Merge pull request #13 from willtemperley/main

Add test to ensure record batch schema nullability and data match.

Author: willtemperley

Sources/Arrow/ArrowArrayBuilder.swift

@@ -19,6 +19,8 @@ import Foundation
 
 /// A type which builds a type-erased `ArrowArray`.
 public protocol AnyArrowArrayBuilder {
+  /// Returns an unparameterised `ArrowArray`.
+  /// - Returns: The type-erased Arrow array.
   func toAnyArrowArray() throws(ArrowError) -> AnyArrowArray
   func appendAny(_ val: Any?)
 }
@@ -73,8 +75,6 @@ extension ArrowArrayBuilderInternal {
     self.arrowType.getStride()
   }
 
-  /// Returns an unparameterised `ArrowArray`.
-  /// - Returns: The type-erased Arrow array.
   public func toAnyArrowArray() throws(ArrowError) -> AnyArrowArray {
     try self.finish()
   }
@@ -109,7 +109,7 @@ public class ArrowArrayBuilderBase<
   }
 }
 
-/// A type which builds an `ArrowArray` with a numeric `ItemType`.
+/// An array builder for numeric types.
 public class NumberArrayBuilder<ItemType>: ArrowArrayBuilderBase<
   FixedBufferBuilder<ItemType>,
   FixedArray<ItemType>
@@ -120,6 +120,7 @@ where ItemType: Numeric, ItemType: BitwiseCopyable {
   }
 }
 
+/// A `String` array builder.
 public class StringArrayBuilder: ArrowArrayBuilderBase<
   VariableBufferBuilder<String>,
   StringArray
@@ -130,6 +131,7 @@ public class StringArrayBuilder: ArrowArrayBuilderBase<
   }
 }
 
+/// A `Data` array builder.
 public class BinaryArrayBuilder: ArrowArrayBuilderBase<
   VariableBufferBuilder<Data>,
   BinaryArray
@@ -140,6 +142,7 @@ public class BinaryArrayBuilder: ArrowArrayBuilderBase<
   }
 }
 
+/// A  `Bool` array builder.
 public class BoolArrayBuilder: ArrowArrayBuilderBase<
   BoolBufferBuilder, BoolArray
 >
@@ -149,6 +152,7 @@ public class BoolArrayBuilder: ArrowArrayBuilderBase<
   }
 }
 
+/// A 32-bit date array builder.
 public class Date32ArrayBuilder: ArrowArrayBuilderBase<
   Date32BufferBuilder,
   Date32Array
@@ -159,6 +163,7 @@ public class Date32ArrayBuilder: ArrowArrayBuilderBase<
   }
 }
 
+/// A 64-bit date array builder.
 public class Date64ArrayBuilder: ArrowArrayBuilderBase<
   Date64BufferBuilder,
   Date64Array
@@ -169,6 +174,7 @@ public class Date64ArrayBuilder: ArrowArrayBuilderBase<
   }
 }
 
+// A 32-bit elaspsed time builder.
 public class Time32ArrayBuilder: ArrowArrayBuilderBase<
   FixedBufferBuilder<Time32>,
   Time32Array
@@ -179,6 +185,7 @@ public class Time32ArrayBuilder: ArrowArrayBuilderBase<
   }
 }
 
+// A 64-bit elaspsed time builder.
 public class Time64ArrayBuilder: ArrowArrayBuilderBase<
   FixedBufferBuilder<Time64>,
   Time64Array
@@ -189,6 +196,7 @@ public class Time64ArrayBuilder: ArrowArrayBuilderBase<
   }
 }
 
+// A Timestamp array builder.
 public class TimestampArrayBuilder: ArrowArrayBuilderBase<
   FixedBufferBuilder<Int64>,
   TimestampArray
@@ -203,6 +211,7 @@ public class TimestampArrayBuilder: ArrowArrayBuilderBase<
 
 // MARK: Struct array builder.
 
+/// Builds an array of structs.
 public class StructArrayBuilder: ArrowArrayBuilderBase<
   StructBufferBuilder,
   NestedArray
@@ -262,7 +271,9 @@ public class StructArrayBuilder: ArrowArrayBuilderBase<
 
 // MARK: List array builder.
 
-/// A type which can build an `NestedArray`containing exactly `ItemType`.
+/// Builds a `NestedArray`containing lists of `ItemType`.
+///
+/// Both lists and items in lists are nullablie.
 public class ListArrayBuilder: ArrowArrayBuilderBase<
   ListBufferBuilder,
   NestedArray
@@ -271,18 +282,17 @@ public class ListArrayBuilder: ArrowArrayBuilderBase<
   let valueBuilder: any AnyArrowArrayBuilder
 
   public override init(_ elementType: ArrowType) throws(ArrowError) {
-
     guard case .list(let field) = elementType else {
       throw .invalid("Expected a field with type .list")
     }
-
     self.valueBuilder = try ArrowArrayBuilders.loadBuilder(
       arrowType: field.type
     )
     try super.init(elementType)
   }
 
-  // Overrides the default
+  // Overrides the protocol extension.
+  // Swift currently provides no marker for this.
   public func append(_ values: [Any?]?) {
     self.bufferBuilder.append(values)
     if let vals = values {

Sources/Arrow/ArrowBuffer.swift

@@ -1,5 +1,5 @@
 // Copyright 2025 The Apache Software Foundation
-// Copyright 2025 The Columnar-Swift Contributors
+// Copyright 2025 The Columnar Swift Contributors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -24,7 +24,8 @@ public class ArrowBuffer {
   let isMemoryOwner: Bool
 
   init(
-    length: UInt, capacity: UInt, rawPointer: UnsafeMutableRawPointer,
+    length: UInt, capacity: UInt,
+    rawPointer: UnsafeMutableRawPointer,
     isMemoryOwner: Bool = true
   ) {
     self.length = length

Sources/Arrow/ArrowSchema.swift

@@ -14,7 +14,8 @@
 
 import Foundation
 
-public struct ArrowSchema: Sendable {
+// Note this is a reference type to reduce copying.
+public final class ArrowSchema: Sendable {
   public let fields: [ArrowField]
   public let fieldLookup: [String: Int]
   init(_ fields: [ArrowField]) {

Sources/Arrow/ArrowTable.swift

@@ -1,4 +1,5 @@
 // Copyright 2025 The Apache Software Foundation
+// Copyright 2025 The Columnar Swift Contributors
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -244,6 +245,12 @@ public class RecordBatch {
       return self
     }
 
+    /// Add a column the `RecordBatch` builder.
+    /// - Parameters:
+    ///   - field: The field describing the array.
+    ///   - arrowArray: The array to add to the reocrd batch.
+    /// - Returns: The `RecordBatch.Builder` with the array appended and the field added to
+    /// the schema.
     @discardableResult
     public func addColumn(
       _ field: ArrowField,
@@ -263,6 +270,17 @@ public class RecordBatch {
           }
         }
       }
+      // Check nullability matches actual data
+      let schema = self.schemaBuilder.finish()
+      for (index, field) in schema.fields.enumerated() {
+        let column = columns[index]
+        if !field.isNullable && column.nullCount > 0 {
+          return .failure(
+            .invalid(
+              "non-nullable column '\(field.name)' contains \(column.nullCount) null values."
+            ))
+        }
+      }
       return .success(
         RecordBatch(self.schemaBuilder.finish(), columns: self.columns)
       )

Sources/Arrow/ArrowType.swift

@@ -210,6 +210,7 @@ public indirect enum ArrowType: Codable, Sendable, Equatable {
   /// of binary data in total.
   case binary
   /// Opaque binary data of fixed size.
+  ///
   /// Enum parameter specifies the number of bytes per value.
   case fixedSizeBinary(Int32)
   /// Opaque binary data of variable length and 64-bit offsets.

Tests/ArrowTests/RecordBatchTests.swift

@@ -54,4 +54,21 @@ struct RecordBatchTests {
       throw error
     }
   }
+
+  // Ensure that invalid record batches can't be built.
+  @Test func schemaNullabilityChecked() throws {
+    let stringBuilder = try ArrowArrayBuilders.loadStringArrayBuilder()
+    stringBuilder.append("test10")
+    stringBuilder.append(nil)
+    stringBuilder.append("test33")
+    let array = try stringBuilder.finish()
+
+    let field = ArrowField(name: "col1", dataType: .utf8, isNullable: false)
+    let result = RecordBatch.Builder()
+      .addColumn(field, arrowArray: array)
+      .finish()
+    if case .success(_) = result {
+      Issue.record("Record batch should have rejected null data.")
+    }
+  }
 }