Added the Porter Stemmer module to improve searches. This doesn't deal with some...

[yaffs-website] / web / modules / contrib / porterstemmer / porterstemmer.module

diff --git a/web/modules/contrib/porterstemmer/porterstemmer.module b/web/modules/contrib/porterstemmer/porterstemmer.module
new file mode 100644 (file)
index 0000000..0a886af
--- /dev/null
+++ b/web/modules/contrib/porterstemmer/porterstemmer.module
@@ -0,0 +1,98 @@
+<?php
+
+/**
+ * @file
+ * Contains porterstemmer.module.
+ */
+
+use Drupal\Core\Routing\RouteMatchInterface;
+use Drupal\porterstemmer\Porter2;
+
+/**
+ * Regular expression defining a word boundary for Porter Stemmer.
+ *
+ * A word boundary is anything not a letter or an apostrophe.
+ */
+define('PORTERSTEMMER_BOUNDARY', "[^a-zA-Z']+");
+
+/**
+ * Implements hook_help().
+ */
+function porterstemmer_help($route_name, RouteMatchInterface $route_match) {
+  switch ($route_name) {
+    // Main module help for the porterstemmer module.
+    case 'help.page.porterstemmer':
+      $output = '';
+      $output .= '<h3>' . t('About') . '</h3>';
+      $output .= '<p>' . t('Improves American English language searching by simplifying related words to their root (conjugations, plurals, ...).') . '</p>';
+      return $output;
+
+    default:
+  }
+}
+
+/**
+ * Implements hook_search_preprocess().
+ *
+ * Stems the words in $text, using the Porter 2 (English) stemming algorithm.
+ */
+function porterstemmer_search_preprocess($text, $langcode = NULL) {
+  // If the language is not set, get it from the language manager.
+  if (!isset($langcode)) {
+    $langcode = \Drupal::languageManager()->getCurrentLanguage()->getId();
+  }
+
+  if ($langcode == 'en') {
+    // Convert text to lower case, and replace special apostrophes with regular
+    // apostrophes.
+    $text = strtolower(str_replace('’', "'", $text));
+
+    // Split into words.
+    $words = preg_split('/(' . PORTERSTEMMER_BOUNDARY . '+)/', $text, -1, PREG_SPLIT_DELIM_CAPTURE);
+
+    if (!count($words)) {
+      return $text;
+    }
+
+    $has_pecl_stem = _porterstemmer_pecl_loaded();
+
+    // Process each word, skipping delimiters.
+    $isword = !preg_match('/' . PORTERSTEMMER_BOUNDARY . '/', $words[0]);
+    foreach ($words as $k => $word) {
+      if ($isword) {
+        if ($has_pecl_stem) {
+          $words[$k] = stem_english($word);
+        }
+        else {
+          $words[$k] = Porter2::stem($word);
+        }
+      }
+      $isword = !$isword;
+    }
+
+    // Put it all back together (note that delimiters are in $words).
+    return implode('', $words);
+  }
+
+  return $text;
+}
+
+/**
+ * Checks to see if the PECL stem extension has been loaded.
+ *
+ * @return bool
+ *    TRUE if the stem_english() function from the PECL stem library can be
+ *    used, FALSE if not.
+ */
+function _porterstemmer_pecl_loaded() {
+  static $has_pecl_stem = FALSE;
+  static $already_checked = FALSE;
+
+  if ($already_checked) {
+    return $has_pecl_stem;
+  }
+
+  $has_pecl_stem = extension_loaded('stem') && function_exists('stem_english');
+  $already_checked = TRUE;
+  return $has_pecl_stem;
+}