refactor(geoarrow-types,geoarrow-pyarrow): Use the geoarrow-types extension definitions in geoarrow-pyarrow (#54)

Author: paleolimbot

geoarrow-pandas/src/geoarrow/pandas/lib.py

@@ -3,7 +3,7 @@
 import pyarrow as _pa
 import pyarrow_hotfix as _  # noqa: F401
 import numpy as _np
-from geoarrow.c import lib
+from geoarrow.types import TypeSpec, type_spec, Encoding
 import geoarrow.pyarrow as _ga
 
 
@@ -83,8 +83,7 @@ class GeoArrowExtensionArray(_pd.api.extensions.ExtensionArray):
     def __init__(self, obj, type=None):
         if type is not None:
             self._dtype = GeoArrowExtensionDtype(type)
-            arrow_type = _ga.GeometryExtensionType._from_ctype(self._dtype._parent)
-            self._data = _ga.array(obj, arrow_type)
+            self._data = _ga.array(obj, self._dtype._parent)
         else:
             self._data = _ga.array(obj)
             self._dtype = GeoArrowExtensionDtype(self._data.type)
@@ -247,8 +246,8 @@ def to_numpy(self, dtype=None, copy=False, na_value=None):
 
         return _np.array(list(self), dtype=object)
 
-    def __array__(self, dtype=None):
-        return self.to_numpy(dtype=dtype)
+    def __array__(self, dtype=None, copy=True):
+        return self.to_numpy(dtype=dtype, copy=copy)
 
 
 @_pd.api.extensions.register_extension_dtype
@@ -271,20 +270,20 @@ class GeoArrowExtensionDtype(_pd.api.extensions.ExtensionDtype):
 
     def __init__(self, parent):
         if isinstance(parent, _ga.GeometryExtensionType):
-            self._parent = parent._type
-        elif isinstance(parent, lib.CVectorType):
             self._parent = parent
+        elif isinstance(parent, TypeSpec):
+            self._parent = _ga.extension_type(parent)
         elif isinstance(parent, GeoArrowExtensionDtype):
             self._parent = parent._parent
         else:
             raise TypeError(
-                "`geoarrow_type` must inherit from geoarrow.pyarrow.VectorType, "
-                "geoarrow.CVectorType, or geoarrow.pandas.GeoArrowExtensionDtype"
+                "`geoarrow_type` must be a pyarrow extension type, "
+                "geoarrow.types.TypeSpec, or geoarrow.pandas.GeoArrowExtensionDtype"
             )
 
     @property
     def pyarrow_dtype(self):
-        return _ga.GeometryExtensionType._from_ctype(self._parent)
+        return self._parent
 
     @property
     def type(self):
@@ -323,9 +322,9 @@ def construct_from_string(cls, string):
         if params["coord_type"] == "[interleaved]":
             coord_type = _ga.CoordType.INTERLEAVED
         elif params["type"] in ("wkt", "wkb"):
-            coord_type = _ga.CoordType.UNKNOWN
+            coord_type = _ga.CoordType.UNSPECIFIED
         else:
-            coord_type = _ga.CoordType.SEPARATE
+            coord_type = _ga.CoordType.SEPARATED
 
         if params["type"] == "point":
             geometry_type = _ga.GeometryType.POINT
@@ -347,7 +346,9 @@ def construct_from_string(cls, string):
         elif params["type"] == "wkt":
             base_type = _ga.wkt()
         else:
-            base_type = _ga.extension_type(geometry_type, dims, coord_type)
+            base_type = _ga.extension_type(
+                type_spec(Encoding.GEOARROW, geometry_type, dims, coord_type)
+            )
 
         try:
             if params["metadata"]:
@@ -368,7 +369,7 @@ def __str__(self):
         ext_name = self._parent.extension_name
         ext_dims = self._parent.dimensions
         ext_coord = self._parent.coord_type
-        ext_meta = self._parent.extension_metadata.decode("UTF-8")
+        ext_meta = self._parent.__arrow_ext_serialize__().decode("UTF-8")
 
         if ext_dims == _ga.Dimensions.XYZ:
             dims_str = "[z]"
@@ -440,7 +441,14 @@ def _wrap_series(self, array_or_chunked):
         )
 
     def _obj_is_geoarrow(self):
-        return isinstance(self._obj.dtype, GeoArrowExtensionDtype)
+        if isinstance(self._obj.dtype, GeoArrowExtensionDtype):
+            return True
+
+        if not isinstance(self._obj.dtype, _pd.ArrowDtype):
+            return False
+
+        arrow_type = self._obj.dtype.pyarrow_dtype
+        return isinstance(arrow_type, _ga.GeometryExtensionType)
 
     def parse_all(self):
         """See :func:`geoarrow.pyarrow.parse_all`"""
@@ -529,9 +537,9 @@ def with_edge_type(self, edge_type):
         """See :func:`geoarrow.pyarrow.with_edge_type`"""
         return self._wrap_series(_ga.with_edge_type(self._obj, edge_type))
 
-    def with_crs(self, crs, crs_type=None):
+    def with_crs(self, crs):
         """See :func:`geoarrow.pyarrow.with_crs`"""
-        return self._wrap_series(_ga.with_crs(self._obj, crs=crs, crs_type=crs_type))
+        return self._wrap_series(_ga.with_crs(self._obj, crs=crs))
 
     def with_dimensions(self, dimensions):
         """See :func:`geoarrow.pyarrow.with_dimensions`"""

geoarrow-pandas/tests/test_geoarrow_pandas.py

@@ -6,7 +6,6 @@
 import pyarrow as pa
 import geoarrow.pandas as gapd
 import geoarrow.pyarrow as ga
-from geoarrow.c import lib
 import numpy as np
 
 
@@ -18,10 +17,10 @@ def test_dtype_constructor():
     from_pyarrow = gapd.GeoArrowExtensionDtype(ga.point())
     assert from_pyarrow.name == "geoarrow.point"
 
-    from_ctype = gapd.GeoArrowExtensionDtype(ga.point()._type)
-    assert from_ctype.name == "geoarrow.point"
+    from_spec = gapd.GeoArrowExtensionDtype(ga.point().spec)
+    assert from_spec.name == "geoarrow.point"
 
-    from_dtype = gapd.GeoArrowExtensionDtype(from_ctype)
+    from_dtype = gapd.GeoArrowExtensionDtype(from_spec)
     assert from_dtype.name == "geoarrow.point"
 
     with pytest.raises(TypeError):
@@ -34,8 +33,8 @@ def test_dtype_strings():
     dtype2 = gapd.GeoArrowExtensionDtype.construct_from_string(str(dtype))
     assert dtype2 == dtype
 
-    dtype = gapd.GeoArrowExtensionDtype(ga.point().with_crs("EPSG:1234"))
-    assert str(dtype) == 'geoarrow.point{"crs":"EPSG:1234"}'
+    dtype = gapd.GeoArrowExtensionDtype(ga.point().with_crs(ga.OGC_CRS84))
+    assert str(dtype) == 'geoarrow.point{"crs": ' + ga.OGC_CRS84.to_json() + "}"
     dtype2 = gapd.GeoArrowExtensionDtype.construct_from_string(str(dtype))
     assert dtype2 == dtype
 
@@ -182,23 +181,10 @@ def test_array_concat():
     assert len(concatenated_diff_type) == 6
 
 
-def test_pyarrow_integration():
-    pa_array = ga.array(["POINT (0 1)", "POINT (1 2)", None])
-    series = pa_array.to_pandas()
-    assert series.dtype == gapd.GeoArrowExtensionDtype(ga.wkt())
-    assert series[0] == gapd.GeoArrowExtensionScalar("POINT (0 1)")
-    assert pa.array(series) is pa_array
-
-    pa_chunked_array = pa.chunked_array([pa_array])
-    series = pa_chunked_array.to_pandas()
-    assert series.dtype == gapd.GeoArrowExtensionDtype(ga.wkt())
-    assert series[0] == gapd.GeoArrowExtensionScalar("POINT (0 1)")
-
-
 def test_accessor_parse_all():
     series = pd.Series(["POINT (0 1)"])
     assert series.geoarrow.parse_all() is series
-    with pytest.raises(lib.GeoArrowCException):
+    with pytest.raises(Exception, match="Expected geometry type at byte 0"):
         pd.Series(["NOT WKT"]).geoarrow.parse_all()
 
 
@@ -278,8 +264,8 @@ def test_accessor_with_edge_type():
 
 
 def test_accessor_with_crs():
-    ga_series = pd.Series(["POINT (0 1)"]).geoarrow.with_crs("EPSG:1234")
-    assert ga_series.dtype.pyarrow_dtype.crs == "EPSG:1234"
+    ga_series = pd.Series(["POINT (0 1)"]).geoarrow.with_crs(ga.OGC_CRS84)
+    assert ga_series.dtype.pyarrow_dtype.crs == ga.OGC_CRS84
 
 
 def test_accessor_with_dimensions():

geoarrow-pyarrow/src/geoarrow/pyarrow/__init__.py

@@ -9,7 +9,14 @@
 
 from geoarrow.types._version import __version__, __version_tuple__  # NOQA: F401
 
-from geoarrow.c.lib import GeometryType, Dimensions, CoordType, EdgeType, CrsType
+from geoarrow.types import (
+    GeometryType,
+    Dimensions,
+    CoordType,
+    EdgeType,
+    Encoding,
+    OGC_CRS84,
+)
 
 from geoarrow.pyarrow._type import (
     GeometryExtensionType,
@@ -33,6 +40,9 @@
     multipolygon,
     extension_type,
     geometry_type_common,
+)
+
+from geoarrow.types.type_pyarrow import (
     register_extension_types,
     unregister_extension_types,
 )
@@ -69,10 +79,11 @@
     "Dimensions",
     "CoordType",
     "EdgeType",
-    "CrsType",
+    "Encoding",
     "GeometryExtensionType",
     "WktType",
     "WkbType",
+    "OGC_CRS84",
     "PointType",
     "LinestringType",
     "PolygonType",

geoarrow-pyarrow/src/geoarrow/pyarrow/_array.py

@@ -55,10 +55,10 @@ def __repr__(self):
         tail_str = [f"<{item.as_py()}>" for item in tail]
         for i in range(len(head)):
             if len(head_str[i]) > max_width:
-                head_str[i] = f"{head_str[i][:(max_width - 4)]}...>"
+                head_str[i] = f"{head_str[i][: (max_width - 4)]}...>"
         for i in range(len(tail)):
             if len(tail_str[i]) > max_width:
-                tail_str[i] = f"{tail_str[i][:(max_width - 4)]}...>"
+                tail_str[i] = f"{tail_str[i][: (max_width - 4)]}...>"
 
         type_name = type(self).__name__
         head_str = "\n".join(head_str)
@@ -138,7 +138,7 @@ def array(obj, type_=None, *args, **kwargs) -> GeometryExtensionArray:
     # Convert GeoPandas to WKB
     if type(obj).__name__ == "GeoSeries":
         if obj.crs:
-            type_ = wkb().with_crs(obj.crs.to_json(), lib.CrsType.PROJJSON)
+            type_ = wkb().with_crs(obj.crs)
         else:
             type_ = wkb()