eensyweensy

a poor man's webspider
git clone https://wehaveforgeathome.hates.computer/eensyweensy.git
Log | Files | Refs | LICENSE

eensyweensy.php (1077B)


      1 <?php
      2 if (!isset($_POST["data"])) {
      3   echo "<pre>You need to feed me some DATA.\nDada Dada Dada.</pre>";
      4   exit;
      5 }
      6 
      7 $data = json_decode(stripslashes($_REQUEST["data"]), true);
      8 $results = array();
      9 
     10 if (is_null($data["href"]) || is_null($data["on_last_page"])
     11       || !is_array($data["urls"])) {
     12   $results["error"] = "validation error: missing on_last_page or urls";
     13 }
     14 else {
     15   // http://www.tizag.com/phpT/fileappend.php
     16   $file_name = "output.txt";
     17   $file_handle = fopen($file_name, "a+");
     18   foreach ($data["urls"] as $url) {
     19     fwrite($file_handle, "$url\n");
     20   }
     21   fclose($file_handle);
     22 
     23   if ($data["on_last_page"] == "true") {
     24     $results["url"] = "http://hampsterdance.com";
     25   }
     26   else {
     27     function next_url ($matches) {
     28       if (count($matches) == 2) {
     29         return $matches[1] . "?page=2";
     30       }
     31       $page = (int) $matches[3];
     32       $page++;
     33       return $matches[1] . "?page=$page";
     34     };
     35     $results["url"] = preg_replace_callback(
     36       "/^([^\?]+)(\?page=(\d*))?$/",
     37       "next_url",
     38       $data["href"]
     39     );
     40   }
     41 }
     42 
     43 echo json_encode($results);
     44 ?>