Skip to content

Commit 91d0f1f

Browse files
committed
Add missing implementation for datetime relevant arrow type
# Conflicts: # src/Microsoft.Data.Analysis/PrimitiveDataFrameColumn.cs
1 parent e88b642 commit 91d0f1f

File tree

5 files changed

+58
-29
lines changed

5 files changed

+58
-29
lines changed

eng/Versions.props

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -30,7 +30,7 @@
3030
<SystemTextJsonVersion>6.0.1</SystemTextJsonVersion>
3131
<SystemThreadingChannelsVersion>4.7.1</SystemThreadingChannelsVersion>
3232
<!-- Other product dependencies -->
33-
<ApacheArrowVersion>2.0.0</ApacheArrowVersion>
33+
<ApacheArrowVersion>11.0.0</ApacheArrowVersion>
3434
<GoogleProtobufVersion>3.19.6</GoogleProtobufVersion>
3535
<LightGBMVersion>2.3.1</LightGBMVersion>
3636
<MicrosoftCodeAnalysisAnalyzersVersion>3.3.0</MicrosoftCodeAnalysisAnalyzersVersion>

src/Microsoft.Data.Analysis/DataFrame.Arrow.cs

Lines changed: 12 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -101,10 +101,18 @@ private static void AppendDataFrameColumnFromArrowArray(Field field, IArrowArray
101101
AppendDataFrameColumnFromArrowArray(fieldsEnumerator.Current, structArrayEnumerator.Current, ret, field.Name + "_");
102102
}
103103
break;
104-
case ArrowTypeId.Decimal:
104+
case ArrowTypeId.Date64:
105+
Date64Array arrowDate64Array = (Date64Array)arrowArray;
106+
dataFrameColumn = new PrimitiveDataFrameColumn<DateTime>(fieldName, arrowDate64Array.Data.Length);
107+
for (int i = 0; i < arrowDate64Array.Data.Length; i++)
108+
{
109+
dataFrameColumn[i] = arrowDate64Array.GetDateTime(i);
110+
}
111+
break;
112+
case ArrowTypeId.Decimal128:
113+
case ArrowTypeId.Decimal256:
105114
case ArrowTypeId.Binary:
106115
case ArrowTypeId.Date32:
107-
case ArrowTypeId.Date64:
108116
case ArrowTypeId.Dictionary:
109117
case ArrowTypeId.FixedSizedBinary:
110118
case ArrowTypeId.HalfFloat:
@@ -114,6 +122,7 @@ private static void AppendDataFrameColumnFromArrowArray(Field field, IArrowArray
114122
case ArrowTypeId.Null:
115123
case ArrowTypeId.Time32:
116124
case ArrowTypeId.Time64:
125+
case ArrowTypeId.Timestamp:
117126
default:
118127
throw new NotImplementedException($"{fieldType.Name}");
119128
}
@@ -145,7 +154,7 @@ public static DataFrame FromArrowRecordBatch(RecordBatch recordBatch)
145154
}
146155

147156
/// <summary>
148-
/// Returns an <see cref="IEnumerable{RecordBatch}"/> without copying data
157+
/// Returns an <see cref="IEnumerable{RecordBatch}"/> mostly without copying data
149158
/// </summary>
150159
public IEnumerable<RecordBatch> ToArrowRecordBatches()
151160
{

src/Microsoft.Data.Analysis/PrimitiveColumnContainer.cs

Lines changed: 0 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -374,18 +374,6 @@ internal int MaxRecordBatchLength(long startIndex)
374374
return Buffers[arrayIndex].Length - (int)startIndex;
375375
}
376376

377-
internal ReadOnlyMemory<byte> GetValueBuffer(long startIndex)
378-
{
379-
int arrayIndex = GetArrayContainingRowIndex(startIndex);
380-
return Buffers[arrayIndex].ReadOnlyBuffer;
381-
}
382-
383-
internal ReadOnlyMemory<byte> GetNullBuffer(long startIndex)
384-
{
385-
int arrayIndex = GetArrayContainingRowIndex(startIndex);
386-
return NullBitMapBuffers[arrayIndex].ReadOnlyBuffer;
387-
}
388-
389377
public IReadOnlyList<T?> this[long startIndex, int length]
390378
{
391379
get

src/Microsoft.Data.Analysis/PrimitiveDataFrameColumn.cs

Lines changed: 44 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -7,6 +7,7 @@
77
using System.Collections.Generic;
88
using System.Diagnostics;
99
using System.Runtime.CompilerServices;
10+
using System.Runtime.InteropServices;
1011
using Apache.Arrow;
1112
using Apache.Arrow.Types;
1213
using Microsoft.ML;
@@ -104,6 +105,8 @@ private IArrowType GetArrowType()
104105
return UInt64Type.Default;
105106
else if (typeof(T) == typeof(ushort))
106107
return UInt16Type.Default;
108+
else if (typeof(T) == typeof(DateTime))
109+
return Date64Type.Default;
107110
else
108111
throw new NotImplementedException(nameof(T));
109112
}
@@ -127,36 +130,64 @@ protected internal override Apache.Arrow.Array ToArrowArray(long startIndex, int
127130
{
128131
int arrayIndex = numberOfRows == 0 ? 0 : _columnContainer.GetArrayContainingRowIndex(startIndex);
129132
int offset = (int)(startIndex - arrayIndex * ReadOnlyDataFrameBuffer<T>.MaxCapacity);
133+
130134
if (numberOfRows != 0 && numberOfRows > _columnContainer.Buffers[arrayIndex].Length - offset)
131135
{
132136
throw new ArgumentException(Strings.SpansMultipleBuffers, nameof(numberOfRows));
133137
}
134-
ArrowBuffer valueBuffer = numberOfRows == 0 ? ArrowBuffer.Empty : new ArrowBuffer(_columnContainer.GetValueBuffer(startIndex));
135-
ArrowBuffer nullBuffer = numberOfRows == 0 ? ArrowBuffer.Empty : new ArrowBuffer(_columnContainer.GetNullBuffer(startIndex));
138+
136139
int nullCount = GetNullCount(startIndex, numberOfRows);
140+
141+
//DateTime requires convertion
142+
if (this.DataType == typeof(DateTime))
143+
{
144+
if (numberOfRows == 0)
145+
return new Date64Array(ArrowBuffer.Empty, ArrowBuffer.Empty, numberOfRows, nullCount, offset);
146+
147+
ReadOnlyDataFrameBuffer<T> valueBuffer = (numberOfRows == 0) ? null : _columnContainer.Buffers[arrayIndex];
148+
ReadOnlyDataFrameBuffer<byte> nullBuffer = (numberOfRows == 0) ? null : _columnContainer.NullBitMapBuffers[arrayIndex];
149+
150+
ReadOnlySpan<DateTime> valueSpan = MemoryMarshal.Cast<T, DateTime>(valueBuffer.ReadOnlySpan);
151+
Date64Array.Builder builder = new Date64Array.Builder().Reserve(valueBuffer.Length);
152+
153+
for (int i = 0; i < valueBuffer.Length; i++)
154+
{
155+
if (BitUtility.GetBit(nullBuffer.ReadOnlySpan, i))
156+
builder.Append(valueSpan[i]);
157+
else
158+
builder.AppendNull();
159+
}
160+
161+
return builder.Build();
162+
}
163+
164+
//No convertion
165+
ArrowBuffer arrowValueBuffer = numberOfRows == 0 ? ArrowBuffer.Empty : new ArrowBuffer(_columnContainer.Buffers[arrayIndex].ReadOnlyBuffer);
166+
ArrowBuffer arrowNullBuffer = numberOfRows == 0 ? ArrowBuffer.Empty : new ArrowBuffer(_columnContainer.NullBitMapBuffers[arrayIndex].ReadOnlyBuffer);
167+
137168
Type type = this.DataType;
138169
if (type == typeof(bool))
139-
return new BooleanArray(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
170+
return new BooleanArray(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
140171
else if (type == typeof(double))
141-
return new DoubleArray(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
172+
return new DoubleArray(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
142173
else if (type == typeof(float))
143-
return new FloatArray(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
174+
return new FloatArray(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
144175
else if (type == typeof(int))
145-
return new Int32Array(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
176+
return new Int32Array(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
146177
else if (type == typeof(long))
147-
return new Int64Array(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
178+
return new Int64Array(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
148179
else if (type == typeof(sbyte))
149-
return new Int8Array(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
180+
return new Int8Array(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
150181
else if (type == typeof(short))
151-
return new Int16Array(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
182+
return new Int16Array(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
152183
else if (type == typeof(uint))
153-
return new UInt32Array(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
184+
return new UInt32Array(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
154185
else if (type == typeof(ulong))
155-
return new UInt64Array(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
186+
return new UInt64Array(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
156187
else if (type == typeof(ushort))
157-
return new UInt16Array(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
188+
return new UInt16Array(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
158189
else if (type == typeof(byte))
159-
return new UInt8Array(valueBuffer, nullBuffer, numberOfRows, nullCount, offset);
190+
return new UInt8Array(arrowValueBuffer, arrowNullBuffer, numberOfRows, nullCount, offset);
160191
else
161192
throw new NotImplementedException(type.ToString());
162193
}

test/Microsoft.Data.Analysis.Tests/ArrowIntegrationTests.cs

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -48,6 +48,7 @@ public void TestArrowIntegration()
4848
.Append("ULongColumn", false, new UInt64Array.Builder().AppendRange(Enumerable.Repeat((ulong)1, 10)).Build())
4949
.Append("ByteColumn", false, new Int8Array.Builder().AppendRange(Enumerable.Repeat((sbyte)1, 10)).Build())
5050
.Append("UByteColumn", false, new UInt8Array.Builder().AppendRange(Enumerable.Repeat((byte)1, 10)).Build())
51+
.Append("Date64Column", false, new Date64Array.Builder().AppendRange(Enumerable.Repeat(DateTime.Now, 10)).Build())
5152
.Build();
5253

5354
DataFrame df = DataFrame.FromArrowRecordBatch(originalBatch);

0 commit comments

Comments
 (0)