TLSV形式のアクセスログを解析する

analyze_access_log.php

<?php

$filePath = "access.log";
// 閾値以下のリクエストは無視する
$countThreshold = 100;
// count | sum_time | avg | req
$sortKey = "count";
// クエリストリング以降を合算する
$sumQueryString = true;

if (isset($argv[1])) {
    $filePath = $argv[1];
}

if (file_exists($filePath)) {
    if ($fp = fopen($filePath, 'r')) {
        $logs = [];
        while ($line = fgets($fp)) {
            $items = explode("\t", $line);
            $kvs = [];
            foreach ($items as $item) {
                $kv = explode(":", $item);
                $key = $kv[0];
                $value = $kv[1];

                $kvs[$key] = $value;
            }

            // URLだけ取り出す
            $reqStrings = explode(' ', $kvs['req']);
            if (count($reqStrings) < 2) {
                // たまに、不正フォーマットがある
                continue;
            }
            $req = $reqStrings[1];
            if ($sumQueryString && strpos($req, '?') !== false) {
                $req = explode('?', $req)[0];
            }

            if (!isset($logs[$req])) {
                $logs[$req] = [
                    'count' => 0,
                    'sum_time' => 0
                ];
            }
            $logs[$req]['count']++;
            $logs[$req]['sum_time'] += $kvs['taken_sec'];
        }
        $rows = [];
        foreach ($logs as $reqLog => $log) {
            $count = $log['count'];
            $sumTime = $log['sum_time'];
            $avgTime = $log['sum_time'] / $count;
            $row = [
                'req' => $reqLog,
                'sum_time' => $sumTime,
                'count' => $count,
                'avg' => $avgTime
            ];
            $rows[] = $row;
        }

        // ソート
        usort($rows, function ($a, $b) use ($sortKey) {
            return $b[$sortKey] - $a[$sortKey];
        });

        foreach ($rows as $row) {
            if ($row['count'] < $countThreshold) {
                continue;
            }
            echo "req:" . $row['req'] . " ";
            echo "sum_time:" . $row['sum_time'] . " ";
            echo "count:" . $row['count'] . " ";
            echo "avg::" . $row['avg'] . " ";
            echo "\n";
        }
    } else {
        echo "can not open file";
    }
} else {
    echo "file does not exists";
}