HeartSaVioR commented on a change in pull request #27025: [SPARK-26560][SQL]
Spark should be able to run Hive UDF using jar regardless of current thread
context classloader
URL: https://github.com/apache/spark/pull/27025#discussion_r361820154
##########
File path:
sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveSessionCatalog.scala
##########
@@ -66,49 +66,52 @@ private[sql] class HiveSessionCatalog(
name: String,
clazz: Class[_],
input: Seq[Expression]): Expression = {
-
- Try(super.makeFunctionExpression(name, clazz, input)).getOrElse {
- var udfExpr: Option[Expression] = None
- try {
- // When we instantiate hive UDF wrapper class, we may throw exception
if the input
- // expressions don't satisfy the hive UDF, such as type mismatch,
input number
- // mismatch, etc. Here we catch the exception and throw
AnalysisException instead.
- if (classOf[UDF].isAssignableFrom(clazz)) {
- udfExpr = Some(HiveSimpleUDF(name, new
HiveFunctionWrapper(clazz.getName), input))
- udfExpr.get.dataType // Force it to check input data types.
- } else if (classOf[GenericUDF].isAssignableFrom(clazz)) {
- udfExpr = Some(HiveGenericUDF(name, new
HiveFunctionWrapper(clazz.getName), input))
- udfExpr.get.dataType // Force it to check input data types.
- } else if
(classOf[AbstractGenericUDAFResolver].isAssignableFrom(clazz)) {
- udfExpr = Some(HiveUDAFFunction(name, new
HiveFunctionWrapper(clazz.getName), input))
- udfExpr.get.dataType // Force it to check input data types.
- } else if (classOf[UDAF].isAssignableFrom(clazz)) {
- udfExpr = Some(HiveUDAFFunction(
- name,
- new HiveFunctionWrapper(clazz.getName),
- input,
- isUDAFBridgeRequired = true))
- udfExpr.get.dataType // Force it to check input data types.
- } else if (classOf[GenericUDTF].isAssignableFrom(clazz)) {
- udfExpr = Some(HiveGenericUDTF(name, new
HiveFunctionWrapper(clazz.getName), input))
- udfExpr.get.asInstanceOf[HiveGenericUDTF].elementSchema // Force it
to check data types.
+ // Current thread context classloader may not be the one loaded the class.
Need to switch
+ // context classloader to initialize instance properly.
+ Utils.withContextClassLoader(clazz.getClassLoader) {
Review comment:
I can't imagine clazz is loaded from other than one of three cases - 1.
available in classpath 2. class dynamically loaded (spark-shell, and more?) 3.
JAR dynamically loaded - and if I understand correctly, Spark classpath (and
Hive dependencies as well if Hive is enabled) is available for the classloader
which loads the clazz for all three cases. This means clazz is the only one we
need to make sure the current context classloader can load it.
----------------------------------------------------------------
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:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
