Merge pull request #84 from G-Research/RowCompressFix

Fixed compression argument being ignored by ParquetRowWriter.
G-Research · Nov 28, 2019 · 958bd3a · 958bd3a
2 parents 424eec4 + aaf3066
commit 958bd3a
Show file tree

Hide file tree

Showing 4 changed files with 39 additions and 7 deletions.
diff --git a/csharp.test/TestRowOrientedParquetFile.cs b/csharp.test/TestRowOrientedParquetFile.cs
@@ -80,6 +80,28 @@ public static void TestWriterDoubleDispose()
             }
         }
 
+        [Test]
+        public static void TestCompressionArgument([Values(Compression.Uncompressed, Compression.Brotli)] Compression compression)
+        {
+            using (var buffer = new ResizableBuffer())
+            {
+                using (var outputStream = new BufferOutputStream(buffer))
+                using (var writer = ParquetFile.CreateRowWriter<(int, float)>(outputStream, compression: compression))
+                {
+                    writer.WriteRows(new[] {(42, 3.14f)});
+                }
+
+                using (var inputStream = new BufferReader(buffer))
+                using (var reader = new ParquetFileReader(inputStream))
+                using (var group = reader.RowGroup(0))
+                {
+                    Assert.AreEqual(2, group.MetaData.NumColumns);
+                    Assert.AreEqual(compression, group.MetaData.GetColumnChunkMetaData(0).Compression);
+                    Assert.AreEqual(compression, group.MetaData.GetColumnChunkMetaData(1).Compression);
+                }
+            }
+        }
+
         private static void TestRoundtrip<TTuple>(TTuple[] rows)
         {
             using (var buffer = new ResizableBuffer())

diff --git a/csharp/ParquetSharp.csproj b/csharp/ParquetSharp.csproj
@@ -12,7 +12,7 @@
     <GenerateDocumentationFile>true</GenerateDocumentationFile>
     <GeneratePackageOnBuild>true</GeneratePackageOnBuild>
     <NoWarn>1591;</NoWarn>
-    <Version>2.0.2-beta3</Version>
+    <Version>2.0.2-beta4</Version>
     <Company>G-Research</Company>
     <Authors>G-Research</Authors>
     <Product>ParquetSharp</Product>

diff --git a/csharp/RowOriented/ParquetFile.cs b/csharp/RowOriented/ParquetFile.cs
@@ -45,7 +45,7 @@ public static ParquetRowWriter<TTuple> CreateRowWriter<TTuple>(
             IReadOnlyDictionary<string, string> keyValueMetadata = null)
         {
             var (columns, writeDelegate) = GetOrCreateWriteDelegate<TTuple>(columnNames);
-            return new ParquetRowWriter<TTuple>(path, columns, writeDelegate);
+            return new ParquetRowWriter<TTuple>(path, columns, compression, keyValueMetadata, writeDelegate);
         }
 
         /// <summary>
@@ -58,7 +58,7 @@ public static ParquetRowWriter<TTuple> CreateRowWriter<TTuple>(
             IReadOnlyDictionary<string, string> keyValueMetadata = null)
         {
             var (columns, writeDelegate) = GetOrCreateWriteDelegate<TTuple>(columnNames);
-            return new ParquetRowWriter<TTuple>(outputStream, columns, writeDelegate);
+            return new ParquetRowWriter<TTuple>(outputStream, columns, compression, keyValueMetadata, writeDelegate);
         }
 
         private static ParquetRowReader<TTuple>.ReadAction GetOrCreateReadDelegate<TTuple>()

diff --git a/csharp/RowOriented/ParquetRowWriter.cs b/csharp/RowOriented/ParquetRowWriter.cs
@@ -13,13 +13,23 @@ public sealed class ParquetRowWriter<TTuple> : IDisposable
     {
         internal delegate void WriteAction(ParquetRowWriter<TTuple> parquetRowWriter, TTuple[] rows, int length);
 
-        internal ParquetRowWriter(string path, Column[] columns, WriteAction writeAction)
-            : this(new ParquetFileWriter(path, columns), writeAction)
+        internal ParquetRowWriter(
+            string path, 
+            Column[] columns, 
+            Compression compression,
+            IReadOnlyDictionary<string, string> keyValueMetadata, 
+            WriteAction writeAction)
+            : this(new ParquetFileWriter(path, columns, compression, keyValueMetadata), writeAction)
         {
         }
 
-        internal ParquetRowWriter(OutputStream outputStream, Column[] columns, WriteAction writeAction)
-            : this(new ParquetFileWriter(outputStream, columns), writeAction)
+        internal ParquetRowWriter(
+            OutputStream outputStream, 
+            Column[] columns,
+            Compression compression,
+            IReadOnlyDictionary<string, string> keyValueMetadata, 
+            WriteAction writeAction)
+            : this(new ParquetFileWriter(outputStream, columns, compression, keyValueMetadata), writeAction)
         {
         }