Services/SMOMatcher.php

   1 <?php
   2
   3 namespace Services;
   4
   5 use Services\ISMOMatcher;
   6 use DOMDocument;
   7 use DOMXPath;
   8 use DOMNodeList;
   9
  10 class SMOMatcher implements ISMOMatcher
  11 {
  12     private $_records;
  13
  14     public function match($title, $filesize)
  15     {
  16         if(!$this->scrapeSmo()) return null;
  17
  18         $most_likely = array('confidence' => 0);
  19         foreach($this->_records as $rowNum => $row) {
  20             $cells = $row->getElementsByTagName('td');
  21
  22             if($rowNum > 0) {
  23                 $candidate = $this->cellToCandidateArray($cells);
  24                 $this->setCandidateSimilarty($title, $filesize, $candidate);
  25                 $most_likely = $candidate['confidence'] > $most_likely['confidence'] ? $candidate : $most_likely;
  26             }
  27         }
  28
  29         return $most_likely;
  30     }
  31
  32     private function scrapeSmo()
  33     {
  34         if($this->_records) return true;
  35
  36         $c = curl_init('http://stepmaniaonline.net/index.php?page=downloads');
  37         curl_setopt($c, CURLOPT_RETURNTRANSFER, true);
  38         //curl_setopt(... other options you want...)
  39
  40         $html = curl_exec($c);
  41
  42         if (curl_error($c)) return false;
  43
  44         // Get the status code
  45         $status = curl_getinfo($c, CURLINFO_HTTP_CODE);
  46
  47         curl_close($c);
  48         //$html = file_get_contents('smo.html');
  49         $dom = new DOMDocument();
  50         @$dom->loadHTML($html);
  51         $xpath = new DOMXPath($dom);
  52
  53         $table = $xpath->query("/html/body/div[@class='container']/div[@class='mid']/div[@class='content']/div[3]/div[@class='blockcontent']/table")->item(0);
  54         $this->_records = $table->getElementsByTagName("tr");
  55
  56         return true;
  57     }
  58
  59     private function cellToCandidateArray(DOMNodeList $cells)
  60     {
  61         return array(
  62             'href' => $cells->item(0)->getElementsByTagName('a')->item(0)->getAttribute('href'),
  63             'title' => $cells->item(0)->nodeValue,
  64             'filesize' => strpos($cells->item(1)->nodeValue, 'Gb') === true ? ($cells->item(1)->nodeValue)*1024*1024*1024 : ($cells->item(1)->nodeValue)*1024*1024
  65         );
  66     }
  67
  68     private function setCandidateSimilarty($title, $filesize, array &$candidate)
  69     {
  70         similar_text($title, $candidate['title'], $percent);
  71         $r = $candidate['filesize'] > $filesize ? $filesize/$candidate['filesize'] : $candidate['filesize']/$filesize;
  72         $candidate['confidence'] = $percent*$r;
  73     }
  74 }