LTSV集計

gistfile1.groovy

// Groovy Version: 2.2.1 JVM: 1.7.0_51 Vendor: Oracle Corporation OS: Mac OS X
// 実行時に引数としてログファイル名を絶対パスで指定すること

// 特定の間隔でアクセス数を算出


@Grab('org.apache.commons:commons-lang3:3.2.1')
import org.apache.commons.lang3.time.FastDateFormat

@Grab('am.ik.ltsv4j:ltsv4j:0.9.0')
import am.ik.ltsv4j.LTSV

def logFile = new File(args[0])
def map = [:].withDefault { 0 }

def interval = 5

println "STARTED: $logFile"
logFile.eachLine { row ->
    def parsed = LTSV.parser().parseLine(row)
    def date = FastDateFormat.getInstance('[dd/MMM/yyyy:HH:mm:ss Z]', Locale.US).parse(parsed['time'])
    def day = FastDateFormat.getInstance('dd').format(date)
    def hour = FastDateFormat.getInstance('HH').format(date)
    def min = FastDateFormat.getInstance('mm').format(date)
    def responseTime = parsed['reqtime_microsec'].toBigDecimal()
    def req = parsed['req']

    // 集計対象時間を指定
    if (req =~ /\.php/) {
        if ('01'.equals(day)) {
//        if (16 <= hour.toInteger() && hour.toInteger() <= 23) {
            // mapのキー名作成
            def minutes = sprintf '%02d', (min.toInteger() / interval).toInteger() * interval
            def key = "${hour}_${minutes}"

            m = map[key]
            if (m) {
                m['rt'] += responseTime
                m['cnt']++
            } else {
                map[key] = [:].withDefault { 0 }
                map[key]['cnt']++
                map[key]['rt'] = responseTime
            }
//        }
        }
    }
}

println "FINISHED: $logFile"
println map.each {
    def average = it.value['rt'] / it.value['cnt']
    println "${it.key},${average}"
}