COPDS-3042: Set file permissions explicitly with cacholote (#152)

* Set file permissions explicitly with cacholote

* cleanup

Author: malmans2

cacholote/config.py

@@ -63,6 +63,7 @@ class Settings(pydantic_settings.BaseSettings):
         "application/netcdf", "application/x-grib", "application/vnd+zarr"
     ] = "application/netcdf"
     io_delete_original: bool = False
+    io_chmod: Optional[int | str] = None
     raise_all_encoding_errors: bool = False
     expiration: Optional[datetime.datetime] = None
     tag: Optional[str] = None
@@ -82,6 +83,17 @@ def validate_create_engine_kwargs(
             create_engine_kwargs["poolclass"] = getattr(sa.pool, poolclass)
         return create_engine_kwargs
 
+    @pydantic.field_validator("io_chmod")
+    def validate_io_chmod(
+        cls: pydantic_settings.BaseSettings, io_chmod: Optional[int]
+    ) -> Optional[int]:
+        if isinstance(io_chmod, str):
+            try:
+                return int(io_chmod, 8)
+            except ValueError:
+                pass
+        return io_chmod
+
     @pydantic.field_validator("expiration")
     def validate_expiration(
         cls: pydantic_settings.BaseSettings, expiration: datetime.datetime | None

cacholote/extra_encoders.py

@@ -374,6 +374,9 @@ def _store_file_object(
                 with fs_out.open(urlpath_out, "wb") as f_out:
                     utils.copy_buffered_file(f_in, f_out)
 
+    if (mode := config.get().io_chmod) is not None:
+        fs_out.chmod(urlpath_out, mode)
+
     if io_delete_original and fs_in.exists(urlpath_in):
         with _logging_timer(
             "remove",

tests/test_50_io_encoder.py

@@ -249,3 +249,28 @@ def cached_in_place_open(path: str) -> io.FileIO:
     output = cached_in_place_open(str(tmpfile))
     assert isinstance(output, fsspec.implementations.local.LocalFileOpener)
     assert output.name == str(tmpfile)
+
+
+@pytest.mark.parametrize(
+    "io_chmod,expected",
+    [
+        (None, 0o100644),
+        (0o100600, 0o100600),
+        ("100600", 0o100600),
+        ("600", 0o100600),
+    ],
+)
+def test_chmod(
+    tmp_path: pathlib.Path,
+    io_chmod: int | None,
+    expected: int,
+) -> None:
+    fs, _ = utils.get_cache_files_fs_dirname()
+
+    # Cache file
+    filename = tmp_path / "test.txt"
+    fsspec.filesystem("file").pipe_file(filename, b"test")
+    with config.set(raise_all_encoding_errors=True, io_chmod=io_chmod):
+        cached_file = cached_open(filename)
+
+    assert fs.info(cached_file)["mode"] == expected