From b580f8a2c33604ee3545304dcd6fbc6a25457e17 Mon Sep 17 00:00:00 2001
From: Steve Taylor <steve_taylor@wsu.edu>
Date: Fri, 21 Jan 2022 13:47:19 -0800
Subject: [PATCH] Pull regex from config

---
 service.inc | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/service.inc b/service.inc
index 0d1bb7a3..4873e432 100644
--- a/service.inc
+++ b/service.inc
@@ -1041,7 +1041,8 @@ class SearchApiDbService extends SearchApiAbstractService {
           $value = format_date($value, 'custom', 'Y y F M n m j d l D');
         }
         $ret = array();
-        foreach (preg_split('/[^\p{L}\p{N}]+/u', $value, -1, PREG_SPLIT_NO_EMPTY) as $v) {
+        $spaces_regex = ma_search_api_get_whitespace_tokenizer_regex();
+        foreach (preg_split('/' . $spaces_regex . '+/u', $value, -1, PREG_SPLIT_NO_EMPTY) as $v) {
           if ($v) {
             $ret[] = array(
               'value' => $v,
@@ -1058,7 +1059,8 @@ class SearchApiDbService extends SearchApiAbstractService {
             $score = $v['score'];
             $v = $v['value'];
             if (drupal_strlen($v) > 50) {
-              $words = preg_split('/[^\p{L}\p{N}]+/u', $v, -1, PREG_SPLIT_NO_EMPTY);
+              $spaces_regex = ma_search_api_get_whitespace_tokenizer_regex();
+              $words = preg_split('/' . $spaces_regex . '+/u', $v, -1, PREG_SPLIT_NO_EMPTY);
               if (count($words) > 1 && max(array_map('drupal_strlen', $words)) <= 50) {
                 // Overlong token is due to bad tokenizing.
                 // Check for "Tokenizer" preprocessor on index.
@@ -1419,7 +1421,8 @@ class SearchApiDbService extends SearchApiAbstractService {
         $words = array_filter(explode(' ', $proc), 'strlen');
       }
       else {
-        $words = preg_split('/[^\p{L}\p{N}]+/u', $proc, -1, PREG_SPLIT_NO_EMPTY);
+        $spaces_regex = ma_search_api_get_whitespace_tokenizer_regex();
+        $words = preg_split('/' . $spaces_regex . '+/u', $proc, -1, PREG_SPLIT_NO_EMPTY);
       }

       if (count($words) > 1) {
@@ -2165,7 +2168,8 @@ class SearchApiDbService extends SearchApiAbstractService {

     // Also collect all keywords already contained in the query so we don't
     // suggest them.
-    $keys = drupal_map_assoc(preg_split('/[^\p{L}\p{N}]+/u', $user_input, -1, PREG_SPLIT_NO_EMPTY));
+    $spaces_regex = ma_search_api_get_whitespace_tokenizer_regex();
+    $keys = drupal_map_assoc(preg_split('/' . $spaces_regex . '+/u', $user_input, -1, PREG_SPLIT_NO_EMPTY));
     if ($incomplete_key) {
       $keys[$incomplete_key] = $incomplete_key;
     }
--
2.25.1