Merge branch 'latest' into try-skipmers

src/core/CHANGELOG.md

@@ -7,6 +7,22 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [unreleased]
 
+## [0.17.2] - 2024-11-15
+
+MSRV: 1.66
+
+Changes/additions:
+
+* enforce a single scaled on a `CollectionSet` (#3397)
+* change `sig_from_record` to use scaled from `Record` to downsample (#3387)
+
+Updates:
+
+* Upgrade rocksdb to 0.22.0, bump MSRV to 1.66  (#3383)
+* Bump thiserror from 1.0.68 to 2.0.3 (#3389)
+* Bump csv from 1.3.0 to 1.3.1 (#3390)
+* Bump tempfile from 3.13.0 to 3.14.0 (#3391)
+
 ## [0.17.1] - 2024-11-11
 
 Changes/additions:

src/core/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "sourmash"
-version = "0.17.1"
+version = "0.17.2"
 authors = ["Luiz Irber <[email protected]>", "N. Tessa Pierce-Ward <[email protected]>", "C. Titus Brown <[email protected]>"]
 description = "tools for comparing biological sequences with k-mer sketches"
 repository = "https://github.com/sourmash-bio/sourmash"

src/core/src/collection.rs

@@ -7,7 +7,7 @@ use crate::encodings::Idx;
 use crate::manifest::{Manifest, Record};
 use crate::prelude::*;
 use crate::storage::{FSStorage, InnerStorage, MemStorage, SigStore, ZipStorage};
-use crate::{Error, Result};
+use crate::{Error, Result, ScaledType};
 
 #[cfg(feature = "parallel")]
 use rayon::prelude::*;
@@ -52,6 +52,11 @@ impl TryFrom<Collection> for CollectionSet {
             return Ok(Self { collection });
         };
 
+        let (min_scaled, max_scaled) = collection.min_max_scaled().expect("empty collection!?");
+        if min_scaled != max_scaled {
+            return Err(Error::MismatchScaled);
+        }
+
         collection
             .manifest
             .iter()
@@ -219,6 +224,17 @@ impl Collection {
     pub fn intersect_manifest(&mut self, mf: &Manifest) {
         self.manifest = self.manifest.intersect_manifest(mf);
     }
+
+    // CTB: question, should we do something about num here?
+    pub fn min_max_scaled(&self) -> Option<(&ScaledType, &ScaledType)> {
+        self.manifest.first().map(|first| {
+            self.manifest
+                .iter()
+                .fold((first.scaled(), first.scaled()), |f, r| {
+                    (f.0.min(r.scaled()), f.1.max(r.scaled()))
+                })
+        })
+    }
 }
 
 impl Select for Collection {
@@ -483,6 +499,50 @@ mod test {
         }
     }
 
+    #[test]
+    fn collection_from_collectionset() -> () {
+        use crate::collection::CollectionSet;
+
+        let base_path = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
+
+        let test_sigs = vec![PathBuf::from("../../tests/test-data/prot/all.zip")];
+
+        let full_paths: Vec<PathBuf> = test_sigs
+            .into_iter()
+            .map(|sig| base_path.join(sig))
+            .collect();
+
+        let collection = Collection::from_zipfile(&full_paths[0]).unwrap();
+
+        let mut selection = Selection::default();
+        selection.set_moltype(HashFunctions::Murmur64Protein);
+        selection.set_scaled(200);
+
+        let collection = collection.select(&selection).expect("should pass");
+        let (min_scaled, max_scaled) = collection.min_max_scaled().expect("not empty");
+        assert_eq!(*min_scaled, *max_scaled);
+        assert_eq!(*min_scaled, 200);
+        let _cs: CollectionSet = collection.try_into().expect("should pass");
+    }
+
+    #[test]
+    #[should_panic]
+    fn collection_from_collectionset_fail() -> () {
+        use crate::collection::CollectionSet;
+
+        let base_path = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
+
+        let test_sigs = vec![PathBuf::from("../../tests/test-data/prot/all.zip")];
+
+        let full_paths: Vec<PathBuf> = test_sigs
+            .into_iter()
+            .map(|sig| base_path.join(sig))
+            .collect();
+
+        let collection = Collection::from_zipfile(&full_paths[0]).unwrap();
+        let _cs: CollectionSet = collection.try_into().expect("should fail");
+    }
+
     #[test]
     #[cfg(all(feature = "branchwater", not(target_arch = "wasm32")))]
     fn collection_from_rocksdb_storage() -> Result<()> {