Quick stats from a git diff file

stat-from-git-diff.php

<?php

// STEP 1: Generate Diff:
// - git diff <old-hash> <newer-hash> > my.diff

// STEP 2: Change the path to the file here:
$filePath = "my.diff";

// STEP 3: List extensions you would like to ignore when doing your stat.
$ignoredExtensions = ['csv', 'txt', 'jar', '.properties'];

// STEP 4: Try to guess a proportion of comment/non logical code (50 for 50%)
$commentProportion = 80;

/**
 * *****************************************************
 * Don't change the code beyond this point
 * *****************************************************
 */

$linesPerDay = 100;

$handle = fopen(__DIR__ . "/" . $filePath, "r");

$stats = [
    'nbFiles' => 0,
    'nbNewFiles' => 0,
    'nbDeletedFiles' => 0,
    'added' => 0,
    'removed' => 0,
    'changed' => 0
];

$currentFileStat = getCurrentStat("", $ignoredExtensions);
while (($line = fgets($handle)) !== false) {

    if (strpos($line, "diff ") === 0) {
        $data = [];
        $result = preg_match("/^diff --git a\/(?'name'.*) .*/", $line, $data);

        $stats['nbFiles']++;
        $diff = $currentFileStat["added"] - $currentFileStat['removed'];
        if ($diff > 0) {
            $stats['added'] += $diff;
        } elseif ($diff < 0) {
            $stats['removed'] += -1 * $diff;
        }
        $stats['changed'] += $currentFileStat["added"] + $currentFileStat['removed'];


        $currentFileStat = getCurrentStat($data['name'], $ignoredExtensions);
        continue;
    } elseif (strpos($line, "+++ ") === 0 || strpos($line, "--- ") === 0) {
        continue;
    }

    if (!$currentFileStat['ignored']) {
        if (strpos($line, "+") === 0) {
            $currentFileStat['added']++;
        } elseif (strpos($line, "-") === 0) {
            $currentFileStat['removed']++;
        } elseif (strpos($line, "deleted file mode") === 0) {
            $stats['nbDeletedFiles']++;
        } elseif (strpos($line, "new file mode") === 0) {
            $stats['nbNewFiles']++;
        }
    }
}

echo "== FILES ==\n";
echo "Number of files concerned: " . $stats['nbFiles'] . "\n";
echo "Number of new files: " . $stats['nbNewFiles'] . "\n";
echo "Number of deleted files: " . $stats['nbDeletedFiles'] . "\n";

echo "\n== LINES OF CODE ==\n";
echo "Number of lines added: " . $stats['added'] . "\n";
echo "Number of lines removed: " . $stats['removed'] . "\n";
echo "Number of lines changed: " . $stats['changed'] . " (contains added & removed lines as well)\n";

echo "\n== TIME ESTIMATION ==\n";
echo "Estimated time spent : " . floor($stats['changed']/$linesPerDay/($commentProportion/100)) . " days \n";

echo "\n== Information ==\n";
echo "Diff file: $filePath\n";
echo "Percent of non logical code used for estimation: " . $commentProportion . "%\n";
echo "Ignored extensions: " . implode(", ", $ignoredExtensions) . "\n\n";

fclose($handle);


function getCurrentStat($fileName, $ignoredExtensions)
{
    $info = pathinfo($fileName);
    $extension = "";
    if (isset($info['extension'])) {
        $extension = $info['extension'];
    }

    return [
        'fileName' => $fileName,
        'ignored' => in_array($extension, $ignoredExtensions),
        'added' => 0,
        'removed' => 0,
    ];
}