From f1dc185bcff67648273874030ebaa9d63ca08a2e Mon Sep 17 00:00:00 2001
From: Simon Van de Vyver <simon.vandevyver@ugent.be>
Date: Tue, 22 Oct 2024 15:07:00 +0200
Subject: [PATCH] allow all sparseness factors

---
 libsais64-rs/src/lib.rs | 20 +++-----------------
 sa-builder/src/lib.rs   | 26 +++++++++++++++++++++++++-
 2 files changed, 28 insertions(+), 18 deletions(-)

diff --git a/libsais64-rs/src/lib.rs b/libsais64-rs/src/lib.rs
index e0ff871..2f95c93 100644
--- a/libsais64-rs/src/lib.rs
+++ b/libsais64-rs/src/lib.rs
@@ -17,30 +17,16 @@ pub mod bitpacking;
 ///
 /// 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(text: &Vec<u8>, sparseness_factor: u8) -> Result<Vec<i64>, &str> {
-    let sparseness_factor = sparseness_factor as usize;
-    let mut libsais_sparseness = sparseness_factor;
-    let mut sa;
+pub fn sais64(text: &Vec<u8>, libsais_sparseness: usize) -> Result<Vec<i64>, &str> {
     let exit_code;
+    let mut sa;
 
-    if sparseness_factor * BITS_PER_CHAR <= 16 {
+    if libsais_sparseness * BITS_PER_CHAR <= 16 {
         // bitpacked values fit in uint16_t
         let packed_text = bitpack_text_16(text, libsais_sparseness);
         sa = vec![0; packed_text.len()];
         exit_code = unsafe { libsais16x64(packed_text.as_ptr(), sa.as_mut_ptr(), packed_text.len() as i64, 0, null_mut()) };
     } else {
-        // bitpacked values do not fit in uint16_t, use 32-bit text
-        // set libsais_sparseness to highest sparseness factor fitting in 32-bit value and sparseness factor divisible by libsais sparseness
-        // max 28 out of 32 bits used, because a bucket is created for every element of the alfabet 8 * 2^28).
-        libsais_sparseness = 28 / BITS_PER_CHAR;
-        while sparseness_factor % libsais_sparseness != 0 && libsais_sparseness * BITS_PER_CHAR > 16 {
-            libsais_sparseness -= 1;
-        }
-
-        if libsais_sparseness * BITS_PER_CHAR <= 16 {
-            return Err("invalid sparseness factor");
-        }
-
         let packed_text = bitpack_text_32(text, libsais_sparseness);
         sa = vec![0; packed_text.len()];
         let k = 1 << (libsais_sparseness * BITS_PER_CHAR);
diff --git a/sa-builder/src/lib.rs b/sa-builder/src/lib.rs
index 2cd4afb..f1fa28e 100644
--- a/sa-builder/src/lib.rs
+++ b/sa-builder/src/lib.rs
@@ -54,7 +54,7 @@ pub fn build_ssa(
 
     // Build the suffix array using the selected algorithm
     let mut sa = match construction_algorithm {
-        SAConstructionAlgorithm::LibSais => libsais64_rs::sais64(&text, sparseness_factor)?,
+        SAConstructionAlgorithm::LibSais => libsais64(&text, sparseness_factor)?,
         SAConstructionAlgorithm::LibDivSufSort => libdivsufsort_rs::divsufsort64(text).ok_or("Building suffix array failed")?
     };
 
@@ -66,6 +66,30 @@ pub fn build_ssa(
     Ok(sa)
 }
 
+const MAX_SPARSENESS: usize = 5;
+fn libsais64(text: &Vec<u8>, sparseness_factor: u8) -> Result<Vec<i64>, &str> {
+    let sparseness_factor = sparseness_factor as usize;
+
+    // set libsais_sparseness to highest sparseness factor fitting in 32-bit value and sparseness factor divisible by libsais sparseness
+    // max 28 out of 32 bits used, because a bucket is created for every element of the alfabet 8 * 2^28).
+    let mut libsais_sparseness = MAX_SPARSENESS;
+    while sparseness_factor % libsais_sparseness != 0 {
+        libsais_sparseness -= 1;
+    }
+    let sample_rate = sparseness_factor / libsais_sparseness;
+    println!("  Sparseness factor: {}", sparseness_factor);
+    println!("  Libsais sparseness factor: {}", libsais_sparseness);
+    println!("  Sample rate: {}", sample_rate);
+
+    let mut sa = libsais64_rs::sais64(&text, libsais_sparseness)?;
+
+    if sample_rate > 1 {
+        sample_sa(&mut sa, sample_rate as u8);
+    }
+
+    Ok(sa)
+}
+
 /// Translate all L's to I's in the given text
 ///
 /// # Arguments