apache · jiayuasu · Aug 2, 2023 · Aug 1, 2023 · Aug 1, 2023 · Aug 1, 2023
@@ -43,6 +43,28 @@ public static Geometry geomFromWKT(String wkt, int srid) throws ParseException {
         return new WKTReader(geometryFactory).read(wkt);
     }
 
+    public static Geometry geomFromEWKT(String ewkt) throws ParseException {
+        if (ewkt == null) {
+            return null;
+        }
+        int SRID = 0;
+        String wkt = ewkt;
+
+        int index = ewkt.indexOf("SRID=");
+        if (index != -1) {
+            int semicolonIndex = ewkt.indexOf(';', index);
+            if (semicolonIndex != -1) {
+                SRID = Integer.parseInt(ewkt.substring(index + 5, semicolonIndex));
+                wkt = ewkt.substring(semicolonIndex + 1);
+            }
+            else {
+                throw new ParseException("Invalid EWKT string");
+            }
+        }
+        GeometryFactory geometryFactory = new GeometryFactory(new PrecisionModel(), SRID);
+        return new WKTReader(geometryFactory).read(wkt);
+    }
+
     public static Geometry geomFromWKB(byte[] wkb) throws ParseException {
         return new WKBReader().read(wkb);
     }

@@ -37,6 +37,23 @@ public void geomFromWKT() throws ParseException {
         ParseException invalid = assertThrows(ParseException.class, () -> Constructors.geomFromWKT("not valid", 0));
         assertEquals("Unknown geometry type: NOT (line 1)", invalid.getMessage());
     }
+
+    @Test
+    public void geomFromEWKT() throws ParseException {
+        assertNull(Constructors.geomFromEWKT(null));
+
+        Geometry geom = Constructors.geomFromEWKT("POINT (1 1)");
+        assertEquals(0, geom.getSRID());
+        assertEquals("POINT (1 1)", geom.toText());
+
+        geom = Constructors.geomFromEWKT("SRID=4269; POINT (1 1)");
+        assertEquals(4269, geom.getSRID());
+        assertEquals("POINT (1 1)", geom.toText());
+
+        ParseException invalid = assertThrows(ParseException.class, () -> Constructors.geomFromEWKT("not valid"));
+        assertEquals("Unknown geometry type: NOT (line 1)", invalid.getMessage());
+    }
+
     @Test
     public void mLineFromWKT() throws ParseException {
         assertNull(Constructors.mLineFromText(null, 0));

@@ -230,6 +230,26 @@ Output:
 POINT(40.7128 -74.006)
 ```
 
+## ST_GeomFromEWKT
+
+Introduction: Construct a Geometry from OGC Extended WKT
+
+Format:
+`ST_GeomFromEWKT (EWkt:string)`
+
+Since: `v1.5.0`
+
+SQL example:
+```sql
+SELECT ST_GeomFromEWKT('SRID=4269;POINT(40.7128 -74.0060)')
+```
+
+Output:
+
+```
+POINT(40.7128 -74.006)
+```
+
 ## ST_LineFromText
 
 Introduction: Construct a LineString from Text

@@ -274,6 +274,26 @@ Output:
 POINT(40.7128 -74.006)
 ```
 
+## ST_GeomFromEWKT
+
+Introduction: Construct a Geometry from OGC Extended WKT
+
+Format:
+`ST_GeomFromEWKT (EWkt:string)`
+
+Since: `v1.5.0`
+
+SQL example:
+```sql
+SELECT ST_GeomFromEWKT('SRID=4269;POINT(40.7128 -74.0060)')
+```
+
+Output:
+
+```
+POINT(40.7128 -74.006)
+```
+
 ## ST_LineFromText
 
 Introduction: Construct a Line from Wkt text

@@ -32,6 +32,7 @@ public static UserDefinedFunction[] getFuncs() {
                 new Constructors.ST_PolygonFromText(),
                 new Constructors.ST_PolygonFromEnvelope(),
                 new Constructors.ST_GeomFromWKT(),
+                new Constructors.ST_GeomFromEWKT(),
                 new Constructors.ST_GeomFromText(),
                 new Constructors.ST_GeomFromWKB(),
                 new Constructors.ST_GeomFromGeoJSON(),

@@ -134,6 +134,13 @@ public Geometry eval(@DataTypeHint("String") String wktString) throws ParseExcep
         }
     }
 
+    public static class ST_GeomFromEWKT extends ScalarFunction {
+        @DataTypeHint(value = "RAW", bridgedTo = org.locationtech.jts.geom.Geometry.class)
+        public Geometry eval(@DataTypeHint("String") String wktString) throws ParseException {
+            return org.apache.sedona.common.Constructors.geomFromEWKT(wktString);
+        }
+    }
+
     public static class ST_GeomFromText extends ScalarFunction {
         @DataTypeHint(value = "RAW", bridgedTo = org.locationtech.jts.geom.Geometry.class)
         public Geometry eval(@DataTypeHint("String") String wktString) throws ParseException {

@@ -148,6 +148,23 @@ public void testGeomFromWKT() {
         assertEquals(data.get(data.size() - 1).getField(0).toString(), result.getField(0).toString());
     }
 
+    @Test
+    public void testGeomFromEWKT() {
+        List<Row> data = new ArrayList<>();
+        data.add(Row.of("SRID=123;MULTILINESTRING((1 2, 3 4), (4 5, 6 7))", "multiline", 0L));
+
+        Table geomTable = createTextTable(data, multilinestringColNames);
+        geomTable = geomTable
+                .select(call(Constructors.ST_GeomFromEWKT.class.getSimpleName(),
+                        $(multilinestringColNames[0]))
+                        .as(multilinestringColNames[0]), $(multilinestringColNames[1]));
+        String result = first(geomTable)
+                .getFieldAs(0)
+                .toString();
+        String expectedGeom = "MULTILINESTRING ((1 2, 3 4), (4 5, 6 7))";
+        assertEquals(expectedGeom, result);
+    }
+
     @Test
     public void testGeomFromText() {
         List<Row> data = createPolygonWKT(testDataSize);

@@ -31,6 +31,7 @@
     "ST_GeomFromText",
     "ST_GeomFromWKB",
     "ST_GeomFromWKT",
+    "ST_GeomFromEWKT",
     "ST_LineFromText",
     "ST_LineStringFromText",
     "ST_Point",
@@ -132,6 +133,17 @@ def ST_GeomFromWKT(wkt: ColumnOrName) -> Column:
     """
     return _call_constructor_function("ST_GeomFromWKT", wkt)
 
+@validate_argument_types
+def ST_GeomFromEWKT(ewkt: ColumnOrName) -> Column:
+    """Generate a geometry column from a OGC Extended Well-Known Text (WKT) string column.
+
+    :param ewkt: OGC Extended WKT string column to generate from.
+    :type ewkt: ColumnOrName
+    :return: Geometry column representing the EWKT string.
+    :rtype: Column
+    """
+    return _call_constructor_function("ST_GeomFromEWKT", ewkt)
+
 
 @validate_argument_types
 def ST_LineFromText(wkt: ColumnOrName) -> Column:

@@ -60,6 +60,12 @@ def test_st_geom_from_wkt(self):
         polygon_df.show(10)
         assert polygon_df.count() == 100
 
+    def test_st_geom_from_ewkt(self):
+        input_df = self.spark.createDataFrame([("SRID=4269;LineString(1 2, 3 4)",)], ["ewkt"])
+        input_df.createOrReplaceTempView("input_ewkt")
+        line_df = self.spark.sql("select ST_LineFromText(ewkt) as geom from input_ewkt")
+        assert line_df.count() == 1
+
     def test_st_geom_from_wkt_3d(self):
         input_df = self.spark.createDataFrame([
             ("Point(21 52 87)",),

@@ -43,6 +43,7 @@ object Catalog {
     function[ST_GeomFromText](0),
     function[ST_LineFromText](),
     function[ST_GeomFromWKT](0),
+    function[ST_GeomFromEWKT](),
     function[ST_GeomFromWKB](),
     function[ST_GeomFromGeoJSON](),
     function[ST_GeomFromGML](),

@@ -91,6 +91,19 @@ case class ST_GeomFromWKT(inputExpressions: Seq[Expression])
   }
 }
 
+/**
+  * Return a Geometry from a OGC Extended WKT string
+  *
+  * @param inputExpressions This function takes a geometry string. The string format must be OGC Extended Well-Known text (EWKT) representation.
+  */
+case class ST_GeomFromEWKT(inputExpressions: Seq[Expression])
+  extends InferredExpression(Constructors.geomFromEWKT _) {
+
+  protected def withNewChildrenInternal(newChildren: IndexedSeq[Expression]) = {
+    copy(inputExpressions = newChildren)
+  }
+}
+
 
 /**
   * Return a Geometry from a WKT string

@@ -55,6 +55,9 @@ object st_constructors extends DataFrameAPI {
 
   def ST_GeomFromWKT(wkt: String, srid: Int): Column = wrapExpression[ST_GeomFromWKT](wkt, srid)
 
+  def ST_GeomFromEWKT(wkt: Column): Column = wrapExpression[ST_GeomFromEWKT](wkt, 0)
+  def ST_GeomFromEWKT(wkt: String): Column = wrapExpression[ST_GeomFromEWKT](wkt, 0)
+
   def ST_LineFromText(wkt: Column): Column = wrapExpression[ST_LineFromText](wkt)
   def ST_LineFromText(wkt: String): Column = wrapExpression[ST_LineFromText](wkt)
 

@@ -93,6 +93,29 @@ class constructorTestScala extends TestBaseScala {
       assert(thrown.getMessage == "Unknown geometry type: NOT (line 1)")
     }
 
+    it("Passed ST_GeomFromEWKT") {
+      var polygonWktDf = sparkSession.read.format("csv").option("delimiter", "\t").option("header", "false").load(mixedWktGeometryInputLocation)
+      polygonWktDf.createOrReplaceTempView("polygontable")
+      var polygonDf = sparkSession.sql("select ST_GeomFromEWKT(polygontable._c0) as countyshape from polygontable")
+      assert(polygonDf.count() == 100)
+      val nullGeom = sparkSession.sql("select ST_GeomFromEWKT(null)")
+      assert(nullGeom.first().isNullAt(0))
+      val pointDf = sparkSession.sql("select ST_GeomFromEWKT('SRID=4269;POINT(-71.064544 42.28787)')")
+      assert(pointDf.count() == 1)
+      // Fail on wrong input type
+      intercept[Exception] {
+        sparkSession.sql("SELECT ST_GeomFromEWKT(0)").collect()
+      }
+    }
+
+    it("Passed ST_GeomFromEWKT invalid input") {
+      // Fail on non wkt strings
+      val thrown = intercept[Exception] {
+        sparkSession.sql("SELECT ST_GeomFromEWKT('not wkt')").collect()
+      }
+      assert(thrown.getMessage == "Unknown geometry type: NOT (line 1)")
+    }
+
     it("Passed ST_LineFromText") {
       val geometryDf = Seq("Linestring(1 2, 3 4)").map(wkt => Tuple1(wkt)).toDF("geom")
       geometryDf.createOrReplaceTempView("linetable")

@@ -94,6 +94,13 @@ class dataFrameAPITestScala extends TestBaseScala {
       assert(actualResult.getSRID == 4326)
     }
 
+    it("passed st_geomfromewkt") {
+      val df = sparkSession.sql("SELECT ST_GeomFromEWKT('SRID=4269;POINT(0.0 1.0)')")
+      val actualResult = df.take(1)(0).get(0).asInstanceOf[Geometry]
+      assert(actualResult.toText == "POINT (0 1)")
+      assert(actualResult.getSRID == 4269)
+    }
+
     it("passed st_geomfromtext") {
       val df = sparkSession.sql("SELECT 'POINT(0.0 1.0)' AS wkt").select(ST_GeomFromText("wkt"))
       val actualResult = df.take(1)(0).get(0).asInstanceOf[Geometry].toText()