Merge https://github.com/friendica/friendica into pull

[friendica.git] / mod / parse_url.php

diff --git a/mod/parse_url.php b/mod/parse_url.php
index 89ef24f8760fa5b99576351169d944429a04145c..cdf2223a866d677c2598614b4c8e185c6b480154 100644 (file)
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -3,9 +3,21 @@
 require_once('library/HTML5/Parser.php');
 require_once('library/HTMLPurifier.auto.php');
 
+function arr_add_hashes(&$item,$k) {
+       $item = '#' . $item;
+}
+
 function parse_url_content(&$a) {
 
        $text = null;
+       $str_tags = '';
+
+       $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']));
@@ -15,13 +27,24 @@ function parse_url_content(&$a) {
        if($_GET['title'])
                $title = strip_tags(trim($_GET['title']));
 
-       if($_GET['desc'])
-               $text = strip_tags(trim($_GET['desc']));
+       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');
+                       $str_tags = $br . implode(' ',$arr_tags) . $br;
+               }
+       }
 
        logger('parse_url: ' . $url);
 
 
-       $template = "<br /><a class=\"bookmark\" href=\"%s\" >%s</a>%s<br />";
+       if($textmode)
+               $template = $br . '[bookmark=%s]%s[/bookmark]%s' . $br;
+       else
+               $template = "<br /><a class=\"bookmark\" href=\"%s\" >%s</a>%s<br />";
 
 
        $arr = array('url' => $url, 'text' => '');
@@ -33,12 +56,17 @@ function parse_url_content(&$a) {
                killme();
        }
 
+
        if($url && $title && $text) {
 
-               $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
+               if($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);
+               $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 
                logger('parse_url (unparsed): returns: ' . $result); 
 
@@ -54,13 +82,26 @@ function parse_url_content(&$a) {
                killme();
        }
 
-       logger('parse_url: data: ' . $s, LOGGER_DATA);
+//     logger('parse_url: data: ' . $s, LOGGER_DATA);
 
        if(! $s) {
-               echo sprintf($template,$url,$url,'');
+               echo sprintf($template,$url,$url,'') . $str_tags;
                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($s,'<title>')+7,64);
@@ -71,16 +112,19 @@ function parse_url_content(&$a) {
 
        $config = HTMLPurifier_Config::createDefault();
        $config->set('Cache.DefinitionImpl', null);
-
        $purifier = new HTMLPurifier($config);
        $s = $purifier->purify($s);
 
-//     logger('parse_url: purified: ' . $s, LOGGER_DATA);
+//     logger('purify_output: ' . $s);
 
-       $dom = @HTML5_Parser::parse($s);
+       try {
+               $dom = HTML5_Parser::parse($s);
+       } catch (DOMException $e) {
+               logger('scrape_dfrn: parse error: ' . $e);
+       }
 
        if(! $dom) {
-               echo sprintf($template,$url,$url,'');
+               echo sprintf($template,$url,$url,'') . $str_tags;
                killme();
        }
 
@@ -142,13 +186,79 @@ function parse_url_content(&$a) {
                }
        }
 
+       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) {
+                                               // guess mimetype from headers or filename
+                                               $type = guess_image_type($image,true);
+                                               
+                                               require_once('include/Photo.php');
+                                               $ph = new Photo($i, $type);
+                                               if($ph->is_valid()) {
+                                                       if($ph->getWidth() > 300 || $ph->getHeight() > 300) {
+                                                               $ph->scaleImage(300);
+                                                               $new_width = $ph->getWidth();
+                                                               $new_height = $ph->getHeight();
+                                                               if($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
+                                                       $image = '';
+                                       
+                                       }
+                               }
+                       }
+               }
+       }
+
        if(strlen($text)) {
-               $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
+               if($textmode)
+                       $text = $br .$br . '[quote]' . $text . '[/quote]' . $br ;
+               else
+                       $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
        }
 
+       if($image) {
+               $text = $image . $br . $text;
+       }
        $title = str_replace(array("\r","\n"),array('',''),$title);
 
-       $result = sprintf($template,$url,($title) ? $title : $url,$text);
+       $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 
        logger('parse_url: returns: ' . $result);