mbeckerle commented on a change in pull request #437:
URL: https://github.com/apache/incubator-daffodil/pull/437#discussion_r505758074
##########
File path:
daffodil-runtime1-unparser/src/main/scala/org/apache/daffodil/processors/unparsers/ElementUnparser.scala
##########
@@ -417,39 +418,56 @@ sealed trait RegularElementUnparserStartEndStrategy
UnparseError(Nope, One(state.currentLocation), "Expected element
start event for %s, but received %s.",
erd.namedQName.toExtendedSyntax, event)
}
- val res = event.info.element
- val mCurNode = state.currentInfosetNodeMaybe
- if (mCurNode.isDefined) {
- val c = mCurNode.get.asComplex
- Assert.invariant(!c.isFinal)
- if (c.maybeIsNilled == MaybeBoolean.True) {
- // cannot add content to a nilled complex element
- UnparseError(One(erd.schemaFileLocation), Nope, "Nilled complex
element %s has content from %s",
- c.erd.namedQName.toExtendedSyntax,
- res.erd.namedQName.toExtendedSyntax)
- }
- c.addChild(res, state.tunable)
- } else {
- val doc = state.documentElement
- doc.addChild(res, state.tunable) // DIDocument, which is never a
current node, must have the child added
- doc.isFinal = true // that's the only child.
- }
- res
+ event.info.element
} else {
Assert.invariant(state.withinHiddenNest)
// Since we never get events for elements in hidden contexts, their
infoset elements
// will have never been created. This means we need to manually
create them
- val e = if (erd.isComplexType) new DIComplex(erd) else new
DISimple(erd)
- e.setHidden()
- state.currentInfosetNode.asComplex.addChild(e, state.tunable)
- e
+ val hiddenElem = if (erd.isComplexType) new DIComplex(erd) else new
DISimple(erd)
+ hiddenElem.setHidden()
+ hiddenElem
+ }
+
+ // now add this new elem to the infoset
+ val parentNodeMaybe = state.currentInfosetNodeMaybe
+ if (parentNodeMaybe.isDefined) {
+ val parentComplex = parentNodeMaybe.get.asComplex
+ Assert.invariant(!parentComplex.isFinal)
+ if (parentComplex.maybeIsNilled == MaybeBoolean.True) {
+ // cannot add content to a nilled complex element
+ UnparseError(One(erd.schemaFileLocation), Nope, "Nilled complex
element %s has content from %s",
+ parentComplex.erd.namedQName.toExtendedSyntax,
+ newElem.erd.namedQName.toExtendedSyntax)
}
+ // We are about to add a child to this complex element. Before we do
+ // that, if the last child added to this complex is a DIArray, and this
+ // new child isn't part of that array, that implies that the DIArray
+ // will have no more children added and should be marked as final, and
Review comment:
So, if I have a format which is a giant array, e.g., PCAP format, the
elements of the array are all referenced by expressions because the header of
each packet needs the length of the packet contents. So we'll never free
anything (pretty much. We might be able to discard a few header elements in
each packet, but we won't be able to discard the packet payload contents) when
parsing or unparsing PCAP? If that's the case, then this isn't really solving
people's streaming issue until this dfdlx extension property telling it to
prune arrays is implemented?
Or.... since the suspension to fill in the length of the packet in the
packet header is evaluated, at that point do we mark everything final and so
that we now know we can release it?
----------------------------------------------------------------
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]
