Easy Markov chain text generator.
Project description
Markovchatter
Easy Markov chain text generator. (CLI / Web API)
Developed on: emptypage / markovchatter — Bitbucket
Table of contents
Install
pip install markovchatter
This installs The markovchatter
Python module and the markovchatter
command line program. You can run the program like both
$ python3 -m markovchatter ...
and
$ markovchatter ...
Usage
Use as a command
usage: markovchatter.py run [-h] [-d] [-l] [-m MAX_CHARS] [-t TRIES] file [file ...]
positional arguments:
file text file(s) to learn
options:
-h, --help show this help message and exit
-d, --no-divider join words without spaces
-l, --new-line files are in new-line text format
-m MAX_CHARS, --max-chars MAX_CHARS
Maximum number of characters [0]
-t TRIES, --tries TRIES
Maximum times to make a sentence [10]
Example:
$ markovchatter run source.txt
Enim ut sem nulla pharetra diam sit amet risus nullam eget felis eget nunc scelerisque viverra.
(Of course, the output text is depends on your source.)
Run as a Web API server
usage: markovchatter.py server [-h] [-l] [-d] [-p PORT] file [file ...]
positional arguments:
file text file(s) to learn
options:
-h, --help show this help message and exit
-l, --new-line files are in new-line text format
-d, --no-divider join words without spaces
-p PORT, --port PORT listen port number [8080]
Start the server:
$ markovchatter server source.txt
* Serving Flask app 'markovchatter'
* Debug mode: off
WARNING: This is a development server. Do not use it in a production deployment. Use a production WSGI server instead.
* Running on all addresses (0.0.0.0)
* Running on http://127.0.0.1:8080
* Running on http://...:8080
Press CTRL+C to quit
And send GET
request to http://hostname:port/text
in another terminal:
$ curl http://localhost:8080/text
Cursus sit amet consectetur adipiscing elit ut aliquam purus sit amet nisl suscipit adipiscing bibendum est.
(The response text is depends on your source, again.)
Maximum length of characters and count of tries can be controlled by query string like /text?max_charas=140&tries=20
.
Options details
Files
The files given to the program is the source text file, which contain sentences it uses as the model. They are normally well-punctuationed text which terminates its sentence with a dot ('.').
If your source text is not suitable for the style; e.g. languages doesn't use dots for its period, you can use new-line format text file with --new-line
option. See option details bellow.
-d
, --no-divider
The markovify library suppose languages that separate words with spaces, and its output is in the same way too. This is annoying when the language doesn't use spaces for its word divider. When the option is set, markovchatter joins words without spaces; which is more suitable for languages like Japanese.
-l
, --new-line
If this option is set, given files are treated as new-line formatted text; which has each sentence on a newline. This is suitable for languages that doesn't use dot (".") for its sentence period.
The feature comes from a library implementation:
Markovify works best with large, well-punctuated texts. If your text does not use .s to delineate sentences, put each sentence on a newline, and use the markovify.NewlineText class instead of markovify.Text class.
— jsvine/markovify: A simple, extensible Markov chain generator.
When the option is set, marcovchatter uses markovify.NewlineText
class instead of markovify.Text
as described above.
-m MAX_CHARS
, --max-chars MAX_CHARS
Available in run
sub-cpmmand.
Tries making a sentence of no more than MAX_CHARS
characters. If it is specified to 0
, Its length is not limited. The default is 0
.
-t TRIES
, --tries TRIES
Available in run
sub-cpmmand.
Maximum count of tries to generate a valid sentense. If it can not be done in specified trials, it returns empty string.
Docker image
The image containing Linux that the library is already installed is also availabe on Docker Hub.
$ docker run --rm -v $(pwd)/text:/etc/markovchatter emptypage/markovchatter run /etc/markovchatter/dict.txt
Vel fringilla est ullamcorper eget nulla facilisi cras fermentum odio eu feugiat pretium nibh ipsum consequat.
Notice that the image doesn't contain any source text file so you have to explicitly mount it and specify its path as an argument.
License
MIT. See LICENSE.
Thanks
This program uses markovify by Jeremy Singer-Vine and Flask by Armin Ronacher. Thank you for these libraries and its developers.
Changes
1.1.0
- Add control ways for maximum character length and count of tries.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for markovchatter-1.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | c3595798477a8e55870a783ed78b528d897f3a0411520c42d2d1da7bc8d788da |
|
MD5 | 9c6eda913e2da55192db35728aa8dbb2 |
|
BLAKE2b-256 | 0b5228185237f6c879e39bc002aadf7fc467b1b045c08eb04b3c54d805d482fb |