friendica/mod/parse_url.php

268 lines
5.9 KiB
PHP
Raw Normal View History

2010-07-23 01:41:45 -04:00
<?php
require_once('library/HTML5/Parser.php');
require_once('library/HTMLPurifier.auto.php');
2010-12-20 22:38:34 -05:00
function arr_add_hashes(&$item,$k) {
$item = '#' . $item;
}
2010-07-23 01:41:45 -04:00
function parse_url_content(&$a) {
2010-07-23 02:17:41 -04:00
$text = null;
$str_tags = '';
2012-02-07 03:13:16 -05:00
$textmode = false;
if(local_user() && intval(get_pconfig(local_user(),'system','plaintext')))
$textmode = true;
if($textmode)
$br = (($textmode) ? "\n" : '<br /?');
if(x($_GET,'binurl'))
$url = trim(hex2bin($_GET['binurl']));
else
$url = trim($_GET['url']);
if($_GET['title'])
$title = strip_tags(trim($_GET['title']));
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 03:13:16 -05:00
$str_tags = $br . implode(' ',$arr_tags) . $br;
}
}
logger('parse_url: ' . $url);
2010-07-23 01:41:45 -04:00
2012-02-07 03:13:16 -05:00
if($textmode)
$template = $br . '[bookmark=%s]%s[/bookmark]%s' . $br;
else
$template = "<br /><a class=\"bookmark\" href=\"%s\" >%s</a>%s<br />";
2010-07-23 01:41:45 -04:00
2010-12-25 18:01:02 -05:00
$arr = array('url' => $url, 'text' => '');
call_hooks('parse_link', $arr);
if(strlen($arr['text'])) {
echo $arr['text'];
killme();
}
if($url && $title && $text) {
2012-02-07 03:13:16 -05:00
if($textmode)
$text = $br . $br . '[quote]' . $text . '[/quote]' . $br;
else
$text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
$title = str_replace(array("\r","\n"),array('',''),$title);
$result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
logger('parse_url (unparsed): returns: ' . $result);
echo $result;
killme();
}
if($url) {
2010-07-23 01:41:45 -04:00
$s = fetch_url($url);
} else {
2010-07-23 02:17:41 -04:00
echo '';
killme();
}
2010-12-20 22:38:34 -05:00
// logger('parse_url: data: ' . $s, LOGGER_DATA);
2010-12-25 18:01:02 -05:00
2010-07-23 01:41:45 -04:00
if(! $s) {
echo sprintf($template,$url,$url,'') . $str_tags;
2010-07-23 01:41:45 -04:00
killme();
}
$matches = '';
$c = preg_match('/\<head(.*?)\>(.*?)\<\/head\>/ism',$s,$matches);
if($c) {
// logger('parse_url: header: ' . $matches[2], LOGGER_DATA);
try {
$domhead = HTML5_Parser::parse($matches[2]);
} catch (DOMException $e) {
logger('scrape_dfrn: parse error: ' . $e);
}
if($domhead)
logger('parsed header');
}
if(! $title) {
if(strpos($s,'<title>')) {
$title = substr($s,strpos($s,'<title>')+7,64);
if(strpos($title,'<') !== false)
$title = strip_tags(substr($title,0,strpos($title,'<')));
}
2011-07-09 18:06:53 -04:00
}
$config = HTMLPurifier_Config::createDefault();
$config->set('Cache.DefinitionImpl', null);
$purifier = new HTMLPurifier($config);
$s = $purifier->purify($s);
// logger('purify_output: ' . $s);
2011-10-20 19:48:07 -04:00
try {
$dom = HTML5_Parser::parse($s);
} catch (DOMException $e) {
logger('scrape_dfrn: parse error: ' . $e);
}
2010-07-23 01:41:45 -04:00
if(! $dom) {
echo sprintf($template,$url,$url,'') . $str_tags;
killme();
}
2010-07-23 01:41:45 -04:00
$items = $dom->getElementsByTagName('title');
2010-07-23 02:17:41 -04:00
if($items) {
foreach($items as $item) {
2010-12-06 19:26:32 -05:00
$title = trim($item->textContent);
2010-07-23 02:17:41 -04:00
break;
}
}
if(! $text) {
$divs = $dom->getElementsByTagName('div');
if($divs) {
foreach($divs as $div) {
$class = $div->getAttribute('class');
if($class && (stristr($class,'article') || stristr($class,'content'))) {
$items = $div->getElementsByTagName('p');
if($items) {
foreach($items as $item) {
$text = $item->textContent;
if(stristr($text,'<script')) {
$text = '';
continue;
}
$text = strip_tags($text);
if(strlen($text) < 100) {
$text = '';
continue;
}
$text = substr($text,0,250) . '...' ;
break;
2011-07-11 20:23:42 -04:00
}
}
}
if($text)
break;
}
}
if(! $text) {
$items = $dom->getElementsByTagName('p');
if($items) {
foreach($items as $item) {
$text = $item->textContent;
if(stristr($text,'<script'))
continue;
$text = strip_tags($text);
if(strlen($text) < 100) {
$text = '';
continue;
}
$text = substr($text,0,250) . '...' ;
break;
2011-07-11 20:23:42 -04:00
}
}
2010-07-23 02:17:41 -04:00
}
}
if(! $text) {
logger('parsing meta');
$items = (isset($domhead) && is_object($domhead) ? $domhead->getElementsByTagName('meta') : null);
if($items) {
foreach($items as $item) {
$property = $item->getAttribute('property');
if($property && (stristr($property,':description'))) {
$text = $item->getAttribute('content');
if(stristr($text,'<script')) {
$text = '';
continue;
}
$text = strip_tags($text);
$text = substr($text,0,250) . '...' ;
}
if($property && (stristr($property,':image'))) {
$image = $item->getAttribute('content');
if(stristr($text,'<script')) {
$image = '';
continue;
}
$image = strip_tags($image);
$i = fetch_url($image);
if($i) {
2012-06-29 00:15:43 -04:00
require_once('include/Photo.php');
2012-06-07 11:42:13 -04:00
// guess mimetype from headers or filename
$type = guess_image_type($image,true);
2012-06-29 00:15:43 -04:00
2012-06-07 11:42:13 -04:00
$ph = new Photo($i, $type);
if($ph->is_valid()) {
if($ph->getWidth() > 300 || $ph->getHeight() > 300) {
$ph->scaleImage(300);
$new_width = $ph->getWidth();
$new_height = $ph->getHeight();
2012-02-07 03:13:16 -05:00
if($textmode)
$image = $br . $br . '[img=' . $new_width . 'x' . $new_height . ']' . $image . '[/img]';
else
$image = '<br /><br /><img height="' . $new_height . '" width="' . $new_width . '" src="' .$image . '" alt="photo" />';
}
else {
if($textmode)
$image = $br . $br . '[img]' . $image . '[/img]';
else
$image = '<br /><br /><img src="' . $image . '" alt="photo" />';
}
}
else
$image = '';
}
}
}
}
}
2010-07-23 02:17:41 -04:00
if(strlen($text)) {
2012-02-07 03:13:16 -05:00
if($textmode)
$text = $br .$br . '[quote]' . $text . '[/quote]' . $br ;
else
$text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
2010-07-23 01:41:45 -04:00
}
if($image) {
2012-02-07 03:13:16 -05:00
$text = $image . $br . $text;
}
$title = str_replace(array("\r","\n"),array('',''),$title);
$result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
2011-09-04 22:58:03 -04:00
logger('parse_url: returns: ' . $result);
echo $result;
2010-07-23 01:41:45 -04:00
killme();
}