Add ListType & ArrowExtensionsEnabled to ArrowReaderProperties

cpp/Enums.cpp

@@ -106,5 +106,9 @@ namespace
 		static_assert((int) ::arrow::Type::type::BINARY == 14);
 		static_assert((int) ::arrow::Type::type::LARGE_BINARY == 35);
 		static_assert((int) ::arrow::Type::type::BINARY_VIEW == 40);
+
+		static_assert((int) ::arrow::Type::type::LIST == 25);
+		static_assert((int) ::arrow::Type::type::LARGE_LIST == 36);
+		static_assert((int) ::arrow::Type::type::LIST_VIEW == 41);
 	}
 }

cpp/arrow/ArrowReaderProperties.cpp

@@ -76,4 +76,24 @@ extern "C"
   {
     TRYCATCH(properties->set_binary_type(value);)
   }
+
+  PARQUETSHARP_EXPORT ExceptionInfo* ArrowReaderProperties_ListType(ArrowReaderProperties* properties, ::arrow::Type::type* value)
+  {
+    TRYCATCH(*value = properties->list_type();)
+  }
+
+  PARQUETSHARP_EXPORT ExceptionInfo* ArrowReaderProperties_SetListType(ArrowReaderProperties* properties, ::arrow::Type::type value)
+  {
+    TRYCATCH(properties->set_list_type(value);)
+  }
+
+  PARQUETSHARP_EXPORT ExceptionInfo* ArrowReaderProperties_GetArrowExtensionEnabled(ArrowReaderProperties* properties, bool* extensions_enabled)
+  {
+    TRYCATCH(*extensions_enabled = properties->get_arrow_extensions_enabled();)
+  }
+
+  PARQUETSHARP_EXPORT ExceptionInfo* ArrowReaderProperties_SetArrowExtensionEnabled(ArrowReaderProperties* properties, bool extensions_enabled)
+  {
+    TRYCATCH(properties->set_arrow_extensions_enabled(extensions_enabled);)
+  }
 }

csharp.test/Arrow/TestArrowReaderProperties.cs

@@ -16,6 +16,9 @@ public void TestDefaultProperties()
             Assert.That(properties.GetReadDictionary(0), Is.False);
             Assert.That(properties.PreBuffer, Is.True);
             Assert.That(properties.CoerceInt96TimestampUnit, Is.EqualTo(Apache.Arrow.Types.TimeUnit.Nanosecond));
+            Assert.That(properties.BinaryType, Is.EqualTo(Apache.Arrow.Types.ArrowTypeId.Binary));
+            Assert.That(properties.ListType, Is.EqualTo(Apache.Arrow.Types.ArrowTypeId.List));
+            Assert.That(properties.ArrowExtensionEnabled, Is.False);
         }
 
         [Test]
@@ -29,13 +32,17 @@ public void TestSetProperties()
             properties.PreBuffer = false;
             properties.CoerceInt96TimestampUnit = Apache.Arrow.Types.TimeUnit.Microsecond;
             properties.BinaryType = Apache.Arrow.Types.ArrowTypeId.LargeBinary;
+            properties.ListType = Apache.Arrow.Types.ArrowTypeId.LargeList;
+            properties.ArrowExtensionEnabled = true;
 
             Assert.That(properties.UseThreads, Is.True);
             Assert.That(properties.BatchSize, Is.EqualTo(789));
             Assert.That(properties.GetReadDictionary(0), Is.True);
             Assert.That(properties.PreBuffer, Is.False);
             Assert.That(properties.CoerceInt96TimestampUnit, Is.EqualTo(Apache.Arrow.Types.TimeUnit.Microsecond));
             Assert.That(properties.BinaryType, Is.EqualTo(Apache.Arrow.Types.ArrowTypeId.LargeBinary));
+            Assert.That(properties.ListType, Is.EqualTo(Apache.Arrow.Types.ArrowTypeId.LargeList));
+            Assert.That(properties.ArrowExtensionEnabled, Is.True);
         }
     }
 }

csharp/Arrow/ArrowReaderProperties.cs

@@ -133,6 +133,44 @@ public Apache.Arrow.Types.ArrowTypeId BinaryType
             }
         }
 
+        /// <summary>
+        /// The Arrow list type to read Parquet list columns as.
+        /// 
+        /// Allowed values are ArrowTypeId.List, ArrowTypeId.LargeList and ArrowTypeId.ListView.
+        /// Default is ArrowTypeId.List.
+        ///
+        /// If a serialized Arrow schema is found in the Parquet metadata,
+        /// this setting is ignored and the Arrow schema takes precedence
+        /// </summary>
+        public Apache.Arrow.Types.ArrowTypeId ListType
+        {
+            get
+            {
+                ParquetSharp.CppTypeId value = ExceptionInfo.Return<ParquetSharp.CppTypeId>(Handle, ArrowReaderProperties_ListType);
+                return value.toPublicEnum();
+            }
+            set
+            {
+                ParquetSharp.CppTypeId cppValue = value.toCppEnum();
+                ExceptionInfo.Check(ArrowReaderProperties_SetListType(Handle.IntPtr, cppValue));
+                GC.KeepAlive(Handle);
+            }
+        }
+
+        /// <summary>
+        /// Whether to enable Parquet-supported Arrow extension types.
+        /// Default is false.
+        /// </summary>
+        public bool ArrowExtensionEnabled
+        {
+            get => ExceptionInfo.Return<bool>(Handle, ArrowReaderProperties_GetArrowExtensionEnabled);
+            set
+            {
+                ExceptionInfo.Check(ArrowReaderProperties_SetArrowExtensionEnabled(Handle.IntPtr, value));
+                GC.KeepAlive(Handle);
+            }
+        }
+
         [DllImport(ParquetDll.Name)]
         private static extern IntPtr ArrowReaderProperties_GetDefault(out IntPtr readerProperties);
 
@@ -175,6 +213,18 @@ public Apache.Arrow.Types.ArrowTypeId BinaryType
         [DllImport(ParquetDll.Name)]
         private static extern IntPtr ArrowReaderProperties_SetBinaryType(IntPtr readerProperties, ParquetSharp.CppTypeId value);
 
+        [DllImport(ParquetDll.Name)]
+        private static extern IntPtr ArrowReaderProperties_ListType(IntPtr readerProperties, out ParquetSharp.CppTypeId value);
+
+        [DllImport(ParquetDll.Name)]
+        private static extern IntPtr ArrowReaderProperties_SetListType(IntPtr readerProperties, ParquetSharp.CppTypeId value);
+
+        [DllImport(ParquetDll.Name)]
+        private static extern IntPtr ArrowReaderProperties_GetArrowExtensionEnabled(IntPtr readerProperties, out bool extensionsEnabled);
+
+        [DllImport(ParquetDll.Name)]
+        private static extern IntPtr ArrowReaderProperties_SetArrowExtensionEnabled(IntPtr readerProperties, bool extensionsEnabled);
+
         internal readonly ParquetHandle Handle;
     }
 }

csharp/ArrowTypeIdExtensions.cs

@@ -10,6 +10,9 @@ internal static class ArrowTypeIdExtensions
             Apache.Arrow.Types.ArrowTypeId.Binary => ParquetSharp.CppTypeId.Binary,
             Apache.Arrow.Types.ArrowTypeId.LargeBinary => ParquetSharp.CppTypeId.LargeBinary,
             Apache.Arrow.Types.ArrowTypeId.BinaryView => ParquetSharp.CppTypeId.BinaryView,
+            Apache.Arrow.Types.ArrowTypeId.List => ParquetSharp.CppTypeId.List,
+            Apache.Arrow.Types.ArrowTypeId.LargeList => ParquetSharp.CppTypeId.LargeList,
+            Apache.Arrow.Types.ArrowTypeId.ListView => ParquetSharp.CppTypeId.ListView,
             _ => throw new ArgumentOutOfRangeException(nameof(arrowTypeId), arrowTypeId, null)
         };
     }

csharp/CppTypeId.cs

@@ -8,6 +8,9 @@ internal enum CppTypeId
         Binary = 14,
         LargeBinary = 35,
         BinaryView = 40,
+        List = 25,
+        LargeList = 36,
+        ListView = 41
     }
 
     internal static class CppTypeIdExtensions
@@ -17,6 +20,9 @@ internal static class CppTypeIdExtensions
             CppTypeId.Binary => Apache.Arrow.Types.ArrowTypeId.Binary,
             CppTypeId.LargeBinary => Apache.Arrow.Types.ArrowTypeId.LargeBinary,
             CppTypeId.BinaryView => Apache.Arrow.Types.ArrowTypeId.BinaryView,
+            CppTypeId.List => Apache.Arrow.Types.ArrowTypeId.List,
+            CppTypeId.LargeList => Apache.Arrow.Types.ArrowTypeId.LargeList,
+            CppTypeId.ListView => Apache.Arrow.Types.ArrowTypeId.ListView,
             _ => throw new ArgumentOutOfRangeException(nameof(binaryType), binaryType, null)
         };
     }

csharp/PublicAPI.Unshipped.txt

@@ -1,6 +1,10 @@
 #nullable enable
+ParquetSharp.Arrow.ArrowReaderProperties.ArrowExtensionEnabled.get -> bool
+ParquetSharp.Arrow.ArrowReaderProperties.ArrowExtensionEnabled.set -> void
 ParquetSharp.Arrow.ArrowReaderProperties.BinaryType.get -> Apache.Arrow.Types.ArrowTypeId
 ParquetSharp.Arrow.ArrowReaderProperties.BinaryType.set -> void
+ParquetSharp.Arrow.ArrowReaderProperties.ListType.get -> Apache.Arrow.Types.ArrowTypeId
+ParquetSharp.Arrow.ArrowReaderProperties.ListType.set -> void
 ParquetSharp.ReaderProperties.ThriftStringSizeLimit.get -> int
 ParquetSharp.ReaderProperties.SetThriftStringSizeLimit(int size) -> void
 ParquetSharp.ReaderProperties.ThriftContainerSizeLimit.get -> int