From 5afff95f4c93dbd05d019da4c50ca4f070aa9d03 Mon Sep 17 00:00:00 2001
From: Jerome Kelleher <jk@well.ox.ac.uk>
Date: Sun, 14 Jul 2024 21:58:43 +0100
Subject: [PATCH] Initial start on cpython interface testing

---
 tests/test_cpython_interface.py | 53 +++++++++++++++++++++++++++++++++
 vcztools/_vcztoolsmodule.c      |  6 ++--
 2 files changed, 56 insertions(+), 3 deletions(-)
 create mode 100644 tests/test_cpython_interface.py

diff --git a/tests/test_cpython_interface.py b/tests/test_cpython_interface.py
new file mode 100644
index 0000000..f406c5a
--- /dev/null
+++ b/tests/test_cpython_interface.py
@@ -0,0 +1,53 @@
+import pytest
+import numpy as np
+
+from vcztools import _vcztools
+
+FIXED_FIELD_NAMES = ["chrom", "pos", "id", "qual", "ref", "alt", "filter"]
+
+
+def example_fixed_data(num_variants, num_samples=0):
+    chrom = np.array(["X"] * num_variants, dtype="S")
+    pos = np.arange(num_variants, dtype=np.int32)
+    id = np.array(["."] * num_variants, dtype="S").reshape((num_variants, 1))
+    ref = np.array(["A"] * num_variants, dtype="S")
+    alt = np.array(["T"] * num_variants, dtype="S").reshape((num_variants, 1))
+    qual = np.arange(num_variants, dtype=np.float32)
+    filter_ = np.ones(num_variants, dtype=bool).reshape((num_variants, 1))
+    filter_id = np.array(["PASS"], dtype="S")
+    return {
+        "chrom": chrom,
+        "pos": pos,
+        "id": id,
+        "qual": qual,
+        "ref": ref,
+        "alt": alt,
+        "filter": filter_,
+        "filter_ids": filter_id,
+    }
+
+
+class TestTypeChecking:
+    @pytest.mark.parametrize("name", FIXED_FIELD_NAMES)
+    def test_field_incorrect_length(self, name):
+        num_variants = 5
+        data = example_fixed_data(num_variants)
+        data[name] = data[name][1:]
+        with pytest.raises(ValueError, match=f"Array {name.upper()} must have "):
+            _vcztools.VcfEncoder(num_variants, 0, **data)
+
+    @pytest.mark.parametrize("name", FIXED_FIELD_NAMES)
+    def test_field_incorrect_dtype(self, name):
+        num_variants = 5
+        data = example_fixed_data(num_variants)
+        data[name] = np.zeros(data[name].shape, dtype=np.int64)
+        with pytest.raises(ValueError, match=f"Wrong dtype for {name.upper()}"):
+            _vcztools.VcfEncoder(num_variants, 0, **data)
+
+    @pytest.mark.parametrize("name", FIXED_FIELD_NAMES)
+    def test_field_incorrect_type(self, name):
+        num_variants = 5
+        data = example_fixed_data(num_variants)
+        data[name] = "A Python string"
+        with pytest.raises(TypeError, match=f"must be numpy.ndarray"):
+            _vcztools.VcfEncoder(num_variants, 0, **data)
diff --git a/vcztools/_vcztoolsmodule.c b/vcztools/_vcztoolsmodule.c
index 077b782..e9c484d 100644
--- a/vcztools/_vcztoolsmodule.c
+++ b/vcztools/_vcztoolsmodule.c
@@ -119,7 +119,7 @@ VcfEncoder_store_fixed_array(VcfEncoder *self, PyArrayObject *array, const char
         goto out;
     }
     if (PyArray_DTYPE(array)->type_num != type) {
-        PyErr_Format(PyExc_ValueError, "Array %s is not of the correct type", name);
+        PyErr_Format(PyExc_ValueError, "Wrong dtype for %s", name);
         goto out;
     }
 
@@ -189,7 +189,7 @@ VcfEncoder_init(VcfEncoder *self, PyObject *args, PyObject *kwds)
         goto out;
     }
 
-    /* NOTE: we generalise this pattern for CHROM also to save a bit of time
+    /* NOTE: we could generalise this pattern for CHROM also to save a bit of time
      * in building numpy String arrays */
     assert(PyArray_CheckExact(filter_ids));
     if (!PyArray_CHKFLAGS(filter_ids, NPY_ARRAY_IN_ARRAY)) {
@@ -209,7 +209,7 @@ VcfEncoder_init(VcfEncoder *self, PyObject *args, PyObject *kwds)
     if (VcfEncoder_add_array(self, "IDS/", "FILTER", filter_ids) != 0) {
         goto out;
     }
-    if (VcfEncoder_store_fixed_array(self, filter, "filter", NPY_BOOL, 2, num_variants)
+    if (VcfEncoder_store_fixed_array(self, filter, "FILTER", NPY_BOOL, 2, num_variants)
         != 0) {
         goto out;
     }