Add test for reading Parquet files using Apache Arrow.

ApplicationLibCode/UnitTests/CMakeLists.txt

@@ -102,6 +102,7 @@ set(SOURCE_UNITTEST_FILES
     ${CMAKE_CURRENT_LIST_DIR}/RifSummaryCalculationIO-Test.cpp
     ${CMAKE_CURRENT_LIST_DIR}/RimEmReader-Test.cpp
     ${CMAKE_CURRENT_LIST_DIR}/RifPolygonReader-Test.cpp
+    ${CMAKE_CURRENT_LIST_DIR}/RifParquetReader-Test.cpp
 )
 
 if(RESINSIGHT_ENABLE_GRPC)

ApplicationLibCode/UnitTests/RifParquetReader-Test.cpp

@@ -0,0 +1,52 @@
+#include "gtest/gtest.h"
+
+#include "RiaTestDataDirectory.h"
+
+#include <arrow/csv/api.h>
+#include <arrow/io/api.h>
+#include <arrow/scalar.h>
+#include <arrow/util/cancel.h>
+#include <parquet/arrow/reader.h>
+
+#include <QDir>
+#include <QString>
+
+TEST( RifParquetReaderTest, ReadValidFile )
+{
+    QDir baseFolder( TEST_DATA_DIR );
+
+    QString filename( "RifParquetReader/example.parquet" );
+    QString filePath = baseFolder.absoluteFilePath( filename );
+    EXPECT_TRUE( QFile::exists( filePath ) );
+
+    arrow::MemoryPool* pool = arrow::default_memory_pool();
+
+    auto openResult = arrow::io::ReadableFile::Open( filePath.toStdString().c_str() );
+    EXPECT_TRUE( openResult.ok() );
+
+    std::shared_ptr<arrow::io::RandomAccessFile> input = std::move( openResult ).ValueOrDie();
+
+    // Open Parquet file reader
+    std::unique_ptr<parquet::arrow::FileReader> arrow_reader;
+    EXPECT_TRUE( parquet::arrow::OpenFile( input, pool, &arrow_reader ).ok() );
+
+    // Read entire file as a single Arrow table
+    std::shared_ptr<arrow::Table> table;
+    EXPECT_TRUE( arrow_reader->ReadTable( &table ).ok() );
+
+    // Expect one column named "col1"
+    EXPECT_EQ( table->columns().size(), 1 );
+    std::vector<std::string> expectedNames = { "col1" };
+    EXPECT_EQ( table->ColumnNames(), expectedNames );
+
+    // Expected the column to contain 100 int64 [0, 99]
+    auto columnData   = table->column( 0 );
+    int  expectedSize = 100;
+    EXPECT_EQ( columnData->length(), expectedSize );
+    for ( int i = 0; i < expectedSize; i++ )
+    {
+        std::shared_ptr<arrow::Scalar>      scalar    = columnData->GetScalar( i ).ValueOrDie();
+        std::shared_ptr<arrow::Int64Scalar> intScalar = std::dynamic_pointer_cast<arrow::Int64Scalar>( scalar );
+        EXPECT_TRUE( scalar->Equals( arrow::Int64Scalar( i ) ) );
+    }
+}

CMakeLists.txt

@@ -636,6 +636,26 @@ add_subdirectory(ThirdParty/tomlplusplus)
 add_subdirectory(ThirdParty/spdlog)
 list(APPEND THIRD_PARTY_LIBRARIES spdlog)
 
+# ##############################################################################
+# Arrow
+# ##############################################################################
+find_package(Arrow CONFIG REQUIRED)
+if(MSVC)
+  list(APPEND THIRD_PARTY_LIBRARIES Arrow::arrow_shared)
+else()
+  list(APPEND THIRD_PARTY_LIBRARIES Arrow::arrow_static)
+endif()
+
+# ##############################################################################
+# Parquet
+# ##############################################################################
+find_package(Parquet CONFIG REQUIRED)
+if(MSVC)
+  list(APPEND THIRD_PARTY_LIBRARIES Parquet::parquet_shared)
+else()
+  list(APPEND THIRD_PARTY_LIBRARIES Parquet::parquet_static)
+endif()
+
 # ##############################################################################
 # Thirdparty libraries are put in ThirdParty solution folder
 # ##############################################################################

vcpkg_x64-linux.txt

@@ -2,5 +2,6 @@ grpc
 boost-filesystem
 boost-spirit
 eigen3
+arrow
 --triplet
 x64-linux

vcpkg_x64-osx.txt

@@ -2,5 +2,6 @@ grpc
 boost-filesystem
 boost-spirit
 eigen3
+arrow
 --triplet
 x64-osx

vcpkg_x64-windows.txt

@@ -2,5 +2,6 @@ grpc
 boost-filesystem
 boost-spirit
 eigen3
+arrow
 --triplet
 x64-windows