From d8b4ca91d90bc8ee85786558ff4fbe669a73872c Mon Sep 17 00:00:00 2001
From: Sam Minnee <sam@silverstripe.com>
Date: Mon, 6 Nov 2017 12:08:29 +1300
Subject: [PATCH] FIX: Fix ContextSummary behaviour with UTF8 chars

Fixes https://github.com/silverstripe/silverstripe-framework/issues/1396

Ensure that DBTextTest has UTF8 test cases
---
 src/ORM/FieldType/DBText.php | 10 +++++-----
 tests/php/ORM/DBTextTest.php | 24 +++++++++++++++++++++++-
 2 files changed, 28 insertions(+), 6 deletions(-)

diff --git a/src/ORM/FieldType/DBText.php b/src/ORM/FieldType/DBText.php
index 1e6066727..494d21986 100644
--- a/src/ORM/FieldType/DBText.php
+++ b/src/ORM/FieldType/DBText.php
@@ -198,7 +198,7 @@ class DBText extends DBString
         $keywords = Convert::raw2xml($keywords);
 
         // Find the search string
-        $position = (int) stripos($text, $keywords);
+        $position = (int) mb_stripos($text, $keywords);
 
         // We want to search string to be in the middle of our block to give it some context
         $position = max(0, $position - ($characters / 2));
@@ -206,19 +206,19 @@ class DBText extends DBString
         if ($position > 0) {
             // We don't want to start mid-word
             $position = max(
-                (int) strrpos(substr($text, 0, $position), ' '),
-                (int) strrpos(substr($text, 0, $position), "\n")
+                (int) mb_strrpos(substr($text, 0, $position), ' '),
+                (int) mb_strrpos(substr($text, 0, $position), "\n")
             );
         }
 
-        $summary = substr($text, $position, $characters);
+        $summary = mb_substr($text, $position, $characters);
         $stringPieces = explode(' ', $keywords);
 
         if ($highlight) {
             // Add a span around all key words from the search term as well
             if ($stringPieces) {
                 foreach ($stringPieces as $stringPiece) {
-                    if (strlen($stringPiece) > 2) {
+                    if (mb_strlen($stringPiece) > 2) {
                         // Maintain case of original string
                         $summary = preg_replace(
                             '/' . preg_quote($stringPiece, '/') . '/i',
diff --git a/tests/php/ORM/DBTextTest.php b/tests/php/ORM/DBTextTest.php
index e5985cc33..2fde108a0 100644
--- a/tests/php/ORM/DBTextTest.php
+++ b/tests/php/ORM/DBTextTest.php
@@ -41,6 +41,7 @@ class DBTextTest extends SapphireTest
             ['The little brown fox jumped over the lazy cow.', 'The little brown fox...'],
             ['<p>Short & Sweet</p>', '&lt;p&gt;Short &amp; Sweet&lt;/p&gt;'],
             ['This text contains &amp; in it', 'This text contains &amp;...'],
+            ['Is an umault in schön?', 'Is an umault in schö...'],
         ];
     }
 
@@ -80,7 +81,10 @@ class DBTextTest extends SapphireTest
             ['<p>Lorem ipsum dolor sit amet</p>', 24, '&lt;p&gt;Lorem ipsum dolor...'],
             ['<p><span>Lorem ipsum dolor sit amet</span></p>', 24, '&lt;p&gt;&lt;span&gt;Lorem ipsum...'],
             ['<p>Lorem ipsum</p>', 24, '&lt;p&gt;Lorem ipsum&lt;/p&gt;'],
-            ['Lorem &amp; ipsum dolor sit amet', 24, 'Lorem &amp;amp; ipsum dolor...']
+            ['Lorem &amp; ipsum dolor sit amet', 24, 'Lorem &amp;amp; ipsum dolor...'],
+
+            ['Is an umault in schön or not?', 22, 'Is an umault in schön...'],
+
         ];
     }
 
@@ -124,6 +128,9 @@ class DBTextTest extends SapphireTest
             // If storing HTML you should use DBHTMLText instead
             ['<p>Text inside a paragraph tag should also work</p>', 3, '&lt;p&gt;Text inside a...'],
             ['<p>Two words</p>', 3, '&lt;p&gt;Two words&lt;/p&gt;'],
+
+            // Check UTF8
+            ['Is an umault in schön or not?', 5, 'Is an umault in schön...'],
         ];
     }
 
@@ -156,6 +163,9 @@ class DBTextTest extends SapphireTest
             // If storing HTML you should use DBHTMLText instead
             ['<p>First sentence.</p>', 2, '&lt;p&gt;First sentence.&lt;/p&gt;'],
             ['<p>First sentence. Second sentence. Third sentence</p>', 2, '&lt;p&gt;First sentence. Second sentence.'],
+
+            // Check UTF8
+            ['Is schön. Isn\'t schön.', 1, 'Is schön.'],
         ];
     }
 
@@ -187,6 +197,9 @@ class DBTextTest extends SapphireTest
             // If storing HTML you should use DBHTMLText instead
             ['<br />First sentence.', '&lt;br /&gt;First sentence.'],
             ['<p>First sentence. Second sentence. Third sentence</p>', '&lt;p&gt;First sentence.'],
+
+            // Check UTF8
+            ['Is schön. Isn\'t schön.', 'Is schön.'],
         ];
     }
 
@@ -243,7 +256,16 @@ class DBTextTest extends SapphireTest
                 'ate',
                 // it should highlight 3 letters or more.
                 'A dog <mark>ate</mark> a cat while looking at a Foobar',
+            ],
+            [
+                'both schön and können have umlauts',
+                21,
+                'schön',
+                // check UTF8 support
+                'both <mark>schön</mark> and können...',
             ]
+
+
         ];
     }