counts number of nodes in tree, logging a warn is the number of child nodes exceeds a certain threshold

childCountWarning.groovy

import java.util.concurrent.atomic.AtomicInteger
import org.apache.jackrabbit.oak.api.Type
import org.apache.jackrabbit.oak.plugins.segment.SegmentBlob
import org.apache.jackrabbit.oak.spi.state.NodeState

def countNodes(NodeState n, String path = "/", Integer flush = 1000000, Long warnAt = 1000, AtomicInteger count = new AtomicInteger(0), AtomicInteger binaries = new AtomicInteger(0), root = true) {
  if(root) {
    println "Counting nodes in tree ${path}"
  }

  cnt = count.incrementAndGet()
  if (cnt % flush == 0) println("  " + cnt)

  try {
    for(prop in n.getProperties()) {
      if(prop.getType() == Type.BINARY || prop.getType() == Type.BINARIES) {
        for(b in prop.getValue(Type.BINARIES)) {
          binaries.incrementAndGet()
          if(b instanceof SegmentBlob) {
            if(!((SegmentBlob)b).isExternal()) {
              b.length()
            }
          } else {
            b.length()
          }
        }
      }
    }
    
    kids = n.getChildNodeCount(warnAt)
    if(kids >= warnAt) {
      println "${path} has ${kids} child nodes"
    }

    for(child in n.getChildNodeEntries()) {
      countNodes(child.getNodeState(), path + child.getName() + "/", flush, warnAt, count, binaries, false)
    }
  } catch(e) {
    println "warning unable to read node ${path}"
  }

  if(root) {
    println "Total nodes in tree ${path}: ${cnt}"
    println "Total binaries in tree ${path}: ${binaries.get()}"
  }

  return cnt
}

My notes: I tried editing this to only look in a specific path, instead of root, but the script was giving me errors (seemed to be thwarted by s3-stored blob paths) so I think I gave up and quit it before finishing.

##my test was to run it only on /content/data
java -Xmx8G -jar oak-run-1.22.11.jar console repository/segmentstore/ :load childCountWarning-content-data.groovy >> countNodes-contentdata.log 2>> countNodes-contentdata-errors.log