From 21d1384e42de953bba5f4bb892fe6fec97597526 Mon Sep 17 00:00:00 2001
From: Josh Chorlton <josh@bugseq.com>
Date: Thu, 2 Oct 2025 22:06:14 +0000
Subject: [PATCH 1/3] optimize performance of array_to_qualitystring

---
 pysam/libcutils.pyx      | 18 ++++++++++++------
 tests/libcutils_bench.py | 10 ++++++++++
 tests/libcutils_test.py  | 10 ++++++++++
 3 files changed, 32 insertions(+), 6 deletions(-)
 create mode 100644 tests/libcutils_bench.py
 create mode 100644 tests/libcutils_test.py

diff --git a/pysam/libcutils.pyx b/pysam/libcutils.pyx
index 7a03f3c44..322cf6757 100644
--- a/pysam/libcutils.pyx
+++ b/pysam/libcutils.pyx
@@ -46,14 +46,20 @@ cpdef array_to_qualitystring(c_array.array qualities, int offset=33):
     """convert an array of quality values to a string."""
     if qualities is None:
         return None
-    cdef int x
 
-    cdef c_array.array result
-    result = c_array.clone(qualities, len(qualities), zero=False)
+    cdef Py_ssize_t n = len(qualities)
 
-    for x from 0 <= x < len(qualities):
-        result[x] = qualities[x] + offset
-    return force_str(result.tobytes())
+    cdef bytearray result_ba = bytearray(n)
+
+    cdef char[:] qualities_view = qualities
+    cdef unsigned char[:] result_view = result_ba
+
+    cdef Py_ssize_t i
+
+    for i in range(n):
+        result_view[i] = qualities_view[i] + offset
+
+    return force_str(bytes(result_ba))
 
 
 cpdef qualities_to_qualitystring(qualities, int offset=33):
diff --git a/tests/libcutils_bench.py b/tests/libcutils_bench.py
new file mode 100644
index 000000000..98d469b7a
--- /dev/null
+++ b/tests/libcutils_bench.py
@@ -0,0 +1,10 @@
+"""Benchmarking the libcutils module. Usage::
+
+pytest tests/libcutils_bench.py
+"""
+import pysam
+
+
+def test_qualitystring_to_array_long_sequences(benchmark):
+    result = benchmark(pysam.array_to_qualitystring, pysam.qualitystring_to_array("123") * 500)
+    assert result == "123" * 500
diff --git a/tests/libcutils_test.py b/tests/libcutils_test.py
new file mode 100644
index 000000000..0febafaa7
--- /dev/null
+++ b/tests/libcutils_test.py
@@ -0,0 +1,10 @@
+"""Benchmarking the libcutils module. Usage::
+
+pytest tests/libcutils_bench.py
+"""
+import pysam
+
+
+def test_qualitystring_to_array_empty():
+    result = pysam.array_to_qualitystring(pysam.qualitystring_to_array(""))
+    assert result == ""

From e0c8a4221adcdf642df91be996e4179279662680 Mon Sep 17 00:00:00 2001
From: John Marshall <jmarshall@hey.com>
Date: Fri, 17 Apr 2026 20:25:45 +1200
Subject: [PATCH 2/3] Extend tests and incorporate into AlignedSegment_test.py

---
 tests/AlignedSegment_test.py | 22 ++++++++++++++++++++++
 tests/libcutils_bench.py     |  2 +-
 tests/libcutils_test.py      | 10 ----------
 3 files changed, 23 insertions(+), 11 deletions(-)
 delete mode 100644 tests/libcutils_test.py

diff --git a/tests/AlignedSegment_test.py b/tests/AlignedSegment_test.py
index 0e850600a..567640580 100644
--- a/tests/AlignedSegment_test.py
+++ b/tests/AlignedSegment_test.py
@@ -1901,5 +1901,27 @@ def test_string_export_import_with_tags(self):
         self.assertEqual(a, b)
 
 
+class TestArrayUtilities(unittest.TestCase):
+    def test_array_to_qualstr(self):
+        data = [
+            "",
+            "Q",
+            """!"#$%&'()*+,-./012...xyz{|}~""",
+            ">>?AB",
+            "ABDDEFGHIJabcdefghij",
+            "ACAFFGGFFFJDFJHHJIJIHKGGHKHHIJHHHJ7123" * 50,
+        ]
+
+        for qual in data:
+            qual_array = pysam.qualitystring_to_array(qual)
+            result = pysam.array_to_qualitystring(qual_array)
+            self.assertEqual(result, qual)
+
+    def test_longarray_to_qualstr(self):
+        qual_array = array.array('l', [64, 65, 66, 67, 68])
+        with self.assertRaises(ValueError):
+            pysam.array_to_qualitystring(qual_array)
+
+
 if __name__ == "__main__":
     unittest.main()
diff --git a/tests/libcutils_bench.py b/tests/libcutils_bench.py
index 98d469b7a..c85983066 100644
--- a/tests/libcutils_bench.py
+++ b/tests/libcutils_bench.py
@@ -5,6 +5,6 @@
 import pysam
 
 
-def test_qualitystring_to_array_long_sequences(benchmark):
+def test_array_to_qualitystring_long_sequences(benchmark):
     result = benchmark(pysam.array_to_qualitystring, pysam.qualitystring_to_array("123") * 500)
     assert result == "123" * 500
diff --git a/tests/libcutils_test.py b/tests/libcutils_test.py
deleted file mode 100644
index 0febafaa7..000000000
--- a/tests/libcutils_test.py
+++ /dev/null
@@ -1,10 +0,0 @@
-"""Benchmarking the libcutils module. Usage::
-
-pytest tests/libcutils_bench.py
-"""
-import pysam
-
-
-def test_qualitystring_to_array_empty():
-    result = pysam.array_to_qualitystring(pysam.qualitystring_to_array(""))
-    assert result == ""

From 24abe663d9fd85611d65fda03b1b75d55dfed36f Mon Sep 17 00:00:00 2001
From: John Marshall <jmarshall@hey.com>
Date: Fri, 17 Apr 2026 20:48:37 +1200
Subject: [PATCH 3/3] Improve memoryview version of array_to_qualitystring()

The data is contiguous so use [::1] to omit stride calculations;
use size_t rather than ssize_t to omit check for end-relative indexing.
---
 pysam/libcutils.pyx | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/pysam/libcutils.pyx b/pysam/libcutils.pyx
index 322cf6757..002df096d 100644
--- a/pysam/libcutils.pyx
+++ b/pysam/libcutils.pyx
@@ -47,14 +47,13 @@ cpdef array_to_qualitystring(c_array.array qualities, int offset=33):
     if qualities is None:
         return None
 
-    cdef Py_ssize_t n = len(qualities)
+    cdef const unsigned char[::1] qualities_view = qualities
+    cdef size_t n = qualities_view.shape[0]
 
     cdef bytearray result_ba = bytearray(n)
+    cdef unsigned char[::1] result_view = result_ba
 
-    cdef char[:] qualities_view = qualities
-    cdef unsigned char[:] result_view = result_ba
-
-    cdef Py_ssize_t i
+    cdef size_t i
 
     for i in range(n):
         result_view[i] = qualities_view[i] + offset