feat: (GH-56) Add arrayApproxEqualString to handle null characters in strings

Signed-off-by: Saurabh Kumar Singh <[email protected]>

Author: singh1203

arrow/array/compare.go

@@ -19,6 +19,7 @@ package array
 import (
 	"fmt"
 	"math"
+	"strings"
 
 	"github.com/apache/arrow-go/v18/arrow"
 	"github.com/apache/arrow-go/v18/arrow/float16"
@@ -487,13 +488,13 @@ func arrayApproxEqual(left, right arrow.Array, opt equalOption) bool {
 		return arrayEqualBinary(l, r)
 	case *String:
 		r := right.(*String)
-		return arrayEqualString(l, r)
+		return arrayApproxEqualString(l, r)
 	case *LargeBinary:
 		r := right.(*LargeBinary)
 		return arrayEqualLargeBinary(l, r)
 	case *LargeString:
 		r := right.(*LargeString)
-		return arrayEqualLargeString(l, r)
+		return arrayApproxEqualLargeString(l, r)
 	case *BinaryView:
 		r := right.(*BinaryView)
 		return arrayEqualBinaryView(l, r)
@@ -644,6 +645,34 @@ func validityBitmapEqual(left, right arrow.Array) bool {
 	return true
 }
 
+func arrayApproxEqualString(left, right *String) bool {
+	for i := 0; i < left.Len(); i++ {
+		if left.IsNull(i) {
+			continue
+		}
+		if stripNulls(left.Value(i)) != stripNulls(right.Value(i)) {
+			return false
+		}
+	}
+	return true
+}
+
+func arrayApproxEqualLargeString(left, right *LargeString) bool {
+	for i := 0; i < left.Len(); i++ {
+		if left.IsNull(i) {
+			continue
+		}
+		if stripNulls(left.Value(i)) != stripNulls(right.Value(i)) {
+			return false
+		}
+	}
+	return true
+}
+
+func stripNulls(s string) string {
+	return strings.TrimRight(s, "\x00")
+}
+
 func arrayApproxEqualFloat16(left, right *Float16, opt equalOption) bool {
 	for i := 0; i < left.Len(); i++ {
 		if left.IsNull(i) {

arrow/array/compare_test.go

@@ -111,6 +111,48 @@ func TestArrayApproxEqual(t *testing.T) {
 	}
 }
 
+func TestArrayApproxEqualStrings(t *testing.T) {
+	for _, tc := range []struct {
+		name string
+		a1   interface{}
+		a2   interface{}
+		want bool
+	}{
+		{
+			name: "string",
+			a1:   []string{"a", "b", "c", "d", "e", "f"},
+			a2:   []string{"a", "b", "c", "d", "e", "f"},
+			want: true,
+		},
+		{
+			name: "string",
+			a1: []string{"a", "b\x00"},
+			a2: []string{"a", "b"},
+			want: true,
+		},
+		{
+			name: "string",
+			a1: []string{"a", "b\x00"},
+			a2: []string{"a\x00", "b"},
+			want: true,
+		},
+	}{
+		t.Run(tc.name, func(t *testing.T) {
+			mem := memory.NewCheckedAllocator(memory.NewGoAllocator())
+			defer mem.AssertSize(t, 0)
+
+			a1 := arrayOf(mem, tc.a1, nil)
+			defer a1.Release()
+			a2 := arrayOf(mem, tc.a2, nil)
+			defer a2.Release()
+
+			if got, want := array.ApproxEqual(a1, a2), tc.want; got != want {
+				t.Fatalf("invalid comparison: got=%v, want=%v\na1: %v\na2: %v\n", got, want, a1, a2)
+			}
+		})
+	}
+}
+
 func TestArrayApproxEqualFloats(t *testing.T) {
 	f16sFrom := func(vs []float64) []float16.Num {
 		o := make([]float16.Num, len(vs))
@@ -445,6 +487,13 @@ func arrayOf(mem memory.Allocator, a interface{}, valids []bool) arrow.Array {
 		bldr.AppendValues(a, valids)
 		return bldr.NewFloat64Array()
 
+	case []string:
+		bldr := array.NewStringBuilder(mem)
+		defer bldr.Release()
+
+		bldr.AppendValues(a, valids)
+		return bldr.NewStringArray()
+
 	default:
 		panic(fmt.Errorf("arrdata: invalid data slice type %T", a))
 	}
@@ -725,4 +774,4 @@ func TestTableEqual(t *testing.T) {
 			}
 		})
 	}
-}
+}