|
| 1 | +/// Compute the Levenshtein edit distance between two strings. |
| 2 | +/// |
| 3 | +/// Used to suggest a likely-intended attribute name when the user mistypes one, |
| 4 | +/// e.g. `validte` -> `validate`. |
| 5 | +pub fn levenshtein(a: &str, b: &str) -> usize { |
| 6 | + let a: Vec<char> = a.chars().collect(); |
| 7 | + let b: Vec<char> = b.chars().collect(); |
| 8 | + let m = a.len(); |
| 9 | + let n = b.len(); |
| 10 | + if m == 0 { |
| 11 | + return n; |
| 12 | + } |
| 13 | + if n == 0 { |
| 14 | + return m; |
| 15 | + } |
| 16 | + |
| 17 | + let mut prev: Vec<usize> = (0..=n).collect(); |
| 18 | + let mut curr: Vec<usize> = vec![0; n + 1]; |
| 19 | + for i in 1..=m { |
| 20 | + curr[0] = i; |
| 21 | + for j in 1..=n { |
| 22 | + let cost = if a[i - 1] == b[j - 1] { 0 } else { 1 }; |
| 23 | + curr[j] = (curr[j - 1] + 1).min(prev[j] + 1).min(prev[j - 1] + cost); |
| 24 | + } |
| 25 | + core::mem::swap(&mut prev, &mut curr); |
| 26 | + } |
| 27 | + prev[n] |
| 28 | +} |
| 29 | + |
| 30 | +/// Return the candidate closest to `query` if its edit distance is `<= max_distance`. |
| 31 | +/// Ties are broken by the order in `candidates`. |
| 32 | +pub fn closest_match<'a>( |
| 33 | + query: &str, |
| 34 | + candidates: &[&'a str], |
| 35 | + max_distance: usize, |
| 36 | +) -> Option<&'a str> { |
| 37 | + let mut best: Option<(usize, &'a str)> = None; |
| 38 | + for cand in candidates { |
| 39 | + let d = levenshtein(query, cand); |
| 40 | + if d <= max_distance && best.map(|(bd, _)| d < bd).unwrap_or(true) { |
| 41 | + best = Some((d, cand)); |
| 42 | + } |
| 43 | + } |
| 44 | + best.map(|(_, s)| s) |
| 45 | +} |
| 46 | + |
| 47 | +#[cfg(test)] |
| 48 | +mod tests { |
| 49 | + use super::*; |
| 50 | + |
| 51 | + #[test] |
| 52 | + fn distance_basics() { |
| 53 | + assert_eq!(levenshtein("", ""), 0); |
| 54 | + assert_eq!(levenshtein("abc", ""), 3); |
| 55 | + assert_eq!(levenshtein("", "abc"), 3); |
| 56 | + assert_eq!(levenshtein("abc", "abc"), 0); |
| 57 | + assert_eq!(levenshtein("validte", "validate"), 1); |
| 58 | + assert_eq!(levenshtein("kitten", "sitting"), 3); |
| 59 | + } |
| 60 | + |
| 61 | + #[test] |
| 62 | + fn picks_closest() { |
| 63 | + let candidates = ["sanitize", "validate", "derive", "default"]; |
| 64 | + assert_eq!(closest_match("validte", &candidates, 2), Some("validate")); |
| 65 | + assert_eq!(closest_match("derve", &candidates, 2), Some("derive")); |
| 66 | + assert_eq!(closest_match("xyz", &candidates, 2), None); |
| 67 | + } |
| 68 | +} |
0 commit comments