Blizzard user data scrapper like a thread from PHP CLI

blizzard-scrapper.php

<?php

/**
 * database.php
 *
 * Database access
 *
 * @category   Data scrapper
 * @package    Blizzard
 * @author     erm3nda
 * @website    http://erm3nda.github.io
 * @copyright  idk
 * @license    Persona natural de liza
 * @version    0.0.1
 * @see        pthreads
 * @link       thebot.net/threads/battle-net-name-number-scraper.368348/#post-4088269
 * @note       This is NOT the right/best way to do that, it's the old style to make a bunch of processes. But works.
 * @platform   Linux
 */

$results_file = "results.txt"; // U can use anything here /tmp/8a9sf8asdf.txt or whatever.

header("Content-type: text/plain, charset=iso-8859-1");
header("Content-Disposition: attachment; filename=$results_file");

function getCurrentSize(){
    $linecount = 0;
    $handle = fopen($file, "r");
    while(!feof($handle)){
        $line = fgets($handle);
        $linecount++;
    }
    fclose($handle);
    return $linecount;
}
    
function getCurrentLines(){
    $lines = count(split(file_get_contents($results_file)));
    return $lines;
}

// this function does check for PID or lock files
function processExists($pid) {
    $return = exec("ps -A | grep $pid");
    if (preg_match("/$pid/", $return)) {
        return $pid;
    }
}

function removeItem($array, $element) {
    return array_diff($array, [$element]);
}

# da loop
if (@$argv[1]) {
    $html2 = file_get_contents('http://us.battle.net'.$argv[1]);
    $doc2 = new DOMDocument();
    if(!empty($html2))
    {
        $doc2->loadHTML($html2);
        libxml_clear_errors();
        $xpath2 = new DOMXPath($doc2);
        $row2 = $xpath2->query('//a[@class="context-link d3-class-1"]/@href');
        if($row2->length > 0)
        {
            foreach($row2 as $row)
            {
                $whole_string = $row->nodeValue;
                $cut_string = explode('/', $whole_string);
                $main_info = $cut_string[4];
                $conv_info = iconv("UTF-8//", "ISO-8859-1//IGNORE", $main_info );
                $main_infos = explode('-', $conv_info);
                // save only if it's not currently present
                if(!preg_match("/$main_infos[0]:$main_infos[1]/", file_get_contents($results_file))){
                    file_put_contents($results_file, $main_infos[0] . ":" .  $main_infos[1] . "\n", FILE_APPEND);
                } else {
                    echo "Skipping already saved result: $main_infos[0]:$main_infos[1]";
                }
                
            }
        }
    }
} else {
    //~ flush();
    echo "Running Blizz ... ";
    $array = array();
    $html1 = file_get_contents('http://us.battle.net/d3/en/forum/3354739/');
    $doc1 = new DOMDocument();
    libxml_use_internal_errors(TRUE);
    if(!empty($html1))
    {
        $doc1->loadHTML($html1);
        libxml_clear_errors();
        $xpath1 = new DOMXPath($doc1);
        $row1 = $xpath1->query('//a[@class="topic-title"]/@href');
        echo "found " . $row1->length . " links" . PHP_EOL;
        if($row1->length > 0)
        $requests = $row1->length;
        {
            echo "Spawning a total of $requests processes, wait a few ..." . PHP_EOL;
            foreach($row1 as $row)
            {   
                $pid[] = shell_exec("php ".  __file__ . " $row->nodeValue > /dev/null 2>&1 & echo $!;"); // run that in a "threaded way"
                $requests = $requests - 1;
                usleep(100);
            }
            // wait for spawned processes
            foreach ($pid as $p)
            {
                if (!processExists($p)){
                    removeItem($pid, $p); // remove finished PIDs from current list
                    usleep(100);
                }
            }
        }
    }
    while (count($pid) == 0){
        echo "Processes did finished, continue ...";
    }
    sleep(4); // wait for disk writes
    $content = array_unique(split("\n", file_get_contents($results_file)));
    echo "Found " . count($content) . " of unique results:" . PHP_EOL;
    file_put_contents($results_file, implode("\n", $content));
    exit();
}
?>