src/libsyntax/util/lev_distance.rs

   1 use std::cmp;
   2 use crate::symbol::Symbol;
   3
   4 #[cfg(test)]
   5 mod tests;
   6
   7 /// Finds the Levenshtein distance between two strings
   8 pub fn lev_distance(a: &str, b: &str) -> usize {
   9     // cases which don't require further computation
  10     if a.is_empty() {
  11         return b.chars().count();
  12     } else if b.is_empty() {
  13         return a.chars().count();
  14     }
  15
  16     let mut dcol: Vec<_> = (0..=b.len()).collect();
  17     let mut t_last = 0;
  18
  19     for (i, sc) in a.chars().enumerate() {
  20         let mut current = i;
  21         dcol[0] = current + 1;
  22
  23         for (j, tc) in b.chars().enumerate() {
  24             let next = dcol[j + 1];
  25             if sc == tc {
  26                 dcol[j + 1] = current;
  27             } else {
  28                 dcol[j + 1] = cmp::min(current, next);
  29                 dcol[j + 1] = cmp::min(dcol[j + 1], dcol[j]) + 1;
  30             }
  31             current = next;
  32             t_last = j;
  33         }
  34     }
  35     dcol[t_last + 1]
  36 }
  37
  38 /// Finds the best match for a given word in the given iterator
  39 ///
  40 /// As a loose rule to avoid the obviously incorrect suggestions, it takes
  41 /// an optional limit for the maximum allowable edit distance, which defaults
  42 /// to one-third of the given word.
  43 ///
  44 /// Besides Levenshtein, we use case insensitive comparison to improve accuracy on an edge case with
  45 /// a lower(upper)case letters mismatch.
  46 pub fn find_best_match_for_name<'a, T>(iter_names: T,
  47                                        lookup: &str,
  48                                        dist: Option<usize>) -> Option<Symbol>
  49     where T: Iterator<Item = &'a Symbol> {
  50     let max_dist = dist.map_or_else(|| cmp::max(lookup.len(), 3) / 3, |d| d);
  51
  52     let (case_insensitive_match, levenstein_match) = iter_names
  53     .filter_map(|&name| {
  54         let dist = lev_distance(lookup, &name.as_str());
  55         if dist <= max_dist {
  56             Some((name, dist))
  57         } else {
  58             None
  59         }
  60     })
  61     // Here we are collecting the next structure:
  62     // (case_insensitive_match, (levenstein_match, levenstein_distance))
  63     .fold((None, None), |result, (candidate, dist)| {
  64         (
  65             if candidate.as_str().to_uppercase() == lookup.to_uppercase() {
  66                 Some(candidate)
  67             } else {
  68                 result.0
  69             },
  70             match result.1 {
  71                 None => Some((candidate, dist)),
  72                 Some((c, d)) => Some(if dist < d { (candidate, dist) } else { (c, d) })
  73             }
  74         )
  75     });
  76
  77     if let Some(candidate) = case_insensitive_match {
  78         Some(candidate) // exact case insensitive match has a higher priority
  79     } else {
  80         levenstein_match.map(|(candidate, _)| candidate)
  81     }
  82 }