mod/parse_url.php

   1 <?php
   2
   3 require_once('library/HTML5/Parser.php');
   4
   5
   6 function parse_url_content(&$a) {
   7
   8         $url = trim($_GET['url']);
   9
  10         $text = null;
  11
  12         $template = "<a href=\"%s\" >%s</a>%s";
  13
  14         if($url)
  15                 $s = fetch_url($url);
  16         else {
  17                 echo '';
  18                 killme();
  19         }
  20
  21         if(! $s) {
  22                 echo sprintf($template,$url,$url,'');
  23                 killme();
  24         }
  25
  26         $dom = @HTML5_Parser::parse($s);
  27
  28         if(! $dom)
  29                 return $ret;
  30
  31         $items = $dom->getElementsByTagName('title');
  32
  33         if($items) {
  34                 foreach($items as $item) {
  35                         $title = trim($item->textContent);
  36                         break;
  37                 }
  38         }
  39
  40
  41         $divs = $dom->getElementsByTagName('div');
  42         if($divs) {
  43                 foreach($divs as $div) {
  44                         $class = $div->getAttribute('class');
  45                         if($class && stristr($class,'article')) {
  46                                 $items = $div->getElementsByTagName('p');
  47                                 if($items) {
  48                                         foreach($items as $item) {
  49                                                 $text = $item->textContent;
  50                                                 $text = strip_tags($text);
  51                                                 if(strlen($text) < 100)
  52                                                         continue;
  53                                                 $text = substr($text,0,250) . '...' ;
  54                                                 break;
  55                                         }
  56                                 }
  57                         }
  58                 }
  59         }
  60
  61         if(! $text) {
  62                 $items = $dom->getElementsByTagName('p');
  63                 if($items) {
  64                         foreach($items as $item) {
  65                                 $text = $item->textContent;
  66                                 $text = strip_tags($text);
  67                                 if(strlen($text) < 100)
  68                                         continue;
  69                                 $text = substr($text,0,250) . '...' ;
  70                                 break;
  71                         }
  72                 }
  73         }
  74
  75         if(strlen($text)) {
  76                 $text = '<br />' . $text;
  77         }
  78
  79         echo sprintf($template,$url,$title,$text);
  80         killme();
  81 }