Skip to content

fix: improve handling variant nulls and nested types #18538

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Mytherin merged 1 commit into from
Aug 7, 2025
+51 −1
Merged

fix: improve handling variant nulls and nested types #18538

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Binary file added data/parquet-testing/variant_unshredded_nested_nulls.parquet
View file
Open in desktop
Binary file not shown.
8 changes: 7 additions & 1 deletion extension/parquet/reader/variant/variant_shredded_conversion.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -320,13 +320,19 @@ static vector<VariantValue> ConvertBinaryEncoding(Vector &metadata, Vector &valu
UnifiedVectorFormat metadata_format;
metadata.ToUnifiedFormat(length, metadata_format);
auto metadata_data = metadata_format.GetData<string_t>(metadata_format);
auto metadata_validity = metadata_format.validity;

vector<VariantValue> ret(length);
if (IS_REQUIRED) {
D_ASSERT(validity.CountValid(length) == length);
for (idx_t i = 0; i < length; i++) {
auto index = value_format.sel->get_index(i + offset);

// Variant itself is NULL
if (!validity.RowIsValid(index) && !metadata_validity.RowIsValid(metadata_format.sel->get_index(i))) {
ret[i] = VariantValue(Value());
continue;
}

D_ASSERT(validity.RowIsValid(index));
auto &metadata_value = metadata_data[metadata_format.sel->get_index(i)];
VariantMetadata variant_metadata(metadata_value);
Expand Down
44 changes: 44 additions & 0 deletions test/parquet/variant/variant_nested_with_nulls.test
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,44 @@
# name: test/parquet/variant/variant_nested_with_nulls.test
# group: [variant]

require parquet

query IIIIII
describe from parquet_scan('data/parquet-testing/variant_unshredded_nested_nulls.parquet')
----
id BIGINT YES NULL NULL NULL
v STRUCT("value" BLOB, metadata BLOB) YES NULL NULL NULL
array_of_variants STRUCT("value" BLOB, metadata BLOB)[] YES NULL NULL NULL
struct_of_variants STRUCT(v STRUCT("value" BLOB, metadata BLOB)) YES NULL NULL NULL
map_of_variants MAP(VARCHAR, STRUCT("value" BLOB, metadata BLOB)) YES NULL NULL NULL
array_of_struct_of_variants STRUCT(v STRUCT("value" BLOB, metadata BLOB))[] YES NULL NULL NULL
struct_of_array_of_variants STRUCT(v STRUCT("value" BLOB, metadata BLOB)[]) YES NULL NULL NULL

statement ok
set variant_legacy_encoding=true;

# Now the variant column gets emitted as JSON
query IIIIII
describe from parquet_scan('data/parquet-testing/variant_unshredded_nested_nulls.parquet')
----
id BIGINT YES NULL NULL NULL
v JSON YES NULL NULL NULL
array_of_variants JSON[] YES NULL NULL NULL
struct_of_variants STRUCT(v JSON) YES NULL NULL NULL
map_of_variants MAP(VARCHAR, JSON) YES NULL NULL NULL
array_of_struct_of_variants STRUCT(v JSON)[] YES NULL NULL NULL
struct_of_array_of_variants STRUCT(v JSON[]) YES NULL NULL NULL

query IIIIIII
select * from parquet_scan('data/parquet-testing/variant_unshredded_nested_nulls.parquet') order by id limit 10;
----
0 {"key":0} ['{"key":0}', NULL, '{"key":0}', NULL, '{"key":0}'] {'v': '{"key":0}'} {0='{"key":0}', nullKey=NULL} [{'v': '{"key":0}'}, {'v': NULL}, NULL, {'v': '{"key":0}'}, NULL, {'v': '{"key":0}'}] {'v': [NULL, '{"key":0}']}
0 {"key":0} ['{"key":0}', NULL, '{"key":0}', NULL, '{"key":0}'] {'v': '{"key":0}'} {0='{"key":0}', nullKey=NULL} [{'v': '{"key":0}'}, {'v': NULL}, NULL, {'v': '{"key":0}'}, NULL, {'v': '{"key":0}'}] {'v': [NULL, '{"key":0}']}
1 {"key":1} ['{"key":1}', NULL, '{"key":1}', NULL, '{"key":1}'] {'v': '{"key":1}'} {1='{"key":1}', nullKey=NULL} [{'v': '{"key":1}'}, {'v': NULL}, NULL, {'v': '{"key":1}'}, NULL, {'v': '{"key":1}'}] {'v': [NULL, '{"key":1}']}
1 {"key":1} ['{"key":1}', NULL, '{"key":1}', NULL, '{"key":1}'] {'v': '{"key":1}'} {1='{"key":1}', nullKey=NULL} [{'v': '{"key":1}'}, {'v': NULL}, NULL, {'v': '{"key":1}'}, NULL, {'v': '{"key":1}'}] {'v': [NULL, '{"key":1}']}
2 {"key":2} ['{"key":2}', NULL, '{"key":2}', NULL, '{"key":2}'] {'v': '{"key":2}'} {2='{"key":2}', nullKey=NULL} [{'v': '{"key":2}'}, {'v': NULL}, NULL, {'v': '{"key":2}'}, NULL, {'v': '{"key":2}'}] {'v': [NULL, '{"key":2}']}
3 {"key":3} ['{"key":3}', NULL, '{"key":3}', NULL, '{"key":3}'] {'v': '{"key":3}'} {3='{"key":3}', nullKey=NULL} [{'v': '{"key":3}'}, {'v': NULL}, NULL, {'v': '{"key":3}'}, NULL, {'v': '{"key":3}'}] {'v': [NULL, '{"key":3}']}
4 {"key":4} ['{"key":4}', NULL, '{"key":4}', NULL, '{"key":4}'] {'v': '{"key":4}'} {4='{"key":4}', nullKey=NULL} [{'v': '{"key":4}'}, {'v': NULL}, NULL, {'v': '{"key":4}'}, NULL, {'v': '{"key":4}'}] {'v': [NULL, '{"key":4}']}
5 {"key":5} ['{"key":5}', NULL, '{"key":5}', NULL, '{"key":5}'] {'v': '{"key":5}'} {5='{"key":5}', nullKey=NULL} [{'v': '{"key":5}'}, {'v': NULL}, NULL, {'v': '{"key":5}'}, NULL, {'v': '{"key":5}'}] {'v': [NULL, '{"key":5}']}
6 {"key":6} ['{"key":6}', NULL, '{"key":6}', NULL, '{"key":6}'] {'v': '{"key":6}'} {6='{"key":6}', nullKey=NULL} [{'v': '{"key":6}'}, {'v': NULL}, NULL, {'v': '{"key":6}'}, NULL, {'v': '{"key":6}'}] {'v': [NULL, '{"key":6}']}
7 {"key":7} ['{"key":7}', NULL, '{"key":7}', NULL, '{"key":7}'] {'v': '{"key":7}'} {7='{"key":7}', nullKey=NULL} [{'v': '{"key":7}'}, {'v': NULL}, NULL, {'v': '{"key":7}'}, NULL, {'v': '{"key":7}'}] {'v': [NULL, '{"key":7}']}
Loading