islandora_usage_stats.install

<?php

/**
 * @file
 * Install/uninstall hooks.
 */

/**
 * Implements hook_uninstall().
 */
function islandora_usage_stats_uninstall() {
  $variables = array(
    'islandora_usage_stats_ip_list',
    'islandora_usage_stats_cooldown_seconds',
    'islandora_usage_stats_exclude_bots',
    'islandora_usage_stats_exclude_bots_regex',
    'islandora_usage_stats_top_downloads_block_display_rows',
    'islandora_usage_stats_top_search_block_display_rows',
  );
  array_walk($variables, 'variable_del');
}

/**
 * Implements hook_schema().
 */
function islandora_usage_stats_schema() {
  $schema['islandora_usage_stats_datastreams'] = array(
    'description' => 'Contains datastream information.',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'dsid' => array(
        'description' => 'The datastream identifier.',
        'type' => 'varchar',
        'length' => 255,
      ),
      'pid_id' => array(
        'description' => 'The pid id foreign relations.',
        'type' => 'int',
        'not NULL' => TRUE,
      ),
      'ds_label' => array(
        'description' => 'The datastream label.',
        'type' => 'varchar',
        'length' => 255,
      ),
    ),
    'foreign keys' => array(
      'pid_id' => array(
        'table' => 'islandora_usage_stats_objects',
        'columns' => array('pid_id' => 'id'),
      ),
    ),
    'primary key' => array('id'),
    'unique keys' => array(
      'pid_dsid' => array('pid_id', 'dsid'),
    ),
  );
  $schema['islandora_usage_stats_top_searches'] = array(
    'description' => 'Used to log top searches.',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'term' => array(
        'description' => 'Search term.',
        'type' => 'text',
        'size' => 'medium',
      ),
      'search_total' => array(
        'description' => 'The total times this term has been searched.',
        'type' => 'int',
        'not NULL' => TRUE,
      ),
    ),
    'indexes' => array(
      // 767 is maximum index size in worst case.
      // But if UTF8 is used then it is 3x. so 255
      'term_index' => array(array('term', 255)),
    ),
    'primary key' => array('id'),
  );

  $schema['islandora_usage_stats_object_access_log'] = array(
    'description' => 'This table is used to store the usage information for Fedora objects.',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'pid_id' => array(
        'description' => 'PID ID of the object.',
        'type' => 'int',
        'not null' => TRUE,
      ),
      'time' => array(
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
      'ip' => array(
        'description' => 'IP of user that accessed the object.',
        'type' => 'varchar',
        'length' => 64,
        'not null' => TRUE,
      ),
      'uid' => array(
        'description' => 'Drupal user that accessed the object.',
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
    ),
    'primary key' => array(
      'id',
    ),
    'indexes' => array(
      'pid_id_index' => array('pid_id'),
    ),
  );

  $schema['islandora_usage_stats_object_ds_access_log'] = array(
    'description' => 'This table is used to store the usage information for Fedora datastreams.',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'ds_id' => array(
        'description' => 'Database datastream ID.',
        'type' => 'int',
        'not null' => TRUE,
      ),
      'time' => array(
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
      'ip' => array(
        'description' => 'IP of user that accessed the object.',
        'type' => 'varchar',
        'length' => 64,
        'not null' => TRUE,
      ),
      'uid' => array(
        'description' => 'Drupal user that accessed the object.',
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
    ),
    'primary key' => array(
      'id',
    ),
    'foreign keys' => array(
      'ds_id' => array(
        'table' => 'islandora_usage_stats_datastreams',
        'columns' => array('ds_id' => 'id'),
      ),
    ),
    'indexes' => array(
      'ds_id' => array('ds_id'),
    ),
  );

  $schema['islandora_usage_stats_objects'] = array(
    'description' => 'Used to specify object information in the DB; primarily for views.',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'label' => array(
        'description' => 'Object label in Fedora.',
        'type' => 'varchar',
        'length' => 255,
      ),
      'pid' => array(
        'description' => 'The PID of the object.',
        'type' => 'varchar',
        'length' => 64,
        'not NULL' => TRUE,
      ),
    ),
    'primary key' => array('id'),
    'unique keys' => array(
      'pid' => array('pid'),
    ),
  );

  $schema['islandora_usage_stats_collection_access_log'] = array(
    'description' => 'Used to log the parent collection(s) of an object being visited.',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'object_access_id' => array(
        'type' => 'int',
        'not null' => TRUE,
      ),
      'collection' => array(
        'description' => 'The PID ID of the collection.',
        'type' => 'int',
        'not NULL' => TRUE,
      ),
    ),
    'foreign keys' => array(
      'object_acces_id' => array(
        'table' => 'islandora_usage_stats_object_access_log',
        'columns' => array('object_access_id' => 'id'),
      ),
      'collection' => array(
        'table' => 'islandora_usage_stats_objects',
        'columns' => array('collection' => 'id'),
      ),
    ),
    'primary key' => array('id'),
    'indexes' => array(
      'object_access_id_index' => array('object_access_id'),
      'collection_index' => array('collection'),
    ),
  );
  $schema['islandora_usage_stats_content_model_access_log'] = array(
    'description' => 'Used to log the content model(s) when an object is visited.',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'object_access_id' => array(
        'type' => 'int',
        'not null' => TRUE,
      ),
      'content_model' => array(
        'description' => 'The PID ID of the content model.',
        'type' => 'int',
        'not NULL' => TRUE,
      ),
    ),
    'foreign keys' => array(
      'object_access_id' => array(
        'table' => 'islandora_usage_stats_object_access_log',
        'columns' => array('object_access_id' => 'id'),
      ),
      'content_model' => array(
        'table' => 'islandora_usage_stats_objects',
        'columns' => array('content_model' => 'id'),
      ),
    ),
    'primary key' => array('id'),
  );
  return $schema;
}

/**
 * Updates schema to support collections and content models.
 */
function islandora_usage_stats_update_7001(&$sandbox) {
  $schema = array();
  $sandbox['#finished'] = 0;
  $pid_tables = array(
    'islandora_usage_stats_object_ds_downloads',
    'islandora_usage_stats_object_access_log',
    'islandora_usage_stats_object_ds_access_log',
    'islandora_usage_stats_object_views',
  );
  // Add new tables.
  if (!isset($sandbox['done_first_pass'])) {
    $sandbox['done_first_pass'] = TRUE;
    $schema['islandora_usage_stats_objects'] = array(
      'description' => 'Used to specify object information in the DB; primarily for views.',
      'fields' => array(
        'id' => array(
          'type' => 'serial',
          'not null' => TRUE,
        ),
        'label' => array(
          'description' => 'Object label in Fedora.',
          'type' => 'varchar',
          'length' => 255,
        ),
        'pid' => array(
          'description' => 'The PID of the object.',
          'type' => 'varchar',
          'length' => 64,
          'not NULL' => TRUE,
        ),
      ),
      'primary key' => array('id'),
      'unique keys' => array(
        'pid' => array('pid'),
      ),
    );
    $schema['islandora_usage_stats_collection_access_log'] = array(
      'description' => 'Used to log the parent collection(s) of an object being visited.',
      'fields' => array(
        'id' => array(
          'type' => 'serial',
          'not null' => TRUE,
        ),
        'object_access_id' => array(
          'type' => 'int',
          'not null' => TRUE,
        ),
        'collection' => array(
          'description' => 'The PID ID of the collection.',
          'type' => 'int',
          'not NULL' => TRUE,
        ),
      ),
      'foreign keys' => array(
        'object_acces_id' => array(
          'table' => 'islandora_usage_stats_object_access_log',
          'columns' => array('object_access_id' => 'id'),
        ),
        'collection' => array(
          'table' => 'islandora_usage_stats_objects',
          'columns' => array('collection' => 'id'),
        ),
      ),
      'primary key' => array('id'),
    );
    $schema['islandora_usage_stats_content_model_access_log'] = array(
      'description' => 'Used to log the content model(s) when an object is visited.',
      'fields' => array(
        'id' => array(
          'type' => 'serial',
          'not null' => TRUE,
        ),
        'object_access_id' => array(
          'type' => 'int',
          'not null' => TRUE,
        ),
        'content_model' => array(
          'description' => 'The PID ID of the content model.',
          'type' => 'int',
          'not NULL' => TRUE,
        ),
      ),
      'foreign keys' => array(
        'object_access_id' => array(
          'table' => 'islandora_usage_stats_object_access_log',
          'columns' => array('object_access_id' => 'id'),
        ),
        'content_model' => array(
          'table' => 'islandora_usage_stats_objects',
          'columns' => array('content_model' => 'id'),
        ),
      ),
      'primary key' => array('id'),
    );
    foreach ($schema as $name => $schema_info) {
      db_create_table($name, $schema_info);
    }
    $pid_queries = array();
    foreach ($pid_tables as $table) {
      db_drop_index($table, 'pid');
      db_add_field(
        $table,
        'pid_id',
        array(
          'description' => 'PID ID of the object.',
          'type' => 'int',
          'not null' => TRUE,
          'initial' => 1,
        )
      );
      $pid_queries[] = db_select($table)
        ->fields($table, array('pid'))
        ->groupBy('pid');
    }
    // Populate PID table with existing PIDS.
    foreach ($pid_queries as $query) {
      if (!isset($pid_query)) {
        $pid_query = $query;
      }
      else {
        $pid_query->union($query);
      }
    }
    db_insert('islandora_usage_stats_objects')
      ->from($pid_query)
      ->execute();
    // Populate PID ID columns, and drop the PID column.
    foreach ($pid_tables as $table) {
      db_update($table)
        ->expression(
          "$table.pid_id",
          "(SELECT islandora_usage_stats_objects.id FROM
            islandora_usage_stats_objects WHERE $table.pid = islandora_usage_stats_objects.pid)"
        )
        ->execute();
      db_drop_field($table, 'pid');
    }
    $sandbox['total'] = db_select('islandora_usage_stats_objects')
      ->fields('islandora_usage_stats_objects', array('pid'))
      ->countQuery()
      ->execute()
      ->fetchField();
  }
  // Populate label info in object table (batch).
  $step = 10;
  if (!isset($sandbox['progress'])) {
    $sandbox['progress'] = 0;
  }
  $query = db_select('islandora_usage_stats_objects')
    ->fields('islandora_usage_stats_objects', array('pid'))
    ->orderBy('islandora_usage_stats_objects.id')
    ->range($sandbox['progress'], $step)
    ->execute();

  $results = $query->fetchAllAssoc('pid');
  $pids = array_keys($results);
  foreach ($pids as $pid) {
    $object = islandora_object_load($pid);
    db_update('islandora_usage_stats_objects')
      ->fields(array('label' => $object->label))
      ->condition('pid', $pid)
      ->execute();
  }
  $sandbox['progress'] = $sandbox['progress'] + $step;
  $sandbox['#finished'] = $sandbox['progress'] / $sandbox['total'];
  if ($sandbox['#finished'] >= 1) {
    return t('Updated database to support logging collection and content model events.');
  }
}

/**
 * Adds support for top search/top downloads monitoring in islandora.
 */
function islandora_usage_stats_update_7002() {
  $schema['islandora_usage_stats_top_searches'] = array(
    'description' => 'Used to log top searches',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'term' => array(
        'description' => 'Search term',
        'type' => 'varchar',
        'length' => 255,
      ),
      'search_total' => array(
        'description' => 'The total times this term has been searched.',
        'type' => 'int',
        'not NULL' => TRUE,
      ),
    ),
    'primary key' => array('id'),
  );

  $schema['islandora_usage_stats_top_downloads'] = array(
    'description' => 'This table is used to store datastream download count for all fedora objects',
    'fields' => array(
      'id' => array(
        'description' => 'Unique id for each row',
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'pid_id' => array(
        'description' => 'PID ID of the object.',
        'type' => 'int',
        'not null' => TRUE,
      ),
      'ds_id' => array(
        'description' => 'The ds relation to the datastreams table',
        'type' => 'int',
        'not null' => TRUE,
      ),
      'downloads' => array(
        'description' => 'The number of views an object has received',
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
      'time' => array(
        'description' => 'The time in number of seconds of the last download of the object',
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
    ),
    'primary key' => array(
      'id',
    ),
    'foreign keys' => array(
      'ds_id' => array(
        'table' => 'islandora_usage_stats_datastreams',
        'columns' => array('ds_id' => 'id'),
      ),
      'pid_id' => array(
        'table' => 'islandora_usage_stats_objects',
        'columns' => array('pid_id' => 'id'),
      ),
    ),
    'indexes' => array(
      'ds_id' => array(
        'ds_id',
      ),
    ),
  );

  $schema['islandora_usage_stats_datastreams'] = array(
    'description' => 'Used to log top searches',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'dsid' => array(
        'description' => 'The datastream identifier',
        'type' => 'varchar',
        'length' => 255,
      ),
      'pid_id' => array(
        'description' => 'The pid id forign relations',
        'type' => 'int',
        'not NULL' => TRUE,
      ),
      'ds_label' => array(
        'description' => 'The datastream label',
        'type' => 'varchar',
        'length' => 255,
      ),
    ),
    'foreign keys' => array(
      'pid_id' => array(
        'table' => 'islandora_usage_stats_objects',
        'columns' => array('pid_id' => 'id'),
      ),
    ),
    'primary key' => array('id'),
  );
  db_create_table('islandora_usage_stats_top_downloads', $schema['islandora_usage_stats_top_downloads']);
  db_create_table('islandora_usage_stats_top_searches', $schema['islandora_usage_stats_top_searches']);
  db_create_table('islandora_usage_stats_datastreams', $schema['islandora_usage_stats_datastreams']);
  return t('Updated database to support top search logging, and top downloads.');
}

/**
 * Large rework of module. All object and datastream records will be lost.
 */
function islandora_usage_stats_update_7003() {
  db_drop_table('islandora_usage_stats_object_views');
  db_drop_table('islandora_usage_stats_object_ds_downloads');
  db_drop_table('islandora_usage_stats_top_downloads');

  db_add_unique_key('islandora_usage_stats_datastreams', 'pid_dsid', array('pid_id', 'dsid'));

  // Just re-create the tables.
  db_drop_table('islandora_usage_stats_object_ds_access_log');
  db_drop_table('islandora_usage_stats_object_access_log');
  $schema = array();
  $schema['islandora_usage_stats_object_access_log'] = array(
    'description' => 'This table is used to store the usage information for Fedora objects.',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'pid_id' => array(
        'description' => 'PID ID of the object.',
        'type' => 'int',
        'not null' => TRUE,
      ),
      'time' => array(
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
      'ip' => array(
        'description' => 'IP of user that accessed the object.',
        'type' => 'varchar',
        'length' => 64,
        'not null' => TRUE,
      ),
      'uid' => array(
        'description' => 'Drupal user that accessed the object.',
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
    ),
    'primary key' => array(
      'id',
    ),
  );

  $schema['islandora_usage_stats_object_ds_access_log'] = array(
    'description' => 'This table is used to store the usage information for Fedora datastreams.',
    'fields' => array(
      'id' => array(
        'type' => 'serial',
        'not null' => TRUE,
      ),
      'ds_id' => array(
        'description' => 'Database datastream ID.',
        'type' => 'int',
        'not null' => TRUE,
      ),
      'time' => array(
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
      'ip' => array(
        'description' => 'IP of user that accessed the object.',
        'type' => 'varchar',
        'length' => 64,
        'not null' => TRUE,
      ),
      'uid' => array(
        'description' => 'Drupal user that accessed the object.',
        'type' => 'int',
        'not null' => TRUE,
        'default' => 0,
      ),
    ),
    'primary key' => array(
      'id',
    ),
    'foreign keys' => array(
      'ds_id' => array(
        'table' => 'islandora_usage_stats_datastreams',
        'columns' => array('ds_id' => 'id'),
      ),
    ),
    'indexes' => array(
      'ds_id' => array('ds_id'),
    ),
  );
  db_create_table('islandora_usage_stats_object_ds_access_log', $schema['islandora_usage_stats_object_ds_access_log']);
  db_create_table('islandora_usage_stats_object_access_log', $schema['islandora_usage_stats_object_access_log']);

  // Now we always log.
  variable_del('islandora_usage_stats_use_access_log');
  variable_del('islandora_usage_stats_alter_pdf_display');
}

/**
 * Adding indexes to speed views.
 */
function islandora_usage_stats_update_7004() {
  db_add_index('islandora_usage_stats_object_access_log', 'pid_id_index', array('pid_id'));
  db_add_index('islandora_usage_stats_collection_access_log', 'object_access_id_index', array('object_access_id'));
  db_add_index('islandora_usage_stats_collection_access_log', 'collection_index', array('collection'));
}

/**
 * Fix to search stats table, increasing the size of search strings allowed.
 */
function islandora_usage_stats_update_7005() {
  db_change_field(
    'islandora_usage_stats_top_searches',
    'term',
    'term',
    array(
      'description' => 'Search term.',
      'type' => 'text',
      'size' => 'medium',
    ),
    array(
      'indexes' => array(
        // 767 is maximum index size in worst case.
        // But if UTF8 is used it is 3x. So 255
        'term_index' => array(array('term', 255)),
      ),
    )
  );
}