Add _print_to_console utility function to handle non-utf-8 console encodings (#16)

Author: DanielNoord

pydocstringformatter/run.py

@@ -85,4 +85,4 @@ def _format_files(self, filepaths: List[Path]) -> None:
             is_changed = self._format_file(file) or is_changed
 
         if not is_changed:
-            print("Nothing to do! All docstrings are correct 🎉")
+            utils._print_to_console("Nothing to do! All docstrings are correct 🎉\n")

pydocstringformatter/utils/__init__.py

@@ -11,6 +11,7 @@
 from pydocstringformatter.utils.file_diference import _generate_diff
 from pydocstringformatter.utils.find_docstrings import _is_docstring
 from pydocstringformatter.utils.find_python_file import _find_python_files
+from pydocstringformatter.utils.output import _print_to_console
 
 __all__ = [
     "_find_python_files",
@@ -22,4 +23,5 @@
     "PydocstringFormatterError",
     "_register_arguments",
     "TomlParsingError",
+    "_print_to_console",
 ]

pydocstringformatter/utils/output.py

@@ -0,0 +1,20 @@
+import sys
+
+
+def _encode_string(string: str) -> bytes:
+    """Encode a string to utf-8.
+
+    This can be used to circumvent the issue of the standard encoding
+    of a windows console not being utf-8.
+    See: https://github.com/DanielNoord/pydocstringformatter/issues/13
+    """
+    return string.encode("utf-8")
+
+
+def _print_to_console(string: str) -> None:
+    """Print a string to the console while handling edge cases.
+
+    This should be used instead of print() whenever we want to
+    print emoji's or non-ASCII characters.
+    """
+    sys.stdout.buffer.write(_encode_string(string))

tests/test_utils.py

@@ -1,7 +1,11 @@
+import sys
 import tokenize
 from pathlib import Path
 from typing import List, Tuple
 
+import pytest
+
+import pydocstringformatter
 from pydocstringformatter.utils import _find_python_files, _is_docstring
 
 HERE = Path(__file__)
@@ -133,3 +137,22 @@ def test_module_docstrings(self) -> None:
                     docstrings.append((tokeninfo.start, tokeninfo.end))
 
         assert docstrings == [((3, 0), (4, 3))]
+
+
+def test_encoding_of_console_messages(
+    capsys: pytest.CaptureFixture[str], test_file: str
+) -> None:
+    """Test that we can print emoji's to non utf-8 consoles.
+
+    Regression test for:
+    https://github.com/DanielNoord/pydocstringformatter/issues/13
+    """
+    sys.stdout.reconfigure(encoding="cp1252")  # type: ignore[attr-defined]
+    with open(test_file, "w", encoding="utf-8") as file:
+        file.write('"""A multi-line\ndocstring\n"""')
+
+    pydocstringformatter.run_docstring_formatter([test_file, "--write"])
+
+    output = capsys.readouterr()
+    assert output.out == "Nothing to do! All docstrings are correct 🎉\n"
+    assert not output.err