Battle.net Armory Scraper

scrape

  function armory_scrape($realm, $character)
  {	// Scrape armory url based on provided realm and character information

		$realm = sanitize_xml(strtolower($realm));
		$character = sanitize_xml(strtolower($character));
		$armoryurl = "http://us.battle.net/wow/en/character/$realm/$character/advanced";
		$debug = false;

		function scrape($armoryurl)
		{	// Get Armory Page
			if($content=@file_get_contents($armoryurl))
			{
				return $content;
			}
			else
			{
				return false;
			}
		}

		function process_page($content)
		{	// Process Armory Data
		/*	eminence.character_cache structure:
			SELECT `id`, `realm`, `name`, `account`, `lastupdate`, `lastcache`, `title`, `guild`, `level`, `race`, `spec`,
			`class`, `points`, `ilvl`, `health`, `power`, `tradeskill1_name`, `tradeskill1_value`, `tradeskill2_name`,
			`tradeskill2_value`, `inv01`, `inv02`, `inv03`, `inv04`, `inv05`, `inv06`, `inv07`, `inv08`, `inv09`, `inv10`,
			`inv11`, `inv12`, `inv13`, `inv14`, `inv15`, `inv16`, `inv17`, `inv18`, `inv19` FROM `character_cache` WHERE 1;
		*/
			$content_array   = array();

			// Realm[1] and Character Name[2]
			$tempcrap = explode("/", substr($content, strpos($content, '<div class="name"><a href="/wow/en/character/')+45, 20));
			$realm = $tempcrap[0];
			$name  = $tempcrap[1];

			// Account[3] Get ID from PHP if logged in...I'm too lazy to code this crap now, I'll do it later.
			#

			// lastupdate[4]
			$tempcrap = explode("<", substr($content, strpos($content, '<div class="summary-lastupdate">')+58, 30));
			$lastupdate = ltrim(rtrim($tempcrap[0]));

			//

			$content_array[0] = NULL;				# id
			$content_array[1] = $realm;				# realm
			$content_array[2] = $name;				# name
			$content_array[3] = NULL;				# account
			$content_array[4] = $lastupdate;		# lastupdate


			if(strstr($content, "<p>This character profile could not be displayed, possibly for one of the following reasons:</p>"))
			{	return 1;	}
			else
			{	return $content_array;	}
		}

		function update_cache($content_array)
		{	// Update SQL Cache

		}

		// DEBUG
		if($debug) { echo "<a href=\"$armoryurl\" target=\"_blank\">$armoryurl</a><br /><br />"; }

		if(!$content = scrape($armoryurl))
		{
			if($debug) { echo "Scrape failed."; }
			return false;
		}
		else
		{
			$content_array = process_page($content);

			if($debug)
			{
				echo "Scrape Succeded.<br /><br />";
				echo "<pre>";
				var_dump($content_array);
				echo "</pre>";
				echo "<div style=\"border:1px solid;\">" . $content . "</div>";
			}
			return true;
		}
	}