feat: add BZIP2 (.bz2) support for reading fasta and fastq (#185)

python/biobear/biobear.pyi

@@ -30,6 +30,7 @@ class FileCompressionType(enum.Enum):
     GZIP = 0
     BGZIP = 1
     NONE = 2
+    BZIP2 = 3
 
 class FastaSequenceDataType(enum.Enum):
     """How to treat the sequence data in a FASTA file."""

python/biobear/compression.py

@@ -25,12 +25,15 @@ class Compression(Enum):
     INFERRED = "INFERRED"
     NONE = "NONE"
     GZIP = "GZIP"
+    BZIP2 = "BZIP2"
 
     @classmethod
     def from_file(cls, path: os.PathLike) -> "Compression":
         """Infer the compression type from the file extension."""
         if Path(path).suffix == ".gz":
             return Compression.GZIP
+        if Path(path).suffix == ".bz2":
+            return Compression.BZIP2
         return Compression.NONE
 
     def infer_or_use(self, path: os.PathLike) -> "Compression":

python/tests/test_session.py

@@ -139,6 +139,14 @@ def test_read_fastq():
 
     assert len(df) == 2
 
+    fastq_path = DATA / "test.fq.bz2"
+    options = FASTQReadOptions(
+        file_extension="fq", file_compression_type=FileCompressionType.BZIP2
+    )
+
+    df = session.read_fastq_file(str(fastq_path), options=options).to_polars()
+
+    assert len(df) == 2
 
 @pytest.mark.skipif(
     not importlib.util.find_spec("polars"), reason="polars not installed"
@@ -285,6 +293,18 @@ def test_read_fasta_gz():
 
     assert len(df) == 2
 
+def test_read_fasta_bz2():
+    """Test reading a fasta.bz2 file."""
+    session = connect()
+
+    fasta_path = DATA / "test.fa.bz2"
+
+    options = FASTAReadOptions(
+        file_extension="fa", file_compression_type=FileCompressionType.BZIP2
+    )
+    df = session.read_fasta_file(str(fasta_path), options=options).to_polars()
+
+    assert len(df) == 2
 
 @pytest.mark.skipif(
     not importlib.util.find_spec("polars"), reason="polars not installed"

src/file_compression_type.rs

@@ -28,6 +28,7 @@ pub enum FileCompressionType {
     GZIP,
     ZSTD,
     UNCOMPRESSED,
+    BZIP2,
 }
 
 impl Default for FileCompressionType {
@@ -56,6 +57,7 @@ impl Display for FileCompressionType {
             Self::GZIP => write!(f, "GZIP"),
             Self::ZSTD => write!(f, "ZSTD"),
             Self::UNCOMPRESSED => write!(f, "UNCOMPRESSED"),
+            Self::BZIP2 => write!(f, "BZIP2"),
         }
     }
 }
@@ -66,6 +68,7 @@ impl From<FileCompressionType> for DFFileCompressionType {
             FileCompressionType::GZIP => DFFileCompressionType::GZIP,
             FileCompressionType::ZSTD => DFFileCompressionType::ZSTD,
             FileCompressionType::UNCOMPRESSED => DFFileCompressionType::UNCOMPRESSED,
+            FileCompressionType::BZIP2 => DFFileCompressionType::BZIP2,
         }
     }
 }
@@ -78,6 +81,7 @@ impl TryFrom<CompressionTypeVariant> for FileCompressionType {
             CompressionTypeVariant::GZIP => Ok(Self::GZIP),
             CompressionTypeVariant::ZSTD => Ok(Self::ZSTD),
             CompressionTypeVariant::UNCOMPRESSED => Ok(Self::UNCOMPRESSED),
+            CompressionTypeVariant::BZIP2 => Ok(Self::BZIP2),
             _ => Err(BioBearError::InvalidCompressionType(value.to_string())),
         }
     }
@@ -91,6 +95,7 @@ impl TryFrom<DFFileCompressionType> for FileCompressionType {
             DFFileCompressionType::GZIP => Ok(Self::GZIP),
             DFFileCompressionType::ZSTD => Ok(Self::ZSTD),
             DFFileCompressionType::UNCOMPRESSED => Ok(Self::UNCOMPRESSED),
+            DFFileCompressionType::BZIP2 => Ok(Self::BZIP2),
             _ => Err(BioBearError::InvalidCompressionType(
                 "Invalid compression type".to_string(),
             )),