Added drush update function to update publications

tripal_pub/api/tripal_pub.api.inc

@@ -12,8 +12,11 @@
  *  keys correspond directly with Tripal Pub CV terms.
  *  
  * @param remote_db
- *    The name of the remote publication database to query. Valid values
- *    include: 'pubmed'.
+ *    The name of the remote publication database to query. These names should
+ *    match the name of the databases in the Chado 'db' table. Currently 
+ *    supported databass include
+ *      'PMID':  PubMed
+ *      
  * @param search_array
  *    An associate array containing the search criteria. The following key 
  *    are expected
@@ -70,6 +73,97 @@ function tripal_pub_get_remote_search_results($remote_db, $search_array,
   
   return $pubs;  
 }
+/*
+ * @ingroup tripal_pub_api
+ */
+function tripal_pub_update_publications() {
+  // get a persistent connection
+  $connection = tripal_db_persistent_chado();
+  if (!$connection) {
+     print "A persistant connection was not obtained. Loading will be slow\n";
+  }
+          
+  // if we cannot get a connection then let the user know the loading will be slow
+  tripal_db_start_transaction();
+  if ($connection) {
+     print "\nNOTE: Loading of publications is performed using a database transaction. \n" .
+           "If the load fails or is terminated prematurely then the entire set of \n" .
+           "insertions/updates is rolled back and will not be found in the database\n\n";
+  }
+  
+  // get a list of all publications that have 
+  // supported databases
+  $sql = "
+    SELECT DB.name as db_name, DBX.accession
+    FROM pub P
+      INNER JOIN pub_dbxref PDBX ON P.pub_id = PDBX.pub_id
+      INNER JOIN dbxref DBX      ON DBX.dbxref_id = PDBX.dbxref_id
+      INNER JOIN db DB           ON DB.db_id = DBX.db_id
+    ORDER BY DB.name
+  ";
+  $results = chado_query($sql);
+  
+  $num_to_retrieve = 100;
+  $i = 0;                 // count the number of IDs. When we hit $num_to_retrieve we'll do the query
+  $curr_db = '';          // keeps track of the current current database
+  $ids = array();         // the list of IDs for the database
+  $search = array();      // the search array passed to the search function
+  
+  // iterate through the pub IDs
+  while ($pub = db_fetch_object($results)) {    
+    $accession = $pub->accession;
+    $remote_db = $pub->db_name;
+    
+    // if we're switching databases then reset the search array    
+    if($remote_db != $curr_db) {
+      // if we had a previous DB then do the update.
+      if ($curr_db) {
+        $search['num_criteria'] = $i - 1;
+        $pubs = tripal_pub_get_remote_search_results($remote_db, $search, $i, 0);
+        tripal_pub_add_publications($pubs); 
+      }
+      $curr_db = $remote_db;      
+      $search = array(
+        'remote_db' => $remote_db,
+        'criteria' => array(),
+      );
+      $ids = array();
+      $i = 0;
+    }
+    
+    // if we've hit the maximum number to retrieve then do the search
+    if($i == $num_to_retrieve) {      
+      $search['num_criteria'] = $i - 1;
+      $pubs = tripal_pub_get_remote_search_results($remote_db, $search, $i, 0);
+      tripal_pub_add_publications($pubs);
+      $search['criteria'] = array();
+      $i = 0;
+    }
+    
+    // add each accession to the search criteria
+    $search['criteria'][] = array(
+      'search_terms' => $accession,
+      'scope' => 'id',
+      'operation' => 'OR'
+    );
+    $i++;   
+  }
+  // now update any remaining in the search criteria array
+  $search['num_criteria'] = $i - 1;
+  $pubs = tripal_pub_get_remote_search_results($remote_db, $search, $i, 0);
+  tripal_pub_add_publications($pubs);
+  
+  // sync the newly added publications with Drupal
+  print "Syncing publications with Drupal...\n";
+  tripal_pub_sync_pubs();
+  print "Syncing contacts with Drupal...\n";
+  tripal_contact_sync_contacts();
+  
+  // transaction is complete
+  tripal_db_commit_transaction();
+  
+  print "Done.\n";
+}
 /*
  * @ingroup tripal_pub_api
  */
@@ -101,25 +195,8 @@ function tripal_pub_import_publications() {
      $criteria = unserialize($import->criteria); 
      $remote_db = $criteria['remote_db'];
      do {       
-       // get the number of records
-       
        // retrieve the pubs for this page. We'll retreive 10 at a time
-       $pubs = tripal_pub_get_remote_search_results($remote_db, $criteria, $num_to_retrieve, $pager_id, $page);
-       
-       // now add the publications
-       foreach ($pubs as $pub) {
-               
-         // add the publication to Chado and sync it with Chado
-         $pub_id = tripal_pub_add_publication($pub);
-         
-         // add the publication cross reference (e.g. to PubMed)
-         if ($pub_id) {         
-           $pub_dbxref = tripal_pub_add_pub_dbxref($pub_id, $pub);
-         }                                      
-
-         $num_pubs++;
-         print $num_pubs . ".  " . $pub['Publication Database'] . ' ' . $pub['Pub Accession'] . "\n";                          
-       } // end for loop       
+       $pubs = tripal_pub_get_remote_search_results($remote_db, $criteria, $num_to_retrieve, $pager_id, $page);             
        $page++;
      } 
      // continue looping until we have a $pubs array that does not have
@@ -138,7 +215,26 @@ function tripal_pub_import_publications() {
   
   print "Done.\n";
 }
-
+/*
+ * 
+ */
+function tripal_pub_add_publications($pubs) {
+  
+  // iterate through the publications and add each one
+  foreach ($pubs as $pub) {
+         
+    // add the publication to Chado and sync it with Chado
+    $pub_id = tripal_pub_add_publication($pub);
+   
+    // add the publication cross reference (e.g. to PubMed)
+    if ($pub_id) {         
+      $pub_dbxref = tripal_pub_add_pub_dbxref($pub_id, $pub);
+    }                                      
+  
+    $num_pubs++;
+    print $num_pubs . ".  " . $pub['Publication Database'] . ' ' . $pub['Pub Accession'] . "\n";                          
+  } // end for loop        
+}
 /*
  * 
  */

tripal_pub/includes/pubmed.inc

@@ -3,14 +3,14 @@
  * @file
  * Tripal Pub PubMed Interface
  *
- * @defgroup tripal_pub_pubmed PubMed Interface
+ * @defgroup tripal_pub_PMID PubMed Interface
  * @ingroup tripal_pub
  */
 
 /**
  *
  */
-function tripal_pub_remote_search_pubmed($search_array, $num_to_retrieve, $pager_id) {
+function tripal_pub_remote_search_PMID($search_array, $num_to_retrieve, $pager_id) {
   
   // convert the terms list provicded by the caller into a string with words
   // separated by a '+' symbol.
@@ -38,14 +38,17 @@ function tripal_pub_remote_search_pubmed($search_array, $num_to_retrieve, $pager
     elseif($scope == 'abstract') {
       $search_str .= '[Title/Abstract]';
     }
+    elseif($scope == 'id') {
+      $search_str .= '[Uid]';
+    }
     $search_str .= ')'; 
   }  
   $search_array['limit'] = $num_to_retrieve;
   $search_array['search_terms'] = $search_str;
 
   // we want to get the list of pubs using the search terms but using a Drupal style pager
-  $pubs = tripal_pager_callback('tripal_pub_pubmed_range',  
-    $num_to_retrieve, $pager_id, 'tripal_pub_pubmed_count', $search_array);
+  $pubs = tripal_pager_callback('tripal_pub_PMID_range',  
+    $num_to_retrieve, $pager_id, 'tripal_pub_PMID_count', $search_array);
  
   return $pubs;
 }
@@ -55,15 +58,15 @@ function tripal_pub_remote_search_pubmed($search_array, $num_to_retrieve, $pager
  * tripal_pager_callback function.  This function returns a count of
  * the dataset to be paged.
  */
-function tripal_pub_pubmed_count($search_array) {
+function tripal_pub_PMID_count($search_array) {
   $terms = $search_array['search_terms'];
-  $days = $search_array['days'];
+  $days  = $search_array['days'];
   $limit = $search_array['limit'];
   
-  $results = tripal_pub_pubmed_search_init($terms, $limit, $days);
-  $_SESSION['tripal_pub_pubmed_query'][$terms]['Count'] = $results['Count'];
-  $_SESSION['tripal_pub_pubmed_query'][$terms]['WebEnv'] = $results['WebEnv'];
-  $_SESSION['tripal_pub_pubmed_query'][$terms]['QueryKey'] = $results['QueryKey'];
+  $results = tripal_pub_PMID_search_init($terms, $limit, $days);
+  $_SESSION['tripal_pub_PMID_query'][$terms]['Count'] = $results['Count'];
+  $_SESSION['tripal_pub_PMID_query'][$terms]['WebEnv'] = $results['WebEnv'];
+  $_SESSION['tripal_pub_PMID_query'][$terms]['QueryKey'] = $results['QueryKey'];
   
   return $results['Count'];
 
@@ -74,38 +77,38 @@ function tripal_pub_pubmed_count($search_array) {
  * tripal_pager_callback function.  This function returns the results
  * within the specified range
  */
-function tripal_pub_pubmed_range($search_array, $start = 0, $limit = 10) {
+function tripal_pub_PMID_range($search_array, $start = 0, $limit = 10) {
   $terms = $search_array['search_terms'];
-  $days = $search_array['days'];
+  $days  = $search_array['days'];
   $limit = $search_array['limit'];
   
-  $count = $_SESSION['tripal_pub_pubmed_query'][$terms]['Count'];
+  $count = $_SESSION['tripal_pub_PMID_query'][$terms]['Count'];
     
   // get the query_key and the web_env from the previous count query.
-  $query_key = $_SESSION['tripal_pub_pubmed_query'][$terms]['QueryKey'];
-  $web_env = $_SESSION['tripal_pub_pubmed_query'][$terms]['WebEnv'];
+  $query_key = $_SESSION['tripal_pub_PMID_query'][$terms]['QueryKey'];
+  $web_env   = $_SESSION['tripal_pub_PMID_query'][$terms]['WebEnv'];
   
   // if this function has been called without calling the count function
   // then we need to do the query.
   if (!$query_key) {
-    $results = tripal_pub_pubmed_search_init($terms, $limit, $days);
-    $_SESSION['tripal_pub_pubmed_query']['WebEnv'] = $results['WebEnv'];
-    $_SESSION['tripal_pub_pubmed_query']['QueryKey'] = $results['QueryKey']; 
+    $results = tripal_pub_PMID_search_init($terms, $limit, $days);
+    $_SESSION['tripal_pub_PMID_query']['WebEnv'] = $results['WebEnv'];
+    $_SESSION['tripal_pub_PMID_query']['QueryKey'] = $results['QueryKey']; 
     $query_key =  $results['QueryKey'];
     $web_env = $results['WebEnv'];
   }
 
-  // now get the list of PMIDs from the previous search
-  $pmids_txt = tripal_pub_pubmed_fetch($query_key, $web_env, 'uilist', 'text', $start, $limit);  
+  // now get the list of PMIDs from the previous search  
+  $pmids_txt = tripal_pub_PMID_fetch($query_key, $web_env, 'uilist', 'text', $start, $limit);  
   
   // iterate through each PMID and get the publication record. This requires a new search and new fetch
   $pmids = explode("\n", trim($pmids_txt));
   $pubs = array();
   foreach ($pmids as $pmid) {
     // now retrieve the individual record
-    $pub_xml = tripal_pub_pubmed_fetch($query_key, $web_env, 'null', 'xml', 0, 1, array('id' => $pmid));
-    $pub = tripal_pub_pubmed_parse_pubxml($pub_xml);
-    $pubs[] = $pub;    
+    $pub_xml = tripal_pub_PMID_fetch($query_key, $web_env, 'null', 'xml', 0, 1, array('id' => $pmid));
+    $pub     = tripal_pub_PMID_parse_pubxml($pub_xml);
+    $pubs[]  = $pub;    
   } 
   return $pubs;
 }
@@ -113,7 +116,7 @@ function tripal_pub_pubmed_range($search_array, $start = 0, $limit = 10) {
 /*
  * 
  */
-function tripal_pub_pubmed_search_init($terms, $retmax, $days = 0){
+function tripal_pub_PMID_search_init($terms, $retmax, $days = 0){
    
   // do a search for a single result so that we can establish a history, and get
   // the number of records. Once we have the number of records we can retrieve
@@ -173,7 +176,7 @@ function tripal_pub_pubmed_search_init($terms, $retmax, $days = 0){
 /*
  * 
  */
-function tripal_pub_pubmed_fetch($query_key, $web_env, $rettype = 'null', 
+function tripal_pub_PMID_fetch($query_key, $web_env, $rettype = 'null', 
   $retmod = 'null', $start = 0, $limit = 10, $args = array()){
 
   // repeat the search performed previously (using WebEnv & QueryKey) to retrieve
@@ -221,7 +224,7 @@ function tripal_pub_pubmed_fetch($query_key, $web_env, $rettype = 'null',
  * Information about PubMed's citation format can be found here
  * http://www.nlm.nih.gov/bsd/policy/cit_format.html
  */
-function tripal_pub_pubmed_parse_pubxml($pub_xml) {
+function tripal_pub_PMID_parse_pubxml($pub_xml) {
   $pub = array();
   
   if (!$pub_xml) {
@@ -244,10 +247,10 @@ function tripal_pub_pubmed_parse_pubxml($pub_xml) {
         case 'Article':
           $pub_model = $xml->getAttribute('PubModel');
           $pub['Publication Model'] = $pub_model;
-          tripal_pub_pubmed_parse_article($xml, $pub);
+          tripal_pub_PMID_parse_article($xml, $pub);
           break;
         case 'MedlineJournalInfo':
-          tripal_pub_pubmed_parse_medline_journal_info($xml, $pub);
+          tripal_pub_PMID_parse_medline_journal_info($xml, $pub);
           break;
         case 'ChemicalList':
           // TODO: handle this
@@ -320,7 +323,7 @@ function tripal_pub_pubmed_parse_pubxml($pub_xml) {
 /*
  * 
  */
-function tripal_pub_pubmed_parse_medline_journal_info($xml, &$pub) {
+function tripal_pub_PMID_parse_medline_journal_info($xml, &$pub) {
   while ($xml->read()) {
     // get this element name
     $element = $xml->name;     
@@ -355,7 +358,7 @@ function tripal_pub_pubmed_parse_medline_journal_info($xml, &$pub) {
 /*
  * 
  */
-function tripal_pub_pubmed_parse_article($xml, &$pub) {
+function tripal_pub_PMID_parse_article($xml, &$pub) {
     
   while ($xml->read()) {
     // get this element name
@@ -368,17 +371,17 @@ function tripal_pub_pubmed_parse_article($xml, &$pub) {
     if ($xml->nodeType == XMLReader::ELEMENT) {    
       switch ($element) {
         case 'Journal':
-          tripal_pub_pubmed_parse_journal($xml, $pub);
+          tripal_pub_PMID_parse_journal($xml, $pub);
           break;
         case 'ArticleTitle':
           $xml->read();
           $pub['Title'] = $xml->value;
           break;
         case 'Abstract':
-          tripal_pub_pubmed_parse_abstract($xml, $pub);
+          tripal_pub_PMID_parse_abstract($xml, $pub);
           break;
         case 'Pagination':
-          tripal_pub_pubmed_parse_pagination($xml, $pub);
+          tripal_pub_PMID_parse_pagination($xml, $pub);
           break;  
         case 'ELocationID':
           $type = $xml->getAttribute('EIdType');
@@ -400,7 +403,7 @@ function tripal_pub_pubmed_parse_article($xml, &$pub) {
           break;
         case 'AuthorList':
           $complete = $xml->getAttribute('CompleteYN');
-          tripal_pub_pubmed_parse_authorlist($xml, $pub);
+          tripal_pub_PMID_parse_authorlist($xml, $pub);
           break;
         case 'InvestigatorList':
           // TODO: perhaps handle this one day.  The investigator list is to list the names of people who 
@@ -420,7 +423,7 @@ function tripal_pub_pubmed_parse_article($xml, &$pub) {
           // TODO: handle this case
           break;
         case 'PublicationTypeList':
-          tripal_pub_pubmed_parse_publication_type($xml, $pub);
+          tripal_pub_PMID_parse_publication_type($xml, $pub);
           break;
         case 'VernacularTitle':
           $xml->read();
@@ -444,7 +447,7 @@ function tripal_pub_pubmed_parse_article($xml, &$pub) {
  * The Tripal Pub ontology doesn't yet have terms for all of the 
  * publication types so we store the value in the 'publication_type' term.
  */
-function tripal_pub_pubmed_parse_publication_type($xml, &$pub) {
+function tripal_pub_PMID_parse_publication_type($xml, &$pub) {
   
   while ($xml->read()) {
     $element = $xml->name;    
@@ -479,7 +482,7 @@ function tripal_pub_pubmed_parse_publication_type($xml, &$pub) {
 /*
  * 
  */
-function tripal_pub_pubmed_parse_abstract($xml, &$pub) {
+function tripal_pub_PMID_parse_abstract($xml, &$pub) {
   $abstract = '';
   
   while ($xml->read()) {
@@ -519,7 +522,7 @@ function tripal_pub_pubmed_parse_abstract($xml, &$pub) {
 /*
  * 
  */
-function tripal_pub_pubmed_parse_pagination($xml, &$pub) {
+function tripal_pub_PMID_parse_pagination($xml, &$pub) {
   while ($xml->read()) {
     $element = $xml->name;
       
@@ -544,7 +547,7 @@ function tripal_pub_pubmed_parse_pagination($xml, &$pub) {
 /*
  * 
  */
-function tripal_pub_pubmed_parse_journal($xml, &$pub) {
+function tripal_pub_PMID_parse_journal($xml, &$pub) {
   
   while ($xml->read()) {
     $element = $xml->name;
@@ -569,7 +572,7 @@ function tripal_pub_pubmed_parse_journal($xml, &$pub) {
         case 'JournalIssue':   
           // valid values of cited_medium are 'Internet' and 'Print'
           $cited_medium = $xml->getAttribute('CitedMedium');                    
-          tripal_pub_pubmed_parse_journal_issue($xml, $pub);       
+          tripal_pub_PMID_parse_journal_issue($xml, $pub);       
           break;        
         case 'Title': 
           $xml->read();
@@ -588,7 +591,7 @@ function tripal_pub_pubmed_parse_journal($xml, &$pub) {
 /*
  * 
  */
-function tripal_pub_pubmed_parse_journal_issue($xml, &$pub) {
+function tripal_pub_PMID_parse_journal_issue($xml, &$pub) {
   
   while ($xml->read()) {
     $element = $xml->name;
@@ -608,7 +611,7 @@ function tripal_pub_pubmed_parse_journal_issue($xml, &$pub) {
           $pub['Issue'] = $xml->value;           
           break;
         case 'PubDate':
-          $date = tripal_pub_pubmed_parse_date($xml, 'PubDate');
+          $date = tripal_pub_PMID_parse_date($xml, 'PubDate');
           $year = $date['year'];
           $month = $date['month'];
           $day = $date['day']; 
@@ -641,7 +644,7 @@ function tripal_pub_pubmed_parse_journal_issue($xml, &$pub) {
 /*
  * 
  */
-function tripal_pub_pubmed_parse_date ($xml, $element_name) {
+function tripal_pub_PMID_parse_date ($xml, $element_name) {
   $date = array();
   
   while ($xml->read()) {
@@ -680,7 +683,7 @@ function tripal_pub_pubmed_parse_date ($xml, $element_name) {
 /*
  * 
  */
-function tripal_pub_pubmed_parse_authorlist($xml, &$pub) {
+function tripal_pub_PMID_parse_authorlist($xml, &$pub) {
   $num_authors = 0;
   
   while ($xml->read()) {

tripal_pub/includes/tripal_pub.admin.inc

@@ -88,7 +88,7 @@ function tripal_pub_importer_setup($action = 'new', $pub_import_id = NULL) {
     $i = $page * $limit + 1;
     if (count($pubs) > 0) {
       foreach ($pubs as $pub) {
-        $rows[] = array(number_format($i), $pub['citation']);
+        $rows[] = array(number_format($i), $pub['Citation']);
         $i++;
       }
     }
@@ -207,7 +207,7 @@ function tripal_pub_importer_setup_form(&$form_state = NULL, $import_id = NULL,
     '#required'      => TRUE,
   );
    
-  $remote_dbs = array('Pubmed' => 'Pubmed');
+  $remote_dbs = array('PMID' => 'Pubmed');
   $form['remote_db'] = array(
     '#title' => t('Remote Database'),
     '#type' => 'select',

tripal_pub/tripal_pub.drush.inc

@@ -26,12 +26,19 @@ function tripal_pub_drush_help($command) {
  */
 function tripal_pub_drush_command() {
   $items = array();
-  $items['tripal-pub-import'] = array(
+  $items['tripal-pubs-import'] = array(
     'description' => dt('Imports publications from remote databases using saved configuration settings.'),
     'examples' => array(
-      'Standard example' => 'drush tripal-pub-import',
+      'Standard example' => 'drush tripal-pubs-import',
     ),
-    'aliases' => array('trp-pubs'),
+    'aliases' => array('tpubs-import'),
+  );
+  $items['tripal-pubs-update'] = array(
+    'description' => dt('Updates publication information for publications with a supported database cross-reference.'),
+    'examples' => array(
+      'Standard example' => 'drush tripal-pubs-update',
+    ),
+    'aliases' => array('tpubs-update'),
   );
   return $items;
 }
@@ -40,6 +47,14 @@ function tripal_pub_drush_command() {
  * Imports publications into Chado
  *
  */
-function drush_tripal_pub_import() {
+function drush_tripal_pub_tripal_pubs_import() {
   tripal_pub_import_publications();
 }
+
+/**
+ * Imports publications into Chado
+ *
+ */
+function drush_tripal_pub_tripal_pubs_update() {
+  tripal_pub_update_publications();
+}