Added rf_agg_reprojected_extent.

Author: metasim

core/src/main/scala/org/locationtech/rasterframes/functions/AggregateFunctions.scala

@@ -20,7 +20,7 @@
  */
 
 package org.locationtech.rasterframes.functions
-import geotrellis.proj4.WebMercator
+import geotrellis.proj4.{CRS, WebMercator}
 import geotrellis.raster.resample.ResampleMethod
 import geotrellis.raster.{IntConstantNoDataCellType, Tile}
 import geotrellis.vector.Extent
@@ -29,6 +29,7 @@ import org.locationtech.rasterframes.expressions.accessors.{ExtractTile, GetCRS,
 import org.locationtech.rasterframes.expressions.aggregates.TileRasterizerAggregate.ProjectedRasterDefinition
 import org.locationtech.rasterframes.expressions.aggregates._
 import org.locationtech.rasterframes.stats._
+import org.locationtech.rasterframes._
 
 /** Functions associated with computing columnar aggregates over tile and geometry columns. */
 trait AggregateFunctions {
@@ -87,17 +88,23 @@ trait AggregateFunctions {
     TileRasterizerAggregate(params, tileCRS, tileExtent, tile)
   }
 
+  import org.apache.spark.sql.functions._
+  import org.locationtech.rasterframes.encoders.StandardEncoders.extentEncoder
+  import org.locationtech.rasterframes.util.NamedColumn
 
-  /** Compute the aggregate extent over a column. */
+  /** Compute the aggregate extent over a column. Assumes CRS homogeneity. */
   def rf_agg_extent(extent: Column): TypedColumn[Any, Extent] = {
-    import org.apache.spark.sql.functions._
-    import org.locationtech.rasterframes.encoders.StandardEncoders.extentEncoder
-    import org.locationtech.rasterframes.util.NamedColumn
     struct(
       min(extent.getField("xmin")) as "xmin",
       min(extent.getField("ymin")) as "ymin",
       max(extent.getField("xmax")) as "xmax",
       max(extent.getField("ymax")) as "ymax"
     ).as(s"rf_agg_extent(${extent.columnName})").as[Extent]
   }
+
+  /** Compute the aggregate extent over a column after reprojecting from the rows source CRS into the given destination CRS . */
+  def rf_agg_reprojected_extent(extent: Column, srcCRS: Column, destCRS: CRS): TypedColumn[Any, Extent] =
+    rf_agg_extent(st_extent(st_reproject(st_geometry(extent), srcCRS, destCRS)))
+    .as(s"rf_agg_reprojected_extent(${extent.columnName}, ${srcCRS.columnName}, $destCRS)")
+    .as[Extent]
 }

core/src/test/scala/org/locationtech/rasterframes/functions/AggregateFunctionsSpec.scala

@@ -202,12 +202,24 @@ class AggregateFunctionsSpec extends TestEnvironment with RasterMatchers {
   }
 
   describe("geometric aggregates") {
+    // SQL docs not available until we re-implement as an expression
+    ignore("should have docs") {
+      checkDocs("rf_agg_extent")
+      checkDocs("rf_agg_reprojected_extent")
+    }
+
     it("should compute an aggregate extent") {
       val src = TestData.l8Sample(1)
       val df = src.toDF(TileDimensions(10, 10))
-      df.show(false)
       val result = df.select(rf_agg_extent($"extent")).first()
       result should be(src.extent)
     }
+
+    it("should compute a reprojected aggregate extent") {
+      val src = TestData.l8Sample(1)
+      val df = src.toDF(TileDimensions(10, 10))
+      val result = df.select(rf_agg_reprojected_extent($"extent", $"crs", WebMercator)).first()
+      result should be(src.extent.reproject(src.crs, WebMercator))
+    }
   }
 }

docs/src/main/paradox/release-notes.md

@@ -9,7 +9,7 @@
 * Added `withSpatialIndex` to RasterSourceDataSource to pre-partition tiles based on tile extents mapped to a Z2 space-filling curve 
 * Add `rf_mask_by_bit`, `rf_mask_by_bits` and `rf_local_extract_bits` to deal with bit packed quality masks. Updated the masking documentation to demonstrate the use of these functions.
 * Added `toDF` extension method to `MultibandGeoTiff`
-* Added `rf_agg_extent` to compute the aggregate extent of a column
+* Added `rf_agg_extent` and `rf_agg_reprojected_extent` to compute the aggregate extent of a column
 * Added `rf_proj_raster` for constructing a `proj_raster` structure from individual CRS, Extent, and Tile columns.
 
 ### 0.8.4

project/RFDependenciesPlugin.scala

@@ -55,7 +55,7 @@ object RFDependenciesPlugin extends AutoPlugin {
       "locationtech-releases" at "https://repo.locationtech.org/content/groups/releases",
       "Azavea Public Builds" at "https://dl.bintray.com/azavea/geotrellis",
       "boundless-releases" at "https://repo.boundlessgeo.com/main/",
-      "Open Source Geospatial Foundation Repository" at "http://download.osgeo.org/webdav/geotools/"
+      "Open Source Geospatial Foundation Repository" at "https://download.osgeo.org/webdav/geotools/"
     ),
 
     // NB: Make sure to update the Spark version in pyrasterframes/python/setup.py

project/build.properties

@@ -1 +1 @@
-sbt.version=1.2.8
+sbt.version=1.3.4

project/plugins.sbt

@@ -1,6 +1,5 @@
 logLevel := sbt.Level.Error
 
-addSbtCoursier
 addSbtPlugin("com.eed3si9n" % "sbt-assembly" % "0.14.6")
 addSbtPlugin("com.eed3si9n" % "sbt-buildinfo" % "0.7.0")
 addSbtPlugin("de.heikoseeberger" % "sbt-header" % "3.0.2")

project/project/plugins.sbt

@@ -27,7 +27,7 @@
 from pyspark.sql.column import Column, _to_java_column
 from pyspark.sql.functions import lit
 from .rf_context import RFContext
-from .rf_types import CellType, Extent
+from .rf_types import CellType, Extent, CRS
 
 THIS_MODULE = 'pyrasterframes'
 
@@ -338,6 +338,11 @@ def rf_agg_extent(extent_col):
     return _apply_column_function('rf_agg_extent', extent_col)
 
 
+def rf_agg_reprojected_extent(extent_col, src_crs_col, dest_crs):
+    """Compute the aggregate extent over a column, first projecting from the row CRS to the destination CRS. """
+    return Column(RFContext.call('rf_agg_reprojected_extent', _to_java_column(extent_col), _to_java_column(src_crs_col),CRS(dest_crs).__jvm__))
+
+
 def rf_agg_overview_raster(tile_col: Column, cols: int, rows: int, aoi: Extent,
                            tile_extent_col: Column = None, tile_crs_col: Column = None):
     """Construct an overview raster of size `cols`x`rows` where data in `proj_raster` intersects the

pyrasterframes/src/main/python/pyrasterframes/rasterfunctions.py

@@ -220,6 +220,19 @@ def buffer(self, amount):
     def __str__(self):
         return self.__jvm__.toString()
 
+class CRS(object):
+    def __init__(self, proj4_str):
+        self.proj4_str = proj4_str
+
+    @cached_property
+    def __jvm__(self):
+        comp = RFContext.active().companion_of("org.locationtech.rasterframes.model.LazyCRS")
+        return comp.apply(self.proj4_str)
+
+    def __str__(self):
+        return self.proj4_str
+
+
 class CellType(object):
     def __init__(self, cell_type_name):
         self.cell_type_name = cell_type_name

pyrasterframes/src/main/python/pyrasterframes/rf_types.py

@@ -195,8 +195,14 @@ def test_z2_index(self):
 
     def test_agg_extent(self):
         r = self.df.select(rf_agg_extent(st_extent('poly_geom')).alias('agg_extent')).select('agg_extent.*').first()
-        self.assertDictEqual(r.asDict(),
-                             Row(xmin=-0.011268955205879273, ymin=-4.011268955205879, xmax=3.0112432169934484,
-                                 ymax=-0.9887567830065516).asDict()
-                             )
+        self.assertDictEqual(
+            r.asDict(),
+            Row(xmin=-0.011268955205879273, ymin=-4.011268955205879, xmax=3.0112432169934484, ymax=-0.9887567830065516).asDict()
+        )
 
+    def test_agg_reprojected_extent(self):
+        r = self.df.select(rf_agg_reprojected_extent(st_extent('poly_geom'), rf_mk_crs("EPSG:4326"), "EPSG:3857")).first()[0]
+        self.assertDictEqual(
+            r.asDict(),
+            Row(xmin=-1254.45435529069, ymin=-446897.63591665257, xmax=335210.0615704097, ymax=-110073.36515944061).asDict()
+        )