quote_randomizer

Randomizes an actual quote by switching 1-3 words with ones of matching type.

You must trust and believe in hardware or life becomes impossible.
--Anton Chekhov

original: You must trust and believe in people or life becomes impossible.

This script uses the Natural Language Toolkit (nltk) module to split an orginal quote to part-of-speech tags (POS tags), such as nouns and verbs. Randomization is then done by replacing randomly chosen words with new words of same POS tags. New words are chosen from an internal nltk dataset. Quotes are chosen from a local SQLite database collected from several sources, see below.

Since a word's POS tag is dependant on the context of the sentence around it, this script analyzes the quote in 3-grams, ie. replacing is done so that the new words orignally appeared in a context where the two adjacent words had the same POS tags as in the quote to randomize.

https://www.nltk.org/

Usage

First, install dependencies in a virtualenv with

python3 -m virtualenv venv
source venv/bin/activate
pip install -r requirements.txt

To utilize the nltk module, some additional data is needed:

python -m nltk.downloader book

This includes a corpora and tokenizer, see https://www.nltk.org/data.html.

Then, build a database of quotes with

python main.py --build-database

Finally, generate a randomized quote with

python main.py --quote

Unit tests

Unit tets can be run with

pytest

Quote sources

The quote database is buit from the quotes in data/quotes.txt. This file contains a collection of quotes from people, books, movies, games as well as facts, song lyrics, poem lines, proverbs and company slogans. These are collected from multiple online sources, including

lajanki / quote_randomizer Goto Github PK