analyse-attribute-usage.php

gistfile1.php

<?php

require_once __DIR__ . '/vendor/autoload.php';

function scan(string $path): array
{
    $o_dir = new RecursiveDirectoryIterator($path);
    $o_iter = new RecursiveIteratorIterator($o_dir);

    $files = [];

    foreach ($o_iter as $o_info) {
        if ($o_info->getExtension() === 'php') {
            $files[] = $o_info->getPathname();
        }
    }

    return $files;
}

function getPerPackageCsv()
{
    $perPackageCsv = fopen(__DIR__ . '/attributes-per-package.csv', 'w');

    fputcsv($perPackageCsv, [
        'package',
        'count',
    ]);

    return $perPackageCsv;
}

function getPerPackagePerAttributeCsv()
{
    $perPackagePerAttributeCsv = fopen(__DIR__ . '/attributes-per-package-per-attribute.csv', 'w');

    fputcsv($perPackagePerAttributeCsv, [
        'package',
        'attribute',
        'count',
    ]);

    return $perPackagePerAttributeCsv;
}

function getPerFileCsv()
{
    $perFileCsv = fopen(__DIR__ . '/attributes-per-file.csv', 'w');

    fputcsv($perFileCsv, [
        'vendor',
        'package',
        'path',
        'count',
    ]);

    return $perFileCsv;
}

function getPerAttributeCsv()
{
    $packageCsv = fopen(__DIR__ . '/attributes.csv', 'w');

    fputcsv($packageCsv, [
        'attribute',
        'count',
    ]);

    return $packageCsv;
}

$packages = glob(__DIR__ . '/sources/**/**');
$totalCount = 0;
$perFileCsv = getPerFileCsv();
$perAttributeCsv = getPerAttributeCsv();
$perPackageCsv = getPerPackageCsv();
$perPackagePerAttributeCsv = getPerPackagePerAttributeCsv();
$packageCount = count($packages);
$perAttributeCount = [];

foreach ($packages as $i => $packagePath) {
    $perPackagePerAttributeCount = [];

    $memory = memory_get_usage() / 1_000_000;

    echo "[{$i}/{$packageCount}] Scanning {$packagePath}\n";
    echo "{$memory}MB\n";

    $packageName = explode('/', $packagePath);

    $lastIndex = array_key_last($packageName);

    [$vendor, $package] = [$packageName[$lastIndex - 1], $packageName[$lastIndex]];

    if ($vendor === 'jetbrains') {
        continue;
    }

    $phpFiles = scan($packagePath);

    $countForPackage = 0;

    foreach ($phpFiles as $phpFile) {
        $contents = file_get_contents($phpFile);

        preg_match_all('/(\s|^)#\[([\\\\\w\s,]+)/', $contents, $matches);

        $attributesInFile = ($matches[2] ?? []);

        $countForPackage += count($attributesInFile);

        $row = [
            'vendor' => $vendor,
            'package' => $package,
            'path' => $phpFile,
            'count' => count($attributesInFile),
        ];

        foreach ($attributesInFile as $attribute) {
            $perAttributeCount[$attribute] ??= 0;
            $perAttributeCount[$attribute]++;
            $perPackagePerAttributeCount[$attribute] ??= 0;
            $perPackagePerAttributeCount[$attribute]++;
        }

        fputcsv($perFileCsv, $row);
    }

    fputcsv($perPackageCsv, [
        'package' => "{$vendor}/{$package}",
        'count' => $countForPackage,
    ]);

    foreach ($perPackagePerAttributeCount as $attribute => $count) {
        fputcsv($perPackagePerAttributeCsv, [
            'package' => "{$vendor}/{$package}",
            'attribute' => $attribute,
            'count' => $count,
        ]);
    }

    echo "{$vendor}/{$package}: {$countForPackage}\n";
    echo "\n";

    $totalCount += count($phpFiles);
}

foreach ($perAttributeCount as $attribute => $count) {
    fputcsv($perAttributeCsv, [
        'attribute' => $attribute,
        'count' => $count,
    ]);
}

fclose($perFileCsv);
fclose($perAttributeCsv);

echo "Done\n";