friendica/mod/parse_url.php

81 lines
1.3 KiB
PHP
Raw Normal View History

2010-07-23 05:41:45 +00:00
<?php
require_once('library/HTML5/Parser.php');
2010-12-21 03:38:34 +00:00
2010-07-23 05:41:45 +00:00
function parse_url_content(&$a) {
2010-07-23 06:17:41 +00:00
2010-07-23 05:41:45 +00:00
$url = trim($_GET['url']);
$text = null;
2010-07-23 06:17:41 +00:00
$template = "<a href=\"%s\" >%s</a>%s";
2010-07-23 05:41:45 +00:00
if($url)
$s = fetch_url($url);
2010-07-23 06:17:41 +00:00
else {
echo '';
killme();
}
2010-12-21 03:38:34 +00:00
2010-07-23 05:41:45 +00:00
if(! $s) {
2010-07-23 06:17:41 +00:00
echo sprintf($template,$url,$url,'');
2010-07-23 05:41:45 +00:00
killme();
}
2010-12-21 03:38:34 +00:00
$dom = @HTML5_Parser::parse($s);
2010-07-23 05:41:45 +00:00
if(! $dom)
return $ret;
$items = $dom->getElementsByTagName('title');
2010-07-23 06:17:41 +00:00
if($items) {
foreach($items as $item) {
2010-12-07 00:26:32 +00:00
$title = trim($item->textContent);
2010-07-23 06:17:41 +00:00
break;
}
}
$divs = $dom->getElementsByTagName('div');
if($divs) {
foreach($divs as $div) {
$class = $div->getAttribute('class');
if($class && stristr($class,'article')) {
$items = $div->getElementsByTagName('p');
if($items) {
foreach($items as $item) {
$text = $item->textContent;
$text = strip_tags($text);
if(strlen($text) < 100)
continue;
$text = substr($text,0,250) . '...' ;
break;
}
}
}
}
}
if(! $text) {
$items = $dom->getElementsByTagName('p');
if($items) {
foreach($items as $item) {
$text = $item->textContent;
$text = strip_tags($text);
if(strlen($text) < 100)
continue;
$text = substr($text,0,250) . '...' ;
break;
}
2010-07-23 06:17:41 +00:00
}
}
if(strlen($text)) {
$text = '<br />' . $text;
2010-07-23 05:41:45 +00:00
}
2010-07-23 06:17:41 +00:00
echo sprintf($template,$url,$title,$text);
2010-07-23 05:41:45 +00:00
killme();
}