From b7e68dbd5cbd0fe144cf5fef5a892835c5b1539e Mon Sep 17 00:00:00 2001
From: Matt Topol <zotthewizard@gmail.com>
Date: Mon, 28 Oct 2024 16:35:13 -0400
Subject: [PATCH] fix(arrow/cdata): handle export struct with no fields (#175)

Fixes #172

Includes a test that reproduced the original reported issue
---
 arrow/cdata/cdata_exports.go        |  7 +++++++
 arrow/cdata/cdata_test.go           | 25 +++++++++++++++++++++++++
 arrow/cdata/cdata_test_framework.go |  8 +++++++-
 3 files changed, 39 insertions(+), 1 deletion(-)

diff --git a/arrow/cdata/cdata_exports.go b/arrow/cdata/cdata_exports.go
index d3673481..6d1e0383 100644
--- a/arrow/cdata/cdata_exports.go
+++ b/arrow/cdata/cdata_exports.go
@@ -401,6 +401,9 @@ func exportArray(arr arrow.Array, out *CArrowArray, outSchema *CArrowSchema) {
 		out.children = (**CArrowArray)(unsafe.Pointer(&childPtrs[0]))
 	case *array.Struct:
 		out.n_children = C.int64_t(arr.NumField())
+		if arr.NumField() == 0 {
+			return
+		}
 		childPtrs := allocateArrowArrayPtrArr(arr.NumField())
 		children := allocateArrowArrayArr(arr.NumField())
 		for i := 0; i < arr.NumField(); i++ {
@@ -421,6 +424,10 @@ func exportArray(arr arrow.Array, out *CArrowArray, outSchema *CArrowSchema) {
 		exportArray(arr.Dictionary(), out.dictionary, nil)
 	case array.Union:
 		out.n_children = C.int64_t(arr.NumFields())
+		if arr.NumFields() == 0 {
+			return
+		}
+
 		childPtrs := allocateArrowArrayPtrArr(arr.NumFields())
 		children := allocateArrowArrayArr(arr.NumFields())
 		for i := 0; i < arr.NumFields(); i++ {
diff --git a/arrow/cdata/cdata_test.go b/arrow/cdata/cdata_test.go
index 2a86ea62..ebd4fcfc 100644
--- a/arrow/cdata/cdata_test.go
+++ b/arrow/cdata/cdata_test.go
@@ -598,6 +598,28 @@ func createTestStructArr() arrow.Array {
 	return bld.NewArray()
 }
 
+func createTestEmptyStructArr() arrow.Array {
+	bld := array.NewStructBuilder(memory.DefaultAllocator, arrow.StructOf())
+	defer bld.Release()
+
+	bld.AppendNull()
+	return bld.NewArray()
+}
+
+func createTestEmptyDenseUnionArr() arrow.Array {
+	bld := array.NewEmptyDenseUnionBuilder(memory.DefaultAllocator)
+	defer bld.Release()
+
+	return bld.NewArray()
+}
+
+func createTestEmptySparseUnionArr() arrow.Array {
+	bld := array.NewEmptySparseUnionBuilder(memory.DefaultAllocator)
+	defer bld.Release()
+
+	return bld.NewArray()
+}
+
 func createTestRunEndsArr() arrow.Array {
 	bld := array.NewRunEndEncodedBuilder(memory.DefaultAllocator,
 		arrow.PrimitiveTypes.Int32, arrow.PrimitiveTypes.Int8)
@@ -687,6 +709,9 @@ func TestNestedArrays(t *testing.T) {
 		{"sparse union", createTestSparseUnion},
 		{"dense union", createTestDenseUnion},
 		{"run-end encoded", createTestRunEndsArr},
+		{"empty struct", createTestEmptyStructArr},
+		{"empty dense union", createTestEmptyDenseUnionArr},
+		{"empty sparse union", createTestEmptySparseUnionArr},
 	}
 
 	for _, tt := range tests {
diff --git a/arrow/cdata/cdata_test_framework.go b/arrow/cdata/cdata_test_framework.go
index c979c3fb..331e80bc 100644
--- a/arrow/cdata/cdata_test_framework.go
+++ b/arrow/cdata/cdata_test_framework.go
@@ -309,6 +309,9 @@ func createCArr(arr arrow.Array, alloc *mallocator.Mallocator) *CArrowArray {
 		children = (**CArrowArray)(unsafe.Pointer(&clist[0]))
 		nchildren += 1
 	case *array.Struct:
+		if arr.NumField() == 0 {
+			break
+		}
 		clist := allocateChildrenPtrArr(alloc, arr.NumField())
 		for i := 0; i < arr.NumField(); i++ {
 			clist[i] = createCArr(arr.Field(i), alloc)
@@ -322,6 +325,9 @@ func createCArr(arr arrow.Array, alloc *mallocator.Mallocator) *CArrowArray {
 		children = (**CArrowArray)(unsafe.Pointer(&clist[0]))
 		nchildren += 2
 	case array.Union:
+		if arr.NumFields() == 0 {
+			break
+		}
 		clist := allocateChildrenPtrArr(alloc, arr.NumFields())
 		for i := 0; i < arr.NumFields(); i++ {
 			clist[i] = createCArr(arr.Field(i), alloc)
@@ -356,7 +362,7 @@ func createCArr(arr arrow.Array, alloc *mallocator.Mallocator) *CArrowArray {
 	tr.bufs = make([][]byte, 0, nbuffers)
 	cbufs := allocateBufferMallocatorPtrArr(alloc, nbuffers)
 	for i, b := range buffers[bufOffset:] {
-		if b != nil {
+		if b != nil && b.Len() > 0 {
 			raw := alloc.Allocate(b.Len())
 			copy(raw, b.Bytes())
 			tr.bufs = append(tr.bufs, raw)