Merge branch 'latest' into py-skipmers

sourmash-bio · Dec 20, 2024 · b63d666 · b63d666
2 parents 246a88f + 5680efc
commit b63d666
Show file tree

Hide file tree

Showing 5 changed files with 159 additions and 49 deletions.
diff --git a/include/sourmash.h b/include/sourmash.h
@@ -40,6 +40,9 @@ enum SourmashErrorCode {
   SOURMASH_ERROR_CODE_INVALID_PROT = 1102,
   SOURMASH_ERROR_CODE_INVALID_CODON_LENGTH = 1103,
   SOURMASH_ERROR_CODE_INVALID_HASH_FUNCTION = 1104,
+  SOURMASH_ERROR_CODE_INVALID_SKIPMER_FRAME = 1105,
+  SOURMASH_ERROR_CODE_INVALID_SKIPMER_SIZE = 1106,
+  SOURMASH_ERROR_CODE_INVALID_TRANSLATE_FRAME = 1107,
   SOURMASH_ERROR_CODE_READ_DATA = 1201,
   SOURMASH_ERROR_CODE_STORAGE = 1202,
   SOURMASH_ERROR_CODE_HLL_PRECISION_BOUNDS = 1301,

diff --git a/src/core/src/errors.rs b/src/core/src/errors.rs
@@ -55,6 +55,15 @@ pub enum SourmashError {
     #[error("Codon is invalid length: {message}")]
     InvalidCodonLength { message: String },
 
+    #[error("Skipmer ksize must be >= n ({n}), but got ksize: {ksize}")]
+    InvalidSkipmerSize { ksize: usize, n: usize },
+
+    #[error("Skipmer frame number must be < n ({n}), but got start: {start}")]
+    InvalidSkipmerFrame { start: usize, n: usize },
+
+    #[error("Frame number must be 0, 1, or 2, but got {frame_number}")]
+    InvalidTranslateFrame { frame_number: usize },
+
     #[error("Set error rate to a value smaller than 0.367696 and larger than 0.00203125")]
     HLLPrecisionBounds,
 
@@ -128,6 +137,9 @@ pub enum SourmashErrorCode {
     InvalidProt = 11_02,
     InvalidCodonLength = 11_03,
     InvalidHashFunction = 11_04,
+    InvalidSkipmerFrame = 11_05,
+    InvalidSkipmerSize = 11_06,
+    InvalidTranslateFrame = 11_07,
     // index-related errors
     ReadData = 12_01,
     Storage = 12_02,
@@ -170,6 +182,9 @@ impl SourmashErrorCode {
             SourmashError::InvalidProt { .. } => SourmashErrorCode::InvalidProt,
             SourmashError::InvalidCodonLength { .. } => SourmashErrorCode::InvalidCodonLength,
             SourmashError::InvalidHashFunction { .. } => SourmashErrorCode::InvalidHashFunction,
+            SourmashError::InvalidSkipmerFrame { .. } => SourmashErrorCode::InvalidSkipmerFrame,
+            SourmashError::InvalidSkipmerSize { .. } => SourmashErrorCode::InvalidSkipmerSize,
+            SourmashError::InvalidTranslateFrame { .. } => SourmashErrorCode::InvalidTranslateFrame,
             SourmashError::ReadDataError { .. } => SourmashErrorCode::ReadData,
             SourmashError::StorageError { .. } => SourmashErrorCode::Storage,
             SourmashError::HLLPrecisionBounds { .. } => SourmashErrorCode::HLLPrecisionBounds,

diff --git a/src/core/src/ffi/minhash.rs b/src/core/src/ffi/minhash.rs
@@ -73,15 +73,26 @@ Result<*const u64> {
 
     let mut output: Vec<u64> = Vec::with_capacity(insize);
 
+    // Call SeqToHashes::new and handle errors
+    let ready_hashes = SeqToHashes::new(
+        buf,
+        mh.ksize(),
+        force,
+        is_protein,
+        mh.hash_function(),
+        mh.seed(),
+    )?;
+
+
     if force && bad_kmers_as_zeroes{
-        for hash_value in SeqToHashes::new(buf, mh.ksize(), force, is_protein, mh.hash_function(), mh.seed()){
+        for hash_value in ready_hashes{
             match hash_value{
                 Ok(x) => output.push(x),
                 Err(err) => return Err(err),
             }
         }
     }else{
-        for hash_value in SeqToHashes::new(buf, mh.ksize(), force, is_protein, mh.hash_function(), mh.seed()){
+        for hash_value in ready_hashes {
             match hash_value{
                 Ok(0) => continue,
                 Ok(x) => output.push(x),