Add Unicode support for SBDF path handling

spotfire/sbdf_helpers.c

@@ -8,33 +8,51 @@
 /* Utility functions for opening FILE pointers Pythonically from ``Union[str,bytes,int]``
  * (similar in behavior to Python's open() function).
  */
-FILE *_pathlike_to_fileptr(PyObject *file, const char* mode) {
+FILE *_pathlike_to_fileptr(PyObject *file, const char *mode) {
     FILE *the_file = NULL;
     int fd;
-    char *filename;
+#if defined(_WIN32)
+    wchar_t wmode[8];
+#endif
     PyObject *filename_obj;
 
     /* int: use the given file descriptor */
-    if(PyLong_Check(file)) {
+    if (PyLong_Check(file)) {
         fd = PyObject_AsFileDescriptor(file);
-        if(fd == -1) return NULL;
+        if (fd == -1) return NULL;
         the_file = fdopen(fd, mode);
-    /* bytes: use the given file name */
-    } else if(PyBytes_Check(file)) {
-        filename = PyBytes_AsString(file);
+#if defined(_WIN32)
+    } else if (PyUnicode_Check(file)) {
+        filename_obj = PyUnicode_AsWideCharString(file, NULL);
+        if (!filename_obj) return NULL;
+        mbstowcs(wmode, mode, sizeof(wmode)/sizeof(wchar_t));
+        the_file = _wfopen((wchar_t *)filename_obj, wmode);
+        PyMem_Free(filename_obj);
+    } else if (PyBytes_Check(file)) {
+        PyObject *unicode_obj = PyUnicode_FromEncodedObject(file, "utf-8", "strict");
+        if (!unicode_obj) return NULL;
+        filename_obj = PyUnicode_AsWideCharString(unicode_obj, NULL);
+        Py_DECREF(unicode_obj);
+        if (!filename_obj) return NULL;
+        mbstowcs(wmode, mode, sizeof(wmode)/sizeof(wchar_t));
+        the_file = _wfopen((wchar_t *)filename_obj, wmode);
+        PyMem_Free(filename_obj);
+#else
+    } else if (PyBytes_Check(file)) {
+        const char *filename = PyBytes_AsString(file);
         the_file = fopen(filename, mode);
     /* unicode/str: decode the given filename as utf-8 */
-    } else if(PyUnicode_Check(file)) {
-        if(!PyUnicode_FSConverter(file, &filename_obj)) return NULL;
-        filename = PyBytes_AsString(filename_obj);
+    } else if (PyUnicode_Check(file)) {
+        if (!PyUnicode_FSConverter(file, &filename_obj)) return NULL;
+        const char *filename = PyBytes_AsString(filename_obj);
         the_file = fopen(filename, mode);
         Py_XDECREF(filename_obj);
-    /* else: raise an exception */
+#endif
     } else {
         PyErr_SetString(PyExc_TypeError, "str, bytes, or integer argument expected");
     }
 
-    if(the_file == NULL) {
+    if (the_file == NULL) {
         PyErr_SetFromErrno(PyExc_IOError);
     }
     return the_file;

spotfire/test/test_sbdf.py

@@ -1,9 +1,11 @@
 """Tests for importing and exporting data to SBDF files."""
 
+from pathlib import Path
 import datetime
 import decimal
 import unittest
 import tempfile
+import os
 import pkg_resources
 
 import pandas
@@ -474,3 +476,21 @@ def test_image_pil(self):
             self.assertEqual(df2.columns[0], 'x')
             val = df2.at[0, "x"]
             self.assertEqual(val[0:8], b'\x89PNG\x0d\x0a\x1a\x0a')
+
+    def test_export_import_unicode_path(self):
+        """Test export and import with a Unicode file path."""
+        dataframe = pandas.DataFrame({"col": [1, 2, 3], "txt": ["a", "b", "c"]})
+        with tempfile.TemporaryDirectory() as tempdir:
+            unicode_filename = Path(tempdir) / "日本語ファイル" / "test.sbdf"
+            os.makedirs(os.path.dirname(unicode_filename), exist_ok=True)
+            # Export to Unicode path
+            sbdf.export_data(dataframe, str(unicode_filename))
+
+            # Import from Unicode path
+            imported = sbdf.import_data(str(unicode_filename))
+
+            # Check roundtrip
+            pandas.testing.assert_frame_equal(imported[["col", "txt"]], dataframe, check_dtype=False)
+            # Check dtype of the column
+            self.assertEqual(dataframe["col"].dtype, "int64")
+            self.assertEqual(dataframe["txt"].dtype, "object")