fix(sort): avoid int64 overflow comparing large integers (#2771)

The int/int comparator returned int(lhs - rhs); the int64 subtraction overflows for large-magnitude values, giving the wrong sign. Compare directly instead.

Co-authored-by: max <max@example.com>

pkg/yqlib/operator_sort.go

@@ -166,7 +166,12 @@ func (a sortableNodeArray) compare(lhs *CandidateNode, rhs *CandidateNode, dateT
 		if err != nil {
 			panic(err)
 		}
-		return int(lhsNum - rhsNum)
+		if lhsNum < rhsNum {
+			return -1
+		} else if lhsNum > rhsNum {
+			return 1
+		}
+		return 0
 	} else if (lhsTag == "!!int" || lhsTag == "!!float") && (rhsTag == "!!int" || rhsTag == "!!float") {
 		lhsNum, err := strconv.ParseFloat(lhs.Value, 64)
 		if err != nil {

pkg/yqlib/operator_sort_test.go

@@ -171,6 +171,15 @@ var sortByOperatorScenarios = []expressionScenario{
 			"D0, P[], (!!seq)::# abc\n- def\n# ghi\n",
 		},
 	},
+	{
+		skipDoc:     true,
+		description: "Sort large integers (no int64 subtraction overflow)",
+		document:    "[5000000000000000000, -5000000000000000000]",
+		expression:  `sort`,
+		expected: []string{
+			"D0, P[], (!!seq)::[-5000000000000000000, 5000000000000000000]\n",
+		},
+	},
 }
 
 func TestSortByOperatorScenarios(t *testing.T) {