Index: modules/content_retriever/content_retriever.module =================================================================== --- modules/content_retriever/content_retriever.module (revision 1) +++ modules/content_retriever/content_retriever.module (working copy) @@ -1,6 +1,6 @@ 'Content Retriever Settings', @@ -35,7 +35,7 @@ 'type' => MENU_NORMAL_ITEM, 'file' => 'content_retriever.admin.inc', ); - + $items['admin/settings/retriever/settings'] = array( 'title' => 'Settings', 'description' => t('Settings for content retrieval over web services from a central distribution server.'), @@ -46,7 +46,7 @@ 'type' => MENU_DEFAULT_LOCAL_TASK, 'file' => 'content_retriever.admin.inc', ); - + $items['admin/settings/retriever/types'] = array( 'title' => 'Content Types', 'description' => t('Settings for content retrieval over web services from a central distribution server.'), @@ -56,7 +56,7 @@ 'type' => MENU_LOCAL_TASK, 'file' => 'content_retriever.admin.inc', ); - + $items['admin/settings/retriever/queues'] = array( 'title' => 'Nodequeues', 'description' => t('Nodequeues to interrogate on remote server to retrieve distributed content.'), @@ -66,7 +66,7 @@ 'type' => MENU_LOCAL_TASK, 'file' => 'content_retriever.admin.inc', ); - + //visual output for debug purposes $last_run = variable_get('content_retriever_last_run', (time() - (24 * 60 * 60)) ); $items['admin/settings/retriever/output'] = array( @@ -87,7 +87,7 @@ function content_retriever_theme() { return array( 'content_retriever_admin_page' => array( - 'arguments' => array(), + 'arguments' => array(), ), 'content_retriever_admin_types_page' => array( 'arguments' => array(), @@ -101,10 +101,10 @@ /** * parses and saves nodes sent back from the web service - * + * * @param $last_run * timestamp - Unix time of last time process executed - * + * * @return * string of HTML for debug purposes only */ @@ -116,7 +116,7 @@ $api_key = variable_get('content_retriever_api_key', ''); //check if screen reporting messages for debug are on or off $screen_reports = variable_get('content_retriever_screen_messages', 0); - + //fetch node data $nodes = _content_retriever_fetch_nodes($last_run); if ($nodes) { @@ -124,33 +124,33 @@ drupal_set_message('_content_retriever_fetch_nodes ran! No. of nodes: '.count($nodes)); } }else{ - if($screen_reports != 0){ - drupal_set_message(t('There are no new nodes to retrieve. Process complete.')); - } - return; + if($screen_reports != 0){ + drupal_set_message(t('There are no new nodes to retrieve. Process complete.')); + } + return; } if ($nodes) { foreach ($nodes as $node) { - // get a handle on the original_nid stored in the local database + // get a handle on the original_nid stored in the local database // to check if we already have a version of the incoming node $result = db_query("SELECT nid FROM {content_retriever} WHERE original_nid = %d",$node->nid); $nid = (int)db_result($result); - + //we need to be a superuser to do this next stuff $temp_user = $user; $user = user_load(array('uid' => 1)); - + //store remote nid $remote_nid = $node->nid; //unset the CMS node ID unset($node->nid); - + //if we have nodes carrying this CMS nid already, use that local ID so we update them as the NID if(!empty($nid)){ - $node->nid = $nid; + $node->nid = $nid; //need to get the latest revision id $result = db_query("SELECT MAX(vid) FROM {node_revisions} WHERE nid = %d",$node->nid); $vid = (int)db_result($result); @@ -158,35 +158,35 @@ // is a new revision $node->vid = $vid; $node->revision = 1; - + if($screen_reports != 0){ drupal_set_message(t('Updating node with nid: '. $nid)); } watchdog(t('integration'), t('Updating node with nid: %nid.'), array('%nid' => $nid), WATCHDOG_NOTICE); }else{ - unset($node->vid); - } - + unset($node->vid); + } + //unset anything unique to each instance of Drupal and save the node unset($node->created); unset($node->changed); - + unset($node->promote); unset($node->sticky); - + unset($node->revision_uid); unset($node->revision_timestamp); unset($node->field_related_items); - + unset($node->path); - + unset($node->comment_count); unset($node->last_comment_timestamp); unset($node->last_comment_name); - + //reset the teaser so a new, safe, teaser is created $node->teaser = node_teaser($node->body, $node->format); - + unset($node->queues); //taxonomy parsing - rewrites the taxonomy object @@ -194,94 +194,121 @@ if ($node->taxonomy) { // construct array of term names from the $node->taxonomy $term_categories = array(); - foreach($node->taxonomy as $term){ + foreach($node->taxonomy as $term){ $term_categories[$term['name']]['vocab'] = $node->vocab_names[$term['tid']]; } // parse the taxonomy and return taxonomy array to be attached to the node object - $node->taxonomy = taxonomy_parser_import_taxonomy($term_categories, $node->type); + $node->taxonomy = taxonomy_parser_import_taxonomy($term_categories, $node->type); } }else{ - // module is not installed so capture the issue - watchdog('integration', 'Failed to call taxonomy_parser_import_taxonomy(). The taxonomy_parser module is not available.', WATCHDOG_WARNING); - unset($node->taxonomy); + // module is not installed so capture the issue + watchdog('integration', 'Failed to call taxonomy_parser_import_taxonomy(). The taxonomy_parser module is not available.', WATCHDOG_WARNING); + unset($node->taxonomy); } - + + // nodereference + if (module_exists('nodereference')) { + $fields = content_fields(); + if (!empty($fields)) { + foreach($fields as $content_type_field){ + if ($content_type_field['type_name'] == $node->type && $content_type_field['module'] == 'nodereference') { + if(!empty($node->$content_type_field['field_name'])){ + $nr_data = $node->$content_type_field['field_name']; + if (isset($nr_data[0]['nid']) && $nr_data[0]['nid'] != '') { + $nr_newids = array(); + // Loop all values + foreach ($nr_data as $nr_key => $nr_item) { + // Try to find the new node id + $result = db_query("select nid, original_nid FROM {content_retriever} as c WHERE c.original_nid = %d", $nr_item['nid']); + $nr_new = (int)db_result($result); + if ($nr_new && $nr_new != 0) { + $nr_newids[] = array (nid => $nr_new); + } + } + $node->$content_type_field['field_name'] = $nr_newids; + } + } + } + } + } + } + //fetch imagefield / filefield files if((module_exists('imagefield')) || (module_exists('filefield'))){ - $fields = content_fields(); + $fields = content_fields(); $files_array = array(); //check the cck field exists on the local content type - + if (!empty($fields)) { foreach($fields as $content_type_field){ if(($content_type_field['type_name'] == $node->type) && (($content_type_field['module'] == 'imagefield')) || ($content_type_field['module'] == 'filefield')){ - if(!empty($node->$content_type_field['field_name'])){ + if(!empty($node->$content_type_field['field_name'])){ $files_array[$content_type_field['field_name']] = $node->$content_type_field['field_name']; - } - } + } + } } } - + $file_count = count($files_array); if ($file_count > 0) { // get a session id from system.service module $connect = content_retriever_xmlrpc($endpoint, 'system.connect', $api_key); - $session_id = $connect['sessid']; - - foreach($files_array as $field => $file_type){ + $session_id = $connect['sessid']; + + foreach($files_array as $field => $file_type){ $field_array = $node->$field; foreach($file_type as $key => $incoming_file){ $file = content_retriever_xmlrpc($endpoint, 'file.get', $api_key, $session_id, (int)$incoming_file['fid']); $file_details = _content_retriever_save_file($file, $node->nid, 'cck', null, $field, $node->type); - + $field_array[$key]['fid'] = $file_details->fid; - $field_array[$key]['filepath'] = $file_details->filepath; + $field_array[$key]['filepath'] = $file_details->filepath; } $node->$field = $field_array; } } - } - - //grab files array and unset files on the node before save + } + + //grab files array and unset files on the node before save $files_array = array(); $files_array = $node->files; unset($node->files); - - // save the node + + // save the node node_save($node); - + //format node->published_date to unix timestamp for DB entry. $published_date = $node->published_date; $published_date = date_convert($published_date, DATE_DATETIME, DATE_UNIX); - + // check whether we are inserting a new node or updating an existing one if(!empty($nid)){ - db_query("UPDATE {content_retriever} SET published_date = %d WHERE nid = %d", $published_date, $node->nid); + db_query("UPDATE {content_retriever} SET published_date = %d WHERE nid = %d", $published_date, $node->nid); }else{ - db_query("INSERT INTO {content_retriever} (nid, original_nid, published_date) VALUES (%d, %d, %d)", $node->nid, $remote_nid, $published_date); + db_query("INSERT INTO {content_retriever} (nid, original_nid, published_date) VALUES (%d, %d, %d)", $node->nid, $remote_nid, $published_date); } - + //fetch ordinary upload module attached files if (module_exists('upload')) { $file_count = count($files_array); if ($file_count > 0) { // get a session id from system.service module $connect = content_retriever_xmlrpc($endpoint, 'system.connect', $api_key); - $session_id = $connect['sessid']; + $session_id = $connect['sessid']; $files = content_retriever_xmlrpc($endpoint, 'file.getNodeFiles', $api_key, $session_id, $remote_nid); if(!empty($files)){ foreach ($files as $file) { _content_retriever_save_file($file, $node->nid, 'file', $node->vid); } }else{ - watchdog(t('integration'), t('Content retriever ran but could not retrieve files on node with node ID %node'), array('%node' => $node->nid), WATCHDOG_NOTICE); + watchdog(t('integration'), t('Content retriever ran but could not retrieve files on node with node ID %node'), array('%node' => $node->nid), WATCHDOG_NOTICE); } } } - + //now stop being the superuser $user = $temp_user; - + //tell watchdog the process has completed with content $output .= '
'. htmlspecialchars(print_r($node, true)) .'
'; } @@ -299,10 +326,10 @@ /** * retrieves and builds an array of nodes from the remote resource, * ready to be saved to the database - * + * * @param $last_run * timestamp - Unix time of last time process executed - * + * * @return * array containing retrieved node objects */ @@ -316,68 +343,68 @@ $api_key = variable_get('content_retriever_api_key', ''); $username = variable_get('content_retriever_webservice_username', ''); $password = variable_get('content_retriever_webservice_pass', ''); - + //warn sysadm via watchdog if no safe content types are found if (count($content_types) == 0) { watchdog('integration', t('Content Retriever found no valid content types - doing nothing.'), array(), WATCHDOG_WARNING); drupal_set_message(t('There are no available content types installed on this instance - cannot retrieve content'), 'error'); return false; } else { - $nodes = array(); - $processed_nids = array(); - - //we need to be logged in as webservice user with + $nodes = array(); + $processed_nids = array(); + + //we need to be logged in as webservice user with //'administer nodes' permissions to be able to //retrieve unpublished nodes via views.get service call $connect = content_retriever_xmlrpc($endpoint, 'system.connect', $api_key); $session_id = $connect['sessid']; - - //login as webservice user + + //login as webservice user $login = content_retriever_xmlrpc($endpoint, 'user.login', $api_key, $session_id, $username, $password); $session_id = $login['sessid']; - + foreach ($content_types as $contenttype) { - + foreach ($queues as $qid => $queue) { - if ($queue != '0') { - //make the call + if ($queue != '0') { + //make the call $result = content_retriever_xmlrpc($endpoint, 'views.get', $api_key, $session_id, $viewname, array((int)$qid, $contenttype, format_date($last_run, 'custom', 'YmdHi'))); if($result) { foreach ($result as $node_info) { //load the remote node $node = _content_retriever_fetch_node($node_info['nid'], $session_id); - //check we haven't already got this node in the $nodes array from another queue + //check we haven't already got this node in the $nodes array from another queue foreach($nodes as $processed_node) { $processed_nids[$processed_node->nid] = $processed_node->nid; - } - if (array_search($node->nid, $processed_nids) === false) { + } + if (array_search($node->nid, $processed_nids) === false) { $nodes[] = $node; - } - $processed_nids = array(); + } + $processed_nids = array(); } } } - } + } } // logout as webservice user from remote site content_retriever_xmlrpc($endpoint, 'user.logout', $api_key, $session_id); return $nodes; } - + } /** * retrieves a specific node from the remote resource, * ready to be saved to the database - * + * * @param $nid * integer node id - * + * * @param $fields * optional array of CCK field names - * + * * @return * node object */ @@ -386,9 +413,9 @@ //variables are set in the admin form submit function $endpoint = variable_get('content_retriever_endpoint', 'http://cms.defaqto.com/services/xmlrpc'); $api_key = variable_get('content_retriever_api_key', ''); - + $result = content_retriever_xmlrpc($endpoint, 'node.get', $api_key, $session_id, (int)$nid, $fields); - + //convert returned array in to a proper object if ($result) { $node = new stdClass(); @@ -399,21 +426,21 @@ } else { return false; } - + } /** * @param $file * array of file data - * + * * @param $nid * integer: the NEW node id on the local instance - * + * * @param $type * string denoting type of file, possible values: - * - image - * - file + * - image + * - file */ function _content_retriever_save_file($file, $nid, $type, $vid = null, $field = '', $node_type = '') { $file_data = base64_decode($file['file']); @@ -421,18 +448,18 @@ case 'file': //set filepath to save to $file['filepath'] = file_directory_path().'/'.$file['filename']; - + //check if the file is already on the system, if it is no need to save it again $result = db_query("SELECT fid FROM {files} WHERE filename = '%s' AND timestamp = %d", $file['filename'], $file['timestamp']); $fid_int = (int)db_result($result); - + if(!$fid_int){ //save file //note, safe to REPLACE because we can be certain we will not have duplicates //we wish to keep - distirbution site will be handling duplicate file names //so we will only ever receive unique ones here file_save_data($file_data, $file['filepath'], FILE_EXISTS_REPLACE); - + //write record in files table drupal_write_record('files', $file); @@ -440,7 +467,7 @@ $result = db_query("SELECT fid FROM {files} WHERE filename = '%s'", $file['filename']); $fid_int = (int)db_result($result); } - + //build array for upload table insertion $upload = array( 'fid' => $fid_int, @@ -453,63 +480,63 @@ //write record in upload table drupal_write_record('upload', $upload); - + return true; break; - + case 'cck': //set filepath to save to $file['filepath'] = file_directory_path().'/'.$file['filename']; - + //set content type db table, and field to search on. $content_table = 'content_type_'.$node_type; $fid_field = $field.'_fid'; - + $file_details = array(); //get timestamp and fid if there is a file on this node $result = db_query("SELECT f.timestamp, f.fid, f.filepath FROM {files} as f - RIGHT JOIN {%s} as ctype - ON f.fid=ctype.%s - WHERE ctype.vid = (SELECT MAX({%s}.vid) - FROM {%s}) AND ctype.nid = %d", $content_table, $fid_field, $content_table, $content_table, $nid); - - + RIGHT JOIN {%s} as ctype + ON f.fid=ctype.%s + WHERE ctype.vid = (SELECT MAX({%s}.vid) + FROM {%s}) AND ctype.nid = %d", $content_table, $fid_field, $content_table, $content_table, $nid); + + $file_details = db_fetch_object($result); - + //check if it's the same file coming over - if($file_details->timestamp == $file['timestamp']){ - //if it is it dont write another record in the files table. - return $file_details; - }else{ + if($file_details->timestamp == $file['timestamp']){ + //if it is it dont write another record in the files table. + return $file_details; + }else{ //save file //note, safe to REPLACE because we can be certain we will not have duplicates //we wish to keep - distribution site will be handling duplicate file names //so we will only ever receive unique ones here file_save_data($file_data, $file['filepath'], FILE_EXISTS_REPLACE); - + //write record in files table drupal_write_record('files', $file); - - $result = db_query("SELECT fid,filepath FROM {files} WHERE filename = '%s'", $file['filename']); + + $result = db_query("SELECT fid,filepath FROM {files} WHERE filename = '%s'", $file['filename']); $file_details = db_fetch_object($result); - + return $file_details; - } + } break; } - + } /** * abstraction of the service calling code needed to retrieve nodes - * + * * makes and xml-rpc call to the Services module of another Drupal * installation, calling the views.get method, and returns the * resulting array - * + * * @param $endpoint * string representing the valid URI for the web service endpoint - * + * * @param $method * string telling us what method to call, examples: * - views.get @@ -517,17 +544,17 @@ * - nodequeue.getQueues * - node.get * - image.getNodeImages - * + * * @param $api_key * string containing a valid API key for this service - * + * * @param $param1 thru $param3 * various: placeholder vars for passing parameters to the web * service methods - can be anything valid for the method call being - * invoked - empty strings by default - * + * invoked - empty strings by default + * * @return - * array containing results of the call + * array containing results of the call */ function content_retriever_xmlrpc($endpoint, $method, $api_key, $session_id = '',$param1 = '', $param2 = '', $param3 = '') { //check if HTTP requests are enabled and report error if not @@ -535,10 +562,10 @@ drupal_set_message(t('Drupal is unable to make HTTP requests. Please reset the HTTP request status.'), 'error', FALSE); watchdog('integration', t('Drupal is unable to make HTTP requests. Please reset the HTTP request status.'), array(), WATCHDOG_CRITICAL); } else { - - if($api_key != ''){ - //use api key to get a hash code for the service. - $timestamp = (string) strtotime("now"); + + if($api_key != ''){ + //use api key to get a hash code for the service. + $timestamp = (string) strtotime("now"); $domain = $_SERVER['SERVER_NAME']; $nonce = content_retriever_get_unique_code('10'); $hash_parameters = array( @@ -548,104 +575,104 @@ $method, ); $hash = hash_hmac("sha256", implode(';', $hash_parameters), $api_key); - } - - switch ($method) { - case 'system.connect': + } + + switch ($method) { + case 'system.connect': $result = xmlrpc($endpoint, $method); - break; - case 'user.login': - //check for an API key - if not supplied, presume not required - if ($api_key == '') { + break; + case 'user.login': + //check for an API key - if not supplied, presume not required + if ($api_key == '') { $result = xmlrpc($endpoint, $method, $session_id, (string) $param1, (string) $param2); } else { $result = xmlrpc($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id, (string) $param1, (string) $param2); } - break; - case 'user.logout': - //check for an API key - if not supplied, presume not required - if ($api_key == '') { + break; + case 'user.logout': + //check for an API key - if not supplied, presume not required + if ($api_key == '') { $result = xmlrpc($endpoint, $method, $session_id); } else { $result = xmlrpc($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id); } - break; - case 'views.get': - //check for an API key - if not supplied, presume not required + break; + case 'views.get': + //check for an API key - if not supplied, presume not required if ($api_key == '') { - //$param1 is view name, $param2 is array of view arguments - //'default' is display_id parameter hardcoded to default. + //$param1 is view name, $param2 is array of view arguments + //'default' is display_id parameter hardcoded to default. $result = xmlrpc($endpoint, $method, $session_id, (string) $param1, 'default', array(), (array) $param2, 0, 0); } else { //var_dump($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id, $param1, $param2); $result = xmlrpc($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id, (string) $param1, 'default', array(), (array) $param2, 0, 0); } - break; - - case 'file.getNodeFiles': - //check for an API key - if not supplied, presume not required + break; + + case 'file.getNodeFiles': + //check for an API key - if not supplied, presume not required if ($api_key == '') { //$param1 is nid $result = xmlrpc($endpoint, $method, $session_id, (int) $param1); } else { $result = xmlrpc($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id, (int) $param1); } - break; - - case 'file.get': - //check for an API key - if not supplied, presume not required + break; + + case 'file.get': + //check for an API key - if not supplied, presume not required if ($api_key == '') { //$param1 is fid $result = xmlrpc($endpoint, $method, $session_id, (int) $param1); } else { $result = xmlrpc($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id, (int) $param1); } - break; - - case 'nodequeue.getQueues': - //check for an API key - if not supplied, presume not required + break; + + case 'nodequeue.getQueues': + //check for an API key - if not supplied, presume not required if ($api_key == '') { $result = xmlrpc($endpoint, $method, $session_id); } else { $result = xmlrpc($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id); } - break; - - case 'node.get': - //check for an API key - if not supplied, presume not required + break; + + case 'node.get': + //check for an API key - if not supplied, presume not required if ($api_key == '') { //$param1 is nid, $param2 is array of cck field names $result = xmlrpc($endpoint, $method, $session_id, (int) $param1, (array) $param2); } else { $result = xmlrpc($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id, (int) $param1, (array) $param2); } - break; - - case 'image.getNodeImages': - //check for an API key - if not supplied, presume not required + break; + + case 'image.getNodeImages': + //check for an API key - if not supplied, presume not required if ($api_key == '') { //$param1 is nid $result = xmlrpc($endpoint, $method, $session_id, (int) $param1); } else { $result = xmlrpc($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id, (int) $param1); } - break; - - case 'node.getAllTypes': - //check for an API key - if not supplied, presume not required + break; + + case 'node.getAllTypes': + //check for an API key - if not supplied, presume not required if ($api_key == '') { //$param1 is nid $result = xmlrpc($endpoint, $method, $session_id); } else { $result = xmlrpc($endpoint, $method, $hash, $domain, $timestamp, $nonce, $session_id); } - break; - - default: - $result = array(); - watchdog('integration', t('Unknown method call attempted.'), array(), WATCHDOG_NOTICE); - break; - } + break; + + default: + $result = array(); + watchdog('integration', t('Unknown method call attempted.'), array(), WATCHDOG_NOTICE); + break; + } return $result; } } @@ -666,7 +693,7 @@ /** * Function extending the available page variables - * + * * This has been implemented to expose the last time the content * retriever has ran to page.tpl. This is purely for admin purposes, * so they can simply visit the homepage of any site retrieving content @@ -680,8 +707,8 @@ /** * Implementation of hook_nodeapi(). - * - * delete the record from the content_retriever table when a node + * + * delete the record from the content_retriever table when a node * is deleted. */ function content_retriever_nodeapi(&$node, $op, $teaser, $page) { @@ -696,6 +723,6 @@ function content_retriever_get_unique_code($length = ''){ $code = md5(uniqid(rand(), true)); if ($length != "") return substr($code, 0, $length); - else return $code; + else return $code; }