fix: skip exhausted blocks before creating an interval (#8180)

* fix: skip exhausted blocks before creating an interval

* refactor: optimize interval creation and fix logic duplication

* docs: add docstring for LocateData

* refactor: extract moveToNextBlock helper to deduplicate logic

* fix: use int64 for block index comparison to prevent overflow

* test: add unit test for LocateData boundary crossing (issue #8179)

* fix: skip exhausted blocks to prevent negative interval size and panics (issue #8179)

* refactor: apply review suggestions for test maintainability and code style

weed/storage/erasure_coding/ec_test.go

@@ -236,3 +236,20 @@ func TestLocateData3(t *testing.T) {
 		{BlockIndex: 8876, InnerBlockOffset: 912752, Size: 112568, IsLargeBlock: false, LargeBlockRowsCount: 2},
 	})
 }
+
+func TestLocateData_Issue8179(t *testing.T) {
+	large := int64(10000)
+	small := int64(100)
+	shardSize := int64(259092) // Resulting in nLargeBlockRows = 25 as seen in panic log
+
+	// Testing range through the large-to-small transition boundary
+	nLargeBlockRows := (shardSize - 1) / large
+	largeAreaSize := nLargeBlockRows * int64(DataShardsCount) * large
+
+	for offset := largeAreaSize - 500; offset < largeAreaSize+500; offset++ {
+		intervals := LocateData(large, small, shardSize, offset, 200)
+		for _, interval := range intervals {
+			assert.True(t, interval.Size > 0, "Interval size must be positive at offset %d, got %+v", offset, interval)
+		}
+	}
+}