cloud-fan commented on a change in pull request #34462:
URL: https://github.com/apache/spark/pull/34462#discussion_r740747727
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/types/StructType.scala
##########
@@ -643,15 +643,14 @@ object StructType extends AbstractDataType {
DecimalType.Fixed(rightPrecision, rightScale)) =>
if ((leftPrecision == rightPrecision) && (leftScale == rightScale)) {
Review comment:
how about this
```
if(leftScale == rightScale) {
DecimalType(leftPrecision.max(rightPrecision), leftScale)
} else {
throw
QueryExecutionErrors.cannotMergeDecimalTypesWithIncompatibleScaleError
}
```
I think we don't need to have two different error messages. We just need to
point out the root cause, which is the incompatible scale.
##########
File path:
sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetSchemaSuite.scala
##########
@@ -388,6 +391,27 @@ class ParquetSchemaSuite extends ParquetSchemaTest {
}
}
+ test("SPARK-37191: Schema merging for DecimalType with different precision
but the same scale") {
+ import testImplicits._
+
+ withTempPath { dir =>
+ val path = dir.getCanonicalPath
+
+ val data1 = spark.sparkContext.parallelize(Seq(Row(new
BigDecimal("123456789.11"))), 1)
+ val schema1 = StructType(StructField("col", DecimalType(12, 2)) :: Nil)
+
+ val data2 = spark.sparkContext.parallelize(Seq(Row(new
BigDecimal("1234567890000.11"))), 1)
+ val schema2 = StructType(StructField("col", DecimalType(17, 2)) :: Nil)
+
+ spark.createDataFrame(data1, schema1).write.parquet(path)
+ spark.createDataFrame(data2, schema2).write.mode("append").parquet(path)
+
+ val res = spark.read.option("mergeSchema", "true").parquet(path)
+ assert(res.schema("col").dataType == DecimalType(17, 2))
+ res.foreach(_ => ()) // must not throw exception
Review comment:
shall we use `checkAnswer` to check the result?
##########
File path:
sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetSchemaSuite.scala
##########
@@ -388,6 +391,27 @@ class ParquetSchemaSuite extends ParquetSchemaTest {
}
}
+ test("SPARK-37191: Schema merging for DecimalType with different precision
but the same scale") {
+ import testImplicits._
+
+ withTempPath { dir =>
+ val path = dir.getCanonicalPath
+
+ val data1 = spark.sparkContext.parallelize(Seq(Row(new
BigDecimal("123456789.11"))), 1)
+ val schema1 = StructType(StructField("col", DecimalType(12, 2)) :: Nil)
+
+ val data2 = spark.sparkContext.parallelize(Seq(Row(new
BigDecimal("1234567890000.11"))), 1)
+ val schema2 = StructType(StructField("col", DecimalType(17, 2)) :: Nil)
+
+ spark.createDataFrame(data1, schema1).write.parquet(path)
+ spark.createDataFrame(data2, schema2).write.mode("append").parquet(path)
+
+ val res = spark.read.option("mergeSchema", "true").parquet(path)
+ assert(res.schema("col").dataType == DecimalType(17, 2))
+ res.foreach(_ => ()) // must not throw exception
Review comment:
maybe we should put the test in `ParquetQuerySuite`, so that we can test
all the parquet readers.
--
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.
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
