Script to get node details and property details from an Oak based setup

nodeStats.groovy

package felixconsole

import com.google.common.base.Stopwatch
import com.google.common.collect.FluentIterable
import com.google.common.collect.TreeTraverser
import org.apache.jackrabbit.JcrConstants
import org.apache.jackrabbit.oak.api.PropertyState
import org.apache.jackrabbit.oak.api.Tree
import org.apache.jackrabbit.oak.plugins.tree.TreeFactory
import org.apache.jackrabbit.oak.spi.state.NodeState
import org.apache.jackrabbit.oak.spi.state.NodeStore
import org.slf4j.LoggerFactory

log = LoggerFactory.getLogger('script-console')

/**
 * This script dump stats related to node and property count. The files
 * are generated under crx-quickstart/logs folder with name node-stats.txt
 * and property-stats.txt
 */

def nodeStatsTxt = new File("crx-quickstart/logs/node-stats.txt")
def propertyStatsTxt = new File("crx-quickstart/logs/property-stats.txt")

NodeStore ns = getStore()
Stopwatch w = Stopwatch.createStarted()

def treeItr = getTraversor(ns.root)
def count = 0

nodeStatsTxt.withPrintWriter {PrintWriter nodeStatsWriter ->
    propertyStatsTxt.withPrintWriter {PrintWriter propsStatsWriter ->
        treeItr.each {Tree tree ->
            def noOfProps = tree.propertyCount
            def path = tree.path

            def type = "NA"
            tree.properties.each {PropertyState ps ->
                propsStatsWriter.println("${ps.name}|${ps.array}|${ps.count()}")
                if (ps.name == JcrConstants.JCR_PRIMARYTYPE){
                    type = ps.name
                }
            }
            count++
            if (count % 10000 == 0){
                print0("Traversed $count so far. Path [$path]")
            }
            nodeStatsWriter.println("$noOfProps|$path|$type")
        }
    }
}

print0("Node stats dumped to  ${nodeStatsTxt.getAbsolutePath()}")
print0("Property stats dumped to  ${propertyStatsTxt.getAbsolutePath()}")
print0("Total time taken : $w")

def print0(def msg){
    log.info(msg)
    println(msg)
}

FluentIterable<Tree> getTraversor(NodeState ns ){
    def traversor = new TreeTraverser<Tree>(){
        @Override
        Iterable<Tree> children(Tree root) {
            return root.children
        }
    }
    return traversor.preOrderTraversal(TreeFactory.createReadOnlyTree(ns))
}

def getStore(){
    osgi.getService(org.apache.sling.jcr.api.SlingRepository.class).manager.store
}