timsaucer opened a new issue, #7315:
URL: https://github.com/apache/arrow-rs/issues/7315
**Describe the bug**
If you have a List of FixedList and the top level list has an offset that is
not zero, the encoded array contains the wrong values in the fixed list. See
the example below to illustrate the problem.
**To Reproduce**
Add the following test within `arrow-ipc/src/writer.rs` test module:
```rust
#[test]
fn test_roundtrip_list_of_fixed_list() -> Result<(), ArrowError> {
let l0_builder = Float32Builder::new();
let l1_builder = FixedSizeListBuilder::new(l0_builder, 3);
let mut l2_builder = ListBuilder::new(l1_builder);
for point in [[1.0, 2.0, 3.0], [4.0, 5.0, 6.0], [7.0, 8.0, 9.0]] {
l2_builder.values().values().append_value(point[0]);
l2_builder.values().values().append_value(point[1]);
l2_builder.values().values().append_value(point[2]);
l2_builder.values().append(true);
}
l2_builder.append(true);
for point in [[10., 11., 12.]] {
l2_builder.values().values().append_value(point[0]);
l2_builder.values().values().append_value(point[1]);
l2_builder.values().values().append_value(point[2]);
l2_builder.values().append(true);
}
l2_builder.append(true);
let array = Arc::new(l2_builder.finish()) as ArrayRef;
let schema = Arc::new(Schema::new_with_metadata(
vec![Field::new(
"points",
DataType::List(Arc::new(Field::new(
"item",
DataType::FixedSizeList(
Arc::new(Field::new("item", DataType::Float32,
true)),
3,
),
true,
))),
true,
)],
HashMap::default(),
));
let subarray_1 = array.slice(0, 1);
let subarray_2 = array.slice(1, 1);
let b1 = RecordBatch::try_new(schema.clone(), vec![subarray_1])?;
let b2 = RecordBatch::try_new(schema.clone(), vec![subarray_2])?;
let mut bytes = Vec::new();
let mut writer = StreamWriter::try_new(&mut bytes, &schema)?;
writer.write(&b1)?;
writer.finish()?;
let mut cursor = std::io::Cursor::new(bytes);
let mut reader = StreamReader::try_new(&mut cursor, None)?;
let b1_return = reader.next().unwrap()?;
assert_eq!(b1, b1_return);
let mut bytes = Vec::new();
let mut writer = StreamWriter::try_new(&mut bytes, &schema)?;
writer.write(&b2)?;
writer.finish()?;
let mut cursor = std::io::Cursor::new(bytes);
let mut reader = StreamReader::try_new(&mut cursor, None)?;
let b2_return = reader.next().unwrap()?;
assert_eq!(b2, b2_return);
Ok(())
}
```
**Expected behavior**
Returned deserialized arrow array should match serialized.
**Additional context**
This is probably related to #6805
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
