497d17f38a
Change `Inline.eval` to return a row of null values rather than a null row in the case of a null input struct.
Consider the following query:
```
set spark.sql.codegen.wholeStage=false;
select inline(array(named_struct('a', 1, 'b', 2), null));
```
This query fails with a `NullPointerException`:
```
22/06/16 15:10:06 ERROR Executor: Exception in task 0.0 in stage 0.0 (TID 0)
java.lang.NullPointerException
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.apply(Unknown Source)
at org.apache.spark.sql.execution.GenerateExec.$anonfun$doExecute$11(GenerateExec.scala:122)
```
(In Spark 3.1.3, you don't need to set `spark.sql.codegen.wholeStage` to false to reproduce the error, since Spark 3.1.3 has no codegen path for `Inline`).
This query fails regardless of the setting of `spark.sql.codegen.wholeStage`:
```
val dfWide = (Seq((1))
.toDF("col0")
.selectExpr(Seq.tabulate(99)(x => s"$x as col${x + 1}"): _*))
val df = (dfWide
.selectExpr("*", "array(named_struct('a', 1, 'b', 2), null) as struct_array"))
df.selectExpr("*", "inline(struct_array)").collect
```
It fails with
```
22/06/16 15:18:55 ERROR Executor: Exception in task 0.0 in stage 0.0 (TID 0)/ 1]
java.lang.NullPointerException
at org.apache.spark.sql.catalyst.expressions.JoinedRow.isNullAt(JoinedRow.scala:80)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.writeFields_0_8$(Unknown Source)
```
When `Inline.eval` returns a null row in the collection, GenerateExec gets a NullPointerException either when joining the null row with required child output, or projecting the null row.
This PR avoids producing the null row and produces a row of null values instead:
```
spark-sql> set spark.sql.codegen.wholeStage=false;
spark.sql.codegen.wholeStage false
Time taken: 3.095 seconds, Fetched 1 row(s)
spark-sql> select inline(array(named_struct('a', 1, 'b', 2), null));
1 2
NULL NULL
Time taken: 1.214 seconds, Fetched 2 row(s)
spark-sql>
```
No.
New unit test.
Closes #36903 from bersprockets/inline_eval_null_struct_issue.
Authored-by: Bruce Robbins <bersprockets@gmail.com>
Signed-off-by: Hyukjin Kwon <gurwls223@apache.org>
(cherry picked from commit
|
||
---|---|---|
.. | ||
catalyst | ||
core | ||
hive | ||
hive-thriftserver | ||
README.md | ||
create-docs.sh | ||
gen-sql-api-docs.py | ||
gen-sql-config-docs.py | ||
gen-sql-functions-docs.py | ||
mkdocs.yml |
README.md
Spark SQL
This module provides support for executing relational queries expressed in either SQL or the DataFrame/Dataset API.
Spark SQL is broken up into four subprojects:
- Catalyst (sql/catalyst) - An implementation-agnostic framework for manipulating trees of relational operators and expressions.
- Execution (sql/core) - A query planner / execution engine for translating Catalyst's logical query plans into Spark RDDs. This component also includes a new public interface, SQLContext, that allows users to execute SQL or LINQ statements against existing RDDs and Parquet files.
- Hive Support (sql/hive) - Includes extensions that allow users to write queries using a subset of HiveQL and access data from a Hive Metastore using Hive SerDes. There are also wrappers that allow users to run queries that include Hive UDFs, UDAFs, and UDTFs.
- HiveServer and CLI support (sql/hive-thriftserver) - Includes support for the SQL CLI (bin/spark-sql) and a HiveServer2 (for JDBC/ODBC) compatible server.
Running ./sql/create-docs.sh
generates SQL documentation for built-in functions under sql/site
, and SQL configuration documentation that gets included as part of configuration.md
in the main docs
directory.