Merge branch 'fix/float-handling'

8 files changed, 438 insertions, 0 deletions

diff --git a/src/BitOps/BinaryConversion.php b/src/BitOps/BinaryConversion.php
new file mode 100644
index 0000000..14b33c9
--- /dev/null
+++ b/src/BitOps/BinaryConversion.php
@@ -0,0 +1,256 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Nsfisis\Waddiwasi\BitOps;
+
+use function assert;
+use function is_float;
+use function is_int;
+use function ord;
+use function pack;
+use function strlen;
+use function unpack;
+
+final readonly class BinaryConversion
+{
+    private function __construct()
+    {
+    }
+
+    /**
+     * @return non-empty-string
+     */
+    public static function serializeI8(int $x): string
+    {
+        return self::packInt(PackIntSpecifiers::Int8, $x);
+    }
+
+    /**
+     * @return non-empty-string
+     */
+    public static function serializeI16(int $x): string
+    {
+        return self::packInt(PackIntSpecifiers::Int16LittleEndian, $x);
+    }
+
+    /**
+     * @return non-empty-string
+     */
+    public static function serializeI32(int $x): string
+    {
+        return self::packInt(PackIntSpecifiers::Int32LittleEndian, $x);
+    }
+
+    /**
+     * @return non-empty-string
+     */
+    public static function serializeI64(int $x): string
+    {
+        return self::packInt(PackIntSpecifiers::Int64LittleEndian, $x);
+    }
+
+    /**
+     * @return non-empty-string
+     */
+    public static function serializeI64InBigEndian(int $x): string
+    {
+        return self::packInt(PackIntSpecifiers::Int64BigEndian, $x);
+    }
+
+    /**
+     * @return non-empty-string
+     */
+    public static function serializeF32(float $x): string
+    {
+        // PHP's pack() does not preserve NaN payload bits, so we have to
+        // manually check if the float is NaN and convert it to a 32-bit float.
+        if (is_nan($x)) {
+            [$sign, , $payload] = FloatOps::destructF64Bits($x);
+            $i = 0
+                | FloatTraits::getF32SignBit(Signedness::fromSignBit($sign))
+                | FloatTraits::F32_EXPONENT_NAN
+                | ($payload >> (FloatTraits::F64_MANTISSA_BITS - FloatTraits::F32_MANTISSA_BITS));
+            return self::packInt(PackIntSpecifiers::Int32LittleEndian, $i);
+        } else {
+            return self::packFloat(PackFloatSpecifiers::Float32LittleEndian, $x);
+        }
+    }
+
+    /**
+     * @return non-empty-string
+     */
+    public static function serializeF64(float $x): string
+    {
+        return self::packFloat(PackFloatSpecifiers::Float64LittleEndian, $x);
+    }
+
+    /**
+     * @param non-empty-string $s
+     */
+    public static function deserializeS8(string $s): int
+    {
+        return self::unpackInt(UnpackIntSpecifiers::SignedInt8, $s);
+    }
+
+    /**
+     * @param non-empty-string $s
+     */
+    public static function deserializeS16(string $s): int
+    {
+        // PHP does not support unpacking signed integer in fixed endian, so we
+        // have to swap byte order if the machine's endianness is not little
+        // endian.
+        return self::unpackInt(UnpackIntSpecifiers::SignedInt16MachineOrder, self::byteSwapIfNeeded($s));
+    }
+
+    /**
+     * @param non-empty-string $s
+     */
+    public static function deserializeS32(string $s): int
+    {
+        // PHP does not support unpacking signed integer in fixed endian, so we
+        // have to swap byte order if the machine's endianness is not little
+        // endian.
+        return self::unpackInt(UnpackIntSpecifiers::SignedInt32MachineOrder, self::byteSwapIfNeeded($s));
+    }
+
+    /**
+     * @param non-empty-string $s
+     */
+    public static function deserializeS64(string $s): int
+    {
+        // PHP does not support unpacking signed integer in fixed endian, so we
+        // have to swap byte order if the machine's endianness is not little
+        // endian.
+        return self::unpackInt(UnpackIntSpecifiers::SignedInt64MachineOrder, self::byteSwapIfNeeded($s));
+    }
+
+    /**
+     * @param non-empty-string $s
+     */
+    public static function deserializeF32(string $s): float
+    {
+        // PHP's unpack() does not preserve NaN payload bits, so we have to
+        // manually check if the float is NaN and convert it to a 32-bit float.
+        $i = self::unpackInt(UnpackIntSpecifiers::UnsignedInt32LittleEndian, $s);
+        if (($i & FloatTraits::F32_EXPONENT_MASK) === FloatTraits::F32_EXPONENT_NAN) {
+            $sign = $i & FloatTraits::F32_SIGN_MASK;
+            $payload = $i & FloatTraits::F32_MANTISSA_MASK;
+            $j = 0 |
+                FloatTraits::getF64SignBit(Signedness::fromSignBit($sign)) |
+                FloatTraits::F64_EXPONENT_NAN |
+                ($payload << (FloatTraits::F64_MANTISSA_BITS - FloatTraits::F32_MANTISSA_BITS));
+            return self::unpackFloat(UnpackFloatSpecifiers::Float64LittleEndian, self::packInt(PackIntSpecifiers::Int64LittleEndian, $j));
+        } else {
+            return self::unpackFloat(UnpackFloatSpecifiers::Float32LittleEndian, $s);
+        }
+    }
+
+    /**
+     * @param non-empty-string $s
+     */
+    public static function deserializeF64(string $s): float
+    {
+        return self::unpackFloat(UnpackFloatSpecifiers::Float64LittleEndian, $s);
+    }
+
+    public static function reinterpretI32AsF32(int $x): float
+    {
+        return self::deserializeF32(self::serializeI32($x));
+    }
+
+    public static function reinterpretI64AsF32(int $x): float
+    {
+        return self::deserializeF32(self::serializeI64($x));
+    }
+
+    public static function reinterpretI32AsF64(int $x): float
+    {
+        return self::deserializeF64(self::serializeI32($x));
+    }
+
+    public static function reinterpretI64AsF64(int $x): float
+    {
+        return self::deserializeF64(self::serializeI64($x));
+    }
+
+    public static function reinterpretF32AsI32(float $x): int
+    {
+        return self::deserializeS32(self::serializeF32($x));
+    }
+
+    public static function reinterpretF64AsI32(float $x): int
+    {
+        return self::deserializeS32(self::serializeF64($x));
+    }
+
+    public static function reinterpretF32AsI64(float $x): int
+    {
+        return self::deserializeS64(self::serializeF32($x));
+    }
+
+    public static function reinterpretF64AsI64(float $x): int
+    {
+        return self::deserializeS64(self::serializeF64($x));
+    }
+
+    /**
+     * @return non-empty-string
+     */
+    private static function packInt(PackIntSpecifiers $spec, int $x): string
+    {
+        $result = pack($spec->value, $x);
+        assert($result !== '');
+        return $result;
+    }
+
+    /**
+     * @return non-empty-string
+     */
+    private static function packFloat(PackFloatSpecifiers $spec, float $x): string
+    {
+        $result = pack($spec->value, $x);
+        assert($result !== '');
+        return $result;
+    }
+
+    /**
+     * @param non-empty-string $s
+     */
+    private static function unpackInt(UnpackIntSpecifiers $spec, string $s): int
+    {
+        assert(strlen($s) === $spec->byteCount());
+        $result = unpack($spec->value, $s);
+        assert($result !== false && isset($result[1]) && is_int($result[1]));
+        return $result[1];
+    }
+
+    /**
+     * @param non-empty-string $s
+     */
+    private static function unpackFloat(UnpackFloatSpecifiers $spec, string $s): float
+    {
+        assert(strlen($s) === $spec->byteCount());
+        $result = unpack($spec->value, $s);
+        assert($result !== false && isset($result[1]) && is_float($result[1]));
+        return $result[1];
+    }
+
+    private static function isLittleEndian(): bool
+    {
+        return pack("s", ord("a"))[0] === "a";
+    }
+
+    /**
+     * @param non-empty-string $s
+     * @return non-empty-string
+     */
+    private static function byteSwapIfNeeded(string $s): string
+    {
+        // note: currently phpstan cannot infer that strrev(non-empty-string) returns non-empty-string.
+        $ret = self::isLittleEndian() ? $s : strrev($s);
+        assert($ret !== '');
+        return $ret;
+    }
+}
diff --git a/src/BitOps/FloatOps.php b/src/BitOps/FloatOps.php
new file mode 100644
index 0000000..cb6cbc4
--- /dev/null
+++ b/src/BitOps/FloatOps.php
@@ -0,0 +1,40 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Nsfisis\Waddiwasi\BitOps;
+
+final readonly class FloatOps
+{
+    private function __construct()
+    {
+    }
+
+    /**
+     * @return array{int, int, int}
+     */
+    public static function destructF64Bits(float $x): array
+    {
+        $i = BinaryConversion::deserializeS64(BinaryConversion::serializeF64($x));
+        return [
+            $i & FloatTraits::F64_SIGN_MASK,
+            $i & FloatTraits::F64_EXPONENT_MASK,
+            $i & FloatTraits::F64_MANTISSA_MASK,
+        ];
+    }
+
+    public static function constructNan(Signedness $sign, float $x): float
+    {
+        [, , $payload] = self::destructF64Bits($x);
+        $i = 0 |
+            FloatTraits::getF64SignBit($sign) |
+            FloatTraits::F64_EXPONENT_NAN |
+            $payload;
+        return BinaryConversion::deserializeF64(BinaryConversion::serializeI64($i));
+    }
+
+    public static function getSignedness(float $x): Signedness
+    {
+        return self::destructF64Bits($x)[0] === 0 ? Signedness::Unsigned : Signedness::Signed;
+    }
+}
diff --git a/src/BitOps/FloatTraits.php b/src/BitOps/FloatTraits.php
new file mode 100644
index 0000000..dca9e18
--- /dev/null
+++ b/src/BitOps/FloatTraits.php
@@ -0,0 +1,50 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Nsfisis\Waddiwasi\BitOps;
+
+final readonly class FloatTraits
+{
+    public const int F32_EXPONENT_BITS = 8;
+    public const int F32_MANTISSA_BITS = 23;
+
+    public const int F32_SIGN_MASK = 0b10000000_00000000_00000000_00000000;
+    public const int F32_EXPONENT_MASK = 0b01111111_10000000_00000000_00000000;
+    public const int F32_MANTISSA_MASK = 0b00000000_01111111_11111111_11111111;
+
+    public const int F32_SIGN_UNSIGNED = 0;
+    public const int F32_SIGN_SIGNED = 0b10000000_00000000_00000000_00000000;
+    public const int F32_EXPONENT_NAN = 0b01111111_10000000_00000000_00000000;
+
+    public const int F64_EXPONENT_BITS = 11;
+    public const int F64_MANTISSA_BITS = 52;
+
+    public const int F64_SIGN_MASK = PHP_INT_MIN;
+    public const int F64_EXPONENT_MASK = 0b01111111_11110000_00000000_00000000_00000000_00000000_00000000_00000000;
+    public const int F64_MANTISSA_MASK = 0b00000000_00001111_11111111_11111111_11111111_11111111_11111111_11111111;
+
+    public const int F64_SIGN_UNSIGNED = 0;
+    public const int F64_SIGN_SIGNED = PHP_INT_MIN;
+    public const int F64_EXPONENT_NAN = 0b01111111_11110000_00000000_00000000_00000000_00000000_00000000_00000000;
+
+    private function __construct()
+    {
+    }
+
+    public static function getF32SignBit(Signedness $sign): int
+    {
+        return match ($sign) {
+            Signedness::Unsigned => self::F32_SIGN_UNSIGNED,
+            Signedness::Signed => self::F32_SIGN_SIGNED,
+        };
+    }
+
+    public static function getF64SignBit(Signedness $sign): int
+    {
+        return match ($sign) {
+            Signedness::Unsigned => self::F64_SIGN_UNSIGNED,
+            Signedness::Signed => self::F64_SIGN_SIGNED,
+        };
+    }
+}
diff --git a/src/BitOps/PackFloatSpecifiers.php b/src/BitOps/PackFloatSpecifiers.php
new file mode 100644
index 0000000..645cae0
--- /dev/null
+++ b/src/BitOps/PackFloatSpecifiers.php
@@ -0,0 +1,11 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Nsfisis\Waddiwasi\BitOps;
+
+enum PackFloatSpecifiers: string
+{
+    case Float32LittleEndian = 'g';
+    case Float64LittleEndian = 'e';
+}
diff --git a/src/BitOps/PackIntSpecifiers.php b/src/BitOps/PackIntSpecifiers.php
new file mode 100644
index 0000000..da10ab2
--- /dev/null
+++ b/src/BitOps/PackIntSpecifiers.php
@@ -0,0 +1,14 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Nsfisis\Waddiwasi\BitOps;
+
+enum PackIntSpecifiers: string
+{
+    case Int8 = 'c';
+    case Int16LittleEndian = 'v';
+    case Int32LittleEndian = 'V';
+    case Int64BigEndian = 'J';
+    case Int64LittleEndian = 'P';
+}
diff --git a/src/BitOps/Signedness.php b/src/BitOps/Signedness.php
new file mode 100644
index 0000000..80cd03b
--- /dev/null
+++ b/src/BitOps/Signedness.php
@@ -0,0 +1,24 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Nsfisis\Waddiwasi\BitOps;
+
+enum Signedness
+{
+    case Unsigned;
+    case Signed;
+
+    public static function fromSignBit(int $b): self
+    {
+        return $b === 0 ? self::Unsigned : self::Signed;
+    }
+
+    public function negated(): self
+    {
+        return match ($this) {
+            self::Unsigned => self::Signed,
+            self::Signed => self::Unsigned,
+        };
+    }
+}
diff --git a/src/BitOps/UnpackFloatSpecifiers.php b/src/BitOps/UnpackFloatSpecifiers.php
new file mode 100644
index 0000000..5a3aa4f
--- /dev/null
+++ b/src/BitOps/UnpackFloatSpecifiers.php
@@ -0,0 +1,19 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Nsfisis\Waddiwasi\BitOps;
+
+enum UnpackFloatSpecifiers: string
+{
+    case Float32LittleEndian = 'g';
+    case Float64LittleEndian = 'e';
+
+    public function byteCount(): int
+    {
+        return match ($this) {
+            self::Float32LittleEndian => 4,
+            self::Float64LittleEndian => 8,
+        };
+    }
+}
diff --git a/src/BitOps/UnpackIntSpecifiers.php b/src/BitOps/UnpackIntSpecifiers.php
new file mode 100644
index 0000000..be210f3
--- /dev/null
+++ b/src/BitOps/UnpackIntSpecifiers.php
@@ -0,0 +1,24 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Nsfisis\Waddiwasi\BitOps;
+
+enum UnpackIntSpecifiers: string
+{
+    case SignedInt8 = 'c';
+    case SignedInt16MachineOrder = 's';
+    case SignedInt32MachineOrder = 'l';
+    case UnsignedInt32LittleEndian = 'V';
+    case SignedInt64MachineOrder = 'q';
+
+    public function byteCount(): int
+    {
+        return match ($this) {
+            self::SignedInt8 => 1,
+            self::SignedInt16MachineOrder => 2,
+            self::SignedInt32MachineOrder, self::UnsignedInt32LittleEndian => 4,
+            self::SignedInt64MachineOrder => 8,
+        };
+    }
+}