Skip to content

[SPARK-39453][SQL][TESTS][FOLLOWUP] Let RAND in filter is more meaningful. #37033

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.

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

Closed
wants to merge 1 commit into from

Conversation

beliefer
Copy link
Contributor

What changes were proposed in this pull request?

#36830 makes DS V2 supports push down misc non-aggregate functions(non ANSI).
But he Rand in test case looks no meaningful.

Why are the changes needed?

Let Rand in filter is more meaningful.

Does this PR introduce any user-facing change?

'No'.
Just update test case.

How was this patch tested?

Just update test case.

@github-actions github-actions bot added the SQL label Jun 30, 2022
@@ -856,11 +856,11 @@ class JDBCV2Suite extends QueryTest with SharedSparkSession with ExplainSuiteHel
val df11 = sql(
"""
|SELECT * FROM h2.test.employee
|WHERE GREATEST(bonus, 1100) > 1200 AND LEAST(salary, 10000) > 9000 AND RAND(1) < 1
Copy link
Contributor Author

Choose a reason for hiding this comment

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

We already test LEAST at

|WHERE IF(SALARY > 10000, SALARY, LEAST(SALARY, 1000)) > 1200

@beliefer
Copy link
Contributor Author

ping @huaxingao cc @cloud-fan

@beliefer beliefer changed the title [SPARK-39453][SQL][FOLLOWUP] Let Rand in filter is more meaningful. [SPARK-39453][SQL][FOLLOWUP] Let RAND in filter is more meaningful. Jul 1, 2022
@HyukjinKwon HyukjinKwon changed the title [SPARK-39453][SQL][FOLLOWUP] Let RAND in filter is more meaningful. [SPARK-39453][SQL][TESTS][FOLLOWUP] Let RAND in filter is more meaningful. Jul 5, 2022
@beliefer
Copy link
Contributor Author

beliefer commented Jul 5, 2022

ping @cloud-fan

@cloud-fan
Copy link
Contributor

thanks, merging to master!

@cloud-fan cloud-fan closed this in 3c9b296 Jul 5, 2022
@beliefer
Copy link
Contributor Author

beliefer commented Jul 5, 2022

@cloud-fan Thank you !

chenzhx pushed a commit to chenzhx/spark that referenced this pull request Jul 21, 2022
…ingful

### What changes were proposed in this pull request?
apache#36830 makes DS V2 supports push down misc non-aggregate functions(non ANSI).
But he `Rand` in test case looks no meaningful.

### Why are the changes needed?
Let `Rand` in filter is more meaningful.

### Does this PR introduce _any_ user-facing change?
'No'.
Just update test case.

### How was this patch tested?
Just update test case.

Closes apache#37033 from beliefer/SPARK-39453_followup.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
chenzhx added a commit to Kyligence/spark that referenced this pull request Jul 27, 2022
…LIMIT (#505)

* [SPARK-39139][SQL] DS V2 supports push down DS V2 UDF

### What changes were proposed in this pull request?
Currently, Spark DS V2 push-down framework supports push down SQL to data sources.
But the DS V2 push-down framework only support push down the built-in functions to data sources.
Each database have a lot very useful functions which not supported by Spark.
If we can push down these functions into data source, it will reduce disk I/O and network I/O and improve the performance when query databases.

### Why are the changes needed?

1. Spark can leverage the functions supported by databases
2. Improve the query performance.

### Does this PR introduce _any_ user-facing change?
'No'.
New feature.

### How was this patch tested?
New tests.

Closes apache#36593 from beliefer/SPARK-39139.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39453][SQL][TESTS][FOLLOWUP] Let `RAND` in filter is more meaningful

### What changes were proposed in this pull request?
apache#36830 makes DS V2 supports push down misc non-aggregate functions(non ANSI).
But he `Rand` in test case looks no meaningful.

### Why are the changes needed?
Let `Rand` in filter is more meaningful.

### Does this PR introduce _any_ user-facing change?
'No'.
Just update test case.

### How was this patch tested?
Just update test case.

Closes apache#37033 from beliefer/SPARK-39453_followup.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-37527][SQL][FOLLOWUP] Cannot compile COVAR_POP, COVAR_SAMP and CORR in `H2Dialect` if them with `DISTINCT`

### What changes were proposed in this pull request?
apache#35145 compile COVAR_POP, COVAR_SAMP and CORR in H2Dialect.
Because H2 does't support COVAR_POP, COVAR_SAMP and CORR works with DISTINCT.
So apache#35145 introduces a bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

### Why are the changes needed?
Fix bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

### Does this PR introduce _any_ user-facing change?
'Yes'.
Bug will be fix.

### How was this patch tested?
New test cases.

Closes apache#37090 from beliefer/SPARK-37527_followup2.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39627][SQL] DS V2 pushdown should unify the compile API

### What changes were proposed in this pull request?
Currently, `JdbcDialect` have two API `compileAggregate` and `compileExpression`, we can unify them.

### Why are the changes needed?
Improve ease of use.

### Does this PR introduce _any_ user-facing change?
'No'.
The two API `compileAggregate` call `compileExpression` not changed.

### How was this patch tested?
N/A

Closes apache#37047 from beliefer/SPARK-39627.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39384][SQL] Compile built-in linear regression aggregate functions for JDBC dialect

### What changes were proposed in this pull request?
Recently, Spark DS V2 pushdown framework translate a lot of standard linear regression aggregate functions.
Currently, only H2Dialect compile these standard linear regression aggregate functions. This PR compile these standard linear regression aggregate functions for other build-in JDBC dialect.

### Why are the changes needed?
Make build-in JDBC dialect support compile linear regression aggregate push-down.

### Does this PR introduce _any_ user-facing change?
'No'.
New feature.

### How was this patch tested?
New test cases.

Closes apache#37188 from beliefer/SPARK-39384.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Sean Owen <srowen@gmail.com>

* [SPARK-39148][SQL] DS V2 aggregate push down can work with OFFSET or LIMIT

### What changes were proposed in this pull request?

This PR refactors the v2 agg pushdown code. The main change is, now we don't build the `Scan` immediately when pushing agg. We did it so before because we want to know the data schema with agg pushed, then we can add cast when rewriting the query plan after pushdown. But the problem is, we build `Scan` too early and can't push down any more operators, while it's common to see LIMIT/OFFSET after agg.

The idea of the refactor is, we don't need to know the data schema with agg pushed. We just give an expectation (the data type should be the same of Spark agg functions), use it to define the output of `ScanBuilderHolder`, and then rewrite the query plan. Later on, when we build the `Scan` and replace `ScanBuilderHolder` with `DataSourceV2ScanRelation`, we check the actual data schema and add a `Project` to do type cast if necessary.

### Why are the changes needed?

support pushing down LIMIT/OFFSET after agg.

### Does this PR introduce _any_ user-facing change?

no

### How was this patch tested?

updated tests

Closes apache#37195 from cloud-fan/agg.

Lead-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

Co-authored-by: Jiaan Geng <beliefer@163.com>
Co-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
yhcast0 pushed a commit to yhcast0/spark that referenced this pull request Aug 8, 2022
…LIMIT (Kyligence#505)

* [SPARK-39139][SQL] DS V2 supports push down DS V2 UDF

### What changes were proposed in this pull request?
Currently, Spark DS V2 push-down framework supports push down SQL to data sources.
But the DS V2 push-down framework only support push down the built-in functions to data sources.
Each database have a lot very useful functions which not supported by Spark.
If we can push down these functions into data source, it will reduce disk I/O and network I/O and improve the performance when query databases.

### Why are the changes needed?

1. Spark can leverage the functions supported by databases
2. Improve the query performance.

### Does this PR introduce _any_ user-facing change?
'No'.
New feature.

### How was this patch tested?
New tests.

Closes apache#36593 from beliefer/SPARK-39139.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39453][SQL][TESTS][FOLLOWUP] Let `RAND` in filter is more meaningful

### What changes were proposed in this pull request?
apache#36830 makes DS V2 supports push down misc non-aggregate functions(non ANSI).
But he `Rand` in test case looks no meaningful.

### Why are the changes needed?
Let `Rand` in filter is more meaningful.

### Does this PR introduce _any_ user-facing change?
'No'.
Just update test case.

### How was this patch tested?
Just update test case.

Closes apache#37033 from beliefer/SPARK-39453_followup.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-37527][SQL][FOLLOWUP] Cannot compile COVAR_POP, COVAR_SAMP and CORR in `H2Dialect` if them with `DISTINCT`

### What changes were proposed in this pull request?
apache#35145 compile COVAR_POP, COVAR_SAMP and CORR in H2Dialect.
Because H2 does't support COVAR_POP, COVAR_SAMP and CORR works with DISTINCT.
So apache#35145 introduces a bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

### Why are the changes needed?
Fix bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

### Does this PR introduce _any_ user-facing change?
'Yes'.
Bug will be fix.

### How was this patch tested?
New test cases.

Closes apache#37090 from beliefer/SPARK-37527_followup2.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39627][SQL] DS V2 pushdown should unify the compile API

### What changes were proposed in this pull request?
Currently, `JdbcDialect` have two API `compileAggregate` and `compileExpression`, we can unify them.

### Why are the changes needed?
Improve ease of use.

### Does this PR introduce _any_ user-facing change?
'No'.
The two API `compileAggregate` call `compileExpression` not changed.

### How was this patch tested?
N/A

Closes apache#37047 from beliefer/SPARK-39627.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39384][SQL] Compile built-in linear regression aggregate functions for JDBC dialect

### What changes were proposed in this pull request?
Recently, Spark DS V2 pushdown framework translate a lot of standard linear regression aggregate functions.
Currently, only H2Dialect compile these standard linear regression aggregate functions. This PR compile these standard linear regression aggregate functions for other build-in JDBC dialect.

### Why are the changes needed?
Make build-in JDBC dialect support compile linear regression aggregate push-down.

### Does this PR introduce _any_ user-facing change?
'No'.
New feature.

### How was this patch tested?
New test cases.

Closes apache#37188 from beliefer/SPARK-39384.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Sean Owen <srowen@gmail.com>

* [SPARK-39148][SQL] DS V2 aggregate push down can work with OFFSET or LIMIT

### What changes were proposed in this pull request?

This PR refactors the v2 agg pushdown code. The main change is, now we don't build the `Scan` immediately when pushing agg. We did it so before because we want to know the data schema with agg pushed, then we can add cast when rewriting the query plan after pushdown. But the problem is, we build `Scan` too early and can't push down any more operators, while it's common to see LIMIT/OFFSET after agg.

The idea of the refactor is, we don't need to know the data schema with agg pushed. We just give an expectation (the data type should be the same of Spark agg functions), use it to define the output of `ScanBuilderHolder`, and then rewrite the query plan. Later on, when we build the `Scan` and replace `ScanBuilderHolder` with `DataSourceV2ScanRelation`, we check the actual data schema and add a `Project` to do type cast if necessary.

### Why are the changes needed?

support pushing down LIMIT/OFFSET after agg.

### Does this PR introduce _any_ user-facing change?

no

### How was this patch tested?

updated tests

Closes apache#37195 from cloud-fan/agg.

Lead-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

Co-authored-by: Jiaan Geng <beliefer@163.com>
Co-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
yhcast0 pushed a commit to Kyligence/spark that referenced this pull request Aug 8, 2022
…LIMIT (#505)

* [SPARK-39139][SQL] DS V2 supports push down DS V2 UDF

### What changes were proposed in this pull request?
Currently, Spark DS V2 push-down framework supports push down SQL to data sources.
But the DS V2 push-down framework only support push down the built-in functions to data sources.
Each database have a lot very useful functions which not supported by Spark.
If we can push down these functions into data source, it will reduce disk I/O and network I/O and improve the performance when query databases.

### Why are the changes needed?

1. Spark can leverage the functions supported by databases
2. Improve the query performance.

### Does this PR introduce _any_ user-facing change?
'No'.
New feature.

### How was this patch tested?
New tests.

Closes apache#36593 from beliefer/SPARK-39139.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39453][SQL][TESTS][FOLLOWUP] Let `RAND` in filter is more meaningful

### What changes were proposed in this pull request?
apache#36830 makes DS V2 supports push down misc non-aggregate functions(non ANSI).
But he `Rand` in test case looks no meaningful.

### Why are the changes needed?
Let `Rand` in filter is more meaningful.

### Does this PR introduce _any_ user-facing change?
'No'.
Just update test case.

### How was this patch tested?
Just update test case.

Closes apache#37033 from beliefer/SPARK-39453_followup.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-37527][SQL][FOLLOWUP] Cannot compile COVAR_POP, COVAR_SAMP and CORR in `H2Dialect` if them with `DISTINCT`

### What changes were proposed in this pull request?
apache#35145 compile COVAR_POP, COVAR_SAMP and CORR in H2Dialect.
Because H2 does't support COVAR_POP, COVAR_SAMP and CORR works with DISTINCT.
So apache#35145 introduces a bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

### Why are the changes needed?
Fix bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

### Does this PR introduce _any_ user-facing change?
'Yes'.
Bug will be fix.

### How was this patch tested?
New test cases.

Closes apache#37090 from beliefer/SPARK-37527_followup2.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39627][SQL] DS V2 pushdown should unify the compile API

### What changes were proposed in this pull request?
Currently, `JdbcDialect` have two API `compileAggregate` and `compileExpression`, we can unify them.

### Why are the changes needed?
Improve ease of use.

### Does this PR introduce _any_ user-facing change?
'No'.
The two API `compileAggregate` call `compileExpression` not changed.

### How was this patch tested?
N/A

Closes apache#37047 from beliefer/SPARK-39627.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39384][SQL] Compile built-in linear regression aggregate functions for JDBC dialect

### What changes were proposed in this pull request?
Recently, Spark DS V2 pushdown framework translate a lot of standard linear regression aggregate functions.
Currently, only H2Dialect compile these standard linear regression aggregate functions. This PR compile these standard linear regression aggregate functions for other build-in JDBC dialect.

### Why are the changes needed?
Make build-in JDBC dialect support compile linear regression aggregate push-down.

### Does this PR introduce _any_ user-facing change?
'No'.
New feature.

### How was this patch tested?
New test cases.

Closes apache#37188 from beliefer/SPARK-39384.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Sean Owen <srowen@gmail.com>

* [SPARK-39148][SQL] DS V2 aggregate push down can work with OFFSET or LIMIT

### What changes were proposed in this pull request?

This PR refactors the v2 agg pushdown code. The main change is, now we don't build the `Scan` immediately when pushing agg. We did it so before because we want to know the data schema with agg pushed, then we can add cast when rewriting the query plan after pushdown. But the problem is, we build `Scan` too early and can't push down any more operators, while it's common to see LIMIT/OFFSET after agg.

The idea of the refactor is, we don't need to know the data schema with agg pushed. We just give an expectation (the data type should be the same of Spark agg functions), use it to define the output of `ScanBuilderHolder`, and then rewrite the query plan. Later on, when we build the `Scan` and replace `ScanBuilderHolder` with `DataSourceV2ScanRelation`, we check the actual data schema and add a `Project` to do type cast if necessary.

### Why are the changes needed?

support pushing down LIMIT/OFFSET after agg.

### Does this PR introduce _any_ user-facing change?

no

### How was this patch tested?

updated tests

Closes apache#37195 from cloud-fan/agg.

Lead-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

Co-authored-by: Jiaan Geng <beliefer@163.com>
Co-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
zheniantoushipashi pushed a commit to Kyligence/spark that referenced this pull request Aug 8, 2022
…LIMIT (#505)

* [SPARK-39139][SQL] DS V2 supports push down DS V2 UDF

### What changes were proposed in this pull request?
Currently, Spark DS V2 push-down framework supports push down SQL to data sources.
But the DS V2 push-down framework only support push down the built-in functions to data sources.
Each database have a lot very useful functions which not supported by Spark.
If we can push down these functions into data source, it will reduce disk I/O and network I/O and improve the performance when query databases.

### Why are the changes needed?

1. Spark can leverage the functions supported by databases
2. Improve the query performance.

### Does this PR introduce _any_ user-facing change?
'No'.
New feature.

### How was this patch tested?
New tests.

Closes apache#36593 from beliefer/SPARK-39139.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39453][SQL][TESTS][FOLLOWUP] Let `RAND` in filter is more meaningful

### What changes were proposed in this pull request?
apache#36830 makes DS V2 supports push down misc non-aggregate functions(non ANSI).
But he `Rand` in test case looks no meaningful.

### Why are the changes needed?
Let `Rand` in filter is more meaningful.

### Does this PR introduce _any_ user-facing change?
'No'.
Just update test case.

### How was this patch tested?
Just update test case.

Closes apache#37033 from beliefer/SPARK-39453_followup.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-37527][SQL][FOLLOWUP] Cannot compile COVAR_POP, COVAR_SAMP and CORR in `H2Dialect` if them with `DISTINCT`

### What changes were proposed in this pull request?
apache#35145 compile COVAR_POP, COVAR_SAMP and CORR in H2Dialect.
Because H2 does't support COVAR_POP, COVAR_SAMP and CORR works with DISTINCT.
So apache#35145 introduces a bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

### Why are the changes needed?
Fix bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

### Does this PR introduce _any_ user-facing change?
'Yes'.
Bug will be fix.

### How was this patch tested?
New test cases.

Closes apache#37090 from beliefer/SPARK-37527_followup2.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39627][SQL] DS V2 pushdown should unify the compile API

### What changes were proposed in this pull request?
Currently, `JdbcDialect` have two API `compileAggregate` and `compileExpression`, we can unify them.

### Why are the changes needed?
Improve ease of use.

### Does this PR introduce _any_ user-facing change?
'No'.
The two API `compileAggregate` call `compileExpression` not changed.

### How was this patch tested?
N/A

Closes apache#37047 from beliefer/SPARK-39627.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39384][SQL] Compile built-in linear regression aggregate functions for JDBC dialect

### What changes were proposed in this pull request?
Recently, Spark DS V2 pushdown framework translate a lot of standard linear regression aggregate functions.
Currently, only H2Dialect compile these standard linear regression aggregate functions. This PR compile these standard linear regression aggregate functions for other build-in JDBC dialect.

### Why are the changes needed?
Make build-in JDBC dialect support compile linear regression aggregate push-down.

### Does this PR introduce _any_ user-facing change?
'No'.
New feature.

### How was this patch tested?
New test cases.

Closes apache#37188 from beliefer/SPARK-39384.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Sean Owen <srowen@gmail.com>

* [SPARK-39148][SQL] DS V2 aggregate push down can work with OFFSET or LIMIT

### What changes were proposed in this pull request?

This PR refactors the v2 agg pushdown code. The main change is, now we don't build the `Scan` immediately when pushing agg. We did it so before because we want to know the data schema with agg pushed, then we can add cast when rewriting the query plan after pushdown. But the problem is, we build `Scan` too early and can't push down any more operators, while it's common to see LIMIT/OFFSET after agg.

The idea of the refactor is, we don't need to know the data schema with agg pushed. We just give an expectation (the data type should be the same of Spark agg functions), use it to define the output of `ScanBuilderHolder`, and then rewrite the query plan. Later on, when we build the `Scan` and replace `ScanBuilderHolder` with `DataSourceV2ScanRelation`, we check the actual data schema and add a `Project` to do type cast if necessary.

### Why are the changes needed?

support pushing down LIMIT/OFFSET after agg.

### Does this PR introduce _any_ user-facing change?

no

### How was this patch tested?

updated tests

Closes apache#37195 from cloud-fan/agg.

Lead-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

Co-authored-by: Jiaan Geng <beliefer@163.com>
Co-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
leejaywei pushed a commit to Kyligence/spark that referenced this pull request Aug 29, 2022
…LIMIT (#505)

* [SPARK-39139][SQL] DS V2 supports push down DS V2 UDF

Currently, Spark DS V2 push-down framework supports push down SQL to data sources.
But the DS V2 push-down framework only support push down the built-in functions to data sources.
Each database have a lot very useful functions which not supported by Spark.
If we can push down these functions into data source, it will reduce disk I/O and network I/O and improve the performance when query databases.

1. Spark can leverage the functions supported by databases
2. Improve the query performance.

'No'.
New feature.

New tests.

Closes apache#36593 from beliefer/SPARK-39139.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39453][SQL][TESTS][FOLLOWUP] Let `RAND` in filter is more meaningful

apache#36830 makes DS V2 supports push down misc non-aggregate functions(non ANSI).
But he `Rand` in test case looks no meaningful.

Let `Rand` in filter is more meaningful.

'No'.
Just update test case.

Just update test case.

Closes apache#37033 from beliefer/SPARK-39453_followup.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-37527][SQL][FOLLOWUP] Cannot compile COVAR_POP, COVAR_SAMP and CORR in `H2Dialect` if them with `DISTINCT`

apache#35145 compile COVAR_POP, COVAR_SAMP and CORR in H2Dialect.
Because H2 does't support COVAR_POP, COVAR_SAMP and CORR works with DISTINCT.
So apache#35145 introduces a bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

Fix bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

'Yes'.
Bug will be fix.

New test cases.

Closes apache#37090 from beliefer/SPARK-37527_followup2.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39627][SQL] DS V2 pushdown should unify the compile API

Currently, `JdbcDialect` have two API `compileAggregate` and `compileExpression`, we can unify them.

Improve ease of use.

'No'.
The two API `compileAggregate` call `compileExpression` not changed.

N/A

Closes apache#37047 from beliefer/SPARK-39627.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39384][SQL] Compile built-in linear regression aggregate functions for JDBC dialect

Recently, Spark DS V2 pushdown framework translate a lot of standard linear regression aggregate functions.
Currently, only H2Dialect compile these standard linear regression aggregate functions. This PR compile these standard linear regression aggregate functions for other build-in JDBC dialect.

Make build-in JDBC dialect support compile linear regression aggregate push-down.

'No'.
New feature.

New test cases.

Closes apache#37188 from beliefer/SPARK-39384.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Sean Owen <srowen@gmail.com>

* [SPARK-39148][SQL] DS V2 aggregate push down can work with OFFSET or LIMIT

This PR refactors the v2 agg pushdown code. The main change is, now we don't build the `Scan` immediately when pushing agg. We did it so before because we want to know the data schema with agg pushed, then we can add cast when rewriting the query plan after pushdown. But the problem is, we build `Scan` too early and can't push down any more operators, while it's common to see LIMIT/OFFSET after agg.

The idea of the refactor is, we don't need to know the data schema with agg pushed. We just give an expectation (the data type should be the same of Spark agg functions), use it to define the output of `ScanBuilderHolder`, and then rewrite the query plan. Later on, when we build the `Scan` and replace `ScanBuilderHolder` with `DataSourceV2ScanRelation`, we check the actual data schema and add a `Project` to do type cast if necessary.

support pushing down LIMIT/OFFSET after agg.

no

updated tests

Closes apache#37195 from cloud-fan/agg.

Lead-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

Co-authored-by: Jiaan Geng <beliefer@163.com>
Co-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
leejaywei pushed a commit to Kyligence/spark that referenced this pull request Aug 29, 2022
…LIMIT (#505)

* [SPARK-39139][SQL] DS V2 supports push down DS V2 UDF

Currently, Spark DS V2 push-down framework supports push down SQL to data sources.
But the DS V2 push-down framework only support push down the built-in functions to data sources.
Each database have a lot very useful functions which not supported by Spark.
If we can push down these functions into data source, it will reduce disk I/O and network I/O and improve the performance when query databases.

1. Spark can leverage the functions supported by databases
2. Improve the query performance.

'No'.
New feature.

New tests.

Closes apache#36593 from beliefer/SPARK-39139.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39453][SQL][TESTS][FOLLOWUP] Let `RAND` in filter is more meaningful

apache#36830 makes DS V2 supports push down misc non-aggregate functions(non ANSI).
But he `Rand` in test case looks no meaningful.

Let `Rand` in filter is more meaningful.

'No'.
Just update test case.

Just update test case.

Closes apache#37033 from beliefer/SPARK-39453_followup.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-37527][SQL][FOLLOWUP] Cannot compile COVAR_POP, COVAR_SAMP and CORR in `H2Dialect` if them with `DISTINCT`

apache#35145 compile COVAR_POP, COVAR_SAMP and CORR in H2Dialect.
Because H2 does't support COVAR_POP, COVAR_SAMP and CORR works with DISTINCT.
So apache#35145 introduces a bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

Fix bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

'Yes'.
Bug will be fix.

New test cases.

Closes apache#37090 from beliefer/SPARK-37527_followup2.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39627][SQL] DS V2 pushdown should unify the compile API

Currently, `JdbcDialect` have two API `compileAggregate` and `compileExpression`, we can unify them.

Improve ease of use.

'No'.
The two API `compileAggregate` call `compileExpression` not changed.

N/A

Closes apache#37047 from beliefer/SPARK-39627.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39384][SQL] Compile built-in linear regression aggregate functions for JDBC dialect

Recently, Spark DS V2 pushdown framework translate a lot of standard linear regression aggregate functions.
Currently, only H2Dialect compile these standard linear regression aggregate functions. This PR compile these standard linear regression aggregate functions for other build-in JDBC dialect.

Make build-in JDBC dialect support compile linear regression aggregate push-down.

'No'.
New feature.

New test cases.

Closes apache#37188 from beliefer/SPARK-39384.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Sean Owen <srowen@gmail.com>

* [SPARK-39148][SQL] DS V2 aggregate push down can work with OFFSET or LIMIT

This PR refactors the v2 agg pushdown code. The main change is, now we don't build the `Scan` immediately when pushing agg. We did it so before because we want to know the data schema with agg pushed, then we can add cast when rewriting the query plan after pushdown. But the problem is, we build `Scan` too early and can't push down any more operators, while it's common to see LIMIT/OFFSET after agg.

The idea of the refactor is, we don't need to know the data schema with agg pushed. We just give an expectation (the data type should be the same of Spark agg functions), use it to define the output of `ScanBuilderHolder`, and then rewrite the query plan. Later on, when we build the `Scan` and replace `ScanBuilderHolder` with `DataSourceV2ScanRelation`, we check the actual data schema and add a `Project` to do type cast if necessary.

support pushing down LIMIT/OFFSET after agg.

no

updated tests

Closes apache#37195 from cloud-fan/agg.

Lead-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

Co-authored-by: Jiaan Geng <beliefer@163.com>
Co-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
RolatZhang pushed a commit to Kyligence/spark that referenced this pull request Aug 29, 2023
…LIMIT (#505)

* [SPARK-39139][SQL] DS V2 supports push down DS V2 UDF

Currently, Spark DS V2 push-down framework supports push down SQL to data sources.
But the DS V2 push-down framework only support push down the built-in functions to data sources.
Each database have a lot very useful functions which not supported by Spark.
If we can push down these functions into data source, it will reduce disk I/O and network I/O and improve the performance when query databases.

1. Spark can leverage the functions supported by databases
2. Improve the query performance.

'No'.
New feature.

New tests.

Closes apache#36593 from beliefer/SPARK-39139.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39453][SQL][TESTS][FOLLOWUP] Let `RAND` in filter is more meaningful

apache#36830 makes DS V2 supports push down misc non-aggregate functions(non ANSI).
But he `Rand` in test case looks no meaningful.

Let `Rand` in filter is more meaningful.

'No'.
Just update test case.

Just update test case.

Closes apache#37033 from beliefer/SPARK-39453_followup.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-37527][SQL][FOLLOWUP] Cannot compile COVAR_POP, COVAR_SAMP and CORR in `H2Dialect` if them with `DISTINCT`

apache#35145 compile COVAR_POP, COVAR_SAMP and CORR in H2Dialect.
Because H2 does't support COVAR_POP, COVAR_SAMP and CORR works with DISTINCT.
So apache#35145 introduces a bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

Fix bug that compile COVAR_POP, COVAR_SAMP and CORR if these aggregate functions with DISTINCT.

'Yes'.
Bug will be fix.

New test cases.

Closes apache#37090 from beliefer/SPARK-37527_followup2.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39627][SQL] DS V2 pushdown should unify the compile API

Currently, `JdbcDialect` have two API `compileAggregate` and `compileExpression`, we can unify them.

Improve ease of use.

'No'.
The two API `compileAggregate` call `compileExpression` not changed.

N/A

Closes apache#37047 from beliefer/SPARK-39627.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

* [SPARK-39384][SQL] Compile built-in linear regression aggregate functions for JDBC dialect

Recently, Spark DS V2 pushdown framework translate a lot of standard linear regression aggregate functions.
Currently, only H2Dialect compile these standard linear regression aggregate functions. This PR compile these standard linear regression aggregate functions for other build-in JDBC dialect.

Make build-in JDBC dialect support compile linear regression aggregate push-down.

'No'.
New feature.

New test cases.

Closes apache#37188 from beliefer/SPARK-39384.

Authored-by: Jiaan Geng <beliefer@163.com>
Signed-off-by: Sean Owen <srowen@gmail.com>

* [SPARK-39148][SQL] DS V2 aggregate push down can work with OFFSET or LIMIT

This PR refactors the v2 agg pushdown code. The main change is, now we don't build the `Scan` immediately when pushing agg. We did it so before because we want to know the data schema with agg pushed, then we can add cast when rewriting the query plan after pushdown. But the problem is, we build `Scan` too early and can't push down any more operators, while it's common to see LIMIT/OFFSET after agg.

The idea of the refactor is, we don't need to know the data schema with agg pushed. We just give an expectation (the data type should be the same of Spark agg functions), use it to define the output of `ScanBuilderHolder`, and then rewrite the query plan. Later on, when we build the `Scan` and replace `ScanBuilderHolder` with `DataSourceV2ScanRelation`, we check the actual data schema and add a `Project` to do type cast if necessary.

support pushing down LIMIT/OFFSET after agg.

no

updated tests

Closes apache#37195 from cloud-fan/agg.

Lead-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

Co-authored-by: Jiaan Geng <beliefer@163.com>
Co-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants