X-Git-Url: http://source.bookstackapp.com/bookstack/blobdiff_plain/f4449928f83748d015a633cdc1cef50fe822648c..refs/pull/5725/head:/app/Search/SearchIndex.php

diff --git a/app/Search/SearchIndex.php b/app/Search/SearchIndex.php
index a8bd2c4b2..844e3584b 100644
--- a/app/Search/SearchIndex.php
+++ b/app/Search/SearchIndex.php
@@ -16,7 +16,7 @@ class SearchIndex
     /**
      * A list of delimiter characters used to break-up parsed content into terms for indexing.
      */
-    public static string $delimiters = " \n\t.-,!?:;()[]{}<>`'\"";
+    public static string $delimiters = " \n\t.-,!?:;()[]{}<>`'\"«»";
 
     /**
      * A list of delimiter which could be commonly used within a single term and also indicate a break between terms.
@@ -160,7 +160,9 @@ class SearchIndex
         /** @var DOMNode $child */
         foreach ($doc->getBodyChildren() as $child) {
             $nodeName = $child->nodeName;
-            $termCounts = $this->textToTermCountMap(trim($child->textContent));
+            $text = trim($child->textContent);
+            $text = str_replace("\u{00A0}", ' ', $text);
+            $termCounts = $this->textToTermCountMap($text);
             foreach ($termCounts as $term => $count) {
                 $scoreChange = $count * ($elementScoreAdjustmentMap[$nodeName] ?? 1);
                 $scoresByTerm[$term] = ($scoresByTerm[$term] ?? 0) + $scoreChange;