Add S3KeyCleaner (#348)

app/Service/Storage/S3KeyCleaner.php

@@ -0,0 +1,54 @@
+<?php
+
+namespace App\Service\Storage;
+
+class S3KeyCleaner
+{
+    public static function sanitize($objectKey, $separator = '-')
+    {
+        return (new self())->sanitizeS3Key($objectKey, $separator);
+    }
+
+    public function replaceLatinCharacters($value)
+    {
+        // Load the latin map
+        $latinMap = json_decode(\File::get(resource_path('data/latin-map.json')), true);
+
+        $result = '';
+        $length = mb_strlen($value);
+        for ($i = 0; $i < $length; $i++) {
+            $char = mb_substr($value, $i, 1);
+            $result .= array_key_exists($char, $latinMap) ? $latinMap[$char] : $char;
+        }
+        return $result;
+    }
+
+    private function removeIllegalCharacters($value)
+    {
+        $SAFE_CHARACTERS = '/[^0-9a-zA-Z! _\\.\\*\'\\(\\)\\-\\/]/';
+        return preg_replace($SAFE_CHARACTERS, '', $value);
+    }
+
+    private function isValidSeparator($separator)
+    {
+        $SAFE_CHARACTERS = '/[^0-9a-zA-Z! _\\.\\*\'\\(\\)\\-\\/]/';
+        return $separator && !preg_match($SAFE_CHARACTERS, $separator);
+    }
+
+    public function sanitizeS3Key($objectKey, $separator = '-')
+    {
+        if (!$this->isValidSeparator($separator)) {
+            throw new \Exception("${separator} is not a valid separator");
+        }
+
+        if (!$objectKey || (!is_string($objectKey) && !is_numeric($objectKey))) {
+            throw new \Exception("Expected non-empty string or number, got ${objectKey}");
+        }
+
+        if (is_numeric($objectKey)) {
+            return strval($objectKey);
+        }
+
+        return str_replace(' ', $separator, $this->removeIllegalCharacters($this->replaceLatinCharacters(trim($objectKey))));
+    }
+}

app/Service/Storage/StorageFileNameParser.php

@@ -30,10 +30,10 @@ class StorageFileNameParser
     public function getMovedFileName(): ?string
     {
         if ($this->fileName && $this->extension) {
-            $fileName = substr($this->fileName, 0, 50).'_'.$this->uuid.'.'.$this->extension;
+            $fileName = substr($this->fileName, 0, 50) . '_' . $this->uuid . '.' . $this->extension;
             $fileName = preg_replace('#\p{C}+#u', '', $fileName); // avoid CorruptedPathDetected exceptions
 
-            return mb_convert_encoding($fileName, 'UTF-8', 'UTF-8');
+            return S3KeyCleaner::sanitize($fileName);
         }
 
         return $this->uuid;
@@ -47,13 +47,15 @@ class StorageFileNameParser
             return;
         }
 
-        if (! str_contains($fileName, '_')) {
+        if (!str_contains($fileName, '_')) {
             return;
         }
 
         $candidateString = substr($fileName, strrpos($fileName, '_') + 1);
-        if (! str_contains($candidateString, '.')
-            || ! Str::isUuid(substr($candidateString, 0, strpos($candidateString, '.')))) {
+        if (
+            !str_contains($candidateString, '.')
+            || !Str::isUuid(substr($candidateString, 0, strpos($candidateString, '.')))
+        ) {
             return;
         }