ftimme
2013-02-27 f4038a2160d55a7f519a3b42be1aa96e29e9a908
interface/lib/classes/aps_crawler.inc.php
@@ -34,6 +34,9 @@
class ApsCrawler extends ApsBase
{
   //public $app_download_url_list = array();
   /**
    * Constructor
    *
@@ -51,6 +54,8 @@
     */
    private function checkRequirements()
    {
        global $app;
        try
        {
            // Check if allow_url_fopen is enabled
@@ -58,7 +63,7 @@
            // Check if the cURL module is available
            if(!function_exists('curl_version')) throw new Exception('cURL is not available');
            
            // Check if used folders are writable (chmod 777)
            // Check if used folders are writable
            if($this->interface_mode)
            {
                if(!is_writable($this->interface_pkg_dir)) 
@@ -74,7 +79,7 @@
        }
        catch(Exception $e)
        {
            $this->app->log($this->log_prefix.'Aborting execution because '.$e->getMessage(), LOGLEVEL_ERROR);
            $app->log($this->log_prefix.'Aborting execution because '.$e->getMessage(), LOGLEVEL_ERROR);
            return false;
        }
    }
@@ -93,7 +98,7 @@
            foreach($files as $file)
            {
                if($file != '.' && $file != '..')
                    if(filetype($dir.'/'.$file) == 'dir') rrmdir($dir.'/'.$file);
                    if(filetype($dir.'/'.$file) == 'dir') $this->removeDirectory($dir.'/'.$file);
                    else @unlink($dir.'/'.$file);
            }
            reset($files);
@@ -197,11 +202,12 @@
     * A method to build query URLs out of a list of vendors
     *
    */
    private function formatVendorCallback(&$array_item, $key)
    private function formatVendorCallback($array_item)
    {
        $array_item = str_replace(' ', '%20', $array_item);
        $array_item = str_replace('http://', '', $array_item);
        $array_item = '/'.$this->aps_version.'.atom?vendor='.$array_item.'&pageSize=100';
      return($array_item);
    }
    
    /**
@@ -209,6 +215,8 @@
     */    
    public function startCrawler() 
    {
        global $app;
        try
        {
            // Make sure the requirements are given so that this script can execute
@@ -216,7 +224,7 @@
            if(!$req_ret) return false;
            
            // Execute the open task and first fetch all vendors (APS catalog API 1.1, p. 12)
            $this->app->log($this->log_prefix.'Fetching data from '.$this->fetch_url);
            $app->log($this->log_prefix.'Fetching data from '.$this->fetch_url);
            $vendor_page = $this->fetchPage('/all-app/'); //$vendor_page = $this->fetchPage('/'.$this->aps_version.'/');
            preg_match_all("/\<a href=\"(.+)\/\" class=\"vendor\"/i", $vendor_page, $matches);
@@ -224,7 +232,12 @@
            if(!$vendors) throw new Exception('Unable to fetch vendors. Aborting');
            // Format all vendors for further processing (i.e. typo3.org -> /1.atom?vendor=typo3.org&pageSize=100
            array_walk($vendors, array($this, 'formatVendorCallback'));
            //array_walk($vendors, array($this, 'formatVendorCallback'));
         if(is_array($vendors)) {
            foreach($vendors as $key => $array_item) {
               $vendors[$key] = $this->formatVendorCallback($array_item);
            }
         }
            
            // Process all vendors in chunks of 50 entries
            $vendor_chunks = array_chunk($vendors, 50);
@@ -233,7 +246,7 @@
            // Get all known apps from the database and the highest known version
            // Note: A dirty hack is used for numerical sorting of the VARCHAR field Version: +0 -> cast
            // A longer but typesafe way would be: ORDER BY CAST(REPLACE(Version, '.', '') AS UNSIGNED) DESC
            $existing_apps = $this->db->queryAllRecords("SELECT * FROM (
            $existing_apps = $app->db->queryAllRecords("SELECT * FROM (
                SELECT name AS Name, CONCAT(version, '-', CAST(`release` AS CHAR)) AS CurrentVersion 
                FROM aps_packages ORDER BY REPLACE(version, '.', '')+0 DESC, `release` DESC
                ) as Versions GROUP BY name");
@@ -270,27 +283,30 @@
                        
                        // Find out a (possibly) existing package version
                        $ex_ver = '';
                  /*
                        array_walk($existing_apps, 
                            create_function('$v, $k, $ex_ver', 'if($v["Name"] == "'.$app_name.'") $ex_ver = $v["CurrentVersion"];'), &$ex_ver);
                        */
                  if(is_array($existing_apps)) {
                     foreach($existing_apps as $k => $v) {
                        if($v["Name"] == $app_name) $ex_ver = $v["CurrentVersion"];
                     }
                  }
                        $new_ver = $app_version.'-'.$app_release;
                        $local_intf_folder = $this->interface_pkg_dir.'/'.$app_name.'-'.$new_ver.'.app.zip/';
                        // Proceed if a newer or at least equal version has been found with server mode or 
                        // interface mode is activated and there's no valid APP-META.xml existing yet
                        if((!$this->interface_mode && version_compare($new_ver, $ex_ver) >= 0)
                         || ($this->interface_mode
                              && (!file_exists($local_intf_folder.'APP-META.xml') || filesize($local_intf_folder.'APP-META.xml') == 0)
                            )
                        )
                        {
                        // interface mode is activated and there are no valid APP-META.xml and PKG_URL existing yet
                        if((!$this->interface_mode && version_compare($new_ver, $ex_ver) >= 0) || ($this->interface_mode && (!file_exists($local_intf_folder.'APP-META.xml') || filesize($local_intf_folder.'APP-META.xml') == 0 || !file_exists($local_intf_folder.'PKG_URL') || filesize($local_intf_folder.'PKG_URL') == 0))){
                            // Check if we already have an old version of this app
                            if(!empty($ex_ver) && version_compare($new_ver, $ex_ver) == 1) $apps_updated++; 
                            $app_dl = parent::getXPathValue($sxe, "entry[position()=1]/link[@a:type='aps']/@href");
                            $app_filesize = parent::getXPathValue($sxe, "entry[position()=1]/link[@a:type='aps']/@length");
                            $app_metafile = parent::getXPathValue($sxe, "entry[position()=1]/link[@a:type='meta']/@href");
                     //$this->app_download_url_list[$app_name.'-'.$new_ver.'.app.zip'] = $app_dl;
                            // Skip ASP.net packages because they can't be used at all
                            $asp_handler = parent::getXPathValue($sxe, '//aspnet:handler');
                            $asp_permissions = parent::getXPathValue($sxe, '//aspnet:permissions');
@@ -307,19 +323,22 @@
                                    if(file_exists($old_folder)) $this->removeDirectory($old_folder);
                                    
                           /*
                                    $this->db->query("UPDATE aps_packages SET package_status = '".PACKAGE_OUTDATED."' WHERE name = '".
                                        $this->db->quote($app_name)."' AND CONCAT(version, '-', CAST(`release` AS CHAR)) = '".
                                        $this->db->quote($ex_ver)."';");
                                    $app->db->query("UPDATE aps_packages SET package_status = '".PACKAGE_OUTDATED."' WHERE name = '".
                                        $app->db->quote($app_name)."' AND CONCAT(version, '-', CAST(`release` AS CHAR)) = '".
                                        $app->db->quote($ex_ver)."';");
                           */
                           $tmp = $this->db->queryOneRecord("SELECT id FROM aps_packages WHERE name = '".
                                        $this->db->quote($app_name)."' AND CONCAT(version, '-', CAST(`release` AS CHAR)) = '".
                                        $this->db->quote($ex_ver)."';");
                           $this->db->datalogUpdate('aps_packages', "package_status = ".PACKAGE_OUTDATED, 'id', $tmp['id']);
                           $tmp = $app->db->queryOneRecord("SELECT id FROM aps_packages WHERE name = '".
                                        $app->db->quote($app_name)."' AND CONCAT(version, '-', CAST(`release` AS CHAR)) = '".
                                        $app->db->quote($ex_ver)."';");
                           $app->db->datalogUpdate('aps_packages', "package_status = ".PACKAGE_OUTDATED, 'id', $tmp['id']);
                           unset($tmp);
                                }
                                
                                // Create the local folder if not yet existing
                                if(!file_exists($local_intf_folder)) @mkdir($local_intf_folder, 0777, true);
                        // Save the package URL in an extra file because it's not part of the APP-META.xml file
                        @file_put_contents($local_intf_folder.'PKG_URL', $app_dl);
                                
                                // Download the meta file
                                $local_metafile = $local_intf_folder.'APP-META.xml';
@@ -428,19 +447,19 @@
                    if($apps_to_dl_chunks[$i][$j]['filesize'] != 0 &&
                       $apps_to_dl_chunks[$i][$j]['filesize'] != filesize($apps_to_dl_chunks[$i][$j]['localtarget']))
                    {
                            $this->app->log($this->log_prefix.' The filesize of the package "'.
                            $app->log($this->log_prefix.' The filesize of the package "'.
                                $apps_to_dl_chunks[$i][$j]['name'].'" is wrong. Download failure?', LOGLEVEL_WARN);
                    }
                }
            }
            
            $this->app->log($this->log_prefix.'Processed '.$apps_in_repo.
            $app->log($this->log_prefix.'Processed '.$apps_in_repo.
                ' apps from the repo. Downloaded '.$apps_updated.
                ' updates, '.$apps_downloaded.' new apps');
        }
        catch(Exception $e)
        {
            $this->app->log($this->log_prefix.$e->getMessage(), LOGLEVEL_ERROR);
            $app->log($this->log_prefix.$e->getMessage(), LOGLEVEL_ERROR);
            return false;
        }
    }
@@ -454,9 +473,11 @@
     */
    public function parseFolderToDB()
    {
        global $app;
        try
        {
            // This method must be used in server mode
            // This method must be used in interface mode
            if(!$this->interface_mode) return false; 
            
            $pkg_list = array();
@@ -473,15 +494,16 @@
            
            // Get registered packages and mark non-existant packages with an error code to omit the install
            $existing_packages = array();
            $path_query = $this->db->queryAllRecords('SELECT path AS Path FROM aps_packages;');
            $path_query = $app->db->queryAllRecords('SELECT path AS Path FROM aps_packages;');
            foreach($path_query as $path) $existing_packages[] = $path['Path']; 
            $diff = array_diff($existing_packages, $pkg_list);
            foreach($diff as $todelete)
                /*$this->db->query("UPDATE aps_packages SET package_status = '".PACKAGE_ERROR_NOMETA."'
                    WHERE path = '".$this->db->quote($todelete)."';");*/
            $tmp = $this->db->queryOneRecord("SELECT id FROM aps_packages WHERE path = '".$this->db->quote($todelete)."';");
            $this->db->datalogUpdate('aps_packages', "package_status = ".PACKAGE_ERROR_NOMETA, 'id', $tmp['id']);
            foreach($diff as $todelete) {
                /*$app->db->query("UPDATE aps_packages SET package_status = '".PACKAGE_ERROR_NOMETA."'
                    WHERE path = '".$app->db->quote($todelete)."';");*/
            $tmp = $app->db->queryOneRecord("SELECT id FROM aps_packages WHERE path = '".$app->db->quote($todelete)."';");
            $app->db->datalogUpdate('aps_packages', "package_status = ".PACKAGE_ERROR_NOMETA, 'id', $tmp['id']);
            unset($tmp);
         }
            
            // Register all new packages
            $new_packages = array_diff($pkg_list, $existing_packages);
@@ -491,7 +513,7 @@
                $metafile = $this->interface_pkg_dir.'/'.$pkg.'/APP-META.xml';
                if(!file_exists($metafile)) 
                {
                    $this->app->log($this->log_prefix.'Cannot read metadata from '.$pkg, LOGLEVEL_ERROR);
                    $app->log($this->log_prefix.'Cannot read metadata from '.$pkg, LOGLEVEL_ERROR);
                    continue;
                }
        
@@ -506,27 +528,63 @@
                $pkg_category = parent::getXPathValue($sxe, '//category');
                $pkg_version = parent::getXPathValue($sxe, 'version');
                $pkg_release = parent::getXPathValue($sxe, 'release');
            //$pkg_url = $this->app_download_url_list[$pkg];
            $pkg_url = @file_get_contents($this->interface_pkg_dir.'/'.$pkg.'/PKG_URL');
                
            /*
                $this->db->query("INSERT INTO `aps_packages`
                $app->db->query("INSERT INTO `aps_packages`
                    (`path`, `name`, `category`, `version`, `release`, `package_status`) VALUES 
                    ('".$this->db->quote($pkg)."', '".$this->db->quote($pkg_name)."',
                    '".$this->db->quote($pkg_category)."', '".$this->db->quote($pkg_version)."',
                    ".$this->db->quote($pkg_release).", ".PACKAGE_ENABLED.");");
                    ('".$app->db->quote($pkg)."', '".$app->db->quote($pkg_name)."',
                    '".$app->db->quote($pkg_category)."', '".$app->db->quote($pkg_version)."',
                    ".$app->db->quote($pkg_release).", ".PACKAGE_ENABLED.");");
            */
            // Insert only if data is complete
            if($pkg != '' && $pkg_name != '' && $pkg_category != '' && $pkg_version != '' && $pkg_release != '' && $pkg_url){
               $insert_data = "(`path`, `name`, `category`, `version`, `release`, `package_url`, `package_status`) VALUES
                    ('".$app->db->quote($pkg)."', '".$app->db->quote($pkg_name)."',
                    '".$app->db->quote($pkg_category)."', '".$app->db->quote($pkg_version)."',
                    ".$app->db->quote($pkg_release).", '".$app->db->quote($pkg_url)."', ".PACKAGE_ENABLED.");";
            
            $insert_data = "(`path`, `name`, `category`, `version`, `release`, `package_status`) VALUES
                    ('".$this->db->quote($pkg)."', '".$this->db->quote($pkg_name)."',
                    '".$this->db->quote($pkg_category)."', '".$this->db->quote($pkg_version)."',
                    ".$this->db->quote($pkg_release).", ".PACKAGE_ENABLED.");";
            $app->db->datalogInsert('aps_packages', $insert_data, 'id');
               $app->db->datalogInsert('aps_packages', $insert_data, 'id');
            } else {
               if(file_exists($this->interface_pkg_dir.'/'.$pkg)) $this->removeDirectory($this->interface_pkg_dir.'/'.$pkg);
            }
            }
        }
        catch(Exception $e)
        {
            $this->app->log($this->log_prefix.$e->getMessage(), LOGLEVEL_ERROR);
         $this->app->error($e->getMessage());
            $app->log($this->log_prefix.$e->getMessage(), LOGLEVEL_ERROR);
         $app->error($e->getMessage());
            return false;
        }
    }
   /**
     * Add missing package URLs to database
     */
    public function fixURLs()
    {
        global $app;
        try
        {
            // This method must be used in interface mode
            if(!$this->interface_mode) return false;
            $incomplete_pkgs = $app->db->queryAllRecords("SELECT * FROM aps_packages WHERE package_url = ''");
         if(is_array($incomplete_pkgs) && !empty($incomplete_pkgs)){
            foreach($incomplete_pkgs as $incomplete_pkg){
               $pkg_url = @file_get_contents($this->interface_pkg_dir.'/'.$incomplete_pkg['path'].'/PKG_URL');
               if($pkg_url != ''){
                  $app->db->datalogUpdate('aps_packages', "package_url = '".$pkg_url."'", 'id', $incomplete_pkg['id']);
               }
            }
         }
        }
        catch(Exception $e)
        {
            $app->log($this->log_prefix.$e->getMessage(), LOGLEVEL_ERROR);
         $app->error($e->getMessage());
            return false;
        }
    }