#16 Improve file handling

Author: roskakori

pyproject.toml

@@ -99,22 +99,6 @@ ignore = [
     "PLR0915",
     # Magic value used in comparison
     "PLR2004",
-    # TODO Enable checks for usage of pathlib.
-    "PTH100",
-    "PTH103",
-    "PTH107",
-    "PTH109",
-    "PTH110",
-    "PTH112",
-    "PTH114",
-    "PTH118",
-    "PTH119",
-    "PTH120",
-    "PTH122",
-    "PTH123",
-    "PTH202",
-    "PTH207",
-    "PTH208",
     # Unneccesarry assign → We regularly use `result = ...; return result` to examine the result in the debugger.
     "RET504",
     # Mutable class attributes should be annotated with `typing.ClassVar`

sanpo/command.py

@@ -4,6 +4,7 @@
 import glob
 import logging
 import sys
+from pathlib import Path
 
 from . import __version__, log
 from .sanitize import sanitize_file
@@ -19,6 +20,7 @@ def _parsed_args(args=None):
         #  However, when passing multiple files this results in
         #  ResourceWarning: unclosed file <_io.TextIOWrapper...>.
         nargs="+",
+        type=Path,
         help='PO file(s) to sanitize; use glob patterns to process multiple files; use "**" for recursive scans',
     )
     return parser.parse_args(args)
@@ -27,13 +29,20 @@ def _parsed_args(args=None):
 def main_without_logging_setup(args=None) -> int:
     result = 0
     po_path = None
+    actual_args = [(str(arg) if isinstance(arg, Path) else arg) for arg in args]
     sanitized_file_count = 0
-    arguments = _parsed_args(args)
+    arguments = _parsed_args(actual_args)
     try:
         for po_path_pattern in arguments.po_paths:
-            for po_path in glob.iglob(po_path_pattern, recursive=True):
-                sanitize_file(po_path)
+            if po_path_pattern.is_file():
+                sanitize_file(po_path_pattern)
                 sanitized_file_count += 1
+            else:
+                for po_path in glob.iglob(str(po_path_pattern), recursive=True):  # noqa: PTH207
+                    # There does not seem to be a way for `path.glob(pattern)` to separate `path` from the `pattern`,
+                    # while `iglob(path_with_pattern)` can process a single combined path and pattern.
+                    sanitize_file(po_path)
+                    sanitized_file_count += 1
     except Exception as error:
         log.error('cannot sanitize "%s": %s', po_path, error)
         result = 1

sanpo/sanitize.py

@@ -2,6 +2,8 @@
 # All rights reserved. Distributed under the BSD 3-Clause License.
 import logging
 import re
+from collections.abc import Iterable
+from pathlib import Path
 
 _LINE_PATTERNS_TO_REMOVE = [
     re.compile(line)
@@ -20,16 +22,17 @@
 _log = logging.getLogger("sanpo")
 
 
-def sanitize_file(po_path: str):
+def sanitize_file(po_path: Path):
+    assert isinstance(po_path, Path), f"po_path must be a Path: {po_path!r}"
     _log.info("sanitizing %s", po_path)
-    with open(po_path, encoding="utf-8") as po_file:
-        lines_to_write = list(sanitize_lines(po_file))
-    with open(po_path, "w", encoding="utf-8") as po_file:
+    with po_path.open(encoding="utf-8") as po_file:
+        lines_to_write = list(sanitized_lines(po_file))
+    with po_path.open("w", encoding="utf-8") as po_file:
         for line_to_write in lines_to_write:
             po_file.write(line_to_write)
 
 
-def sanitize_lines(source_lines):
+def sanitized_lines(source_lines: Iterable[str]) -> Iterable[str]:
     for line in source_lines:
         line_has_to_be_removed = any(
             line_pattern_to_remove.match(line) for line_pattern_to_remove in _LINE_PATTERNS_TO_REMOVE

tests/_common.py

@@ -1,10 +1,10 @@
 # Copyright (c) 2021, Thomas Aglassinger
 # All rights reserved. Distributed under the BSD 3-Clause License.
-import os
+from pathlib import Path
 from typing import Optional
 from unittest import TestCase
 
-TEST_TEMP_FOLDER = os.path.join(os.path.dirname(__file__), "temp")
+TEST_TEMP_FOLDER = Path(__file__).parent / "temp"
 
 
 class PoFileTest(TestCase):
@@ -26,17 +26,14 @@ def write_po_file(self, test_name: str):
             'msgid "spam"',
             'msgstr ""',
         ]
-        self.po_path = os.path.join(TEST_TEMP_FOLDER, test_name + ".po")
-        with open(self.po_path, "w", encoding="utf-8") as po_file:
-            for po_line in po_lines:
-                po_file.write(f"{po_line}\n")
+        self.po_path = TEST_TEMP_FOLDER / (test_name + ".po")
+        self.po_path.write_text("\n".join(po_lines) + "\n")
         self._po_paths_to_remove.append(self.po_path)
 
-    def po_lines(self, po_path: Optional[str] = None) -> list[str]:
+    def po_lines(self, po_path: Optional[Path] = None) -> list[str]:
         actual_po_path = po_path if po_path is not None else self.po_path
-        with open(actual_po_path, encoding="utf-8") as po_file:
-            return [line.rstrip("\n") for line in po_file]
+        return actual_po_path.read_text().split("\n")
 
     def tearDown(self):
         for po_path_to_remove in self._po_paths_to_remove:
-            os.remove(po_path_to_remove)
+            po_path_to_remove.unlink()

tests/test_sanitize.py

@@ -2,7 +2,7 @@
 # All rights reserved. Distributed under the BSD 3-Clause License.
 from unittest import TestCase
 
-from sanpo.sanitize import sanitize_file, sanitize_lines
+from sanpo.sanitize import sanitize_file, sanitized_lines
 from tests._common import PoFileTest
 
 
@@ -17,7 +17,7 @@ def test_can_remove_pointless_lines(self):
             r'"Language-Team: LANGUAGE [email protected]\n"',
             "PRESERVE",
         ]
-        self.assertEqual(list(sanitize_lines(source_lines)), ["PRESERVE"])
+        self.assertEqual(list(sanitized_lines(source_lines)), ["PRESERVE"])
 
     def test_can_preserve_metadata_lines(self):
         source_lines = [
@@ -26,7 +26,7 @@ def test_can_preserve_metadata_lines(self):
             r'"Last-Translator: John Doe [email protected]\n"',
             r'"Language-Team: en [email protected]\n"',
         ]
-        self.assertEqual(list(sanitize_lines(source_lines)), source_lines)
+        self.assertEqual(list(sanitized_lines(source_lines)), source_lines)
 
 
 class TransformFileTest(PoFileTest):