codespell_lib/_codespell.py: read additional args from file with with… (#3841)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: julian-smith-artifex-com

README.rst

@@ -229,6 +229,13 @@ instead of these invalid entries:
 
 .. _tomli: https://pypi.org/project/tomli/
 
+Reading arguments from file
+---------------------------
+
+Additional arguments can be read from a file with ``@PATH``. Arguments are
+extracted using ``shlex.split()``.
+
+
 pre-commit hook
 ---------------
 

codespell_lib/_codespell.py

@@ -23,6 +23,7 @@
 import itertools
 import os
 import re
+import shlex
 import sys
 import textwrap
 from collections.abc import Iterable, Sequence
@@ -388,7 +389,27 @@ def _supports_ansi_colors() -> bool:
 def parse_options(
     args: Sequence[str],
 ) -> tuple[argparse.Namespace, argparse.ArgumentParser, list[str]]:
-    parser = argparse.ArgumentParser(formatter_class=NewlineHelpFormatter)
+    # Split lines read from `@PATH` using shlex.split(), otherwise default
+    # behaviour is to have one arg per line. See:
+    # https://docs.python.org/3/library/argparse.html#argparse.ArgumentParser.convert_arg_line_to_args
+    class ArgumentParser2(argparse.ArgumentParser):
+        def convert_arg_line_to_args(self, arg_line: str) -> list[str]:
+            if sys.platform == "win32":
+                # On Windows, shlex.split() seems to be messed up by back
+                # slashes. Temporarily changing them to forward slashes seems
+                # to make things work better.
+                arg_line = arg_line.replace("\\", "/")
+                ret = shlex.split(arg_line)
+                ret = [p.replace("/", "\\") for p in ret]
+            else:
+                ret = shlex.split(arg_line)
+            return ret
+
+    parser = ArgumentParser2(
+        formatter_class=NewlineHelpFormatter,
+        fromfile_prefix_chars="@",
+        epilog="Use @PATH to read additional arguments from file PATH.",
+    )
 
     parser.set_defaults(colors=_supports_ansi_colors())
     parser.add_argument("--version", action="version", version=VERSION)

codespell_lib/tests/test_basic.py

@@ -1483,3 +1483,44 @@ def test_stdin(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
         code, stdout, _ = result
         assert stdout == "1: Thsi ==> This\n"
         assert code == 1
+
+
+def test_args_from_file(
+    tmp_path: Path,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    import textwrap
+
+    print()
+    fname1 = tmp_path / "tmp1"
+    fname2 = tmp_path / "tmp2"
+    fname3 = tmp_path / "tmp3"
+    fname_list = tmp_path / "tmp_list"
+    fname_list.write_text(f"{fname1} {fname2}\n{fname3}")
+    fname1.write_text("abandonned\ncode")
+    fname2.write_text("exmaple\n")
+    fname3.write_text("abilty\n")
+    print(f"{fname_list=}")
+    args = ["codespell", f"@{fname_list}"]
+    print(f"Running: {args=}")
+    cp = subprocess.run(  # noqa: S603
+        args,
+        check=False,
+        text=True,
+        capture_output=True,
+    )
+    code = cp.returncode
+    stdout = cp.stdout
+    stderr = cp.stderr
+    print(f"{code=}")
+    print(f"stdout:\n{textwrap.indent(stdout, '    ')}")
+    print(f"stderr:\n{textwrap.indent(stderr, '    ')}")
+    assert "tmp1:1: abandonned ==> abandoned\n" in stdout, f"{stdout=}"
+    assert "tmp2:1: exmaple ==> example\n" in stdout, f"{stdout=}"
+    assert "tmp3:1: abilty ==> ability\n" in stdout, f"{stdout=}"
+    assert code, f"{code=}"
+
+    # Run same test via cs_.main() so code coverage checks work.
+    print("Testing with direct call to cs_.main()")
+    r = cs_.main(*args[1:])
+    print(f"{r=}")