Add chi-square test and fix a typo that mane it not uniform

Cargo.toml

@@ -11,3 +11,6 @@ eyre = "0.6.8"
 lazy_static = "1.4.0"
 rand = "0.8.5"
 regex = "1.9.1"
+
+[dev-dependencies]
+statrs = "0.16.0"

src/main.rs

@@ -1,8 +1,8 @@
+mod passphrase;
 mod words;
 
 use clap::Parser;
 use eyre::Result;
-use rand::{rngs::ThreadRng, Rng};
 
 #[derive(Parser, Debug)]
 #[command(author, version, about, long_about = None)]
@@ -24,34 +24,8 @@ fn main() -> Result<()> {
 
  println!(
  "{}",
- passphrase(&mut rng, &mut words, args.num_words, &args.separator)
+ passphrase::new(&mut rng, &mut words, args.num_words, &args.separator)
  );
 
  Ok(())
 }
-
-fn passphrase(
- rng: &mut ThreadRng,
- words: &mut Vec<String>,
- num_words: usize,
- separator: &str,
-) -> String {
- if words.len() < num_words {
- eprintln!(
- "Your dictionary only has {} suitable words, but you asked for {} words.",
- words.len(),
- num_words
- );
- return "".to_string();
- }
-
- (0..num_words).for_each(|i| {
- let j = rng.gen_range(0..words.len());
- words.swap(i, j)
- });
-
- (0..num_words)
- .map(|i| words[i].to_owned())
- .collect::<Vec<String>>()
- .join(separator)
-}

src/passphrase.rs

@@ -0,0 +1,67 @@
+use rand::{rngs::ThreadRng, Rng};
+
+pub fn new(
+ rng: &mut ThreadRng,
+ words: &mut Vec<String>,
+ num_words: usize,
+ separator: &str,
+) -> String {
+ if words.len() < num_words {
+ eprintln!(
+ "Your dictionary only has {} suitable words, but you asked for {} words.",
+ words.len(),
+ num_words
+ );
+ return "".to_string();
+ }
+
+ (0..num_words).for_each(|i| {
+ let j = rng.gen_range(i..words.len());
+ words.swap(i, j)
+ });
+
+ (0..num_words)
+ .map(|i| words[i].to_owned())
+ .collect::<Vec<String>>()
+ .join(separator)
+}
+
+mod test {
+ #[test]
+ fn read_words() {
+ use crate::{passphrase, words};
+ use statrs::distribution::{ChiSquared, ContinuousCDF};
+ use std::collections::HashMap;
+
+ let n = 12_000_000;
+ let mut rng = rand::thread_rng();
+
+ // this test file has 4 words, which can have 24 permutations
+ let words = words::list(Some("src/fixtures/test")).unwrap();
+
+ let mut histogram: HashMap<String, u32> = HashMap::new();
+ (1..n).for_each(|_| {
+ let mut words = words.clone();
+ let s = passphrase::new(&mut rng, &mut words, 4, " ");
+ *histogram.entry(s).or_insert(0) += 1;
+ });
+
+ assert_eq!(histogram.len(), 24);
+
+ let expected_frequency = n as f64 / 24.0;
+ let chi_squared_stat: f64 = histogram
+ .iter()
+ .map(|(_, v)| (*v as f64 - expected_frequency).powi(2) / expected_frequency)
+ .sum();
+ let df = ((2 - 1) * (24 - 1)) as f64;
+ let dist = ChiSquared::new(df).unwrap();
+ let p = dist.cdf(chi_squared_stat);
+
+ eprintln!("χ^2: {}", chi_squared_stat);
+ eprintln!("p: {}", p);
+
+ // the p-value should be greater than 0.05 so that we can't reject the null hypothesis
+ // if we can reject the null hypothesis, then the passphrase generator is not uniform
+ assert_eq!(p > 0.05, true);
+ }
+}