Receive an email when a new item pops at Leboncoin.fr

leboncoin.php

<?php

/* ******************************************************************************* */
/* Script to track search results on leboncoin                                     */
/* Receive an email when a new item pops                                           */
/* ******************************************************************************* */

/* ******************************************************************************* */
// - Add this cron job to check for new items every 30 minutes                     */
// */30 * * * * curl http://yoursite.com/leboncoin.php                             */

/* - Make the folder containing this script writable                               */
/* ******************************************************************************* */

/* ******************************************************************************* */
/* Add search pages here :                                                         */
/* ******************************************************************************* */

$targets = array(
    'http://www.leboncoin.fr/annonces/offres/ile_de_france/occasions/?f=a&th=1&q=lapin+nain',
    'http://www.leboncoin.fr/annonces/offres/ile_de_france/occasions/?f=a&th=1&q=lapin+belier'
);

/* ******************************************************************************* */
/* Add/Remove proxies here (mainly because leboncoin blacklisted OVH) :            */
/* ******************************************************************************* */

$proxies = array(
    '62.210.56.250:8089',
    '62.210.56.250:7808',
    '62.210.56.250:3127',
    '188.165.244.5:8080',
    '195.154.231.43:3128'
);

/* ******************************************************************************* */
/* Add your email address here :                                                   */
/* ******************************************************************************* */

define('EMAIL_ADDRESS', '[email protected]');

/* ******************************************************************************* */
/* Stop editing here (unless you know what you're doing) :                         */
/* ******************************************************************************* */

function checkProxy($proxy){
    $timeout = 5;
    $splited = explode(':',$proxy);
    if(!@fsockopen($splited[0], $splited[1], $errorNumber, $errorMessage, $timeout)){
        return false;
    }
    else{
        return true;
    }
}

function file_get_contents_curl($url) {
    global $proxies;
    $curl = curl_init();

    curl_setopt($curl, CURLOPT_AUTOREFERER, TRUE);
    curl_setopt($curl, CURLOPT_HEADER, 0);
    curl_setopt($curl, CURLOPT_RETURNTRANSFER, 1);
    curl_setopt($curl, CURLOPT_URL, $url);
    curl_setopt($curl, CURLOPT_FOLLOWLOCATION, TRUE);       
    curl_setopt($curl, CURLOPT_CONNECTTIMEOUT, 60);    

    if(count($proxies)){
        $proxiesDown = true;

        foreach($proxies as $proxy){
            if(checkProxy($proxy)){
                $proxiesDown = false;
                curl_setopt($curl, CURLOPT_PROXY, $proxy);
                break;
            }
        }
        if($proxiesDown){
            sendMail('Leboncoin — proxies are down', "Yo,\n\n Proxies to access leboncoin are down :(");
        }
        
    }

    $data = curl_exec($curl);
    curl_close($curl);

    return $data;
}

function sendMail($subject, $body){
    $to = EMAIL_ADDRESS;
    mail($to, $subject, $body);
}

function getItems($src){
    $html = file_get_contents_curl($src);
    $dom = new DOMDocument();
    @$dom->loadHTML($html);

    $result = array();

    $parent = "list-lbc";
    $a = new DOMXPath($dom);
    $items = $a->query("//*[contains(concat(' ', normalize-space(@class), ' '), ' $parent ')]/a");

    foreach ($items as $item) {
        $result[] = $item->getAttribute('href');
    }
    return $result;
}

function getKnown($file){
    $content = @file_get_contents($file);
    return explode(',', $content);
}

function detectNew($items, $known){
    foreach($items as $item){
        if(!in_array($item, $known)){
            sendMail('Leboncoin — New item', "Yo,\n\n New item yolo : \n\n ".$item);
        }
    }
}

function updateKnown($file, $items){
    $fh   = fopen($file, 'w') or die("can't open file");
    $data = implode(',', $items);
    fwrite($fh, $data);
    fclose($fh);
}


foreach($targets as $k => $url){
    $dataFile = 'known'.$k.'.txt';

    $items = getItems($url);
    $known = getKnown($dataFile);

    detectNew($items, $known);
    updateKnown($dataFile, $items);
}
?>