From: Dan Brown <redacted>
Date: Sat, 30 Sep 2017 13:14:23 +0000 (+0100)
Subject: Updated search indexer to split words better
X-Git-Tag: v0.18.2~1^2~8
X-Git-Url: http://source.bookstackapp.com/bookstack/commitdiff_plain/5fd04fa470e5438d3afba43ac01f853cca977e5a?ds=inline

Updated search indexer to split words better

Will now split up words based on more chars than just spaces.
Not takes into account newlines, tabs, periods & commas.

Fixed #531
---

diff --git a/app/Services/SearchService.php b/app/Services/SearchService.php
index bb92a1d7c..aebeace1e 100644
--- a/app/Services/SearchService.php
+++ b/app/Services/SearchService.php
@@ -382,11 +382,13 @@ class SearchService
     protected function generateTermArrayFromText($text, $scoreAdjustment = 1)
     {
         $tokenMap = []; // {TextToken => OccurrenceCount}
-        $splitText = explode(' ', $text);
-        foreach ($splitText as $token) {
-            if ($token === '') continue;
+        $splitChars = " \n\t.,";
+        $token = strtok($text, $splitChars);
+
+        while ($token !== false) {
             if (!isset($tokenMap[$token])) $tokenMap[$token] = 0;
             $tokenMap[$token]++;
+            $token = strtok($splitChars);
         }
 
         $terms = [];