2010-07-23 05:41:45 +00:00
|
|
|
<?php
|
2012-07-18 19:06:38 +00:00
|
|
|
/* 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>
|
|
|
|
|
*/
|
|
|
|
|
|
2012-07-11 23:17:33 +00:00
|
|
|
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 21:41:04 +00:00
|
|
|
function completeurl($url, $scheme) {
|
2015-03-07 22:14:26 +00:00
|
|
|
$urlarr = parse_url($url);
|
2012-07-12 21:41:04 +00:00
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
if (isset($urlarr["scheme"]))
|
|
|
|
|
return($url);
|
2012-07-12 21:41:04 +00:00
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
$schemearr = parse_url($scheme);
|
2012-07-12 21:41:04 +00:00
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
$complete = $schemearr["scheme"]."://".$schemearr["host"];
|
2012-07-12 21:41:04 +00:00
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
if (@$schemearr["port"] != "")
|
|
|
|
|
$complete .= ":".$schemearr["port"];
|
2012-07-12 21:41:04 +00:00
|
|
|
|
2012-09-05 23:26:11 +00:00
|
|
|
if(strpos($urlarr['path'],'/') !== 0)
|
|
|
|
|
$complete .= '/';
|
|
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
$complete .= $urlarr["path"];
|
2012-07-12 21:41:04 +00:00
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
if (@$urlarr["query"] != "")
|
|
|
|
|
$complete .= "?".$urlarr["query"];
|
2012-07-12 21:41:04 +00:00
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
if (@$urlarr["fragment"] != "")
|
|
|
|
|
$complete .= "#".$urlarr["fragment"];
|
2012-07-12 21:41:04 +00:00
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
return($complete);
|
2012-07-12 21:41:04 +00:00
|
|
|
}
|
|
|
|
|
|
2015-07-16 23:08:28 +00:00
|
|
|
function parseurl_getsiteinfo_cached($url, $no_guessing = false, $do_oembed = true) {
|
|
|
|
|
|
|
|
|
|
$data = Cache::get("parse_url:".$no_guessing.":".$do_oembed.":".$url);
|
|
|
|
|
if (!is_null($data)) {
|
|
|
|
|
$data = unserialize($data);
|
|
|
|
|
return $data;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
$data = parseurl_getsiteinfo($url, $no_guessing, $do_oembed);
|
|
|
|
|
|
2015-08-24 15:24:14 +00:00
|
|
|
Cache::set("parse_url:".$no_guessing.":".$do_oembed.":".$url,serialize($data), CACHE_DAY);
|
2015-07-16 23:08:28 +00:00
|
|
|
|
|
|
|
|
return $data;
|
|
|
|
|
}
|
|
|
|
|
|
2014-07-14 06:37:40 +00:00
|
|
|
function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1) {
|
2014-10-03 11:02:15 +00:00
|
|
|
require_once("include/network.php");
|
2014-05-03 10:04:54 +00:00
|
|
|
|
2014-09-07 08:47:37 +00:00
|
|
|
$a = get_app();
|
|
|
|
|
|
2012-07-11 23:17:33 +00:00
|
|
|
$siteinfo = array();
|
2014-02-22 14:46:19 +00:00
|
|
|
|
2014-07-14 06:37:40 +00:00
|
|
|
if ($count > 10) {
|
|
|
|
|
logger("parseurl_getsiteinfo: Endless loop detected for ".$url, LOGGER_DEBUG);
|
|
|
|
|
return($siteinfo);
|
|
|
|
|
}
|
|
|
|
|
|
2014-04-04 08:58:31 +00:00
|
|
|
$url = trim($url, "'");
|
|
|
|
|
$url = trim($url, '"');
|
2014-10-03 11:02:15 +00:00
|
|
|
|
|
|
|
|
$url = original_url($url);
|
|
|
|
|
|
2014-04-04 08:58:31 +00:00
|
|
|
$siteinfo["url"] = $url;
|
2014-02-22 14:46:19 +00:00
|
|
|
$siteinfo["type"] = "link";
|
|
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
$stamp1 = microtime(true);
|
|
|
|
|
|
2012-07-11 23:17:33 +00:00
|
|
|
$ch = curl_init();
|
|
|
|
|
curl_setopt($ch, CURLOPT_URL, $url);
|
|
|
|
|
curl_setopt($ch, CURLOPT_HEADER, 1);
|
2015-01-04 19:04:59 +00:00
|
|
|
curl_setopt($ch, CURLOPT_NOBODY, 1);
|
2012-07-11 23:17:33 +00:00
|
|
|
curl_setopt($ch, CURLOPT_TIMEOUT, 3);
|
|
|
|
|
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
|
2013-03-02 13:46:06 +00:00
|
|
|
//curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
|
2014-09-07 08:47:37 +00:00
|
|
|
curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
|
2012-07-11 23:17:33 +00:00
|
|
|
|
|
|
|
|
$header = curl_exec($ch);
|
2013-03-02 13:46:06 +00:00
|
|
|
$curl_info = @curl_getinfo($ch);
|
2015-03-07 22:14:26 +00:00
|
|
|
$http_code = $curl_info['http_code'];
|
2012-07-11 23:17:33 +00:00
|
|
|
curl_close($ch);
|
|
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
$a->save_timestamp($stamp1, "network");
|
|
|
|
|
|
2014-04-04 08:58:31 +00:00
|
|
|
if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302") OR ($curl_info['http_code'] == "303") OR ($curl_info['http_code'] == "307"))
|
2013-03-02 13:46:06 +00:00
|
|
|
AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
|
|
|
|
|
if ($curl_info['redirect_url'] != "")
|
2014-07-14 06:37:40 +00:00
|
|
|
$siteinfo = parseurl_getsiteinfo($curl_info['redirect_url'], $no_guessing, $do_oembed, ++$count);
|
2013-03-02 13:46:06 +00:00
|
|
|
else
|
2014-07-14 06:37:40 +00:00
|
|
|
$siteinfo = parseurl_getsiteinfo($curl_info['location'], $no_guessing, $do_oembed, ++$count);
|
2013-03-02 13:46:06 +00:00
|
|
|
return($siteinfo);
|
|
|
|
|
}
|
|
|
|
|
|
2015-01-04 19:04:59 +00:00
|
|
|
// if the file is too large then exit
|
|
|
|
|
if ($curl_info["download_content_length"] > 1000000)
|
|
|
|
|
return($siteinfo);
|
|
|
|
|
|
|
|
|
|
// if it isn't a HTML file then exit
|
|
|
|
|
if (($curl_info["content_type"] != "") AND !strstr(strtolower($curl_info["content_type"]),"html"))
|
|
|
|
|
return($siteinfo);
|
|
|
|
|
|
2015-04-05 18:40:31 +00:00
|
|
|
if ($do_oembed) {
|
|
|
|
|
require_once("include/oembed.php");
|
|
|
|
|
|
|
|
|
|
$oembed_data = oembed_fetch_url($url);
|
|
|
|
|
|
|
|
|
|
if ($oembed_data->type != "error")
|
|
|
|
|
$siteinfo["type"] = $oembed_data->type;
|
|
|
|
|
|
|
|
|
|
if (($oembed_data->type == "link") AND ($siteinfo["type"] != "photo")) {
|
|
|
|
|
if (isset($oembed_data->title))
|
|
|
|
|
$siteinfo["title"] = $oembed_data->title;
|
|
|
|
|
if (isset($oembed_data->description))
|
|
|
|
|
$siteinfo["text"] = trim($oembed_data->description);
|
|
|
|
|
if (isset($oembed_data->thumbnail_url))
|
|
|
|
|
$siteinfo["image"] = $oembed_data->thumbnail_url;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2015-04-06 11:49:57 +00:00
|
|
|
$stamp1 = microtime(true);
|
|
|
|
|
|
2015-01-04 19:04:59 +00:00
|
|
|
// Now fetch the body as well
|
|
|
|
|
$ch = curl_init();
|
|
|
|
|
curl_setopt($ch, CURLOPT_URL, $url);
|
|
|
|
|
curl_setopt($ch, CURLOPT_HEADER, 1);
|
|
|
|
|
curl_setopt($ch, CURLOPT_NOBODY, 0);
|
|
|
|
|
curl_setopt($ch, CURLOPT_TIMEOUT, 10);
|
|
|
|
|
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
|
|
|
|
|
curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
|
|
|
|
|
|
|
|
|
|
$header = curl_exec($ch);
|
|
|
|
|
$curl_info = @curl_getinfo($ch);
|
2015-03-07 22:14:26 +00:00
|
|
|
$http_code = $curl_info['http_code'];
|
2015-01-04 19:04:59 +00:00
|
|
|
curl_close($ch);
|
|
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
$a->save_timestamp($stamp1, "network");
|
|
|
|
|
|
2012-07-12 21:41:04 +00:00
|
|
|
// Fetch the first mentioned charset. Can be in body or header
|
2014-04-04 08:58:31 +00:00
|
|
|
$charset = "";
|
2012-07-12 21:41:04 +00:00
|
|
|
if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
|
2014-07-24 20:50:56 +00:00
|
|
|
$charset = trim(trim(trim(array_pop($matches)), ';,'));
|
2014-04-04 08:58:31 +00:00
|
|
|
|
|
|
|
|
if ($charset == "")
|
2012-07-11 23:17:33 +00:00
|
|
|
$charset = "utf-8";
|
|
|
|
|
|
|
|
|
|
$pos = strpos($header, "\r\n\r\n");
|
|
|
|
|
|
|
|
|
|
if ($pos)
|
|
|
|
|
$body = trim(substr($header, $pos));
|
|
|
|
|
else
|
|
|
|
|
$body = $header;
|
|
|
|
|
|
2014-07-24 20:50:56 +00:00
|
|
|
if (($charset != '') AND (strtoupper($charset) != "UTF-8")) {
|
|
|
|
|
logger("parseurl_getsiteinfo: detected charset ".$charset, LOGGER_DEBUG);
|
|
|
|
|
//$body = mb_convert_encoding($body, "UTF-8", $charset);
|
|
|
|
|
$body = iconv($charset, "UTF-8//TRANSLIT", $body);
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-11 23:17:33 +00:00
|
|
|
$body = mb_convert_encoding($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);
|
|
|
|
|
|
2013-02-24 11:54:53 +00:00
|
|
|
$list = $xpath->query("//meta[@content]");
|
2015-03-07 22:14:26 +00:00
|
|
|
foreach ($list as $node) {
|
|
|
|
|
$attr = array();
|
|
|
|
|
if ($node->attributes->length)
|
|
|
|
|
foreach ($node->attributes as $attribute)
|
|
|
|
|
$attr[$attribute->name] = $attribute->value;
|
|
|
|
|
|
|
|
|
|
if (@$attr["http-equiv"] == 'refresh') {
|
|
|
|
|
$path = $attr["content"];
|
|
|
|
|
$pathinfo = explode(";", $path);
|
|
|
|
|
$content = "";
|
|
|
|
|
foreach ($pathinfo AS $value) {
|
|
|
|
|
if (substr(strtolower($value), 0, 4) == "url=")
|
|
|
|
|
$content = substr($value, 4);
|
|
|
|
|
}
|
|
|
|
|
if ($content != "") {
|
|
|
|
|
$siteinfo = parseurl_getsiteinfo($content, $no_guessing, $do_oembed, ++$count);
|
|
|
|
|
return($siteinfo);
|
|
|
|
|
}
|
|
|
|
|
}
|
2013-02-24 11:54:53 +00:00
|
|
|
}
|
|
|
|
|
|
2012-07-12 21:41:04 +00:00
|
|
|
//$list = $xpath->query("head/title");
|
|
|
|
|
$list = $xpath->query("//title");
|
2012-07-11 23:17:33 +00:00
|
|
|
foreach ($list as $node)
|
|
|
|
|
$siteinfo["title"] = html_entity_decode($node->nodeValue, ENT_QUOTES, "UTF-8");
|
|
|
|
|
|
2012-07-12 21:41:04 +00:00
|
|
|
//$list = $xpath->query("head/meta[@name]");
|
|
|
|
|
$list = $xpath->query("//meta[@name]");
|
2012-07-11 23:17:33 +00:00
|
|
|
foreach ($list as $node) {
|
|
|
|
|
$attr = array();
|
|
|
|
|
if ($node->attributes->length)
|
2015-03-07 22:14:26 +00:00
|
|
|
foreach ($node->attributes as $attribute)
|
|
|
|
|
$attr[$attribute->name] = $attribute->value;
|
2012-07-11 23:17:33 +00:00
|
|
|
|
2014-03-02 00:00:36 +00:00
|
|
|
$attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
|
2012-07-11 23:17:33 +00:00
|
|
|
|
2014-04-04 08:58:31 +00:00
|
|
|
if ($attr["content"] != "")
|
|
|
|
|
switch (strtolower($attr["name"])) {
|
|
|
|
|
case "fulltitle":
|
|
|
|
|
$siteinfo["title"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "description":
|
|
|
|
|
$siteinfo["text"] = $attr["content"];
|
|
|
|
|
break;
|
2014-12-01 21:54:01 +00:00
|
|
|
case "thumbnail":
|
|
|
|
|
$siteinfo["image"] = $attr["content"];
|
|
|
|
|
break;
|
2014-04-04 08:58:31 +00:00
|
|
|
case "twitter:image":
|
|
|
|
|
$siteinfo["image"] = $attr["content"];
|
|
|
|
|
break;
|
2014-09-27 10:49:00 +00:00
|
|
|
case "twitter:image:src":
|
|
|
|
|
$siteinfo["image"] = $attr["content"];
|
|
|
|
|
break;
|
2014-04-04 08:58:31 +00:00
|
|
|
case "twitter:card":
|
2014-05-03 10:04:54 +00:00
|
|
|
if (($siteinfo["type"] == "") OR ($attr["content"] == "photo"))
|
2014-04-04 08:58:31 +00:00
|
|
|
$siteinfo["type"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "twitter:description":
|
|
|
|
|
$siteinfo["text"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "twitter:title":
|
|
|
|
|
$siteinfo["title"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "dc.title":
|
|
|
|
|
$siteinfo["title"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "dc.description":
|
|
|
|
|
$siteinfo["text"] = $attr["content"];
|
|
|
|
|
break;
|
2014-10-03 10:18:33 +00:00
|
|
|
case "keywords":
|
|
|
|
|
$keywords = explode(",", $attr["content"]);
|
|
|
|
|
break;
|
|
|
|
|
case "news_keywords":
|
|
|
|
|
$keywords = explode(",", $attr["content"]);
|
|
|
|
|
break;
|
2014-04-04 08:58:31 +00:00
|
|
|
}
|
2014-03-16 16:48:28 +00:00
|
|
|
if ($siteinfo["type"] == "summary")
|
|
|
|
|
$siteinfo["type"] = "link";
|
2012-07-11 23:17:33 +00:00
|
|
|
}
|
|
|
|
|
|
2014-10-03 10:18:33 +00:00
|
|
|
if (isset($keywords)) {
|
|
|
|
|
$siteinfo["keywords"] = array();
|
|
|
|
|
foreach ($keywords as $keyword)
|
2015-11-28 12:15:00 +00:00
|
|
|
if (!in_array(trim($keyword), $siteinfo["keywords"]))
|
|
|
|
|
$siteinfo["keywords"][] = trim($keyword);
|
2014-10-03 10:18:33 +00:00
|
|
|
}
|
|
|
|
|
|
2012-07-12 21:41:04 +00:00
|
|
|
//$list = $xpath->query("head/meta[@property]");
|
|
|
|
|
$list = $xpath->query("//meta[@property]");
|
2012-07-11 23:17:33 +00:00
|
|
|
foreach ($list as $node) {
|
|
|
|
|
$attr = array();
|
|
|
|
|
if ($node->attributes->length)
|
2015-03-07 22:14:26 +00:00
|
|
|
foreach ($node->attributes as $attribute)
|
|
|
|
|
$attr[$attribute->name] = $attribute->value;
|
2012-07-11 23:17:33 +00:00
|
|
|
|
2014-03-02 00:00:36 +00:00
|
|
|
$attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
|
2012-07-11 23:17:33 +00:00
|
|
|
|
2014-04-04 08:58:31 +00:00
|
|
|
if ($attr["content"] != "")
|
|
|
|
|
switch (strtolower($attr["property"])) {
|
|
|
|
|
case "og:image":
|
|
|
|
|
$siteinfo["image"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "og:title":
|
|
|
|
|
$siteinfo["title"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
case "og:description":
|
|
|
|
|
$siteinfo["text"] = $attr["content"];
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2013-12-02 19:24:41 +00:00
|
|
|
if ((@$siteinfo["image"] == "") AND !$no_guessing) {
|
2015-03-07 22:14:26 +00:00
|
|
|
$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;
|
2012-07-11 23:17:33 +00:00
|
|
|
|
2012-07-12 21:41:04 +00:00
|
|
|
$src = completeurl($attr["src"], $url);
|
2012-09-06 01:24:06 +00:00
|
|
|
$photodata = @getimagesize($src);
|
2012-07-11 23:17:33 +00:00
|
|
|
|
2012-09-06 01:24:06 +00:00
|
|
|
if (($photodata) && ($photodata[0] > 150) and ($photodata[1] > 150)) {
|
2012-07-12 21:41:04 +00:00
|
|
|
if ($photodata[0] > 300) {
|
2012-07-14 11:59:42 +00:00
|
|
|
$photodata[1] = round($photodata[1] * (300 / $photodata[0]));
|
2012-07-12 21:41:04 +00:00
|
|
|
$photodata[0] = 300;
|
|
|
|
|
}
|
|
|
|
|
if ($photodata[1] > 300) {
|
2012-07-14 11:59:42 +00:00
|
|
|
$photodata[0] = round($photodata[0] * (300 / $photodata[1]));
|
2012-07-12 21:41:04 +00:00
|
|
|
$photodata[1] = 300;
|
|
|
|
|
}
|
|
|
|
|
$siteinfo["images"][] = array("src"=>$src,
|
|
|
|
|
"width"=>$photodata[0],
|
|
|
|
|
"height"=>$photodata[1]);
|
2012-07-12 06:20:27 +00:00
|
|
|
}
|
2012-07-12 21:41:04 +00:00
|
|
|
|
2015-03-07 22:14:26 +00:00
|
|
|
}
|
2012-09-06 01:24:06 +00:00
|
|
|
} else {
|
2012-07-12 21:41:04 +00:00
|
|
|
$src = completeurl($siteinfo["image"], $url);
|
2012-08-02 08:25:08 +00:00
|
|
|
|
|
|
|
|
unset($siteinfo["image"]);
|
|
|
|
|
|
2012-09-06 01:24:06 +00:00
|
|
|
$photodata = @getimagesize($src);
|
2012-07-12 06:20:27 +00:00
|
|
|
|
2012-09-06 01:24:06 +00:00
|
|
|
if (($photodata) && ($photodata[0] > 10) and ($photodata[1] > 10))
|
2012-07-12 21:41:04 +00:00
|
|
|
$siteinfo["images"][] = array("src"=>$src,
|
|
|
|
|
"width"=>$photodata[0],
|
|
|
|
|
"height"=>$photodata[1]);
|
2012-07-12 06:20:27 +00:00
|
|
|
}
|
2012-07-11 23:17:33 +00:00
|
|
|
|
2013-12-02 19:24:41 +00:00
|
|
|
if ((@$siteinfo["text"] == "") AND (@$siteinfo["title"] != "") AND !$no_guessing) {
|
2012-07-11 23:17:33 +00:00
|
|
|
$text = "";
|
|
|
|
|
|
|
|
|
|
$list = $xpath->query("//div[@class='article']");
|
|
|
|
|
foreach ($list as $node)
|
2012-07-12 21:41:04 +00:00
|
|
|
if (strlen($node->nodeValue) > 40)
|
|
|
|
|
$text .= " ".trim($node->nodeValue);
|
2012-07-11 23:17:33 +00:00
|
|
|
|
|
|
|
|
if ($text == "") {
|
|
|
|
|
$list = $xpath->query("//div[@class='content']");
|
|
|
|
|
foreach ($list as $node)
|
2012-07-12 21:41:04 +00:00
|
|
|
if (strlen($node->nodeValue) > 40)
|
|
|
|
|
$text .= " ".trim($node->nodeValue);
|
2012-07-11 23:17:33 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// If none text was found then take the paragraph content
|
|
|
|
|
if ($text == "") {
|
|
|
|
|
$list = $xpath->query("//p");
|
|
|
|
|
foreach ($list as $node)
|
2012-07-12 21:41:04 +00:00
|
|
|
if (strlen($node->nodeValue) > 40)
|
|
|
|
|
$text .= " ".trim($node->nodeValue);
|
2012-07-11 23:17:33 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ($text != "") {
|
|
|
|
|
$text = trim(str_replace(array("\n", "\r"), array(" ", " "), $text));
|
|
|
|
|
|
|
|
|
|
while (strpos($text, " "))
|
|
|
|
|
$text = trim(str_replace(" ", " ", $text));
|
|
|
|
|
|
2014-03-02 00:00:36 +00:00
|
|
|
$siteinfo["text"] = trim(html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...');
|
2012-07-11 23:17:33 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2014-04-04 08:58:31 +00:00
|
|
|
logger("parseurl_getsiteinfo: Siteinfo for ".$url." ".print_r($siteinfo, true), LOGGER_DEBUG);
|
|
|
|
|
|
2014-10-20 06:03:47 +00:00
|
|
|
call_hooks('getsiteinfo', $siteinfo);
|
|
|
|
|
|
2012-07-11 23:17:33 +00:00
|
|
|
return($siteinfo);
|
|
|
|
|
}
|
2010-12-21 03:38:34 +00:00
|
|
|
|
2011-09-20 23:31:45 +00:00
|
|
|
function arr_add_hashes(&$item,$k) {
|
|
|
|
|
$item = '#' . $item;
|
|
|
|
|
}
|
|
|
|
|
|
2010-07-23 05:41:45 +00:00
|
|
|
function parse_url_content(&$a) {
|
2010-07-23 06:17:41 +00:00
|
|
|
|
2011-09-20 05:21:55 +00:00
|
|
|
$text = null;
|
2011-09-20 23:31:45 +00:00
|
|
|
$str_tags = '';
|
2011-09-20 05:21:55 +00:00
|
|
|
|
2012-02-07 08:13:16 +00:00
|
|
|
$textmode = false;
|
|
|
|
|
|
2012-12-04 00:31:28 +00:00
|
|
|
if(local_user() && (! feature_enabled(local_user(),'richtext')))
|
|
|
|
|
$textmode = true;
|
|
|
|
|
|
2012-07-12 06:20:27 +00:00
|
|
|
//if($textmode)
|
|
|
|
|
$br = (($textmode) ? "\n" : '<br />');
|
2012-02-07 08:13:16 +00:00
|
|
|
|
2011-09-20 05:21:55 +00:00
|
|
|
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 10:36:12 +00:00
|
|
|
|
2011-09-20 23:31:45 +00:00
|
|
|
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 08:13:16 +00:00
|
|
|
$str_tags = $br . implode(' ',$arr_tags) . $br;
|
2011-09-20 23:31:45 +00:00
|
|
|
}
|
|
|
|
|
}
|
2011-04-10 10:36:12 +00:00
|
|
|
|
2015-03-22 09:53:24 +00:00
|
|
|
// add url scheme if missing
|
|
|
|
|
$arrurl = parse_url($url);
|
|
|
|
|
if (!x($arrurl, 'scheme')) {
|
|
|
|
|
if (x($arrurl, 'host'))
|
|
|
|
|
$url = "http:".$url;
|
|
|
|
|
else
|
|
|
|
|
$url = "http://".$url;
|
|
|
|
|
}
|
|
|
|
|
|
2011-04-10 10:36:12 +00:00
|
|
|
logger('parse_url: ' . $url);
|
2010-07-23 05:41:45 +00:00
|
|
|
|
2012-02-07 08:13:16 +00:00
|
|
|
if($textmode)
|
2014-02-22 14:46:19 +00:00
|
|
|
$template = '[bookmark=%s]%s[/bookmark]%s';
|
2012-02-07 08:13:16 +00:00
|
|
|
else
|
2014-02-22 14:46:19 +00:00
|
|
|
$template = "<a class=\"bookmark\" href=\"%s\" >%s</a>%s";
|
2010-07-23 05:41:45 +00:00
|
|
|
|
2010-12-25 23:01:02 +00:00
|
|
|
$arr = array('url' => $url, 'text' => '');
|
|
|
|
|
|
|
|
|
|
call_hooks('parse_link', $arr);
|
|
|
|
|
|
|
|
|
|
if(strlen($arr['text'])) {
|
|
|
|
|
echo $arr['text'];
|
|
|
|
|
killme();
|
|
|
|
|
}
|
|
|
|
|
|
2011-10-28 09:50:00 +00:00
|
|
|
|
2011-09-20 05:21:55 +00:00
|
|
|
if($url && $title && $text) {
|
|
|
|
|
|
2014-02-16 16:35:01 +00:00
|
|
|
$title = str_replace(array("\r","\n"),array('',''),$title);
|
|
|
|
|
|
2012-02-07 08:13:16 +00:00
|
|
|
if($textmode)
|
2014-01-05 15:22:42 +00:00
|
|
|
$text = '[quote]' . trim($text) . '[/quote]' . $br;
|
2014-02-16 16:35:01 +00:00
|
|
|
else {
|
|
|
|
|
$text = '<blockquote>' . htmlspecialchars(trim($text)) . '</blockquote><br />';
|
|
|
|
|
$title = htmlspecialchars($title);
|
|
|
|
|
}
|
2011-09-20 05:21:55 +00:00
|
|
|
|
2011-09-20 23:31:45 +00:00
|
|
|
$result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
|
2011-09-20 05:21:55 +00:00
|
|
|
|
2012-07-11 23:17:33 +00:00
|
|
|
logger('parse_url (unparsed): returns: ' . $result);
|
2011-09-20 05:21:55 +00:00
|
|
|
|
|
|
|
|
echo $result;
|
|
|
|
|
killme();
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-11 23:17:33 +00:00
|
|
|
$siteinfo = parseurl_getsiteinfo($url);
|
2011-09-20 05:21:55 +00:00
|
|
|
|
2014-10-20 06:03:47 +00:00
|
|
|
// if ($textmode) {
|
|
|
|
|
// require_once("include/items.php");
|
|
|
|
|
//
|
|
|
|
|
// echo add_page_info_data($siteinfo);
|
|
|
|
|
// killme();
|
|
|
|
|
// }
|
|
|
|
|
|
2014-04-04 08:58:31 +00:00
|
|
|
$url= $siteinfo["url"];
|
|
|
|
|
|
2014-12-01 21:54:01 +00:00
|
|
|
// If the link contains BBCode stuff, make a short link out of this to avoid parsing problems
|
|
|
|
|
if (strpos($url, '[') OR strpos($url, ']')) {
|
|
|
|
|
require_once("include/network.php");
|
|
|
|
|
$url = short_link($url);
|
|
|
|
|
}
|
|
|
|
|
|
2014-01-05 15:22:42 +00:00
|
|
|
$sitedata = "";
|
|
|
|
|
|
2015-03-22 10:03:46 +00:00
|
|
|
if($siteinfo["title"] != "") {
|
2012-07-11 23:17:33 +00:00
|
|
|
$text = $siteinfo["text"];
|
|
|
|
|
$title = $siteinfo["title"];
|
2010-07-23 05:41:45 +00:00
|
|
|
}
|
|
|
|
|
|
2012-07-12 06:20:27 +00:00
|
|
|
$image = "";
|
|
|
|
|
|
2014-04-04 08:58:31 +00:00
|
|
|
if (($siteinfo["type"] != "video") AND (sizeof($siteinfo["images"]) > 0)){
|
2012-09-14 06:01:10 +00:00
|
|
|
/* Execute below code only if image is present in siteinfo */
|
|
|
|
|
|
|
|
|
|
$total_images = 0;
|
|
|
|
|
$max_images = get_config('system','max_bookmark_images');
|
|
|
|
|
if($max_images === false)
|
|
|
|
|
$max_images = 2;
|
2012-07-12 06:20:27 +00:00
|
|
|
else
|
2012-09-14 06:01:10 +00:00
|
|
|
$max_images = intval($max_images);
|
|
|
|
|
|
|
|
|
|
foreach ($siteinfo["images"] as $imagedata) {
|
|
|
|
|
if($textmode)
|
|
|
|
|
$image .= '[img='.$imagedata["width"].'x'.$imagedata["height"].']'.$imagedata["src"].'[/img]' . "\n";
|
|
|
|
|
else
|
|
|
|
|
$image .= '<img height="'.$imagedata["height"].'" width="'.$imagedata["width"].'" src="'.$imagedata["src"].'" alt="photo" /><br />';
|
|
|
|
|
$total_images ++;
|
|
|
|
|
if($max_images && $max_images >= $total_images)
|
|
|
|
|
break;
|
2014-03-16 16:48:28 +00:00
|
|
|
}
|
2012-09-14 06:01:10 +00:00
|
|
|
}
|
2012-07-12 06:20:27 +00:00
|
|
|
|
2010-07-23 06:17:41 +00:00
|
|
|
if(strlen($text)) {
|
2012-02-07 08:13:16 +00:00
|
|
|
if($textmode)
|
2014-01-05 15:22:42 +00:00
|
|
|
$text = '[quote]'.trim($text).'[/quote]';
|
2012-02-07 08:13:16 +00:00
|
|
|
else
|
2014-02-16 16:35:01 +00:00
|
|
|
$text = '<blockquote>'.htmlspecialchars(trim($text)).'</blockquote>';
|
2010-07-23 05:41:45 +00:00
|
|
|
}
|
|
|
|
|
|
2014-02-22 14:46:19 +00:00
|
|
|
if($image)
|
2012-07-12 21:41:04 +00:00
|
|
|
$text = $br.$br.$image.$text;
|
2014-02-22 14:46:19 +00:00
|
|
|
else
|
|
|
|
|
$text = $br.$text;
|
2014-01-05 15:22:42 +00:00
|
|
|
|
2011-09-20 05:21:55 +00:00
|
|
|
$title = str_replace(array("\r","\n"),array('',''),$title);
|
2011-09-04 07:48:45 +00:00
|
|
|
|
2011-09-20 23:31:45 +00:00
|
|
|
$result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
|
2011-09-05 02:58:03 +00:00
|
|
|
|
2012-07-11 23:17:33 +00:00
|
|
|
logger('parse_url: returns: ' . $result);
|
2011-09-05 02:58:03 +00:00
|
|
|
|
2014-01-05 15:22:42 +00:00
|
|
|
$sitedata .= trim($result);
|
|
|
|
|
|
2014-03-16 16:48:28 +00:00
|
|
|
if (($siteinfo["type"] == "video") AND ($url != ""))
|
2014-04-04 08:58:31 +00:00
|
|
|
echo "[class=type-video]".$sitedata."[/class]";
|
2014-03-16 16:48:28 +00:00
|
|
|
elseif (($siteinfo["type"] != "photo"))
|
2014-02-22 14:46:19 +00:00
|
|
|
echo "[class=type-link]".$sitedata."[/class]";
|
|
|
|
|
else
|
|
|
|
|
echo "[class=type-photo]".$title.$br.$image."[/class]";
|
2014-01-05 15:22:42 +00:00
|
|
|
|
2010-07-23 05:41:45 +00:00
|
|
|
killme();
|
2011-05-23 08:37:09 +00:00
|
|
|
}
|
2014-05-03 10:04:54 +00:00
|
|
|
?>
|
