[SPARK-28521][SQL] Fix error message for built-in functions

[wangyum]

What changes were proposed in this pull request?

spark-sql> select cast(1);
19/07/26 00:54:17 ERROR SparkSQLDriver: Failed in [select cast(1)]
java.lang.UnsupportedOperationException: empty.init
	at scala.collection.TraversableLike$class.init(TraversableLike.scala:451)
	at scala.collection.mutable.ArrayOps$ofInt.scala$collection$IndexedSeqOptimized$$super$init(ArrayOps.scala:234)
	at scala.collection.IndexedSeqOptimized$class.init(IndexedSeqOptimized.scala:135)
	at scala.collection.mutable.ArrayOps$ofInt.init(ArrayOps.scala:234)
	at org.apache.spark.sql.catalyst.analysis.FunctionRegistry$$anonfun$7$$anonfun$11.apply(FunctionRegistry.scala:565)
	at org.apache.spark.sql.catalyst.analysis.FunctionRegistry$$anonfun$7$$anonfun$11.apply(FunctionRegistry.scala:558)
	at scala.Option.getOrElse(Option.scala:121)

The reason is that we did not handle the case validParametersCount.length == 0 because the parameter types can be Expression, DataType and Option. This PR makes it handle the case validParametersCount.length == 0.

How was this patch tested?

unit tests

[wangyum]

cc @mgaido91

[mgaido91]

but here we're still filtering by expressions..am I missing something?

[wangyum]

Two reasons:

1. It will throw exception at CheckAnalysis if parameter is a DataType. So we do not need handle it here:

scala> spark.sql("select cast(int)")
org.apache.spark.sql.AnalysisException: cannot resolve '`int`' given input columns: []; line 1 pos 12;
'Project [unresolvedalias('cast('int), None)]
+- OneRowRelation

  at org.apache.spark.sql.catalyst.analysis.package$AnalysisErrorAt.failAnalysis(package.scala:42)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$$anonfun$$nestedInanonfun$checkAnalysis$1$2.applyOrElse(CheckAnalysis.scala:113)
  at org.apache.spark.sql.catalyst.analysis.CheckAnalysis$$anonfun$$nestedInanonfun$checkAnalysis$1$2.applyOrElse(CheckAnalysis.scala:110)
  at org.apache.spark.sql.catalyst.trees.TreeNode.$anonfun$transformUp$2(TreeNode.scala:306)

2. It's hard to know a parameter is DataType or Expression here.

[mgaido91]

So, I think the problem here is not that we are not considering the DataType args, but the problem IMHO is that we are not handling the case validParametersCount == 0. I think in that case we should throw something like: Invalid arguments for function $name.. Indeed, in the cast case, I'd argue that it is not so clear to me whether the arguments are 1 or 2, since it is not cast(1, string) but cast(1 as string)... So I don't think that it is correct to state that the number of arguments for cast is 2.

[SparkQA]

Test build #108210 has finished for PR 25261 at commit 32a671c.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #108438 has finished for PR 25261 at commit 94af203.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[mgaido91]

LGTM, just a comment on wording. Thanks.

[mgaido91]

nit: in case of 0, I'd prefer a more generic Invalid arguments for function $name, instead of Invalid number of arguments for function $name. Because as I mentioned earlier, I think it is questionable which is the number of arguments in this case...

[wangyum]

Done

[SparkQA]

Test build #108500 has finished for PR 25261 at commit 3c46a84.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[srowen]

Merged to master