twitterbot 0.3.4

Features

You can use Twitterbot to:

• Post a new message.

• Reply to any twitter mentions with a message.

Installation

You can get twitterbot from PyPI with:

pip install twitterbot

The development version can be installed with:

pip install -e git://github.com/jessamynsmith/twitterbot.git#egg=twitterbot

If you are developing locally, your version can be installed from the project directory with:

python setup.py.install

Usage

Quick Start

By default, settings are populated from environment variables. The authentication variables are required and can be obtained from your Twitter account. It is recommended that you read Twitter’s Automation rules and best practices before setting up a bot.

• TWITTER_CONSUMER_KEY

• TWITTER_CONSUMER_SECRET

• TWITTER_OAUTH_SECRET

• TWITTER_OAUTH_TOKEN

You can optionally set the following environment variables:

• TWITTER_MESSAGE_PROVIDER

  Provides messages to be posted. Defaults to ‘messages.HelloWorldMessageProvider’, a simple provider that always returns “Hello World!”

• TWITTER_SINCE_ID_PROVIDER

  Provides storage for since_id. Twitter uses sinFile in which to store last retrieved since_id. Defaults to using the filesystem (‘./.since_id.txt’). You may set a value in the file to start handling mentions at a particular message id.

• TWITTER_DRY_RUN

  If set to True, messages will be logged rather than actually posting them to Twitter.

The underquotedbot project is a working example of using the twitterbot library to build a bot that is deployed to heroku and runs the twitter account @the_underquoted.

Setting a Custom Message Provider

You can inject your own message provider by setting the following environment variable:

export TWITTER_MESSAGE_PROVIDER='bot.messages.MyMessageProvider'

You would then need to create a bot.messages module with a MyMessageProvider class that implements the create() method, e.g.

class MyMessageProvider(object):

    def create(self, mention, max_message_length):
        """
        Create a message
        :param mention: JSON object containing mention details from Twitter
        :param max_message_length: Maximum allowable length for created message
        :return: a message
        """
        return "This is my message!"

Setting a Custom Since_id Provider

The default is to use the FileSystemSinceIdProvider. Using the file system will NOT work correctly on Heroku or any other host with an ephemeral file system. If you cannot rely on the file system, you MUST specify a different SinceIdProvider.

TwitterBot comes with a Redis provider. By default, localhost will be used for redis. To use it, you must first install redis on your system. I recommend using homebrew on OSX:

brew install redis
brew services start redis

Once you have redis installed or available as a service, you can install the python redis package and set environment variables to configure the provider.

pip install redis
export TWITTER_SINCE_ID_PROVIDER='twitter_bot.since_id.redis_provider.RedisSinceIdProvider'
export REDIS_URL=redis://:@somehost:someport # Optional, defaults to localhost

You can inject a custom since_id provider by setting the following environment variable:

export TWITTER_SINCE_ID_PROVIDER='bot.since_id.MySinceIdProvider'

You would then need to create a bot.since_id module with a MySinceIdProvider class that implements the get(), set(), and delete() methods, e.g.

# since_id.py
import os
from twitter_bot import SettingsError
from twitter_bot import BaseSinceIdProvider

class MySinceIdProvider(BaseSinceIdProvider):

    def __init__(self, source=None):
        if not source:
            source = os.environ.get('TWITTER_SINCE_ID_SOURCE')
            if not source:
                raise SettingsError("You must supply source or set the TWITTER_SINCE_ID_SOURCE "
                                    "environment variable.")
        self.source = source

    def get(self):
        return self.source.get('since_id')

    def set(self, since_id):
        return self.source.set('since_id', since_id)

    def delete(self):
        return self.source.delete('since_id')

Overriding Settings

If you require more control over settings, you can subclass Settings:

from twitter_bot import Settings

class MyBotSettings(Settings):
    def __init__(self):
        super(MyBotSettings, self).__init__()
        self.MESSAGE_PROVIDER = 'bot.messages.MyProvider'

Automating the bot

To run the bot as a cron job or Heroku scheduler task, you can make make a small script that uses the provided runner. If you have customized settings, import your own settings class rather than the provided settings.

#!/usr/bin/env python
# runner.py

import sys

from twitter_bot import BotRunner, Settings

if __name__ == '__main__':
    if len(sys.argv) != 2:
        print("You must specify a single command, either 'post_message' or 'reply_to_mentions'")
        result = 1
    else:
        result = BotRunner().go(Settings(), sys.argv[1])
    sys.exit(result)

Then call the script as follows:

$ ./runner.py post_message
$ ./runner.py reply_to_mentions

Development

Fork the project on github and git clone your fork, e.g.:

git clone https://github.com/<username>/twitterbot.git

Set up virtualenv:

mkvirtualenv twitterbot
pip install -r requirements/package.txt -r requirements/test.txt

Run tests with coverage (should be 100%) and check code style:

coverage run -m nose
coverage report -m
flake8

Verify all supported Python versions:

pip install tox
tox

Run bot:

$ ./bin/runner.py reply_to_mentions  # Check twitter stream for mentions, and reply
$ ./bin/runner.py post_message       # Post a message to twitter

Validating The Project Locally

The CircleCI build can be validated locally, using the CircleCI CLI and docker.

First, install Docker Desktop

Install the CircleCI CLI, e.g. using homebrew on OSX:

$ brew install circleci

Then, you can validate it by running this command in the terminal:

$ circleci config validate

Once you know your config is valid, you can test it. The CLI allows you to run a single job from CircleCI on your desktop using docker:

$ circleci local execute --job build

For more information, see the [CircleCI docs](https://circleci.com/docs/2.0/local-cli/#validate-a-circleci-config)