Merge pull request #67 from ngoldbaum/static-pandas-stringdtype

Add a statically defined PandasStringDType

Author: ngoldbaum

.github/workflows/ci.yml

@@ -19,7 +19,7 @@ jobs:
       - name: Install build and test dependencies
         run: |
           pip install -i https://pypi.anaconda.org/scientific-python-nightly-wheels/simple numpy
-          python -m pip install -U pip build pytest unyt wheel meson ninja meson-python patchelf
+          python -m pip install -U pip build pytest unyt wheel meson ninja meson-python patchelf pandas
       - name: Install asciidtype
         working-directory: asciidtype
         run: |
@@ -78,3 +78,5 @@ jobs:
         working-directory: stringdtype
         run: |
           ASAN_OPTIONS=detect_leaks=false LD_PRELOAD=/usr/lib/gcc/x86_64-linux-gnu/11/libasan.so pytest -s -vvv --color=yes
+          pip uninstall -y pandas
+          ASAN_OPTIONS=detect_leaks=false LD_PRELOAD=/usr/lib/gcc/x86_64-linux-gnu/11/libasan.so pytest -s -vvv --color=yes

stringdtype/pyproject.toml

@@ -11,6 +11,10 @@ build-backend = "mesonpy"
 [tool.black]
 line-length = 79
 
+[tool.isort]
+profile = "black"
+line_length = 79
+
 [project]
 name = "stringdtype"
 description = "A dtype for storing UTF-8 strings"

stringdtype/stringdtype/__init__.py

@@ -3,12 +3,23 @@
 """
 
 from .missing import NA  # isort: skip
-from .scalar import StringScalar  # isort: skip
+from .scalar import StringScalar, PandasStringScalar  # isort: skip
 from ._main import StringDType, _memory_usage
 
+try:
+    from ._main import PandasStringDType
+except ImportError:
+    PandasStringDType = None
+
 __all__ = [
     "NA",
     "StringDType",
     "StringScalar",
     "_memory_usage",
 ]
+
+# this happens when pandas isn't importable
+if PandasStringDType is None:
+    del PandasStringDType
+else:
+    __all__.extend("PandasStringDType")

stringdtype/stringdtype/scalar.py

@@ -1,5 +1,9 @@
-"""A scalar type needed by the dtype machinery."""
+"""Scalar types needed by the dtype machinery."""
 
 
 class StringScalar(str):
     pass
+
+
+class PandasStringScalar(str):
+    pass

stringdtype/stringdtype/src/casts.c

@@ -80,17 +80,18 @@ static char *s2s_name = "cast_StringDType_to_StringDType";
 
 static NPY_CASTING
 unicode_to_string_resolve_descriptors(PyObject *NPY_UNUSED(self),
-                                      PyArray_DTypeMeta *NPY_UNUSED(dtypes[2]),
+                                      PyArray_DTypeMeta *dtypes[2],
                                       PyArray_Descr *given_descrs[2],
                                       PyArray_Descr *loop_descrs[2],
                                       npy_intp *NPY_UNUSED(view_offset))
 {
     if (given_descrs[1] == NULL) {
-        StringDTypeObject *new = new_stringdtype_instance(NA_OBJ);
+        PyArray_Descr *new = (PyArray_Descr *)new_stringdtype_instance(
+                (PyTypeObject *)dtypes[1]);
         if (new == NULL) {
             return (NPY_CASTING)-1;
         }
-        loop_descrs[1] = (PyArray_Descr *)new;
+        loop_descrs[1] = new;
     }
     else {
         Py_INCREF(given_descrs[1]);