Struct lipsum::MarkovChain

pub struct MarkovChain<'a, R: Rng> { /* fields omitted */ }

Simple order two Markov chain implementation.

The Markov chain is a chain of order two, which means that it will use the previous two words (a bigram) when predicting the next word. This is normally enough to generate random text that looks somewhat plausible. The implementation is based on Generating arbitrary text with Markov chains in Rust.

Implementations

impl<'a> MarkovChain<'a, ThreadRng>

pub fn new() -> MarkovChain<'a, ThreadRng>

Create a new empty Markov chain. It will use a default thread-local random number generator.

Examples

use lipsum::MarkovChain;

let chain = MarkovChain::new();
assert!(chain.is_empty());

impl<'a, R: Rng> MarkovChain<'a, R>

pub fn new_with_rng(rng: R) -> MarkovChain<'a, R>

Create a new empty Markov chain that uses the given random number generator.

Examples

use rand::SeedableRng;
use rand_chacha::ChaCha20Rng;
use lipsum::MarkovChain;

let rng = ChaCha20Rng::seed_from_u64(0);
let mut chain = MarkovChain::new_with_rng(rng);
chain.learn("infra-red red orange yellow green blue indigo x-ray");

// The chain jumps consistently like this:
assert_eq!(chain.generate(1), "Orange.");
assert_eq!(chain.generate(1), "Infra-red.");
assert_eq!(chain.generate(1), "Yellow.");

pub fn learn(&mut self, sentence: &'a str)

Add new text to the Markov chain. This can be called several times to build up the chain.

Examples

use lipsum::MarkovChain;

let mut chain = MarkovChain::new();
chain.learn("red green blue");
assert_eq!(chain.words(("red", "green")), Some(&vec!["blue"]));

chain.learn("red green yellow");
assert_eq!(chain.words(("red", "green")), Some(&vec!["blue", "yellow"]));

pub fn len(&self) -> usize

Returs the number of states in the Markov chain.

Examples

use lipsum::MarkovChain;

let mut chain = MarkovChain::new();
assert_eq!(chain.len(), 0);

chain.learn("red orange yellow green blue indigo");
assert_eq!(chain.len(), 4);

pub fn is_empty(&self) -> bool

Returns true if the Markov chain has no states.

Examples

use lipsum::MarkovChain;

let mut chain = MarkovChain::new();
assert!(chain.is_empty());

chain.learn("foo bar baz");
assert!(!chain.is_empty());

pub fn words(&self, state: Bigram<'a>) -> Option<&Vec<&str>>

Get the possible words following the given bigram, or None if the state is invalid.

Examples

use lipsum::MarkovChain;

let mut chain = MarkovChain::new();
chain.learn("red green blue");
assert_eq!(chain.words(("red", "green")), Some(&vec!["blue"]));
assert_eq!(chain.words(("foo", "bar")), None);

pub fn generate(&mut self, n: usize) -> String

Generate a sentence with n words of lorem ipsum text. The sentence will start from a random point in the Markov chain and a . will be added as necessary to form a full sentence.

See generate_from if you want to control the starting point for the generated text and see iter if you simply want a sequence of words.

Examples

Generating the sounds of a grandfather clock:

use lipsum::MarkovChain;

let mut chain = MarkovChain::new();
chain.learn("Tick, Tock, Tick, Tock, Ding! Tick, Tock, Ding! Ding!");
println!("{}", chain.generate(15));

The output looks like this:

Ding! Tick, Tock, Tick, Tock, Ding! Ding! Tock, Ding! Tick, Tock, Tick, Tock, Tick, Tock.

pub fn generate_from(&mut self, n: usize, from: Bigram<'a>) -> String

Generate a sentence with n words of lorem ipsum text. The sentence will start from the given bigram and a . will be added as necessary to form a full sentence.

Use generate if the starting point is not important. See iter_from if you want a sequence of words that you can format yourself.

pub fn iter(&mut self) -> Words<R>

Make a never-ending iterator over the words in the Markov chain. The iterator starts at a random point in the chain.

pub fn iter_from(&mut self, from: Bigram<'a>) -> Words<R>

Make a never-ending iterator over the words in the Markov chain. The iterator starts at the given bigram.

Trait Implementations

impl<'a> Default for MarkovChain<'a, ThreadRng>

fn default() -> Self

Create a new empty Markov chain. It will use a default thread-local random number generator.