Skip to content

skip empty batch while inserting #4

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
merged 1 commit into from
Jan 26, 2022
Merged

skip empty batch while inserting #4

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
69 changes: 41 additions & 28 deletions datafusion/src/physical_plan/sorts/sort_preserving_merge.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -410,40 +410,53 @@ impl SortPreservingMergeStream {
// Cursor is not finished - don't need a new RecordBatch yet
return Poll::Ready(Ok(()));
}
let mut streams = self.streams.streams.lock().unwrap();
let mut empty_batch = false;
{
let mut streams = self.streams.streams.lock().unwrap();

let stream = &mut streams[idx];
if stream.is_terminated() {
return Poll::Ready(Ok(()));
}

// Fetch a new input record and create a cursor from it
match futures::ready!(stream.poll_next_unpin(cx)) {
None => return Poll::Ready(Ok(())),
Some(Err(e)) => {
return Poll::Ready(Err(e));
let stream = &mut streams[idx];
if stream.is_terminated() {
return Poll::Ready(Ok(()));
}
Some(Ok(batch)) => {
let cursor = match SortKeyCursor::new(
idx,
self.next_batch_id, // assign this batch an ID
&batch,
&self.column_expressions,
self.sort_options.clone(),
) {
Ok(cursor) => cursor,
Err(e) => {
return Poll::Ready(Err(ArrowError::ExternalError(Box::new(e))));

// Fetch a new input record and create a cursor from it
match futures::ready!(stream.poll_next_unpin(cx)) {
None => return Poll::Ready(Ok(())),
Some(Err(e)) => {
return Poll::Ready(Err(e));
}
Some(Ok(batch)) => {
if batch.num_rows() > 0 {
let cursor = match SortKeyCursor::new(
idx,
self.next_batch_id, // assign this batch an ID
&batch,
&self.column_expressions,
self.sort_options.clone(),
) {
Ok(cursor) => cursor,
Err(e) => {
return Poll::Ready(Err(ArrowError::ExternalError(
Box::new(e),
)));
}
};
self.next_batch_id += 1;
self.min_heap.push(cursor);
self.cursor_finished[idx] = false;
self.batches[idx].push_back(batch)
} else {
empty_batch = true;
}
};
self.next_batch_id += 1;
self.min_heap.push(cursor);
self.cursor_finished[idx] = false;
self.batches[idx].push_back(batch)
}
}
}

Poll::Ready(Ok(()))
if empty_batch {
self.maybe_poll_stream(cx, idx)
} else {
Poll::Ready(Ok(()))
}
Copy link
Author

@yjshen yjshen Jan 26, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The changes above seem scary, but indent thing, an empty_batch flag is introduced, and call this method again to skip empty batches.

Copy link
Owner

@alamb alamb Jan 26, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍 The changes may be clearer by using "whitespace blind" diff: https://github.com/alamb/arrow-datafusion/pull/4/files?w=1

}

/// Drains the in_progress row indexes, and builds a new RecordBatch from them
Expand Down