CQ5findReplace.groovy

import groovy.json.JsonSlurper

def nodesToProcess = getNodes("admin" , "admin" , "localhost:4502" , "geometrixx")
def sites = getSites("https://mavendc-utils.s3.amazonaws.com/test/data.json")

processSite("geometrixx" , nodesToProcess, sites , false)

def getSites(json){
    new JsonSlurper().parseText(new URL(json).getText("UTF-8"))
}

def processSite(key , nodesToProcess, sites, dryRun = true){
    nodesToProcess.get(key).each{
        processNode(getNode(it.path) , sites , dryRun)
    }.grep()
}

def getNodes(usr, pwd, host, key){
    def url = """http://HOSTTOREPLACE/crx/de/query.jsp?_dc=1388628075229&_charset_=utf-8&type=xpath&stmt=%2Fjcr%3Aroot%2Fcontent%2FSITETOREPLACE%2F%2F*%5Bjcr%3Acontains(%40text%2C'what')%5D&showResults=true"""
    url = url.replace("HOSTTOREPLACE", host).replace("SITETOREPLACE",key)
    def authString = "${usr}:${pwd}".getBytes().encodeBase64().toString()
    def conn = url.toURL().openConnection()
    conn.setRequestProperty("Authorization" , "Basic ${authString}")
    new JsonSlurper().parseText(conn.content.getText("UTF-8")).results.groupBy{
        def matched = it.path =~/(\/content\/)([a-zA-Z]+)/
        matched[0][2]
    }
}

def processNode(node , mappy, dryRun){
    def wasProcessed = false
    ['text'].each{ prop ->
        if(processProperty(node , prop, mappy)) wasProcessed = true
    }
    if(!dryRun){
       println "SAVING"
       save()
    }
    wasProcessed
}

def processProperty(node , prop, mappy){
    if(!node.hasProperty(prop)) return false
    def text = node.get(prop)
    def wasProcessed = false
    mappy.each{rewrite ->
        if(text.contains(rewrite.fromurl)){
            println "Rewrite match found!"
            println node.path
            println "Rewriting ${rewrite.fromurl} to ${rewrite.tourl}\n\n"
            node.set(prop, text.replaceAll(rewrite.fromurl , rewrite.tourl))
            wasProcessed = true
        }
        wasProcessed
    }
}