tustvold commented on code in PR #4280:
URL: https://github.com/apache/arrow-rs/pull/4280#discussion_r1205529669
##########
parquet/src/arrow/arrow_writer/mod.rs:
##########
@@ -284,156 +247,271 @@ impl<W: Write> RecordBatchWriter for ArrowWriter<W> {
}
}
-fn write_leaves<W: Write>(
- row_group_writer: &mut SerializedRowGroupWriter<'_, W>,
- arrays: &[ArrayRef],
- levels: &mut [Vec<LevelInfo>],
-) -> Result<()> {
- assert_eq!(arrays.len(), levels.len());
- assert!(!arrays.is_empty());
-
- let data_type = arrays.first().unwrap().data_type().clone();
- assert!(arrays.iter().all(|a| a.data_type() == &data_type));
-
- match &data_type {
- ArrowDataType::Null
- | ArrowDataType::Boolean
- | ArrowDataType::Int8
- | ArrowDataType::Int16
- | ArrowDataType::Int32
- | ArrowDataType::Int64
- | ArrowDataType::UInt8
- | ArrowDataType::UInt16
- | ArrowDataType::UInt32
- | ArrowDataType::UInt64
- | ArrowDataType::Float32
- | ArrowDataType::Float64
- | ArrowDataType::Timestamp(_, _)
- | ArrowDataType::Date32
- | ArrowDataType::Date64
- | ArrowDataType::Time32(_)
- | ArrowDataType::Time64(_)
- | ArrowDataType::Duration(_)
- | ArrowDataType::Interval(_)
- | ArrowDataType::Decimal128(_, _)
- | ArrowDataType::Decimal256(_, _)
- | ArrowDataType::FixedSizeBinary(_) => {
- let mut col_writer = row_group_writer.next_column()?.unwrap();
- for (array, levels) in arrays.iter().zip(levels.iter_mut()) {
- write_leaf(col_writer.untyped(), array,
levels.pop().expect("Levels exhausted"))?;
+/// A list of [`Bytes`] comprising a single column chunk
+#[derive(Default)]
+struct ArrowColumnChunk {
+ length: usize,
+ data: Vec<Bytes>,
Review Comment:
By buffering separate chunks, we should be performing no more memcpy of data
than we were before :tada:
--
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]
