v/vlib/strings/similarity.v

61 lines
1.7 KiB
V
Raw Normal View History

module strings
2019-09-14 22:48:30 +02:00
#-js
// use levenshtein distance algorithm to calculate
// the distance between between two strings (lower is closer)
pub fn levenshtein_distance(a, b string) int {
2019-09-14 22:48:30 +02:00
mut f := [0].repeat2(b.len+1)
for ca in a {
mut j := 1
mut fj1 := f[0]
f[0]++
for cb in b {
mut mn := if f[j]+1 <= f[j-1]+1 { f[j]+1 } else { f[j-1]+1 }
if cb != ca {
mn = if mn <= fj1+1 { mn } else { fj1+1 }
} else {
mn = if mn <= fj1 { mn } else { fj1 }
}
fj1 = f[j]
f[j] = mn
j++
}
}
return f[f.len-1]
}
// use levenshtein distance algorithm to calculate
// how similar two strings are as a percentage (higher is closer)
2019-09-13 15:15:30 +02:00
pub fn levenshtein_distance_percentage(a, b string) f32 {
d := levenshtein_distance(a, b)
l := if a.len >= b.len { a.len } else { b.len }
2019-09-13 15:15:30 +02:00
return (1.00 - f32(d)/f32(l)) * 100.00
}
// implementation of SørensenDice coefficient.
// find the similarity between two strings.
// returns f64 between 0.0 (not similar) and 1.0 (exact match).
2019-09-13 15:15:30 +02:00
pub fn dice_coefficient(s1, s2 string) f32 {
if s1.len == 0 || s2.len == 0 { return 0.0 }
if s1 == s2 { return 1.0 }
if s1.len < 2 || s2.len < 2 { return 0.0 }
2019-09-15 19:07:12 +02:00
a := if s1.len > s2.len { s1 } else { s2 }
b := if a == s1 { s2 } else { s1 }
mut first_bigrams := map[string]int
2019-09-15 19:07:12 +02:00
for i := 0; i < a.len-1; i++ {
bigram := a.substr(i, i+2)
first_bigrams[bigram] = if bigram in first_bigrams { first_bigrams[bigram]+1 } else { 1 }
}
mut intersection_size := 0
2019-09-15 19:07:12 +02:00
for i := 0; i < b.len-1; i++ {
bigram := b.substr(i, i+2)
count := if bigram in first_bigrams { first_bigrams[bigram] } else { 0 }
if count > 0 {
2019-09-15 19:07:12 +02:00
first_bigrams[bigram] = count - 1
intersection_size++
}
}
2019-09-15 19:07:12 +02:00
return (2.0 * intersection_size) / (f32(a.len) + f32(b.len) - 2)
}