Chattymarkov is a python module that lets you generate random sentences through a markov chain algorithm.
It is useful mostly for bots which are aimed at learning from user’s chat and generate totally w.t.f. answers.
The library can support multiple databases, especially redis which is quite suitable to store relevant information for markov chains.
pip install chattymarkov
From examples/memory.py
#!/usr/bin/env python3
from chattymarkov import ChattyMarkov
markov = ChattyMarkov("memory://")
markov.learn("My favorite animal is the crocodile")
markov.learn("The word animal is six letters long")
print(markov.generate())
Here the markov
instance learns two sentences (presumably gathered
from a chat network such as IRC or Discord). What is interesting is that
the ‘animal is’ sequence appears twice. So the generate()
method,
which returns a completely random result, may return an entirely built
sentence which hasn’t been ever written by anyone:
$ ./memory.py
The word animal is the crocodile
$ ./memory
My favorite animal is six letters long
The more sentences, the funnier generated ones.
- Redis (recommended): you can either provide a unix socket path (e.g.
redis:///path/to/unix_socket.sock;db=0;password=foobar
or an URL (e.gredis://user:password@localhost:6739/0
). For the async version, either useredis_async://
or instantiate aChattyMarkovAsync
instance. - JSON: you can provide a path to a file that will be formated with JSON.
Example:
json:///path/to/file.json
- Memory: in-memory database, just provide
memory://
as a connect string. For the async version, usememory_async://
instead.
If you want to add some support to a database or redesign the library, please make a pull request so we can discuss about it.
- Support other databases?