Merge pull request #14 from willtemperley/main

Reinstate C import / export

Author: willtemperley

Sources/Arrow/ArrowArray.swift

@@ -13,6 +13,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+import ArrowC
 import Foundation
 
 /// A type-erased ArrowArray.
@@ -25,9 +26,10 @@ public protocol AnyArrowArray {
   var bufferDataSizes: [Int] { get }
   func asAny(_ index: UInt) -> Any?
   func asString(_ index: UInt) -> String
+  func setCArrayPtr(_ cArrayPtr: UnsafePointer<ArrowC.ArrowArray>?)
 }
 
-// MARK: - Core Protocol
+// MARK: Core Protocol
 
 /// The interface for Arrow array types.
 public protocol ArrowArray<ItemType>: AnyArrowArray {
@@ -37,8 +39,42 @@ public protocol ArrowArray<ItemType>: AnyArrowArray {
   subscript(_ index: UInt) -> ItemType? { get }
 }
 
-// MARK: - Default Implementations
-extension ArrowArray {
+public class ArrowArrayBase<T>: ArrowArray {
+
+  public var arrowData: ArrowData
+  public var cArrayPtr: UnsafePointer<ArrowC.ArrowArray>? = nil
+
+  required public init(_ arrowData: ArrowData) throws(ArrowError) {
+    self.arrowData = arrowData
+  }
+
+  public subscript(_ index: UInt) -> T? {
+    fatalError("Base class is abstract.")
+  }
+
+  public func asString(_ index: UInt) -> String {
+    guard let value = self[index] else {
+      return ""
+    }
+    return "\(value)"
+  }
+
+  public func asAny(_ index: UInt) -> Any? {
+    self[index]
+  }
+
+  public func setCArrayPtr(_ cArrayPtr: UnsafePointer<ArrowC.ArrowArray>?) {
+    self.cArrayPtr = cArrayPtr
+  }
+
+  deinit {
+    if let cArrayPtr = cArrayPtr {
+      ArrowCImporter.release(cArrayPtr)
+    }
+  }
+}
+
+extension ArrowArrayBase {
   public var nullCount: UInt {
     arrowData.nullCount
   }
@@ -63,32 +99,19 @@ extension ArrowArray {
     arrowData.buffers.map { Int($0.capacity) }
   }
 
-  public func isNull(at index: UInt) throws -> Bool {
+  public func isNull(at index: UInt) throws(ArrowError) -> Bool {
     if index >= self.length {
-      throw ArrowError.outOfBounds(index: Int64(index))
+      throw .outOfBounds(index: Int64(index))
     }
     return arrowData.isNull(index)
   }
-
-  public func asString(_ index: UInt) -> String {
-    guard let value = self[index] else {
-      return ""
-    }
-    return "\(value)"
-  }
-
-  public func asAny(_ index: UInt) -> Any? {
-    self[index]
-  }
 }
 
 // MARK: Fixed Arrays
 
-public protocol FixedArrayProtocol: ArrowArray where ItemType: BitwiseCopyable {
-}
+public class FixedArray<T>: ArrowArrayBase<T> where T: BitwiseCopyable {
 
-extension FixedArrayProtocol {
-  public subscript(_ index: UInt) -> ItemType? {
+  public override subscript(_ index: UInt) -> ItemType? {
     if arrowData.isNull(index) {
       return nil
     }
@@ -106,28 +129,9 @@ extension FixedArrayProtocol {
   }
 }
 
-public struct FixedArray<T>: FixedArrayProtocol where T: BitwiseCopyable {
-  public typealias ItemType = T
-  public let arrowData: ArrowData
+public class StringArray: ArrowArrayBase<String> {
 
-  public init(arrowData: ArrowData) {
-    self.arrowData = arrowData
-  }
-
-  public init(_ arrowData: ArrowData) {
-    self.arrowData = arrowData
-  }
-}
-
-public struct StringArray: ArrowArray {
-  public typealias ItemType = String
-  public let arrowData: ArrowData
-
-  public init(_ arrowData: ArrowData) {
-    self.arrowData = arrowData
-  }
-
-  public subscript(_ index: UInt) -> String? {
+  public override subscript(_ index: UInt) -> String? {
     let offsetIndex = MemoryLayout<Int32>.stride * Int(index)
     if self.arrowData.isNull(index) {
       return nil
@@ -138,8 +142,8 @@ public struct StringArray: ArrowArray {
 
     var startIndex: Int32 = 0
     if index > 0 {
-      startIndex = offsets.rawPointer.advanced(by: offsetIndex).load(
-        as: Int32.self)
+      startIndex = offsets.rawPointer.advanced(by: offsetIndex)
+        .load(as: Int32.self)
     }
 
     let endIndex = offsets.rawPointer.advanced(
@@ -150,21 +154,16 @@ public struct StringArray: ArrowArray {
     let rawPointer = values.rawPointer.advanced(by: Int(startIndex))
       .bindMemory(to: UInt8.self, capacity: arrayLength)
     let buffer = UnsafeBufferPointer<UInt8>(
-      start: rawPointer, count: arrayLength)
-    let byteArray = Array(buffer)
-    return String(data: Data(byteArray), encoding: .utf8)
+      start: rawPointer,
+      count: arrayLength
+    )
+    return String(bytes: buffer, encoding: .utf8)
   }
 }
 
-public struct BoolArray: ArrowArray {
-  public typealias ItemType = Bool
-  public let arrowData: ArrowData
+public class BoolArray: ArrowArrayBase<Bool> {
 
-  public init(_ arrowData: ArrowData) {
-    self.arrowData = arrowData
-  }
-
-  public subscript(_ index: UInt) -> Bool? {
+  public override subscript(_ index: UInt) -> Bool? {
     if self.arrowData.isNull(index) {
       return nil
     }
@@ -173,15 +172,9 @@ public struct BoolArray: ArrowArray {
   }
 }
 
-public struct Date32Array: ArrowArray {
-  public typealias ItemType = Date
-  public let arrowData: ArrowData
+public class Date32Array: ArrowArrayBase<Date> {
 
-  public init(_ arrowData: ArrowData) {
-    self.arrowData = arrowData
-  }
-
-  public subscript(_ index: UInt) -> Date? {
+  public override subscript(_ index: UInt) -> Date? {
     if self.arrowData.isNull(index) {
       return nil
     }
@@ -193,15 +186,9 @@ public struct Date32Array: ArrowArray {
   }
 }
 
-public struct Date64Array: ArrowArray {
-  public typealias ItemType = Date
-  public let arrowData: ArrowData
+public class Date64Array: ArrowArrayBase<Date> {
 
-  public init(_ arrowData: ArrowData) {
-    self.arrowData = arrowData
-  }
-
-  public subscript(_ index: UInt) -> Date? {
+  public override subscript(_ index: UInt) -> Date? {
     if self.arrowData.isNull(index) {
       return nil
     }
@@ -217,13 +204,7 @@ public typealias Time64Array = FixedArray<Time64>
 
 public typealias Time32Array = FixedArray<Time32>
 
-public struct TimestampArray: FixedArrayProtocol {
-  public typealias ItemType = Timestamp
-  public let arrowData: ArrowData
-
-  public init(_ arrowData: ArrowData) {
-    self.arrowData = arrowData
-  }
+public class TimestampArray: FixedArray<Timestamp> {
 
   public struct FormattingOptions: Equatable {
     public var dateFormat: String = "yyyy-MM-dd HH:mm:ss.SSS"
@@ -256,7 +237,7 @@ public struct TimestampArray: FixedArrayProtocol {
   private var cachedFormatter: DateFormatter?
   private var cachedOptions: FormattingOptions?
 
-  public mutating func formattedDate(
+  public func formattedDate(
     at index: UInt,
     options: FormattingOptions = FormattingOptions()
   ) -> String? {
@@ -300,8 +281,7 @@ public struct TimestampArray: FixedArrayProtocol {
     return Date(timeIntervalSince1970: timeInterval)
   }
 
-  // TODO: Mutating function to hack around cached formatter
-  public mutating func asString(_ index: UInt) -> String {
+  public override func asString(_ index: UInt) -> String {
     if let formatted = formattedDate(at: index) {
       return formatted
     } else {
@@ -310,13 +290,7 @@ public struct TimestampArray: FixedArrayProtocol {
   }
 }
 
-public struct BinaryArray: ArrowArray {
-  public typealias ItemType = Data
-  public let arrowData: ArrowData
-
-  public init(_ arrowData: ArrowData) {
-    self.arrowData = arrowData
-  }
+public class BinaryArray: ArrowArrayBase<Data> {
 
   public struct Options {
     public var printAsHex = false
@@ -325,7 +299,7 @@ public struct BinaryArray: ArrowArray {
 
   public var options = Options()
 
-  public subscript(_ index: UInt) -> Data? {
+  public override subscript(_ index: UInt) -> Data? {
     let offsetIndex = MemoryLayout<Int32>.stride * Int(index)
     if self.arrowData.isNull(index) {
       return nil
@@ -350,7 +324,7 @@ public struct BinaryArray: ArrowArray {
     return Data(byteArray)
   }
 
-  public func asString(_ index: UInt) -> String {
+  public override func asString(_ index: UInt) -> String {
     guard let data = self[index] else { return "" }
     if options.printAsHex {
       return data.hexEncodedString()
@@ -364,13 +338,14 @@ public struct BinaryArray: ArrowArray {
   }
 }
 
-public struct NestedArray: ArrowArray, AnyArrowArray {
-  public typealias ItemType = [Any?]
-  public let arrowData: ArrowData
+public class NestedArray: ArrowArrayBase<[Any?]> {
+
   private var children: [AnyArrowArray]?
 
-  public init(_ arrowData: ArrowData) throws(ArrowError) {
-    self.arrowData = arrowData
+  public required init(
+    _ arrowData: ArrowData
+  ) throws(ArrowError) {
+    try super.init(arrowData)
 
     switch arrowData.type {
     case .list(let field):
@@ -398,7 +373,7 @@ public struct NestedArray: ArrowArray, AnyArrowArray {
     }
   }
 
-  public subscript(_ index: UInt) -> [Any?]? {
+  public override subscript(_ index: UInt) -> [Any?]? {
     if self.arrowData.isNull(index) {
       return nil
     }
@@ -432,7 +407,7 @@ public struct NestedArray: ArrowArray, AnyArrowArray {
     }
   }
 
-  public func asString(_ index: UInt) -> String {
+  public override func asString(_ index: UInt) -> String {
     switch arrowData.type {
     case .list(let _):
       if self.arrowData.isNull(index) {

Sources/Arrow/ArrowArrayBuilder.swift

@@ -82,9 +82,10 @@ extension ArrowArrayBuilderInternal {
 
 // MARK: Base implementation.
 
-// Note: It would be preferable to move all of this to a protocol, however
+// Note: It would be preferable to move this to a protocol, however
 // ListArrayBuilder overrides finish. This is delicate because protocol
-// extension method dispatching means the
+// extension method dispatching uses static dispatch, so overrides are not
+// called when type erasure is used in nested types.
 public class ArrowArrayBuilderBase<
   BufferBuilder: ArrowBufferBuilder,
   ArrayType: ArrowArray<BufferBuilder.ItemType>

Sources/Arrow/ArrowArrayLoader.swift

@@ -21,41 +21,41 @@ struct ArrowArrayLoader {
   ) throws(ArrowError) -> any AnyArrowArray {
     switch arrowType {
     case .int8:
-      return FixedArray<Int8>(arrowData)
+      return try FixedArray<Int8>(arrowData)
     case .int16:
-      return FixedArray<Int16>(arrowData)
+      return try FixedArray<Int16>(arrowData)
     case .int32:
-      return FixedArray<Int32>(arrowData)
+      return try FixedArray<Int32>(arrowData)
     case .int64:
-      return FixedArray<Int64>(arrowData)
+      return try FixedArray<Int64>(arrowData)
     case .uint8:
-      return FixedArray<UInt8>(arrowData)
+      return try FixedArray<UInt8>(arrowData)
     case .uint16:
-      return FixedArray<UInt16>(arrowData)
+      return try FixedArray<UInt16>(arrowData)
     case .uint32:
-      return FixedArray<UInt32>(arrowData)
+      return try FixedArray<UInt32>(arrowData)
     case .uint64:
-      return FixedArray<UInt64>(arrowData)
+      return try FixedArray<UInt64>(arrowData)
     case .float64:
-      return FixedArray<Double>(arrowData)
+      return try FixedArray<Double>(arrowData)
     case .float32:
-      return FixedArray<Float>(arrowData)
+      return try FixedArray<Float>(arrowData)
     case .date32:
-      return Date32Array(arrowData)
+      return try Date32Array(arrowData)
     case .date64:
-      return Date64Array(arrowData)
+      return try Date64Array(arrowData)
     case .time32:
-      return Time32Array(arrowData)
+      return try Time32Array(arrowData)
     case .time64:
-      return Time64Array(arrowData)
+      return try Time64Array(arrowData)
     case .timestamp:
-      return TimestampArray(arrowData)
+      return try TimestampArray(arrowData)
     case .utf8:
-      return StringArray(arrowData)
+      return try StringArray(arrowData)
     case .boolean:
-      return BoolArray(arrowData)
+      return try BoolArray(arrowData)
     case .binary:
-      return BinaryArray(arrowData)
+      return try BinaryArray(arrowData)
     case .strct(let _):
       return try NestedArray(arrowData)
     case .list(let _):