more tests and comments

Signed-off-by: Robert Kruszewski <github@robertk.io>

Author: robert3005

cpp/src/arrow/util/byte_size.cc

@@ -201,6 +201,12 @@ struct GetByteRangesArray {
     RETURN_NOT_OK(range_offsets->Append(sizeof(c_type) * offset));
     RETURN_NOT_OK(range_lengths->Append(sizeof(c_type) * length));
 
+    // The following calculation is an over/under estimate of the size since views buffer
+    // might
+    // 1. Not reference all the values in data buffers (the array was filtered without gc)
+    // 2. Reference a value multiple times without repeating it in the data buffer
+    //
+    // Producing exact byte size would require linear scan of all values in view buffer
     for (int i = 2; i < input.buffers.size(); i++) {
       const Buffer& buf = *input.buffers[i];
       RETURN_NOT_OK(range_starts->Append(reinterpret_cast<uint64_t>(buf.data())));
@@ -255,6 +261,12 @@ struct GetByteRangesArray {
     RETURN_NOT_OK(range_offsets->Append(sizeof(offset_type) * offset));
     RETURN_NOT_OK(range_lengths->Append(sizeof(offset_type) * length));
 
+    // The following calculation is an over/under estimate of the byte size since views
+    // buffer might
+    // 1. Not reference all the values in data buffers (the array was filtered without gc)
+    // 2. Reference a value multiple times without repeating it in the data buffer
+    //
+    // Producing exact byte size would require linear scan of all values in view buffer
     GetByteRangesArray child{*input.child_data[0],
                              0,
                              (*input.child_data[0]).length,

cpp/src/arrow/util/byte_size_test.cc

@@ -326,6 +326,11 @@ TEST(ByteRanges, StringViewType) {
   std::shared_ptr<Array> str_view_with_nulls = ArrayFromJSON(
       utf8_view(), R"(["short", null, "another long string that requires a buffer"])");
   CheckBufferRanges(str_view_with_nulls, {{0, 0, 1}, {1, 0, 48}, {2, 0, 42}});
+
+  CheckBufferRanges(str_view_arr->Slice(1, 1), {{0, 16, 16}, {1, 0, 38}});
+  CheckBufferRanges(str_view_arr->Slice(0, 1), {{0, 0, 16}, {1, 0, 38}});
+  CheckBufferRanges(str_view_with_nulls->Slice(2, 1),
+                    {{0, 0, 1}, {1, 32, 16}, {2, 0, 42}});
 }
 
 TEST(ByteRanges, BinaryViewType) {
@@ -337,6 +342,10 @@ TEST(ByteRanges, BinaryViewType) {
   std::shared_ptr<Array> bin_view_with_nulls =
       ArrayFromJSON(binary_view(), R"(["AB", null, "CDEFGHIJKLMNOPQRSTUVWXYZ"])");
   CheckBufferRanges(bin_view_with_nulls, {{0, 0, 1}, {1, 0, 48}, {2, 0, 24}});
+
+  CheckBufferRanges(bin_view_arr->Slice(1, 1), {{0, 16, 16}, {1, 0, 22}});
+  CheckBufferRanges(bin_view_with_nulls->Slice(2, 1),
+                    {{0, 0, 1}, {1, 32, 16}, {2, 0, 24}});
 }
 
 using ListViewArrowTypes = ::testing::Types<ListViewType, LargeListViewType>;
@@ -359,6 +368,15 @@ TYPED_TEST(ByteRangesListView, Basic) {
                                            {1, 0, 3 * sizeof(offset_type)},
                                            {2, 0, 3 * sizeof(offset_type)},
                                            {3, 0, 20}});
+  CheckBufferRanges(list_view_arr->Slice(2, 1),
+                    {{0, 2 * sizeof(offset_type), sizeof(offset_type)},
+                     {1, 2 * sizeof(offset_type), sizeof(offset_type)},
+                     {2, 0, 16}});
+  CheckBufferRanges(list_view_with_nulls->Slice(2, 1),
+                    {{0, 0, 1},
+                     {1, 2 * sizeof(offset_type), sizeof(offset_type)},
+                     {2, 2 * sizeof(offset_type), sizeof(offset_type)},
+                     {3, 0, 20}});
 }
 
 TYPED_TEST(ByteRangesListView, NestedListView) {