pydiverse-common 0.3.10__tar.gz → 0.3.12__tar.gz

{pydiverse_common-0.3.10 → pydiverse_common-0.3.12}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pydiverse-common
-Version: 0.3.10
+Version: 0.3.12
 Summary: Common functionality shared between pydiverse libraries
 Author: QuantCo, Inc.
 Author-email: Martin Trautmann <windiana@users.sf.net>, Finn Rudolph <finn.rudolph@t-online.de>

{pydiverse_common-0.3.10 → pydiverse_common-0.3.12}/docs/source/changelog.md

@@ -1,5 +1,12 @@
 # Changelog
 
+## 0.3.12 (2025-08-21)
+- fixed Dtype.to_pandas() for parameterized String and Decimal types
+
+## 0.3.11 (2025-08-21)
+- fixed __eq__, __hash__, and __repr__ for types with parameters
+- import string length from sqlalchemy VARCHAR(n) type
+
 ## 0.3.10 (2025-08-21)
 - implemented String with max_length argument for SQL VARCHAR(n) generation
 - implemented Decimal with precision and scale arguments

{pydiverse_common-0.3.10 → pydiverse_common-0.3.12}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pydiverse-common"
-version = "0.3.10"
+version = "0.3.12"
 description = "Common functionality shared between pydiverse libraries"
 authors = [
   { name = "QuantCo, Inc." },

{pydiverse_common-0.3.10 → pydiverse_common-0.3.12}/src/pydiverse/common/dtypes.py

@@ -23,6 +23,10 @@ class Dtype:
         """Return a string representation of this dtype."""
         return self.__class__.__name__
 
+    def __str__(self):
+        """Return a string representation of this dtype."""
+        return self.__repr__()
+
     @classmethod
     def is_int(cls):
         """Return ``True`` if this dtype is an integer type."""
@@ -71,7 +75,7 @@ class Dtype:
             # otherwise.
             return Float64()
         if isinstance(sql_type, sa.String):
-            return String()
+            return String(sql_type.length)
         if isinstance(sql_type, sa.Boolean):
             return Bool()
         if isinstance(sql_type, sa.Date):
@@ -307,6 +311,15 @@ class Dtype:
 
         if isinstance(self, Enum):
             return pd.CategoricalDtype(self.categories)
+        if isinstance(self, String):
+            return pd.StringDtype()  # max_length not needed for dataframes
+        if isinstance(self, Decimal):
+            # NumericDtype exists but is not used because fixpoint is more common
+            # in SQL than in dataframes.
+            return pd.Float64Dtype()
+        if isinstance(self, List):
+            # we don't want to produce object columns
+            raise TypeError("pandas doesn't have a native list dtype")
 
         return {
             Int(): pd.Int64Dtype(),  # we default to 64 bit
@@ -321,8 +334,6 @@ class Dtype:
             Float(): pd.Float64Dtype(),  # we default to 64 bit
             Float32(): pd.Float32Dtype(),
             Float64(): pd.Float64Dtype(),
-            Decimal(): pd.Float64Dtype(),  # NumericDtype exists but is not used
-            String(): pd.StringDtype(),
             Bool(): pd.BooleanDtype(),
             Date(): "datetime64[s]",
             Datetime(): "datetime64[us]",
@@ -416,6 +427,20 @@ class Decimal(Float):
         self.precision = precision or 31
         self.scale = scale or (self.precision // 3 + 1)
 
+    def __eq__(self, rhs):
+        return (
+            isinstance(rhs, self.__class__)
+            and self.precision == rhs.precision
+            and self.scale == rhs.scale
+        )
+
+    def __hash__(self):
+        return hash((self.__class__.__name__, self.precision, self.scale))
+
+    def __repr__(self):
+        """Return a string representation of this dtype."""
+        return f"{self.__class__.__name__}({self.precision}, {self.scale})"
+
     def to_sql(self):
         import sqlalchemy as sa
 
@@ -480,6 +505,16 @@ class String(Dtype):
         """
         self.max_length = max_length
 
+    def __eq__(self, rhs):
+        return isinstance(rhs, self.__class__) and self.max_length == rhs.max_length
+
+    def __hash__(self):
+        return hash((self.__class__.__name__, self.max_length))
+
+    def __repr__(self):
+        """Return a string representation of this dtype."""
+        return f"{self.__class__.__name__}({self.max_length})"
+
     def to_sql(self):
         """Convert this Dtype to a SQL type."""
         import sqlalchemy as sa
@@ -523,10 +558,11 @@ class List(Dtype):
         return isinstance(rhs, List) and self.inner == rhs.inner
 
     def __hash__(self):
-        return hash((0, hash(self.inner)))
+        return hash((self.__class__.__name__, hash(self.inner)))
 
     def __repr__(self):
-        return f"List[{repr(self.inner)}]"
+        """Return a string representation of this dtype."""
+        return f"{self.__class__.__name__}[{self.inner}]"
 
     def to_sql(self):
         import sqlalchemy as sa
@@ -556,11 +592,13 @@ class Enum(String):
     def __eq__(self, rhs):
         return isinstance(rhs, Enum) and self.categories == rhs.categories
 
-    def __repr__(self) -> str:
-        return f"Enum[{', '.join(repr(c) for c in self.categories)}]"
-
     def __hash__(self):
-        return hash(tuple(self.categories))
+        return hash((self.__class__.__name__, tuple(self.categories)))
+
+    def __repr__(self) -> str:
+        return (
+            f"{self.__class__.__name__}[{', '.join(repr(c) for c in self.categories)}]"
+        )
 
     def to_polars(self):
         import polars as pl

{pydiverse_common-0.3.10 → pydiverse_common-0.3.12}/tests/dtypes/test_dtype_pandas.py

@@ -113,7 +113,15 @@ def test_dtype_to_pandas_numpy():
     assert_conversion(UInt16(), pd.UInt16Dtype())
     assert_conversion(UInt8(), pd.UInt8Dtype())
 
+    assert_conversion(Float64(), pd.Float64Dtype())
+    assert_conversion(Float32(), pd.Float32Dtype())
+
+    assert_conversion(Decimal(), pd.Float64Dtype())
+    assert_conversion(Decimal(15), pd.Float64Dtype())
+    assert_conversion(Decimal(15, 2), pd.Float64Dtype())
+
     assert_conversion(String(), pd.StringDtype())
+    assert_conversion(String(10), pd.StringDtype())
     assert_conversion(Bool(), pd.BooleanDtype())
 
     assert_conversion(Date(), "datetime64[s]")
@@ -142,7 +150,17 @@ def test_dtype_to_pandas_pyarrow():
     assert_conversion(UInt16(), pa.uint16())
     assert_conversion(UInt8(), pa.uint8())
 
+    assert_conversion(Float64(), pa.float64())
+    assert_conversion(Float32(), pa.float32())
+
+    assert_conversion(Decimal(), pa.decimal128(31, 11))
+    assert_conversion(Decimal(76), pa.decimal256(76, 76 // 3 + 1))
+    assert_conversion(Decimal(15), pa.decimal64(15, 6))
+    assert_conversion(Decimal(18, 2), pa.decimal64(18, 2))
+    assert_conversion(Decimal(9, 9), pa.decimal32(9, 9))
+
     assert_conversion(String(), pd.StringDtype(storage="pyarrow"))
+    assert_conversion(String(10), pd.StringDtype(storage="pyarrow"))
     assert_conversion(Bool(), pa.bool_())
 
     assert_conversion(Date(), pa.date32())

{pydiverse_common-0.3.10 → pydiverse_common-0.3.12}/tests/dtypes/test_dtype_sqlalchemy.py

@@ -111,7 +111,7 @@ def test_all_types(type_):
     if type_ is pdc.List:
         type_obj = type_(pdc.Int64())
     elif type_ is pdc.Enum:
-        type_obj = type_("a", "b", "c")
+        type_obj = type_("a", "bbb", "cc")
     else:
         type_obj = type_()
     dst_type = type_obj.to_sql()
@@ -129,6 +129,6 @@ def test_all_types(type_):
         Float: Float64(),
         Int: Int64(),
         # there is no Enum
-        Enum: String(),
+        Enum: String(3),
     }
     assert back_type == acceptance_map.get(type_, type_obj)