perf: improve alignment::distance::levenshtein and alignment::distanc…

…e::bounded_levenshtein on strrings where distance is small (#522)

* added O(nk) algorithm for edit distance with simd

* small edits

* small edits

* format fixing

* update dependencies

* increase MSRV

---------

Co-authored-by: Johannes Köster <johannes.koester@tu-dortmund.de>
Co-authored-by: Johannes Koester <johannes.koester@uk-essen.de>

.github/workflows/rust.yml

@@ -86,7 +86,7 @@ jobs:
     needs: Formatting
     runs-on: ubuntu-latest
     env:
-      MSRV_VERSION: 1.60.0
+      MSRV_VERSION: 1.62.0
     steps:
       - name: Checkout repository
         uses: actions/checkout@v2

Cargo.toml

@@ -53,6 +53,7 @@ triple_accel = ">=0.3, <0.5"
 thiserror = "1"
 anyhow = "1"
 rand = ">=0.7.3, < 0.9"
+editdistancek = ">=1.0.1, <2"
 
 [dependencies.vec_map]
 version = "0.8"

README.md

@@ -57,7 +57,7 @@ For extra credit, feel free to familiarize yourself with:
 
 ## Minimum supported Rust version
 
-Currently the minimum supported Rust version is 1.60.0.
+Currently the minimum supported Rust version is 1.62.0.
 
 ## License
 

src/alignment/distance.rs

@@ -6,8 +6,6 @@
 //! Various subroutines for computing a distance between sequences. Features
 //! both scalar and efficient vectorized distance functions with SIMD.
 
-use std::cmp::min;
-
 use crate::utils::TextSlice;
 
 /// Compute the Hamming distance between two strings. Complexity: O(n).
@@ -59,30 +57,7 @@ pub fn hamming(alpha: TextSlice<'_>, beta: TextSlice<'_>) -> u64 {
 /// ```
 #[allow(unused_assignments)]
 pub fn levenshtein(alpha: TextSlice<'_>, beta: TextSlice<'_>) -> u32 {
-    let mut columns = [vec![0u32; alpha.len() + 1], vec![0u32; alpha.len() + 1]];
-    let mut i_prev = 0;
-    let mut i_cur = 1;
-
-    for i in 0..columns[0].len() {
-        columns[0][i] = i as u32;
-    }
-
-    for (j, item) in beta.iter().enumerate() {
-        i_cur %= 2;
-        i_prev = 1 - i_cur;
-
-        columns[i_cur][0] = 1 + j as u32;
-        for i in 1..columns[0].len() {
-            columns[i_cur][i] = min(
-                columns[i_prev][i - 1] + if alpha[i - 1] == *item { 0 } else { 1 },
-                min(columns[i_cur][i - 1] + 1, columns[i_prev][i] + 1),
-            );
-        }
-
-        i_cur += 1;
-    }
-
-    columns[i_cur - 1][columns[0].len() - 1]
+    editdistancek::edit_distance(alpha, beta) as u32
 }
 
 pub mod simd {
@@ -106,6 +81,7 @@ pub mod simd {
     //! smaller vectors.
 
     use crate::utils::TextSlice;
+    use std::cmp::{max, min};
 
     /// SIMD-accelerated Hamming distance between two strings. Complexity: O(n / w), for
     /// SIMD vectors of length w (usually w = 16 or w = 32).
@@ -187,7 +163,15 @@ pub mod simd {
     /// assert_eq!(ldist, None);
     /// ```
     pub fn bounded_levenshtein(alpha: TextSlice<'_>, beta: TextSlice<'_>, k: u32) -> Option<u32> {
-        triple_accel::levenshtein::levenshtein_simd_k(alpha, beta, k)
+        if let Some(x) = editdistancek::edit_distance_bounded(
+            alpha,
+            beta,
+            min(k as usize, max(alpha.len(), beta.len())),
+        ) {
+            Some(x as u32)
+        } else {
+            None
+        }
     }
 }