Humans vs Zombies v2

<?php

  /*
   * @author: Andrew Horsman
   * @description: Humans vs. Zombies Extractor
   */
   
  // Configuration, meh at a second config file.
  $grabPage = 'http://www.hvzgatech.com/?page=killboard';
  $sqlServ  = 'localhost';
  $sqlUser  = 'root';
  $sqlPass  = 'pass';
  $sqlDb    = 'humans_zombies';
   
  // Grab the markup up from Humans vs. Zombies website.
  echo "[+] Downloading page...\n";
  $humansZombiesPage = file_get_contents($grabPage);
  
  // Define a Hash, everytime we find a table data cell (<td>) we'll insert the
  // value into the current iterated key (go through the keys in a loop).  
  // An itearator variable will keep track of the current key we're on,
  // we can use array_keys to list all of the $killboard keys in a standard 
  // array (one that is indexed from 0 thus allowing us to easily iterate 
  // through keys.
  echo "[+] Creating table structure...\n";
  $killboard = array(
    "name" => array(),
    "team" => array(),
    "kills" => array(),
    "lastfed" => array(),
    "slogan" => array()
  );  
  $boardKeys = array_keys($killboard);
  
  // Initial processing of the page, deleting all instances of \n and \r so we
  // don't have to do a multiline RegEx.  Then to avoid confusing actual data
  // with the <td> tags above the fold, delete any part of the page before
  // the `Slogan` column.
  echo "[+] Grabbing specific page parts...\n";
  $humansZombiesPage = str_replace(array("\n", "\r"), '', $humansZombiesPage);
  $humansZombiesPage = substr($humansZombiesPage, strpos($humansZombiesPage, "Slogan</td></tr>") + 16);
  $humansZombiesPage = substr($humansZombiesPage, 0, strpos($humansZombiesPage, "</table>"));
  
  // Yes, I'm using RegEx to parse HTML
  // http://www.urbandictionary.com/define.php?term=ctfd
  // This RegEx matches any sub string that contains a to z, A to Z, 0 to 9 or
  // whitespace, and is wrapped in `<td>` followed by `</td>` (although these
  // wrapped tags are not included in the actual match data).
  echo "[+] Parsing...\n";
  if (!preg_match_all('/(?<=\<td\>)[a-zA-Z0-9\s:\-]*?(?=\<\/td\>)/', $humansZombiesPage, $matches))
    die("[-] No data found.");
    
  // preg_match_all puts each RegEx _group_ match into an array, therefore we
  // only want the first group at $matches[0].
  echo "[+] Poorly coded markup parsed...\n";
  $matches = $matches[0];
    
  // Here we're going to iterate through every match.  Each time we find a match
  // we will place it in the next key of the $killboards Hash.  Therefore, the
  // first match will go into $killboard["name"], the next into $killboard["team"]
  // etc...  $curCell keeps track of the current key, we increment this per loop
  // and reset it when it hits the number of keys in $boardKeys.  $boardKeys
  // contains all the keys in $killboard.
  echo "[+] Inserting into Hash...\n";
  $curCell = 0;
  for ($i = 0; $i < count($matches); ++$i) {
    $killboard[$boardKeys[$curCell]][] = $matches[$i]; // Append the match ([]) to the proper key in $killboard.
    
    ++$curCell;
    if ($curCell == count($boardKeys)) $curCell = 0;
  }
  
  // Print the table header (each column double tab separated), followed by
  // each group of $killboard's keys.  Count the number of values in lastfed 
  // because this is the last column, if a group was not fully filled up, we 
  // won't get an out of bounds issue.
  echo "[+] Displaying table...\n";
  echo "Name\t\tTeam\t\tKills\t\tLast Fed\t\tSlogan\n";
  for ($i = 0; $i < count($killboard[end($boardKeys)]); ++$i) 
    echo $killboard["name"][$i] . "\t\t" . $killboard["team"][$i] . "\t\t" . $killboard["kills"][$i] . "\t\t" . $killboard["lastfed"][$i] . "\t\t" . $killboard["slogan"][$i] . "\n";

  echo "[+] Connecting to the database...\n";
  
  // Let's put this in MySQL, shall we? Top of the mornin' SQL!
  $dbc = mysql_connect($sqlServ, $sqlUser, $sqlPass) or
    die("[-] Error connecting to MySQL - " . mysql_error());
    
  mysql_select_db($sqlDb, $dbc) or 
    die("[-] Error selecting database - " . mysql_error());
    
  // Create the records table just in case this is the first time ran.
  $createHumansZombieTableQueryString = 
    'CREATE TABLE IF NOT EXISTS `records` (' .
    '`id` INT(11) NOT NULL AUTO_INCREMENT,' .
    '`name` VARCHAR(255) NOT NULL,' .
    '`team` VARCHAR(255) NOT NULL,' . 
    '`kills` INT(11) NOT NULL,' .
    '`last_fed` VARCHAR(255) NOT NULL,' .
    '`slogan` VARCHAR(255) NOT NULL,' . 
    '`recorded_time` DATETIME NOT NULL,' .
    'PRIMARY KEY (`id`)' .
    ')';
  mysql_query($createHumansZombieTableQueryString, $dbc) or
    die("[-] Could not create table - " . mysql_error($dbc));
   
  // Time to populate... 
  for ($i = 0; $i < count($killboard[end($boardKeys)]); ++$i) {
    $tmp_name  = mysql_real_escape_string($killboard["name"][$i], $dbc);
    $tmp_team  = mysql_real_escape_string($killboard["team"][$i], $dbc);
    $tmp_kills = mysql_real_escape_string($killboard["kills"][$i], $dbc);
    $tmp_last  = mysql_real_escape_string($killboard["lastfed"][$i], $dbc); 
    $tmp_slog  = mysql_real_escape_string($killboard["slogan"][$i], $dbc);
  
    $recordInsertQueryString = 
      "INSERT INTO `records` (`name`, `team`, `kills`, `last_fed`, `slogan`, `recorded_time`) VALUES (" .
      "'" . $tmp_name . "', '" .
      $tmp_team . "', " .
      $tmp_kills . ", '" .
      $tmp_last . "', '" .
      $tmp_slog . "', NOW())";
    mysql_query($recordInsertQueryString, $dbc) or error_log("Could not add record - " . mysql_error($dbc));
  }
  
  echo "[+] Added records...\n";
  
  mysql_close($dbc);
  echo "[~] All done chaps!\n";
  
?>