FriskBy

This web service is part of the FriskBy project.

Air quality measurement devices can POST measurements to the server. The server has a REST api which can be used to query the stored data.

The web service is based on the Python web framework Django: http://www.djangoproject.com.

Getting started

If you want to make modifications to the friskby web-server you must go through some initial setup before you can start hacking. The specific commandlines illustrated in this README assume that you are using a Debian based Linux distribution, but there is nothing Debian or even Linux specific to the software as such.

Installing the dependencies (I)

You need to install the following packages: git, postgresql, postgresql-server-dev-all, python-dev and python-pip:

bash% sudo apt-get install git postgresql postgresql-server-dev-all python-pip python-dev

After pip has been installed, we install all the required Python packages by giving the file requirements.txt to pip:

bash% sudo pip install -r requirements.txt

Setting up the source code

Development of the source code is done on GitHub using a model with personal forks. So to set up your environment for working with the friskby web service:

Create an account on GitHub.
Fork the friskby repository to your personal account. When you have done this you should have a personal repository: https://github.com/$USER/friskby.
Clone your repository down to your personal computer:
```
bash% git clone git@github.com:$USER/friskby
```
Add the FriskByBergen/friskby repository as a remote repository:
```
bash% git remote add upstream git@github.com:FriskByBergen/friskby
```
This remote repository will be used when you should update your local git repository with the changes done by other developers.

The git/github workflows used by FriskBy are very common, and extensive explanations are only a google search away.

Installing the dependencies (II)

In addition to the binary packages listed above you must also install several Python packages. The Python packages can be installed with normal pip install:

bash% sudo pip install -r requirements.txt

Creating the database

A database for storing the measurements is an essential part of the friskby web service, and to develop on the source code for the web server you need to have your own local database. Observe that Django is quite database agnostic, and you could in principle use MySQL or Sqlite instead of postgres for your own personal development. However the friskby web server currently uses postgres in production, and there is also a possibility that we would like to use postgres extensions to Django in the future. The following guideline is therefor based on postgres:

Change identity to posgtgres:
```
bash% sudo su - postgres
```

Create a new user (role):

bash% createuser friskby-user -P
Enter password for new role: <friskby-pwd>
Enter it again: <friskby-pwd>

As indicated the createuser program will prompt for a password.

Create a new database - owned by the new user:

bash% createdb friskby-db --owner=friskby-user

After these steps you should have made a database with name friskby-db and user with credentials (friskby-user, friskby-pwd). These three values should be part of the DATABASE_URL connection string - see the section about environment variables. Log out of the postgres account and test the connection:

bash% psql friskby-db -U friskby-user -h localhost

Environment variables

The configuration of the FriskBy web server is handled through the use of environment variables. In Django configuration settings are read from the settings namespace in the root of the project, in this project the settings/__init__.py file contains several calls of the type:

SETTING_VARIABLE = os.getenv("SETTING_ENV_VARIABLE")

The file init_env.sh.template is a template file which includes the environment variables you should set to run the FriskBy web server. Follow the instructions in this file and create a personal init_env.sh file, the init_env.sh file should not be under version control.

Testing the code

When you updated your environment you are ready to actually run the friskby code. Before starting Django, we need to migrate:

    bash% ./manage.py migrate

To run all the tests:

    bash% ./manage.py test

To start the development server:

   bash% ./manage.py runserver

Then you go to http://127.0.0.1:8000 in your browser and interact with your personal development copy of the FriskBy web server.

Developing and getting the code merged

When you are finished with your changes make a Pull Request on GitHub.

Deploying the code

The FriskBy web server is deployed on Heroku. It should be quite simple to deploy using an alternative platform.

Creating testdata

To get some testdata to work with there are management commands. To create three testsensors with random data:

   bash% ./manage.py add_testdevice TestDev1 TestDev2 TestDev3

This will by default add 100 random datapoints to each of the sensors, but by passing --num=nnn you can add a different number of points. The random devices can be removed with:

   bash% ./manage.py drop_testdevice TestDev1 TestDev3

which will remove the devices 'TestDev1' and 'TestDev3'. If you pass the special argument '--all' to the 'drop_testdevice' managament command all devices will be removed.

In addition to the 'add_testdevice' and 'drop_testdevice' commands there are commands 'add_testdata' and 'drop_testdata' which will only add or drop testdata, not the actual devices.

   bash% ./manage.py add_testdata

Will add 100 datapoints to each available sensor, by passing '--num=' you can control the number of points, and by passing '--device=' or '--sensor=' you can control which sensor gets the data.

   bash% ./manage.py drop_testdata

will drop all the testdata. Pass '--device' or '--sensor'.

Synchronizing with production data

To obtain the production data, follow these steps:

First we let the user friskby-user create databases:

$ sudo -iu postgres
$ psql
> ALTER USER "friskby-user" CREATEDB;
> \q

Then, whenever we want to replace our data with the production data:

$ sudo -iu postgres
$ dropdb friskby-db ; PGUSER=friskby-user PGPASSWORD=fby heroku pg:pull HEROKU_POSTGRESQL_SILVER_URL friskby-db --app friskby

where fby, <friskby-pwd> of course is replaced with your chosen 8+ character password.

semafor/friskby