This is an automated email from the ASF dual-hosted git repository.
zeroshade pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/arrow-go.git
The following commit(s) were added to refs/heads/main by this push:
new 7e537909 fix(array): support sparse union sliced type ids (#881)
7e537909 is described below
commit 7e537909f4d6f6160e075fe29cd8bd04283113ad
Author: Minh Vu <[email protected]>
AuthorDate: Thu Jul 2 19:01:58 2026 +0200
fix(array): support sparse union sliced type ids (#881)
Normalize the type-id value buffer to the input array's logical window
before constructing sparse union data, so a sparse union can be built from
sliced type ids without panicking when children are full logical length. The
union is built at offset 0. Parity follow-up to #877.
---
arrow/array/union.go | 12 ++++++++++--
arrow/array/union_test.go | 18 ++++++++++++++++++
2 files changed, 28 insertions(+), 2 deletions(-)
diff --git a/arrow/array/union.go b/arrow/array/union.go
index f3159242..58524a76 100644
--- a/arrow/array/union.go
+++ b/arrow/array/union.go
@@ -289,7 +289,15 @@ func NewSparseUnionFromArraysWithFieldCodes(typeIDs
arrow.Array, children []arro
return nil, errors.New("arrow/array: type codes must have same
length as children")
}
- buffers := []*memory.Buffer{nil, typeIDs.Data().Buffers()[1]}
+ typeIDBuffer := typeIDs.Data().Buffers()[1]
+ if typeIDBuffer != nil {
+ typeIDBuffer = memory.SliceBuffer(typeIDBuffer,
+ arrow.Int8Traits.BytesRequired(typeIDs.Data().Offset()),
+ arrow.Int8Traits.BytesRequired(typeIDs.Len()))
+ defer typeIDBuffer.Release()
+ }
+
+ buffers := []*memory.Buffer{nil, typeIDBuffer}
ty := arrow.SparseUnionFromArrays(children, fields, codes)
childData := make([]arrow.ArrayData, len(children))
@@ -300,7 +308,7 @@ func NewSparseUnionFromArraysWithFieldCodes(typeIDs
arrow.Array, children []arro
}
}
- data := NewData(ty, typeIDs.Len(), buffers, childData, 0,
typeIDs.Data().Offset())
+ data := NewData(ty, typeIDs.Len(), buffers, childData, 0, 0)
defer data.Release()
return NewSparseUnionData(data), nil
}
diff --git a/arrow/array/union_test.go b/arrow/array/union_test.go
index 7421b1bc..e20f0247 100644
--- a/arrow/array/union_test.go
+++ b/arrow/array/union_test.go
@@ -622,6 +622,24 @@ func (s *UnionFactorySuite) TestMakeSparse() {
s.Error(result.ValidateFull())
})
+ s.Run("sliced type ids", func() {
+ expected, err := array.NewSparseUnionFromArrays(s.typeIDs,
children)
+ s.NoError(err)
+ defer expected.Release()
+
+ baseTypeIDs := s.typeidsFromSlice(3, 0, 1, 2, 0, 1, 3, 2, 0, 2,
1)
+ defer baseTypeIDs.Release()
+ slicedTypeIDs := array.NewSlice(baseTypeIDs, 1,
int64(baseTypeIDs.Len()))
+ defer slicedTypeIDs.Release()
+
+ result, err := array.NewSparseUnionFromArrays(slicedTypeIDs,
children)
+ s.NoError(err)
+ defer result.Release()
+ s.Zero(result.Data().Offset())
+ s.NoError(result.ValidateFull())
+ s.True(array.Equal(expected, result))
+ })
+
s.Run("invalid child length", func() {
children[3], _, _ = array.FromJSON(s.mem,
arrow.PrimitiveTypes.Int8,
strings.NewReader(`[0, 0, 0, 0, 0, -12, 0, 0, 0]`))
