2010-07-23 07:41:45 +02:00
|
|
|
<?php
|
2012-07-18 21:06:38 +02:00
|
|
|
/* To-Do
|
|
|
|
|
https://developers.google.com/+/plugins/snippet/
|
|
|
|
|
|
|
|
|
|
<meta itemprop="name" content="Toller Titel">
|
|
|
|
|
<meta itemprop="description" content="Eine tolle Beschreibung">
|
|
|
|
|
<meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
|
|
|
|
|
|
|
|
|
|
<body itemscope itemtype="http://schema.org/Product">
|
|
|
|
|
<h1 itemprop="name">Shiny Trinket</h1>
|
|
|
|
|
<img itemprop="image" src="{image-url}" />
|
|
|
|
|
<p itemprop="description">Shiny trinkets are shiny.</p>
|
|
|
|
|
</body>
|
|
|
|
|
*/
|
|
|
|
|
|
2012-07-12 01:17:33 +02:00
|
|
|
if(!function_exists('deletenode')) {
|
|
|
|
|
function deletenode(&$doc, $node)
|
|
|
|
|
{
|
|
|
|
|
$xpath = new DomXPath($doc);
|
|
|
|
|
$list = $xpath->query("//".$node);
|
|
|
|
|
foreach ($list as $child)
|
|
|
|
|
$child->parentNode->removeChild($child);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-12 23:41:04 +02:00
|
|
|
function completeurl($url, $scheme) {
|
|
|
|
|
$urlarr = parse_url($url);
|
|
|
|
|
|
|
|
|
|
if (isset($urlarr["scheme"]))
|
|
|
|
|
return($url);
|
|
|
|
|
|
|
|
|
|
$schemearr = parse_url($scheme);
|
|
|
|
|
|
|
|
|
|
$complete = $schemearr["scheme"]."://".$schemearr["host"];
|
|
|
|
|
|
|
|
|
|
if ($schemearr["port"] != "")
|
|
|
|
|
$complete .= ":".$schemearr["port"];
|
|
|
|
|
|
|
|
|
|
$complete .= $urlarr["path"];
|
|
|
|
|
|
|
|
|
|
if ($urlarr["query"] != "")
|
|
|
|
|
$complete .= "?".$urlarr["query"];
|
|
|
|
|
|
|
|
|
|
if ($urlarr["fragment"] != "")
|
|
|
|
|
$complete .= "#".$urlarr["fragment"];
|
|
|
|
|
|
|
|
|
|
return($complete);
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-12 01:17:33 +02:00
|
|
|
function parseurl_getsiteinfo($url) {
|
|
|
|
|
$siteinfo = array();
|
2010-07-23 07:41:45 +02:00
|
|
|
|
2012-07-12 01:17:33 +02:00
|
|
|
$ch = curl_init();
|
|
|
|
|
curl_setopt($ch, CURLOPT_URL, $url);
|
|
|
|
|
curl_setopt($ch, CURLOPT_HEADER, 1);
|
|
|
|
|
curl_setopt($ch, CURLOPT_NOBODY, 0);
|
|
|
|
|
curl_setopt($ch, CURLOPT_TIMEOUT, 3);
|
|
|
|
|
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
|
|
|
|
|
curl_setopt($ch,CURLOPT_USERAGENT,'Opera/9.64(Windows NT 5.1; U; de) Presto/2.1.1');
|
|
|
|
|
|
|
|
|
|
$header = curl_exec($ch);
|
|
|
|
|
curl_close($ch);
|
|
|
|
|
|
2012-07-12 23:41:04 +02:00
|
|
|
// Fetch the first mentioned charset. Can be in body or header
|
|
|
|
|
if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
|
2012-07-12 01:17:33 +02:00
|
|
|
$charset = trim(array_pop($matches));
|
|
|
|
|
else
|
|
|
|
|
$charset = "utf-8";
|
|
|
|
|
|
|
|
|
|
$pos = strpos($header, "\r\n\r\n");
|
|
|
|
|
|
|
|
|
|
if ($pos)
|
|
|
|
|
$body = trim(substr($header, $pos));
|
|
|
|
|
else
|
|
|
|
|
$body = $header;
|
|
|
|
|
|
|
|
|
|
$body = mb_convert_encoding($body, "UTF-8", $charset);
|
|
|
|
|
$body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8");
|
|
|
|
|
|
|
|
|
|
$doc = new DOMDocument();
|
|
|
|
|
@$doc->loadHTML($body);
|
|
|
|
|
|
|
|
|
|
deletenode($doc, 'style');
|
|
|
|
|
deletenode($doc, 'script');
|
|
|
|
|
deletenode($doc, 'option');
|
|
|
|
|
deletenode($doc, 'h1');
|
|
|
|
|
deletenode($doc, 'h2');
|
|
|
|
|
deletenode($doc, 'h3');
|
|
|
|
|
deletenode($doc, 'h4');
|
|
|
|
|
deletenode($doc, 'h5');
|
|
|
|
|
deletenode($doc, 'h6');
|
|
|
|
|
deletenode($doc, 'ol');
|
|
|
|
|
deletenode($doc, 'ul');
|
|
|
|
|
|
|
|
|
|
$xpath = new DomXPath($doc);
|
|
|
|
|
|
2012-07-12 23:41:04 +02:00
|
|
|
//$list = $xpath->query("head/title");
|
|
|
|
|
$list = $xpath->query("//title");
|
2012-07-12 01:17:33 +02:00
|
|
|
foreach ($list as $node)
|
|
|
|
|
$siteinfo["title"] = html_entity_decode($node->nodeValue, ENT_QUOTES, "UTF-8");
|
|
|
|
|
|
2012-07-12 23:41:04 +02:00
|
|
|
//$list = $xpath->query("head/meta[@name]");
|
|
|
|
|
$list = $xpath->query("//meta[@name]");
|
2012-07-12 01:17:33 +02:00
|
|
|
foreach ($list as $node) {
|
|
|
|
|
$attr = array();
|
|
|
|
|
if ($node->attributes->length)
|
|
|
|
|
foreach ($node->attributes as $attribute)
|
|
|
|
|
$attr[$attribute->name] = $attribute->value;
|
|
|
|
|
|
|
|
|
|
$attr["content"] = html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8");
|
|
|
|
|
|
|
|
|
|
switch (strtolower($attr["name"])) {
|
|
|
|
|
case "fulltitle":
|
|
|
|
|
$siteinfo["title"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "description":
|
|
|
|
|
$siteinfo["text"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "dc.title":
|
|
|
|
|
$siteinfo["title"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "dc.description":
|
|
|
|
|
$siteinfo["text"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-12 23:41:04 +02:00
|
|
|
//$list = $xpath->query("head/meta[@property]");
|
|
|
|
|
$list = $xpath->query("//meta[@property]");
|
2012-07-12 01:17:33 +02:00
|
|
|
foreach ($list as $node) {
|
|
|
|
|
$attr = array();
|
|
|
|
|
if ($node->attributes->length)
|
|
|
|
|
foreach ($node->attributes as $attribute)
|
|
|
|
|
$attr[$attribute->name] = $attribute->value;
|
|
|
|
|
|
|
|
|
|
$attr["content"] = html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8");
|
|
|
|
|
|
|
|
|
|
switch (strtolower($attr["property"])) {
|
|
|
|
|
case "og:image":
|
|
|
|
|
$siteinfo["image"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "og:title":
|
|
|
|
|
$siteinfo["title"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "og:description":
|
|
|
|
|
$siteinfo["text"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ($siteinfo["image"] == "") {
|
|
|
|
|
$list = $xpath->query("//img[@src]");
|
|
|
|
|
foreach ($list as $node) {
|
|
|
|
|
$attr = array();
|
|
|
|
|
if ($node->attributes->length)
|
|
|
|
|
foreach ($node->attributes as $attribute)
|
|
|
|
|
$attr[$attribute->name] = $attribute->value;
|
|
|
|
|
|
2012-07-12 23:41:04 +02:00
|
|
|
$src = completeurl($attr["src"], $url);
|
|
|
|
|
$photodata = getimagesize($src);
|
2012-07-12 01:17:33 +02:00
|
|
|
|
2012-07-12 23:41:04 +02:00
|
|
|
if (($photodata[0] > 150) and ($photodata[1] > 150)) {
|
|
|
|
|
if ($photodata[0] > 300) {
|
2012-07-14 13:59:42 +02:00
|
|
|
$photodata[1] = round($photodata[1] * (300 / $photodata[0]));
|
2012-07-12 23:41:04 +02:00
|
|
|
$photodata[0] = 300;
|
|
|
|
|
}
|
|
|
|
|
if ($photodata[1] > 300) {
|
2012-07-14 13:59:42 +02:00
|
|
|
$photodata[0] = round($photodata[0] * (300 / $photodata[1]));
|
2012-07-12 23:41:04 +02:00
|
|
|
$photodata[1] = 300;
|
|
|
|
|
}
|
|
|
|
|
$siteinfo["images"][] = array("src"=>$src,
|
|
|
|
|
"width"=>$photodata[0],
|
|
|
|
|
"height"=>$photodata[1]);
|
2012-07-12 08:20:27 +02:00
|
|
|
}
|
2012-07-12 23:41:04 +02:00
|
|
|
|
2012-07-12 01:17:33 +02:00
|
|
|
}
|
2012-07-12 08:20:27 +02:00
|
|
|
} else {
|
2012-07-12 23:41:04 +02:00
|
|
|
$src = completeurl($siteinfo["image"], $url);
|
|
|
|
|
$photodata = getimagesize($src);
|
2012-07-12 08:20:27 +02:00
|
|
|
|
2012-07-12 23:41:04 +02:00
|
|
|
if (($photodata[0] > 10) and ($photodata[1] > 10))
|
|
|
|
|
$siteinfo["images"][] = array("src"=>$src,
|
|
|
|
|
"width"=>$photodata[0],
|
|
|
|
|
"height"=>$photodata[1]);
|
2012-07-12 08:20:27 +02:00
|
|
|
}
|
2012-07-12 01:17:33 +02:00
|
|
|
|
|
|
|
|
if ($siteinfo["text"] == "") {
|
|
|
|
|
$text = "";
|
|
|
|
|
|
|
|
|
|
$list = $xpath->query("//div[@class='article']");
|
|
|
|
|
foreach ($list as $node)
|
2012-07-12 23:41:04 +02:00
|
|
|
if (strlen($node->nodeValue) > 40)
|
|
|
|
|
$text .= " ".trim($node->nodeValue);
|
2012-07-12 01:17:33 +02:00
|
|
|
|
|
|
|
|
if ($text == "") {
|
|
|
|
|
$list = $xpath->query("//div[@class='content']");
|
|
|
|
|
foreach ($list as $node)
|
2012-07-12 23:41:04 +02:00
|
|
|
if (strlen($node->nodeValue) > 40)
|
|
|
|
|
$text .= " ".trim($node->nodeValue);
|
2012-07-12 01:17:33 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// If none text was found then take the paragraph content
|
|
|
|
|
if ($text == "") {
|
|
|
|
|
$list = $xpath->query("//p");
|
|
|
|
|
foreach ($list as $node)
|
2012-07-12 23:41:04 +02:00
|
|
|
if (strlen($node->nodeValue) > 40)
|
|
|
|
|
$text .= " ".trim($node->nodeValue);
|
2012-07-12 01:17:33 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ($text != "") {
|
|
|
|
|
$text = trim(str_replace(array("\n", "\r"), array(" ", " "), $text));
|
|
|
|
|
|
|
|
|
|
while (strpos($text, " "))
|
|
|
|
|
$text = trim(str_replace(" ", " ", $text));
|
|
|
|
|
|
|
|
|
|
$siteinfo["text"] = html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...';
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return($siteinfo);
|
|
|
|
|
}
|
2010-12-21 04:38:34 +01:00
|
|
|
|
2011-09-21 01:31:45 +02:00
|
|
|
function arr_add_hashes(&$item,$k) {
|
|
|
|
|
$item = '#' . $item;
|
|
|
|
|
}
|
|
|
|
|
|
2010-07-23 07:41:45 +02:00
|
|
|
function parse_url_content(&$a) {
|
2010-07-23 08:17:41 +02:00
|
|
|
|
2011-09-20 07:21:55 +02:00
|
|
|
$text = null;
|
2011-09-21 01:31:45 +02:00
|
|
|
$str_tags = '';
|
2011-09-20 07:21:55 +02:00
|
|
|
|
2012-02-07 09:13:16 +01:00
|
|
|
$textmode = false;
|
|
|
|
|
if(local_user() && intval(get_pconfig(local_user(),'system','plaintext')))
|
|
|
|
|
$textmode = true;
|
|
|
|
|
|
2012-07-12 08:20:27 +02:00
|
|
|
//if($textmode)
|
|
|
|
|
$br = (($textmode) ? "\n" : '<br />');
|
2012-02-07 09:13:16 +01:00
|
|
|
|
2011-09-20 07:21:55 +02:00
|
|
|
if(x($_GET,'binurl'))
|
|
|
|
|
$url = trim(hex2bin($_GET['binurl']));
|
|
|
|
|
else
|
|
|
|
|
$url = trim($_GET['url']);
|
|
|
|
|
|
|
|
|
|
if($_GET['title'])
|
|
|
|
|
$title = strip_tags(trim($_GET['title']));
|
2011-04-10 12:36:12 +02:00
|
|
|
|
2011-09-21 01:31:45 +02:00
|
|
|
if($_GET['description'])
|
|
|
|
|
$text = strip_tags(trim($_GET['description']));
|
|
|
|
|
|
|
|
|
|
if($_GET['tags']) {
|
|
|
|
|
$arr_tags = str_getcsv($_GET['tags']);
|
|
|
|
|
if(count($arr_tags)) {
|
|
|
|
|
array_walk($arr_tags,'arr_add_hashes');
|
2012-02-07 09:13:16 +01:00
|
|
|
$str_tags = $br . implode(' ',$arr_tags) . $br;
|
2011-09-21 01:31:45 +02:00
|
|
|
}
|
|
|
|
|
}
|
2011-04-10 12:36:12 +02:00
|
|
|
|
|
|
|
|
logger('parse_url: ' . $url);
|
2010-07-23 07:41:45 +02:00
|
|
|
|
2012-02-07 09:13:16 +01:00
|
|
|
if($textmode)
|
|
|
|
|
$template = $br . '[bookmark=%s]%s[/bookmark]%s' . $br;
|
|
|
|
|
else
|
|
|
|
|
$template = "<br /><a class=\"bookmark\" href=\"%s\" >%s</a>%s<br />";
|
2010-07-23 07:41:45 +02:00
|
|
|
|
2010-12-26 00:01:02 +01:00
|
|
|
$arr = array('url' => $url, 'text' => '');
|
|
|
|
|
|
|
|
|
|
call_hooks('parse_link', $arr);
|
|
|
|
|
|
|
|
|
|
if(strlen($arr['text'])) {
|
|
|
|
|
echo $arr['text'];
|
|
|
|
|
killme();
|
|
|
|
|
}
|
|
|
|
|
|
2011-10-28 11:50:00 +02:00
|
|
|
|
2011-09-20 07:21:55 +02:00
|
|
|
if($url && $title && $text) {
|
|
|
|
|
|
2012-02-07 09:13:16 +01:00
|
|
|
if($textmode)
|
2012-07-12 23:41:04 +02:00
|
|
|
$text = $br . '[quote]' . trim($text) . '[/quote]' . $br;
|
2012-02-07 09:13:16 +01:00
|
|
|
else
|
2012-07-12 23:41:04 +02:00
|
|
|
$text = '<br /><blockquote>' . trim($text) . '</blockquote><br />';
|
2012-02-07 09:13:16 +01:00
|
|
|
|
2011-09-20 07:21:55 +02:00
|
|
|
$title = str_replace(array("\r","\n"),array('',''),$title);
|
|
|
|
|
|
2011-09-21 01:31:45 +02:00
|
|
|
$result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
|
2011-09-20 07:21:55 +02:00
|
|
|
|
2012-07-12 01:17:33 +02:00
|
|
|
logger('parse_url (unparsed): returns: ' . $result);
|
2011-09-20 07:21:55 +02:00
|
|
|
|
|
|
|
|
echo $result;
|
|
|
|
|
killme();
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-12 01:17:33 +02:00
|
|
|
$siteinfo = parseurl_getsiteinfo($url);
|
2011-09-20 07:21:55 +02:00
|
|
|
|
2012-07-12 01:17:33 +02:00
|
|
|
if($siteinfo["title"] == "") {
|
2012-07-12 23:41:04 +02:00
|
|
|
echo print_r($siteinfo, true);
|
|
|
|
|
//echo sprintf($template,$url,$url,'') . $str_tags;
|
2010-07-23 07:41:45 +02:00
|
|
|
killme();
|
2012-07-12 01:17:33 +02:00
|
|
|
} else {
|
|
|
|
|
$text = $siteinfo["text"];
|
|
|
|
|
$title = $siteinfo["title"];
|
2010-07-23 07:41:45 +02:00
|
|
|
}
|
|
|
|
|
|
2012-07-12 08:20:27 +02:00
|
|
|
$image = "";
|
|
|
|
|
|
|
|
|
|
foreach ($siteinfo["images"] as $imagedata)
|
|
|
|
|
if($textmode)
|
|
|
|
|
$image .= '[img='.$imagedata["width"].'x'.$imagedata["height"].']'.$imagedata["src"].'[/img]';
|
|
|
|
|
else
|
|
|
|
|
$image .= '<img height="'.$imagedata["height"].'" width="'.$imagedata["width"].'" src="'.$imagedata["src"].'" alt="photo" />';
|
|
|
|
|
|
2010-07-23 08:17:41 +02:00
|
|
|
if(strlen($text)) {
|
2012-02-07 09:13:16 +01:00
|
|
|
if($textmode)
|
2012-07-12 08:20:27 +02:00
|
|
|
$text = $br.'[quote]'.trim($text).'[/quote]'.$br ;
|
2012-02-07 09:13:16 +01:00
|
|
|
else
|
2012-07-12 08:20:27 +02:00
|
|
|
$text = '<br /><blockquote>'.trim($text).'</blockquote><br />';
|
2010-07-23 07:41:45 +02:00
|
|
|
}
|
|
|
|
|
|
2011-10-28 11:50:00 +02:00
|
|
|
if($image) {
|
2012-07-12 23:41:04 +02:00
|
|
|
$text = $br.$br.$image.$text;
|
2011-10-28 11:50:00 +02:00
|
|
|
}
|
2011-09-20 07:21:55 +02:00
|
|
|
$title = str_replace(array("\r","\n"),array('',''),$title);
|
2011-09-04 09:48:45 +02:00
|
|
|
|
2011-09-21 01:31:45 +02:00
|
|
|
$result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
|
2011-09-05 04:58:03 +02:00
|
|
|
|
2012-07-12 01:17:33 +02:00
|
|
|
logger('parse_url: returns: ' . $result);
|
2011-09-05 04:58:03 +02:00
|
|
|
|
2012-07-12 23:41:04 +02:00
|
|
|
echo trim($result);
|
2010-07-23 07:41:45 +02:00
|
|
|
killme();
|
2011-05-23 10:37:09 +02:00
|
|
|
}
|
