Generate Largest-Possible Tiles

locationtech · Oct 17, 2017 · 81052f5 · 81052f5
1 parent c67c787
commit 81052f5
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 7 deletions.
diff --git a/s3/src/test/scala/geotrellis/spark/io/s3/S3GeoTiffRDDSpec.scala b/s3/src/test/scala/geotrellis/spark/io/s3/S3GeoTiffRDDSpec.scala
@@ -67,7 +67,7 @@ class S3GeoTiffRDDSpec
       val geoTiffBytes = Files.readAllBytes(Paths.get(testGeoTiffPath))
       mockClient.putObject(bucket, key, geoTiffBytes)
 
-      val options = S3GeoTiffRDD.Options(getS3Client = () => new MockS3Client, partitionBytes=1<<20)
+      val options = S3GeoTiffRDD.Options(getS3Client = () => new MockS3Client, partitionBytes=1<<20, maxTileSize = Some(64))
       val geometry = Line(Point(141.7066667, -17.5200000), Point(142.1333333, -17.7))
       val fn = {( _: Any, key: ProjectedExtent) => key }
       val source1 =

diff --git a/spark/src/main/scala/geotrellis/spark/io/RasterReader.scala b/spark/src/main/scala/geotrellis/spark/io/RasterReader.scala
@@ -77,8 +77,20 @@ object RasterReader {
     cols: Int, rows: Int, maxSize: Int,
     segCols: Int, segRows: Int
   ): Array[GridBounds] = {
-    val colSize: Int = if (maxSize >= segCols) segCols; else best(maxSize, segCols)
-    val rowSize: Int = if (maxSize >= segRows) segRows; else best(maxSize, segRows)
+    val colSize: Int =
+      if (maxSize >= segCols * 2) {
+        math.floor(maxSize.toDouble / segCols).toInt * segCols
+      } else if (maxSize >= segCols) {
+        segCols
+      } else best(maxSize, segCols)
+
+    val rowSize: Int =
+      if (maxSize >= segRows * 2) {
+        math.floor(maxSize.toDouble / segRows).toInt * segRows
+      } else if (maxSize >= segRows) {
+        segRows
+      } else best(maxSize, segRows)
+
     val windows = listWindows(cols, rows, colSize, rowSize)
 
     windows
@@ -90,10 +102,22 @@ object RasterReader {
     extent: Extent, segCols: Int, segRows: Int, geometry: Geometry,
     options: Rasterizer.Options = Rasterizer.Options.DEFAULT
   ): Array[GridBounds] = {
+    val maxColSize: Int =
+      if (maxSize >= segCols * 2) {
+        math.floor(maxSize.toDouble / segCols).toInt * segCols
+      } else if (maxSize >= segCols) {
+        segCols
+      } else best(maxSize, segCols)
+
+    val maxRowSize: Int =
+      if (maxSize >= segRows) {
+        math.floor(maxSize.toDouble / segRows).toInt * segRows
+      } else if (maxSize >= segRows) {
+        segRows
+      } else best(maxSize, segRows)
+
     val result = scala.collection.mutable.ArrayBuffer[GridBounds]()
-    val maxColSize: Int = if (maxSize >= segCols) segCols; else best(maxSize, segCols)
-    val maxRowSize: Int = if (maxSize >= segRows) segRows; else best(maxSize, segRows)
-    val re = RasterExtent(extent, cols/maxColSize, rows/maxRowSize)
+    val re = RasterExtent(extent, math.max(cols/maxColSize,1), math.max(rows/maxRowSize,1))
 
     Rasterizer.foreachCellByGeometry(geometry, re, options)({ (col: Int, row: Int) =>
       result +=

diff --git a/spark/src/test/scala/geotrellis/spark/io/hadoop/HadoopGeoTiffRDDSpec.scala b/spark/src/test/scala/geotrellis/spark/io/hadoop/HadoopGeoTiffRDDSpec.scala
@@ -44,7 +44,7 @@ class HadoopGeoTiffRDDSpec
 
     it("should filter by geometry") {
       val testGeoTiffPath = new Path(localFS.getWorkingDirectory, "spark/src/test/resources/all-ones.tif")
-      val options = HadoopGeoTiffRDD.Options(partitionBytes=Some(1<<20))
+      val options = HadoopGeoTiffRDD.Options(partitionBytes=Some(1<<20), maxTileSize = Some(64))
       val geometry = Line(Point(141.7066667, -17.5200000), Point(142.1333333, -17.7))
       val fn = {( _: URI, key: ProjectedExtent) => key }
       val source1 =