fix: Add support for multibyte strings

system/Commands/Utilities/Namespaces.php

@@ -120,10 +120,10 @@ private function outputAllNamespaces(array $params): array
 
     private function truncate(string $string, int $max): string
     {
-        $length = strlen($string);
+        $length = mb_strlen($string);
 
         if ($length > $max) {
-            return substr($string, 0, $max - 3) . '...';
+            return mb_substr($string, 0, $max - 3) . '...';
         }
 
         return $string;

system/Helpers/text_helper.php

@@ -44,35 +44,40 @@ function word_limiter(string $str, int $limit = 100, string $endChar = '…'
     /**
      * Character Limiter
      *
-     * Limits the string based on the character count.  Preserves complete words
+     * Limits the string based on the character count. Preserves complete words
      * so the character count may not be exactly as specified.
      *
      * @param string $endChar the end character. Usually an ellipsis
      */
-    function character_limiter(string $str, int $n = 500, string $endChar = '…'): string
+    function character_limiter(string $string, int $limit = 500, string $endChar = '…'): string
     {
-        if (mb_strlen($str) < $n) {
-            return $str;
+        if (mb_strlen($string) < $limit) {
+            return $string;
         }
 
         // a bit complicated, but faster than preg_replace with \s+
-        $str = preg_replace('/ {2,}/', ' ', str_replace(["\r", "\n", "\t", "\x0B", "\x0C"], ' ', $str));
+        $string       = preg_replace('/ {2,}/', ' ', str_replace(["\r", "\n", "\t", "\x0B", "\x0C"], ' ', $string));
+        $stringLength = mb_strlen($string);
 
-        if (mb_strlen($str) <= $n) {
-            return $str;
+        if ($stringLength <= $limit) {
+            return $string;
         }
 
-        $out = '';
+        $output       = '';
+        $outputLength = 0;
+        $words        = explode(' ', trim($string));
 
-        foreach (explode(' ', trim($str)) as $val) {
-            $out .= $val . ' ';
-            if (mb_strlen($out) >= $n) {
-                $out = trim($out);
+        foreach ($words as $word) {
+            $output .= $word . ' ';
+            $outputLength = mb_strlen($output);
+
+            if ($outputLength >= $limit) {
+                $output = trim($output);
                 break;
             }
         }
 
-        return (mb_strlen($out) === mb_strlen($str)) ? $out : $out . $endChar;
+        return ($outputLength === $stringLength) ? $output : $output . $endChar;
     }
 }
 
@@ -712,38 +717,44 @@ function alternator(...$args): string
     function excerpt(string $text, ?string $phrase = null, int $radius = 100, string $ellipsis = '...'): string
     {
         if (isset($phrase)) {
-            $phrasePos = stripos($text, $phrase);
-            $phraseLen = strlen($phrase);
+            $phrasePosition = mb_stripos($text, $phrase);
+            $phraseLength   = mb_strlen($phrase);
         } else {
-            $phrasePos = $radius / 2;
-            $phraseLen = 1;
+            $phrasePosition = $radius / 2;
+            $phraseLength   = 1;
         }
 
-        $pre = explode(' ', substr($text, 0, $phrasePos));
-        $pos = explode(' ', substr($text, $phrasePos + $phraseLen));
+        $beforeWords = explode(' ', mb_substr($text, 0, $phrasePosition));
+        $afterWords  = explode(' ', mb_substr($text, $phrasePosition + $phraseLength));
 
-        $prev  = ' ';
-        $post  = ' ';
-        $count = 0;
+        $firstPartOutput = ' ';
+        $endPartOutput   = ' ';
+        $count           = 0;
 
-        foreach (array_reverse($pre) as $e) {
-            if ((strlen($e) + $count + 1) < $radius) {
-                $prev = ' ' . $e . $prev;
+        foreach (array_reverse($beforeWords) as $beforeWord) {
+            $beforeWordLength = mb_strlen($beforeWord);
+
+            if (($beforeWordLength + $count + 1) < $radius) {
+                $firstPartOutput = ' ' . $beforeWord . $firstPartOutput;
             }
-            $count = ++$count + strlen($e);
+
+            $count = ++$count + $beforeWordLength;
         }
 
         $count = 0;
 
-        foreach ($pos as $s) {
-            if ((strlen($s) + $count + 1) < $radius) {
-                $post .= $s . ' ';
+        foreach ($afterWords as $afterWord) {
+            $afterWordLength = mb_strlen($afterWord);
+
+            if (($afterWordLength + $count + 1) < $radius) {
+                $endPartOutput .= $afterWord . ' ';
             }
-            $count = ++$count + strlen($s);
+
+            $count = ++$count + $afterWordLength;
         }
 
         $ellPre = $phrase !== null ? $ellipsis : '';
 
-        return str_replace('  ', ' ', $ellPre . $prev . $phrase . $post . $ellipsis);
+        return str_replace('  ', ' ', $ellPre . $firstPartOutput . $phrase . $endPartOutput . $ellipsis);
     }
 }

system/View/View.php

@@ -339,7 +339,7 @@ public function renderString(string $view, ?array $options = null, ?bool $saveDa
      */
     public function excerpt(string $string, int $length = 20): string
     {
-        return (strlen($string) > $length) ? substr($string, 0, $length - 3) . '...' : $string;
+        return (mb_strlen($string) > $length) ? mb_substr($string, 0, $length - 3) . '...' : $string;
     }
 
     /**

tests/system/Commands/Utilities/NamespacesTest.php

@@ -14,6 +14,7 @@
 namespace CodeIgniter\Commands\Utilities;
 
 use CodeIgniter\Test\CIUnitTestCase;
+use CodeIgniter\Test\ReflectionHelper;
 use CodeIgniter\Test\StreamFilterTrait;
 use PHPUnit\Framework\Attributes\Group;
 
@@ -24,6 +25,7 @@
 final class NamespacesTest extends CIUnitTestCase
 {
     use StreamFilterTrait;
+    use ReflectionHelper;
 
     protected function setUp(): void
     {
@@ -84,4 +86,14 @@ public function testNamespacesCommandAllNamespaces(): void
             str_replace(' ', '', $this->getBuffer())
         );
     }
+
+    public function testTruncateNamespaces(): void
+    {
+        $commandObject  = new Namespaces(service('logger'), service('commands'));
+        $truncateRunner = $this->getPrivateMethodInvoker($commandObject, 'truncate');
+
+        $this->assertSame('App\Controllers\...', $truncateRunner('App\Controllers\Admin', 19));
+        // multibyte namespace
+        $this->assertSame('App\Контроллеры\...', $truncateRunner('App\Контроллеры\Админ', 19));
+    }
 }

tests/system/Helpers/TextHelperTest.php

@@ -24,7 +24,8 @@
 #[Group('Others')]
 final class TextHelperTest extends CIUnitTestCase
 {
-    private string $_long_string = 'Once upon a time, a framework had no tests. It sad. So some nice people began to write tests. The more time that went on, the happier it became. Everyone was happy.';
+    private string $longString   = 'Once upon a time, a framework had no tests. It sad. So some nice people began to write tests. The more time that went on, the happier it became. Everyone was happy.';
+    private string $mbLongString = 'Давным-давно во фреймворке не было тестов. Это печально. И вот несколько хороших людей начали писать тесты. Чем больше времени проходило, тем счастливее становилось. Все были счастливы.';
 
     protected function setUp(): void
     {
@@ -165,19 +166,29 @@ public function testIncrementString(): void
 
     public function testWordLimiter(): void
     {
-        $this->assertSame('Once upon a time,…', word_limiter($this->_long_string, 4));
-        $this->assertSame('Once upon a time,…', word_limiter($this->_long_string, 4, '…'));
+        $this->assertSame('Once upon a time,…', word_limiter($this->longString, 4));
+        $this->assertSame('Once upon a time,…', word_limiter($this->longString, 4, '…'));
         $this->assertSame('', word_limiter('', 4));
-        $this->assertSame('Once upon a…', word_limiter($this->_long_string, 3, '…'));
+        $this->assertSame('Once upon a…', word_limiter($this->longString, 3, '…'));
         $this->assertSame('Once upon a time', word_limiter('Once upon a time', 4, '…'));
+
+        $this->assertSame('Давным-давно во фреймворке не было тестов.…', word_limiter($this->mbLongString, 6));
+        $this->assertSame('Давным-давно во фреймворке не было тестов.…', word_limiter($this->mbLongString, 6, '…'));
+        $this->assertSame('Давным-давно во фреймворке…', word_limiter($this->mbLongString, 3, '…'));
+        $this->assertSame('Давным-давно во фреймворке не было тестов.', word_limiter('Давным-давно во фреймворке не было тестов.', 6, '…'));
     }
 
     public function testCharacterLimiter(): void
     {
-        $this->assertSame('Once upon a time, a…', character_limiter($this->_long_string, 20));
-        $this->assertSame('Once upon a time, a…', character_limiter($this->_long_string, 20, '…'));
+        $this->assertSame('Once upon a time, a…', character_limiter($this->longString, 20));
+        $this->assertSame('Once upon a time, a…', character_limiter($this->longString, 20, '…'));
         $this->assertSame('Short', character_limiter('Short', 20));
         $this->assertSame('Short', character_limiter('Short', 5));
+
+        $this->assertSame('Давным-давно во фреймворке не было тестов.…', character_limiter($this->mbLongString, 41));
+        $this->assertSame('Давным-давно во фреймворке не было тестов.…', character_limiter($this->mbLongString, 41, '…'));
+        $this->assertSame('Короткий', character_limiter('Короткий', 20));
+        $this->assertSame('Короткий', character_limiter('Короткий', 8));
     }
 
     public function testAsciiToEntities(): void
@@ -391,17 +402,25 @@ public function testDefaultWordWrapCharlim(): void
 
     public function testExcerpt(): void
     {
-        $string = $this->_long_string;
+        $string = $this->longString;
         $result = ' Once upon a time, a framework had no tests. It sad  So some nice people began to write tests. The more time that went on, the happier it became. ...';
-        $this->assertSame(excerpt($string), $result);
+        $this->assertSame($result, excerpt($string));
+
+        $multibyteResult = ' Давным-давно во фреймворке не было тестов. Это печ льно. И вот несколько хороших людей начали писать тесты. Чем больше времени проходило, тем ...';
+
+        $this->assertSame($multibyteResult, excerpt($this->mbLongString));
     }
 
     public function testExcerptRadius(): void
     {
-        $string = $this->_long_string;
+        $string = $this->longString;
         $phrase = 'began';
         $result = '... people began to ...';
         $this->assertSame(excerpt($string, $phrase, 10), $result);
+
+        $multibyteResult = '... Это печально . И вот ...';
+
+        $this->assertSame($multibyteResult, excerpt($this->mbLongString, 'печально', 10));
     }
 
     public function testAlternator(): void

tests/system/View/ViewTest.php

@@ -405,4 +405,12 @@ public function testRenderSectionSavingData(): void
         $view->setVar('testString', 'Hello World');
         $this->assertStringContainsString($expected, $view->render('extend_reuse_section'));
     }
+
+    public function testViewExcerpt(): void
+    {
+        $view = new View($this->config, $this->viewsDir, $this->loader);
+
+        $this->assertSame('CodeIgniter is a PHP full-stack web framework...', $view->excerpt('CodeIgniter is a PHP full-stack web framework that is light, fast, flexible and secure.', 48));
+        $this->assertSame('CodeIgniter - это полнофункциональный веб-фреймворк...', $view->excerpt('CodeIgniter - это полнофункциональный веб-фреймворк на PHP, который является легким, быстрым, гибким и безопасным.', 54));
+    }
 }

user_guide_src/source/changelogs/v4.6.0.rst

@@ -168,6 +168,7 @@ Method Signature Changes
 - **Time:** The first parameter type of the ``createFromTimestamp()`` has been
   changed from ``int`` to ``int|float``, and the return type ``static`` has been
   added.
+- **Helpers:** ``character_limiter()`` parameter names have been updated. If you use named arguments, you need to update the function calls.
 
 Removed Type Definitions
 ------------------------
@@ -350,6 +351,8 @@ Bugs Fixed
 - **Response:**
     - Headers set using the ``Response`` class are now prioritized and replace headers
       that can be set manually using the PHP ``header()`` function.
+- **View:** Added support for multibyte strings for ``View::excerpt()``.
+- **Helpers:** Added support for multibyte strings for ``excerpt()``.
 
 See the repo's
 `CHANGELOG.md <https://github.com/codeigniter4/CodeIgniter4/blob/develop/CHANGELOG.md>`_

user_guide_src/source/helpers/text_helper.rst

@@ -166,11 +166,11 @@ The following functions are available:
 
     .. literalinclude:: text_helper/012.php
 
-.. php:function:: word_limiter($str[, $limit = 100[, $end_char = '…']])
+.. php:function:: word_limiter($str[, $limit = 100[, $endChar = '…']])
 
     :param    string    $str: Input string
     :param    int    $limit: Limit
-    :param    string    $end_char: End character (usually an ellipsis)
+    :param    string    $endChar: End character (usually an ellipsis)
     :returns:    Word-limited string
     :rtype:    string
 
@@ -181,11 +181,11 @@ The following functions are available:
     The third parameter is an optional suffix added to the string. By
     default it adds an ellipsis.
 
-.. php:function:: character_limiter($str[, $n = 500[, $end_char = '…']])
+.. php:function:: character_limiter($string[, $limit = 500[, $endChar = '…']])
 
-    :param    string    $str: Input string
-    :param    int    $n: Number of characters
-    :param    string    $end_char: End character (usually an ellipsis)
+    :param    string    $string: Input string
+    :param    int    $limit: Number of characters
+    :param    string    $endChar: End character (usually an ellipsis)
     :returns:    Character-limited string
     :rtype:    string