Github user JoshRosen commented on a diff in the pull request:
https://github.com/apache/spark/pull/12750#discussion_r61665640
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/json/InferSchema.scala
---
@@ -246,12 +263,39 @@ private[sql] object InferSchema {
}
case (StructType(fields1), StructType(fields2)) =>
- val newFields = (fields1 ++ fields2).groupBy(field =>
field.name).map {
- case (name, fieldTypes) =>
- val dataType =
fieldTypes.view.map(_.dataType).reduce(compatibleType)
- StructField(name, dataType, nullable = true)
+ // Both fields1 and fields2 should be sorted by name, since
inferField performs sorting.
+ // Therefore, we can take advantage of the fact that we're
merging sorted lists and skip
+ // building a hash map or performing additional sorting.
+ val newFields = new mutable.ArrayBuffer[StructField]()
+
+ var f1Idx = 0
+ var f2Idx = 0
+
+ while (f1Idx < fields1.length && f2Idx < fields2.length) {
+ val f1Name = fields1(f1Idx).name
+ val f2Name = fields2(f2Idx).name
+ if (f1Name == f2Name) {
+ val dataType = compatibleType(fields1(f1Idx).dataType,
fields2(f2Idx).dataType)
+ newFields += StructField(f1Name, dataType, nullable = true)
+ f1Idx += 1
+ f2Idx += 1
+ } else if (f1Name < f2Name) {
+ newFields += fields1(f1Idx)
+ f1Idx += 1
+ } else { // f1Name > f2Name
+ newFields += fields2(f2Idx)
+ f2Idx += 1
+ }
+ }
+ while (f1Idx < fields1.length) {
+ newFields += fields1(f1Idx)
+ f1Idx += 1
+ }
+ while (f2Idx < fields2.length) {
+ newFields += fields2(f2Idx)
+ f2Idx += 1
}
- StructType(newFields.toSeq.sortBy(_.name))
+ StructType(newFields.toSeq)
--- End diff --
Agreed; good catch.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
