there anywhere complete examples of accessing files via this protocol?

[vadimkantorov]

Is there a separate script / example of getting an API key for our Google Workspace's Drive, so that I can paste it on my ssh machine?

[vadimkantorov]

I managed to make it run with python catfsspec.py gdrive://hello.txt, but there are a few problems with it:

1. fsspec.open seems to not cache the auth token, and it's requested every time I run it. How can I make it auto-cache and use the token if it's present / not expired? and maybe request it again if it's not good?
2. How can I use the file ID in the url? It is useful to be able to address via global file ID to keep the access code even if the file was moved around. It's the benefit of the global file IDs and of the drive URLs which stay the same even if file is moved around. It would be good to have gdrivefs also have this property

[vadimkantorov]

@martindurant It would be very useful to access the file via its ID, because:

• it is persistent even if the file is renamed (in addition to being persistent when the file is moved around)
• the file ID is present in the URL, so one can easily transform https://drive.google.com/file/d/1I_bokg0Fejy7dVxNkZcJD4gCagawLIVY/view?usp=drive_link into gdrive://1I_bokg0Fejy7dVxNkZcJD4gCagawLIVY without having to open the file in the browser, find its full location and the file name

[martindurant]

OK, you have a fair point about both auth and potential renaming. So, to backtrack, we would need either to recognise filenames that are in fact file IDs, or have a special way to express the URL for these.

[vadimkantorov]

maybe use some prefix like hash-sign gdrivefs://#1I_bokg0Fejy7dVxNkZcJD4gCagawLIVY or gdrivefs://hello.txt#1I_bokg0Fejy7dVxNkZcJD4gCagawLIVY (to override the true file name)? (or use some other character which is forbidden to appear in file names / google drive file names)

[martindurant]

That could be OK, yes.

[vadimkantorov]

One problem is whether we also want to specify the drive ID in this URL? If so, it could be extended as /myfolder/hello.txt#1I_bokg0Fejy7dVxNkZcJD4gCagawLIVY@MyDrive#driveid with the at-sign