weatherBot

A Twitter bot for weather. Powered by Dark Sky.

Note: Any language or wording suggestions are appreciated and should be submitted as an issue or pull request.

Example bots can be found at @MorrisMNWeather and @WeatherByBrian.

weatherBot tweets the current weather condition at scheduled times. If a special weather event is happening, it will tweet that (outside of the scheduled times). weatherBot can also tweet the current day's forecast.

Features

• Current conditions at scheduled times
• Daily forecast at a scheduled time
• Severe weather alerts issued by a governmental authority
• Real time "special" events (precipitation, fog, extreme temperatures, wind, etc.)
• Granular throttling of special events
• Variable location for all tweets based on the locations of a user's recent tweets
• Fully customizable text for tweets via a YAML file
• International support for timezones, units, and languages
• Twitter geolocation in each tweet
• Console and file based logging
• Send the traceback of a crash as a direct message
• Cache runtime information to a file for easy resuming
• Configuration file
• Deploy via Heroku or Docker

Install Dependencies

Run the following from the repository root directory to install the needed dependencies.

# The minimum dependencies needed to run weatherBot
pip3 install -r requirements.txt
# Additional dependencies needed for testing, linting, and validating
pip3 install -r requirements-dev.txt

Use

weatherBot.py has been built for Python 3 (tested with 3.3 and above). Legacy Python is not supported.

1. Set your location and other settings in weatherBot.conf
2. Set your API keys and secrets as environmental variables (recommended) or in keys.py

python3 weatherBot.py weatherBot.conf

You're all set!

Settings and Customizing

Configuration File

Many features of weatherBot can be customized in a conf file. This ships with a file named weatherBot.conf, but can be called whatever you'd like. Each option has a comment above it describing its purpose. If you want a clean conf file, feel free to remove all but the settings you set, they are all optional. The section headers must remain in the file.

API Keys

The Twitter app consumer key and secret as well as the access token and token secret are located either in environmental variables (recommended) or in the keys.py file. The script will pull in the keys from the environmental variables over the keys.py file. See https://apps.twitter.com to get your keys and secrets. The names of the environmental variables are as follows: WEATHERBOT_CONSUMER_KEY, WEATHERBOT_CONSUMER_SECRET, WEATHERBOT_ACCESS_TOKEN, WEATHERBOT_ACCESS_TOKEN_SECRET, and WEATHERBOT_DARKSKY_KEY. Entering keys into keys.py is not required if you have entered them as environmental variables.

Strings

The language as well as the text used for all tweets can be edited or added in strings.yml. Remember to set the units and path/filename (defaults to strings.yml) in the configuration file.

Variable Location

Enable variable location to have the location for weather change. The Twitter username in the variable location user setting will be used to determine this location. The specified user must tweet with location fairly regularly (at least every 20 tweets, not including retweets), or the manually entered location will be used. The most recent tweet with a location will be used to get the location for weather. For example, say the given user tweets from Minneapolis, MN one day. Minneapolis will be used as the location indefinitely until a new tweet with location is posted or if 20 new tweets have been posted that do not contain a location. weatherBot checked the user's timeline every 30 minutes for updated in location. The human readable Twitter location will also be added to the beginning of each tweet. For example, in the same case as earlier, "Minneapolis, MN: " would be prefixed to every tweet.

Deploying to Heroku

weatherBot can easily be deployed to Heroku. Install the heroku-toolbelt and run the following to get started:

heroku login
heroku create
git push heroku master

The Twitter and Dark Sky keys need to be added. The format to do so is:

heroku config:set \
WEATHERBOT_CONSUMER_KEY=xxx \
WEATHERBOT_CONSUMER_SECRET=xxx \
WEATHERBOT_ACCESS_TOKEN=xxx \
WEATHERBOT_ACCESS_TOKEN_SECRET=xxx \
WEATHERBOT_DARKSKY_KEY=xxx

You can also add keys/environmental variables on the Heroku project's settings page.

Deploying with Docker

weatherBot can easily be deployed using Docker.

weatherBot from Docker Hub

To download the image, run:

docker pull brianmitchl/weatherbot

Start the bot with the following, replacing the API keys and secrets with the correct strings. You will also need to add your conf file to the container (I use the -v flag).

docker run --name weatherBot -d \
-v /home/brianmitchell/weatherBot/dist/custom.conf:/src/custom.conf \
-e WEATHERBOT_CONSUMER_KEY=xxx \
-e WEATHERBOT_CONSUMER_SECRET=xxx \
-e WEATHERBOT_ACCESS_TOKEN=xxx \
-e WEATHERBOT_ACCESS_TOKEN_SECRET=xxx \
-e WEATHERBOT_DARKSKY_KEY=xxx \
brianmitchl/weatherbot python weatherBot.py custom.conf

Building Your Own weatherBot

Alternatively, if you wish to build the Docker image yourself, run:

docker build -t weatherbot .

Start the bot using the same run command as above, while replacing the image name with what you call yours.

Tools Used

• Tweepy
• Dark Sky API
• python-forecast.io