relaxes typed set to allow unused fspaths (will be dropped on copy if trimmed is provided)

Author: tclose

fileformats/generic/set.py

@@ -1,10 +1,9 @@
 import typing as ty
 import itertools
 from pathlib import Path
-from fileformats.core import FileSet, validated_property
+from fileformats.core import FileSet
 from fileformats.core.mixin import WithClassifiers
 from fileformats.core.collection import TypedCollection
-from fileformats.core.exceptions import FormatMismatchError
 
 
 class TypedSet(TypedCollection):
@@ -27,16 +26,8 @@ def __repr__(self) -> str:
             paths_repr += ", ..."
         return f"{self.type_name}({paths_repr})"
 
-    @validated_property
-    def _all_paths_used(self) -> None:
-        all_contents_paths = set(itertools.chain(*(c.fspaths for c in self.contents)))
-        missing = self.fspaths - all_contents_paths
-        if missing:
-            contents_str = "\n".join(repr(c) for c in self.contents)
-            raise FormatMismatchError(
-                f"Paths {[str(p) for p in missing]} are not used by any of the "
-                f"contents of {self.type_name}:\n{contents_str}"
-            )
+    def required_paths(self) -> ty.FrozenSet[Path]:
+        return frozenset(itertools.chain(*(c.required_paths() for c in self.contents)))
 
 
 class SetOf(WithClassifiers, TypedSet):  # type: ignore[misc]

fileformats/generic/tests/test_directory.py renamed to fileformats/generic/tests/test_collection.py

@@ -34,9 +34,11 @@ def test_directory_optional_contents(tmp_path):
 
     optional_dir = DirectoryOf[MyFormatGz, ty.Optional[YourFormat]](sample_dir)
     assert optional_dir.contents == [my_format]
+
     your_format = YourFormat.sample(dest_dir=tmp_path)
     optional_dir = DirectoryOf[MyFormatGz, ty.Optional[YourFormat]](sample_dir)
     assert optional_dir.contents == [my_format, your_format]
+
     required_dir = DirectoryOf[MyFormatGz, YourFormat](sample_dir)
     assert required_dir.contents == [my_format, your_format]
 
@@ -47,15 +49,24 @@ def test_set_optional_contents():
 
     sample_set = SetOf[MyFormatGz, YourFormat](my_format, your_format)
     assert sample_set.contents == [my_format, your_format]
-    with pytest.raises(
-        FormatMismatchError, match="are not used by any of the contents of "
-    ):
-        SetOf[MyFormatGz](my_format, your_format)
+    assert set(sample_set.required_paths()) == {my_format.fspath, your_format.fspath}
+
+    sample_set = SetOf[MyFormatGz](my_format, your_format)
+    assert list(sample_set.required_paths()) == [my_format.fspath]
+
     with pytest.raises(
         FormatMismatchError, match="Did not find the required content types"
     ):
         SetOf[MyFormatGz, YourFormat](my_format)
+
     sample_set = SetOf[MyFormatGz, ty.Optional[YourFormat]](my_format)
     assert sample_set.contents == [my_format]
+    assert list(sample_set.required_paths()) == [my_format.fspath]
+
     sample_set = SetOf[MyFormatGz, ty.Optional[YourFormat]](my_format, your_format)
     assert sample_set.contents == [my_format, your_format]
+    assert set(sample_set.required_paths()) == {my_format.fspath, your_format.fspath}
+
+    sample_set = SetOf[ty.Optional[MyFormatGz]](my_format)
+    assert sample_set.contents == [my_format]
+    assert list(sample_set.required_paths()) == [my_format.fspath]