You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@arrow.apache.org by GitBox <gi...@apache.org> on 2021/05/24 14:55:58 UTC
[GitHub] [arrow-datafusion] Jimexist opened a new pull request #413: finish implementation of window accumulation and func `row_number()` and add `sql.rs` test
Jimexist opened a new pull request #413:
URL: https://github.com/apache/arrow-datafusion/pull/413
# Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes #123` indicates that this PR will close issue #123.
-->
Closes #298 .
# Rationale for this change
with #381 we can finally close up this. based on #375
# What changes are included in this PR?
<!--
There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR.
-->
# Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be updated before approving the PR.
-->
<!--
If there are any breaking changes to public APIs, please add the `api change` label.
-->
--
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] Jimexist commented on pull request #413: finish implementation of window accumulation and func `row_number()` and add `sql.rs` test
Posted by GitBox <gi...@apache.org>.
Jimexist commented on pull request #413:
URL: https://github.com/apache/arrow-datafusion/pull/413#issuecomment-847538012
converting back to draft because most changes are already incorporated in #375
--
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] codecov-commenter commented on pull request #413: finish implementation of window accumulation and func `row_number()` and add `sql.rs` test
Posted by GitBox <gi...@apache.org>.
codecov-commenter commented on pull request #413:
URL: https://github.com/apache/arrow-datafusion/pull/413#issuecomment-847113389
# [Codecov](https://codecov.io/gh/apache/arrow-datafusion/pull/413?src=pr&el=h1&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation) Report
> Merging [#413](https://codecov.io/gh/apache/arrow-datafusion/pull/413?src=pr&el=desc&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation) (9ba25a0) into [master](https://codecov.io/gh/apache/arrow-datafusion/commit/7359e4b4df0836b8de970d551c6eeae22d1cc810?el=desc&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation) (7359e4b) will **increase** coverage by `0.57%`.
> The diff coverage is `81.81%`.
[](https://codecov.io/gh/apache/arrow-datafusion/pull/413?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation)
```diff
@@ Coverage Diff @@
## master #413 +/- ##
==========================================
+ Coverage 74.85% 75.43% +0.57%
==========================================
Files 146 148 +2
Lines 24498 24744 +246
==========================================
+ Hits 18339 18666 +327
+ Misses 6159 6078 -81
```
| [Impacted Files](https://codecov.io/gh/apache/arrow-datafusion/pull/413?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation) | Coverage Δ | |
|---|---|---|
| [datafusion/src/physical\_plan/expressions/mod.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvcGh5c2ljYWxfcGxhbi9leHByZXNzaW9ucy9tb2QucnM=) | `71.42% <ø> (ø)` | |
| [datafusion/src/physical\_plan/hash\_aggregate.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvcGh5c2ljYWxfcGxhbi9oYXNoX2FnZ3JlZ2F0ZS5ycw==) | `85.21% <ø> (ø)` | |
| [datafusion/src/physical\_plan/sort.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvcGh5c2ljYWxfcGxhbi9zb3J0LnJz) | `92.07% <ø> (ø)` | |
| [datafusion/src/physical\_plan/mod.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvcGh5c2ljYWxfcGxhbi9tb2QucnM=) | `80.58% <68.75%> (-2.18%)` | :arrow_down: |
| [...fusion/src/physical\_plan/expressions/row\_number.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvcGh5c2ljYWxfcGxhbi9leHByZXNzaW9ucy9yb3dfbnVtYmVyLnJz) | `69.23% <69.23%> (ø)` | |
| [datafusion/src/physical\_plan/window\_functions.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvcGh5c2ljYWxfcGxhbi93aW5kb3dfZnVuY3Rpb25zLnJz) | `87.68% <70.58%> (-1.05%)` | :arrow_down: |
| [...afusion/src/physical\_plan/expressions/nth\_value.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvcGh5c2ljYWxfcGxhbi9leHByZXNzaW9ucy9udGhfdmFsdWUucnM=) | `70.76% <70.76%> (ø)` | |
| [datafusion/src/physical\_plan/windows.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvcGh5c2ljYWxfcGxhbi93aW5kb3dzLnJz) | `84.00% <90.98%> (+84.00%)` | :arrow_up: |
| [datafusion/src/execution/context.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvZXhlY3V0aW9uL2NvbnRleHQucnM=) | `92.07% <100.00%> (+0.02%)` | :arrow_up: |
| [datafusion/src/physical\_plan/planner.rs](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation#diff-ZGF0YWZ1c2lvbi9zcmMvcGh5c2ljYWxfcGxhbi9wbGFubmVyLnJz) | `80.45% <100.00%> (+3.94%)` | :arrow_up: |
| ... and [14 more](https://codecov.io/gh/apache/arrow-datafusion/pull/413/diff?src=pr&el=tree-more&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation) | |
------
[Continue to review full report at Codecov](https://codecov.io/gh/apache/arrow-datafusion/pull/413?src=pr&el=continue&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation).
> **Legend** - [Click here to learn more](https://docs.codecov.io/docs/codecov-delta?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation)
> `Δ = absolute <relative> (impact)`, `ø = not affected`, `? = missing data`
> Powered by [Codecov](https://codecov.io/gh/apache/arrow-datafusion/pull/413?src=pr&el=footer&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation). Last update [7359e4b...9ba25a0](https://codecov.io/gh/apache/arrow-datafusion/pull/413?src=pr&el=lastupdated&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation). Read the [comment docs](https://docs.codecov.io/docs/pull-request-comments?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=The+Apache+Software+Foundation).
--
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] Jimexist closed pull request #413: finish implementation of window accumulation and func `row_number()` and add `sql.rs` test
Posted by GitBox <gi...@apache.org>.
Jimexist closed pull request #413:
URL: https://github.com/apache/arrow-datafusion/pull/413
--
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] Dandandan commented on a change in pull request #413: finish implementation of window accumulation and func `row_number()` and add `sql.rs` test
Posted by GitBox <gi...@apache.org>.
Dandandan commented on a change in pull request #413:
URL: https://github.com/apache/arrow-datafusion/pull/413#discussion_r639013631
##########
File path: datafusion/src/physical_plan/windows.rs
##########
@@ -186,10 +301,252 @@ impl ExecutionPlan for WindowAggExec {
));
}
- // let input = self.input.execute(0).await?;
+ let input = self.input.execute(partition).await?;
+
+ let stream = Box::pin(WindowAggStream::new(
+ self.schema.clone(),
+ self.window_expr.clone(),
+ input,
+ ));
+ Ok(stream)
+ }
+}
+
+pin_project! {
+ /// stream for window aggregation plan
+ pub struct WindowAggStream {
+ schema: SchemaRef,
+ #[pin]
+ output: futures::channel::oneshot::Receiver<ArrowResult<RecordBatch>>,
+ finished: bool,
+ }
+}
+
+type WindowAccumulatorItem = Box<dyn WindowAccumulator>;
+
+fn window_expressions(
+ window_expr: &[Arc<dyn WindowExpr>],
+) -> Result<Vec<Vec<Arc<dyn PhysicalExpr>>>> {
+ Ok(window_expr
+ .iter()
+ .map(|expr| expr.expressions())
+ .collect::<Vec<_>>())
+}
+
+fn window_aggregate_batch(
+ batch: &RecordBatch,
+ window_accumulators: &mut [WindowAccumulatorItem],
+ expressions: &[Vec<Arc<dyn PhysicalExpr>>],
+) -> Result<Vec<Option<Vec<ScalarValue>>>> {
+ // 1.1 iterate accumulators and respective expressions together
+ // 1.2 evaluate expressions
+ // 1.3 update / merge window accumulators with the expressions' values
+
+ // 1.1
+ window_accumulators
+ .iter_mut()
+ .zip(expressions)
+ .map(|(window_acc, expr)| {
+ // 1.2
+ let values = &expr
+ .iter()
+ .map(|e| e.evaluate(batch))
+ .map(|r| r.map(|v| v.into_array(batch.num_rows())))
+ .collect::<Result<Vec<_>>>()?;
+
+ window_acc.scan_batch(batch.num_rows(), values)
+ })
+ .into_iter()
+ .collect::<Result<Vec<_>>>()
+}
+
+/// returns a vector of ArrayRefs, where each entry corresponds to either the
+/// final value (mode = Final) or states (mode = Partial)
+fn finalize_window_aggregation(
+ window_accumulators: &[WindowAccumulatorItem],
+) -> Result<Vec<Option<ScalarValue>>> {
+ window_accumulators
+ .iter()
+ .map(|window_accumulator| window_accumulator.evaluate())
+ .collect::<Result<Vec<_>>>()
+}
+
+fn create_window_accumulators(
+ window_expr: &[Arc<dyn WindowExpr>],
+) -> Result<Vec<WindowAccumulatorItem>> {
+ window_expr
+ .iter()
+ .map(|expr| expr.create_accumulator())
+ .collect::<Result<Vec<_>>>()
+}
+
+async fn compute_window_aggregate(
+ schema: SchemaRef,
+ window_expr: Vec<Arc<dyn WindowExpr>>,
+ mut input: SendableRecordBatchStream,
+) -> ArrowResult<RecordBatch> {
+ let mut window_accumulators = create_window_accumulators(&window_expr)
+ .map_err(DataFusionError::into_arrow_external_error)?;
+
+ let expressions = window_expressions(&window_expr)
+ .map_err(DataFusionError::into_arrow_external_error)?;
+
+ let expressions = Arc::new(expressions);
+
+ // TODO each element shall have some size hint
+ let mut accumulator: Vec<Vec<ScalarValue>> =
+ iter::repeat(vec![]).take(window_expr.len()).collect();
+
+ let mut original_batches: Vec<RecordBatch> = vec![];
+
+ let mut total_num_rows = 0;
+
+ while let Some(batch) = input.next().await {
+ let batch = batch?;
+ total_num_rows += batch.num_rows();
+ original_batches.push(batch.clone());
+
+ let batch_aggregated =
+ window_aggregate_batch(&batch, &mut window_accumulators, &expressions)
+ .map_err(DataFusionError::into_arrow_external_error)?;
+ accumulator.iter_mut().zip(batch_aggregated).for_each(
+ |(acc_for_window, window_batch)| {
+ if let Some(data) = window_batch {
+ acc_for_window.extend(data);
+ }
+ },
+ );
+ }
+
+ let aggregated_mapped = finalize_window_aggregation(&window_accumulators)
+ .map_err(DataFusionError::into_arrow_external_error)?;
+
+ let mut columns: Vec<ArrayRef> = accumulator
+ .iter()
+ .zip(aggregated_mapped)
+ .map(|(acc, agg)| match (acc, agg) {
+ // either accumulator values or the aggregated values are non-empty, but not both
+ (acc, Some(scalar_value)) if acc.is_empty() => {
+ Ok(scalar_value.to_array_of_size(total_num_rows))
+ }
+ (acc, None) if !acc.is_empty() => ScalarValue::iter_to_array(acc),
+ _ => Err(DataFusionError::Execution(
+ "Invalid window function behavior".to_owned(),
+ )),
+ })
+ .collect::<Result<Vec<ArrayRef>>>()
+ .map_err(DataFusionError::into_arrow_external_error)?;
+
+ for i in 0..(schema.fields().len() - window_expr.len()) {
+ let col = concat(
Review comment:
I guess this is only needed for the empty `over` clause right, as in other cases we probably only want to access the rows.
Also for functions like RowNumber we could be streaming/emit batches instead of "caching" them?
--
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org