A command-line for mite that gets out of your way!
Do you track time in mite, but wish you could control the clock with a few keystrokes from the nearest terminal window? Then give miteclock a try!
The goal of this program is to address aspects of using mite that I, as a terminal and keyboard user, have found inconvenient in daily use:
- Having to search for the mite browser tab or opening a new one when I always have terminal windows handy.
- Having to switch between mouse (for projects/services) and keyboard (for notes) to create an new entry.
- Surprising narrow-by-typing behavior in the projects/services menus.
- Inflexible support for templates, no ability to compose an entry from pre-defined parts.
See here for more context.
This program is tested with Python versions 3.7-3.11. To minimize "works on my machine"
problems, the dependencies have been pinned to specific versions. Therefore it's
strongly recommended to install it in a dedicated virtualenv and then add a symbolic
link to the executable somewhere in your PATH. An even better option is to use the
pipx wrapper which automatically takes care of
these two steps.
Install with a standard pip command:
pip install miteclockNow you should be able to run the following in your terminal:
mThe first time you run it, it will prompt you for your account information and create a
TOML configuration file in your home directory
named ~/.config/miteclock/config.toml. Then it will show you the help message for the
program. This message and the help for sub-commands should provide enough general
documentation, so the rest of this README is a tutorial to get you started.
There are only two commands to interact with the timer: m start starts a clock,
m stop stops it. That's it, that simple. m stop is self-explanatory (run it with
--help), so here we focus on m start.
Let's say your mite account has the following projects:
- ACME – Self-healing container deployments
- OCP: ED-209
- CHAZ 2020
In these projects you perform the following services (Dienstleistungen):
- Development
- Regular Maintenance
- Irregular Maintenance
- QA
- Backend QA
From your experience with the mite webapp, you know that a time entry has the following three fields:
- project
- service
- note
What if instead of selecting the project and the service from a drop-down you did so by pressing just one key? This is much faster, especially if you have more realistic (i.e. larger) sets of projects and services that you'd have to sift through with the drop-down.
These keys are known as shortcuts and you can define them in your configuration file.
For our example here, let's create a few mappings from keys to project/service names. We
open our ~/.config/miteclock/config.toml in a text editor and add the following in the
[shortcuts] table:
[shortcuts]
# Shortcuts for projects.
a = "ACME -- Self-healing container deployments"
o = "OCP: ED-209"
h = "CHAZ 2020"
t = "Team-Internal"
# Shortcuts for services.
d = "Development"
r = "Regular Maintenance"
i = "Irregular Maintenance"
c = "Communication/Coordination"Now we can add an activity and start the clock for it with this one command:
m start a d 'writing some code'The first two arguments to start are expanded into "ACME – Self-healing
container deployments" and "Development" respectively. The last argument is the note. We
put it in quotes so that it is treated as a single argument.
Note that order matters for the expanded items. It must be like in the webapp:
- project
- service
- note
If you want to leave any field unspecified, enter an empty string for it. For example if you're working for "ACME" but haven't narrowed your work down to an exact service or task, run this:
m start a '' ''Leaving notes empty and filling them out later is so common, that an empty note can be completely omitted. We can shorten the command above to:
m start a ''Another way to avoid writing out a note is to put it into a shortcut definition. This works well for recurring meetings or tasks where the note stays the same. Let's add some shortcuts that describe recurring activities for many programmers:
daily = ["t", "c", "daily stand-up"]
retro = ["t", "c", "retrospective"]
server = ['a', 'r', "regular server maintenance"]Notice how we used the shortcuts we had already defined to create new shortcuts? It's shortcuts all the way down!
These nested shortcuts can span any consecutive part of an activity definition. This is valid…
kickoff = ["c", "kickoff meeting for project"]… and can be used with all your projects, for example:
m start h kickoff # Tracks kickoff meeting for CHAZ 2020
m start o kickoff # Tracks kickoff meeting for OCP: ED-209This is also valid:
acmedev = ["a", "d"]This, however, is invalid:
invalid = ["a", "some ACME-related note"]Often you might have to stop the clock for some activity and then start it back up later.
If you have clocked in some entries for the day and run m start without any arguments,
you will be presented with a list of the activities you recorded for the day paired with
keys you can press to select one of the entries. Note that unlike in the mite webapp,
time entries are sorted by the time they were updated last not by the time when they
were created.
You can skip this menu by passing the -l flag (or --last if you like typing) which
automatically starts the last entry for which you had a clock running.
You can even run the same exact command a second time, e.g.
m start a d 'writing some code'
# ... some other commands...
m start a d 'writing some code'There is also m resume which is just an alias for m start -l.
m status will report the current status of the tracker: whether or not the clock is
running and for which entry, which entries have been created today.
m show shows you a list of shortcuts. m show projects shows the projects your
account has access to. m show services does the same for services. Note that the list
of projects can be long enough that you may want to save it to a file or filter it with
grep.
So far we have not discussed how we match shortcuts for projects and services to the
right entries in your mite account. Let's say we expand the shortcut
d = "Development". Based on its position we know it is a service, so we search all
services associated with your mite account. We collect those services that contain
the "Development" in them. If we collect exactly one service, we specify it in the
time entry. Otherwise we have to change the shortcut definition to address one of two
issues: either the shortcut matched multiple entries or it did not match any entries at
all.
This works fine for most cases, especially if we use the complete name of a service. It
fails, however, for the following service from our example: "QA". This is because its
full name is contained in another service "Backend QA", so we will always match both
with a definition like q = "QA".
What we would like is a way to say that service names must strictly match the shortcut definition. Exactly for these cases there is another way to specify shortcut expansions. The form we have been using so far is actually a shorthand for the following:
q = {pattern = "QA", match = "substring"}To switch to strict matching, we change the value of match:
q = {pattern = "QA", match = "strict"}The process of matching projects is the same with one addition. In mite it is possible for projects to have the same names and different customers. For instance, you may be building a prototype for two companies, so your projects would be:
- Prototype (ACME Inc.)
- Prototype (King Inc.)
In this case you could define the following shortcuts:
pa = {project = "Prototype", customer = "ACME"}
pk = {project = "Prototype", customer = "King"}If you need support for strict matches for either the project name or the customer name,
you can use all that was mentioned above, just place it in the project and customer
keys:
pa = {project = {pattern = "Prototype", match = "strict"}, customer = {pattern = "ACME", match = "strict"}}Expanding the contents of the project and customer keys is currently not supported.
In other words, this won't work:
a = "ACME"
pa = {project = "Prototype", customer = "a"}If you find a problem with the program, please open an issue here.
If you want to submit changes, fork this repo, create a branch in your fork that
contains your work, open a pull request against the master branch in this repo.
For local development, install the dependencies using poetry.
poetry install
poetry run pre-commit installPlease make sure to add tests for any code changes. Assuming the commands above succeeded, run this:
poetry run pytestYou can also use tox to test your changes against all supported Python versions:
poetry run toxThere already are almost half a dozen command-line interfaces in several languages (Ruby, JavaScript, Go, Python). There's even a PHP wrapper library. What is the need for yet another cli?
I find that all the existing interfaces provide both too much functionality and too little. They try to cover all possible tasks, exposing all the gory details of the underlying data along the way. If you regularly import and export time records or manage projects and services for an account, these tools can be very helpful.
The way most of us use mite though is to start and stop the clock for activities "on the go" throughout the day. This takes advantage of mite's built-in tracking capabilities. A lot of the activities are recurring, like check-in meetings with clients or team members. Moreover, most activities on a given day revolve around a handful of projects and services.
This program aims to reduce the book-keeping cost of specifying activities. It lets the user focus on their work while instructing mite to do what it does best: track time. We deliberately expose a simple interface and deal only in relevant concepts.
This project would not have been possible at all without the folks who run mite making their API accessible. Many thanks to them for that. I am also grateful to the people who wrote client libraries and cli tools based on the API. This provided context to my efforts and thus helped me define what I wanted to focus on.
MIT