<?php
// $Id: apachesolr_attachments.module,v 1.23 2010-11-26 14:07:08 wmostrey Exp $

/**
 * @file
 * Provides a file attachment search implementation for use with the Apache Solr module
 */

define ('EXTRACTING_SERVLET', 'extract/tika');

/**
 * Implementation of hook_menu().
 */
function apachesolr_attachments_menu() {
  $items = array();
  $items['admin/settings/apachesolr/attachments'] = array(
    'title' => 'File attachments',
    'description' => 'Administer Apache Solr Attachments.',
    'page callback' => 'apachesolr_attachments_admin_page',
    'access arguments' => array('administer search'),
    'file' => 'apachesolr_attachments.admin.inc',
    'type' => MENU_LOCAL_TASK,
  );
  $items['admin/settings/apachesolr/attachments/confirm/reindex'] = array(
    'title' => 'Reindex all files',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('apachesolr_attachments_confirm', 5),
    'access arguments' => array('administer search'),
    'file' => 'apachesolr_attachments.admin.inc',
    'type' => MENU_CALLBACK,
  );
  $items['admin/settings/apachesolr/attachments/confirm/delete'] = array(
    'title' => 'Delete and reindex all files',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('apachesolr_attachments_confirm', 5),
    'access arguments' => array('administer search'),
    'file' => 'apachesolr_attachments.admin.inc',
    'type' => MENU_CALLBACK,
  );
  $items['admin/settings/apachesolr/attachments/confirm/clear-cache'] = array(
    'title' => 'Delete the local cache of file text',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('apachesolr_attachments_confirm', 5),
    'access arguments' => array('administer search'),
    'file' => 'apachesolr_attachments.admin.inc',
    'type' => MENU_CALLBACK,
  );
  return $items;
}

/**
 * Implementation of hook_help().
 */
function apachesolr_attachments_help($section) {
  switch ($section) {
    case 'admin/settings/apachesolr/index':
      if (!variable_get('apachesolr_read_only', 0)) {
        $remaining = 0;
        $total = 0;
        // Collect the stats
        $status = apachesolr_attachments_search('status');
        $remaining += $status['remaining'];
        $total += $status['total'];
        return t('<br />There @items remaining to be examined for attachments out of @total total.', array(
          '@items' => format_plural($remaining, t('is 1 post'), t('are @count posts')),
          '@total' => $total,
        ));
      }
      break;
  }
}

/**
 * Implementation of hook_search().
 */
function apachesolr_attachments_search($op = 'search', $keys = NULL) {

  switch ($op) {
    case 'name':
      return ''; // We dont want a tab
    case 'reset':
      apachesolr_clear_last_index('apachesolr_attachments');
      return;
    case 'status':
      // TODO: Figure out a way to know how many actual files are left to update.
      return apachesolr_index_status('apachesolr_attachments');
    case 'search':
      return array();
  }
}

/**
 * Implementation of hook_apachesolr_types_exclude().
 */
function apachesolr_attachments_apachesolr_types_exclude($namespace) {
  if ($namespace == 'apachesolr_attachments') {
    if (variable_get('apachesolr_attachments_exclude_types', 1)) {
      $excluded_types = variable_get('apachesolr_search_excluded_types', array());
      return array_filter($excluded_types);
    }
    return array();
  }
}

/**
 * Implementation of hook_apachesolr_document_handlers().
 *
 * @param string $type
 *   Entity type. 'node', 'comment', and so forth. Used to evaluate whether this module
 *   should be interested in creating documents.
 * @param string $namespace
 *   Usually the name of the module that is initiating indexing. In this case
 *   we want to register a handler if the namespace is 'apachesolr_search'.
 * @return array $handlers
 *   An array of strings that are function names. Each function returns a $document from
 *   an entity (of type $type).
 */
function apachesolr_attachments_apachesolr_document_handlers($type, $namespace) {
  if ($type == 'node' && $namespace == 'apachesolr_attachments') {
    return array('apachesolr_attachments_add_documents');
  }
}

/**
 * Hook is called by search.module to add things to the search index.
 * In our case we will search content types and add any CCK type that
 * is a file type that we know how to parse and any uploaded file
 * attachments.
 */
function apachesolr_attachments_update_index() {
  if(!variable_get('apachesolr_attachments_tika_path', '') && variable_get('apachesolr_attachment_extract_using', 'tika') == 'tika') {
    return;
  }
  $start = time();
  $cron_try = variable_get('apachesolr_attachements_cron_try', 20);
  $cron_limit = variable_get('apachesolr_attachments_cron_limit', 100);
  $cron_time_limit = variable_get('apachesolr_attachements_cron_time_limit', 15);
  $num_tried = 0;
  module_load_include('inc', 'apachesolr_attachments', 'apachesolr_attachments.admin');
  do {
    $rows = apachesolr_get_nodes_to_index('apachesolr_attachments', $cron_try);
    $success = apachesolr_index_nodes($rows, 'apachesolr_attachments');
    $num_tried += $cron_try;
  } while ($success && ($num_tried < $cron_limit) && (time() - $start < $cron_time_limit));
}

/**
 * Implementation of hook_nodeapi().
 *
 * For a delete: mark all associated attachments as removed.
 */
function apachesolr_attachments_nodeapi($node, $op) {

  switch ($op) {
    case 'delete':
      apachesolr_attachments_remove_attachments_from_index($node->nid);
      // Mark attachments for later re-deletion in case the query fails.
      db_query("UPDATE {apachesolr_attachments_files} SET removed = 1 WHERE nid = %d", $node->nid);
      break;
  }
}

/**
 * Implementation of hook_cron().
 *
 * Delete all removed attachments from the Solr store.
 */
function apachesolr_attachments_cron() {
  try {
    $solr = apachesolr_get_solr();
    $result = db_query("SELECT fid, nid FROM {apachesolr_attachments_files} WHERE removed = 1");
    $ids = array();
    $fids = array();
    while ($file = db_fetch_object($result)) {
      $ids[] = apachesolr_document_id($file->fid .'-'. $file->nid, 'file');
      $fids[] = $file->fid;
    }
    if ($ids) {
      $solr->deleteByMultipleIds($ids);
      $solr->commit();
      // There was no exception, so update the table.
      db_query("DELETE FROM {apachesolr_attachments_files} WHERE fid IN (". db_placeholders($fids) .")", $fids);
    }
  }
  catch (Exception $e) {
    // Shortened project name because the watchdog limits type to 16 characters.
    watchdog('ApacheSolrAttach', nl2br(check_plain($e->getMessage())) . ' in apachesolr_attachments_cron', NULL, WATCHDOG_ERROR);
  }
}

/**
 * Implementation of hook_apachesolr_modify_query().
 */
function apachesolr_attachments_apachesolr_modify_query(&$query, &$params, $caller) {
  // Fetch the extra file data on searches.
  if ($caller == 'apachesolr_search') {
    $params['hl.fl'] .= ',ss_filemime,ss_file_node_title,ss_file_node_url';
  }
  elseif ($caller == 'apachesolr_mlt') {
    // Exclude files from MLT results.
    $query->add_filter('entity', 'file', TRUE);
  }
}

/**
 * Implementation of hook_apachesolr_process_results().
 *
 * When using the Apache Solr search module, everything is treated as a node
 * and as such values like the link and type won't be configured correctly if
 * it is a file attachement. We override such values here as needed.
 */
function apachesolr_attachments_apachesolr_process_results(&$results) {

  foreach ($results as &$item) {
    if (isset($item['node']->ss_filemime)) {
      $nid = $item['node']->nid;
      $item['link'] = file_create_url($item['node']->path);
      $node_link = t('<em>attached to:</em> !node_link', array('!node_link' => l($item['node']->ss_file_node_title, 'node/'. $nid)));
      $icon = theme('filefield_icon', array('filemime' => $item['node']->ss_filemime));
      $file_type = t('!icon @filemime', array('@filemime' => $item['node']->ss_filemime, '!icon' => $icon));
      $item['snippet'] .= '<div>' . $file_type .' '. $node_link . '</div>';
      $item['extra'] = array();
      $item['type'] = t('File attachment');
    }
  }
}

/**
 * For a particular node id, remove all file attachments from the solr index.
 */
function apachesolr_attachments_remove_attachments_from_index($nid) {
  try {
    $solr = apachesolr_get_solr();
    $solr->deleteByQuery("nid:{$nid} AND entity:file AND hash:". apachesolr_site_hash());
    $solr->commit();
  }
  catch (Exception $e) {
    // Shortened project name because the watchdog limits type to 16 characters.
    watchdog('ApacheSolrAttach', nl2br(check_plain($e->getMessage())), NULL, WATCHDOG_ERROR);
  }
}