Demonstrates that max-hits can be specified with values greater than a short size

pybwa/libbwaindex.pxd

@@ -43,7 +43,6 @@ cdef extern from "bntseq.h":
     void bns_destroy(bntseq_t *bns)
 
 cdef bytes force_bytes(object s)
-cdef bytes force_bytes_with(object s, encoding: str | None = *, errors: str | None = *)
 cpdef bint _set_bwa_idx_verbosity(int level)
 
 cdef class BwaIndex:

pybwa/libbwaindex.pyx

@@ -28,19 +28,13 @@ cdef str TEXT_ENCODING = 'utf-8'
 
 
 cdef bytes force_bytes(object s):
-    return force_bytes_with(s, None, None)
-
-
-cdef bytes force_bytes_with(
-    object s, encoding: str | None = None, errors: str | None = None
-): # pragma: no cover
     """convert string or unicode object to bytes, assuming utf8 encoding."""
     if s is None:
         return None
     elif PyBytes_Check(s):
         return s
     elif PyUnicode_Check(s):
-        return s.encode(encoding or TEXT_ENCODING, errors or ERROR_HANDLER)
+        return s.encode(TEXT_ENCODING, ERROR_HANDLER)
     else:
         raise TypeError("Argument must be a string, bytes or unicode.")
 

pyproject.toml

@@ -95,7 +95,7 @@ sphinx-rtd-theme = ">=3.0.2"
 sphinx-autodoc-typehints = "2"
 
 [build-system]
-requires = ["poetry-core>=2.1.1", "cython>=3.0.11", "setuptools>=75.1.0", "pysam >=0.22.1"]
+requires = ["poetry-core>=2.1.1", "cython>=3.0.11", "setuptools>=75.1.0", "pysam ==0.22.1"]
 build-backend = "poetry.core.masonry.api"
 
 [tool.poe]

tests/_test_libbwaindex.pyx

@@ -1,12 +1,12 @@
 # cython: language_level=3
 
-from pybwa.libbwaindex cimport force_bytes_with
+from pybwa.libbwaindex cimport force_bytes
 import pytest
 
 
-def test_force_bytes_with() -> None:
-    assert force_bytes_with(None) is None
-    assert force_bytes_with(b"foo-bar") == b"foo-bar"
-    assert force_bytes_with("foo-bar") == b"foo-bar"
+def test_force_bytes() -> None:
+    assert force_bytes(None) is None
+    assert force_bytes(b"foo-bar") == b"foo-bar"
+    assert force_bytes("foo-bar") == b"foo-bar"
     with pytest.raises(TypeError, match="Argument must be a string, bytes or unicode"):
-        force_bytes_with(42)
+        force_bytes(42)

tests/test_libbwaindex.py

@@ -11,10 +11,10 @@
 import_test_lib(libname="libbwaindex")
 
 
-def test_force_bytes_with() -> None:
+def test_force_bytes() -> None:
     import _test_libbwaindex
 
-    _test_libbwaindex.test_force_bytes_with()
+    _test_libbwaindex.test_force_bytes()
 
 
 @pytest.mark.parametrize("in_place", [True, False])

tests/test_libbwaln.py

@@ -373,3 +373,27 @@ def test_bwa_aln_ambiguous_bases(num_amb: int, tmp_path_factory: pytest.TempPath
     else:
         assert rec.has_tag("XN"), str(rec)
         assert rec.get_tag("XN") == num_amb
+
+
+@pytest.mark.parametrize("limit", [1, 32767, 32768, 100000])
+def test_bwa_aln_max_hits(limit: int, tmp_path_factory: pytest.TempPathFactory) -> None:
+    query = "G" * 25
+    target = ("A" * 25) + query + ("T" * 25)
+
+    src_dir = Path(str(tmp_path_factory.mktemp("test_bwa_aln_max_hits")))
+    fasta = src_dir / "ref.fasta"
+    with fasta.open("w") as writer:
+        writer.write(">ref\n")
+        for _ in range(limit):
+            writer.write(f"{target}\n")
+    BwaIndex.index(fasta=fasta)
+
+    bwa = BwaAln(prefix=fasta)
+    queries = [FastxRecord(name="NA", sequence=query)]
+    options = BwaAlnOptions(seed_length=len(query), max_mismatches=0, max_hits=limit)
+    recs = bwa.align(queries=queries, opt=options)
+    assert len(recs) == 1
+    rec = recs[0]
+    print(rec)
+    assert rec.is_unmapped is False
+    assert rec.get_tag("HN") == limit, f"limit was {limit}"