IfcTriangulatedFaceSet indices get treated as oids

[muren400]

Hi,

we recently observed this Exception:

12:18:16 ERROR [GeometryRunner]:                                                                                  (GeometryRunner.java:850) 
org.bimserver.BimserverDatabaseException: No class for cid 2075 (cid came from oid: 67611)
	at org.bimserver.database.Database.getEClassForOid(Database.java:460)
	at org.bimserver.database.DatabaseSession.getEClassForOid(DatabaseSession.java:2201)
	at org.bimserver.database.queries.FollowReferenceStackFrame.process(FollowReferenceStackFrame.java:63)
	at org.bimserver.database.queries.QueryObjectProvider.next(QueryObjectProvider.java:169)
	at org.bimserver.geometry.GeometryRunner.run(GeometryRunner.java:159)
	at java.base/java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:539)
	at java.base/java.util.concurrent.FutureTask.run(FutureTask.java:264)
	at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1136)
	at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:635)
	at java.base/java.lang.Thread.run(Thread.java:833)

It was thrown when importing an IFC-File containing terrain.

The terrain is represented as an IfcTriangulatedFaceSet and apparently the indices in CoordIndex are treated like oids in QueryIncludeStackFrame.process().

BIMserver/BimServer/src/org/bimserver/database/queries/QueryIncludeStackFrame.java

Lines 110 to 117 in 91e2fa1

	} else if (element instanceof HashMapVirtualObject && feature.getEAnnotation("twodimensionalarray")!=null) {
	for ( Object nestedElement :(List)((HashMapVirtualObject) element).get("List")){
	if(nestedElement instanceof Long){
	processReference((Long) nestedElement);
	// TODO process as direct feature?
	}
	}
	} else {

This doesn't seem to cause too much problems until the list of vertices (and thus the indices) is big enough to cause the above mentioned exception.

Unfortunately I can't share the mentioned IFC-File but one can observe the behavior (processReference being called for indices) with this example file.

[muren400]

A simple fix seems to be skipping all instances of ListOfELong.

HashMapVirtualObject virtualObject = (HashMapVirtualObject) element;
if(virtualObject.eClass().getName().equals("ListOfELong")) {
	return !featureIterator.hasNext();
}
					
for ( Object nestedElement :(List)virtualObject.get("List")){
	if(nestedElement instanceof Long){
		processReference((Long) nestedElement);
		// TODO process as direct feature?
	}
}

[hlg]

Another issue with the sample file appears when serializing with a non-streaming serializer - the coord indizes are missing from the triangulated faces set.

This might be somewhat related to #1109 (comment). There are various cases in the schema where positive integers in lists or two-dimensional lists are actual values, not references. It seems like those are affected in different ways by a bug in how the database queries and the object provider work.