tripal_seq.module

<?php
/**
 *    Let's include some files
 */
// Admin
require_once 'includes/admin/tseq.dblist.inc';
require_once 'includes/admin/tseq.settings.inc';
require_once 'includes/admin/tseq.categories.inc';

// Classes
require_once 'includes/tseq_databases.inc';

// API
require_once 'api/tseq.api.inc';

// Forms (allow for more than just basic form)
require_once 'includes/tseq_submit.form.inc';

// Theme
require_once 'theme/tseq.theme.inc';

// Output
require_once 'includes/tseq.review.inc';
require_once 'includes/tseq.download.inc';

/**
 * Implements hook_permission
 */
function tripal_seq_permission() {
    return array (
        'administer diamond' => array (
        'title' => t ( 'Administer Diamond Module' ),
        'description' => t ( 'Allows an admin to modify Tripal Sequence Similarity search configuration' )
        ),
    );
}

function tripal_seq_init()
{
}

/*
 * Implements hook_menu
 */
function tripal_seq_menu()
{
    $items = array();

    // TSeq Job submission form (for URL)
    $items['TSeq/submit'] = array(
        'title' => 'Sequence Similarity Search: Diamond/BLAST',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_submit_form'),
        'access callback' => TRUE,
        'access arguments' => array('access content'),
        'type' => MENU_NORMAL_ITEM,
        );

    // TSeq Results page
    $items['TSeq/results/%'] = array(
        'title' => 'Diamond Job Results',
        'page callback' => 'tseq_review',
        'page arguments' => array(2),
        'access callback' => TRUE,
        'access arguments' => array('access content'),
        'type' => MENU_CALLBACK
        );

    // TSeq Download page
    $items['TSeq/results/download/%/%'] = array(
        'title' => 'TSeq Job Results Download',
        'page callback' => 'tseq_download',
        'page arguments' => array(3,4),
        'access callback' => TRUE,
        'access arguments' => array('access content'),
        'type' => MENU_CALLBACK
        );

    // Configuration pages
    // Configuration: Main page
    $items['admin/tripal/extension/tseq/config'] = array(
        'title' =>  'TSeq Configuration',
        'description'   =>  'Configuration for the Tripal Sequence Similarity Search module',
        'page callback' => 'tseq_config_dblist_page_build',
        'page arguments'=>  array('tseq_config_dblist_page'),
        'type'  =>  MENU_NORMAL_ITEM,
        'file'  =>  'includes/admin/tseq.dblist.inc',
        'file path'     =>  drupal_get_path('module','tripal_seq'),
        'access arguments' => array('administer diamond'),
    );

    // Configuration: Database list, first page
    $items['admin/tripal/extension/tseq/config/dblist'] = array(
        'title' => 'Available Databases',
        'description' => 'Tripal Sequence Similarity Search Admin Configuration page',
        'page callback' => 'tseq_config_dblist_page_build',
        'page arguments' => array('tseq_config_dblist_page'),
        'access arguments' => array('administer diamond'),
        'type' => MENU_DEFAULT_LOCAL_TASK ,
        'file' => 'includes/admin/tseq.dblist.inc',
        'file path' => drupal_get_path ('module', 'tripal_seq')
        );


    // Configuration: Categories Tab
    $items['admin/tripal/extension/tseq/config/categories'] = array(
        'title' => 'Categories',
        'type'  => MENU_LOCAL_TASK,
        'weight' => 1,
        'priority' => 1,
        'page callback' =>  'tseq_config_categories_page_build',
        'page arguments'=>  array('tseq_config_categories_page'),
        'access arguments' => array('administer diamond'),
        'file' => 'includes/admin/tseq.categories.inc',
        'file path' => drupal_get_path ('module', 'tripal_seq')
    );

    // Configuration: Settings Tab
    $items['admin/tripal/extension/tseq/config/settings'] = array(
        'title' => 'Job Settings',
        'type'  => MENU_LOCAL_TASK,
        'weight' => 2,
        'priority' => 2,
        'page callback' =>  'drupal_get_form',
        'page arguments'=>  array('tseq_configuration_settings_form'),
        'access arguments' => array('administer diamond'),
        'file' => 'includes/admin/tseq.settings.inc',
        'file path' => drupal_get_path ('module', 'tripal_seq')
    );

    // Configuration: Defaults Tab
    $items['admin/tripal/extension/tseq/config/defaults'] = array(
        'title' => 'Submission Defaults',
        'type'  => MENU_LOCAL_TASK,
        'weight' => 3,
        'priority' => 3,
        'page callback' =>  'drupal_get_form',
        'page arguments'=>  array('tseq_configuration_defaults_form'),
        'access arguments' => array('administer diamond'),
        'file' => 'includes/admin/tseq.defaults.inc',
        'file path' => drupal_get_path ('module', 'tripal_seq')
    );

    //Admin: Genome Browser Link Scheme Tab
    /*$items['admin/tripal/extension/tseq/config/browseLink'] = array(
        'title' => 'Genome Browser Link',
        'type'  => MENU_LOCAL_TASK,
        'weight' => 4,
        'priority' => 4,
        'page callback' =>  'tseq_config_browseLink_page_build',
        'page arguments'=>  array('1', 'tseq_config_browseLink_page'),
        'access callback' => TRUE,
        'access arguments' => array('administer diamond'),
        'file' => 'includes/admin/tseq.browseLink.inc',
        'file path' => drupal_get_path ('module', 'tripal_seq')
    );
     *
     */

    // Admin: add a new target database
    $items['admin/tripal/extension/tseq/config/add_db'] = array(
        'title' => 'Add an Existing Database',
        'description' => 'Tripal Diamond Admin Configuration page',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_database_add_form',5),
        'access arguments' => array('administer diamond'),
        'type' => MENU_LOCAL_ACTION,
        );

    // Admin: edit an existing database
    $items['admin/tripal/extension/tseq/edit_db/%'] = array(
        'title' => 'Tripal Diamond Database Edit',
        'description' => 'Tripal Diamond Admin Configuration page',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_database_edit_form',5),
        'access arguments' => array('administer diamond'),
        'type' => MENU_CALLBACK
        );
    // Admin: delete an existing database
    $items['admin/tripal/extension/tseq/delete_db/%'] = array(
        'title' => 'Tripal Diamond Database Delete',
        'description' => 'Tripal Diamond Admin Configuration page',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_database_delete_form', 5),
        'access arguments' => array('administer diamond'),
        'type' => MENU_CALLBACK
        );

    //  Add/Edit/Delete Categories

    // Admin: add a new category
    $items['admin/tripal/extension/tseq/add_category'] = array(
        'title' => 'Add Category',
        'description' => 'Tripal Diamond Admin Configuration page',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_category_add_form',5),
        'access arguments' => array('administer diamond'),
        'type' => MENU_CALLBACK,
        );

   // Admin: edit an existing category
    $items['admin/tripal/extension/tseq/edit_category/%'] = array(
        'title' => 'Tripal Diamond Category Edit',
        'description' => 'Tripal Diamond Admin Configuration page',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_category_edit_form',5),
        'access arguments' => array('administer diamond'),
        'type' => MENU_CALLBACK
        );

    // Admin: delete an existing category
    $items['admin/tripal/extension/tseq/delete_category/%'] = array(
        'title' => 'Tripal Diamond Database Delete',
        'description' => 'Tripal Diamond Admin Configuration page',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_category_delete_form', 5),
        'access arguments' => array('administer diamond'),
        'type' => MENU_CALLBACK
        );

    /*
    // Admin: add a new Genome Browser Link Scheme
    $items['admin/tripal/extension/diamond/add_browseLink'] = array(
        'title' => 'TSeq Genome Browser Link Scheme Add',
        'description' => 'Tripal Diamond Admin Configuration page',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_browseLink_add_form',5),
        'access callback' => TRUE,
        'access arguments' => array('administer diamond'),
        'type' => MENU_CALLBACK,
        );

    // Admin: edit an existing Genome Browser Link Scheme
    $items['admin/tripal/extension/diamond/edit_browseLink/%'] = array(
        'title' => 'TSeq Genome Browser Link Scheme Edit',
        'description' => 'Tripal Diamond Admin Configuration page',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_browseLink_edit_form',5),
        'access callback' => TRUE,
        'access arguments' => array('administer diamond'),
        'type' => MENU_CALLBACK
        );
    // Admin: delete an existing Genome Browser Link Scheme
    $items['admin/tripal/extension/diamond/delete_browseLink/%'] = array(
        'title' => 'TSeq Genome Browser Link Scheme Delete',
        'description' => 'Tripal Diamond Admin Configuration page',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_browseLink_delete_form', 5),
        'access callback' => TRUE,
        'access arguments' => array('administer diamond'),
        'type' => MENU_CALLBACK
        );
     *
     */

    //Admin: Export the database list
    $items['admin/tripal/extension/tseq/config/export'] = array(
        'title' => 'TSeq Database Export (CSV)',
        'description' => 'Export the database list to a CSV file for backup/transfer',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_database_export_form'),
        'access callback' => TRUE,
        'access arguments' => array('administer diamond'),
        'type' => MENU_LOCAL_ACTION,
        'weight' => 1,
        );

    //Admin: Import a saved database list
    $items['admin/tripal/extension/tseq/config/import'] = array(
        'title' => 'TSeq Database Import (CSV)',
        'description' => 'Import to the database list from a CSV file',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_database_import_form'),
        'access callback' => TRUE,
        'access arguments' => array('administer diamond'),
        'type' => MENU_LOCAL_ACTION,
        'weight' => 2,
        );

    //Admin: Refresh the database list
    $items['admin/tripal/extension/tseq/config/refresh'] = array(
        'title' => 'Refresh this list',
        'description' => 'Check every database file to update the status.',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('tseq_database_list_refresh_form'),
        'access callback' => TRUE,
        'access arguments' => array('administer diamond'),
        'type' => MENU_LOCAL_ACTION,
        'weight' => 3,
        );

    // Analysis page
    $items['user/%/tseq_analysis'] = array (
        'title' => 'TSeq Analysis',
        'description' => 'Show results of Diamond or BLAST jobs belonging to the user',
        'page callback' => 'tseq_user_analysis_page',
        'page arguments' => array(1),
        'access callback' => TRUE,
        'type' => MENU_LOCAL_TASK,
        'file' => 'includes/tseq.user.inc',
        'file path' => drupal_get_path('module', 'tripal_seq'),
    );

    return $items;
}


function tripal_seq_theme()
{
    $items = array();
    $path = drupal_get_path('module', 'tripal_seq');

    // Displays the BLAST results for each job
    $items['tseq_results'] = array(
        'template' => 'tseq_review',
        'path' => "$path/theme",
    );
    return $items;
}


// Email to users


/**
 * Implements hook_mail
 */
function tripal_seq_mail($key, &$message, $params) {
    $site_name = variable_get('site_name', "");

    /*
    * Get information from the job (tseq_job_information)
    */
    $job_id = $params['job']->getJobID();
    $job_information  = tseq_get_job_information($job_id);    // from table: tseq_job_information
    $job_results      = tseq_get_job_results($job_id);        // from table: tseq_results
    $job_human_readable = tseq_get_program($job_information['sequence_query_type'],$job_information['database_search_type']);
    global $base_url;
    switch ($key) {

        case 'job-started':
            $user = $params['user'];
            $message['subject'] = t("Your Sequence Similary Search Job has started");
            $message['body'][] = "The ".$job_human_readable[0]." job you submitted to ".$site_name." has started.";
            $message['body'][] = "You can view the progress by <a href=\"".$base_url."/TSeq/results/".$job_information['tripal_job_id']."\">clicking here</a>.";
            $message['body'][] = "or by visiting this URL: ".$base_url."/TSeq/results/".$job_information['tripal_job_id'];
            $message['body'][] = "Job details:";
            $message['body'][] = "Job ID: ".$job_id;
            $message['body'][] = "Submission date: ".format_date($job_information['submit_time'], 'short');
            //$message['body'][] = "Number of sequences submitted: ".$job_results['sequence_count'];

            break;

        case 'job-complete':
            $user = $params['user'];
            $message['subject'] = t("Your Sequence Similary Search Job has finished");
            $message['body'][] = "The ".$job_human_readable[0]." job you submitted to ".$site_name." has finished.";
            $message['body'][] = "You can view the progress by <a href=\"".$base_url."/TSeq/results/".$job_information['tripal_job_id']."\">clicking here</a>.";
            $message['body'][] = "or by visiting this URL: ".$base_url."/TSeq/results/".$job_information['tripal_job_id'];
            $message['body'][] = "Job details:";
            $message['body'][] = "Job ID: ".$job_id;
            $message['body'][] = "Submission date: ".format_date($job_information['submit_time'], 'short');
            $message['body'][] = "Number of sequences submitted: ".$job_results['sequence_count'];
            $message['body'][] = "Number of matches found: ".$job_results['matches'];

            break;

        case 'job-error':
        $message['subject'] = t("TSeq Job: Error");
        $message['body'][] = "There was an issue with the TSeq job";
        break;
    }
}

/**
 * @param integer $tripal_job_id Job ID that TSeq is being notified about
 * @param string $key one-word status of job
 *
 * Allow Tripal Remote Job to notify this module that something has happened
 */

function tripal_seq_trj_notify($tripal_job_id,$key)
{
    switch ($key)
    {
        case "started":
            break;
        case "completed":
            publish_results($tripal_job_id);
            break;
        default:
            // handle errors PUMPKIN
            break;
    }
}

/**
 * @param int $job_id Tripal Job ID of the completed job
 *
 * Publish the results to the database so they can be viewed on the results page and in email notifications immediately
 */
function publish_results($job_id)
{
    // Get the Job information to be used throughout this function
    $job_information = tseq_get_job_information($job_id);

    $outputPath = DRUPAL_ROOT.'/sites/default/files/tripal/jobs/';
    // Do the results of this job exist in the database?
    $resultsFound = FALSE;
    if (!tseq_results_exist($job_id)) // No
    {
        //Prepare the data to be inserted into table `tseq_results`
        /*
         *  tripal_job_id       $job_id
         *  user_id             $job_information['user_id'];
         *  tseq_db_id          tseq_get_db_id_by_location($job_information['database_file'])
         *  summary             ?
         *  result_type         assume BLAST XML
         *  data                ...
         *  matches             ...
         *  database_used
         */
        $user_id = $job_information['user_id'];
        $tseq_db_id = tseq_get_db_id_by_location($job_information['database_file']);
        $result_type = "BLAST XML";
        // Get status and data
        $matches_found = 0;
        //Database used
        if ($job_information['database_file_type'] == 'database')
        {
            // Make some spaghetti getting this information because the sequence database name isn't stored in the tseq db
            $tseq_db_id = tseq_get_db_id_by_location($job_information['database_file']);
            $db_info = tseq_get_db_info($tseq_db_id);
            //drupal_set_message("Database used: ".$db_info['name'].", version ".$db_info['version']." (".$db_info['category'].")");
            $database_used = $db_info['name'].", version ".$db_info['version']." (".$db_info['type']." - ".$db_info['category'].")";

        }
        else
        {
            $database_used = '';
        }
        //Sequences queried
        $query_file_no_path = explode('/',$job_information['sequence_file']);
        $query_file = $outputPath.$job_id.'/'.$query_file_no_path[count($query_file_no_path)-1];
        if (file_exists($query_file))
        {
            $sequence_count = tseq_get_query_sequence_count($query_file);
        }
        else
        {
            $sequence_count = -1;
        }

        // Look in the results files to see what has been returned.
        $blastXML_filename = '/results_blastxml.txt';
        $pairwise_filename = '/results_pairwise.txt';
        if (file_exists($outputPath.$job_id.$blastXML_filename) && file_exists($outputPath.$job_id.'/STDERR.txt'))
        {
            //Prepare output if there is any data to show in $blastXML_filename
            if (filesize($outputPath.$job_id.'/STDERR.txt') > 0)
            {
                $jobResults = file($outputPath.$job_id.'/STDERR.txt');
                $resultsFound = TRUE;
                foreach($jobResults as $resultLine)
                {
                    // Check to see if there were any matches.
                    if (strcmp($resultLine, "0 queries aligned.") == 0)
                    {   
                        $resultsFound = FALSE;
                        $summary = "no results found";
                        $data = "no results found";
                    }
                }
                if ($resultsFound)
                {
                    $summary = "results found";
                    $data = 'Look for data in tseq_results_data table';
                }
                // Gather other details
                //Matches found
                $matches_found = tseq_get_matches_count($outputPath.$job_id.$blastXML_filename);

            }
            else if (filesize($outputPath.$job_id.'/STDERR.txt') > 0)
            {
                $data = file_get_contents($outputPath.$job_id.'/STDERR.txt');
                $summary = "error";
            }
            else
            {
                $data = "There were no matches found.";
                $summary = "error";
            }
        }
        else
        {
            $data = "Results not found. Please contact an adminstrator and reference Tripal Job Number: ".$job_id;        // This means there was no STDERR or STDOUT files, and possibly even the job dir (most likely due to system)
            $summary = "error";
            if (file_exists($query_file))
            {
                $sequence_count = 0;
            }
            $database_used ='';
        }

        // Pairwise/alignment view
        /*
        if (file_exists($outputPath.$job_id.$pairwise_filename)) // File exists
        {
            if (filesize($outputPath.$job_id.$pairwise_filename) > 0) // File has contents
            {
                // TODO Determine contents of pairwise results with 0 matches
                $pairwise_results = file($outputPath.$job_id.$pairwise_filename);
                foreach ($pairwise_results as $pairwise_result)
            }
        }*/


        // and finally, insert (if there is anything to insert)

        // Results details (always insert these)
        $newDB = array(
            'tripal_job_id' => $job_id,
            'user_id'       => $user_id,
            'tseq_db_id'    => $tseq_db_id,
            'data'          => $data,
            'summary'       => $summary,
            'result_type'   => $result_type,
            'sequence_count'=> $sequence_count,
            'matches'       => $matches_found,
            'database_used' => $database_used,
        );
        drupal_write_record('tseq_results', $newDB); // Insert the job details

        // Get the actual results file, not the STDOUT file
        $jobResults = file($outputPath.$job_id.$blastXML_filename);

        // Results data (Insert these only if they exist)
        if ($resultsFound)
        {
            foreach($jobResults as $key => $jobResult)
            {
                $columns = explode("\t", $jobResult);
                $job_results_data[$key] = array(
                    'tripal_job_id' => $job_id,
                    'match_id'      => $key,
                    'query_label'   => $columns[0],
                    'target'        => $columns[1],
                    'percent_identity'  => $columns[2],
                    'alignment_length'  => $columns[3],
                    'mismatches'        => $columns[4],
                    'gap_opens'         => $columns[5],
                    'start_position_query'  => $columns[6],
                    'end_position_query'     => $columns[7],
                    'start_position_database'  => $columns[8],
                    'end_position_database'     => $columns[9],
                    'e_value'   =>  $columns[10],
                    'bit_score' =>  $columns[11],
                );
                drupal_write_record('tseq_results_data',$job_results_data[$key]);
            }
        }
    }
}

/**
 * Update existing db location status
 */
function tripal_seq_update_db_status() {
// Get the list of databases stored in the database
$databases = tseq_get_all_dbs();
$status = array();
foreach($databases as $database) {
    $status[$database->db_id] = [$database->db_id, TSeqDatabases::db_check($database->db_id, $database->location)];
}
foreach($status as $update) {
    db_update('tseq_db_existing_locations')
        ->fields(array(
            'status' => $update[1],
        ))
        ->condition('db_id', $update[0], '=')
        ->execute();
    }

// Save the current time in the settings table (tseq_settings.db_last_refresh)
$now = time();
db_update('tseq_settings')
    ->fields(array(
        'db_last_refresh' => $now,
    ))
    ->condition('settings_id', 0, '=')
    ->execute();
}

/**
 * Implements hook_cron()
 */
function tripal_seq_cron() {
    $args = [];
    $includes = [];

    tripal_add_job('Cron: Checking TSeq database file status',
        'tripal_seq', 'tripal_seq_update_db_status', $args, 0, 1, $includes, TRUE);
}