You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Bruce Robbins (JIRA)" <ji...@apache.org> on 2018/06/23 01:35:00 UTC
[jira] [Created] (SPARK-24633) arrays_zip function's code generator
splits input processing incorrectly
Bruce Robbins created SPARK-24633:
-------------------------------------
Summary: arrays_zip function's code generator splits input processing incorrectly
Key: SPARK-24633
URL: https://issues.apache.org/jira/browse/SPARK-24633
Project: Spark
Issue Type: Bug
Components: SQL
Affects Versions: 2.4.0
Environment: Mac OS High Sierra
Reporter: Bruce Robbins
This works:
{noformat}
scala> val df = spark.read.parquet("many_arrays_per_row")
df: org.apache.spark.sql.DataFrame = [k0: array<bigint>, k1: array<bigint> ... 98 more fields]
scala> df.selectExpr("arrays_zip(k0, k1, k2)").show(truncate=false)
+----------------------------------------+
|arrays_zip(k0, k1, k2) |
+----------------------------------------+
|[[6583, 1312, 7460], [668, 1626, 4129]] |
|[[5415, 5251, 1514], [1631, 2224, 2553]]|
+----------------------------------------+
{noformat}
If I add one more array to the parameter list, I get this:
{noformat}
scala> df.selectExpr("arrays_zip(k0, k1, k2, k3)").show(truncate=false)
18/06/22 18:06:41 ERROR CodeGenerator: failed to compile: org.codehaus.commons.compiler.CompileException: File 'generated.java', Line 92, Column 35: Unknown variable or type "scan_row_0"
org.codehaus.commons.compiler.CompileException: File 'generated.java', Line 92, Column 35: Unknown variable or type "scan_row_0"
at org.codehaus.janino.UnitCompiler.compileError(UnitCompiler.java:11821)
at org.codehaus.janino.UnitCompiler.getType2(UnitCompiler.java:6521)
at org.codehaus.janino.UnitCompiler.access$13100(UnitCompiler.java:212)
at org.codehaus.janino.UnitCompiler$18.visitPackage(UnitCompiler.java:6133)
.. much exception trace...
18/06/22 18:06:41 WARN WholeStageCodegenExec: Whole-stage codegen disabled for plan (id=1):
*(1) LocalLimit 21
+- *(1) Project [cast(arrays_zip(k0#375, k1#376, k2#387, k3#398) as string) AS arrays_zip(k0, k1, k2, k3)#619]
+- *(1) FileScan parquet [k0#375,k1#376,k2#387,k3#398] Batched: false, Format: Parquet, Location: InMemoryFileIndex[file:/Users/brobbins/github/spark_upstream/many_arrays_per_row], PartitionFilters: [], PushedFilters: [], ReadSchema: struct<k0:array<bigint>,k1:array<bigint>,k2:array<bigint>,k3:array<bigint>>
+----------------------------------------------------+
|arrays_zip(k0, k1, k2, k3) |
+----------------------------------------------------+
|[[6583, 1312, 7460, 3712], [668, 1626, 4129, 2815]] |
|[[5415, 5251, 1514, 1580], [1631, 2224, 2553, 7555]]|
+----------------------------------------------------+
{noformat}
I still got the answer!
I add a 5th parameter:
{noformat}
scala> df.selectExpr("arrays_zip(k0, k1, k2, k3, k4)").show(truncate=false)
18/06/22 18:07:53 ERROR CodeGenerator: failed to compile: org.codehaus.commons.compiler.CompileException: File 'generated.java', Line 97, Column 35: Unknown variable or type "scan_row_0"
org.codehaus.commons.compiler.CompileException: File 'generated.java', Line 97, Column 35: Unknown variable or type "scan_row_0"
at org.codehaus.janino.UnitCompiler.compileError(UnitCompiler.java:11821)
at org.codehaus.janino.UnitCompiler.getType2(UnitCompiler.java:6521)
at org.codehaus.janino.UnitCompiler.access$13100(UnitCompiler.java:212)
at org.codehaus.janino.UnitCompiler$18.visitPackage(UnitCompiler.java:6133)
at org.codehaus.janino.UnitCompiler$18.visitPackage(UnitCompiler.java:6130)
at org.codehaus.janino.Java$Package.accept(Java.java:4077)
.. much exception trace...
Caused by: org.codehaus.commons.compiler.CompileException: File 'generated.java', Line 73, Column 21: failed to compile: org.codehaus.commons.compiler.CompileException: File 'generated.java', Line 73, Column 21: Unknown variable or type "i"
at org.apache.spark.sql.catalyst.expressions.codegen.CodeGenerator$.org$apache$spark$sql$catalyst$expressions$codegen$CodeGenerator$$doCompile(CodeGenerator.scal\
a:1361)
at org.apache.spark.sql.catalyst.expressions.codegen.CodeGenerator$$anon$1.load(CodeGenerator.scala:1423)
at org.apache.spark.sql.catalyst.expressions.codegen.CodeGenerator$$anon$1.load(CodeGenerator.scala:1420)
at com.google.common.cache.LocalCache$LoadingValueReference.loadFuture(LocalCache.java:3599)
at com.google.common.cache.LocalCache$Segment.loadSync(LocalCache.java:2379)
... 31 more
scala>
{noformat}
This time, no result.
It looks like the generated code is expecting the input row to be in a parameter (either i or scan_row_x), but that parameter is not passed to the input handler function (see lines 069 and 073)
{noformat}
/* 069 */ private int getValuesAndCardinalities_0_1(ArrayData[] arrVals_0, int biggestCardinality_0) {
/* 070 */
/* 071 */
/* 072 */ if (biggestCardinality_0 != -1) {
/* 073 */ boolean isNull_6 = i.isNullAt(4);
/* 074 */ ArrayData value_6 = isNull_6 ?
/* 075 */ null : (i.getArray(4));
/* 076 */ if (!isNull_6) {
/* 077 */ arrVals_0[4] = value_6;
/* 078 */ biggestCardinality_0 = Math.max(biggestCardinality_0, value_6.numElements());
/* 079 */ } else {
/* 080 */ biggestCardinality_0 = -1;
/* 081 */ }
/* 082 */ }
/* 083 */
/* 084 */ return biggestCardinality_0;
/* 085 */
/* 086 */ }
{noformat}
Here's the scan_row_0 case:
{noformat}
/* 095 */ private int project_getValuesAndCardinalities_0_1(ArrayData[] project_arrVals_0, int project_biggestCardinality_0) {
/* 096 */ if (project_biggestCardinality_0 != -1) {
/* 097 */ boolean scan_isNull_3 = scan_row_0.isNullAt(3);
/* 098 */ ArrayData scan_value_3 = scan_isNull_3 ?
/* 099 */ null : (scan_row_0.getArray(3));
/* 100 */ if (!scan_isNull_3) {
/* 101 */ project_arrVals_0[3] = scan_value_3;
/* 102 */ project_biggestCardinality_0 = Math.max(project_biggestCardinality_0, scan_value_3.numElements());
/* 103 */ } else {
/* 104 */ project_biggestCardinality_0 = -1;
/* 105 */ }
/* 106 */ }
{noformat}
I am marking this as minor since this function is new and not in a released version of Spark.
cc [~DylanGuedes]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org