[String] bytesAt() and codePointsAt()

symfony · Oct 14, 2019 · f8d2faa · f8d2faa
1 parent 9d9d095
commit f8d2faa
Show file tree

Hide file tree

Showing 9 changed files with 142 additions and 6 deletions.
diff --git a/src/Symfony/Component/String/AbstractString.php b/src/Symfony/Component/String/AbstractString.php
@@ -225,6 +225,16 @@ public function beforeLast($needle, bool $includeNeedle = false, int $offset = 0
         return $this->slice(0, $i);
     }
 
+    /**
+     * @return int[]
+     */
+    public function bytesAt(int $offset): array
+    {
+        $str = $this->slice($offset, 1);
+
+        return '' === $str->string ? [] : array_values(unpack('C*', $str->string));
+    }
+
     /**
      * @return static
      */

diff --git a/src/Symfony/Component/String/AbstractUnicodeString.php b/src/Symfony/Component/String/AbstractUnicodeString.php
@@ -159,11 +159,14 @@ public function camel(): parent
         return $str;
     }
 
-    public function codePoint(int $offset = 0): ?int
+    /**
+     * @return int[]
+     */
+    public function codePointsAt(int $offset): array
     {
-        $str = $offset ? $this->slice($offset, 1) : $this;
+        $str = $this->slice($offset, 1);
 
-        return '' === $str->string ? null : mb_ord($str->string);
+        return '' === $str->string ? [] : array_map('mb_ord', preg_split('//u', $str->string, -1, PREG_SPLIT_NO_EMPTY));
     }
 
     public function folded(bool $compat = true): parent

diff --git a/src/Symfony/Component/String/ByteString.php b/src/Symfony/Component/String/ByteString.php
@@ -43,11 +43,11 @@ public static function fromRandom(int $length = 16): self
         return new static(substr($string, 0, $length));
     }
 
-    public function byteCode(int $offset = 0): ?int
+    public function bytesAt(int $offset): array
     {
-        $str = $offset ? $this->slice($offset, 1) : $this;
+        $str = $this->string[$offset] ?? '';
 
-        return '' === $str->string ? null : \ord($str->string);
+        return '' === $str ? [] : [\ord($str)];
     }
 
     public function append(string ...$suffix): parent

diff --git a/src/Symfony/Component/String/CodePointString.php b/src/Symfony/Component/String/CodePointString.php
@@ -75,6 +75,13 @@ public function chunk(int $length = 1): array
         return $chunks;
     }
 
+    public function codePointsAt(int $offset): array
+    {
+        $str = $offset ? $this->slice($offset, 1) : $this;
+
+        return '' === $str->string ? [] : [mb_ord($str->string)];
+    }
+
     public function endsWith($suffix): bool
     {
         if ($suffix instanceof AbstractString) {

diff --git a/src/Symfony/Component/String/Tests/AbstractAsciiTestCase.php b/src/Symfony/Component/String/Tests/AbstractAsciiTestCase.php
@@ -34,6 +34,27 @@ public function testCreateFromEmptyString()
         $this->assertTrue($instance->isEmpty());
     }
 
+    /**
+     * @dataProvider provideBytesAt
+     */
+    public function testBytesAt(array $expected, string $string, int $offset, int $form = null)
+    {
+        $instance = static::createFromString($string);
+        $instance = $form ? $instance->normalize($form) : $instance;
+
+        $this->assertSame($expected, $instance->bytesAt($offset));
+    }
+
+    public static function provideBytesAt(): array
+    {
+        return [
+            [[], '', 0],
+            [[], 'a', 1],
+            [[0x62], 'abc', 1],
+            [[0x63], 'abcde', -3],
+        ];
+    }
+
     /**
      * @dataProvider provideWrap
      */

diff --git a/src/Symfony/Component/String/Tests/AbstractUnicodeTestCase.php b/src/Symfony/Component/String/Tests/AbstractUnicodeTestCase.php
@@ -34,6 +34,40 @@ public function provideCreateFromCodePoint(): array
         ];
     }
 
+    public static function provideBytesAt(): array
+    {
+        return array_merge(
+            parent::provideBytesAt(),
+            [
+                [[0xC3, 0xA4], 'Späßchen', 2],
+                [[0xC3, 0x9F], 'Späßchen', -5],
+            ]
+        );
+    }
+
+    /**
+     * @dataProvider provideCodePointsAt
+     */
+    public function testCodePointsAt(array $expected, string $string, int $offset, int $form = null)
+    {
+        $instance = static::createFromString($string);
+        $instance = $form ? $instance->normalize($form) : $instance;
+
+        $this->assertSame($expected, $instance->codePointsAt($offset));
+    }
+
+    public static function provideCodePointsAt(): array
+    {
+        return [
+            [[], '', 0],
+            [[], 'a', 1],
+            [[0x53], 'Späßchen', 0],
+            [[0xE4], 'Späßchen', 2],
+            [[0xDF], 'Späßchen', -5],
+            [[0x260E], '☢☎❄', 1],
+        ];
+    }
+
     public static function provideLength(): array
     {
         return [

diff --git a/src/Symfony/Component/String/Tests/ByteStringTest.php b/src/Symfony/Component/String/Tests/ByteStringTest.php
@@ -21,6 +21,19 @@ protected static function createFromString(string $string): AbstractString
         return new ByteString($string);
     }
 
+    public static function provideBytesAt(): array
+    {
+        return array_merge(
+            parent::provideBytesAt(),
+            [
+                [[0xC3], 'Späßchen', 2],
+                [[0x61], "Spa\u{0308}ßchen", 2],
+                [[0xCC], "Spa\u{0308}ßchen", 3],
+                [[0xE0], 'नमस्ते', 6],
+            ]
+        );
+    }
+
     public static function provideLength(): array
     {
         return array_merge(

diff --git a/src/Symfony/Component/String/Tests/CodePointStringTest.php b/src/Symfony/Component/String/Tests/CodePointStringTest.php
@@ -31,4 +31,28 @@ public static function provideLength(): array
             ]
         );
     }
+
+    public static function provideBytesAt(): array
+    {
+        return array_merge(
+            parent::provideBytesAt(),
+            [
+                [[0x61], "Spa\u{0308}ßchen", 2],
+                [[0xCC, 0x88], "Spa\u{0308}ßchen", 3],
+                [[0xE0, 0xA5, 0x8D], 'नमस्ते', 3],
+            ]
+        );
+    }
+
+    public static function provideCodePointsAt(): array
+    {
+        return array_merge(
+            parent::provideCodePointsAt(),
+            [
+                [[0x61], "Spa\u{0308}ßchen", 2],
+                [[0x0308], "Spa\u{0308}ßchen", 3],
+                [[0x094D], 'नमस्ते', 3],
+            ]
+        );
+    }
 }
diff --git a/src/Symfony/Component/String/Tests/UnicodeStringTest.php b/src/Symfony/Component/String/Tests/UnicodeStringTest.php
@@ -90,6 +90,30 @@ public static function provideChunk(): array
         );
     }
 
+    public static function provideBytesAt(): array
+    {
+        return array_merge(
+            parent::provideBytesAt(),
+            [
+                [[0xC3, 0xA4], "Spa\u{0308}ßchen", 2],
+                [[0x61, 0xCC, 0x88], "Spa\u{0308}ßchen", 2, UnicodeString::NFD],
+                [[0xE0, 0xA4, 0xB8, 0xE0, 0xA5, 0x8D], 'नमस्ते', 2],
+            ]
+        );
+    }
+
+    public static function provideCodePointsAt(): array
+    {
+        return array_merge(
+            parent::provideCodePointsAt(),
+            [
+                [[0xE4], "Spa\u{0308}ßchen", 2],
+                [[0x61, 0x0308], "Spa\u{0308}ßchen", 2, UnicodeString::NFD],
+                [[0x0938, 0x094D], 'नमस्ते', 2],
+            ]
+        );
+    }
+
     public static function provideLower(): array
     {
         return array_merge(