daktari 0.0.95

Assist in setting up and maintaining developer environments

Daktari is a tool to help the initial setup and ongoing maintenance of developer environments. It runs a series of checks (for example, that required software is installed) and provides suggestions on how to fix the issue if the check fails.

Configuration

In the root of the project repository, create a .daktari.py configuration file listing the checks you want run. For example,

from daktari.checks.git import *

version = "0.0.95"
title = "My Project"

checks = [
    GitInstalled(),
    GitLfsInstalled(),
    GitLfsSetUpForUser(),
    GitLfsFilesDownloaded(),
    GitCryptInstalled(),
]

Then run daktari to diagnose your environment:

$ daktari
✅ [git.installed] Git is installed
✅ [git.lfs.installed] Git LFS is installed
✅ [git.lfs.setUpForUser] Git LFS is set up for the current user
✅ [git.lfs.filesDownloaded] Git LFS files have been downloaded
❌ [git.crypt.installed] git-crypt is not installed
┌─💡 Suggestion ─────────┐
│ brew install git-crypt │
└────────────────────────┘

Custom Check

You can write a custom check as a Python class within .daktari.py, and include it in your list of checks. Example of a check implementation:

class GitCryptInstalled(Check):
    name = "git.crypt.installed"
    depends_on = [GitInstalled]

    suggestions = {
        OS.OS_X: "<cmd>brew install git-crypt</cmd>",
        OS.UBUNTU: "<cmd>sudo apt install git-crypt</cmd>",
        OS.GENERIC: "Install git-crypt (https://www.agwa.name/projects/git-crypt/)",
    }

    def check(self):
        return self.verify(can_run_command("git crypt version"), "git-crypt is <not/> installed")

Testing Daktari changes locally

Having cloned the repo into ~/daktari, you can make use of PYTHONPATH to run daktari using your local changes.

To do this, navigate into a directory that has a .daktari.py (e.g. another repository intending to use your change) and run:

PYTHONPATH=~/daktari python3 -m daktari --debug

Release instructions

Daktari is continuously deployed via a github action - see release.yaml. In case of a need to manually release, the steps are:

bumpversion --verbose patch
python setup.py sdist bdist_wheel
twine check dist/*
twine upload dist/*