Script to clean up unpublished broken links

s = ctx.getJCRSession("foflinks");
println ("at root")

saveCount = 0
// batch of nodes to be processed in next run
nextBatch = []
// batch of nodes to be processed in current run
currentBatch = []
// a node to start with
currentBatch.add("/")
// batch of nodes with SO error ... to be deleted once some of their children had been deleted and their evaluation subpath had been made less deep
stackOverflowBatch = []

println ("cleanup starting")
while (currentBatch.size() > 0) {
    cleanLinks(currentBatch);
    // prepare next batch for processing
    currentBatch.clear()
    currentBatch.addAll(nextBatch)
    nextBatch.clear()
}
println ("cleaning stack overflow nodes")
cleanLinks(stactOverflowBatch.reverse())
println ("cleanup finished")


def cleanLinks(list) {
    println ("  about to start new batch with ${list.size()} elements")
    // parent loop
    while (list.size() > 0) {
        parent = s.getNode(list.remove(0))
        children = parent.getNodes().iterator()
        println("   checking ${parent.path} child nodes")
        // child nodes loop
        while (children.hasNext()) {
            n = children.next()
            if (n.name.startsWith("jcr:")) {
                // ignore system nodes
                continue;
            }
            // delete all unpublished children
            if (!n.hasProperty("mgnl:activationStatus") || !n.getProperty("mgnl:activationStatus").getBoolean()) {
                println("     ${n.path} not published, will remove")

                try {
                    n.remove();
                } catch (StackOverflowError e) {
                    println("     failed to remove ${n.path}, will try again later")

                    // try again once child nodes had been deleted
                    stackOverflowBatch.add(n.path)
                    // mark child nodes for next round of deletion
                    nextBatch.add(n.path)
                }
                saveCount++
            } else {
                    println("     will check child nodes of ${n.path} in next round")

                // published, need to evaluate it's children, but due to potentially very big depth of the nodes we can't run it recursively
                nextBatch.add(n.path)
            }
        
            if (saveCount % 500 == 0) {
                println("      save()");
                // save only every x nodos to balance amount of memory used for modified nodes vs speed of running due to time necessary for each save
                s.save()
            }
        }
    }
}