Support synonyms better

tripal · Feb 26, 2019 · 9c196ef · 9c196ef
1 parent cd1fd97
commit 9c196ef
Show file tree

Hide file tree

Showing 7 changed files with 110 additions and 84 deletions.
diff --git a/includes/Common/Instance.php b/includes/Common/Instance.php
@@ -296,17 +296,16 @@ public function setIndexParams($index_name, $shards = 5, $replicas = 0, $tokeniz
       'number_of_replicas' => $replicas,
       'analysis' => [
         'analyzer' => [
-          $index_name => [
-            'type' => 'custom',
-            'tokenizer' => $tokenizer,
-            'filter' => array_keys($filters),
-          ],
+          //$index_name => [
+          //  'type' => 'custom',
+          //  'tokenizer' => $tokenizer,
+          //  'filter' => array_keys($filters),
+          //],
           'synonym' => [
             'tokenizer' => 'whitespace',
-            'filter' => [
-              'synonym',
-              'lowercase'
-            ],
+            'filter' => array_unique(
+              ['synonym', 'lowercase'] + array_keys($filters)
+            ),
           ],
         ],
         'filter' => [
@@ -315,6 +314,10 @@ public function setIndexParams($index_name, $shards = 5, $replicas = 0, $tokeniz
             'format' => 'wordnet',
             'synonyms_path' => 'analysis/wn_s.pl',
           ],
+          'stemmer' => [
+            'type' => 'stemmer',
+            'language' => 'english',
+          ],
         ],
       ],
       'max_result_window' => 1000000,

diff --git a/includes/Models/Model.php b/includes/Models/Model.php
@@ -10,11 +10,6 @@
  */
 class Model{
 
-  /**
-   * @var array
-   */
-  protected $attributes = [];
-
   /**
    * The ES instance.
    *
@@ -362,46 +357,35 @@ public function search() {
   }
 
   /**
-   * Get query paramaters.
+   * Paginate the results.
+   *
+   * @param int $per_page
    *
    * @return array
    * @throws \Exception
    */
-  public function getQuery() {
-    return $this->builder->build();
+  public function paginate($per_page = 10) {
+    $start = microtime(true);
+    $results = $this->search();
+    $total = $results['hits']['total'];
+    $current_page = pager_default_initialize($total, $per_page);
+    return [
+      'results' => $results,
+      'total' => $total,
+      'page' => $current_page + 1,
+      'pages' => ceil($total / $per_page),
+      'pager' => theme('pager', ['quantity', $total]),
+      'time' => microtime(true) - $start
+    ];
   }
 
   /**
-   * Get an attribute.
-   *
-   * @param string $name
-   *   The name of attribute.
-   *
-   * @return mixed
-   *   The value of the attribute if it exists.
-   */
-  public function __get($name) {
-    if (array_key_exists($name, $this->attributes)) {
-      return $this->attributes[$name];
-    }
-
-    if (method_exists(static::class, $name)) {
-      return;
-    }
-
-    return $this->{$name};
-  }
-
-  /**
-   * Check if an attribute isset.
-   *
-   * @param string $name
-   *   The name of the attribute.
+   * Get query paramaters.
    *
-   * @return bool
-   *   Whether the attribute has been set.
+   * @return array
+   * @throws \Exception
    */
-  public function __isset($name) {
-    return isset($this->attributes[$name]);
+  public function getQuery() {
+    return $this->builder->build();
   }
 }
diff --git a/includes/Query/SimpleQueryClause.php b/includes/Query/SimpleQueryClause.php
@@ -62,8 +62,8 @@ public function multiMatch(array $fields, $value) {
         'fields' => $fields,
         'query' => $value,
         'lenient' => TRUE,
-        //'type' => 'phrase',
         'analyzer' => 'synonym',
+        //'type' => 'phrase',
         'fuzziness' => 'AUTO',
       ],
     ];

diff --git a/includes/tripal_elasticsearch.advanced_search.form.inc b/includes/tripal_elasticsearch.advanced_search.form.inc
@@ -64,11 +64,11 @@ function tripal_elasticsearch_advanced_search_form($form, &$form_state) {
 
   $form['#method'] = 'GET';
 
-  if (isset($_GET['term']) && !empty($_GET['term'])) {
-    $form['wrapper']['results'] = [
-      '#markup' => tripal_elasticsearch_advanced_search_results($_GET),
-    ];
-  }
+  //if (isset($_GET['term']) && !empty($_GET['term'])) {
+  $form['wrapper']['results'] = [
+    '#markup' => tripal_elasticsearch_advanced_search_results($_GET),
+  ];
+  //}
 
   return $form;
 }
@@ -79,26 +79,18 @@ function tripal_elasticsearch_advanced_search_form($form, &$form_state) {
  * @return string
  */
 function tripal_elasticsearch_advanced_search_results(array $values) {
-  $fields = ['content.*', 'title'];
-
-  if (isset($_GET['field']) && !empty($_GET['field'])) {
-    $fields = ["content.{$_GET['field']}"];
+  if (empty($values['category']) && empty($values['term'])) {
+    return '';
   }
 
   try {
     $instance = new \ES\Common\Instance();
-    $model = new \ES\Models\Model();
-    $model->setIndexName('entities');
-    $model->where($fields, trim($values['term']));
-
-    $category = trim($values['category'] ?? '');
-    if (!empty($category)) {
-      $model->where('bundle_label', "\"$category\"");
+    $results = tripal_elasticsearch_perform_advanced_search($values, 15);
+    if($results['total'] == 0 && !empty($values['term'])) {
+      $values['term'] = strtolower($values['term']);
+      $results = tripal_elasticsearch_perform_advanced_search($values, 15);
     }
-
-    $model->highlight(['content.*', 'title']);
-    $data = $model->search();
-    $hits = $instance->formatHits($data);
+    $hits = $instance->formatHits($results['results']);
 
     if (count($hits) === 0) {
       return theme(
@@ -109,7 +101,16 @@ function tripal_elasticsearch_advanced_search_results(array $values) {
       );
     }
 
-    return tripal_elasticsearch_get_website_search_result_table($hits);
+    $content = theme('elasticsearch_results_header', [
+      'page' => $results['page'],
+      'total' => $results['total'],
+      'pages' => $results['pages'],
+      'time' => $results['time']
+    ]);
+    $content .= tripal_elasticsearch_get_website_search_result_table($hits, false);
+    $content .= $results['pager'];
+
+    return $content;
   } catch (Exception $exception) {
     watchdog(
       'tripal_elasticsearch',
@@ -123,6 +124,40 @@ function tripal_elasticsearch_advanced_search_results(array $values) {
   }
 }
 
+/**
+ * @param $values
+ *
+ * @return array
+ * @throws \Exception
+ */
+function tripal_elasticsearch_perform_advanced_search($values, $per_page = NULL) {
+  $fields = ['content.*', 'title'];
+
+  if (isset($values['field']) && !empty($values['field'])) {
+    $fields = ["content.{$_GET['field']}"];
+  }
+
+  $model = new \ES\Models\Model();
+  $model->setIndexName('entities');
+
+  if (!empty($values['term'])) {
+    $model->where($fields, trim($values['term']));
+  }
+
+  $category = trim($values['category'] ?? '');
+  if (!empty($category)) {
+    $model->where('bundle_label', "\"$category\"");
+  }
+
+  $model->highlight(['content.*', 'title']);
+
+  if (is_null($per_page)) {
+    return $model->search();
+  }
+
+  return $model->paginate($per_page);
+}
+
 /**
  * AJAX Callback.
  *

diff --git a/theme/templates/elasticsearch_results_header.tpl.php b/theme/templates/elasticsearch_results_header.tpl.php
@@ -0,0 +1,9 @@
+<p>
+    <span>
+        <b><?php print number_format($total) ?> results found</b>
+      <?php if (isset($time)): ?>
+          (<?php print number_format($time, 2) ?> seconds)
+      <?php endif; ?>
+    </span>
+    <span style="float: right">Page <?php print $page ?> of <?php print $pages ?></span>
+</p>
diff --git a/tripal_elasticsearch.api.inc b/tripal_elasticsearch.api.inc
@@ -238,10 +238,6 @@ function tripal_elasticsearch_get_website_search_result_table($search_results, $
     $row = new stdClass();
     $row->title = $item->title;
     if (isset($item->nid)) {
-      //      $row = '<h3>' . l($item->title, 'node/' . $item->nid) . '</h3>';
-      //      if ($base_url) {
-      //        $row = '<h3><a href="' . $base_url . '/node/' . $item->nid . '">' . $item->title . '</a></h3>';
-      //      }
 
       if ($base_url) {
         $row->url = "{$base_url}/node/{$item->nid}";
@@ -251,11 +247,6 @@ function tripal_elasticsearch_get_website_search_result_table($search_results, $
       }
     }
     else {
-      //      $row = '<h3>' . l($item->title, 'bio_data/' . $item->entity_id) . '</h3>';
-      //      if ($base_url) {
-      //        $row = '<h3><a href="' . $base_url . '/bio_data/' . $item->entity_id . '">' . $item->title . '</a></h3>';
-      //      }
-
       if ($base_url) {
         $row->url = "{$base_url}/bio_data/{$item->entity_id}";
       }
@@ -285,13 +276,6 @@ function tripal_elasticsearch_get_website_search_result_table($search_results, $
     $rows[] = $row;
   }
 
-  //  $output = theme('table', [
-  //    'header' => [],
-  //    'rows' => $rows,
-  //    'attributes' => [
-  //      'id' => 'es-search-results-table',
-  //    ],
-  //  ]);
   $output = theme('elasticsearch_results', [
     'rows' => $rows,
     'base_url' => $base_url

diff --git a/tripal_elasticsearch.module b/tripal_elasticsearch.module
@@ -388,6 +388,17 @@ function tripal_elasticsearch_theme($existing, $type, $theme, $path) {
         'message' => ''
       ]
     ],
+    'elasticsearch_results_header' => [
+      'template' => 'elasticsearch_results_header',
+      'render element' => 'content',
+      'path' => "$path/theme/templates",
+      'variables' => [
+        // Current page
+        'page' => 1,
+        'total' => 0,
+        'pages' => 1
+      ]
+    ],
     'elasticsearch_results' => [
       'template' => 'elasticsearch_results',
       // 'render element' => 'elements',