Ported CbValue from Unreal to Zen (#10)

CompactBinary: Ported CbValue changes from UE5

3 files changed, 355 insertions, 21 deletions

diff --git a/zencore/include/zencore/compactbinary.h b/zencore/include/zencore/compactbinary.h
index 4fce129ea..60e9ec7aa 100644
--- a/zencore/include/zencore/compactbinary.h
+++ b/zencore/include/zencore/compactbinary.h
@@ -30,6 +30,7 @@ class CbArrayView;
 class BinaryReader;
 class BinaryWriter;
 class CompressedBuffer;
+class CbValue;
 
 class DateTime
 {
@@ -318,6 +319,9 @@ public:
 	static constexpr inline bool IsUuid(CbFieldType Type) { return GetType(Type) == CbFieldType::Uuid; }
 	static constexpr inline bool IsObjectId(CbFieldType Type) { return GetType(Type) == CbFieldType::ObjectId; }
 
+	static constexpr inline bool IsCustomById(CbFieldType Type) { return GetType(Type) == CbFieldType::CustomById; }
+	static constexpr inline bool IsCustomByName(CbFieldType Type) { return GetType(Type) == CbFieldType::CustomByName; }
+
 	static constexpr inline bool IsDateTime(CbFieldType Type) { return GetType(Type) == CbFieldType::DateTime; }
 	static constexpr inline bool IsTimeSpan(CbFieldType Type) { return GetType(Type) == CbFieldType::TimeSpan; }
 
@@ -365,6 +369,46 @@ public:
 	virtual void VisitTimeSpan(TimeSpan Value)				= 0;
 };
 
+/** A custom compact binary field type with an integer identifier. */
+struct CbCustomById
+{
+	/** An identifier for the sub-type of the field. */
+	uint64_t Id = 0;
+	/** A view of the value. Lifetime is tied to the field that the value is associated with. */
+	MemoryView Data;
+};
+
+/** A custom compact binary field type with a string identifier. */
+struct CbCustomByName
+{
+	/** An identifier for the sub-type of the field. Lifetime is tied to the field that the name is associated with. */
+	std::u8string_view Name;
+	/** A view of the value. Lifetime is tied to the field that the value is associated with. */
+	MemoryView Data;
+};
+
+namespace CompactBinaryPrivate {
+	/** Parameters for converting to an integer. */
+	struct IntegerParams
+	{
+		/** Whether the output type has a sign bit. */
+		uint32_t IsSigned : 1;
+		/** Bits of magnitude. (7 for int8) */
+		uint32_t MagnitudeBits : 31;
+	};
+
+	/** Make integer params for the given integer type. */
+	template<typename IntType>
+	static constexpr inline IntegerParams MakeIntegerParams()
+	{
+		IntegerParams Params;
+		Params.IsSigned		 = IntType(-1) < IntType(0);
+		Params.MagnitudeBits = 8 * sizeof(IntType) - Params.IsSigned;
+		return Params;
+	}
+
+}  // namespace CompactBinaryPrivate
+
 /**
  * An atom of data in the compact binary format.
  *
@@ -393,13 +437,20 @@ public:
 
 	ZENCORE_API CbFieldView(const void* DataPointer, CbFieldType FieldType = CbFieldType::HasFieldType);
 
+	/** Construct a field from a value, without access to the name. */
+	inline explicit CbFieldView(const CbValue& Value);
+
 	/** Returns the name of the field if it has a name, otherwise an empty view. */
 	constexpr inline std::string_view GetName() const { return std::string_view(static_cast<const char*>(Payload) - NameLen, NameLen); }
 
+	/** Returns the value for unchecked access. Prefer the typed accessors below. */
+	inline CbValue GetValue() const;
+
 	ZENCORE_API MemoryView	 AsBinaryView(MemoryView Default = MemoryView());
 	ZENCORE_API CbObjectView AsObjectView();
 	ZENCORE_API CbArrayView	 AsArrayView();
 	ZENCORE_API std::string_view AsString(std::string_view Default = std::string_view());
+	ZENCORE_API std::u8string_view AsU8String(std::u8string_view Default = std::u8string_view());
 
 	ZENCORE_API void IterateAttachments(std::function<void(CbFieldView)> Visitor) const;
 
@@ -448,6 +499,11 @@ public:
 	/** Access the field as a OID. Returns the provided default on error. */
 	ZENCORE_API Oid AsObjectId(const Oid& Default);
 
+	/** Access the field as a custom sub-type with an integer identifier. Returns the provided default on error. */
+	ZENCORE_API CbCustomById AsCustomById(CbCustomById Default);
+	/** Access the field as a custom sub-type with a string identifier. Returns the provided default on error. */
+	ZENCORE_API CbCustomByName AsCustomByName(CbCustomByName Default);
+
 	/** Access the field as a date/time tick count. Returns the provided default on error. */
 	ZENCORE_API int64_t AsDateTimeTicks(int64_t Default = 0);
 
@@ -590,25 +646,6 @@ protected:
 	}
 
 private:
-	/** Parameters for converting to an integer. */
-	struct IntegerParams
-	{
-		/** Whether the output type has a sign bit. */
-		uint32_t IsSigned : 1;
-		/** Bits of magnitude. (7 for int8) */
-		uint32_t MagnitudeBits : 31;
-	};
-
-	/** Make integer params for the given integer type. */
-	template<typename IntType>
-	static constexpr inline IntegerParams MakeIntegerParams()
-	{
-		IntegerParams Params;
-		Params.IsSigned		 = IntType(-1) < IntType(0);
-		Params.MagnitudeBits = 8 * sizeof(IntType) - Params.IsSigned;
-		return Params;
-	}
-
 	/**
 	 * Access the field as the given integer type.
 	 *
@@ -617,11 +654,12 @@ private:
 	template<typename IntType>
 	inline IntType AsInteger(IntType Default)
 	{
-		return IntType(AsInteger(uint64_t(Default), MakeIntegerParams<IntType>()));
+		return IntType(AsInteger(uint64_t(Default), CompactBinaryPrivate::MakeIntegerParams<IntType>()));
 	}
 
-	ZENCORE_API uint64_t AsInteger(uint64_t Default, IntegerParams Params);
+	ZENCORE_API uint64_t AsInteger(uint64_t Default, CompactBinaryPrivate::IntegerParams Params);
 
+private:
 	/** The field type, with the transient HasFieldType flag if the field contains its type. */
 	CbFieldType Type = CbFieldType::None;
 	/** The error (if any) that occurred on the last field access. */
diff --git a/zencore/include/zencore/compactbinaryvalue.h b/zencore/include/zencore/compactbinaryvalue.h
new file mode 100644
index 000000000..5795ef957
--- /dev/null
+++ b/zencore/include/zencore/compactbinaryvalue.h
@@ -0,0 +1,290 @@
+// Copyright Epic Games, Inc. All Rights Reserved.
+
+#pragma once
+
+#include <zencore/compactbinary.h>
+#include <zencore/endian.h>
+#include <zencore/iobuffer.h>
+#include <zencore/iohash.h>
+#include <zencore/memory.h>
+
+namespace zen {
+
+namespace CompactBinaryPrivate {
+
+	template<typename T>
+	static constexpr inline T ReadUnaligned(const void* const Memory)
+	{
+#if PLATFORM_SUPPORTS_UNALIGNED_LOADS
+		return *static_cast<const T*>(Memory);
+#else
+		T Value;
+		memcpy(&Value, Memory, sizeof(Value));
+		return Value;
+#endif
+	}
+}  // namespace CompactBinaryPrivate
+/**
+ * A type that provides unchecked access to compact binary values.
+ *
+ * The main purpose of the type is to efficiently switch on field type. For every other use case,
+ * prefer to use the field, array, and object types directly. The accessors here do not check the
+ * type before reading the value, which means they can read out of bounds even on a valid compact
+ * binary value if the wrong accessor is used.
+ */
+class CbValue
+{
+public:
+	CbValue(CbFieldType Type, const void* Value);
+
+	CbObjectView AsObjectView() const;
+	CbArrayView	 AsArrayView() const;
+
+	MemoryView AsBinary() const;
+
+	/** Access as a string. Checks for range errors and uses the default if OutError is not null. */
+	std::string_view AsString(CbFieldError* OutError = nullptr, std::string_view Default = std::string_view()) const;
+
+	/** Access as a string as UTF8. Checks for range errors and uses the default if OutError is not null. */
+	std::u8string_view AsU8String(CbFieldError* OutError = nullptr, std::u8string_view Default = std::u8string_view()) const;
+
+	/**
+	 * Access as an integer, with both positive and negative values returned as unsigned.
+	 *
+	 * Checks for range errors and uses the default if OutError is not null.
+	 */
+	uint64_t AsInteger(CompactBinaryPrivate::IntegerParams Params, CbFieldError* OutError = nullptr, uint64_t Default = 0) const;
+
+	uint64_t AsIntegerPositive() const;
+	int64_t	 AsIntegerNegative() const;
+
+	float  AsFloat32() const;
+	double AsFloat64() const;
+
+	bool AsBool() const;
+
+	inline IoHash AsObjectAttachment() const { return AsHash(); }
+	inline IoHash AsBinaryAttachment() const { return AsHash(); }
+	inline IoHash AsAttachment() const { return AsHash(); }
+
+	IoHash AsHash() const;
+	Guid   AsUuid() const;
+
+	int64_t AsDateTimeTicks() const;
+	int64_t AsTimeSpanTicks() const;
+
+	Oid AsObjectId() const;
+
+	CbCustomById   AsCustomById() const;
+	CbCustomByName AsCustomByName() const;
+
+	inline CbFieldType GetType() const { return Type; }
+	inline const void* GetData() const { return Data; }
+
+private:
+	const void* Data;
+	CbFieldType Type;
+};
+
+inline CbFieldView::CbFieldView(const CbValue& InValue) : Type(InValue.GetType()), Payload(InValue.GetData())
+{
+}
+
+inline CbValue
+CbFieldView::GetValue() const
+{
+	return CbValue(CbFieldTypeOps::GetType(Type), Payload);
+}
+
+inline CbValue::CbValue(CbFieldType InType, const void* InValue) : Data(InValue), Type(InType)
+{
+}
+
+inline CbObjectView
+CbValue::AsObjectView() const
+{
+	return CbObjectView(*this);
+}
+
+inline CbArrayView
+CbValue::AsArrayView() const
+{
+	return CbArrayView(*this);
+}
+
+inline MemoryView
+CbValue::AsBinary() const
+{
+	const uint8_t* const Bytes = static_cast<const uint8_t*>(Data);
+	uint32_t			 ValueSizeByteCount;
+	const uint64_t		 ValueSize = ReadVarUInt(Bytes, ValueSizeByteCount);
+	return MakeMemoryView(Bytes + ValueSizeByteCount, ValueSize);
+}
+
+inline std::string_view
+CbValue::AsString(CbFieldError* OutError, std::string_view Default) const
+{
+	const char* const Chars = static_cast<const char*>(Data);
+	uint32_t		  ValueSizeByteCount;
+	const uint64_t	  ValueSize = ReadVarUInt(Chars, ValueSizeByteCount);
+
+	if (OutError)
+	{
+		if (ValueSize >= (uint64_t(1) << 31))
+		{
+			*OutError = CbFieldError::RangeError;
+			return Default;
+		}
+		*OutError = CbFieldError::None;
+	}
+
+	return std::string_view(Chars + ValueSizeByteCount, int32_t(ValueSize));
+}
+
+inline std::u8string_view
+CbValue::AsU8String(CbFieldError* OutError, std::u8string_view Default) const
+{
+	const char8_t* const Chars = static_cast<const char8_t*>(Data);
+	uint32_t			 ValueSizeByteCount;
+	const uint64_t		 ValueSize = ReadVarUInt(Chars, ValueSizeByteCount);
+
+	if (OutError)
+	{
+		if (ValueSize >= (uint64_t(1) << 31))
+		{
+			*OutError = CbFieldError::RangeError;
+			return Default;
+		}
+		*OutError = CbFieldError::None;
+	}
+
+	return std::u8string_view(Chars + ValueSizeByteCount, int32_t(ValueSize));
+}
+
+inline uint64_t
+CbValue::AsInteger(CompactBinaryPrivate::IntegerParams Params, CbFieldError* OutError, uint64_t Default) const
+{
+	// A shift of a 64-bit value by 64 is undefined so shift by one less because magnitude is never zero.
+	const uint64_t OutOfRangeMask = uint64_t(-2) << (Params.MagnitudeBits - 1);
+	const uint64_t IsNegative	  = uint8_t(Type) & 1;
+
+	uint32_t	   MagnitudeByteCount;
+	const uint64_t Magnitude = ReadVarUInt(Data, MagnitudeByteCount);
+	const uint64_t Value	 = Magnitude ^ -int64_t(IsNegative);
+
+	if (OutError)
+	{
+		const uint64_t IsInRange = (!(Magnitude & OutOfRangeMask)) & ((!IsNegative) | Params.IsSigned);
+		*OutError				 = IsInRange ? CbFieldError::None : CbFieldError::RangeError;
+
+		const uint64_t UseValueMask = -int64_t(IsInRange);
+		return (Value & UseValueMask) | (Default & ~UseValueMask);
+	}
+
+	return Value;
+}
+
+inline uint64_t
+CbValue::AsIntegerPositive() const
+{
+	uint32_t MagnitudeByteCount;
+	return ReadVarUInt(Data, MagnitudeByteCount);
+}
+
+inline int64_t
+CbValue::AsIntegerNegative() const
+{
+	uint32_t MagnitudeByteCount;
+	return int64_t(ReadVarUInt(Data, MagnitudeByteCount)) ^ -int64_t(1);
+}
+
+inline float
+CbValue::AsFloat32() const
+{
+	const uint32_t Value = FromNetworkOrder(CompactBinaryPrivate::ReadUnaligned<uint32_t>(Data));
+	return reinterpret_cast<const float&>(Value);
+}
+
+inline double
+CbValue::AsFloat64() const
+{
+	const uint64_t Value = FromNetworkOrder(CompactBinaryPrivate::ReadUnaligned<uint64_t>(Data));
+	return reinterpret_cast<const double&>(Value);
+}
+
+inline bool
+CbValue::AsBool() const
+{
+	return uint8_t(Type) & 1;
+}
+
+inline IoHash
+CbValue::AsHash() const
+{
+	return IoHash::MakeFrom(Data);
+}
+
+inline Guid
+CbValue::AsUuid() const
+{
+	Guid Value;
+	memcpy(&Value, Data, sizeof(Guid));
+	Value.A = FromNetworkOrder(Value.A);
+	Value.B = FromNetworkOrder(Value.B);
+	Value.C = FromNetworkOrder(Value.C);
+	Value.D = FromNetworkOrder(Value.D);
+	return Value;
+}
+
+inline int64_t
+CbValue::AsDateTimeTicks() const
+{
+	return FromNetworkOrder(CompactBinaryPrivate::ReadUnaligned<int64_t>(Data));
+}
+
+inline int64_t
+CbValue::AsTimeSpanTicks() const
+{
+	return FromNetworkOrder(CompactBinaryPrivate::ReadUnaligned<int64_t>(Data));
+}
+
+inline Oid
+CbValue::AsObjectId() const
+{
+	return Oid::FromMemory(Data);
+}
+
+inline CbCustomById
+CbValue::AsCustomById() const
+{
+	const uint8_t* Bytes = static_cast<const uint8_t*>(Data);
+	uint32_t	   DataSizeByteCount;
+	const uint64_t DataSize = ReadVarUInt(Bytes, DataSizeByteCount);
+	Bytes += DataSizeByteCount;
+
+	CbCustomById Value;
+	uint32_t	 TypeIdByteCount;
+	Value.Id   = ReadVarUInt(Bytes, TypeIdByteCount);
+	Value.Data = MakeMemoryView(Bytes + TypeIdByteCount, DataSize - TypeIdByteCount);
+	return Value;
+}
+
+inline CbCustomByName
+CbValue::AsCustomByName() const
+{
+	const uint8_t* Bytes = static_cast<const uint8_t*>(Data);
+	uint32_t	   DataSizeByteCount;
+	const uint64_t DataSize = ReadVarUInt(Bytes, DataSizeByteCount);
+	Bytes += DataSizeByteCount;
+
+	uint32_t	   TypeNameLenByteCount;
+	const uint64_t TypeNameLen = ReadVarUInt(Bytes, TypeNameLenByteCount);
+	Bytes += TypeNameLenByteCount;
+
+	CbCustomByName Value;
+	Value.Name = std::u8string_view(reinterpret_cast<const char8_t*>(Bytes), static_cast<std::u8string_view::size_type>(TypeNameLen));
+	Value.Data = MakeMemoryView(Bytes + TypeNameLen, DataSize - TypeNameLen - TypeNameLenByteCount);
+	return Value;
+}
+
+}  // namespace zen
diff --git a/zencore/include/zencore/memory.h b/zencore/include/zencore/memory.h
index 3d4db1081..aba391c85 100644
--- a/zencore/include/zencore/memory.h
+++ b/zencore/include/zencore/memory.h
@@ -354,6 +354,12 @@ MakeMemoryView(const void* Data, const void* DataEnd)
 	return MemoryView(Data, DataEnd);
 }
 
+[[nodiscard]] inline MemoryView
+MakeMemoryView(const void* Data, uint64_t Size)
+{
+	return MemoryView(Data, reinterpret_cast<const uint8_t*>(Data) + Size);
+}
+
 /**
  * Make a non-owning mutable view of the memory of the initializer list.
  *