getmata_curl.php

<?php
// ---------------------------------------------------------------
// ----------------- file: getmata_curl.php ----------------------
// ---------------------------------------------------------------

header("Content-Type: text/html; charset=utf-8");

error_reporting(E_ALL);
ini_set("display_errors", 1);

## $myscript = 'getmata_curl.php';
$myscript = $_SERVER['PHP_SELF'];


// ------------------------------------------------------------

print "<br />\n";
print "<br /> --- <a href=\"".$myscript."\">HOME</a> (RESET) --- \n";
print "<br />\n";

// ----------------------------------------------------

if (!empty($_REQUEST['url'])) {
    $go_url = trim(chop($_REQUEST['url']));
    $go_url = str_replace('http://http://','http://',$go_url);
    if (strpos($go_url,'http://') === FALSE && strpos($go_url,'https://') === FALSE) { $go_url = 'http://'.$go_url; }
    // ----- to-do: filters and/or escape url string -----
}
else {
    $go_url = 'http://';
}

// ----------------------------------------------------

function pingSite($url, $length=4096) {

    $fp = curl_init($url);

    curl_setopt($fp, CURLOPT_TIMEOUT, 10);
    curl_setopt($fp, CURLOPT_FAILONERROR, 1);
    curl_setopt($fp, CURLOPT_RANGE, '0-4096');
    curl_setopt($fp, CURLOPT_RETURNTRANSFER, 1);
    curl_setopt($fp, CURLOPT_USERAGENT, 'Mozilla/4.0 (compatible; MSIE 5.01; Windows NT 5.0)');

    $result = curl_exec($fp);

##  if (curl_errno($fp) != 0) { $status = false; }
    if (curl_errno($fp) != 0) { $status = curl_error($fp); }
    else { $status = true; }

    curl_close($fp);

    return array('status' => $status, 'html' => $result);

}

// ----------------------------------------------------


$online = pingSite($go_url);

if ( $online['status'] !== true && !empty($go_url) && $go_url != 'http://' ) {

print "<br />\n";
print "<br /> ........ Die URL <span style=\"margin:8px; padding:4px; background:#DDEEDD;\">".$go_url."</span> ist <strong style=\"color:red;\">nicht</strong> online !!! ... <a href=\"".$myscript."\">zur&uuml;ck</a>\n";
print "<br />\n";
print "<br />ERROR: <span style=\"margin:8px; padding:4px; background:#DEDEDE\">".$online['status']."</span>\n";
print "<br />\n";
exit;
}
else {


$hsize = strlen($online['html']);

print "<br />\n";
print "<br />Size: ".$hsize."\n";
print "<br />\n";


// ------------------------------------------------------------

if ( !empty($_REQUEST['save']) && !empty($go_url) && $go_url != 'http://' ) {

print "<br />\n";
print "<br /> To-Do ........ Speichern \n";
print "<br />\n";
print "<br /> ###### DEBUG ###### _POST ###### \n";
print "<br />\n";
print "<br />\n";

print "<pre style=\"margin:4px; padding:4px; background:#DEDEDE; text-align:left;\">\n";
print_r($_POST);
print "</pre>\n";

print "<br />\n";
print "<br />\n";
print "<br />\n";

}
else {
// ------------------------------------------------------------

$formular1 = '
<form name="form1" method="post" action="">
  <br /> URL: <input type="text" name="url" size="48" maxlength="180" value="'.$go_url.'" />
  <br />
  <br /> RUN: <input type="submit" name="meta" value="Get MetaTags" /> &nbsp;&nbsp; INFO: <span id="minfo" style="margin:12px; padding:4px; background:#DDEEDD;">Bitte eine URL eingeben.</span>
  <br />
  <br /> Titel: <input type="text" name="title" size="48" maxlength="90" />
  <br />
  <br /> Beschreibung:
  <br /> <textarea name="description" cols="48" rows="4" wrap="VIRTUAL"></textarea>
  <br />
  <br /> Keywords:
  <br /> <textarea name="keywords" cols="48" rows="4" wrap="VIRTUAL"></textarea>
  <br />
  <br /> OK:  <input type="submit" name="save" value="Speichern" />
  <br />
</form>
';


// ----------------------------------------------------

print "<br />\n";
print "<br />\n";

print $formular1;

print "<br />\n";
print "<br />\n";

}


// ----------------------------------------------------


function str_utf8_encode($str) {
    if (mb_detect_encoding($str, 'UTF-8', true) === FALSE) {
    $str = utf8_encode($str);
    }
    return $str;
}

// ----------------------------------------------------------------
// ----------------------------------------------------------------

## function get_meta_from_url($url) {

function get_meta_from_url($html) {

/* ---------------
$html = '';

if($fh = @fopen($url,"r")){
    while (!feof($fh)){
        $buffer = fgets($fh, 4096);
        if (strpos($buffer,'/head') !== FALSE) { break; }
        $html .= $buffer;
    }
    fclose($fh);
}
else {
print "<br />ERROR ... fopen\n";
print "<br /> ........ Die URL <span style=\"margin:8px; padding:4px; background:#DDEEDD;\">".$go_url."</span> ist <strong style=\"color:red;\">nicht</strong> online !!! ... <a href=\"".$myscript."\">zur&uuml;ck</a>\n";
exit;
}
--------------- */


// --------------------------------------------

$pattern1 = '/<title>(.*)<\/title>/iU';
preg_match_all ($pattern1, $html, $matches_title);

## print "<pre>\n";
## print_r($matches_title);
## print "</pre>\n";

// --------------------------------------------

$pattern2 = '/<meta name=(["\']?)(description|keywords)\\1\\s+content=(["\']?)(.+?)\\3[ \/>]/i';
preg_match_all ($pattern2, $html, $matches_meta, PREG_SET_ORDER);

## print "<pre>\n";
## print_r($matches_meta);
## print "</pre>\n";

// --------------------------------------------

$matchdata = array();

$matchdata['title'] = (isset($matches_title[1][0])) ? str_utf8_encode($matches_title[1][0]) : '';

foreach($matches_meta as $matchitem) {
    $mkey = strtolower($matchitem[2]);
    $matchdata[$mkey] = (isset($matchitem[4])) ? str_utf8_encode($matchitem[4]) : '';
}

// --------------------------------------------

return $matchdata;


}

// ----------------------------------------------------------------
// ----------------------------------------------------------------


// ----------------------------------------------------
if ( !empty($_REQUEST['meta']) ) {


// ---------------------------------
if ( !empty($go_url) && $go_url != 'http://' ) {

## $data = get_meta_from_url($go_url);

$data = get_meta_from_url($online['html']);



print "<p>###### DEBUG ######</p>\n";
print "<br />URL: ".$go_url."\n";
print "<br />\n";

print "<pre style=\"margin:4px; padding:4px; background:#DEDEDE; text-align:left;\">\n";
print_r($data);
print "</pre>\n";

print "<br />\n";
print "<br />\n";


$jscode = '';

$jscode .= '
<script type="text/javascript">
';

$jscode .= '
document.form1.title.value = "'.$data['title'].'";
document.form1.description.value = "'.$data['description'].'";
document.form1.keywords.value = "'.$data['keywords'].'";
';

$jscode .= '
document.getElementById("minfo").innerHTML = "OK ... MetaTags wurden gelesen.";
';

$jscode .= '
</script>
';


print $jscode;


}
else {
// ---------------------------------

print '
<script type="text/javascript">
document.getElementById("minfo").innerHTML = "Kannst Du nicht Lesen? ... zuerst eine URL eintragen !!!";
</script>
';

}
// ---------------------------------

}
// ----------------------------------------------------

}
// ------------------------------------------------------------


?>