twitter-markov 0.2.3

Install

Download or clone the package and run python setup.py install. Feel free to use a virtualenv, if you’re into that.

Brain Train

Train the brain with the twittermarkov_learn command.

The twittermarkov_learn comes with options to ignore replies or retweets, and to filter out mentions, urls, media, and/or hashtags.

When reading an archive, these arguments use the tweet’s metadata to precisely strip the offending content. This may not work well for tweets posted before 2011 or so. For text files or older tweets, a regular expression search is used.

# Usage is twittermarkov_learn ARCHIVE BRAIN
$ twittermarkov_learn twitter/archive/path archive.brain

# teach the brain from a text file
$ twittermarkov_learn --txt file.txt txt.brain

$ twittermarkov_learn --no-replies twitter/archive/path archive-no-replies.brain
# Text like this will be ignored:
# @sample I ate a sandwich

# Text like this will be read in:
# I ate a sandwich with @sample

If you’re using a Twitter archive, the ARCHIVE argument should be the top-level folder of the archive (usually a long name like 16853453_3f21d17c73166ef3c77d7994c880dd93a8159c88). If you have a text file, the argument should be a file name

Config

See the bots.yaml file for a full list of settings. Plug your settings in and save the file as bots.yaml to your home directory or ~/bots. You can also use JSON, if that’s your thing.

At a minimum, your config file will need to look like this:

apps:
    example_app_name:
        consumer_key: ...
        consumer_secret: ...

users:
    example_screen_name:

        key: ...
        secret: ...

        app: example_app_name

        # If you want your bot to continue to learn, include this
        parent: your_screen_name

Read up on dev.twitter.com on obtaining authentication tokens.

First Tweet

Tweeting is easy. By default, the twittermarkov application will learn recent tweets from your parent and send one tweet.

The very first time you tweet, you should use:

$ twittermarkov --tweet --no-learn example_screen_name

After that, use:

$ twittermarkov --tweet example_screen_name

To have your bot reply to mentions, use:

$ twittermarkov --reply example_screen_name

Automating

On a *nix system, set up a cron job like so:

0 10-20 * * * twittermarkov --tweet example_screen_name
15,45 10-20 * * * twittermarkov --reply example_screen_name