1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
//! Deterministic prose readability metrics — lexical diversity and sentence
//! rhythm — for the `inkhaven style` report (1.8.1). A companion to the
//! span-based [`style_warnings`](crate::tui::style_warnings) detectors: where
//! those flag individual spots, this measures the whole text quantitatively.
//!
//! Multilingual by construction: word and sentence segmentation are
//! Unicode-aware, so the numbers are meaningful for Russian, French, German, or
//! any script — no English-only assumptions. Pure, offline, zero AI.
use std::collections::HashSet;
use unicode_segmentation::UnicodeSegmentation;
/// Accumulates readability statistics across many paragraphs; call [`add`] for
/// each and [`finish`] once at the end.
///
/// [`add`]: Readability::add
/// [`finish`]: Readability::finish
#[derive(Default)]
pub struct Readability {
tokens: u64,
types: HashSet<String>,
sentence_lengths: Vec<u32>,
}
/// The finished per-scope readability summary.
#[derive(Debug, Clone, PartialEq)]
pub struct ReadabilityReport {
/// Total word tokens.
pub tokens: u64,
/// Distinct word types (case-folded).
pub types: u64,
/// Type–token ratio — lexical diversity, 0..1. Length-sensitive, so most
/// useful compared across texts of similar size.
pub ttr: f64,
/// Sentence count.
pub sentences: u64,
/// Mean words per sentence.
pub mean_sentence_len: f64,
/// Standard deviation of sentence length — the *rhythm*. A very low value
/// relative to the mean is monotone prose.
pub stdev_sentence_len: f64,
/// The longest single sentence, in words.
pub longest_sentence: u32,
}
impl ReadabilityReport {
/// Coefficient of variation of sentence length (stdev / mean) — a
/// scale-free measure of rhythm. Near 0 is metronomic; healthy narrative
/// prose is typically 0.5 or higher. `0.0` when there is nothing to measure.
pub fn rhythm_cv(&self) -> f64 {
if self.mean_sentence_len > 0.0 {
self.stdev_sentence_len / self.mean_sentence_len
} else {
0.0
}
}
}
impl Readability {
/// Fold one paragraph (or any run of text) into the accumulator.
pub fn add(&mut self, text: &str) {
for w in text.unicode_words() {
self.tokens += 1;
self.types.insert(w.to_lowercase());
}
for sentence in split_sentences(text) {
let n = sentence.unicode_words().count() as u32;
if n > 0 {
self.sentence_lengths.push(n);
}
}
}
/// Whether any prose was seen at all.
pub fn is_empty(&self) -> bool {
self.tokens == 0
}
/// Compute the summary.
pub fn finish(&self) -> ReadabilityReport {
let tokens = self.tokens;
let types = self.types.len() as u64;
let ttr = if tokens > 0 { types as f64 / tokens as f64 } else { 0.0 };
let sentences = self.sentence_lengths.len() as u64;
let (mean, stdev, longest) = if sentences > 0 {
let sum: u64 = self.sentence_lengths.iter().map(|&x| u64::from(x)).sum();
let mean = sum as f64 / sentences as f64;
let var = self
.sentence_lengths
.iter()
.map(|&x| {
let d = f64::from(x) - mean;
d * d
})
.sum::<f64>()
/ sentences as f64;
(mean, var.sqrt(), self.sentence_lengths.iter().copied().max().unwrap_or(0))
} else {
(0.0, 0.0, 0)
};
ReadabilityReport {
tokens,
types,
ttr,
sentences,
mean_sentence_len: mean,
stdev_sentence_len: stdev,
longest_sentence: longest,
}
}
}
/// Split text into sentences at terminal punctuation, Unicode-aware and
/// language-agnostic. `.`, `!`, `?`, `…` and their fullwidth CJK variants
/// terminate a sentence; a line break does too (headings, list items, verse
/// lines). Deliberately simple — a deterministic approximation, not a parser;
/// abbreviations may over-split, which is acceptable for an aggregate metric.
fn split_sentences(text: &str) -> Vec<String> {
let mut out = Vec::new();
let mut cur = String::new();
for ch in text.chars() {
if matches!(ch, '.' | '!' | '?' | '…' | '\n' | '\r' | '。' | '！' | '？' | '‽') {
if !cur.trim().is_empty() {
out.push(std::mem::take(&mut cur));
} else {
cur.clear();
}
} else {
cur.push(ch);
}
}
if !cur.trim().is_empty() {
out.push(cur);
}
out
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn ttr_counts_types_over_tokens_case_folded() {
let mut r = Readability::default();
r.add("The cat sat. The Cat ran.");
let rep = r.finish();
// tokens: the cat sat the cat ran = 6; types: the cat sat ran = 4.
assert_eq!(rep.tokens, 6);
assert_eq!(rep.types, 4);
assert!((rep.ttr - 4.0 / 6.0).abs() < 1e-9);
}
#[test]
fn sentence_lengths_and_rhythm() {
let mut r = Readability::default();
// Two sentences of 3 and 5 words.
r.add("one two three. four five six seven eight!");
let rep = r.finish();
assert_eq!(rep.sentences, 2);
assert!((rep.mean_sentence_len - 4.0).abs() < 1e-9);
assert_eq!(rep.longest_sentence, 5);
// stdev of {3,5} about mean 4 is 1.0; CV = 0.25.
assert!((rep.stdev_sentence_len - 1.0).abs() < 1e-9);
assert!((rep.rhythm_cv() - 0.25).abs() < 1e-9);
}
#[test]
fn russian_prose_is_measured() {
let mut r = Readability::default();
// Cyrillic words and a Russian question mark; two sentences.
r.add("Кот сидит на окне. Что он видит?");
let rep = r.finish();
assert_eq!(rep.sentences, 2);
// Кот сидит на окне (4) · Что он видит (3) = 7 tokens, all distinct.
assert_eq!(rep.tokens, 7);
assert_eq!(rep.types, 7);
}
#[test]
fn empty_text_is_safe() {
let r = Readability::default();
assert!(r.is_empty());
let rep = r.finish();
assert_eq!(rep.tokens, 0);
assert_eq!(rep.ttr, 0.0);
assert_eq!(rep.rhythm_cv(), 0.0);
}
}