Fix alt text stripping accents and single quotes (#2514)

* Fix alt text stripping accents and single quotes

The altTextFrom method doesn't work well for non english characters and accents, it also uppercases every word which is weird

* Add tests and fix replace accents

Author: Tofandel

src/Helpers/media_library_helpers.php

@@ -3,6 +3,7 @@
 use Illuminate\Support\Facades\Storage;
 use Aws\S3\S3Client;
 use Aws\S3\PostObjectV4;
+use Illuminate\Support\Str;
 
 if (!function_exists('s3Endpoint')) {
     /**
@@ -60,13 +61,11 @@ function bytesToHuman($bytes)
     /**
      * @param string $str
      * @return bool|string
+     * @deprecated Use Str::ascii instead
      */
     function replaceAccents($str)
     {
-        if (function_exists('mb_convert_encoding')) {
-            return mb_convert_encoding($str, 'ASCII', 'UTF-8');
-        }
-        return iconv('UTF-8', 'ASCII//TRANSLIT', $str);
+        return Str::ascii($str);
     }
 }
 
@@ -77,7 +76,7 @@ function replaceAccents($str)
      */
     function sanitizeFilename($filename)
     {
-        $sanitizedFilename = replaceAccents($filename);
+        $sanitizedFilename = Str::ascii($filename);
 
         $invalid = array(
             ' ' => '-',
@@ -87,8 +86,8 @@ function sanitizeFilename($filename)
 
         $sanitizedFilename = str_replace(array_keys($invalid), array_values($invalid), $sanitizedFilename);
 
-        $sanitizedFilename = preg_replace('#[^A-Za-z0-9-\. ]#', '', $sanitizedFilename); // Remove all non-alphanumeric except .
-        $sanitizedFilename = preg_replace('#\.(?=.*\.)#', '', $sanitizedFilename); // Remove all but last .
+        $sanitizedFilename = preg_replace('#[^A-Za-z0-9-. ]#', '', $sanitizedFilename); // Remove all non-alphanumeric except .
+        $sanitizedFilename = preg_replace('#\.(?=.*\.)#', '-', $sanitizedFilename); // Remove all but last .
         $sanitizedFilename = preg_replace('#-+#', '-', $sanitizedFilename); // Replace any more than one - in a row
         $sanitizedFilename = str_replace('-.', '.', $sanitizedFilename); // Remove last - if at the end
         $sanitizedFilename = strtolower($sanitizedFilename); // Lowercase

src/Models/Media.php

@@ -49,10 +49,10 @@ public function altTextFrom($filename)
     {
         $filename = pathinfo($filename, PATHINFO_FILENAME);
         if (Str::endsWith($filename, '@2x')) {
-            $filename = substr($filename, 0, -2);
+            $filename = substr($filename, 0, -3);
         }
 
-        return ucwords(preg_replace('/[^a-zA-Z0-9]/', ' ', sanitizeFilename($filename)));
+        return Str::ucfirst(preg_replace('/[-_]/', ' ', $filename));
     }
 
     public function canDeleteSafely()

tests/unit/Helpers/MediaLibraryHelpersTest.php

@@ -0,0 +1,19 @@
+<?php
+
+namespace A17\Twill\Tests\Unit\Helpers;
+
+use A17\Twill\Tests\Unit\TestCase;
+
+class MediaLibraryHelpersTest extends TestCase
+{
+    public function testReplaceAccents()
+    {
+        $this->assertEquals('aeeiou', replaceAccents('àéèïôû'));
+    }
+
+    public function testSanitizeFilename()
+    {
+        $this->assertEquals('happy-paques-xo-png.jpg', sanitizeFilename('Happy_Pâques - XO.png.jpg'));
+    }
+
+}

tests/unit/Models/MediaTest.php

@@ -0,0 +1,18 @@
+<?php
+
+namespace A17\Twill\Tests\Unit\Models;
+
+use A17\Twill\Models\Media;
+use A17\Twill\Tests\Unit\TestCase;
+
+class MediaTest extends TestCase
+{
+    public function testAltText()
+    {
+        $m = new Media();
+
+        $this->assertEquals("Happy Holidays", $m->altTextFrom('Happy_Holidays.jpg'));
+        $this->assertEquals("Happy Holidays", $m->altTextFrom('Happy_Holidays@2x.jpg'));
+        $this->assertEquals("J'aime la pièce", $m->altTextFrom('J\'aime-la-pièce.jpg'));
+    }
+}