Advertisement
Guest User

Untitled

a guest
Jun 30th, 2015
203
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
  1. <?
  2.  
  3. function curl($url) {
  4.     $ch = curl_init();
  5.     curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
  6.     curl_setopt($ch, CURLOPT_VERBOSE, 0);
  7.     curl_setopt($ch, CURLOPT_USERAGENT, 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.0.3705; .NET CLR 1.1.4322)');
  8.     curl_setopt($ch, CURLOPT_FOLLOWLOCATION, TRUE);
  9.     curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
  10.     curl_setopt($ch, CURLOPT_COOKIEJAR, 'cookie.txt');
  11.     curl_setopt($ch, CURLOPT_REFERER, $referer);
  12.     curl_setopt($ch, CURLOPT_URL, $url);
  13.     $return = curl_exec($ch);
  14.     return $return;
  15. }
  16.  
  17. // EKSEMPEL PÅ BRUK:
  18. // sett $søkelink under til kategorien som trengs.
  19. // så er det bare å bytte ut insertProduct funksjonen helt nederst med logikken for å putte det i databasen deres.
  20.  
  21. $søkelink = "http://www.focalprice.com/categorylist/001029?orderby=newarrivals&pagesize=72&page=1";
  22. $html = curl($søkelink, "http://www.focalprice.com");
  23.  
  24. preg_match('~<span>Pages: 1/([0-9]{1,})</span>~', $html, $out);
  25. $pages = $out[1];
  26. $page = 0;
  27. while ($page < $pages) {
  28.     $page++;
  29.     $søkelink = preg_replace("/page=[0-9]{1,}/", "page=$page", $søkelink);
  30.     echo "FOCALPRICE SØKELINK: $søkelink<br>";
  31.     if ($page > 1) $html = curl($søkelink);
  32.  
  33.     $products = explode('<ul class="infobox">', $html);
  34.     array_shift($products);
  35.     while (count($products) > 72) {
  36.         array_pop($products);
  37.     }
  38.     echo "found " . count($products) . " products on page $page..<br>";
  39.  
  40.     foreach ($products as $product) {
  41.         if (!preg_match('~href="(http://www.focalprice.com/(.*)\.html)"~', $product, $out)) {
  42.             var_dump($product);
  43.             echo "found no url in product<br>";
  44.             continue;
  45.         }
  46.         $producturl = $out[1];
  47.  
  48.         preg_match("#(.*?//.*?/.*?/)#", $producturl, $match);
  49.         $producturl = $match[1];
  50.  
  51.         if (productExists($producturl)) {
  52.             echo "$producturl exists. continue<br>";
  53.             continue;
  54.         }
  55.  
  56.         echo "Producturl[$producturl]<br>";
  57.         $html = curl($producturl);
  58.         preg_match_all('~(http://img.focalprice.com//550x426/(.*?)-(.*?))"~', $html, $out);
  59.         $bilder = array();
  60.         foreach ($out[1] as $tmp) {
  61.             $bilder[] = str_replace("//550", "/550", $tmp);
  62.         }
  63.         var_dump($bilder);
  64.  
  65.         preg_match('~<h1 id="productName" class="c00">(.*?)<~s', $html, $out);
  66.         $produktnavn = $out[1];
  67.  
  68.         preg_match('/ecomm_totalvalue: (\d{1,3}\.\d{2}),/', $html, $out);
  69.         $kinapris = $out[1];
  70.  
  71.         preg_match('~<div class="description_m">(.*?)<div class="description_m hide">~s', $html, $out);
  72.         $beskrivelse = $out[1];
  73.  
  74.         reporting($producturl, $bilder, $produktnavn, $kinapris, $beskrivelse);
  75.         insertProduct(mysql_real_escape_string($produktnavn), $kinapris, $producturl, $storeurl, mysql_real_escape_string($beskrivelse), $kategori, $subkategori, $butikk, $bilder);
  76.         $focal_product_count++;
  77.     }
  78. }
  79. ?>
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement