Specify utf-8 explicitly on behalf of windows

ctypesgen/printer_python/printer.py

@@ -32,7 +32,7 @@ class WrapperPrinter:
     def __init__(self, outpath, options, data, argv):
         outpath = Path(outpath).resolve()
         status_message(f"Writing to {outpath}.")
-        self.file = outpath.open("w")
+        self.file = outpath.open("w", encoding="utf-8")
 
         try:
             self.options = options
@@ -122,9 +122,7 @@ def print_info(self, argv):
         # TODO(py38) consider shlex.join()
         argv_str = ' '.join([f'"{a}"' if ' ' in a else a for a in argv])
         argv_str = self._strip_private_paths(argv_str)
-        self.file.write(
-            f"# Auto-generated by: ctypesgen {argv_str}\n"
-        )
+        self.file.write(f"# Auto-generated by: ctypesgen {argv_str}\n")
 
 
     def print_preamble(self):
@@ -267,16 +265,16 @@ def print_enum(self, enum):
         self.file.write("enum_%s = c_int" % enum.tag)
 
 
-    def print_typedef(self, typedef):
-        self.srcinfo(typedef.src, wants_nl=False)
-        self.file.write("%s = %s" % (typedef.name, typedef.ctype.py_string()))
-
-
     def print_constant(self, constant):
         self.srcinfo(constant.src, wants_nl=False)
         self.file.write("%s = %s" % (constant.name, constant.value.py_string(False)))
 
 
+    def print_typedef(self, typedef):
+        self.srcinfo(typedef.src, wants_nl=False)
+        self.file.write("%s = %s" % (typedef.name, typedef.ctype.py_string()))
+
+
     def print_variable(self, variable):
         assert self.options.library, "Binary symbol requires --library LIBNAME"
         self.srcinfo(variable.src)

tests/ctypesgentest.py

@@ -47,18 +47,20 @@ def module_from_code(name, python_code):
 
 def generate(header_str, args=[], lang="py"):
 
-    # use custom tempfiles scoping so we may retain data for inspection
-    # also note that python stdlib tempfiles don't play well with windows
+    # Use custom tempfiles scoping so we may retain data for inspection
+    # Windows notes:
+    # - Avoid stdlib tempfiles, they're not usable by anyone except the direct creator, otherwise you'll get permission errors.
+    # - The default file encoding seems to be cp1252, which is problematic with special chars (such as the banana in the constants test). Need to specify UTF-8 explicitly. PEP 686 should hopefully improve this.
 
     global COUNTER
     COUNTER += 1
 
     tmp_in = TMP_DIR/f"in_header_{COUNTER:02d}.h"
-    tmp_in.write_text(header_str.strip() + "\n")
+    tmp_in.write_text(header_str.strip() + "\n", encoding="utf-8")
     try:
         tmp_out = TMP_DIR/f"out_bindings_{COUNTER:02d}.{lang}"
         ctypesgen_main(["-i", tmp_in, "-o", tmp_out, "--output-language", lang, *args])
-        content = tmp_out.read_text()
+        content = tmp_out.read_text(encoding="utf-8")
     finally:
         if CLEANUP_OK:
             tmp_in.unlink()