WeatherServer

Full Stack Embedded 2016 features a cheap weather station. This Django app serves the observed weather data in human and machine readable formats.

Please refer to LICENSE if you want to copy or use this software.

Authors

The members of Full Stack Embedded 2016 are, in alphabetical order:

• Frederic Afadjigla
• Nico Caspari
• William Herndon
• Daniel Lee
• Gregor Schnee

This component was authored mainly by Daniel Lee and Gregor Schnee.

Setup

The following instructions detail how to install a minimal web server in a development environment. If you are wanting to deploy the weather app in a production environment, we suggest you read the Django documentation for deploying apps.

For a quick minimal setup without worrying about the dependencies yourself, run:

sudo ./configure

from the repository's root directory. Be warned, however, that this will install all dependencies and install the app systemwide on your machine. If you do this, you can skip all following steps in the tutorial.

Dependencies

The following dependencies must be installed (e.g. with apt-get or zypper:

• python3
• python3-pip (for installing Python dependencies)

Additionally, the following Python packages are required. Note that if you're installing using setup.py you will not need to install anything manually for running the app. Also, pandas is required to instal the random mock data. This dependency will be removed along with the generation utility as soon as real sample data has been integrated into the package.

• django
• pandas (only required for generating random sample data)

These packages can be installed without using setup.py e.g. with:

pip3 install $package

Installation for deployment

It is suggested - but not necessary - to install the WeatherServer package if you are wanting to deploy it in some kind of a web server.

If you haven't downloaded the sources you can use PyPI. Note: This is a planned capability, the package won't be registered with PyPI until a release version is available. Once the package has been registered, it can be installed as follows:

pip install WeatherServer

If you have downloaded the sources you can install the package by navigating into the package folder and using the following command:

python3 setup.py install

Remember to preface the commands with sudo if you wish to install the package system wide.

Setting up the project

Setup the app by initializing the database and setting up its tables. If you don't configure weather_server/weather_server/settings.py this will automatically use a SQLite database in the app's root folder.

me@host:~/WeatherServer/weather_server> # Setup database
me@host:~/WeatherServer/weather_server> python3 manage.py makemigrations
me@host:~/WeatherServer/weather_server> python3 manage.py migrate
me@host:~/WeatherServer/weather_server> # Setup super user - follow prompts
me@host:~/WeatherServer/weather_server> python3 manage.py createsuperuser

Now that the database is setup, you can start the development server as follows:

me@host:~/WeatherServer/weather_server> python3 manage.py runserver

Now you can access the server by navigating e.g. to http://localhost:8000.

Using test data

Test data is currently a hack because currently no actual data can enter the database - the weather station hardware does not exist yet. This section will be removed in the future.

Creating and importing test data

Test data can be created by executing show_weather/utils/create_mock_dataseries.py. Note that this step requires pandas.

me@host:~/WeatherServer/weather_server/show_weather/utils/> python3 create_mock_dataseries.py

This generates a randomized series of observations, stored as a CSV under weather_server/show_weather/utils/data_series.csv. You can load this into the database using weather_server/show_weather/utils/import_generated_series.py. This is slightly complicated because it uses Django's ORM, so it has to either be started inside a Django shell, which takes care of all the bookkeeping legwork for you (establishing DB connection, etc.) or you have to do Django's administrative work for it. This example reduces the workload by using a Django shell, but it's a bit esoteric for the same reason - the Django shell is started, then the script is loaded as a string and executed inside that environment. Don't worry, I washed my hands after writing it and it will soon die.

weather_server/show_weather/utils/import_generated_series.py creates an observing station and saves all of the observations as belonging to that station. It also expects the sample data series to be stored inside the utils folder.

me@host:~/WeatherServer/weather_server> python3 manage.py shell
Python 3.4.1 (default, May 23 2014, 17:48:28) [GCC] on linux
Type "help", "copyright", "credits" or "license" for more information.
(InteractiveConsole)
>>> with open("show_weather/utils/import_generated_series.py") as script:
...     exec(script.read())

Lots of warnings follow because the datetime object used is timezone naive. Now your database is full of observations for my apartment, at the intersection of Constitution Hill, The Mall and Birdcage Walk.

Examining test data

Now your development server is running. If you navigate to localhost:8000/admin/ you can login to the admin page and see the entries in your database.

API for viewing data

Currently, only CSVs are served as raw data. The following endpoints are supported:

• /data/csv/stations/ - return a CSV of known station metadata
• /data/csv/$station_id/$start_date - $end_date - return a CSV of observations made at the station with $station_id between the dates (inclusively) specified. Dates are to be formated as YYYY-MM-DD HH:MM.