README.md

slackfs

Browse and upload slack files directly from your filesystem.

How does it work?

This is a simple FUSE application that talks to Slack using API in the backend. Right now it runs in a single-threaded foreground mode.

Usage

$ pip install -r requirements.txt
$ export SLACKFS_TOKEN=xoxp-your-token
$ ./slackfs.py /path/to/mountpoint

It was tested with legacy tokens that can be generated at https://api.slack.com/legacy/custom-integrations/legacy-tokens.

Environment variables

• SLACKFS_TOKEN -- token for the underlying Slack client
• SLACKFS_LOG_LEVEL -- the usual Python logging level, defaults to ERROR

Filesystem layout

In the root of the filesystem there are directories that represent channels you are a member of. Inside a channel you can find files that have been shared to it. There is a following naming convention for files: SLACK-FILE-ID_ORIGNAL-FILE-NAME to avoid name overlapping. stat on a file reports its real creation time so it is possible to list files by it (e.g. ls -lat | head).

$ tree
.
├── some-channel
│   ├── FU0BQXXXX_Screenshot_2020-02-21_21-21-29.png
│   ├── FUCKUXXXX_somefile
│   └── FUCL2XXXX_Screenshot_2020-02-21_21-23-41.png
├── empty-channel
├── other-channel
│   ├── FJPC5XXXX_dh.png
│   ├── FJS36XXXX_gh.png
...

Current limitations

Only some of the FUSE callbacks are implemented to provide the basic functionality. More sophisticated are provided by the fusepy library and most likely return unconditional success or some garbage. There is hardly any error handling in slackfs which in turn cause syscalls to return errors to applications.

New directory creation is UB. Creating files in root of the filesystem is UB, as is modifying existing ones.

There is absolutely no slack api rate limit respecting implemented on slackfs side. Unless official slack python api library handles that itself, your key or even account might be temporarily blocked.

At the moment there will be at most some 200 channels visible and each of them will contain at most 200 files (both arbitrarily chosen by the slack's api).

Channels and files are fetched lazily only once, there is no way to invalidate cache other than remounting the whole filesystem. However, after a successful upload of a new file, it's local name will change to reflect the aformentioned naming convention.