Don't create a GeoParquet metadata string if the SFT has no geometries

locationtech · Mar 28, 2024 · 09588e8 · 09588e8
1 parent 137dcb5
commit 09588e8
Show file tree

Hide file tree

Showing 2 changed files with 37 additions and 1 deletion.
diff --git a/...main/scala/org/locationtech/geomesa/fs/storage/parquet/io/SimpleFeatureWriteSupport.scala b/...main/scala/org/locationtech/geomesa/fs/storage/parquet/io/SimpleFeatureWriteSupport.scala
@@ -20,6 +20,7 @@ import org.locationtech.geomesa.fs.storage.common.observer.BoundsObserver
 import org.locationtech.geomesa.fs.storage.parquet.io.SimpleFeatureParquetSchema.{GeoParquetSchemaKey, SchemaVersionKey, geoParquetMetadata}
 import org.locationtech.geomesa.utils.geotools.ObjectType
 import org.locationtech.geomesa.utils.geotools.ObjectType.ObjectType
+import org.locationtech.geomesa.utils.geotools.RichSimpleFeatureType.RichSimpleFeatureType
 import org.locationtech.geomesa.utils.text.WKBUtils
 import org.locationtech.jts.geom._
 
@@ -81,9 +82,14 @@ class SimpleFeatureWriteSupport extends WriteSupport[SimpleFeature] {
 
   // called once at the end after all SimpleFeatures are written
   override def finalizeWrite(): FinalizedWriteContext = {
-    // Get the bounding box that spans all the SimpleFeatures' geometries
+    // Get the bounding boxes that span each geometry type
     val bboxes = observer.getBoundingBoxes
 
+    // If the SFT has no geometries, then there's no need to create GeoParquet metadata
+    if (bboxes.isEmpty) {
+      return new FinalizedWriteContext(schema.metadata)
+    }
+
     // TODO: not an elegant way to do it
     //  somehow trying to mutate the map, e.g. by calling metadata.put(GeoParquetSchemaKey, result), causes empty parquet files to be written
     val newMetadata: java.util.Map[String, String] = Map(

diff --git a/...torage-parquet/src/test/scala/org/locationtech/geomesa/parquet/ParquetReadWriteTest.scala b/...torage-parquet/src/test/scala/org/locationtech/geomesa/parquet/ParquetReadWriteTest.scala
@@ -221,6 +221,36 @@ class ParquetReadWriteTest extends Specification with AllExpectations with LazyL
       metadata.contains(polygonsBboxString) must beTrue
     }
 
+    "write parquet files with no geometries" >> {
+      val f = Files.createTempFile("geomesa", ".parquet")
+      val sft = SimpleFeatureTypes.createType("test", "name:String,age:Int,dtg:Date")
+      val sftConf = {
+        val c = new Configuration()
+        StorageConfiguration.setSft(c, sft)
+        // Use GZIP in tests but snappy in prod due to license issues
+        c.set(ParquetCompressionOpt, CompressionCodecName.GZIP.toString)
+        c
+      }
+
+      val features = {
+        Seq(
+          ScalaSimpleFeature.create(sft, "1", "first", 100, "2017-01-01T00:00:00Z"),
+          ScalaSimpleFeature.create(sft, "2", null, 200, "2017-01-02T00:00:00Z"),
+          ScalaSimpleFeature.create(sft, "3", "third", 300, "2017-01-03T00:00:00Z")
+        )
+      }
+
+      val writer = SimpleFeatureParquetWriter.builder(new Path(f.toUri), sftConf).build()
+      WithClose(writer) { writer =>
+        features.foreach(writer.write)
+      }
+
+      Files.size(f) must beGreaterThan(0L)
+
+      val metadata = writer.getFooter.getFileMetaData.getKeyValueMetaData.get(GeoParquetSchemaKey)
+      metadata must beNull
+    }
+
     "read geoparquet files" >> {
       val result = readFile(FilterCompat.NOOP, sftConf)
       result mustEqual features