Simple script to analyse the standard output results of Percona Toolkit pt-duplicate-key-checker

analyse-pt-duplicate-key-checker.php

<?php

$types = [];

$fp = @fopen("/downloads/db1-duplicate-keys.txt", "r");
if ($fp) {
    $i = 0;
    $status = 0;
    while (($buffer = fgets($fp, 4096)) !== false) {

        if ($status == 0) {
            if (line_is_comment($buffer)) {
                $status = 1;
            }
            continue;
        }

        if ($status == 1) {
            if (line_is_database_with_table($buffer)) {
                $status = 2;
            }
            continue;
        }

        if ($status == 2) {
            if (line_is_comment($buffer)) {
                $status = 3;
            }
            continue;
        }

        if ($status == 3) {
            if (line_is_empty($buffer)) {
                $status = 4;
            }
            continue;
        }

        if ($status == 4) {
            $type = recognize_type(trim($buffer));

            if (isset($types[$type])) {
                $types[$type] += 1;
            } else {
                $types[$type] = 1;
            }

            $status = 3;
            continue;
        }

        $status = 0;
        $i++;
    }
    if (!feof($fp)) {
        echo "Error: unexpected fgets() fail\n";
    }
    fclose($fp);
}

echo json_encode($types, JSON_PRETTY_PRINT);

function line_is_comment(string $line): bool
{
    return str_contains($line, '# ###');
}

function line_is_empty(string $line): bool
{
    return trim($line) == '';
}

function line_is_database_with_table(string $line): bool
{
    return preg_match('/^# [a-zA-Z0-9_]+\.[a-zA-Z0-9_]+/', $line);
}

function recognize_type(string $line): string
{
    if (str_contains($line, 'is a duplicate of')) {
        $type = 'is a duplicate of';
    } elseif (str_contains($line, 'is a left-prefix of')) {
        $type = 'is a left-prefix of';
    } elseif (str_contains($line, 'is a duplicate constraint')) {
        $type = 'is a duplicate constraint';
    } else {
        $type = 'unknown';
    }
    return $type;
}

// Outputs something like
// {
//     "is a duplicate of": 82491,
//     "is a left-prefix of": 17174,
//     "is a duplicate constraint": 2
// }