[APC-3] CBOR Implementation (TS)

src/CBOR/CBOR.php

@@ -11,6 +11,7 @@
 
 namespace ATProto\Core\CBOR;
 
+use ATProto\Core\CBOR\MajorTypes\TextString;
 use ATProto\Core\CBOR\MajorTypes\UnsignedInteger;
 
 class CBOR
@@ -21,13 +22,24 @@ public static function encode(string|int|array $data): string
             case 'integer':
                 return UnsignedInteger::encode($data);
             break;
+            case 'string':
+                return TextString::encode($data);
+            break;
         }
 
         throw new \ValueError("Unsupported type: " . gettype($data));
     }
 
-    public static function decode(string $data): int
+    public static function decode(string $data): int|string
     {
-        return UnsignedInteger::decode((string) $data);
+        if (TextString::validate($data)) {
+            return TextString::decode($data);
+        }
+
+        if (UnsignedInteger::validate($data)) {
+            return UnsignedInteger::decode($data);
+        }
+
+        throw new \ValueError("Unsupported type.");
     }
 }

src/CBOR/MajorTypes/TextString.php

@@ -0,0 +1,75 @@
+<?php declare(strict_types = 1);
+
+/**
+ * This file is part of the ATProto Core package.
+ *
+ * (c) Core Branch
+ *
+ * This source code is licensed under the MIT license found in the
+ * LICENSE file in the root directory of this source tree.
+ */
+
+namespace ATProto\Core\CBOR\MajorTypes;
+
+class TextString
+{
+    public static function encode(string $input): string
+    {
+        $length = strlen($input);
+
+        if ($length <= 23) {
+            $header = chr((0x03 << 5) | $length);
+        } elseif ($length <= 0xFF) {
+            $header = chr((0x03 << 5) | 24) . chr($length);
+        } elseif ($length <= 0xFFFF) {
+            $header = chr((0x03 << 5) | 25) . pack('n', $length);
+        } elseif ($length <= 0xFFFFFFFF) {
+            $header = chr((0x03 << 5) | 26) . pack('N', $length);
+        } else {
+            $header = chr((0x03 << 5) | 27) . pack('J', $length);
+        }
+
+        return $header . $input;
+    }
+
+    public static function decode(string $input): string
+    {
+        if (! self::validate($input)) {
+            throw new \ValueError('Invalid CBOR TextString major type.');
+        }
+
+        $additionalInfo = ord($input[0]) & 0x1F;
+        $offset = 1;
+
+        if ($additionalInfo <= 23) {
+            $length = $additionalInfo;
+        } elseif ($additionalInfo === 24) {
+            $length = ord($input[$offset]);
+            $offset += 1;
+        } elseif ($additionalInfo === 25) {
+            $length = unpack('n', substr($input, $offset, 2))[1];
+            $offset += 2;
+        } elseif ($additionalInfo === 26) {
+            $length = unpack('N', substr($input, $offset, 4))[1];
+            $offset += 4;
+        } elseif ($additionalInfo === 27) {
+            $length = unpack('J', substr($input, $offset, 8))[1];
+            $offset += 8;
+        } else {
+            throw new \ValueError('Invalid CBOR TextString length information.');
+        }
+
+        $text = substr($input, $offset, $length);
+
+        if (strlen($text) !== $length) {
+            throw new \ValueError('Invalid CBOR TextString length mismatch.');
+        }
+
+        return $text;
+    }
+
+    public static function validate(string $input): bool
+    {
+        return ((ord($input[0]) >> 5) & 0x07) === 0x03;
+    }
+}

src/CBOR/MajorTypes/UnsignedInteger.php

@@ -17,14 +17,12 @@ class UnsignedInteger
 {
     public static function decode(string $data): int
     {
-        $firstByte = ord($data[0]);
-        $majorType = ($firstByte >> 5) & 0x07;
-        $additionalInfo = $firstByte & 0x1F;
-
-        if ($majorType !== 0) {
-            throw new ValueError("Invalid major type for unsigned integer: $majorType");
+        if (! self::validate($data)) {
+            throw new ValueError("Invalid major type for unsigned integer.");
         }
 
+        $additionalInfo = ord($data[0]) & 0x1F;
+
         if ($additionalInfo <= 23) {
             $value = $additionalInfo;
         } elseif ($additionalInfo === 24) {
@@ -73,4 +71,9 @@ public static function encode(int $value): string
 
         return $prefixedPack('J', "\x1B");
     }
+
+    public static function validate(string $input): bool
+    {
+        return ((ord($input[0]) >> 5) & 0x07) === 0x00;
+    }
 }

tests/Unit/CBOR/CBORTest.php

@@ -18,20 +18,36 @@
 class CBORTest extends TestCase
 {
     #[DataProvider('validCases')]
-    public function testEncode(int $data, string $expected): void
+    public function testEncode(int|string $data, string $expected): void
     {
         $encoded = CBOR::encode($data);
         $this->assertSame($expected, $encoded);
     }
 
     #[DataProvider('validCases')]
-    public function testDecode(int $expected, string $data): void
+    public function testDecode(int|string $expected, string $data): void
     {
         $actual = CBOR::decode($data);
 
         $this->assertSame($expected, $actual);
     }
 
+    public function testCBORDecodeThrowsExceptionWhenPassedUnsupportedType(): void
+    {
+        $this->expectException(\ValueError::class);
+        $this->expectExceptionMessage('Unsupported type.');
+
+        CBOR::decode("\x80"); // CBOR array
+    }
+
+    public function testCBOREncodeThrowsExceptionWhenPassedUnsupportedType(): void
+    {
+        $this->expectException(\ValueError::class);
+        $this->expectExceptionMessage('Unsupported type: array');
+
+        CBOR::encode(array());
+    }
+
     /**
      * @return array[]
      */
@@ -42,8 +58,8 @@ public static function validCases(): array
             [1, hex2bin('01')], // 1 encoded as CBOR unsigned integer
             [10, hex2bin('0a')], // 10 encoded as CBOR unsigned integer
 
-//            // String test cases
-//            [['hello'], hex2bin('6568656c6c6f')], // "hello" encoded as CBOR text string
+            // String test cases
+            ['hello', hex2bin('6568656C6C6F')], // "hello" encoded as CBOR text string
 //
 //            // Boolean test cases
 //            [[true], hex2bin('f5')], // true encoded as CBOR special type

tests/Unit/CBOR/MajorTypes/TextStringTest.php

@@ -0,0 +1,87 @@
+<?php declare(strict_types = 1);
+
+/**
+ * This file is part of the ATProto Core package.
+ *
+ * (c) Core Branch
+ *
+ * This source code is licensed under the MIT license found in the
+ * LICENSE file in the root directory of this source tree.
+ */
+
+namespace Tests\Unit\CBOR\MajorTypes;
+
+use ATProto\Core\CBOR\MajorTypes\TextString;
+use PHPUnit\Framework\Attributes\DataProvider;
+use PHPUnit\Framework\TestCase;
+
+class TextStringTest extends TestCase
+{
+    #[DataProvider('provideValidCases')]
+    public function testEncodeProducesCorrectCBORRepresentation(string $input, string $expectedHeader): void
+    {
+        $actual = bin2hex(TextString::encode($input));
+        $expected = bin2hex($expectedHeader . $input);
+
+        $this->assertSame($expected, $actual);
+    }
+
+    #[DataProvider('provideValidCases')]
+    public function testDecodeExtractsOriginalStringFromCBORRepresentation(string $input, string $header): void
+    {
+        $encoded = $header . $input;
+
+        $actual = TextString::decode($encoded);
+        $expected = $input;
+
+        $this->assertSame($expected, $actual);
+    }
+
+    public static function provideValidCases(): array
+    {
+        return [
+            ["f", "\x61"],
+            ["fo", "\x62"],
+            ["foo", "\x63"],
+            ["foob", "\x64"],
+            ["fooba", "\x65"],
+            ["foobar", "\x66"],
+            [
+                "This is a longer string. This is a longer string. This is a longer string. This is a longer string.",
+                "\x78\x63"
+            ],
+            [
+                "This is a longer string. This is a longer string. This is a longer string. This is a longer string. 
+                This is a longer string. This is a longer string. This is a longer string. This is a longer string. 
+                This is a longer string. This is a longer string. This is a longer string. This is a longer string.",
+                "\x79\x01\x4D"
+            ]
+        ];
+    }
+
+    public function testDecodeThrowsExceptionForInvalidMajorType(): void
+    {
+        $this->expectException(\ValueError::class);
+        $this->expectExceptionMessage("Invalid CBOR TextString major type.");
+
+        TextString::decode("\x0C");
+    }
+
+    public function testDecodeThrowsExceptionForLengthMismatch(): void
+    {
+        $this->expectException(\ValueError::class);
+        $this->expectExceptionMessage("Invalid CBOR TextString length mismatch.");
+
+        // Encoded length is 5, but actual length is 4
+        TextString::decode("\x65\x66\x6F\x6F\x62");
+    }
+
+    public function testDecodeThrowsExceptionForInvalidAdditionalInformation(): void
+    {
+        $this->expectException(\ValueError::class);
+        $this->expectExceptionMessage("Invalid CBOR TextString length information.");
+
+        // Additional info 28 is invalid for text strings
+        TextString::decode("\x7C\x01");
+    }
+}

tests/Unit/CBOR/MajorTypes/UnsignedIntegerTest.php

@@ -47,7 +47,7 @@ public function testEncodeThrowsAnExceptionForValueGreaterThanIntMax(): void
     public function testDecodeThrowsAnExceptionWhenPassedInvalidValue(string $case): void
     {
         $this->expectException(\ValueError::class);
-        $this->expectExceptionMessage("Invalid major type for unsigned integer: ");
+        $this->expectExceptionMessage("Invalid major type for unsigned integer.");
 
         UnsignedInteger::decode($case);
     }