tripal/tripal_bulk_loader/tripal_bulk_loader.module

<?php

/**
 * @file
 * The functions for the Tripal bulk loader.
 *
 * @defgroup tripal_bulk_loader Tripal Bulk Loader Module
 * @ingroup tripal_modules
 */
include('includes/tripal_bulk_loader.loader.inc');
include('includes/tripal_bulk_loader.constants.inc');
include('includes/tripal_bulk_loader.admin.inc');
include('includes/tripal_bulk_loader.admin.templates.inc');

// API
include('api/tripal_bulk_loader.api.templates.inc');

/**
 * Implements hook_init
 * Used to add stylesheets and javascript files to the header
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_init() {
  // Add javascript and style sheet
  drupal_add_css(drupal_get_path('theme', 'tripal') . '/css/tripal_bulk_loader.css');
  drupal_add_js(drupal_get_path('theme', 'tripal') . '/js/tripal_bulk_loader.js');
}

/**
 * Implements hook_menu
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_menu() {
  $items = array();
  // Bulk Loading Job Node
  $items['node/%node/constants/%/edit'] = array(
    'title' => 'Edit Constant Set',
    'description' => 'Edit a group of constants associated with the current bulk loader',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_edit_constant_set_form', 1, 3),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_CALLBACK,
  );
  $items['node/%node/constants/%/delete'] = array(
    'title' => 'Delete Constant Set',
    'description' => 'Delete a group of constants associated with the current bulk loader',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_delete_constant_set_form', 1, 3),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_CALLBACK,
  );

  // Admin pages -----------------
  $items['admin/tripal/tripal_bulk_loader_template'] = array(
    'title' => 'Bulk Loader',
    'description' => 'Templates for loading tab-delimited data',
    'page callback' => 'theme',
    'page arguments' => array('tripal_bulk_loader_admin'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_NORMAL_ITEM,
  );
  $items['admin/tripal/tripal_bulk_loader_template/configure'] = array(
    'title' => 'Configure',
    'description' => 'Configuration of global options related to bulk loading jobs',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_configuration_form'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_NORMAL_ITEM,
  );
  $items['admin/tripal/tripal_bulk_loader_template/manage_templates'] = array(
    'title' => 'Manage Templates',
    'description' => 'Create/Update/Delete/Import/Export Templates',
    'page callback' => 'tripal_bulk_loader_admin_manage_templates',
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_NORMAL_ITEM,
  );
  $items['admin/tripal/tripal_bulk_loader_template/jobs'] = array(
    'title' => 'Bulk Loader Jobs',
    'description' => 'Listing of Bulk Loading Jobs',
    'page callback' => 'tripal_bulk_loader_admin_jobs',
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_NORMAL_ITEM,
  );

  // Create/Edit Template --------
  $items['admin/tripal/tripal_bulk_loader_template/manage_templates/create'] = array(
    'title' => 'Create Template',
    'description' => 'Create loader template for loading tab-delimited data',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_modify_template_base_form', 'create'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_NORMAL_ITEM,
  );
  $items['admin/tripal/tripal_bulk_loader_template/manage_templates/edit'] = array(
    'title' => 'Edit Template',
    'description' => 'Edit loader template for loading tab-delimited data',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_modify_template_base_form', 'edit'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_NORMAL_ITEM,
  );
  $items['admin/tripal/tripal_bulk_loader_template/edit_record'] = array(
    'title' => 'Edit Template Record',
    'description' => 'Edit a record in an existing tripal bulk loader template.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_edit_template_record_form'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_CALLBACK,
  );
  $items['admin/tripal/tripal_bulk_loader_template/add_field'] = array(
    'title' => 'Add Template Field',
    'description' => 'Add a template field to an existing tripal bulk loader template.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_add_template_field_form'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_CALLBACK,
  );
  $items['admin/tripal/tripal_bulk_loader_template/edit_field'] = array(
    'title' => 'Edit Template Field',
    'description' => 'Edit an existing field from a tripal bulk loader template.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_edit_template_field_form'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_CALLBACK,
  );
  // Delete Template -----
  $items['admin/tripal/tripal_bulk_loader_template/manage_templates/delete'] = array(
    'title' => 'Delete Template',
    'description' => 'Delete bulk loader template',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_delete_template_base_form'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_NORMAL_ITEM,
  );
  // Import/Export ---------
  $items['admin/tripal/tripal_bulk_loader_template/manage_templates/import'] = array(
    'title' => 'Import Template',
    'description' => 'Import Loaders',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_import_export_template_form', 'import'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_NORMAL_ITEM,
  );
  $items['admin/tripal/tripal_bulk_loader_template/manage_templates/export'] = array(
    'title' => 'Export Template',
    'description' => 'Export Loaders',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('tripal_bulk_loader_import_export_template_form', 'export'),
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_NORMAL_ITEM,
  );
  // AHAH ---------
  $items['admin/tripal/tripal_bulk_loader_template/add_field_ahah'] = array(
    'page callback' => 'tripal_bulk_loader_add_field_ahah',
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_CALLBACK,
  );
  $items['admin/tripal/tripal_bulk_loader_template/edit_field_ahah'] = array(
    'page callback' => 'tripal_bulk_loader_edit_field_ahah',
    'access arguments' => array('administer tripal_bulk_loader'),
    'type' => MENU_CALLBACK,
  );

  return $items;
}


/**
 * Implements hook_theme
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_theme() {
  return array(
    'node' => array(
      'variables' => array('node' => NULL),
      'path' => drupal_get_path('module', 'tripal_bulk_loader') . '/theme',
    ),
    'tripal_bulk_loader_set_constants_form' => array(
      'function' => 'theme_tripal_bulk_loader_set_constants_form',
      'variables' => array('form' => NULL),
    ),
    'tripal_bulk_loader_template' => array(
      'variables' => array('template_id' => NULL),
      'template' => 'tripal_bulk_loader_template',
      'path' => drupal_get_path('module', 'tripal_bulk_loader') . '/theme',
    ),
    'tripal_bulk_loader_modify_template_base_form' => array(
      'variables' => array('form' => NULL),
      'template' => 'tripal_bulk_loader_modify_template_base_form',
      'path' => drupal_get_path('module', 'tripal_bulk_loader') . '/theme',
    ),
    'tripal_bulk_loader_edit_template_field_form' => array(
      'variables' => array('form' => NULL),
      'template' => 'tripal_bulk_loader_edit_template_field_form',
      'path' => drupal_get_path('module', 'tripal_bulk_loader') . '/theme',
    ),
    'tripal_bulk_loader_add_template_field_form' => array(
      'variables' => array('form' => NULL),
      'template' => 'tripal_bulk_loader_add_template_field_form',
      'path' => drupal_get_path('module', 'tripal_bulk_loader') . '/theme',
    ),
    'tripal_bulk_loader_admin' => array(
      'template' => 'tripal_bulk_loader_admin',
      'variables' =>  array(NULL),
      'path' => drupal_get_path('module', 'tripal_bulk_loader') . '/theme',
    ),
  );
}

/**
 * Implement hook_access().
 *
 * This hook allows node modules to limit access to the node types they define.
 *
 *  @param $op
 *  The operation to be performed
 *
 *  @param $node
 *  The node on which the operation is to be performed, or, if it does not yet exist, the
 *  type of node to be created
 *
 *  @param $account
 *  A user object representing the user for whom the operation is to be performed
 *
 *  @return
 *  If the permission for the specified operation is not set then return FALSE. If the
 *  permission is set then return NULL as this allows other modules to disable
 *  access.  The only exception is when the $op == 'create'.  We will always
 *  return TRUE if the permission is set.
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_node_access($node, $op, $account) {
  if ($op == 'create') {
    if (!user_access('create tripal_bulk_loader', $account)) {
      return FALSE;
    }
    return TRUE;
  }
  if ($op == 'update') {
    if (!user_access('edit tripal_bulk_loader', $account)) {
      return FALSE;
    }
  }
  if ($op == 'delete') {
    if (!user_access('delete tripal_bulk_loader', $account)) {
      return FALSE;
    }
  }
  if ($op == 'view') {
    if (!user_access('access tripal_bulk_loader', $account)) {
      return FALSE;
    }
  }
  return NULL;
}

/**
 * Implements hook_perm
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_permission() {
  return array(
      'access tripal_bulk_loader',
      'create tripal_bulk_loader',
      'delete tripal_bulk_loader',
      'edit tripal_bulk_loader',
      'administer tripal_bulk_loader',
  );
}

//////////////////////////////////////////////////////////////////////////////////////////////
// Node Functions
//////////////////////////////////////////////////////////////////////////////////////////////

/**
 * Implements hook_node_info
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_node_info() {
  $nodes = array();
  $nodes['tripal_bulk_loader'] = array(
      'name' => t('Bulk Loading Job'),
      'base' => 'tripal_bulk_loader',
      'description' => t('A bulk loader for inserting tab-delimited data into chado database'),
      'has_title' => TRUE,
      'has_body' => FALSE,
      'locked' => TRUE
  );
  return $nodes;
}

/**
 * Implements node_form
 * Used to gather the extra details stored with a Bulk Loading Job Node
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_form($node, $form_state) {
  $form = array();

  if (isset($form_state['values'])) {
    $node = $form_state['values'] + (array)$node;
    $node = (object) $node;
  }

  $sql = "SELECT * FROM {tripal_bulk_loader_template}";
  $results = db_query($sql)->execute();
  $templates = array();
  foreach ($results as $template) {
    $templates [$template->template_id] = $template->name;
  }

  if (!$templates) {
    $form['label'] = array(
    '#type' => 'item',
      '#description' => t("Loader template needs to be created before any bulk loader can be added. Go to 'Tripal Management > Bulk Loader Template' to create the template."),
      '#weight'        => -10,
    );

    return $form;
  }

  $form['loader'] = array(
    '#type' => 'fieldset',
    '#title' => t('Basic Details'),
  );

  $form['loader']['loader_name'] = array(
    '#type'          => 'textfield',
    '#title'         => t('Loading Job Name'),
    '#weight'        => -10,
    '#required'      => TRUE,
    '#default_value' => $node->loader_name
  );

  $form['loader']['template_id'] = array(
    '#type' => 'select',
    '#title' => t('Template'),
    '#description'   => t('Please specify a template for this loader'),
    '#options'       => $templates,
    '#weight'        => -9,
    '#required'      => TRUE,
    '#default_value' => $node->template_id,
  );

  $form['loader']['file']= array(
    '#type'          => 'textfield',
    '#title'         => t('Data File'),
    '#description'   => t('Please specify the data file to be loaded. This must be a tab-delimited text file with UNIX line endings.'),
    '#weight'        => -8,
    '#default_value' => $node->file,
    '#maxlength'     => 1024,
  );

  $form['loader']['has_header'] = array(
    '#type' => 'radios',
    '#title' => t('File has a Header'),
    '#options' => array( 1 => 'Yes', 2 => 'No'),
    '#weight' => -7,
    '#default_value' => $node->file_has_header,
  );

  $form['loader']['keep_track_inserted'] = array(
    '#type' => 'radios',
    '#title' => t('Keep track of inserted record IDs'),
    '#description' => t('This enables the ability to revert an entire loading job even if '
      .'it completed successfully. Furthermore, it displays the number of records '
      .'successfully inserted into each table.'),
    '#options' => array( 1 => 'Yes', 0 => 'No'),
    '#weight' => -7,
    '#default_value' => (isset($node->keep_track_inserted)) ? $node->keep_track_inserted : variable_get('tripal_bulk_loader_keep_track_inserted', FALSE),
  );

  return $form;
}


/**
 * Implements node_load
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_load($node) {
  $sql = "SELECT * FROM {tripal_bulk_loader} WHERE nid = :nid";
  $node = db_query($sql, array(':nid' => $node->nid))->fetchObject();

  $node->title = 'Bulk Loading Job: ' . $node->loader_name;

  // Add job details
  $progress = tripal_bulk_loader_progess_file_get_progress($node->job_id);
  $sql = "SELECT * FROM {tripal_jobs} WHERE job_id=:job_id";
  $node->job = db_query($sql, array(':job_id' => $node->job_id))->fetchObject();

  // Add the loader template
  $sql = "SELECT * FROM {tripal_bulk_loader_template} WHERE template_id=:template";
  $results = db_query($sql, array(':template' => $node->template_id))->fetchObject();
  $template = unserialize($results->template_array);
  $node->template = $results;
  $node->template->template_array = $template;

  // Add inserted records
  $sql = 'SELECT * FROM {tripal_bulk_loader_inserted} WHERE nid=:nid';
  $resource = db_query($sql, array(':nid' => $node->nid))->execute();
  foreach ($resource as $r) {
    $r->num_inserted = sizeof(preg_split('/,/', $r->ids_inserted));
    $node->inserted_records->{$r->table_inserted_into} = $r;
  }

  // Add exposed field list
  $node->exposed_fields = array();
  if ($template) {
    foreach ($template as $record_id => $record) {
      foreach ($record['fields'] as $field_id => $field) {
        if ($field['exposed']) {
          $node->exposed_fields[] = array(
            'record_id' => $record_id,
            'field_id' => $field_id,
            'title' => $field['title'],
          );
        }
      }
    }

    if (empty($node->exposed_fields)) {
      $node->exposed_fields[] = array();
    }
  }


  // Add constants
  $sql = 'SELECT * FROM {tripal_bulk_loader_constants} WHERE nid=:nid ORDER BY group_id, record_id, field_id';
  $resource = db_query($sql, array(':nid' => $node->nid))->execute();
  foreach ($resource as $r) {
    $node->constants[$r->group_id][$r->record_id][$r->field_id] = array(
      'constant_id' => $r->constant_id,
      'group_id' => $r->group_id,
      'chado_table' => $r->chado_table,
      'chado_field' => $r->chado_field,
      'record_id' => $r->record_id,
      'field_id' => $r->field_id,
      'value' => $r->value
    );
  }
  if (!$node->constants) {
    $node->constants[] = array();
  }

  return $node;
}

/**
 * Implements node_insert
 * Insert the data from the node form on Create content
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_insert($node) {

  // Insert into tripal_bulk_loader
  $sql = "INSERT INTO {tripal_bulk_loader} (nid, loader_name, template_id, file, file_has_header, job_status, keep_track_inserted) VALUES (:nid, :loader, :template, :file, :header, :status, :ids)";
  db_query($sql,
    array(':nid' => $node->nid,
      ':loader' => $node->loader_name,
      ':template' => $node->template_id,
      ':file' => $node->file,
      ':header' => $node->has_header,
      ':status' => 'Initialized',
      ':ids' => $node->keep_track_inserted
    )
  )->execute();

  // Update title
  $node->title =$node->loader_name;
  drupal_write_record('node', $node, 'nid');
  drupal_write_record('node_revision', $node, 'nid');

  drupal_set_message(t('After reviewing the details, please Submit this Job (by clicking the "Submit Job" button below). No data will be loaded until the submitted job is reached in the queue.'));

}

/**
 * Implements node_delete
 * Deletes the data when the delete button on the node form is clicked
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_delete($node) {
  $sql = "DELETE FROM {tripal_bulk_loader} WHERE nid = :nid";
  db_query($sql, array(':nid' => $node->nid))->execute();
}

/**
 * Implements node_update
 * Updates the data submitted by the node form on edit
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_update($node) {

  // Update tripal_bulk_loader
  $sql = "UPDATE {tripal_bulk_loader} SET nid = :nid, loader_name = :loader, template_id = :template, file = :file, file_has_header = :header, keep_track_inserted = :ids WHERE nid = :wherenid";
  db_query($sql,
    array(':nid' => $node->nid,
      ':loader' => $node->loader_name,
      ':template' => $node->template_id,
      ':file' => $node->file,
      ':header' => $node->has_header,
      ':ids' => $node->keep_track_inserted,
      ':wherenid' => $node->nid
    )
  )->execute();

  // Add a job if the user want to load the data
  global $user;
  if ($node->job) {
    $job_args[0] =$node->loader_name;
    $job_args[1] = $node->template_id;
    $job_args[2] = $node->file;
    if (is_readable($node->file)) {
      $fname = preg_replace("/.*\/(.*)/", "$1", $node->file);
      tripal_add_job("Bulk Load: $fname", 'tripal_bulk_loader', 'tripal_bulk_loader_load_data', $job_args, $user->uid);
    }
    else {
      drupal_set_message(t("Can not open %file. Job not scheduled.", array('%file' => $node->file)));
    }
  }

}

///////////////////////////////////////////////////////////

/**
 * Preprocessor function for the tripal_bulk_loader template
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_preprocess_tripal_bulk_loader_template(&$variables) {

  $sql = "SELECT * FROM {tripal_bulk_loader_template} WHERE template_id=%d";
  $template = db_query($sql, array(':template' => $variables['template_id']))->fetchObject();
  $template->template_array = unserialize($template->template_array);
  $variables['template'] = $template;

}

/**
 * Get the progress of the current constant set from the progress file
 *
 * When transactions are used, database updates to drupal cannot be made. Thus a separate
 * method to keep track of progress was implemented: save a period to the file for each
 * record successfully inserted; each line in the file represents a processed line.
 *
 * @param $job_id
 *   The id of the tripal job to check the progress of
 * @param $node
 *   The tripal_bulk_loader node associated with the job
 *
 * @return
 *   An array with the following keys:
 *     num_lines = the number of lines in the file processed so far
 *     total_lines = the total number of lines in the input file
 *     percent_file = the percent the input file has been loaded
 *     num_records = the number of records successfully inserted
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_progess_file_get_progress($job_id, $update_progress = TRUE) {
  $filename = '/tmp/tripal_bulk_loader_progress-' . $job_id . '.out';
  if (!file_exists($filename)) {
    return (object) array();
  }

  $num_lines = trim(`wc --lines < $filename`);
  $num_records = trim(`grep -o "." $filename | wc --lines`);

  $job = db_query("SELECT j.*, b.file, b.file_has_header, c.num as num_constant_sets
                              FROM {tripal_jobs} j
                              LEFT JOIN {tripal_bulk_loader} b ON b.job_id=j.job_id
                              LEFT JOIN (
                                  SELECT nid, count(distinct(group_id)) as num
                                  FROM {tripal_bulk_loader_constants}
                                  GROUP BY nid
                                ) c ON c.nid=b.nid
                              WHERE j.job_id=:job", array(':job' =>$job_id))->execute();
  if ($job->num_constant_sets) {
    $num_constant_sets_loaded = round($job->progress / (100 / $job->num_constant_sets), 4);

    // If the next constant set has started loading
    if ($job->num_constant_sets != $num_constant_sets_loaded) {

      // total lines in input file
      $total_lines = trim(`wc --lines < $job->file`);
      if ($job->file_has_header) {
        $total_lines--;
      }

      // percent of the current constant set loaded
      $percent = round($num_lines/$total_lines * 100, 2);

      // percent of the total job = (<# fully loaded constant sets> * 100 )
      //                           + <percent of current constant set>
      //                           / <total number of constant sets>
      $total_percent = (($num_constant_sets_loaded * 100) + $percent) / $job->num_constant_sets;

      // update the progress of the job
      if ($update_progress AND ($percent != 0 OR $percent != 100)) {
        tripal_job_set_progress($job_id, round($total_percent, 0));
      }
    }
  }

  return (object) array(
    'num_lines' => $num_lines,
    'total_lines' => $total_lines,
    'percent_file' => $percent,
    'num_constant_sets_loaded' => $num_constant_sets_loaded,
    'total_percent' => $total_percent,
    'num_records' => $num_records
  );
}

/**
 * Implements hook_job_describe_args()
 * Specifically to make viewing past tripal jobs more readable for jobs registered by this module
 *
 * @params $callback
 *   The callback passed into tripal_add_job()
 * @param $args
 *   The arguements passed into tripal_add_job()
 * @return
 *   An array where keys are the human readable headers describing each arguement
 *   and the value is the aguement passed in after formatting
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_job_describe_args($callback, $args) {

  $new_args = array();
  if ($callback == 'tripal_bulk_loader_load_data') {
    //1st arg is the nid for a bulk loader node
    $node = node_load($args[0]);
    $new_args['Bulk Loading Job'] = l($node->title, 'node/' . $args[0]);
    return $new_args;
  }

}

/**
 * Implements hook_coder_ignore().
 * Defines the path to the file (tripal_bulk_loader.coder_ignores.txt) where ignore rules for coder are stored
 *
 * @ingroup tripal_bulk_loader
 */
function tripal_bulk_loader_coder_ignore() {
  return array(
    'path' => drupal_get_path('module', 'tripal_bulk_loader'),
    'line prefix' => drupal_get_path('module', 'tripal_bulk_loader'),
  );
}