Category: Tutorial

Creating a Sitemap Based on all the Links within a Website

I built this web crawler because I wanted a way to create a sitemap of this website I was building. I know there are a few websites out there that will do this for you but I didn’t want to rely on someone else and I wanted to change a few things. So in order to do this I used php and cURL.
I started out creating a class for the crawler. When I create a new crawler class I pass in the url of the website I want to start with. This also uses cURL to access the webpage and get the content and headers. Inside this class are also methods to get all the links of a page, the page title, the entire content, just the body content, and the headers. But you could easily add more to say grab all the images on a page.

The Crawler Class

  

class Crawler {
  protected $markup='';
  protected $httpinfo='';

  public function __construct($uri, $justheaders=0){
    $output = $this->getMarkup($uri, $justheaders);
    $this->markup = $output['output'];
    $this->httpinfo = $output['code'];
  }

  public function getMarkup($uri, $justheaders) {
    $ch = curl_init($uri);
    curl_setopt($ch, CURLOPT_HEADER, 1);
    if($justheaders){
      curl_setopt($ch, CURLOPT_NOBODY, 1);
      curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
      curl_setopt($ch, CURLOPT_FAILONERROR, 1);
    }
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
    curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 30);
    curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
    curl_setopt($ch, CURLOPT_MAXREDIRS, 5);

    $output['output'] = curl_exec($ch);
    $output['code'] = curl_getinfo($ch);
    curl_close($ch);
    return $output;
  }

  public function get($type){
    $method = "_get_{$type}";
    if (method_exists($this, $method)){
      return call_user_method($method, $this);
    }
  }

  protected function _get_info(){
    return $this->httpinfo;
  }

  protected function _get_links(){
    if(!empty($this->markup)){
      preg_match_all('/<a(?:.*?)href=(["|\'].*?["|\'])(.*?)>(.*?)\<\/a\>/i',
                               $this->markup, $links);
      return !empty($links[1]) ? array_flip(array_flip($links[1])) : FALSE;
    }
  }

  protected function _get_body(){
    if(!empty($this->markup)){
      preg_match('/\<body\>(.*?)\<\/body\>/msU', $this->markup, $body);
      return $body[1];
    }
  }
  protected function _get_content(){
    if(!empty($this->markup)){
      return $this->markup;
    }
  }

  protected function _get_pagetitle() {
    if (!empty($this->markup)){
     preg_match_all('/<title>(.*?)\<\/title\>/si', $this->markup, $pagetitles);
     return !empty($pagetitles[1]) ? $pagetitles[1] : FALSE;
    }
  }
}

After this I create a recursive function that will follow each of the links. Each time I call this function I create a new instance of the Crawler class. If the url isn’t valid I just return. If the url is redirected curl has an option to follow links, the CURLOPT_FOLLOWLOCATION option. Since this is set to on, you need to get the actual url which is contained in the header information. After this I call the get links function. This will return all the unique links on a page. (Calling the array_flip twice makes them unique).

I then get the title tag for each page. This is used when creating the sitemap. I remove all the script tags and all the html tags. The next thing I do is get a base url. Since I’m creating a sitemap of just one website I don’t need any links to external pages. So if any of the links don’t contain this base url I wont follow it.

Now I begin looping through all the links on the page. If its an external link I return. If it is an absolute link and it contains a “/” I replace the whole thing with a slash. If it doesn’t have the slash but is an absolute link then the new val is “”. I do this because I am prepending the base url that we got earlier to it.

I then explode on the “/”. If the first element is empty then I put the base url in there. Otherwise I prepend the url to it. This is done to get the correct link no matter if it is a relative link, root relative, or absolute.
The complete link is formed and checked to see if it already exist in the globallinkarr. If it doesn’t I add it and then begin getting the different levels. Basically everytime there is a “/” in the url then that is a different level. This is used when I am creating the html sitemap. Also to create this array of levels, I have to call array_merge_recursive. Well the regular php function didn’t quite work. If a url has numbers as one of its levels for example blog/2009/12/post then that function would turn the 2009 to its own key. So I needed it to keep the keys the same so I just got another function off of php.net.

The Function to Get all the Links

$dontfollow = array('pdf', 'jpg', 'png', 'jpeg','zip', 'gz', 'tar', 'txt');

function findAllLinks($url){
  global $globallinkarr;
  global $depthlinks;
  global $dontfollow;
  global $pagetitles;
  global $contentarr;

  $crawl = new Crawler($url);
  $info = $crawl->get('info');

  $validcodes = array(200,301,302);
  if(!in_array($info['http_code'], $validcodes))
    return;
  $url = $info['url'];
  $links = $crawl->get('links');
  $title = $crawl->get('pagetitle');
  $title = $title[0];
  $body = $crawl->get('body');
  $count++;

  $content = strip_tags(preg_replace('//msU', '', $body));

  if(!array_key_exists($url, $contentarr))
    $contentarr[$url] = array('title'=>"$title", 'pagecontent'=>"$content");

  if(!count($links) || !is_array($links)) return;
  else{
    if(preg_match('/http(?:s)?:\/\/(.*?)\/(.*)/', $url, $pattern)){
      $baseurl = $pattern[1];
    }else{
      $baseurl = $url;
    }

    foreach($links as $val){
      if(preg_match('/.*?javascript:void\(0\)/', $val) || ereg('#', $val)){
        continue;
      }
      if(!preg_match('/[0-9a-zA-Z]/', $val)) continue;
      $val = trim($val, '"\'');

      /**
       * CHECK IF LINK IS GOING TO ANOTHER DOMAIN.  IF SO DONT FOLLOW IT.
      */

      if(preg_match('/^http(s)?:\/\//', $val) &&
               !strpos($val, preg_replace('/http(s)?:\/\//', '', $baseurl))){
        continue;
      }

      if(ereg('http', $val) && preg_match('/^http(s)?:\/\/.*?\//', $val)){
        $val = preg_replace('/^http(s)?:\/\/.*?\//', '/', $val);
      }else if(ereg('http', $val)){
        $val = '';
      }

      $sl = explode('/', $val);

      if(!preg_match('/[0-9a-zA-Z]/', $sl[0])){
        $sl[0] = preg_replace('/^http(s)?:\/\//', '', $baseurl);
        $complink = implode('/', $sl);
        $sl = explode('/', $complink);

      }else{
        $prepend = explode('/', preg_replace('/^http(s)?:\/\//', '', $url));
        if(count($prepend)>1){
          array_pop($prepend);
          $prep = implode('/', $prepend);
        }else $prep = $prepend[0];
        $sl[0] = $prep.'/'.$sl[0];
        $complink = implode('/', $sl);

        $sl = explode('/', $complink);

      }
      if(!end($sl)) array_pop($sl);

      if(!in_array($complink, $globallinkarr)){
        $globallinkarr[] = $complink;
        $pagetitles[$complink] = $title;

        $depth = count($sl);
        $templinks = array();
        $newlinks = array();
        if($depth > 1){
           if(!$sl[$depth-1]) $sl[$depth-1] = 'index';
           $templinks[$sl[$depth-2]][] = $sl[$depth-1];

           if($depth > 2){
	     for($i=$depth-2; $i>0; $i--){
               $hold = $templinks;
               $templinks = array();
               $templinks[$sl[$i-1]] = $hold;

             }
          }

          $temp = $templinks[$sl[0]];
          $newlinks[$sl[0]] = $temp;

        }
        $depthlinks = array_merge_recursive2($newlinks,$depthlinks);
        $end = strtolower(end(explode(".", $complink)));

        if(!preg_match('/^http(s)?:\/\//', $complink))
          $complink = 'http://'.$complink;
          if(!in_array($end, $dontfollow) && !ereg("sitemap", $complink)){
            findAllLinks($complink);
          }
       }
    }
    return;
  }
}

The Array Merge Recursive Function I Used From php.net

function array_merge_recursive2($array1, $array2){
  $arrays = func_get_args();
  $narrays = count($arrays);

  // check arguments
  // comment out if more performance is necessary
  //   (in this case the foreach loop will trigger a warning if the argument is not an array)
  for ($i = 0; $i < $narrays; $i ++) {
   if (!is_array($arrays[$i])) {
   // also array_merge_recursive returns nothing in this case
     trigger_error('Argument #' . ($i+1) . ' is not an array - trying to merge array with scalar! Returning null!', E_USER_WARNING);
     return;
    }
  }

    // the first array is in the output set in every case
  $ret = $arrays[0];

  // merege $ret with the remaining arrays
  for ($i = 1; $i < $narrays; $i ++) {
    foreach ($arrays[$i] as $key => $value) {
     /***  KEEP THIS COMMENTED OUT TO KEEP THE ORIGINAL KEYS
     //if (((string) $key) === ((string) intval($key))) { // integer or string as integer key - append
     //   $ret[] = $value;
    // }
    // else { // string key - merge
      if (is_array($value) && isset($ret[$key])) {
        // if $ret[$key] is not an array you try to merge an scalar
        // value with an array - the result is not defined (incompatible arrays)
        // in this case the call will trigger an E_USER_WARNING and the $ret[$key] will be null.
        $ret[$key] = array_merge_recursive2($ret[$key], $value);
      }
      else {
        $ret[$key] = $value;
      }
           // }
    }
  }
  return $ret;
}

So after I created the arrays with all the links I create the sitemaps. The first one here is an xml sitemap used for the robots.txt file.
Its really simple and used the globallinkarr array.

XML Sitemap

function createXMLSiteMap($globallinkarr){
  $xml = '<?xml version="1.0" encoding="UTF-8"?>
    <urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">';
   if(count($globallinkarr)){
   foreach($globallinkarr as $val){
     if(!preg_match('/http(s)?:\/\//', $val)){
       $val = 'http://'.$val;
     }
     $xml .= '
       <url>
          <loc>'.str_replace('&', '&amp',$val).'</loc>
       </url>';
    }
  }
  $xml .= '</urlset>';
  return $xml;
}

The html sitemap is created using a recursive function that goes through depthlinkarr. Also it checks if the link is valid if it isn’t in the globallinkarr since all those are already checked. When it does check it only needs the headers so to speed things up I set the curl option CURLOPT_NOBODY to true. The page was timing out on me a lot but with this option set and checking to see if it already existed in the globallinkarr helped stop it from timing out. But if you have a whole lot of links there is a good chance this will cause your page to timeout.

The HTML Sitemap

function createSiteMap($depthlinks, $before = ''){
  global $globallinkarr;
  global $pagetitles;
  $validcodes = array(200,301,302);
  if(count($depthlinks)){
  $sitetree = '<ul style="padding:5px; margin:5px;">';
  foreach($depthlinks as $key=>$val){
    if(is_array($val)){
      if($before) $newbefore = $before.'/';
      $newbefore .= $key;

      $newkey = preg_replace('/^http(s)?:\/\//', '', $newbefore);

    $title = ($pagetitles[$newkey] != "") ? $pagetitles[$newkey] : $newbefore;
      if(!preg_match('/^http(s)?:\/\//', $newbefore))
         $newbefore = 'http://'.$newbefore;
      $exist = 0;
      if(in_array($newbefore, $globallinkarr)){
        $exist = 1;
      }else{
        $test = new Crawler($newbefore, 1);
        $info = $test->get('info');
        if(in_array($info['http_code'], $validcodes))
          $exist = 1;
        }
        if($exist){
          $sitetree .= '
           <li><a style="display:block;" href="'.$newbefore.'"
           target="_blank" title="'.$title.'" />'.$title.'</a></li>';
        }else{
          $sitetree .= '
             <li>'.$title;
        }
        $temp = createSiteMap($val, $newbefore);
        if($temp){
          $sitetree .= $temp;
          $sitetree .= '</li>';
        }
      }else{
        if($before != '') $newval = $before.'/'.$val;
        else $newval = $val;
        $newkey = preg_replace('/^http(s)?:\/\//', '',$newval);
        $title = $pagetitles[$newkey] ? $pagetitles[$newkey] : $newval;
        if(!preg_match('/^http(s)?:\/\//', $newval))
           $newval = 'http://'.$newval;

        $exist = 0;
        if(in_array($newval, $globallinkarr)){
            $exist = 1;
        }else{
          $test = new Crawler($newval, 1);
          $info = $test->get('info');
          if(in_array($info['http_code'], $validcodes))
            $exist = 1;
        }

        if($exist){
          $sitetree .= '
            <li><a href="'.$newval.'" title="'.$title.'"
                target="_blank">'.$title.'</a></li>';
        }
      }
    }
  $sitetree .= '</ul>';
  return $sitetree;
  }
}