MRG: update sourmash to r0.17.0

Cargo.toml

@@ -12,7 +12,7 @@ crate-type = ["cdylib"]
 pyo3 = { version = "0.22.5", features = ["extension-module", "anyhow"] }
 rayon = "1.10.0"
 serde = { version = "1.0.213", features = ["derive"] }
-sourmash = { version = "0.16.0", features = ["branchwater"] }
+sourmash = { version = "0.17.0", features = ["branchwater"] }
 serde_json = "1.0.132"
 niffler = "2.4.0"
 log = "0.4.22"

src/fastgather.rs

@@ -110,7 +110,7 @@ pub fn fastgather(
         query_name,
         query_filename,
         query_mh,
-        scaled as u64,
+        scaled as u32,
         matchlist,
         threshold_hashes,
         gather_output,

src/fastmultigather.rs

@@ -29,7 +29,7 @@ use crate::utils::{
 pub fn fastmultigather(
     query_filepath: String,
     against_filepath: String,
-    threshold_bp: usize,
+    threshold_bp: u32,
     scaled: Option<usize>,
     selection: Selection,
     allow_failed_sigpaths: bool,
@@ -47,9 +47,9 @@ pub fn fastmultigather(
     )?;
 
     let scaled = match scaled {
-        Some(s) => s,
+        Some(s) => s as u32,
         None => {
-            let scaled = *query_collection.max_scaled().expect("no records!?") as usize;
+            let scaled = *query_collection.max_scaled().expect("no records!?");
             eprintln!(
                 "Setting scaled={} based on max scaled in query collection",
                 scaled
@@ -64,12 +64,11 @@ pub fn fastmultigather(
     let threshold_hashes: u64 = {
         let x = threshold_bp / scaled;
         if x > 0 {
-            x
+            x as u64
         } else {
             1
         }
-    }
-    .try_into()?;
+    };
 
     println!("threshold overlap: {} {}", threshold_hashes, threshold_bp);
 
@@ -164,7 +163,7 @@ pub fn fastmultigather(
                         query_name,
                         query_filename,
                         query_mh,
-                        scaled as u64,
+                        scaled,
                         matchlist,
                         threshold_hashes,
                         Some(gather_output),

src/fastmultigather_rocksdb.rs

@@ -16,7 +16,7 @@ pub fn fastmultigather_rocksdb(
     queries_file: String,
     index: PathBuf,
     selection: Selection,
-    threshold_bp: usize,
+    threshold_bp: u32,
     output: Option<String>,
     allow_failed_sigpaths: bool,
 ) -> Result<(), Box<dyn std::error::Error>> {
@@ -55,7 +55,7 @@ pub fn fastmultigather_rocksdb(
     let send = query_collection
         .par_iter()
         .filter_map(|(coll, _idx, record)| {
-            let threshold = threshold_bp / selection.scaled()? as usize;
+            let threshold = threshold_bp / selection.scaled()?;
             let ksize = selection.ksize()?;
 
             // query downsampling happens here
@@ -76,7 +76,7 @@ pub fn fastmultigather_rocksdb(
                             counter,
                             query_colors,
                             hash_to_color,
-                            threshold,
+                            threshold as usize,
                             &query_mh,
                             Some(selection.clone()),
                         );
@@ -104,8 +104,8 @@ pub fn fastmultigather_rocksdb(
                                     query_bp: query_mh.n_unique_kmers() as usize,
                                     ksize: ksize as usize,
                                     moltype: query_mh.hash_function().to_string(),
-                                    scaled: query_mh.scaled() as usize,
-                                    query_n_hashes: query_mh.size(),
+                                    scaled: query_mh.scaled(),
+                                    query_n_hashes: query_mh.size() as u64,
                                     query_abundance: query_mh.track_abundance(),
                                     query_containment_ani: match_.query_containment_ani(),
                                     match_containment_ani: match_.match_containment_ani(),

src/lib.rs

@@ -140,7 +140,7 @@ fn do_fastmultigather(
             query_filenames,
             againstfile_path,
             selection.clone(),
-            threshold_bp,
+            threshold_bp as u32,
             output_path,
             allow_failed_sigpaths,
         ) {
@@ -157,7 +157,7 @@ fn do_fastmultigather(
         match fastmultigather::fastmultigather(
             query_filenames,
             siglist_path,
-            threshold_bp,
+            threshold_bp as u32,
             scaled,
             selection,
             allow_failed_sigpaths,

src/manysearch.rs

@@ -205,8 +205,8 @@ fn downsample_and_inflate_abundances(
     against: &KmerMinHash,
 ) -> Result<
     (
-        Option<usize>,
-        Option<usize>,
+        Option<u64>,
+        Option<u64>,
         Option<f64>,
         Option<f64>,
         Option<f64>,
@@ -218,7 +218,7 @@ fn downsample_and_inflate_abundances(
 
     let abunds: Vec<u64>;
     let sum_weighted: u64;
-    let sum_all_abunds: usize;
+    let sum_all_abunds: u64;
 
     // avoid downsampling if we can
     if against_scaled != query_scaled {
@@ -227,10 +227,10 @@ fn downsample_and_inflate_abundances(
             .downsample_scaled(query.scaled())
             .expect("cannot downsample sketch");
         (abunds, sum_weighted) = query.inflated_abundances(&against_ds)?;
-        sum_all_abunds = against_ds.sum_abunds() as usize;
+        sum_all_abunds = against_ds.sum_abunds();
     } else {
         (abunds, sum_weighted) = query.inflated_abundances(against)?;
-        sum_all_abunds = against.sum_abunds() as usize;
+        sum_all_abunds = against.sum_abunds();
     }
 
     let average_abund = sum_weighted as f64 / abunds.len() as f64;
@@ -239,7 +239,7 @@ fn downsample_and_inflate_abundances(
 
     Ok((
         Some(sum_all_abunds),
-        Some(sum_weighted as usize),
+        Some(sum_weighted),
         Some(average_abund),
         Some(median_abund),
         Some(std_abund),