Merge branch 'master' of https://github.com/aloneguid/parquet-dotnet

aloneguid · Oct 11, 2022 · 14b15e5 · 14b15e5
2 parents c1db3be + 8795b4b
commit 14b15e5
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 1 deletion.
diff --git a/src/Parquet.Test/ParquetReaderTest.cs b/src/Parquet.Test/ParquetReaderTest.cs
@@ -82,6 +82,17 @@ public async Task Reads_multi_page_file() {
             }
         }
 
+        [Fact]
+        public async Task Reads_rle_dictionary_encoded_columns() {
+            using(ParquetReader reader = await ParquetReader.CreateAsync(OpenTestFile("rle_dictionary_encoded_columns.parquet"), leaveStreamOpen: false)) {
+                DataColumn[] data = await reader.ReadEntireRowGroupAsync();
+
+                //If we made it this far we were able to read all the columns
+                Assert.Single(data[0].Data);
+                Assert.Equal(40539, ((float?[])data[0].Data)[0]);
+            }
+        }
+
         [Fact]
         public async Task Reads_byte_arrays() {
             byte[] nameValue;

diff --git a/src/Parquet.Test/data/rle_dictionary_encoded_columns.parquet b/src/Parquet.Test/data/rle_dictionary_encoded_columns.parquet
diff --git a/src/Parquet/File/DataColumnReader.cs b/src/Parquet/File/DataColumnReader.cs
@@ -240,6 +240,7 @@ private void ReadColumn(BinaryReader reader, Thrift.Encoding encoding, long tota
                     break;
 
                 case Thrift.Encoding.PLAIN_DICTIONARY:
+                case Thrift.Encoding.RLE_DICTIONARY:
                     if(cd.indexes == null)
                         cd.indexes = new int[(int)totalValues];
                     indexCount = ReadPlainDictionary(reader, maxReadCount, cd.indexes, 0);
@@ -277,4 +278,4 @@ private static int GetRemainingLength(BinaryReader reader) {
             return (int)(reader.BaseStream.Length - reader.BaseStream.Position);
         }
     }
-}
+}