feat: add support for nupy array shape validation

Author: sg495

test/test_00_validate.py

@@ -695,6 +695,27 @@ def test_numpy_array_error() -> None:
     with pytest.raises(TypeError):
         validate(val, npt.NDArray[np.str_])
 
+def test_numpy_array_shape() -> None:
+    # pylint: disable = import-outside-toplevel
+    import numpy as np
+    val = np.zeros(5, dtype=np.uint8)
+    validate(val, np.ndarray[typing.Any, np.dtype[np.uint8]])
+    validate(val, np.ndarray[tuple, np.dtype[np.uint8]])
+    validate(val, np.ndarray[tuple[typing.Any, ...], np.dtype[np.uint8]])
+    validate(val, np.ndarray[tuple[typing.Any], np.dtype[np.uint8]])
+    validate(val, np.ndarray[tuple[int, ...], np.dtype[np.uint8]])
+    validate(val, np.ndarray[tuple[int], np.dtype[np.uint8]])
+    validate(val, np.ndarray[tuple[Literal[5], ...], np.dtype[np.uint8]])
+    validate(val, np.ndarray[tuple[Literal[5]], np.dtype[np.uint8]])
+    with pytest.raises(TypeError):
+        validate(val, np.ndarray[tuple[int, int], np.dtype[np.uint8]])
+    with pytest.raises(TypeError):
+        validate(val, np.ndarray[tuple[typing.Any, typing.Any], np.dtype[np.uint8]])
+    with pytest.raises(TypeError):
+        validate(val, np.ndarray[tuple[Literal[5], int], np.dtype[np.uint8]])
+    with pytest.raises(TypeError):
+        validate(val, np.ndarray[tuple[int, Literal[5]], np.dtype[np.uint8]])
+
 
 def test_typevar() -> None:
     T = typing.TypeVar("T")

typing_validation/validation.py

@@ -228,7 +228,7 @@ def _unsupported_type_error(
     """
     Error for unsupported types, with optional explanation.
     """
-    msg = "Unsupported validation for type {t!r}."
+    msg = f"Unsupported validation for type {t!r}."
     if explanation is not None:
         msg += " " + explanation
     return UnsupportedTypeError(msg)
@@ -658,6 +658,7 @@ def _extract_dtypes(t: Any) -> typing.Sequence[Any]:
 
 def _validate_numpy_array(val: Any, t: Any) -> None:
     import numpy as np  # pylint: disable = import-outside-toplevel
+
     if not isinstance(val, TypeInspector):
         _validate_type(val, np.ndarray)
     # assert hasattr(t, "__args__"), _missing_args_msg(t)
@@ -676,7 +677,7 @@ def _validate_numpy_array(val: Any, t: Any) -> None:
             val._record_unsupported_type(t)
             return
         raise _unsupported_type_error(
-            t, f"Unsupported NumPy dtype {dtype_t!r}"
+            t, f"Unsupported NumPy dtype {dtype_t!r}."
         ) from None
     if isinstance(val, TypeInspector):
         val._record_pending_type_generic(t.__origin__)
@@ -685,9 +686,10 @@ def _validate_numpy_array(val: Any, t: Any) -> None:
             validate(val, arg)
         return
     val_dtype = val.dtype
-    if any(dtype is Any or np.issubdtype(val_dtype, dtype) for dtype in dtypes):
-        return
-    raise _numpy_dtype_error(val, t)
+    if not any(dtype is Any or np.issubdtype(val_dtype, dtype) for dtype in dtypes):
+        raise _numpy_dtype_error(val, t)
+    validate(val.shape, t__args__[0])
+    return
 
 
 def _validate_typevar(val: Any, t: TypeVar) -> None: