use adapted libsais library

libsais64-rs/builder.rs

@@ -53,7 +53,7 @@ fn main() -> Result<(), Box<dyn Error>> {
 
     // clone the c library
     Command::new("git")
-        .args(["clone", "https://github.com/IlyaGrebnov/libsais.git", "--depth=1"])
+        .args(["clone", "git@github.com:unipept/unipept-libsais.git", "libsais", "--depth=1"])
         .status()
         .expect("Failed to clone the libsais repository");
 

libsais64-rs/libsais-wrapper.h

@@ -1,4 +1,4 @@
-#include "libsais/include/libsais64.h"
+#include "libsais/include/libsais16x64.h"
 
 
-int64_t libsais64(const uint8_t * T, int64_t * SA, int64_t n, int64_t fs, int64_t * freq);
+int64_t libsais16x64(const uint16_t * T, int64_t * SA, int64_t n, int64_t fs, int64_t * freq);

libsais64-rs/src/lib.rs

@@ -2,6 +2,7 @@
 #![allow(non_upper_case_globals)]
 #![allow(non_camel_case_types)]
 #![allow(non_snake_case)]
+use std::ptr::null_mut;
 include!(concat!(env!("OUT_DIR"), "/bindings.rs"));
 
 /// Builds the suffix array over the `text` using the libsais64 algorithm
@@ -13,9 +14,9 @@ include!(concat!(env!("OUT_DIR"), "/bindings.rs"));
 ///
 /// Returns Some with the suffix array build over the text if construction succeeds
 /// Returns None if construction of the suffix array failed
-pub fn sais64_long(text: &mut Vec<i64>, alphabet_size: i64, sparseness_factor: u8) -> Option<Vec<i64>> {
+pub fn sais64(text: &Vec<u16>, sparseness_factor: u8) -> Option<Vec<i64>> {
     let mut sa = vec![0; text.len()];
-    let exit_code = unsafe { libsais64_long(text.as_mut_ptr(), sa.as_mut_ptr(), text.len() as i64, alphabet_size, 0) };
+    let exit_code = unsafe { libsais16x64(text.as_ptr(), sa.as_mut_ptr(), text.len() as i64, 0, null_mut()) };
     if exit_code == 0 {
         let sparseness_factor = sparseness_factor as i64;
         for elem in sa.iter_mut() {

sa-builder/src/bitpacking.rs

@@ -10,7 +10,7 @@ fn get_rank(c: u8) -> u8 {
 }
 
 pub const BITS_PER_CHAR: usize = 5;
-pub fn bitpack_text(text: &Vec<u8>, sparseness_factor: u8) -> Vec<i64> {
+pub fn bitpack_text(text: &Vec<u8>, sparseness_factor: u8) -> Vec<u16> {
     let sparseness_factor = sparseness_factor as usize;
     let num_ints = (text.len() + (sparseness_factor-1)) / sparseness_factor;
     let mut text_packed = vec![0; num_ints];
@@ -21,19 +21,19 @@ pub fn bitpack_text(text: &Vec<u8>, sparseness_factor: u8) -> Vec<i64> {
 
     for i in 0..(num_ints-1) {
         let ti = i * sparseness_factor;
-        let mut element = 0i64;
+        let mut element = 0u16;
         for j in 0..sparseness_factor {
-            let rank_c = get_rank(text[ti + j]) as i64;
+            let rank_c = get_rank(text[ti + j]) as u16;
             element |= rank_c << (BITS_PER_CHAR * (sparseness_factor - 1 - j));
         }
         text_packed[i] = element;
     }
 
     // Handle the last element
-    let mut last_element = 0i64;
+    let mut last_element = 0u16;
     let last_el_start = sparseness_factor * (num_ints - 1);
     for i in 0..((text.len() - 1) % sparseness_factor + 1) {
-        let rank_c = get_rank(text[last_el_start + i]) as i64;
+        let rank_c = get_rank(text[last_el_start + i]) as u16;
         last_element |= rank_c << (BITS_PER_CHAR * (sparseness_factor - 1 - i));
     }
     text_packed[num_ints - 1] = last_element;

sa-builder/src/lib.rs

@@ -1,6 +1,5 @@
 use std::error::Error;
 use crate::bitpacking::bitpack_text;
-use crate::bitpacking::BITS_PER_CHAR;
 use clap::{Parser, ValueEnum};
 
 pub mod bitpacking;
@@ -59,9 +58,9 @@ pub fn build_ssa(
     // Build the suffix array using the selected algorithm
     let mut sa = match construction_algorithm {
         SAConstructionAlgorithm::LibSais => {
-            let mut packed_text = bitpack_text(text, sparseness_factor);
+            let packed_text = bitpack_text(text, sparseness_factor);
 
-            libsais64_rs::sais64_long(&mut packed_text, (1 as i64) << (BITS_PER_CHAR * sparseness_factor as usize), sparseness_factor)
+            libsais64_rs::sais64(&packed_text, sparseness_factor)
         },
         SAConstructionAlgorithm::LibDivSufSort => libdivsufsort_rs::divsufsort64(text)
     }