Add ST_SetCrs (#37)

* Split into ST_SetSRID and ST_SetCRS

* Fix doc, update integration tests and allow null args

* updates parquet integration tests to use st_setcrs

Author: jesspav

python/sedonadb/tests/functions/test_transforms.py

@@ -50,19 +50,17 @@ def test_st_setsrid(eng, geom, srid, expected_srid):
         assert df.crs == pyproj.CRS(expected_srid)
 
 
-# PostGIS does not handle String CRS input to ST_SetSrid
+# PostGIS does not have an API ST_SetCrs
 @pytest.mark.parametrize("eng", [SedonaDB])
 @pytest.mark.parametrize(
-    ("geom", "srid", "expected_srid"),
+    ("geom", "crs", "expected_srid"),
     [
         ("POINT (1 1)", "EPSG:26920", 26920),
         ("POINT (1 1)", pyproj.CRS("EPSG:26920").to_json(), 26920),
     ],
 )
-def test_st_setsrid_sedonadb(eng, geom, srid, expected_srid):
+def test_st_setcrs_sedonadb(eng, geom, crs, expected_srid):
     eng = eng.create_or_skip()
-    result = eng.execute_and_collect(
-        f"SELECT ST_SetSrid({geom_or_null(geom)}, '{srid}')"
-    )
+    result = eng.execute_and_collect(f"SELECT ST_SetCrs({geom_or_null(geom)}, '{crs}')")
     df = eng.result_to_pandas(result)
     assert df.crs.to_epsg() == expected_srid

python/sedonadb/tests/io/test_parquet.py

@@ -102,7 +102,7 @@ def test_read_geoparquet_pruned(geoarrow_data, name):
         result = eng.execute_and_collect(
             f"""
             SELECT "OBJECTID", geometry FROM tab
-            WHERE ST_Intersects(geometry, ST_SetSRID({geom_or_null(wkt_filter)}, '{gdf.crs.to_json()}'))
+            WHERE ST_Intersects(geometry, ST_SetCRS({geom_or_null(wkt_filter)}, '{gdf.crs.to_json()}'))
             ORDER BY "OBJECTID";
         """
         )
@@ -127,7 +127,7 @@ def test_read_geoparquet_pruned(geoarrow_data, name):
         result = eng.execute_and_collect(
             f"""
             SELECT * FROM tab_dataset
-            WHERE ST_Intersects(geometry, ST_SetSRID({geom_or_null(wkt_filter)}, '{gdf.crs.to_json()}'))
+            WHERE ST_Intersects(geometry, ST_SetCRS({geom_or_null(wkt_filter)}, '{gdf.crs.to_json()}'))
             ORDER BY "OBJECTID";
         """
         )

rust/sedona-expr/src/scalar_udf.rs

@@ -207,6 +207,11 @@ impl ArgMatcher {
         Arc::new(IsGeography {})
     }
 
+    /// Matches a null argument
+    pub fn is_null() -> Arc<dyn TypeMatcher + Send + Sync> {
+        Arc::new(IsNull {})
+    }
+
     /// Matches any numeric argument
     pub fn is_numeric() -> Arc<dyn TypeMatcher + Send + Sync> {
         Arc::new(IsNumeric {})
@@ -371,6 +376,14 @@ impl TypeMatcher for IsBoolean {
     }
 }
 
+#[derive(Debug)]
+struct IsNull {}
+impl TypeMatcher for IsNull {
+    fn match_type(&self, arg: &SedonaType) -> bool {
+        matches!(arg, SedonaType::Arrow(DataType::Null))
+    }
+}
+
 /// Type definition for a Scalar kernel implementation function
 pub type SedonaScalarKernelImpl =
     Arc<dyn Fn(&[SedonaType], &[ColumnarValue]) -> Result<ColumnarValue> + Send + Sync>;
@@ -602,6 +615,9 @@ mod tests {
 
         assert!(ArgMatcher::is_boolean().match_type(&SedonaType::Arrow(DataType::Boolean)));
         assert!(!ArgMatcher::is_boolean().match_type(&SedonaType::Arrow(DataType::Int32)));
+
+        assert!(ArgMatcher::is_null().match_type(&SedonaType::Arrow(DataType::Null)));
+        assert!(!ArgMatcher::is_null().match_type(&SedonaType::Arrow(DataType::Int32)));
     }
 
     #[test]

rust/sedona-functions/src/register.rs

@@ -85,6 +85,7 @@ pub fn default_function_set() -> FunctionSet {
         crate::st_pointzm::st_pointm_udf,
         crate::st_pointzm::st_pointzm_udf,
         crate::st_transform::st_transform_udf,
+        crate::st_setsrid::st_set_crs_udf,
         crate::st_setsrid::st_set_srid_udf,
         crate::st_srid::st_srid_udf,
         crate::st_xyzm::st_m_udf,
@@ -124,6 +125,7 @@ pub mod stubs {
     pub use crate::st_area::st_area_udf;
     pub use crate::st_length::st_length_udf;
     pub use crate::st_perimeter::st_perimeter_udf;
+    pub use crate::st_setsrid::st_set_crs_with_engine_udf;
     pub use crate::st_setsrid::st_set_srid_with_engine_udf;
     pub use crate::st_transform::st_transform_udf;
 }