stevedlawrence commented on a change in pull request #262: Unordered sequences
URL: https://github.com/apache/incubator-daffodil/pull/262#discussion_r324350310
##########
File path:
daffodil-runtime1/src/main/scala/org/apache/daffodil/infoset/InfosetImpl.scala
##########
@@ -1415,6 +1423,50 @@ sealed class DIComplex(override val erd:
ElementRuntimeData, val tunable: Daffod
}
}
+ final def sortChildNodesByPosition(): Unit = {
+ // TODO: Once we upgrade to Scala 2.13 we can call the inplace sort
method: DAFFODIL-2152
+ // childNodes = childNodes.sortInPlace(_.erd.position)
+ childNodes = childNodes.sortBy(_.erd.position)
+ }
+
+ /*
+ * When parsing unordered sequences it is possible to have non-contiguous
arrays.
+ * When parsed, these arrays show up as separate DIArrays in the infoset. We
need
+ * to combine these separate arrays.
+ * */
+ final def flattenAndValidateChildNodes(): Maybe[List[ElementRuntimeData]] = {
+ var ret: List[ElementRuntimeData] = List()
+ var groups = childNodes.groupBy(_.erd)
+ groups foreach {
+ case (erd, nodes) => {
+ // Check min/maxOccurs validity while iterating over childNodes
+ val min0 = erd.minOccurs
+ val max0 = erd.maxOccurs
+ val isUnbounded = max0 == -1
+
+ // Flatten multiple DIArrays into the first one
+ if (erd.isArray) {
+ nodes.reduceLeft((a, b) =>
a.asInstanceOf[DIArray].concat(b.asInstanceOf[DIArray]))
+ nodes.reduceToSize(1)
+ }
+
+ val occurrence = if (erd.isArray) nodes(0).contents.length else
nodes.length
+ if (isUnbounded && occurrence < min0)
+ ret = ret :+ erd
+ else if (!isUnbounded && (occurrence < min0 || occurrence > max0))
+ ret = ret :+ erd
+ }
+ }
+
+ val (erds, nodes) = groups.unzip
+ childNodes.clear()
+ nodes.foreach( childNodes ++= _ )
+ if (ret.isEmpty)
+ Maybe.Nope
+ else
+ Maybe(ret)
+ }
+
override def addChild(e: InfosetElement): Unit = {
if (!e.isHidden && !hasVisibleChildren) hasVisibleChildren = true
if (e.runtimeData.isArray) {
Review comment:
I feel like there's an edge case where that wouldn't work. Imagine we parse
an array, and then parse a scalar. Now say we try to parse another of the same
array. WIth your change that will contact to the previous array. But then if
that array element fails for some reason we need to be able to remove that
array we concated. But I *think* the way we backtrack arrays is by looking the
last element in the infoset and reducing the array size. But the array is not
the last in the childNodes seq now, since the array we added is some previous
array.
----------------------------------------------------------------
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
