use std::cmp;
use symbol::Symbol;
pub fn lev_distance(a: &str, b: &str) -> usize {
if a.is_empty() {
return b.chars().count();
} else if b.is_empty() {
return a.chars().count();
}
let mut dcol: Vec<_> = (0..b.len() + 1).collect();
let mut t_last = 0;
for (i, sc) in a.chars().enumerate() {
let mut current = i;
dcol[0] = current + 1;
for (j, tc) in b.chars().enumerate() {
let next = dcol[j + 1];
if sc == tc {
dcol[j + 1] = current;
} else {
dcol[j + 1] = cmp::min(current, next);
dcol[j + 1] = cmp::min(dcol[j + 1], dcol[j]) + 1;
}
current = next;
t_last = j;
}
} dcol[t_last + 1]
}
pub fn find_best_match_for_name<'a, T>(iter_names: T,
lookup: &str,
dist: Option<usize>) -> Option<Symbol>
where T: Iterator<Item = &'a Symbol> {
let max_dist = dist.map_or_else(|| cmp::max(lookup.len(), 3) / 3, |d| d);
iter_names
.filter_map(|&name| {
let dist = lev_distance(lookup, &name.as_str());
if dist <= max_dist { Some((name, dist))
} else {
None
}
})
.min_by_key(|&(_, val)| val) .map(|(s, _)| s) }
#[test]
fn test_lev_distance() {
use std::char::{from_u32, MAX};
for c in (0..MAX as u32)
.filter_map(|i| from_u32(i))
.map(|i| i.to_string()) {
assert_eq!(lev_distance(&c[..], &c[..]), 0);
}
let a = "\nMäry häd ä little lämb\n\nLittle lämb\n";
let b = "\nMary häd ä little lämb\n\nLittle lämb\n";
let c = "Mary häd ä little lämb\n\nLittle lämb\n";
assert_eq!(lev_distance(a, b), 1);
assert_eq!(lev_distance(b, a), 1);
assert_eq!(lev_distance(a, c), 2);
assert_eq!(lev_distance(c, a), 2);
assert_eq!(lev_distance(b, c), 1);
assert_eq!(lev_distance(c, b), 1);
}