first pass on view types

Author: paleolimbot

geoarrow-types/src/geoarrow/types/__init__.py

@@ -17,6 +17,8 @@
     large_wkb,
     wkt,
     large_wkt,
+    wkb_view,
+    wkt_view,
     box,
     point,
     linestring,
@@ -42,6 +44,8 @@
     "large_wkb",
     "wkt",
     "large_wkt",
+    "wkb_view",
+    "wkt_view",
     "geoarrow",
     "box",
     "point",

geoarrow-types/src/geoarrow/types/constants.py

@@ -91,7 +91,13 @@ class Encoding(TypeSpecEnum):
     LARGE_WKT = 4
     """Well-known text encoding with 64-bit offsets"""
 
-    GEOARROW = 5
+    WKB_VIEW = 5
+    """Well-known binary encoding using binary views as a storage type"""
+
+    WKT_VIEW = 6
+    """Well-known binary encoding using string views as a storage type"""
+
+    GEOARROW = 7
     """GeoArrow native nested list encoding"""
 
     def is_serialized(self):
@@ -100,6 +106,8 @@ def is_serialized(self):
             Encoding.LARGE_WKB,
             Encoding.WKT,
             Encoding.LARGE_WKT,
+            Encoding.WKB_VIEW,
+            Encoding.WKT_VIEW
         )
 
 

geoarrow-types/src/geoarrow/types/type_pyarrow.py

@@ -558,6 +558,10 @@ def _parse_storage(storage_type):
         return [("string", ())]
     elif pa_types.is_large_string(storage_type):
         return [("large_string", ())]
+    elif hasattr(pa_types, "is_binary_view") and pa_types.is_binary_view(storage_type):
+        return [("binary_view", ())]
+    elif hasattr(pa_types, "is_string_view") and pa_types.is_string_view(storage_type):
+        return [("string_view", ())]
     elif pa_types.is_float64(storage_type):
         return [("double", ())]
     elif isinstance(storage_type, pa.ListType):
@@ -925,9 +929,9 @@ def _add_union_types_to_native_storage_types():
 
     for coord_type in ALL_COORD_TYPES:
         for dimension in ALL_DIMENSIONS:
-            _NATIVE_STORAGE_TYPES[
-                (GeometryType.GEOMETRY, coord_type, dimension)
-            ] = _generate_union_storage(coord_type=coord_type, dimensions=[dimension])
+            _NATIVE_STORAGE_TYPES[(GeometryType.GEOMETRY, coord_type, dimension)] = (
+                _generate_union_storage(coord_type=coord_type, dimensions=[dimension])
+            )
 
         # With unknown dimensions, we reigster the massive catch-all union
         _NATIVE_STORAGE_TYPES[
@@ -1014,6 +1018,10 @@ def _spec_short_repr(spec, ext_name):
     Encoding.LARGE_WKB: pa.large_binary(),
 }
 
+if hasattr(pa, "binary_view"):
+    _SERIALIZED_STORAGE_TYPES[Encoding.WKT_VIEW] = pa.string_view()
+    _SERIALIZED_STORAGE_TYPES[Encoding.WKB_VIEW] = pa.binary_view()
+
 _NATIVE_STORAGE_TYPES = _generate_storage_types()
 _add_union_types_to_native_storage_types()
 
@@ -1022,6 +1030,8 @@ def _spec_short_repr(spec, ext_name):
     ("large_binary",): Encoding.LARGE_WKB,
     ("string",): Encoding.WKT,
     ("large_string",): Encoding.LARGE_WKT,
+    ("binary_view",): Encoding.WKB_VIEW,
+    ("string_view",): Encoding.WKT_VIEW,
     ("struct",): TypeSpec(
         encoding=Encoding.GEOARROW,
         geometry_type=GeometryType.POINT,

geoarrow-types/src/geoarrow/types/type_spec.py

@@ -357,12 +357,22 @@ def wkb(*, edge_type=None, crs=crs.UNSPECIFIED) -> TypeSpec:
 def large_wkb(*, edge_type=None, crs=crs.UNSPECIFIED) -> TypeSpec:
     """Large well-known binary encoding
 
-    Create a :class:`TypeSpec` denoting a well-known binary type  with
+    Create a :class:`TypeSpec` denoting a well-known binary type with
     64-bit data offsets. See :func:`type_spec` for parameter definitions.
     """
     return type_spec(encoding=Encoding.LARGE_WKB, edge_type=edge_type, crs=crs)
 
 
+def wkb_view(*, edge_type=None, crs=crs.UNSPECIFIED) -> TypeSpec:
+    """Well-known binary view encoding
+
+    Create a :class:`TypeSpec` denoting a well-known binary type using
+    binary views as the underlying storage type. See :func:`type_spec`
+    for parameter definitions.
+    """
+    return type_spec(encoding=Encoding.WKB_VIEW, edge_type=edge_type, crs=crs)
+
+
 def wkt(*, edge_type=None, crs=crs.UNSPECIFIED) -> TypeSpec:
     """Well-known text encoding
 
@@ -381,6 +391,16 @@ def large_wkt(*, edge_type=None, crs=crs.UNSPECIFIED) -> TypeSpec:
     return type_spec(encoding=Encoding.LARGE_WKT, edge_type=edge_type, crs=crs)
 
 
+def wkt_view(*, edge_type=None, crs=crs.UNSPECIFIED) -> TypeSpec:
+    """Well-known text encoding
+
+    Create a :class:`TypeSpec` denoting a well-known text type using
+    string views as the underlying storage type. See :func:`type_spec`
+    for parameter definitions.
+    """
+    return type_spec(encoding=Encoding.WKT_VIEW, edge_type=edge_type, crs=crs)
+
+
 def geoarrow(
     *,
     geometry_type=None,
@@ -619,6 +639,8 @@ def type_spec(
     Encoding.LARGE_WKB: "geoarrow.wkb",
     Encoding.WKT: "geoarrow.wkt",
     Encoding.LARGE_WKT: "geoarrow.wkt",
+    Encoding.WKB_VIEW: "geoarrow.wkb",
+    Encoding.WKT_VIEW: "geoarrow.wkt",
 }
 
 _GEOARROW_EXT_NAMES = {

geoarrow-types/tests/test_type_pyarrow.py

@@ -441,6 +441,8 @@ def test_multipolygon_array_from_geobuffers():
         gt.large_wkt(),
         gt.wkb(),
         gt.large_wkb(),
+        gt.wkt_view(),
+        gt.wkb_view(),
         # Geometry types
         gt.box(),
         gt.point(),