Merge branch 'main' of github.com:zarr-developers/zarr-python into feat/v2-v3-codecs

Author: d-v-b

src/zarr/core/dtype/npy/common.py

@@ -58,6 +58,12 @@
 IntishFloat = NewType("IntishFloat", float)
 """A type for floats that represent integers, like 1.0 (but not 1.1)."""
 
+IntishStr = NewType("IntishStr", str)
+"""A type for strings that represent integers, like "0" or "42"."""
+
+FloatishStr = NewType("FloatishStr", str)
+"""A type for strings that represent floats, like "3.14" or "-2.5"."""
+
 NumpyEndiannessStr = Literal[">", "<", "="]
 NUMPY_ENDIANNESS_STR: Final = ">", "<", "="
 
@@ -488,6 +494,59 @@ def check_json_intish_float(data: JSON) -> TypeGuard[IntishFloat]:
     return isinstance(data, float) and data.is_integer()
 
 
+def check_json_intish_str(data: JSON) -> TypeGuard[IntishStr]:
+    """
+    Check if a JSON value is a string that represents an integer, like "0", "42", or "-5".
+
+    Parameters
+    ----------
+    data : JSON
+        The JSON value to check.
+
+    Returns
+    -------
+    bool
+        True if the data is a string representing an integer, False otherwise.
+    """
+    if not isinstance(data, str):
+        return False
+
+    try:
+        int(data)
+    except ValueError:
+        return False
+    else:
+        return True
+
+
+def check_json_floatish_str(data: JSON) -> TypeGuard[FloatishStr]:
+    """
+    Check if a JSON value is a string that represents a float, like "3.14", "-2.5", or "0.0".
+
+    Note: This function is intended to be used AFTER check_json_float_v2/v3, so it only
+    handles regular string representations that those functions don't cover.
+
+    Parameters
+    ----------
+    data : JSON
+        The JSON value to check.
+
+    Returns
+    -------
+    bool
+        True if the data is a string representing a regular float, False otherwise.
+    """
+    if not isinstance(data, str):
+        return False
+
+    try:
+        float(data)
+    except ValueError:
+        return False
+    else:
+        return True
+
+
 def check_json_str(data: JSON) -> TypeGuard[str]:
     """
     Check if a JSON value is a string.

src/zarr/core/dtype/npy/float.py

@@ -19,6 +19,7 @@
     TFloatScalar_co,
     check_json_float_v2,
     check_json_float_v3,
+    check_json_floatish_str,
     endianness_to_numpy_str,
     float_from_json_v2,
     float_from_json_v3,
@@ -270,13 +271,17 @@ def from_json_scalar(self, data: JSON, *, zarr_format: ZarrFormat) -> TFloatScal
         if zarr_format == 2:
             if check_json_float_v2(data):
                 return self._cast_scalar_unchecked(float_from_json_v2(data))
+            elif check_json_floatish_str(data):
+                return self._cast_scalar_unchecked(float(data))
             else:
                 raise TypeError(
                     f"Invalid type: {data}. Expected a float or a special string encoding of a float."
                 )
         elif zarr_format == 3:
             if check_json_float_v3(data):
                 return self._cast_scalar_unchecked(float_from_json_v3(data))
+            elif check_json_floatish_str(data):
+                return self._cast_scalar_unchecked(float(data))
             else:
                 raise TypeError(
                     f"Invalid type: {data}. Expected a float or a special string encoding of a float."

src/zarr/core/dtype/npy/int.py

@@ -26,6 +26,7 @@
 from zarr.core.dtype.npy.common import (
     check_json_int,
     check_json_intish_float,
+    check_json_intish_str,
     endianness_to_numpy_str,
     get_endianness_from_numpy_dtype,
 )
@@ -209,6 +210,10 @@ def from_json_scalar(self, data: JSON, *, zarr_format: ZarrFormat) -> TIntScalar
             return self._cast_scalar_unchecked(data)
         if check_json_intish_float(data):
             return self._cast_scalar_unchecked(int(data))
+
+        if check_json_intish_str(data):
+            return self._cast_scalar_unchecked(int(data))
+
         raise TypeError(f"Invalid type: {data}. Expected an integer.")
 
     def to_json_scalar(self, data: object, *, zarr_format: ZarrFormat) -> int:

tests/test_dtype/test_npy/test_float.py

@@ -167,3 +167,47 @@ class TestFloat64(_BaseTestFloat):
         ("0x3ff0000000000000", 1.0),
     )
     item_size_params = (Float64(),)
+
+
+def test_check_json_floatish_str() -> None:
+    """Test the check_json_floatish_str function."""
+    from zarr.core.dtype.npy.common import check_json_floatish_str
+
+    # Test valid string floats
+    assert check_json_floatish_str("3.14")
+    assert check_json_floatish_str("0.0")
+    assert check_json_floatish_str("-2.5")
+    assert check_json_floatish_str("1.0")
+
+    # Test invalid cases
+    assert not check_json_floatish_str("not_a_number")
+    assert not check_json_floatish_str("")
+    assert not check_json_floatish_str(3.14)  # actual float, not string
+    assert not check_json_floatish_str(42)  # int
+    assert not check_json_floatish_str(None)
+
+    # Test that special cases still work via float() conversion
+    # (these will be handled by existing functions first in practice)
+    assert check_json_floatish_str("NaN")
+    assert check_json_floatish_str("Infinity")
+    assert check_json_floatish_str("-Infinity")
+
+
+def test_string_float_from_json_scalar() -> None:
+    """Test that string representations of floats can be parsed by from_json_scalar."""
+    # Test with Float32
+    dtype_instance = Float32()
+    result = dtype_instance.from_json_scalar("3.14", zarr_format=3)
+    assert abs(result - np.float32(3.14)) < 1e-6
+    assert isinstance(result, np.float32)
+
+    # Test other cases
+    result = dtype_instance.from_json_scalar("0.0", zarr_format=3)
+    assert result == np.float32(0.0)
+
+    result = dtype_instance.from_json_scalar("-2.5", zarr_format=3)
+    assert result == np.float32(-2.5)
+
+    # Test that it works for v2 format too
+    result = dtype_instance.from_json_scalar("1.5", zarr_format=2)
+    assert result == np.float32(1.5)

tests/test_dtype/test_npy/test_int.py

@@ -281,3 +281,42 @@ class TestUInt64(BaseTestZDType):
     )
     invalid_scalar_params = ((UInt64(), {"set!"}), (UInt64(), ("tuple",)))
     item_size_params = (UInt64(),)
+
+
+def test_check_json_intish_str() -> None:
+    """Test the check_json_intish_str function."""
+    from zarr.core.dtype.npy.common import check_json_intish_str
+
+    # Test valid string integers
+    assert check_json_intish_str("0")
+    assert check_json_intish_str("42")
+    assert check_json_intish_str("-5")
+    assert check_json_intish_str("123")
+
+    # Test invalid cases
+    assert not check_json_intish_str("3.14")
+    assert not check_json_intish_str("not_a_number")
+    assert not check_json_intish_str("")
+    assert not check_json_intish_str(42)  # actual int, not string
+    assert not check_json_intish_str(3.14)  # float
+    assert not check_json_intish_str(None)
+
+
+def test_string_integer_from_json_scalar() -> None:
+    """Test that string representations of integers can be parsed by from_json_scalar."""
+    # Test the specific reproducer case
+    dtype_instance = Int32()
+    result = dtype_instance.from_json_scalar("0", zarr_format=3)
+    assert result == np.int32(0)
+    assert isinstance(result, np.int32)
+
+    # Test other cases
+    result = dtype_instance.from_json_scalar("42", zarr_format=3)
+    assert result == np.int32(42)
+
+    result = dtype_instance.from_json_scalar("-5", zarr_format=3)
+    assert result == np.int32(-5)
+
+    # Test that it works for v2 format too
+    result = dtype_instance.from_json_scalar("123", zarr_format=2)
+    assert result == np.int32(123)