Web Sitesinden E-Mail Adresini Alma

<?php

function get_web_page($url) {
    $options = array(
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_HEADER         => false,
        CURLOPT_FOLLOWLOCATION => true,
        CURLOPT_ENCODING       => "",
        CURLOPT_USERAGENT      => "Mozilla/5.0 (compatible; email-bot/1.0)",
        CURLOPT_AUTOREFERER    => true,
        CURLOPT_CONNECTTIMEOUT => 30,
        CURLOPT_TIMEOUT        => 30,
        CURLOPT_MAXREDIRS      => 10,
        CURLOPT_SSL_VERIFYPEER => false
    );
    $ch = curl_init($url);
    curl_setopt_array($ch, $options);

    try {
        $content = curl_exec($ch);
    } catch (Exception $e) {
        curl_close($ch);
        return false;
    }

    $header = curl_getinfo($ch);
    curl_close($ch);

    $header["content"] = $content;
    return $header;
}

// E-posta adresini bulmak için regex ve cfemail desteği
function getMailAdressRegex($string) {
    preg_match('/([a-zA-Z0-9._%+\-]+@[a-zA-Z0-9.\-]+\.[a-zA-Z]{2,})/', $string, $mailAdress);
    if (count($mailAdress) > 0) {
        return $mailAdress[0];
    } else {
        $cfCheck = checkCfMail($string);
        if ($cfCheck != false) {
            return cfDecodeEmail($cfCheck);
        }
        return false;
    }
}
function checkCfMail($input) {
    preg_match('/data-cfemail="(.*?)"/', $input, $mailAdressArray);
    return $mailAdressArray[1] ?? false;
}
function cfDecodeEmail($encodedString) {
    $k = hexdec(substr($encodedString, 0, 2));
    $email = '';
    for ($i = 2; $i < strlen($encodedString) - 1; $i += 2) {
        $email .= chr(hexdec(substr($encodedString, $i, 2)) ^ $k);
    }
    return $email;
}

// Href’i normalize et (mutlak/bağıl link vs)
function normalizeLink($url, $site) {
    if (strpos($url, 'http') === 0) {
        return $url;
    } else if (strpos($url, '/') === 0) {
        // domain.com/iletisim gibi
        $parsed = parse_url($site);
        $domain = $parsed['scheme'] . '://' . $parsed['host'];
        return $domain . $url;
    } else {
        // domain.com/iletisim gibi
        return rtrim($site, '/') . '/' . ltrim($url, '/');
    }
}

// İletişim/contact linkini bulup mail arar
function getContactLink($HTML, $site) {
    $contactPattern = '/iletisim|contact|bize-ulas|bizeulas|bize-yaz|bizeyaz/i';
    $dom = new DOMDocument();
    if (empty($HTML)) return false;

    @$dom->loadHTML($HTML);
    $crdx = new DOMXPath($dom);
    $linkler = $crdx->evaluate("//a");

    foreach ($linkler as $href) {
        $url = $href->getAttribute('href');
        if (preg_match($contactPattern, $url)) {
            $link = normalizeLink($url, $site);
            $contactdata = get_web_page($link);
            if ($contactdata && isset($contactdata["content"])) {
                $mailAdress = getMailAdressRegex($contactdata["content"]);
                if ($mailAdress) return $mailAdress;
            }
        }
    }
    return false;
}

// Ana sayfada e-posta bulamazsa iletişimde ara
function getMailAdressFromWebSite($website) {
    $data = get_web_page($website);

    if (!$data || !isset($data["content"])) return false;

    // HTTP meta refresh ile yönlendirme kontrolü
    preg_match('|content="\d+;url=(.*?)"|i', $data["content"], $refreshLink);
    if (count($refreshLink) > 0) {
        $data = get_web_page($refreshLink[1]);
        $website = $refreshLink[1];
    }

    $mail = getMailAdressRegex($data["content"]);
    if (!$mail) {
        $mailContact = getContactLink($data["content"], $website);
        return $mailContact ?: false;
    }
    return $mail;
}

// TEST
$mail = getMailAdressFromWebSite("https://www.tasarimrehberi.com/");
$returnArray["mail_adress"] = $mail ?: "";
echo json_encode($returnArray, JSON_UNESCAPED_UNICODE);

?>