tustvold commented on code in PR #5056:
URL: https://github.com/apache/arrow-datafusion/pull/5056#discussion_r1092166166
##########
datafusion/core/src/physical_plan/file_format/csv.rs:
##########
@@ -224,20 +240,38 @@ impl FileOpener for CsvOpener {
match config.object_store.get(file_meta.location()).await? {
GetResult::File(file, _) => {
let decoder = file_compression_type.convert_read(file)?;
- Ok(futures::stream::iter(config.open(decoder,
true)).boxed())
+ Ok(futures::stream::iter(config.open(decoder)).boxed())
}
GetResult::Stream(s) => {
- let mut first_chunk = true;
- let s = s.map_err(Into::<DataFusionError>::into);
- let decoder = file_compression_type.convert_stream(s)?;
- Ok(newline_delimited_stream(decoder)
- .map_ok(move |bytes| {
- let reader = config.open(bytes.reader(),
first_chunk);
- first_chunk = false;
- futures::stream::iter(reader)
- })
- .try_flatten()
- .boxed())
+ let mut decoder = config.builder().build_decoder();
+ let s = s.map_err(DataFusionError::from);
+ let mut input =
file_compression_type.convert_stream(s)?.fuse();
+ let mut buffered = Bytes::new();
+
+ let s = futures::stream::poll_fn(move |cx| {
+ loop {
+ if buffered.is_empty() {
+ buffered = match
ready!(input.poll_next_unpin(cx)) {
+ Some(Ok(b)) => b,
+ Some(Err(e)) => {
+ return Poll::Ready(Some(Err(e.into())))
+ }
+ None => break,
+ };
+ }
+ let decoded = match
decoder.decode(buffered.as_ref()) {
+ // Note: the decoder needs to be called with
an empty
+ // array to delimt the final record
Review Comment:
You are quite correct, I'm investigating how this is working...
--
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: github-unsubscr...@arrow.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
