job

Job runner with logging

Installation

Install system-wide:

$ pip install shell-jobrunner

Install just for the current user:

$ pip install --user shell-jobrunner

Uninstallation

$ pip uninstall jobrunner

Examples

• Run sleep 5 in the background

  $ job sleep 5

• Run ls when the last job finishes and it passed (exit code 0)

  $ job -B. ls

• Run ls when last job finishes (pass / fail)

  $ job -b. ls

• Monitor job execution

  $ job -W
  Sat Aug 10, 2019 20:48:23  No jobs running, load: 0/0/0

• Retry a job

  $ job --retry ls

Query Examples

NOTE . is available as an alias to the most recently executed job (as in the Examples above).

• View recently executed job log file

  $ job ls
  $ view `job`   # Opens the output from ls using "view"

• View two most recently executed

  $ job echo 1
  $ job echo 2
  $ view `job -n0 -n1`

• Query by job name

  $ job echo foo
  $ job echo bar
  $ view `job -g foo`

• Show job info by name

  $ job ls
  $ job -s ls

Configuration

The default configuration file location is ~/.config/jobrc, but can be
overridden using the --rc-file option.

Sample rcfile:

[mail]
program = mail
# For notifications over chat applications (like Google Chat), use chatmail as
# your mail program instead. "chatmail" must be specified rather than a differently
# named link to the script, else some options provided to job (such as --rc-file)
# will not be passed through to it.
# program = chatmail
domain = example.com
[ui]
watch reminder = full|summary  # default=summary
[chatmail]
at all = all|none|no id # default=none
reuse threads = true|false # default true
[chatmail.google-chat-userhooks]
user1 = https://chat.googleapis.com/v1/spaces/...
[chatmail.google-chat-userids]
# Retrieve this using your browser inspector on an existing mention of this user.
# It should show up as "user/some_long_integer" somewhere in the span's metadata.
user1 = <long integer>

System Notifications (Systemd user service example)

If you want to enable notifications when jobs finish, one way to do this is to use the --notifier argument.

~/.config/systemd/user/job-notify.service:

[Unit]
Description=Jobrunner Notifier

[Service]
Type=simple
ExecStart=env job --notifier jsonNotify.py
RestartSec=30
Restart=always

[Install]
WantedBy=default.target

~/.local/bin/jsonNotify.py:

#!/usr/bin/env python3

from json import load
import subprocess
from sys import stdin

cmd = ["notify-send"]
data = load(stdin)
rc = data.get("rc", 0)
if rc != 0:
    cmd += ["--urgency=critical"]
cmd += [data["subject"], data["body"]]
subprocess.run(cmd)

Hacking

Primary workflow

It's highly recommend to work inside a virtualenv using pipenv.

Create new virtualenv and install an editable version of jobrunner:

pipenv --three install --dev
pipenv run pip install -e .

Autoformat the code and check linters:

pipenv run ./format.sh

Run tests:

pipenv run pytest

Run CI checks locally

This allows you to run something similar to the azure pipelines locally using docker.
It will use PIP_INDEX_URL and / or ~/.config/pip/pip.conf to configure a pypi mirror.
This will also update Pipfile*.lock.

./test-docker.py [--versions 2.7 3.7 3.8] [--upgrade] [--ignore-unclean]