Fixed renaming functionality to work with i18n characters (#1607)

Author: stloyd

src/core/etl/src/Flow/ETL/Transformer/RenameAllCaseTransformer.php

@@ -21,23 +21,56 @@ public function transform(Rows $rows, FlowContext $context) : Rows
         return $rows->map(function (Row $row) : Row {
             foreach ($row->entries()->all() as $entry) {
                 if ($this->upper) {
-                    $row = $row->rename($entry->name(), \strtoupper($entry->name()));
+                    $row = $row->rename($entry->name(), \mb_strtoupper($entry->name()));
                 }
 
                 if ($this->lower) {
-                    $row = $row->rename($entry->name(), \strtolower($entry->name()));
+                    $row = $row->rename($entry->name(), \mb_strtolower($entry->name()));
                 }
 
                 if ($this->ucfirst) {
-                    $row = $row->rename($entry->name(), \ucfirst($entry->name()));
+                    $row = $row->rename($entry->name(), $this->ucFirst($entry->name()));
                 }
 
                 if ($this->ucwords) {
-                    $row = $row->rename($entry->name(), \ucwords($entry->name()));
+                    $row = $row->rename($entry->name(), $this->ucWords($entry->name()));
                 }
             }
 
             return $row;
         });
     }
+
+    private function ucFirst(string $string) : string
+    {
+        // Available from PHP 8.4+
+        if (\function_exists('mb_ucfirst')) {
+            return \mb_ucfirst($string);
+        }
+
+        $encoding = \mb_internal_encoding();
+
+        return \mb_strtoupper(\mb_substr($string, 0, 1, $encoding), $encoding) . \mb_substr($string, 1, null, $encoding);
+    }
+
+    private function ucWords(string $string) : string
+    {
+        $result = '';
+        $previousCharacter = ' ';
+
+        $encoding = \mb_internal_encoding();
+
+        for ($i = 0, $length = \mb_strlen($string, $encoding); $i < $length; $i++) {
+            $currentCharacter = \mb_substr($string, $i, 1, $encoding);
+
+            if (' ' === $previousCharacter) {
+                $currentCharacter = \mb_strtoupper($currentCharacter, $encoding);
+            }
+
+            $result .= $currentCharacter;
+            $previousCharacter = $currentCharacter;
+        }
+
+        return $result;
+    }
 }

src/core/etl/tests/Flow/ETL/Tests/Integration/DataFrame/RenameTest.php

@@ -61,6 +61,24 @@ public function test_rename_all_lower_case() : void
         );
     }
 
+    public function test_rename_all_lower_case_i18n() : void
+    {
+        $rows = rows(row(int_entry('ILOŚĆ PRZEDMIOTÓW', 0)), row(int_entry('ILOŚĆ PRZEDMIOTÓW', 10)));
+
+        $ds = df()
+            ->read(from_rows($rows))
+            ->renameAllLowerCase()
+            ->getEachAsArray();
+
+        self::assertEquals(
+            [
+                ['ilość przedmiotów' => 0],
+                ['ilość przedmiotów' => 10],
+            ],
+            \iterator_to_array($ds)
+        );
+    }
+
     public function test_rename_all_to_snake_case() : void
     {
         $rows = rows(row(int_entry('id', 1), str_entry('UserName', 'name'), bool_entry('isActive', true)), row(int_entry('id', 2), str_entry('UserName', 'name'), bool_entry('isActive', false)));
@@ -133,4 +151,22 @@ public function test_rename_all_upper_case_word() : void
             \iterator_to_array($ds)
         );
     }
+
+    public function test_rename_all_upper_case_word_i18n() : void
+    {
+        $rows = rows(row(int_entry('ósmy', 8)), row(int_entry('dziewiąty', 9)));
+
+        $ds = df()
+            ->read(from_rows($rows))
+            ->renameAllUpperCaseWord()
+            ->getEachAsArray();
+
+        self::assertEquals(
+            [
+                ['Ósmy' => 8],
+                ['Dziewiąty' => 9],
+            ],
+            \iterator_to_array($ds)
+        );
+    }
 }