From aa6ef9ffbc4682ebc5cd33d1beb641bc4d20ee7e Mon Sep 17 00:00:00 2001
From: Simon Van de Vyver <simon.vandevyver@ugent.be>
Date: Thu, 24 Oct 2024 11:17:08 +0200
Subject: [PATCH] add comments + do not bitpack text sparseness is 1

---
 libsais64-rs/src/bitpacking.rs |  8 ++++++++
 libsais64-rs/src/lib.rs        | 14 +++++---------
 2 files changed, 13 insertions(+), 9 deletions(-)

diff --git a/libsais64-rs/src/bitpacking.rs b/libsais64-rs/src/bitpacking.rs
index 9ee002e..d6d4cc8 100644
--- a/libsais64-rs/src/bitpacking.rs
+++ b/libsais64-rs/src/bitpacking.rs
@@ -9,8 +9,12 @@ fn get_rank(c: u8) -> u8 {
     }
 }
 
+// Amount of bits necessary to represent one character in the protein text.
 pub const BITS_PER_CHAR: usize = 5;
+
+// Bitpack text in a vector of u8 elements. BITS_PER_CHAR * sparseness_factor <= 8.
 pub fn bitpack_text_8(text: &Vec<u8>, sparseness_factor: usize) -> Vec<u8> {
+    assert!(BITS_PER_CHAR * sparseness_factor <= 8);
 
     let num_ints = (text.len() + (sparseness_factor-1)) / sparseness_factor;
     let mut text_packed = vec![0; num_ints];
@@ -42,7 +46,9 @@ pub fn bitpack_text_8(text: &Vec<u8>, sparseness_factor: usize) -> Vec<u8> {
 
 }
 
+// Bitpack text in a vector of u16 elements. BITS_PER_CHAR * sparseness_factor <= 16.
 pub fn bitpack_text_16(text: &Vec<u8>, sparseness_factor: usize) -> Vec<u16> {
+    assert!(BITS_PER_CHAR * sparseness_factor <= 16);
 
     let num_ints = (text.len() + (sparseness_factor-1)) / sparseness_factor;
     let mut text_packed = vec![0; num_ints];
@@ -74,7 +80,9 @@ pub fn bitpack_text_16(text: &Vec<u8>, sparseness_factor: usize) -> Vec<u16> {
 
 }
 
+// Bitpack text in a vector of u16 elements. BITS_PER_CHAR * sparseness_factor <= 32.
 pub fn bitpack_text_32(text: &Vec<u8>, sparseness_factor: usize) -> Vec<u32> {
+    assert!(BITS_PER_CHAR * sparseness_factor <= 32);
 
     let num_ints = (text.len() + (sparseness_factor-1)) / sparseness_factor;
     let mut text_packed = vec![0; num_ints];
diff --git a/libsais64-rs/src/lib.rs b/libsais64-rs/src/lib.rs
index f6b6783..d9eca5a 100644
--- a/libsais64-rs/src/lib.rs
+++ b/libsais64-rs/src/lib.rs
@@ -8,7 +8,7 @@ include!(concat!(env!("OUT_DIR"), "/bindings.rs"));
 
 pub mod bitpacking;
 
-/// Builds the suffix array over the `text` using the libsais64 algorithm
+/// Builds the suffix array over the `text` using the libsais algorithm
 ///
 /// # Arguments
 /// * `text` - The text used for suffix array construction
@@ -24,7 +24,7 @@ pub fn sais64(text: &Vec<u8>, libsais_sparseness: usize) -> Result<Vec<i64>, &st
     let required_bits = libsais_sparseness * BITS_PER_CHAR;
     if required_bits <= 8 {
         // bitpacked values fit in uint8_t
-        let packed_text = bitpack_text_8(text, libsais_sparseness);
+        let packed_text = if libsais_sparseness == 1 { text } else { &bitpack_text_8(text, libsais_sparseness) };
         sa = vec![0; packed_text.len()];
         exit_code = unsafe { libsais64(packed_text.as_ptr(), sa.as_mut_ptr(), packed_text.len() as i64, 0, null_mut()) };
     } else if required_bits <= 16 {
@@ -54,14 +54,10 @@ mod tests {
 
     #[test]
     fn check_build_sa_with_libsais64() {
-        let bits_per_char = 5;
         let sparseness_factor = 4;
-        let mut text = [100834,   // BANA
-            493603,                     // NA-B
-            80975,                      // ANAN
-            65536                       // A$
-        ].to_vec();
+        let mut text = "BANANA-BANANA$".as_bytes().to_vec();
         let sa = sais64(&mut text, sparseness_factor);
-        assert_eq!(sa, Some(vec![12, 8, 0, 4]));
+        let correct_sa: Vec<i64> = vec![12, 8, 0, 4];
+        assert_eq!(sa, Ok(correct_sa));
     }
 }