PHP Self Stats

self_stats.php

<?php

// PHP Self Stats

/* 
 * get_loaded_extensions() returns all the extension names 
 * currently loaded.
 *
 * Using array_intersect(), we can get only the extensions
 * we want from those which are really available.
 */
$extensions = array_intersect(get_loaded_extensions(), array(
    'Core', 
    'ctype', 
    'date', 
    'filter', 
    'hash', 
    'iconv', 
    'json', 
    'SPL', 
    'pcre', 
    'Reflection',
    'session',
    'standard',
    'pdo',
    'bcmath',
    'mbstring',
    'gettext',
    'mcrypt',
    'sockets',
    'tokenizer',
    'intl',
    'fileinfo',
    'Phar',
    'apc'
));
/* 
 * get_extension_funcs() returns all functions from an extension.
 *
 * array_map() will pass the function get_extension_funcs() on
 * all extension names. (get_extension_funcs('Core'), 
 * get_extension_funcs('ctype'), etc).
 *
 * After this, we'll have a multi-dimensional array of 
 * functions by extension. Similar to this:
 * array(
 *     'Core' => array(...),
 *     'ctype' => array(...),
 *     ...
 * )
 *
 * In order to merge these groups, we need to
 * array_merge($core, $ctype, $etc), but no so fast!
 *
 * call_user_func_array() lets you pass an array as individual
 * arguments to a function.
 *
 * function sum($a, $b) { return $a + $b; }
 * call_user_func_array('sum', array(3, 4)); // 7
 *
 * Since each extension index in the array now is an array
 * with more functions, we pass these to array_merge so 
 * we can get a full list. 
 *
 * Both call_user_func_array() and array_merge works with 
 * N parameters, not only 2 like our sum() sample.
 */
$all_functions = call_user_func_array(
    'array_merge', 
    array_map('get_extension_funcs', $extensions)
);
/*
 * This is another trick to speed up conversions. 
 * 
 * array_fill($start, $end, $value) creates an array from
 * $start to $end and fill it with $value. We're creating
 * here an array of 1s: array(1,1,1,1,1,1,etc).
 *
 * This array os 1s will have the same size of the 
 * $all_functions. So we can combine them with array_combine.
 *
 * Before we had array('strlen', 'strpos', etc...)
 * Now we have array('strlen' => 1, 'strpos' => 1);
 *
 */
$prefixes      = array_combine($all_functions, array_fill(0, count($all_functions), 1));
$prefix_sd     = 0;
$best          = $prefixes;
$prefix_size    = array_reduce(array_map('strlen', array_keys($prefixes)), 'max', 1);
do {
    $top            = count($prefixes);
    $best_prefix_sd = $prefix_sd;
    error_log("Trying to find top {$prefix_size}-char prefixes on $top names with more than $best_prefix_sd score...");
    $processed = array();
    array_walk(
        $prefixes,
        function ($usages, $prefix_name) use ($prefixes, $prefix_size, &$processed) {
            $prefix_length = strlen($prefix_name);
            if ($prefix_length > $prefix_size) {
                foreach (array_keys($prefixes) as $another_prefix) {
                    $another_length = strlen($another_prefix);
                    if ($prefix_name !== $another_prefix 
                        || $prefix_length < $another_length
                        || $another_length > $prefix_size) {
                        foreach (range(min($prefix_length, $prefix_size), $another_length) as $i) {
                            if (!substr_compare($prefix_name, $another_prefix, 0, $i)) {
                                $prefix_length = $i;
                            } else {
                                break;
                            }
                        }            
                    }
                }
                $prefix_name = substr($prefix_name, 0, $prefix_length) ?: $prefix_name;
            }
            if (!isset($processed[$prefix_name])) {
                $processed[$prefix_name] = 0;
            }
            $processed[$prefix_name] = max($processed[$prefix_name] + $usages, 1);
        }
    );
    $prefixes = array_filter($processed);
    $prefvals = array_values($prefixes);
    $sample_count = count($prefvals);
    for ($current_sample = 0; $sample_count > $current_sample; ++$current_sample) {
        $sample_square[$current_sample] = pow($prefvals[$current_sample], 2);
    }
    $prefix_sd = sqrt(
        array_sum($sample_square) / $sample_count 
        - pow((array_sum($prefvals) / $sample_count), 2)
    );
    if ($prefix_sd >= $best_prefix_sd) {
        $best = $prefixes;
        $best_prefix_sd = $prefix_sd;
    } else {
        $prefix_size = array_reduce(array_map('strlen', array_keys($prefixes)), 'max', 1);
    }
    $prefix_size--;
} while ($prefix_size >= 3);

arsort($best);
print_r(array_slice($best, 0, 10));