GvoT est un logiciel libre pour organiser des votations à large échelle.

Table of content

• Give a try
• Installation
• Deployment
• Structure
• Development

Give a try

On a Debian-based host - running at least Debian Buster:

$ sudo apt install python3 virtualenv git make
$ git clone https://forge.cliss21.org/cliss21/gvot
$ cd gvot

$ make init
# A configuration file will be created interactively; you can uncomment:
#  ENV=development

#$ make test # optional
$ make serve

Then visit http://127.0.0.1:8000/ in your web browser.

Installation

Requirements

On a Debian-based host - running at least Debian Stretch, you will need the following packages:

• python3
• virtualenv
• make
• git (recommended for getting the source)
• python3-mysqldb (optional, in case of a MySQL / MariaDB database)
• python3-psycopg2 (optional, in case of a PostgreSQL database)

Quick start

It assumes that you already have the application source code locally - the best way is by cloning this repository - and that you are in this folder.

1. Define your local configuration in a file named config.env, which can be copied from config.env.example and edited to suits your needs.

   Depending on your environment, you will have to create your database and the user at first.

2. Run make init.

   Note that if there is no config.env file, it will be created interactively.

That's it! Your environment is now initialized with the application installed. To update it, once the source code is checked out, simply run make update.

You can also check that your application is well configured by running make check.

In a production setup, you will need to add cron jobs in order to flush mail queues. See below. If you don't want to be able to send hundreds of emails you might drop EMAIL_BACKEND from the settings and forget about it.

Deployment

Web application

NginX with uWSGI

Here is an example deployment using NGINX - as the Web server - and uWSGI - as the application server.

uWSGI

The uWSGI configuration doesn't require a special configuration, except that we are using Python 3 and a virtual environment. Note that if you serve the application on a sub-location, you will have to add route-run = fixpathinfo: to your uWSGI configuration (available since v2.0.11).

NGINX

In the server block of your NGINX configuration, add the following blocks and set the path to your application instance and to the uWSGI socket:

location / {
    include uwsgi_params;
    uwsgi_pass unix:<uwsgi_socket_path>;
}
location /media {
    alias <app_instance_path>/var/media;
}
location /static {
    alias <app_instance_path>/var/static;
    # Optional: don't log access to assets
    access_log off;
}
location = /favicon.ico {
    alias <app_instance_path>/var/static/favicon/favicon.ico;
    # Optional: don't log access to the favicon
    access_log off;
}

Apache with mod_wsgi

Here is an example deployment using Apache as the Web server.

The WSGI configuration doesn't require a special configuration, except that we are using Python 3 and a virtual environment. See e.g. Django's docs for more details.

    WSGIProcessGroup gvot
    WSGIDaemonProcess gvot python-home=<app_instance_path>/venv/ python-path=<app_instance_path>/

    WSGIScriptAlias / <app_instance_path>/gvot/wsgi.py process-group=gvot

    <Directory <app_instance_path>/gvot/>
        <Files wsgi.py>
            Require all granted
        </Files>
    </Directory>

    Alias /static/ <app_instance_path>/var/static/

    <Directory <app_instance_path>/var/static/>
        Require all granted
    </Directory>

    Alias /media/ <app_instance_path>/var/media/

    <Directory <app_instance_path>/var/media/>
        Require all granted
    </Directory>

    Alias /favicon.ico <app_instance_path>/var/static/favicon/favicon.ico

Emails

In a production setup, you will need to add cron jobs in order to flush mail queues:

* * * * *       (<app_instance_path>/venv/bin/python <app_instance_path>/manage.py send_mail        >> <app_instance_path>/var/log/cron_mail.log 2>&1)
0,20,40 * * * * (<app_instance_path>/venv/bin/python <app_instance_path>/manage.py retry_deferred   >> <app_instance_path>/var/log/cron_mail_deferred.log 2>&1)
0 0 * * *       (<app_instance_path>/venv/bin/python <app_instance_path>/manage.py purge_mail_log 7 >> <app_instance_path>/var/log/cron_mail_purge.log 2>&1)

See https://github.com/pinax/django-mailer/blob/master/docs/usage.rst#clear-queue-with-command-extensions for more on this.

Structure

Overview

All the application files - e.g. Django code including settings, templates and statics - are located into gvot/.

Two environments are defined - either for requirements and settings:

• development: for local application development and testing. It uses a SQLite3 database and enable debugging by default, add some useful settings and applications for development purpose - i.e. the django-debug-toolbar.
• production: for production. It checks that configuration is set and correct, try to optimize performances and enforce some settings - i.e. HTTPS related ones.

Local changes

You can override and extend statics and templates locally. This can be useful if you have to change the logo for a specific instance for example. For that, just put your files under the local/static/ and local/templates/ folders.

Regarding the statics, do not forget to collect them after that. Note also that the local/ folder is ignored by git.

Variable content

All the variable content - e.g. user-uploaded media, collected statics - are stored inside the var/ folder. It is also ignored by git as it's specific to each application installation.

So, you will have to configure your Web server to serve the var/media/ and var/static/ folders, which should point to /media/ and /static/, respectively.

Development

The easiest way to deploy a development environment is by using the Makefile.

Before running make init, ensure that you have either set ENV=development in the config.env file or have this environment variable. Note that you can still change this variable later and run make init again.

There is some additional rules when developing, which are mainly wrappers for manage.py. You can list all of them by running make help. Here are the main ones:

• make serve: run a development server
• make test: test the whole application
• make lint: check the Python code syntax

Assets

The assets - e.g. CSS, JavaScript, images, fonts - are generated using a Gulp-powered build system with these features:

• SCSS compilation and prefixing
• JavaScript module bundling with webpack
• Styleguide and components preview
• Built-in BrowserSync server
• Compression for production builds

The source files live in assets/, and the styleguide in styleguide/.

Requirements

You will need to have npm installed on your system. If you are running Debian, do not rely on the npm package which is either outdated or removed - starting from Debian Stretch. Instead, here is a way to install the last version as a regular user:

1. Ensure that you have the following Debian packages installed, from at least stretch-backports:

   • nodejs
   • node-rimraf

2. Set the npm's installation prefix as an environment variable:

     $ export npm_config_prefix=~/.node_modules
   

3. Retrieve and execute the last npm's installation script:

     $ curl -L https://www.npmjs.com/install.sh | sh
   

4. Add the npm's binary folder to your environment variables:

   $ export PATH="${HOME}/.node_modules/bin:${PATH}"
   

   In order to keep those environment variables the next time you will log in, you can append the following lines to the end of your ~/.profile file:

   if [ -d "${HOME}/.node_modules/bin" ] ; then
       PATH="${HOME}/.node_modules/bin:${PATH}"
       export npm_config_prefix=~/.node_modules
   fi
   

5. That's it! You can check that npm is now installed by running the following:

   $ npm --version
   

Usage

Start by installing the application dependencies - which are defined in package.json - by running: npm install.

The following tasks are then available:

• npm run build: build all the assets for development and production use, and put them in the static folder - e.g gvot/static.
• npm run styleguide: run a server with the styleguide and watch for file changes.
• npm run serve: run a proxy server to the app - which must already be served on localhost:8000 - with the styleguide on /styleguide and watch for file changes.
• npm run lint: lint the JavaScript and the SCSS code.

In production, only the static files will be used. It is recommended to commit the compiled assets just before a new release only. This will prevent to have a growing repository due to the minified files.

License

GvoT is developed by Cliss XXI and licensed under the AGPLv3+.