(datasketches-spark) branch simplify_registration updated: Clean up codegen errors a bit -- still one for pmf/cdf but fallback works ok

[jm...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

jmalkin pushed a commit to branch simplify_registration
in repository https://gitbox.apache.org/repos/asf/datasketches-spark.git


The following commit(s) were added to refs/heads/simplify_registration by this push:
     new c689e44  Clean up codegen errors a bit -- still one for pmf/cdf but fallback works ok
c689e44 is described below

commit c689e44242b87f7573d4d5714dfef754a243f1a7
Author: Jon Malkin <78...@users.noreply.github.com>
AuthorDate: Tue Jan 7 14:50:04 2025 -0800

    Clean up codegen errors a bit -- still one for pmf/cdf but fallback works ok
---
 src/main/scala/org/apache/spark/sql/expressions/KllExpressions.scala | 5 +----
 src/main/scala/org/apache/spark/sql/types/KllDoublesSketchType.scala | 4 ++--
 2 files changed, 3 insertions(+), 6 deletions(-)

diff --git a/src/main/scala/org/apache/spark/sql/expressions/KllExpressions.scala b/src/main/scala/org/apache/spark/sql/expressions/KllExpressions.scala
index 34b6296..9eec024 100644
--- a/src/main/scala/org/apache/spark/sql/expressions/KllExpressions.scala
+++ b/src/main/scala/org/apache/spark/sql/expressions/KllExpressions.scala
@@ -256,15 +256,12 @@ case class KllGetPmfCdf(first: Expression,
       s"""
          |${sketchEval.code}
          |${splitPointsEval.code}
-         |if (${sketchEval.isNull} || ${splitPointsEval.isNull}) {
-         |  boolean ${ev.isNull} = true;
-         |} else {
+         |if (!${sketchEval.isNull} && !${splitPointsEval.isNull}) {
          |  org.apache.datasketches.quantilescommon.QuantileSearchCriteria searchCriteria = ${if (isInclusive) "org.apache.datasketches.quantilescommon.QuantileSearchCriteria.INCLUSIVE" else "org.apache.datasketches.quantilescommon.QuantileSearchCriteria.EXCLUSIVE"};
          |  final org.apache.datasketches.kll.KllDoublesSketch $sketch = org.apache.spark.sql.types.KllDoublesSketchType.wrap(${sketchEval.value});
          |  final double[] splitPoints = ((org.apache.spark.sql.catalyst.util.GenericArrayData)${splitPointsEval.value}).toDoubleArray();
          |  final double[] result = ${if (isPmf) s"$sketch.getPMF(splitPoints, searchCriteria)" else s"$sketch.getCDF(splitPoints, searchCriteria)"};
          |  org.apache.spark.sql.catalyst.util.GenericArrayData ${ev.value} = new org.apache.spark.sql.catalyst.util.GenericArrayData(result);
-         |  boolean ${ev.isNull} = false;
          |}
        """.stripMargin
     ev.copy(code = CodeBlock(Seq(code), Seq.empty))
diff --git a/src/main/scala/org/apache/spark/sql/types/KllDoublesSketchType.scala b/src/main/scala/org/apache/spark/sql/types/KllDoublesSketchType.scala
index fda471d..2fb01e9 100644
--- a/src/main/scala/org/apache/spark/sql/types/KllDoublesSketchType.scala
+++ b/src/main/scala/org/apache/spark/sql/types/KllDoublesSketchType.scala
@@ -51,11 +51,11 @@ case object KllDoublesSketchType extends KllDoublesSketchType {
   })
 
   // non-udf version
-  val wrap = ((bytes: Array[Byte]) => {
+  def wrap(bytes: Array[Byte]): KllDoublesSketch = {
     if (bytes == null) {
       null
     } else {
       deserialize(bytes)
     }
-  })
+  }
 }


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@datasketches.apache.org
For additional commands, e-mail: commits-help@datasketches.apache.org