query-string-usage-report

query-string-usage-report.php

<?php

$handle = fopen("ssl_access_log", "r");
$i = 0;

$reportParamCount = [];
$reportParamParts = [];

$giveFeedbackEvery = 1000;

while (!feof($handle)) {
    $buffer = fgets($handle, 4096);
    $parts = explode(" ", $buffer);
    // Only GET requests
    if ($parts[5] !== "\"GET") {
        continue;
    }

    // Skip URLs that match...
    if (preg_match('/\.js|\.css|\.jpeg|\.jpg|\.svg|\.png|\.gif|status\.php/', $parts[6])) {
        continue;
    }

    $i++;
    $pos = strpos($parts[6], '?');
    // There was no query string...
    if ($pos === false) {
        $reportParamCount[0]++;
        continue;
    }
    $queryString = substr($parts[6], $pos + 1);

    $params = explode('&', $queryString);
    $reportParamCount[count($params)]++;

    foreach ($params as $param) {
        $paramParts = explode('=', $param);
        $reportParamParts[$paramParts[0]]['usageCount']++;
        $reportParamParts[$paramParts[0]]['values'][$paramParts[1]]++;
    }

    if ($i % $giveFeedbackEvery === 0) {
        echo 'Analyzed ' . $i . ' lines' . PHP_EOL;
    }
}

fclose($handle);
ksort($reportParamCount);
uasort($reportParamParts, function ($a, $b) {
    if ($a['usageCount'] === $b['usageCount']) {
        return 0;
    }
    return $a['usageCount'] > $b['usageCount'] ? -1 : 1;
});

echo '-----------------' . PHP_EOL;
echo 'Report' . PHP_EOL;
echo '-----------------' . PHP_EOL;

echo 'Param Count Report'. PHP_EOL;
foreach ($reportParamCount as $key => $val) {
    echo $key . ': ' . $val . PHP_EOL;
}
echo 'Param Usage Report' . PHP_EOL;
foreach ($reportParamParts as $key => $val) {
    echo '> ' . $key . PHP_EOL;
    echo '>>> Used ' . $val['usageCount'] . ' times' . PHP_EOL;
    echo '>>> ' . count($val['values']) . ' unique values' . PHP_EOL;
}