<?phprequire('class-IXR.php');require('headzoo.php');/** * DiggSig API Scraper Class * * The class that does the task of grabbing user pages from digg.com, and * regular expressing the pages for useful information. * * @author Sean Hickey <seanhickey@gmail.com> * @version 1.0 */class DiggSigAPIScraper { /** * The user's Digg username. * @protected string */ protected $user_name; /** * Array of data pulled from grabbed digg.com pages using regexp. * @protected array */ protected $user_profile_data; /** * HTML from the user's profile and dugg page. * @protected string */ protected $user_profile_dugg_html; /** * HTML from the user's submitted page. * @protected string */ protected $user_submitted_html; /** * HTML from the user's commented on page. * @protected string */ protected $user_commented_html; /** * HTML from the user's homepage page. * @protected string */ protected $user_homepage_html; /** * HTML from the user's friends page. * @protected string */ protected $user_friends_html; /** * Class constructor * * @param string $user_name The user's Digg username * @return void */ public function __construct($user_name = null) { if ($user_name) $this->SetUserName($user_name); } /** * Sets the user's Digg username * * Sets the user's Digg username. Always returns TRUE. * * @param string $user_name The user's Digg username * @return bool */ public function SetUserName($user_name) { $this->user_name = str_replace(' ', '_', $user_name); return true; } /** * Grabs the HTML from digg.com, and runs regexp on it * * Grabs all the required pages from digg.com, and runs regexp on the * contents to get all the user's Digg data. Returns an array with * all the data, or FALSE if there was a failure. * * @return mixed */ public function Parse($query = null) { $query_protecteds = array ( 'profile' => '', 'submitted' => '', 'commented' => '', 'homepage' => '', 'friends' => '' ); if ($query) { parse_str($query, $query_protecteds); } if (isset($query_protecteds['profile'])) { if (!$this->GetUserProfileDugg()) { return false; } if (!$this->ParseUserProfileDuggHTML()) { return false; } } if (isset($query_protecteds['submitted'])) { if (!$this->GetUserSubmitted()) { return false; } if (!$this->ParseUserSubmittedHTML()) { return false; } } if (isset($query_protecteds['commented'])) { if (!$this->GetUserCommentedOn()) { return false; } if (!$this->ParseUserCommentedHTML()) { return false; } } if (isset($query_protecteds['homepage'])) { if (!$this->GetUserHomepage()) { return false; } if (!$this->ParseUserHomepageHTML()) { return false; } } if (isset($query_protecteds['friends'])) { if (!$this->GetUserFriends()) { return false; } if (!$this->ParseUserFriendsHTML()) { return false; } } return $this->user_profile_data; } /** * Gets the user's profile information from digg.com * * Grabs the Digg profile page for the user, and the dugg page for the * user. Stores the HTML in the class property $user_profile_dugg_html. * Returns TRUE if the HTML was grabbed successfully, and FALSE if it * wasn't. * * @return bool */ private function GetUserProfileDugg() { if (!$this->user_name) return false; $profile_html = $this->GetUserPage('http://digg.com/users/' . $this->user_name . '/profile'); if (!$profile_html) return false; $dugg_html = $this->GetUserPage('http://digg.com/users/' . $this->user_name . '/dugg'); if (!$dugg_html) return false; $this->user_profile_dugg_html = $profile_html . $dugg_html; return true; } /** * Gets the user's submitted information from digg.com * * Grabs the Digg submitted page for the user. Stores the HTML * in the class property $user_submitted_html. * Returns TRUE if the HTML was grabbed successfully, and FALSE if it * wasn't. * * @return bool */ private function GetUserSubmitted() { if (!$this->user_name) return false; $this->user_submitted_html = $this->GetUserPage('http://digg.com/users/' . $this->user_name . '/submitted'); return $this->user_submitted_html; } /** * Gets the user's commented on information from digg.com * * Grabs the Digg commented on page for the user. Stores the HTML * in the class property $user_commented_html. * Returns TRUE if the HTML was grabbed successfully, and FALSE if it * wasn't. * * @return bool */ private function GetUserCommentedOn() { if (!$this->user_name) return false; $this->user_commented_html = $this->GetUserPage('http://digg.com/users/' . $this->user_name . '/commented'); return $this->user_commented_html; } /** * Gets the user's homepage information from digg.com * * Grabs the Digg homepage page for the user. Stores the HTML * in the class property $user_homepage_html. * Returns TRUE if the HTML was grabbed successfully, and FALSE if it * wasn't. * * @return bool */ private function GetUserHomepage() { if (!$this->user_name) return false; $this->user_homepage_html = $this->GetUserPage('http://digg.com/users/' . $this->user_name . '/homepage'); return $this->user_homepage_html; } /** * Gets the user's friends information from digg.com * * Grabs the Digg friends page for the user. Stores the HTML * in the class property $user_homepage_html. * Returns TRUE if the HTML was grabbed successfully, and FALSE if it * wasn't. * * @return bool */ private function GetUserFriends() { if (!$this->user_name) return false; $this->user_friends_html = $this->GetUserPage('http://digg.com/users/' . $this->user_name . '/friends/list'); return $this->user_friends_html; } /** * Grabs a page from digg.com * * Grabs a page from digg.com, and if the $property parameter is set, * stores the HTML in the variable. Returns the HTML on success and * FALSE on failure. * * @param string $url The URL of the page to grab * @param string $property The member property to store the HTML in * @return mixed */ private function GetUserPage($url) { $html = Headzoo::get_http_document($url); if (!$html) return false; return $html; } /** * Parse the $user_profile_dugg_html content * * Runs regular expressions on the $user_profile_dugg_html property, and * adds matched data to the $user_profile_data property. Returns TRUE * if all the regular expressions matched, returns FALSE otherwise. * * @return bool */ private function ParseUserProfileDuggHTML() { $regexps = array ( 'GetRealName' => "~<dt>Real Name:</dt>.<dd>(.*)<\/dd>~isU", 'GetMemberName' => "~<dt>Member Name:</dt>.*<dd>(.*)</dd>~isU", 'GetAvatarURL' => "~<img src=\"(.*)\" width=\"48\" height=\"48\" alt=\"\" class=\"user-image\" />~iU", 'GetLocation' => "~<dt>Location:</dt>.*<dd>(.*)</dd>~isU", 'GetRanking' => "~<dt>Overall Ranking:</dt>.*<dd><a href=\"/tsort\?user={$this->user_name}\">(.*)</a>~isU", 'GetNumberOfStoriesDugg' => "~<dt>Stories Dugg:</dt>.*<dd>(.*)\s{4}\(~isU", 'GetNumberOfStoriesSubmitted' => "~<dt>Stories Submitted:</dt>.*<dd>(.*)\s{4}\(~isU", 'GetNumberOfPromotedStories' => "~<dt>Promoted Stories:</dt>.*<dd>(.*)\s{4}\(~isU", 'GetNumberOfProfileViews' => "~<dt>Profile Views:</dt>.*<dd>(.*)</dd>~isU" ); foreach ($regexps as $method_name => $regexp) { if (!preg_match($regexp, $this->user_profile_dugg_html, $matches)) { return false; } $this->user_profile_data[$method_name] = $matches[1]; } if (!preg_match_all($this->_GetStoriesPageRegExp(), $this->user_profile_dugg_html, $matches)) { return false; } for($i = 0; $i < count($matches[1]); $i++) { $this->user_profile_data['GetLastDuggStories'][] = array('title' => $matches[2][$i], 'url' => $matches[1][$i], 'diggs' => $matches[3][$i]); } return true; } /** * Parse the $user_submitted_html content * * Runs regular expressions on the $user_submitted_html property, and * adds matched data to the $user_profile_data property. Returns TRUE * if all the regular expressions matched, returns FALSE otherwise. * * @return bool */ private function ParseUserSubmittedHTML() { if (!preg_match_all($this->_GetStoriesPageRegExp(), $this->user_submitted_html, $matches)) { return false; } for($i = 0; $i < count($matches[1]); $i++) { $this->user_profile_data['GetSubmittedStories'][] = array('title' => $matches[2][$i], 'url' => $matches[1][$i], 'diggs' => $matches[3][$i]); } return true; } /** * Parse the $user_commented_html content * * Runs regular expressions on the $user_commented_html property, and * adds matched data to the $user_profile_data property. Returns TRUE * if all the regular expressions matched, returns FALSE otherwise. * * @return bool */ private function ParseUserCommentedHTML() { if (!preg_match_all($this->_GetStoriesPageRegExp(), $this->user_commented_html, $matches)) { return false; } for($i = 0; $i < count($matches[1]); $i++) { $this->user_profile_data['GetCommentedStories'][] = array('title' => $matches[2][$i], 'url' => $matches[1][$i], 'diggs' => $matches[3][$i]); } return true; } /** * Parse the $user_homepage_html content * * Runs regular expressions on the $user_homepage_html property, and * adds matched data to the $user_profile_data property. Returns TRUE * if all the regular expressions matched, returns FALSE otherwise. * * @return bool */ private function ParseUserHomepageHTML() { if (!preg_match_all($this->_GetStoriesPageRegExp(), $this->user_homepage_html, $matches)) { return false; } for($i = 0; $i < count($matches[1]); $i++) { $this->user_profile_data['GetPromotedStories'][] = array('title' => $matches[2][$i], 'url' => $matches[1][$i], 'diggs' => $matches[3][$i]); } return true; } /** * Parse the $user_friends_html content * * Runs regular expressions on the $user_friends_html property, and * adds matched data to the $user_profile_data property. Returns TRUE * if all the regular expressions matched, returns FALSE otherwise. * * @return bool */ private function ParseUserFriendsHTML() { if (!preg_match_all('~<h3><a href="(.*)">(.*)</a></h3>~isU', $this->user_friends_html, $matches)) { return false; } for($i = 0; $i < count($matches[1]); $i++) { $this->user_profile_data['GetFriendsList'][] = array('name' => $matches[2][$i], 'url' => $matches[1][$i]); } return true; } /** * Returns the regexp needed for many of the methods * * Many of the methods run the same exact regular expression. Because of * that, the regexp is stored in this function, and retrieved by those * methods. That way if the regexp needs to be changed, it can be changed * in this one spot. * * @return string */ private function _GetStoriesPageRegExp() { return '~<h3 id="title[0-9]+"><a href="(.*)".*>(.*)</a></h3>.*<strong id="diggs-strong-[0-9]+">(.*)</strong>~isU'; }}/** * DiggSig API Get Methods Class * * The class is used to get values created by the DiggSigAPIScraper class * via method calls. This class extends the DiggSigAPIScraper class, and * this is the class that should be initiated to get user Digg information. * * @author Sean Hickey <seanhickey@gmail.com> * @version 1.0 */class DiggSigAPI extends DiggSigAPIScraper { /** * PHP's magic __call() method to call members methods * * PHP's magic __call() method to call member methods on this class * that doen't exists. * Faux methods available through this method: * * string|null GetRealName() * string|null GetMemberName() * string|null GetAvatarURL() * string|null GetLocation() * string|null GetRanking() * string|null GetNumberOfStoriesDugg() * string|null GetNumberOfStoriesSubmitted() * string|null GetNumberOfPromotedStories() * string|null GetNumberOfProfileViews() * array|null GetLastDuggStories($number_to_return = null) * array|null GetSubmittedStories($number_to_return = null) * array|null GetCommentedStories($number_to_return = null) * array|null GetPromotedStories($number_to_return = null) * array|null GetFriendsList($number_to_return = null) * * @param string $method The name of the method being called * @param array $args Arguments passed to the method * @return mixed */ public function __call($method, $args) { if (isset($args[0])) { $arg = $args[0]; } else { $arg = null; } return $this->GetUserData($method, $arg); } /** * Returns a value from the $user_profile_data array * * Given the index key for the $user_profile_data, returns the value * at that index, or NULL if the index key doesn't exists. * * @param string $key The index key * @return mixed */ public function GetUserData($key) { return (isset($this->user_profile_data[$key])) ? $this->user_profile_data[$key] : null; } /** * Returns an array of stories, or friends for the user * * Returns an array of stories, or friends for the user. Can return a * list of dugg stories, promoted stories, commented on stories, * submitted stories, or the list of friends for the user. * The number of array indixes to be returned can be limited with the * second parameter. Returns NULL if the index doesn't exists. * * @param string $key The key for the $user_profile_data array * @param int $number_to_return The number of array indixes to return * @return null */ public function GetStoryData($key, $number_to_return = null) { if (!isset($this->user_profile_data[$key]) || empty($this->user_profile_data[$key])) { return null; } $story_list = $this->user_profile_data[$key]; if ($number_to_return) { if ($number_to_return > count($story_list)) $number_to_return = count($story_list); return array_slice($story_list, 0, $number_to_return); } else { return $story_list; } }}class DiggSigAPIXMLRPCServer extends IXR_IntrospectionServer { private $DiggSigAPI; private $parse_pages = null; public function SetPages($parse_pages) { $this->parse_pages = $parse_pages; return true; } public function SetUserName($user_name) { $this->DiggSigAPI->SetUserName($user_name); return $this->DiggSigAPI->Parse($this->parse_pages); } public function GetRealName($user_name = null) { if ($user_name) $this->SetUserName($user_name); return $this->DiggSigAPI->GetRealName(); } public function GetMemberName($user_name = null) { if ($user_name) $this->SetUserName($user_name); return $this->DiggSigAPI->GetMemberName(); } public function GetAvatarURL($user_name = null) { if ($user_name) $this->SetUserName($user_name); return $this->DiggSigAPI->GetAvatarURL(); } public function GetLocation($user_name = null) { if ($user_name) $this->SetUserName($user_name); return $this->DiggSigAPI->GetLocation(); } public function GetRanking($user_name = null) { if ($user_name) $this->SetUserName($user_name); return $this->DiggSigAPI->GetRanking(); } public function GetNumberOfStoriesDugg($user_name = null) { if ($user_name) $this->SetUserName($user_name); return $this->DiggSigAPI->GetNumberOfStoriesDugg(); } public function GetNumberOfStoriesSubmitted($user_name = null) { if ($user_name) $this->SetUserName($user_name); return $this->DiggSigAPI->GetNumberOfStoriesSubmitted(); } public function GetNumberOfPromotedStories($user_name = null) { if ($user_name) $this->SetUserName($user_name); return $this->DiggSigAPI->GetNumberOfPromotedStories(); } public function GetNumberOfProfileViews($user_name = null) { if ($user_name) $this->SetUserName($user_name); return $this->DiggSigAPI->GetNumberOfProfileViews(); } public function GetLastDuggStories($args = null) { if ($args[0]) $this->SetUserName($args[0]); return $this->DiggSigAPI->GetLastDuggStories($args[1]); } public function GetSubmittedStories($args = null) { if ($args[0]) $this->SetUserName($args[0]); return $this->DiggSigAPI->GetSubmittedStories($args[1]); } public function GetCommentedStories($args = null) { if ($args[0]) $this->SetUserName($args[0]); return $this->DiggSigAPI->GetCommentedStories($args[1]); } public function GetPromotedStories($args = null) { if ($args[0]) $this->SetUserName($args[0]); return $this->DiggSigAPI->GetPromotedStories($args[1]); } public function GetFriendsList($args = null) { if ($args[0]) $this->SetUserName($args[0]); return $this->DiggSigAPI->GetFriendsList($args[1]); } public function __construct(DiggSigAPI $DiggSigAPI) { $this->DiggSigAPI = $DiggSigAPI; $this->addCallback ( 'digg.SetPages', 'this:SetPages', array('bool', 'string'), "Sets the pages to download from Digg" ); $this->addCallback ( 'digg.SetUserName', 'this:SetUserName', array('bool', 'string'), "Sets the user's Digg username" ); $this->addCallback ( 'digg.GetRealName', 'this:GetRealName', array('string', 'string'), "Returns the user's real name" ); $this->addCallback ( 'digg.GetMemberName', 'this:GetMemberName', array('string', 'string'), "Returns the user's member name" ); $this->addCallback ( 'digg.GetAvatarURL', 'this:GetAvatarURL', array('string', 'string'), "Returns the user's avatar URL" ); $this->addCallback ( 'digg.GetLocation', 'this:GetLocation', array('string', 'string'), "Returns the user's location" ); $this->addCallback ( 'digg.GetRanking', 'this:GetRanking', array('string', 'string'), "Returns the user's Digg ranking" ); $this->addCallback ( 'digg.GetNumberOfStoriesDugg', 'this:GetNumberOfStoriesDugg', array('string', 'string'), "Returns the number of stories dugg by the user" ); $this->addCallback ( 'digg.GetNumberOfStoriesSubmitted', 'this:GetNumberOfStoriesSubmitted', array('string', 'string'), "Returns the number of stories submitted by the user" ); $this->addCallback ( 'digg.GetNumberOfPromotedStories', 'this:GetNumberOfPromotedStories', array('string', 'string'), "Returns the number of stories submitted by the user that were promoted" ); $this->addCallback ( 'digg.GetNumberOfProfileViews', 'this:GetNumberOfProfileViews', array('string', 'string'), "Returns the number of times the user's profile was viewed" ); $this->addCallback ( 'digg.GetLastDuggStories', 'this:GetLastDuggStories', array('struct', 'string', 'int'), "Returns an array of the last 20 stories dugg by the user" ); $this->addCallback ( 'digg.GetSubmittedStories', 'this:GetSubmittedStories', array('struct', 'string', 'int'), "Returns an array of the last 20 stories submitted by the user" ); $this->addCallback ( 'digg.GetCommentedStories', 'this:GetCommentedStories', array('struct', 'string', 'int'), "Returns an array of the last 20 stories the user commented on" ); $this->addCallback ( 'digg.GetPromotedStories', 'this:GetPromotedStories', array('struct', 'string', 'int'), "Returns an array of the last 20 stories the user submitted that were promoted" ); $this->addCallback ( 'digg.GetFriendsList', 'this:GetFriendsList', array('struct', 'string', 'int'), "Returns an array of the last 20 friends added to the user's friends list" ); $this->serve(); }}$DiggSigAPIXMLRPCServer = new DiggSigAPIXMLRPCServer(new DiggSigAPI);?>