First of all, thank you for your interest in the project! We really appreciate the fact that you're about to take some time to read this and hack on the project.
This document will guide you through common operations such as:
- Setup your development environment
- Working on your first issue
- Writing unit tests to validate your work
- Submit your work
The next sections of this document include a full installation guide to help you setup a local, development version of Funkwhale. If you only want to fix small things on the front-end, and don't want to manage a full development environment, there is another way.
As the front-end can work with any Funkwhale server, you can work with the front-end only, and make it talk with an existing instance (like the demo one, or you own instance, if you have one).
If even that is too much for you, you can also make your changes without any development environment, and open a merge request. We will be able to review your work easily by spawning automatically a live version of your changes, thanks to Gitlab Review apps.
Clone the repository:
git clone ssh://[email protected]/funkwhale/funkwhale.git cd funkwhale cd front
Install the dependencies:
yarn install
Compile the translations:
yarn i18n-compile
Launch the development server:
# this will serve the front-end on http://localhost:8000/front/ VUE_PORT=8000 yarn serve
Make the front-end talk with an existing server (like https://demo.funkwhale.audio or https://open.audio), by clicking on the corresponding link in the footer
Start hacking!
If you want to fix a bug or implement a feature, you'll need to run a local, development copy of funkwhale.
We provide a docker based development environment, which should be both easy to setup and work similarly regardless of your development machine setup.
Instructions for bare-metal setup will come in the future (Merge requests are welcome).
This is already cover in the relevant documentations:
Visit https://dev.funkwhale.audio/funkwhale/funkwhale and clone the repository using SSH or HTTPS. Example using SSH:
git clone ssh://[email protected]/funkwhale/funkwhale.git cd funkwhale
Note
As of January 2020, the SSH fingerprints of our Gitlab server are the following:
$ ssh-keyscan dev.funkwhale.audio | ssh-keygen -lf - # dev.funkwhale.audio:22 SSH-2.0-OpenSSH_7.4p1 Debian-10+deb9u6 # dev.funkwhale.audio:22 SSH-2.0-OpenSSH_7.4p1 Debian-10+deb9u6 # dev.funkwhale.audio:22 SSH-2.0-OpenSSH_7.4p1 Debian-10+deb9u6 2048 SHA256:WEZ546nkMhB9yV9lyDZZcEeN/IfriyhU8+mj7Cz/+sU dev.funkwhale.audio (RSA) 256 SHA256:dEhAo+1ImjC98hSqVdnkwVleheCulV8xIsV1eKUcig0 dev.funkwhale.audio (ECDSA) 256 SHA256:/AxZwOSP74hlNKCHzmu9Trlp9zVGTrsJOV+zet1hYyQ dev.funkwhale.audio (ED25519)
Next release development occurs on the "develop" branch, and releases are made on the "master" branch. Therefore, when submitting Merge Requests, ensure you are merging on the develop branch.
In development, we use the docker-compose file named dev.yml, and this is why all our docker-compose commands will look like this:
docker-compose -f dev.yml logs
If you do not want to add the -f dev.yml snippet every time, you can run this command before starting your work:
export COMPOSE_FILE=dev.yml
We provide a working .env.dev configuration file that is suitable for development. However, to enable customization on your machine, you should also create a .env file that will hold your personal environment variables (those will not be commited to the project).
Create it like this:
touch .env
Create the federation network:
docker network create federation
On your initial clone, or if there have been some changes in the app dependencies, you will have to rebuild your containers. This is done via the following command:
docker-compose -f dev.yml build
To setup funkwhale's database schema, run this:
docker-compose -f dev.yml run --rm api python manage.py migrate
This will create all the tables needed for the API to run properly. You will also need to run this whenever changes are made on the database schema.
It is safe to run this command multiple times, so you can run it whenever you fetch develop.
You'll need at least an admin user and some artists/tracks/albums to work locally.
Create an admin user with the following command:
docker-compose -f dev.yml run --rm api python manage.py createsuperuser
Injecting fake data is done by running the following script:
artists=25 command="from funkwhale_api.music import fake_data; fake_data.create_data($artists)" echo $command | docker-compose -f dev.yml run --rm api python manage.py shell -i python
The previous command will create 25 artists with random albums, tracks and metadata.
Before the first Funkwhale launch, it is required to run this:
docker-compose -f dev.yml run --rm front yarn run i18n-compile
Then you can run everything with:
docker-compose -f dev.yml up front api nginx celeryworker
This will launch all services, and output the logs in your current terminal window.
If you prefer to launch them in the background instead, use the -d flag, and access the logs when you need it via docker-compose -f dev.yml logs --tail=50 --follow.
Once everything is up, you can access the various funkwhale's components:
- The Vue webapp, on http://localhost:8000
- The API, on http://localhost:8000/api/v1/
- The django admin, on http://localhost:800/api/admin/
Once you're down with your work, you can stop running containers, if any, with:
docker-compose -f dev.yml stop
If you want to wipe your development environment completely (e.g. if you want to start over from scratch), just run:
docker-compose -f dev.yml down -v
This will wipe your containers and data, so please be careful before running it.
You can keep your data by removing the -v flag.
This is not needed unless you need to work on federation-related features.
To achieve that, you'll need:
- to update your dns resolver to resolve all your .dev hostnames locally
- a reverse proxy (such as traefik) to catch those .dev requests and and with https certificate
- two instances (or more) running locally, following the regular dev setup
If you use dnsmasq, this is as simple as doing:
echo "address=/test/172.17.0.1" | sudo tee /etc/dnsmasq.d/test.conf sudo systemctl restart dnsmasq
If you use NetworkManager with dnsmasq integration, use this instead:
echo "address=/test/172.17.0.1" | sudo tee /etc/NetworkManager/dnsmasq.d/test.conf sudo systemctl restart NetworkManager
Simply copy bundled certificates:
sudo cp docker/ssl/test.crt /usr/local/share/ca-certificates/ sudo update-ca-certificates
This certificate is a wildcard for *.funkwhale.test
Launch the traefik proxy:
docker-compose -f docker/traefik.yml up -d
Then, in separate terminals, you can setup as many different instances as you need:
export COMPOSE_PROJECT_NAME=node2 export VUE_PORT=1234 # this has to be unique for each instance docker-compose -f dev.yml run --rm api python manage.py migrate docker-compose -f dev.yml run --rm api python manage.py createsuperuser docker-compose -f dev.yml up nginx api front nginx api celeryworker
Note that by default, if you don't export the COMPOSE_PROJECT_NAME, we will default to node1 as the name of your instance.
Assuming your project name is node1, your server will be reachable
at https://node1.funkwhale.test/. Not that you'll have to trust
the SSL Certificate as it's self signed.
When working on federation with traefik, ensure you have this in your env:
# This will ensure we don't bind any port on the host, and thus enable # multiple instances of funkwhale to be spawned concurrently. VUE_PORT_BINDING= # This disable certificate verification EXTERNAL_REQUESTS_VERIFY_SSL=false # this ensure you don't have incorrect urls pointing to http resources FUNKWHALE_PROTOCOL=https # Disable host ports binding for the nginx container, as traefik is serving everything NGINX_PORTS_MAPPING=80
- Fork the project if you did not already or if you do not have access to the main repository
- Checkout the development branch and pull most recent changes:
git checkout develop && git pull - If working on an issue, assign yourself to the issue. Otherwise, consider open an issue before starting to work on something, especially for new features.
- Create a dedicated branch for your work
42-awesome-fix. It is good practice to prefix your branch name with the ID of the issue you are solving. - Work on your stuff
- Commit small, atomic changes to make it easier to review your contribution
- Add a changelog fragment to summarize your changes:
echo "Implemented awesome stuff (#42)" > changes/changelog.d/42.feature - Push your branch
- Create your merge request
- Take a step back and enjoy, we're really grateful you did all of this and took the time to contribute!
To ensure we have extensive and well-structured changelog, any significant work such as closing an issue must include a changelog fragment. Small changes may include a changelog fragment as well but this is not mandatory. If you're not sure about what to do, do not panic, open your merge request normally and we'll figure everything during the review ;)
Changelog fragments are text files that can contain one or multiple lines
that describe the changes occurring in a bunch of commits. Those files reside
in changes/changelog.d.
A typical fragment looks like that:
Fixed broken audio player on Chrome 42 for ogg files (#567)
If the work fixes one or more issues, the issue number should be included at the
end of the fragment ((#567) is the issue number in the previous example).
If your work is not related to a specific issue, use the merge request identifier instead, like this:
Fixed a typo in landing page copy (!342)
Fragment files should respect the following naming pattern: changes/changelog.d/<name>.<category>.
Name can be anything describing your work, or simply the identifier of the issue number you are fixing.
Category can be one of:
feature: for new featuresenhancement: for enhancements on existing featuresbugfix: for bugfixesdoc: for documentationi18n: for internationalization-related workmisc: for anything else
Here is a shortcut you can use/adapt to easily create new fragments from command-line:
issue="42"
content="Fixed an overflowing issue on small resolutions (#$issue)"
category="bugfix"
echo $content > changes/changelog.d/$issue.$categoryYou can of course create fragments by hand in your text editor, or from Gitlab's interface as well.
We're using https://github.com/Polyconseil/vue-gettext to manage i18n in the project.
When working on the front-end, any end-user string should be marked as a translatable string, with the proper context, as described below.
Translations in HTML use the <translate> tag:
<template>
<div>
<h1><translate translate-context="Content/Profile/Header">User profile</translate></h1>
<p>
<translate
translate-context="Content/Profile/Paragraph"
:translate-params="{username: 'alice'}">
You are logged in as %{ username }
</translate>
</p>
<p>
<translate
translate-context="Content/Profile/Paragraph"
translate-plural="You have %{ count } new messages, that's a lot!"
:translate-n="unreadMessagesCount"
:translate-params="{count: unreadMessagesCount}">
You have 1 new message
</translate>
</p>
</div>
</template>
Anything between the <translate> and </translate> delimiters will be considered as a translatable string.
You can use variables in the translated string via the :translate-params="{var: 'value'}" directive, and reference them like this:
val value is %{ value }.
For pluralization, you need to use translate-params in conjunction with translate-plural and translate-n:
translate-paramsshould contain the variable you're using for pluralization (which is usually shown to the user)translate-nshould match the same variable- The
<translate>delimiters contain the non-pluralized version of your string - The
translate-pluraldirective contains the pluralized version of your string
Translations in javascript work by calling the this.$*gettext functions:
export default {
computed: {
strings () {
let tracksCount = 42
let playButton = this.$pgettext('Sidebar/Player/Button/Verb, Short', 'Play')
let loginMessage = this.$pgettext('*/Login/Message', 'Welcome back %{ username }')
let addedMessage = this.$npgettext('*/Player/Message', 'One track was queued', '%{ count } tracks were queued', tracksCount)
console.log(this.$gettextInterpolate(addedMessage, {count: tracksCount}))
console.log(this.$gettextInterpolate(loginMessage, {username: 'alice'}))
}
}
}
The first argument of the $pgettext and $npgettext functions is the string context.
Translation contexts provided via the translate-context directive and the $pgettext and $npgettext are never shown to end users
but visible by Funkwhale translators. They help translators where and how the strings are used,
especially with short or ambiguous strings, like May, which can refer a month or a verb.
While we could in theory use free form context, like This string is inside a button, in the main page, and is a call to action,
Funkwhale use a hierarchical structure to write contexts and keep them short and consistents accross the app. The previous context,
rewritten correctly would be: Content/Home/Button/Call to action.
This hierarchical structure is made of several parts:
- The location part, which is required and refers to the big blocks found in Funkwhale UI where the translated string is displayed:
ContentFooterHeadMenuPopupSidebar*for strings that are not tied to a specific location
- The feature part, which is required, and refers to the feature associated with the translated string:
AboutAdminAlbumArtistEmbedHomeLoginLibraryModerationPlayerPlaylistProfileFavoritesNotificationsRadioSearchSettingsSignupTrackQueue*for strings that are not tied to a specific feature
- The component part, which is required and refers to the type of element that contain the string:
ButtonCardCheckboxDropdownError messageFormHeaderHelp textHidden textIconInputImageLabelLinkList itemMenuMessageParagraphPlaceholderTabTableTitleTooltip*for strings that are not tied to a specific component
- The detail part, which is optional and refers to the contents of the string itself, such as:
AdjectiveCall to actionNounShortUnitVerb
Here are a few examples of valid context hierarchies:
Sidebar/Player/ButtonContent/Home/Button/Call to actionFooter/*/Help text*/*/*/Verb, ShortPopup/Playlist/ButtonContent/Admin/Table.Label/Short, Noun (Value is a date)
It's possible to nest multiple component parts to reach a higher level of detail. The component parts are then separated by a dot:
Sidebar/Queue/Tab.TitleContent/*/Button.TitleContent/*/Table.HeaderFooter/*/List item.LinkContent/*/Form.Help text
If you want to ensure your translatable strings are correctly marked for translation, you can try to extract them.
Extraction is done by calling yarn run i18n-extract, which
will pull all the strings from source files and put them in a PO files.
You can then inspect the PO files to ensure everything is fine (but don't commit them, it's not needed).
tree api -L 2 -d
api
├── config # configuration directory (settings, urls, wsgi server)
│ └── settings # Django settings files
├── funkwhale_api # project directory, all funkwhale logic is here
├── requirements # python requirements files
└── tests # test files, matches the structure of the funkwhale_api directoryNote
Unless trivial, API contributions must include unittests to ensure your fix or feature is working as expected and won't break in the future
To run the pytest test suite, use the following command:
docker-compose -f dev.yml run --rm api pytest
This is regular pytest, so you can use any arguments/options that pytest usually accept:
# get some help docker-compose -f dev.yml run --rm api pytest -h # Stop on first failure docker-compose -f dev.yml run --rm api pytest -x # Run a specific test file docker-compose -f dev.yml run --rm api pytest tests/music/test_models.py
Although teaching you how to write unit tests is outside of the scope of this document, you'll find below a collection of tips, snippets and resources you can use if you want to learn on that subject.
Useful links:
- A quick introduction to unit test writing with pytest
- A complete guide to Test-Driven Development (although not using Pytest)
- pytest: documentation of our testing engine and runner
- pytest-mock: project page of our mocking engine
- factory-boy: documentation of factory-boy, which we use to easily generate fake objects and data
Recommendations:
- Test files must target a module and mimic
funkwhale_apidirectory structure: if you're writing tests forfunkwhale_api/myapp/views.py, you should put thoses tests intests/myapp/test_views.py - Tests should be small and test one thing. If you need to test multiple things, write multiple tests.
We provide a lot of utils and fixtures to make the process of writing tests as painless as possible. You'll find some usage examples below.
Use factories to create arbitrary objects:
# funkwhale_api/myapp/users.py
def downgrade_user(user):
"""
A simple function that remove superuser status from users
and return True if user was actually downgraded
"""
downgraded = user.is_superuser
user.is_superuser = False
user.save()
return downgraded
# tests/myapp/test_users.py
from funkwhale_api.myapp import users
def test_downgrade_superuser(factories):
user = factories['users.User'](is_superuser=True)
downgraded = users.downgrade_user(user)
assert downgraded is True
assert user.is_superuser is False
def test_downgrade_normal_user_does_nothing(factories):
user = factories['users.User'](is_superuser=False)
downgraded = something.downgrade_user(user)
assert downgraded is False
assert user.is_superuser is FalseNote
We offer factories for almost if not all models. Factories are located
in a factories.py file inside each app.
Mocking: mocking is the process of faking some logic in our code. This is useful when testing components that depend on each other:
# funkwhale_api/myapp/notifications.py
def notify(email, message):
"""
A function that sends an email to the given recipient
with the given message
"""
# our email sending logic here
# ...
# funkwhale_api/myapp/users.py
from . import notifications
def downgrade_user(user):
"""
A simple function that remove superuser status from users
and return True if user was actually downgraded
"""
downgraded = user.is_superuser
user.is_superuser = False
user.save()
if downgraded:
notifications.notify(user.email, 'You have been downgraded!')
return downgraded
# tests/myapp/test_users.py
def test_downgrade_superuser_sends_email(factories, mocker):
"""
Your downgrade logic is already tested, however, we want to ensure
an email is sent when user is downgraded, but we don't have any email
server available in our testing environment. Thus, we need to mock
the email sending process.
"""
mocked_notify = mocker.patch('funkwhale_api.myapp.notifications.notify')
user = factories['users.User'](is_superuser=True)
users.downgrade_user(user)
# here, we ensure our notify function was called with proper arguments
mocked_notify.assert_called_once_with(user.email, 'You have been downgraded')
def test_downgrade_not_superuser_skips_email(factories, mocker):
mocked_notify = mocker.patch('funkwhale_api.myapp.notifications.notify')
user = factories['users.User'](is_superuser=True)
users.downgrade_user(user)
# here, we ensure no email was sent
mocked_notify.assert_not_called()Views: you can find some readable views tests in file: api/tests/users/test_views.py
Note
A complete list of available-fixtures is available by running
docker-compose -f dev.yml run --rm api pytest --fixtures
To run the front-end test suite, use the following command:
docker-compose -f dev.yml run --rm front yarn test:unit
We also support a "watch and test" mode were we continually relaunch tests when changes are recorded on the file system:
docker-compose -f dev.yml run --rm front yarn test:unit -w
The latter is especially useful when you are debugging failing tests.
Note
The front-end test suite coverage is still pretty low
To make a new 3.4 release:
# setup export NEXT_RELEASE=3.4 # replace with the next release number export PREVIOUS_RELEASE=3.3 # replace with the previous release number # ensure you have an up-to-date repo git checkout develop # use master if you're doing a hotfix release git pull # compile changelog towncrier --version $NEXT_RELEASE --yes # polish changelog # - update the date # - look for typos # - add list of contributors via `python3 scripts/get-contributions-stats.py develop $PREVIOUS_RELEASE` nano CHANGELOG # Set the `__version__` variable to $NEXT_RELEASE nano nano api/funkwhale_api/__init__.py # commit git add . git commit -m "Version bump and changelog for $NEXT_RELEASE" # tag git tag $NEXT_RELEASE # publish git push --tags && git push # if you're doing a hotfix release from master git checkout develop && git merge master && git push # if you're doing a non-hotfix release, and a real release (not a real release) from develop git checkout master && git merge develop && git push
Then, visit https://dev.funkwhale.audio/funkwhale/funkwhale/-/tags, copy-paste the changelog on the corresponding tag, and announce the good news ;)