cryptlib.h

Go to the documentation of this file.

// cryptlib.h - originally written and placed in the public domain by Wei Dai

/// \file cryptlib.h
/// \brief Abstract base classes that provide a uniform interface to this library.

/*! \mainpage Crypto++ Library 8.8 API Reference
<dl>
<dt>Abstract Base Classes<dd>
    cryptlib.h
<dt>Authenticated Encryption Modes<dd>
    CCM, EAX, \ref GCM "GCM (2K tables)", \ref GCM "GCM (64K tables)"
<dt>Block Ciphers<dd>
    \ref Rijndael "AES", ARIA, Weak::ARC4, Blowfish, BTEA, \ref CHAM128 "CHAM (64/128)", Camellia,
    \ref CAST128 "CAST (128/256)", DES, \ref DES_EDE2 "2-key Triple-DES", \ref DES_EDE3 "3-key Triple-DES",
    \ref DES_XEX3 "DESX", GOST, HIGHT, IDEA, LEA, \ref LR "Luby-Rackoff", \ref Kalyna128 "Kalyna (128/256/512)",
    MARS, RC2, RC5, RC6, \ref SAFER_K "SAFER-K", \ref SAFER_SK "SAFER-SK", SEED, Serpent,
    \ref SHACAL2 "SHACAL-2", SHARK, \ref SIMECK64 "SIMECK (32/64)" SKIPJACK, SM4, Square, TEA,
    \ref ThreeWay "3-Way", \ref Threefish256 "Threefish (256/512/1024)", Twofish, XTEA
<dt>Stream Ciphers<dd>
    \ref ChaCha "ChaCha (8/12/20)", \ref HC128 "HC-128/256", \ref Panama "Panama-LE", \ref Panama "Panama-BE",
    Rabbit, Salsa20, \ref SEAL "SEAL-LE", \ref SEAL "SEAL-BE", WAKE, XSalsa20
<dt>Hash Functions<dd>
    BLAKE2s, BLAKE2b, \ref Keccak "Keccak (F1600)", SHA1, SHA224, SHA256, SHA384, SHA512,
    \ref SHA3 "SHA-3", SM3, LSH (256/512), Tiger, RIPEMD160, RIPEMD256, SipHash, Whirlpool,
    Weak::MD2, Weak::MD4, Weak::MD5
<dt>Non-Cryptographic Checksums<dd>
    CRC32, CRC32C, Adler32
<dt>Message Authentication Codes<dd>
    BLAKE2b, BLAKE2s, CBC_MAC, CMAC, DMAC, \ref GCM "GCM (GMAC)", HMAC, Poly1305, TTMAC, VMAC
<dt>Random Number Generators<dd>
    NullRNG, LC_RNG, RandomPool, BlockingRng, NonblockingRng, AutoSeededRandomPool, AutoSeededX917RNG,
    NIST Hash_DRBG and HMAC_DRBG, \ref MersenneTwister "MersenneTwister (MT19937 and MT19937-AR)",
    DARN, RDRAND, RDSEED
<dt>Key Derivation and Password-based Cryptography<dd>
    HKDF, \ref PKCS12_PBKDF "PBKDF (PKCS #12)", \ref PKCS5_PBKDF1 "PBKDF-1 (PKCS #5)",
    \ref PKCS5_PBKDF2_HMAC "PBKDF-2/HMAC (PKCS #5)"
<dt>Public Key Cryptosystems<dd>
    DLIES, ECIES, LUCES, RSAES, RabinES, LUC_IES
<dt>Public Key Signature Schemes<dd>
    DSA, DSA2, \ref ed25519 "Ed25519", GDSA, ECDSA, NR, ECNR, LUCSS, RSASS, RSASS_ISO,
    RabinSS, RWSS, ESIGN
<dt>Key Agreement<dd>
    DH, DH2, \ref x25519 "X25519", \ref MQV_Domain "MQV", \ref HMQV_Domain "HMQV",
    \ref FHMQV_Domain "FHMQV", ECDH, x25519, ECMQV, ECHMQV, ECFHMQV, XTR_DH
<dt>Algebraic Structures<dd>
    Integer, PolynomialMod2, PolynomialOver, RingOfPolynomialsOver,
    ModularArithmetic, MontgomeryRepresentation, GFP2_ONB, GF2NP, GF256, GF2_32, EC2N, ECP
<dt>Secret Sharing and Information Dispersal<dd>
    SecretSharing, SecretRecovery, InformationDispersal, InformationRecovery
<dt>Compression<dd>
    Deflator, Inflator, Gzip, Gunzip, ZlibCompressor, ZlibDecompressor
<dt>Input Source Classes<dd>
    StringSource, ArraySource, VectorSource, FileSource, RandomNumberSource
<dt>Output Sink Classes<dd>
    StringSinkTemplate, StringSink, VectorSink, ArraySink, FileSink, RandomNumberSink
<dt>Filter Wrappers<dd>
    StreamTransformationFilter, AuthenticatedEncryptionFilter, AuthenticatedDecryptionFilter, HashFilter,
    HashVerificationFilter, SignerFilter, SignatureVerificationFilter
<dt>Binary to Text Encoders and Decoders<dd>
    HexEncoder, HexDecoder, Base64Encoder, Base64Decoder, Base64URLEncoder, Base64URLDecoder, Base32Encoder,
    Base32Decoder
<dt>Wrappers for OS features<dd>
    Timer, ThreadUserTimer

</dl>

<!--

<dt>FIPS 140 validated cryptography<dd>
    fips140.h

In the DLL version of Crypto++, only the following implementation class are available.
<dl>
<dt>Block Ciphers<dd>
    AES, \ref DES_EDE2 "2-key Triple-DES", \ref DES_EDE3 "3-key Triple-DES", SKIPJACK
<dt>Cipher Modes (replace template parameter BC with one of the block ciphers above)<dd>
    \ref ECB_Mode "ECB_Mode<BC>", \ref CTR_Mode "CTR_Mode<BC>", \ref CBC_Mode "CBC_Mode<BC>",
    \ref CFB_FIPS_Mode "CFB_FIPS_Mode<BC>", \ref OFB_Mode "OFB_Mode<BC>", \ref GCM "GCM<AES>"
<dt>Hash Functions<dd>
    SHA1, SHA224, SHA256, SHA384, SHA512
<dt>Public Key Signature Schemes (replace template parameter H with one of the hash functions above)<dd>
    RSASS<PKCS1v15, H>, RSASS<PSS, H>, RSASS_ISO<H>, RWSS<P1363_EMSA2, H>, DSA, ECDSA<ECP, H>,
    ECDSA<EC2N, H>
<dt>Message Authentication Codes (replace template parameter H with one of the hash functions above)<dd>
    HMAC<H>, CBC_MAC<DES_EDE2>, CBC_MAC<DES_EDE3>, GCM<AES>
<dt>Random Number Generators<dd>
    DefaultAutoSeededRNG (AutoSeededX917RNG<AES>)
<dt>Key Agreement<dd>
    DH, DH2
<dt>Public Key Cryptosystems<dd>
    RSAES<OAEP<SHA1> >
</dl>

-->

<p>This reference manual is a work in progress. Some classes lack detailed descriptions.
<p>Click <a href="CryptoPPRef.zip">here</a> to download a zip archive containing this manual.
<p>Thanks to Ryan Phillips for providing the Doxygen configuration file
and getting us started on the manual.
*/

#ifndef CRYPTOPP_CRYPTLIB_H
#define CRYPTOPP_CRYPTLIB_H

#include "config.h"
#include "stdcpp.h"
#include "trap.h"

// C5264 new for VS2022/v17.4, MSC v17.3.4
// https://github.com/weidai11/cryptopp/issues/1185
#if CRYPTOPP_MSC_VERSION
# pragma warning(push)
# pragma warning(disable: 4127 4189 4505 4702)
# if (CRYPTOPP_MSC_VERSION >= 1933)
#  pragma warning(disable: 5264)
# endif
#endif

NAMESPACE_BEGIN(CryptoPP)

// forward declarations
class Integer;
class RandomNumberGenerator;
class BufferedTransformation;

/// \brief Specifies a direction for a cipher to operate
/// \sa BlockTransformation::IsForwardTransformation(), BlockTransformation::IsPermutation(), BlockTransformation::GetCipherDirection()
enum CipherDir {
    /// \brief the cipher is performing encryption
    ENCRYPTION,
    /// \brief the cipher is performing decryption
    DECRYPTION};

/// \brief Represents infinite time
CRYPTOPP_CONST_OR_CONSTEXPR unsigned long INFINITE_TIME = ULONG_MAX;

// VC60 workaround: using enums as template parameters causes problems
/// \brief Converts an enumeration to a type suitable for use as a template parameter
template <typename ENUM_TYPE, int VALUE>
struct EnumToType
{
    static ENUM_TYPE ToEnum() {return static_cast<ENUM_TYPE>(VALUE);}
};

/// \brief Provides the byte ordering
/// \details Big-endian and little-endian modes are supported. Bi-endian and PDP-endian modes
///  are not supported.
enum ByteOrder {
    /// \brief byte order is little-endian
    LITTLE_ENDIAN_ORDER = 0,
    /// \brief byte order is big-endian
    BIG_ENDIAN_ORDER = 1};

/// \brief Provides a constant for LittleEndian
typedef EnumToType<ByteOrder, LITTLE_ENDIAN_ORDER> LittleEndian;
/// \brief Provides a constant for BigEndian
typedef EnumToType<ByteOrder, BIG_ENDIAN_ORDER> BigEndian;

/// \brief Base class for all exceptions thrown by the library
/// \details All library exceptions directly or indirectly inherit from the Exception class.
///  The Exception class itself inherits from std::exception. The library does not use
///  std::runtime_error derived classes.
class CRYPTOPP_DLL Exception : public std::exception
{
public:
    /// \enum ErrorType
    /// \brief Error types or categories
    enum ErrorType {
        /// \brief A method was called which was not implemented
        NOT_IMPLEMENTED,
        /// \brief An invalid argument was detected
        INVALID_ARGUMENT,
        /// \brief BufferedTransformation received a Flush(true) signal but can't flush buffers
        CANNOT_FLUSH,
        /// \brief Data integerity check, such as CRC or MAC, failed
        DATA_INTEGRITY_CHECK_FAILED,
        /// \brief Input data was received that did not conform to expected format
        INVALID_DATA_FORMAT,
        /// \brief Error reading from input device or writing to output device
        IO_ERROR,
        /// \brief Some other error occurred not belonging to other categories
        OTHER_ERROR
    };

    virtual ~Exception() throw() {}

    /// \brief Construct a new Exception
    explicit Exception(ErrorType errorType, const std::string &s) : m_errorType(errorType), m_what(s) {}

    /// \brief Retrieves a C-string describing the exception
    const char *what() const throw() {return (m_what.c_str());}
    /// \brief Retrieves a string describing the exception
    const std::string &GetWhat() const {return m_what;}
    /// \brief Sets the error string for the exception
    void SetWhat(const std::string &s) {m_what = s;}
    /// \brief Retrieves the error type for the exception
    ErrorType GetErrorType() const {return m_errorType;}
    /// \brief Sets the error type for the exceptions
    void SetErrorType(ErrorType errorType) {m_errorType = errorType;}

private:
    ErrorType m_errorType;
    std::string m_what;
};

/// \brief An invalid argument was detected
class CRYPTOPP_DLL InvalidArgument : public Exception
{
public:
    /// \brief Construct an InvalidArgument
    /// \param s the message for the exception
    /// \details The member function <tt>what()</tt> returns <tt>s</tt>.
    explicit InvalidArgument(const std::string &s) : Exception(INVALID_ARGUMENT, s) {}
};

/// \brief Input data was received that did not conform to expected format
class CRYPTOPP_DLL InvalidDataFormat : public Exception
{
public:
    /// \brief Construct an InvalidDataFormat
    /// \param s the message for the exception
    /// \details The member function <tt>what()</tt> returns <tt>s</tt>.
    explicit InvalidDataFormat(const std::string &s) : Exception(INVALID_DATA_FORMAT, s) {}
};

/// \brief A decryption filter encountered invalid ciphertext
class CRYPTOPP_DLL InvalidCiphertext : public InvalidDataFormat
{
public:
    /// \brief Construct an InvalidCiphertext
    /// \param s the message for the exception
    /// \details The member function <tt>what()</tt> returns <tt>s</tt>.
    explicit InvalidCiphertext(const std::string &s) : InvalidDataFormat(s) {}
};

/// \brief A method was called which was not implemented
class CRYPTOPP_DLL NotImplemented : public Exception
{
public:
    /// \brief Construct an NotImplemented
    /// \param s the message for the exception
    /// \details The member function <tt>what()</tt> returns <tt>s</tt>.
    explicit NotImplemented(const std::string &s) : Exception(NOT_IMPLEMENTED, s) {}
};

/// \brief Flush(true) was called but it can't completely flush its buffers
class CRYPTOPP_DLL CannotFlush : public Exception
{
public:
    /// \brief Construct an CannotFlush
    /// \param s the message for the exception
    /// \details The member function <tt>what()</tt> returns <tt>s</tt>.
    explicit CannotFlush(const std::string &s) : Exception(CANNOT_FLUSH, s) {}
};

/// \brief The operating system reported an error
class CRYPTOPP_DLL OS_Error : public Exception
{
public:
    virtual ~OS_Error() throw() {}

    /// \brief Construct an OS_Error
    /// \param errorType the error type
    /// \param s the message for the exception
    /// \param operation the operation for the exception
    /// \param errorCode the error code
    /// \details The member function <tt>what()</tt> returns <tt>s</tt>.
    OS_Error(ErrorType errorType, const std::string &s, const std::string& operation, int errorCode)
        : Exception(errorType, s), m_operation(operation), m_errorCode(errorCode) {}

    /// \brief Retrieve the operating system API that reported the error
    const std::string & GetOperation() const {return m_operation;}
    /// \brief Retrieve the error code returned by the operating system
    int GetErrorCode() const {return m_errorCode;}

protected:
    std::string m_operation;
    int m_errorCode;
};

/// \brief Returns a decoding results
struct CRYPTOPP_DLL DecodingResult
{
    /// \brief Constructs a DecodingResult
    /// \details isValidCoding is initialized to false and messageLength is
    ///  initialized to 0.
    explicit DecodingResult() : isValidCoding(false), messageLength(0) {}
    /// \brief Constructs a DecodingResult
    /// \param len the message length
    /// \details isValidCoding is initialized to true.
    explicit DecodingResult(size_t len) : isValidCoding(true), messageLength(len) {}

    /// \brief Compare two DecodingResult
    /// \param rhs the other DecodingResult
    /// \return true if either isValidCoding or messageLength is \a not equal,
    ///  false otherwise
    bool operator==(const DecodingResult &rhs) const {return isValidCoding == rhs.isValidCoding && messageLength == rhs.messageLength;}
    /// \brief Compare two DecodingResult
    /// \param rhs the other DecodingResult
    /// \return true if either isValidCoding or messageLength is \a not equal,
    ///  false otherwise
    /// \details Returns <tt>!operator==(rhs)</tt>.
    bool operator!=(const DecodingResult &rhs) const {return !operator==(rhs);}

    /// \brief Flag to indicate the decoding is valid
    bool isValidCoding;
    /// \brief Recovered message length if isValidCoding is true, undefined otherwise
    size_t messageLength;
};

/// \brief Interface for retrieving values given their names
/// \details This class is used to safely pass a variable number of arbitrarily
///  typed arguments to functions and to read values from keys and crypto parameters.
/// \details To obtain an object that implements NameValuePairs for the purpose of
///  parameter passing, use the MakeParameters() function.
/// \details To get a value from NameValuePairs, you need to know the name and the
///  type of the value. Call GetValueNames() on a NameValuePairs object to obtain a
///  list of value names that it supports. then look at the Name namespace
///  documentation to see what the type of each value is, or alternatively, call
///  GetIntValue() with the value name, and if the type is not int, a
///  ValueTypeMismatch exception will be thrown and you can get the actual type from
///  the exception object.
/// \sa NullNameValuePairs, g_nullNameValuePairs,
///  <A HREF="http://www.cryptopp.com/wiki/NameValuePairs">NameValuePairs</A> on the
///  Crypto++ wiki
class NameValuePairs
{
public:
    virtual ~NameValuePairs() {}

    /// \brief Thrown when an unexpected type is encountered
    /// \details Exception thrown when trying to retrieve a value using a different
    ///  type than expected
    class CRYPTOPP_DLL ValueTypeMismatch : public InvalidArgument
    {
    public:
        /// \brief Construct a ValueTypeMismatch
        /// \param name the name of the value
        /// \param stored the \a actual type of the value stored
        /// \param retrieving the \a presumed type of the value retrieved
        ValueTypeMismatch(const std::string &name, const std::type_info &stored, const std::type_info &retrieving)
            : InvalidArgument("NameValuePairs: type mismatch for '" + name + "', stored '" + stored.name() + "', trying to retrieve '" + retrieving.name() + "'")
            , m_stored(stored), m_retrieving(retrieving) {}

        /// \brief Provides the stored type
        /// \return the C++ mangled name of the type
        const std::type_info & GetStoredTypeInfo() const {return m_stored;}

        /// \brief Provides the retrieveing type
        /// \return the C++ mangled name of the type
        const std::type_info & GetRetrievingTypeInfo() const {return m_retrieving;}

    private:
        const std::type_info &m_stored;
        const std::type_info &m_retrieving;
    };

    /// \brief Get a copy of this object or subobject
    /// \tparam T class or type
    /// \param object reference to a variable that receives the value
    template <class T>
    bool GetThisObject(T &object) const
    {
        return GetValue((std::string("ThisObject:")+typeid(T).name()).c_str(), object);
    }

    /// \brief Get a pointer to this object
    /// \tparam T class or type
    /// \param ptr reference to a pointer to a variable that receives the value
    template <class T>
    bool GetThisPointer(T *&ptr) const
    {
        return GetValue((std::string("ThisPointer:")+typeid(T).name()).c_str(), ptr);
    }

    /// \brief Get a named value
    /// \tparam T class or type
    /// \param name the name of the object or value to retrieve
    /// \param value reference to a variable that receives the value
    /// \return true if the value was retrieved, false otherwise
    /// \sa GetValue(), GetValueWithDefault(), GetIntValue(), GetIntValueWithDefault(),
    ///  GetRequiredParameter() and GetRequiredIntParameter()
    template <class T>
    bool GetValue(const char *name, T &value) const
    {
        return GetVoidValue(name, typeid(T), &value);
    }

    /// \brief Get a named value
    /// \tparam T class or type
    /// \param name the name of the object or value to retrieve
    /// \param defaultValue the default value of the class or type if it does not exist
    /// \return the object or value
    /// \sa GetValue(), GetValueWithDefault(), GetIntValue(), GetIntValueWithDefault(),
    ///  GetRequiredParameter() and GetRequiredIntParameter()
    template <class T>
    T GetValueWithDefault(const char *name, T defaultValue) const
    {
        T value;
        bool result = GetValue(name, value);
        // No assert... this recovers from failure
        if (result) {return value;}
        return defaultValue;
    }

    /// \brief Get a list of value names that can be retrieved
    /// \return a list of names available to retrieve
    /// \details the items in the list are delimited with a colon.
    CRYPTOPP_DLL std::string GetValueNames() const
        {std::string result; GetValue("ValueNames", result); return result;}

    /// \brief Get a named value with type int
    /// \param name the name of the value to retrieve
    /// \param value the value retrieved upon success
    /// \return true if an int value was retrieved, false otherwise
    /// \details GetIntValue() is used to ensure we don't accidentally try to get an
    ///  unsigned int or some other type when we mean int (which is the most common case)
    /// \sa GetValue(), GetValueWithDefault(), GetIntValue(), GetIntValueWithDefault(),
    ///  GetRequiredParameter() and GetRequiredIntParameter()
    CRYPTOPP_DLL bool GetIntValue(const char *name, int &value) const
        {return GetValue(name, value);}

    /// \brief Get a named value with type int, with default
    /// \param name the name of the value to retrieve
    /// \param defaultValue the default value if the name does not exist
    /// \return the value retrieved on success or the default value
    /// \sa GetValue(), GetValueWithDefault(), GetIntValue(), GetIntValueWithDefault(),
    ///  GetRequiredParameter() and GetRequiredIntParameter()
    CRYPTOPP_DLL int GetIntValueWithDefault(const char *name, int defaultValue) const
        {return GetValueWithDefault(name, defaultValue);}

    /// \brief Get a named value with type word64
    /// \param name the name of the value to retrieve
    /// \param value the value retrieved upon success
    /// \return true if an word64 value was retrieved, false otherwise
    /// \sa GetValue(), GetValueWithDefault(), GetWord64ValueWithDefault(), GetIntValue(),
    ///  GetIntValueWithDefault(), GetRequiredParameter() and GetRequiredIntParameter()
    CRYPTOPP_DLL bool GetWord64Value(const char *name, word64 &value) const
        {return GetValue(name, value);}

    /// \brief Get a named value with type word64, with default
    /// \param name the name of the value to retrieve
    /// \param defaultValue the default value if the name does not exist
    /// \return the value retrieved on success or the default value
    /// \sa GetValue(), GetValueWithDefault(), GetWord64Value(), GetIntValue(),
    ///  GetIntValueWithDefault(), GetRequiredParameter() and GetRequiredWord64Parameter()
    CRYPTOPP_DLL word64 GetWord64ValueWithDefault(const char *name, word64 defaultValue) const
        {return GetValueWithDefault(name, defaultValue);}

    /// \brief Ensures an expected name and type is present
    /// \param name the name of the value
    /// \param stored the type that was stored for the name
    /// \param retrieving the type that is being retrieved for the name
    /// \throw ValueTypeMismatch
    /// \details ThrowIfTypeMismatch() effectively performs a type safety check.
    ///  stored and retrieving are C++ mangled names for the type.
    /// \sa GetValue(), GetValueWithDefault(), GetIntValue(), GetIntValueWithDefault(),
    ///  GetRequiredParameter() and GetRequiredIntParameter()
    CRYPTOPP_DLL static void CRYPTOPP_API ThrowIfTypeMismatch(const char *name, const std::type_info &stored, const std::type_info &retrieving)
        {if (stored != retrieving) throw ValueTypeMismatch(name, stored, retrieving);}

    /// \brief Retrieves a required name/value pair
    /// \tparam T class or type
    /// \param className the name of the class
    /// \param name the name of the value
    /// \param value reference to a variable to receive the value
    /// \throw InvalidArgument
    /// \details GetRequiredParameter() throws InvalidArgument if the name
    ///  is not present or not of the expected type T.
    /// \sa GetValue(), GetValueWithDefault(), GetIntValue(), GetIntValueWithDefault(),
    ///  GetRequiredParameter() and GetRequiredIntParameter()
    template <class T>
    void GetRequiredParameter(const char *className, const char *name, T &value) const
    {
        if (!GetValue(name, value))
            throw InvalidArgument(std::string(className) + ": missing required parameter '" + name + "'");
    }

    /// \brief Retrieves a required name/value pair
    /// \param className the name of the class
    /// \param name the name of the value
    /// \param value reference to a variable to receive the value
    /// \throw InvalidArgument
    /// \details GetRequiredParameter() throws InvalidArgument if the name
    ///  is not present or not of the expected type T.
    /// \sa GetValue(), GetValueWithDefault(), GetIntValue(), GetIntValueWithDefault(),
    ///  GetRequiredParameter() and GetRequiredIntParameter()
    CRYPTOPP_DLL void GetRequiredIntParameter(const char *className, const char *name, int &value) const
    {
        if (!GetIntValue(name, value))
            throw InvalidArgument(std::string(className) + ": missing required parameter '" + name + "'");
    }

    /// \brief Get a named value
    /// \param name the name of the object or value to retrieve
    /// \param valueType reference to a variable that receives the value
    /// \param pValue void pointer to a variable that receives the value
    /// \return true if the value was retrieved, false otherwise
    /// \details GetVoidValue() retrieves the value of name if it exists.
    /// \note GetVoidValue() is an internal function and should be implemented
    ///  by derived classes. Users should use one of the other functions instead.
    /// \sa GetValue(), GetValueWithDefault(), GetIntValue(), GetIntValueWithDefault(),
    ///  GetRequiredParameter() and GetRequiredIntParameter()
    CRYPTOPP_DLL virtual bool GetVoidValue(const char *name, const std::type_info &valueType, void *pValue) const =0;
};

// Doxygen cannot handle initialization
#if CRYPTOPP_DOXYGEN_PROCESSING
/// \brief Default channel for BufferedTransformation
/// \details DEFAULT_CHANNEL is equal to an empty string
/// \details The definition for DEFAULT_CHANNEL is in <tt>cryptlib.cpp</tt>.
///  It can be subject to <A HREF="https://isocpp.org/wiki/faq/ctors">Static
///  Initialization Order Fiasco</A>. If you experience a crash in
///  DEFAULT_CHANNEL where the string object is NULL, then you probably have
///  a global object using DEFAULT_CHANNEL before it has been constructed.
const std::string DEFAULT_CHANNEL;

/// \brief Channel for additional authenticated data
/// \details AAD_CHANNEL is equal to "AAD"
/// \details The definition for AAD_CHANNEL is in <tt>cryptlib.cpp</tt>.
///  It can be subject to <A HREF="https://isocpp.org/wiki/faq/ctors">Static
///  Initialization Order Fiasco</A>. If you experience a crash in
///  AAD_CHANNEL where the string object is NULL, then you probably have a
///  global object using AAD_CHANNEL before it has been constructed.
const std::string AAD_CHANNEL;

/// \brief An empty set of name-value pairs
/// \details The definition for g_nullNameValuePairs is in <tt>cryptlib.cpp</tt>.
///  It can be subject to <A HREF="https://isocpp.org/wiki/faq/ctors">Static
///  Initialization Order Fiasco</A>. If you experience a crash in
///  g_nullNameValuePairs where the string object is NULL, then you probably
///  have a global object using g_nullNameValuePairs before it has been
///  constructed.
const NameValuePairs& g_nullNameValuePairs;

#else
extern CRYPTOPP_DLL const std::string DEFAULT_CHANNEL;
extern CRYPTOPP_DLL const std::string AAD_CHANNEL;
extern CRYPTOPP_DLL const NameValuePairs& g_nullNameValuePairs;
#endif

// Document additional name spaces which show up elsewhere in the sources.
#if CRYPTOPP_DOXYGEN_PROCESSING
/// \brief Namespace containing value name definitions.
/// \details Name is part of the CryptoPP namespace.
/// \details The semantics of value names, types are:
/// <pre>
///     ThisObject:ClassName (ClassName, copy of this object or a subobject)
///     ThisPointer:ClassName (const ClassName *, pointer to this object or a subobject)
/// </pre>
DOCUMENTED_NAMESPACE_BEGIN(Name)
// more names defined in argnames.h
DOCUMENTED_NAMESPACE_END

/// \brief Namespace containing weak and wounded algorithms.
/// \details Weak is part of the CryptoPP namespace. Schemes and algorithms are moved into Weak
///  when their security level is reduced to an unacceptable level by contemporary standards.
/// \details To use an algorithm in the Weak namespace, you must <tt>\c \#define
///  CRYPTOPP_ENABLE_NAMESPACE_WEAK 1</tt> before including a header for a weak or wounded
///  algorithm. For example:
///  <pre>  \c \#define CRYPTOPP_ENABLE_NAMESPACE_WEAK 1
///   \c \#include <md5.h>
///   ...
///   CryptoPP::Weak::MD5 md5;
///   </pre>
DOCUMENTED_NAMESPACE_BEGIN(Weak)
// weak and wounded algorithms
DOCUMENTED_NAMESPACE_END
#endif

/// \brief Namespace containing NaCl library functions
/// \details TweetNaCl is a compact and portable reimplementation of the NaCl library.
DOCUMENTED_NAMESPACE_BEGIN(NaCl)
// crypto_box, crypto_box_open, crypto_sign, and crypto_sign_open (and friends)
DOCUMENTED_NAMESPACE_END

/// \brief Namespace containing testing and benchmark classes.
/// \details Source files for classes in the Test namespaces include
///  <tt>test.cpp</tt>, <tt>validat#.cpp</tt> and <tt>bench#.cpp</tt>.
DOCUMENTED_NAMESPACE_BEGIN(Test)
// testing and benchmark classes
DOCUMENTED_NAMESPACE_END

// ********************************************************

/// \brief Interface for cloning objects
/// \note this is \a not implemented by most classes
/// \sa ClonableImpl, NotCopyable
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE Clonable
{
public:
    virtual ~Clonable() {}

    /// \brief Copies  this object
    /// \return a copy of this object
    /// \throw NotImplemented
    /// \note this is \a not implemented by most classes
    /// \sa NotCopyable
    virtual Clonable* Clone() const {throw NotImplemented("Clone() is not implemented yet.");}  // TODO: make this =0
};

/// \brief Interface for all crypto algorithms
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE Algorithm : public Clonable
{
public:
    virtual ~Algorithm() {}

    /// \brief Interface for all crypto algorithms
    /// \param checkSelfTestStatus determines whether the object can proceed if the self
    ///  tests have not been run or failed.
    /// \details When FIPS 140-2 compliance is enabled and checkSelfTestStatus == true,
    ///  this constructor throws SelfTestFailure if the self test hasn't been run or fails.
    /// \details FIPS 140-2 compliance is disabled by default. It is only used by certain
    ///  versions of the library when the library is built as a DLL on Windows. Also see
    ///  CRYPTOPP_ENABLE_COMPLIANCE_WITH_FIPS_140_2 in config.h.
    Algorithm(bool checkSelfTestStatus = true);

    /// \brief Provides the name of this algorithm
    /// \return the standard algorithm name
    /// \details The standard algorithm name can be a name like <tt>AES</tt> or <tt>AES/GCM</tt>.
    ///  Some algorithms do not have standard names yet. For example, there is no standard
    ///  algorithm name for Shoup's ECIES.
    /// \note AlgorithmName is not universally implemented yet.
    virtual std::string AlgorithmName() const {return "unknown";}

    /// \brief Retrieve the provider of this algorithm
    /// \return the algorithm provider
    /// \details The algorithm provider can be a name like "C++", "SSE", "NEON", "AESNI",
    ///  "ARMv8" and "Power8". C++ is standard C++ code. Other labels, like SSE,
    ///  usually indicate a specialized implementation using instructions from a higher
    ///  instruction set architecture (ISA). Future labels may include external hardware
    ///  like a hardware security module (HSM).
    /// \details Generally speaking Wei Dai's original IA-32 ASM code falls under "SSE2".
    ///  Labels like "SSSE3" and "SSE4.1" follow after Wei's code and use intrinsics
    ///  instead of ASM.
    /// \details Algorithms which combine different instructions or ISAs provide the
    ///  dominant one. For example on x86 <tt>AES/GCM</tt> returns "AESNI" rather than
    ///  "CLMUL" or "AES+SSE4.1" or "AES+CLMUL" or "AES+SSE4.1+CLMUL".
    /// \note Provider is not universally implemented yet.
    /// \since Crypto++ 8.0
    virtual std::string AlgorithmProvider() const {return "C++";}
};

/// \brief Interface for algorithms that take byte strings as keys
/// \sa FixedKeyLength(), VariableKeyLength(), SameKeyLengthAs(), SimpleKeyingInterfaceImpl()
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE SimpleKeyingInterface
{
public:
    virtual ~SimpleKeyingInterface() {}

    /// \brief Returns smallest valid key length
    /// \return the minimum key length, in bytes
    virtual size_t MinKeyLength() const =0;

    /// \brief Returns largest valid key length
    /// \return the maximum key length, in bytes
    virtual size_t MaxKeyLength() const =0;

    /// \brief Returns default key length
    /// \return the default key length, in bytes
    virtual size_t DefaultKeyLength() const =0;

    /// \brief Returns a valid key length for the algorithm
    /// \param keylength the size of the key, in bytes
    /// \return the valid key length, in bytes
    /// \details keylength is provided in bytes, not bits. If keylength is less than MIN_KEYLENGTH,
    ///  then the function returns MIN_KEYLENGTH. If keylength is greater than MAX_KEYLENGTH,
    ///  then the function returns MAX_KEYLENGTH. if If keylength is a multiple of KEYLENGTH_MULTIPLE,
    ///  then keylength is returned. Otherwise, the function returns a \a lower multiple of
    ///  KEYLENGTH_MULTIPLE.
    virtual size_t GetValidKeyLength(size_t keylength) const =0;

    /// \brief Returns whether keylength is a valid key length
    /// \param keylength the requested keylength
    /// \return true if keylength is valid, false otherwise
    /// \details Internally the function calls GetValidKeyLength()
    virtual bool IsValidKeyLength(size_t keylength) const
        {return keylength == GetValidKeyLength(keylength);}

    /// \brief Sets or reset the key of this object
    /// \param key the key to use when keying the object
    /// \param length the size of the key, in bytes
    /// \param params additional initialization parameters to configure this object
    virtual void SetKey(const byte *key, size_t length, const NameValuePairs &params = g_nullNameValuePairs);

    /// \brief Sets or reset the key of this object
    /// \param key the key to use when keying the object
    /// \param length the size of the key, in bytes
    /// \param rounds the number of rounds to apply the transformation function,
    ///  if applicable
    /// \details SetKeyWithRounds() calls SetKey() with a NameValuePairs
    ///  object that only specifies rounds. rounds is an integer parameter,
    ///  and <tt>-1</tt> means use the default number of rounds.
    void SetKeyWithRounds(const byte *key, size_t length, int rounds);

    /// \brief Sets or reset the key of this object
    /// \param key the key to use when keying the object
    /// \param length the size of the key, in bytes
    /// \param iv the initialization vector to use when keying the object
    /// \param ivLength the size of the iv, in bytes
    /// \details SetKeyWithIV() calls SetKey() with a NameValuePairs
    ///  that only specifies IV. The IV is a byte buffer with size ivLength.
    ///  ivLength is an integer parameter, and <tt>-1</tt> means use IVSize().
    void SetKeyWithIV(const byte *key, size_t length, const byte *iv, size_t ivLength);

    /// \brief Sets or reset the key of this object
    /// \param key the key to use when keying the object
    /// \param length the size of the key, in bytes
    /// \param iv the initialization vector to use when keying the object
    /// \details SetKeyWithIV() calls SetKey() with a NameValuePairs() object
    ///  that only specifies iv. iv is a byte buffer, and it must have
    ///  a size IVSize().
    void SetKeyWithIV(const byte *key, size_t length, const byte *iv)
        {SetKeyWithIV(key, length, iv, IVSize());}

    /// \brief Secure IVs requirements as enumerated values.
    /// \details Provides secure IV requirements as a monotonically increasing enumerated values.
    ///  Requirements can be compared using less than (&lt;) and greater than (&gt;). For example,
    ///  <tt>UNIQUE_IV &lt; RANDOM_IV</tt> and <tt>UNPREDICTABLE_RANDOM_IV &gt; RANDOM_IV</tt>.
    /// \details Objects that use SimpleKeyingInterface do not support an optional IV. That is,
    ///  an IV must be present or it must be absent. If you wish to support an optional IV then
    ///  provide two classes - one with an IV and one without an IV.
    /// \sa IsResynchronizable(), CanUseRandomIVs(), CanUsePredictableIVs(), CanUseStructuredIVs()
    enum IV_Requirement {
        /// \brief The IV must be unique
        UNIQUE_IV = 0,
        /// \brief The IV must be random and possibly predictable
        RANDOM_IV,
        /// \brief The IV must be random and unpredictable
        UNPREDICTABLE_RANDOM_IV,
        /// \brief The IV is set by the object
        INTERNALLY_GENERATED_IV,
        /// \brief The object does not use an IV
        NOT_RESYNCHRONIZABLE
    };

    /// \brief Minimal requirement for secure IVs
    /// \return the secure IV requirement of the algorithm
    virtual IV_Requirement IVRequirement() const =0;

    /// \brief Determines if the object can be resynchronized
    /// \return true if the object can be resynchronized (i.e. supports initialization vectors), false otherwise
    /// \note If this function returns true, and no IV is passed to SetKey() and <tt>CanUseStructuredIVs()==true</tt>,
    ///  an IV of all 0's will be assumed.
    bool IsResynchronizable() const {return IVRequirement() < NOT_RESYNCHRONIZABLE;}

    /// \brief Determines if the object can use random IVs
    /// \return true if the object can use random IVs (in addition to ones returned by GetNextIV), false otherwise
    bool CanUseRandomIVs() const {return IVRequirement() <= UNPREDICTABLE_RANDOM_IV;}

    /// \brief Determines if the object can use random but possibly predictable IVs
    /// \return true if the object can use random but possibly predictable IVs (in addition to ones returned by
    ///  GetNextIV), false otherwise
    bool CanUsePredictableIVs() const {return IVRequirement() <= RANDOM_IV;}

    /// \brief Determines if the object can use structured IVs
    /// \return true if the object can use structured IVs, false otherwise
    /// \details CanUseStructuredIVs() indicates whether the object can use structured IVs; for example a counter
    ///  (in addition to ones returned by GetNextIV).
    bool CanUseStructuredIVs() const {return IVRequirement() <= UNIQUE_IV;}

    /// \brief Returns length of the IV accepted by this object
    /// \return the size of an IV, in bytes
    /// \throw NotImplemented() if the object does not support resynchronization
    /// \details The default implementation throws NotImplemented
    virtual unsigned int IVSize() const
        {throw NotImplemented(GetAlgorithm().AlgorithmName() + ": this object doesn't support resynchronization");}

    /// \brief Provides the default size of an IV
    /// \return default length of IVs accepted by this object, in bytes
    unsigned int DefaultIVLength() const {return IVSize();}

    /// \brief Provides the minimum size of an IV
    /// \return minimal length of IVs accepted by this object, in bytes
    /// \throw NotImplemented() if the object does not support resynchronization
    virtual unsigned int MinIVLength() const {return IVSize();}

    /// \brief Provides the maximum size of an IV
    /// \return maximal length of IVs accepted by this object, in bytes
    /// \throw NotImplemented() if the object does not support resynchronization
    virtual unsigned int MaxIVLength() const {return IVSize();}

    /// \brief Resynchronize with an IV
    /// \param iv the initialization vector
    /// \param ivLength the size of the initialization vector, in bytes
    /// \details Resynchronize() resynchronizes with an IV provided by the caller. <tt>ivLength=-1</tt> means use IVSize().
    /// \throw NotImplemented() if the object does not support resynchronization
    virtual void Resynchronize(const byte *iv, int ivLength=-1) {
        CRYPTOPP_UNUSED(iv); CRYPTOPP_UNUSED(ivLength);
        throw NotImplemented(GetAlgorithm().AlgorithmName() + ": this object doesn't support resynchronization");
    }

    /// \brief Retrieves a secure IV for the next message
    /// \param rng a RandomNumberGenerator to produce keying material
    /// \param iv a block of bytes to receive the IV
    /// \details The IV must be at least IVSize() in length.
    /// \details This method should be called after you finish encrypting one message and are ready
    ///  to start the next one. After calling it, you must call SetKey() or Resynchronize().
    ///  before using this object again.
    /// \details Internally, the base class implementation calls RandomNumberGenerator's GenerateBlock()
    /// \note This method is not implemented on decryption objects.
    virtual void GetNextIV(RandomNumberGenerator &rng, byte *iv);

protected:
    /// \brief Returns the base class Algorithm
    /// \return the base class Algorithm
    virtual const Algorithm & GetAlgorithm() const =0;

    /// \brief Sets the key for this object without performing parameter validation
    /// \param key a byte buffer used to key the cipher
    /// \param length the length of the byte buffer
    /// \param params additional parameters passed as NameValuePairs
    /// \details key must be at least DEFAULT_KEYLENGTH in length.
    virtual void UncheckedSetKey(const byte *key, unsigned int length, const NameValuePairs &params) =0;

    /// \brief Validates the key length
    /// \param length the size of the keying material, in bytes
    /// \throw InvalidKeyLength if the key length is invalid
    void ThrowIfInvalidKeyLength(size_t length);

    /// \brief Validates the object
    /// \throw InvalidArgument if the IV is present
    /// \details Internally, the default implementation calls IsResynchronizable() and throws
    ///  InvalidArgument if the function returns  true.
    /// \note called when no IV is passed
    void ThrowIfResynchronizable();

    /// \brief Validates the IV
    /// \param iv the IV with a length of IVSize, in bytes
    /// \throw InvalidArgument on failure
    /// \details Internally, the default implementation checks the iv. If iv is not NULL or nullptr,
    ///  then the function succeeds. If iv is NULL, then IVRequirement is checked against
    ///  UNPREDICTABLE_RANDOM_IV. If IVRequirement is UNPREDICTABLE_RANDOM_IV, then
    ///  then the function succeeds. Otherwise, an exception is thrown.
    void ThrowIfInvalidIV(const byte *iv);

    /// \brief Validates the IV length
    /// \param length the size of an IV, in bytes
    /// \throw InvalidArgument if the IV length is invalid
    size_t ThrowIfInvalidIVLength(int length);

    /// \brief Retrieves and validates the IV
    /// \param params NameValuePairs with the IV supplied as a ConstByteArrayParameter
    /// \param size the length of the IV, in bytes
    /// \return a pointer to the first byte of the IV
    /// \throw InvalidArgument if the number of rounds are invalid
    const byte * GetIVAndThrowIfInvalid(const NameValuePairs &params, size_t &size);

    /// \brief Validates the key length
    /// \param length the size of the keying material, in bytes
    inline void AssertValidKeyLength(size_t length) const
        {CRYPTOPP_UNUSED(length); CRYPTOPP_ASSERT(IsValidKeyLength(length));}
};

/// \brief Interface for the data processing part of block ciphers
/// \details Classes derived from BlockTransformation are block ciphers
///  in ECB mode (for example the DES::Encryption class), which are stateless.
///  These classes should not be used directly, but only in combination with
///  a mode class (see CipherModeDocumentation in modes.h).
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE BlockTransformation : public Algorithm
{
public:
    virtual ~BlockTransformation() {}

    /// \brief Encrypt or decrypt a block
    /// \param inBlock the input message before processing
    /// \param outBlock the output message after processing
    /// \param xorBlock an optional XOR mask
    /// \details ProcessAndXorBlock encrypts or decrypts inBlock, xor with xorBlock, and write to outBlock.
    /// \details The size of the block is determined by the block cipher and its documentation. Use
    ///  BLOCKSIZE at compile time, or BlockSize() at runtime.
    /// \note The message can be transformed in-place, or the buffers must \a not overlap
    /// \sa FixedBlockSize, BlockCipherFinal from seckey.h and BlockSize()
    virtual void ProcessAndXorBlock(const byte *inBlock, const byte *xorBlock, byte *outBlock) const =0;

    /// \brief Encrypt or decrypt a block
    /// \param inBlock the input message before processing
    /// \param outBlock the output message after processing
    /// \details ProcessBlock encrypts or decrypts inBlock and write to outBlock.
    /// \details The size of the block is determined by the block cipher and its documentation.
    ///  Use BLOCKSIZE at compile time, or BlockSize() at runtime.
    /// \sa FixedBlockSize, BlockCipherFinal from seckey.h and BlockSize()
    /// \note The message can be transformed in-place, or the buffers must \a not overlap
    void ProcessBlock(const byte *inBlock, byte *outBlock) const
        {ProcessAndXorBlock(inBlock, NULLPTR, outBlock);}

    /// \brief Encrypt or decrypt a block in place
    /// \param inoutBlock the input message before processing
    /// \details ProcessBlock encrypts or decrypts inoutBlock in-place.
    /// \details The size of the block is determined by the block cipher and its documentation.
    ///  Use BLOCKSIZE at compile time, or BlockSize() at runtime.
    /// \sa FixedBlockSize, BlockCipherFinal from seckey.h and BlockSize()
    void ProcessBlock(byte *inoutBlock) const
        {ProcessAndXorBlock(inoutBlock, NULLPTR, inoutBlock);}

    /// Provides the block size of the cipher
    /// \return the block size of the cipher, in bytes
    virtual unsigned int BlockSize() const =0;

    /// \brief Provides input and output data alignment for optimal performance.
    /// \return the input data alignment that provides optimal performance
    /// \sa GetAlignment() and OptimalBlockSize()
    virtual unsigned int OptimalDataAlignment() const;

    /// \brief Determines if the transformation is a permutation
    /// \return true if this is a permutation (i.e. there is an inverse transformation)
    virtual bool IsPermutation() const {return true;}

    /// \brief Determines if the cipher is being operated in its forward direction
    /// \return true if DIR is ENCRYPTION, false otherwise
    /// \sa IsForwardTransformation(), IsPermutation(), GetCipherDirection()
    virtual bool IsForwardTransformation() const =0;

    /// \brief Determines the number of blocks that can be processed in parallel
    /// \return the number of blocks that can be processed in parallel, for bit-slicing implementations
    /// \details Bit-slicing is often used to improve throughput and minimize timing attacks.
    virtual unsigned int OptimalNumberOfParallelBlocks() const {return 1;}

    /// \brief Bit flags that control AdvancedProcessBlocks() behavior
    enum FlagsForAdvancedProcessBlocks {
        /// \brief inBlock is a counter
        BT_InBlockIsCounter=1,
        /// \brief should not modify block pointers
        BT_DontIncrementInOutPointers=2,
        /// \brief Xor inputs before transformation
        BT_XorInput=4,
        /// \brief perform the transformation in reverse
        BT_ReverseDirection=8,
        /// \brief Allow parallel transformations
        BT_AllowParallel=16};

    /// \brief Encrypt and xor multiple blocks using additional flags
    /// \param inBlocks the input message before processing
    /// \param xorBlocks an optional XOR mask
    /// \param outBlocks the output message after processing
    /// \param length the size of the blocks, in bytes
    /// \param flags additional flags to control processing
    /// \details Encrypt and xor multiple blocks according to FlagsForAdvancedProcessBlocks flags.
    /// \note If BT_InBlockIsCounter is set, then the last byte of inBlocks may be modified.
    virtual size_t AdvancedProcessBlocks(const byte *inBlocks, const byte *xorBlocks, byte *outBlocks, size_t length, word32 flags) const;

    /// \brief Provides the direction of the cipher
    /// \return ENCRYPTION if IsForwardTransformation() is true, DECRYPTION otherwise
    /// \sa IsForwardTransformation(), IsPermutation()
    inline CipherDir GetCipherDirection() const {return IsForwardTransformation() ? ENCRYPTION : DECRYPTION;}
};

/// \brief Interface for the data processing portion of stream ciphers
/// \sa StreamTransformationFilter()
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE StreamTransformation : public Algorithm
{
public:
    virtual ~StreamTransformation() {}

    /// \brief Provides a reference to this object
    /// \return A reference to this object
    /// \details Useful for passing a temporary object to a function that takes a non-const reference
    StreamTransformation& Ref() {return *this;}

    /// \brief Provides the mandatory block size of the cipher
    /// \return The block size of the cipher if input must be processed in blocks, 1 otherwise
    /// \details Stream ciphers and some block ciphers modes of operation return 1. Modes that
    ///  return 1 must be able to process a single byte at a time, like counter mode. If a
    ///  mode of operation or block cipher cannot stream then it must not return 1.
    /// \details When filters operate the mode or cipher, ProcessData will be called with a
    ///  string of bytes that is determined by MandatoryBlockSize and OptimalBlockSize. When a
    ///  policy is set, like 16-byte strings for a 16-byte block cipher, the filter will buffer
    ///  bytes until the specified number of bytes is available to the object.
    /// \sa ProcessData, ProcessLastBlock, MandatoryBlockSize, MinLastBlockSize, BlockPaddingSchemeDef, IsLastBlockSpecial
    virtual unsigned int MandatoryBlockSize() const {return 1;}

    /// \brief Provides the input block size most efficient for this cipher
    /// \return The input block size that is most efficient for the cipher
    /// \details The base class implementation returns MandatoryBlockSize().
    /// \note Optimal input length is
    ///  <tt>n * OptimalBlockSize() - GetOptimalBlockSizeUsed()</tt> for any <tt>n > 0</tt>.
    virtual unsigned int OptimalBlockSize() const {return MandatoryBlockSize();}

    /// \brief Provides the number of bytes used in the current block when processing at optimal block size.
    /// \return the number of bytes used in the current block when processing at the optimal block size
    virtual unsigned int GetOptimalBlockSizeUsed() const {return 0;}

    /// \brief Provides input and output data alignment for optimal performance
    /// \return the input data alignment that provides optimal performance
    /// \sa GetAlignment() and OptimalBlockSize()
    virtual unsigned int OptimalDataAlignment() const;

    /// \brief Encrypt or decrypt an array of bytes
    /// \param outString the output byte buffer
    /// \param inString the input byte buffer
    /// \param length the size of the input and output byte buffers, in bytes
    /// \details ProcessData is called with a string of bytes whose size depends on MandatoryBlockSize.
    ///  Either <tt>inString == outString</tt>, or they must not overlap.
    /// \sa ProcessData, ProcessLastBlock, MandatoryBlockSize, MinLastBlockSize, BlockPaddingSchemeDef, IsLastBlockSpecial
    virtual void ProcessData(byte *outString, const byte *inString, size_t length) =0;

    /// \brief Encrypt or decrypt the last block of data
    /// \param outString the output byte buffer
    /// \param outLength the size of the output byte buffer, in bytes
    /// \param inString the input byte buffer
    /// \param inLength the size of the input byte buffer, in bytes
    /// \return the number of bytes used in outString
    /// \details ProcessLastBlock is used when the last block of data is special and requires handling
    ///  by the cipher. The current implementation provides an output buffer with a size
    ///  <tt>inLength+2*MandatoryBlockSize()</tt>. The return value allows the cipher to expand cipher
    ///  text during encryption or shrink plain text during decryption.
    /// \details This member function is used by CBC-CTS and OCB modes.
    /// \sa ProcessData, ProcessLastBlock, MandatoryBlockSize, MinLastBlockSize, BlockPaddingSchemeDef, IsLastBlockSpecial
    virtual size_t ProcessLastBlock(byte *outString, size_t outLength, const byte *inString, size_t inLength);

    /// \brief Provides the size of the last block
    /// \return the minimum size of the last block
    /// \details MinLastBlockSize() returns the minimum size of the last block. 0 indicates the last
    ///  block is not special.
    /// \details MandatoryBlockSize() enlists one of two behaviors. First, if MandatoryBlockSize()
    ///  returns 1, then the cipher can be streamed and ProcessData() is called with the tail bytes.
    ///  Second, if MandatoryBlockSize() returns non-0, then the string of bytes is padded to
    ///  MandatoryBlockSize() according to the padding mode. Then, ProcessData() is called with the
    ///  padded string of bytes.
    /// \details Some authenticated encryption modes are not expressed well with MandatoryBlockSize()
    ///  and MinLastBlockSize(). For example, AES/OCB uses 16-byte blocks (MandatoryBlockSize = 16)
    ///  and the last block requires special processing (MinLastBlockSize = 0). However, 0 is a valid
    ///  last block size for OCB and the special processing is custom padding, and not standard PKCS
    ///  padding. In response an unambiguous IsLastBlockSpecial() was added.
    /// \sa ProcessData, ProcessLastBlock, MandatoryBlockSize, MinLastBlockSize, BlockPaddingSchemeDef, IsLastBlockSpecial
    virtual unsigned int MinLastBlockSize() const {return 0;}

    /// \brief Determines if the last block receives special processing
    /// \return true if the last block reveives special processing, false otherwise.
    /// \details Some authenticated encryption modes are not expressed well with
    ///  MandatoryBlockSize() and MinLastBlockSize(). For example, AES/OCB uses
    ///  16-byte blocks (MandatoryBlockSize = 16) and the last block requires special processing
    ///  (MinLastBlockSize = 0). However, 0 is a valid last block size for OCB and the special
    ///  processing is custom padding, and not standard PKCS padding. In response an
    ///  unambiguous IsLastBlockSpecial() was added.
    /// \details When IsLastBlockSpecial() returns false nothing special happens. All the former
    ///  rules and behaviors apply. This is the default behavior of IsLastBlockSpecial().
    /// \details When IsLastBlockSpecial() returns true four things happen. First, MinLastBlockSize = 0
    ///  means 0 is a valid block size that should be processed. Second, standard block cipher padding is
    ///  \a not \a applied. Third, the caller supplies an outString is larger than inString by
    ///  <tt>2*MandatoryBlockSize()</tt>. That is, there's a reserve available when processing the last block.
    ///  Fourth, the cipher is responsible for finalization like custom padding. The cipher will tell
    ///  the library how many bytes were processed or used by returning the appropriate value from
    ///  ProcessLastBlock().
    /// \details The return value of ProcessLastBlock() indicates how many bytes were written to
    ///  <tt>outString</tt>. A filter pipelining data will send <tt>outString</tt> and up to <tt>outLength</tt>
    ///  to an <tt>AttachedTransformation()</tt> for additional processing. Below is an example of the code
    ///  used in <tt>StreamTransformationFilter::LastPut</tt>.
    /// <pre>  if (m_cipher.IsLastBlockSpecial())
    ///   {
    ///     size_t reserve = 2*m_cipher.MandatoryBlockSize();
    ///     space = HelpCreatePutSpace(*AttachedTransformation(), DEFAULT_CHANNEL, length+reserve);
    ///     length = m_cipher.ProcessLastBlock(space, length+reserve, inString, length);
    ///     AttachedTransformation()->Put(space, length);
    ///     return;
    ///   }</pre>
    /// \sa ProcessData, ProcessLastBlock, MandatoryBlockSize, MinLastBlockSize, BlockPaddingSchemeDef, IsLastBlockSpecial
    /// \since Crypto++ 6.0
    virtual bool IsLastBlockSpecial() const {return false;}

    /// \brief Encrypt or decrypt a string of bytes
    /// \param inoutString the string to process
    /// \param length the size of the inoutString, in bytes
    /// \details Internally, the base class implementation calls ProcessData().
    inline void ProcessString(byte *inoutString, size_t length)
        {ProcessData(inoutString, inoutString, length);}

    /// \brief Encrypt or decrypt a string of bytes
    /// \param outString the output string to process
    /// \param inString the input string to process
    /// \param length the size of the input and output strings, in bytes
    /// \details Internally, the base class implementation calls ProcessData().
    inline void ProcessString(byte *outString, const byte *inString, size_t length)
        {ProcessData(outString, inString, length);}

    /// \brief Encrypt or decrypt a byte
    /// \param input the input byte to process
    /// \details Internally, the base class implementation calls ProcessData() with a size of 1.
    inline byte ProcessByte(byte input)
        {ProcessData(&input, &input, 1); return input;}

    /// \brief Determines whether the cipher supports random access
    /// \return true if the cipher supports random access, false otherwise
    virtual bool IsRandomAccess() const =0;

    /// \brief Seek to an absolute position
    /// \param pos position to seek
    /// \throw NotImplemented
    /// \details The base class implementation throws NotImplemented. The function
    ///  \ref CRYPTOPP_ASSERT "asserts" IsRandomAccess() in debug builds.
    virtual void Seek(lword pos)
    {
        CRYPTOPP_UNUSED(pos);
        CRYPTOPP_ASSERT(!IsRandomAccess());
        throw NotImplemented("StreamTransformation: this object doesn't support random access");
    }

    /// \brief Determines whether the cipher is self-inverting
    /// \return true if the cipher is self-inverting, false otherwise
    /// \details IsSelfInverting determines whether this transformation is
    ///  self-inverting (e.g. xor with a keystream).
    virtual bool IsSelfInverting() const =0;

    /// \brief Determines if the cipher is being operated in its forward direction
    /// \return true if DIR is ENCRYPTION, false otherwise
    /// \sa IsForwardTransformation(), IsPermutation(), GetCipherDirection()
    virtual bool IsForwardTransformation() const =0;
};

/// \brief Interface for hash functions and data processing part of MACs
/// \details HashTransformation objects are stateful. They are created in an initial state,
///  change state as Update() is called, and return to the initial
///  state when Final() is called. This interface allows a large message to
///  be hashed in pieces by calling Update() on each piece followed by
///  calling Final().
/// \sa HashFilter(), HashVerificationFilter()
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE HashTransformation : public Algorithm
{
public:
    virtual ~HashTransformation() {}

    /// \brief Provides a reference to this object
    /// \return A reference to this object
    /// \details Useful for passing a temporary object to a function that takes a non-const reference
    HashTransformation& Ref() {return *this;}

    /// \brief Updates a hash with additional input
    /// \param input the additional input as a buffer
    /// \param length the size of the buffer, in bytes
    virtual void Update(const byte *input, size_t length) =0;

    /// \brief Request space which can be written into by the caller
    /// \param size the requested size of the buffer
    /// \details The purpose of this method is to help avoid extra memory allocations.
    /// \details size is an \a IN and \a OUT parameter and used as a hint. When the call is made,
    ///  size is the requested size of the buffer. When the call returns, size is the size of
    ///  the array returned to the caller.
    /// \details The base class implementation sets size to 0 and returns NULL or nullptr.
    /// \note Some objects, like ArraySink, cannot create a space because its fixed.
    virtual byte * CreateUpdateSpace(size_t &size) {size=0; return NULLPTR;}

    /// \brief Computes the hash of the current message
    /// \param digest a pointer to the buffer to receive the hash
    /// \details Final() restarts the hash for a new message.
    /// \pre <tt>COUNTOF(digest) <= DigestSize()</tt> or <tt>COUNTOF(digest) <= HASH::DIGESTSIZE</tt> ensures
    ///  the output byte buffer is large enough for the digest.
    virtual void Final(byte *digest)
        {TruncatedFinal(digest, DigestSize());}

    /// \brief Restart the hash
    /// \details Discards the current state, and restart for a new message
    virtual void Restart()
        {TruncatedFinal(NULLPTR, 0);}

    /// Provides the digest size of the hash
    /// \return the digest size of the hash.
    virtual unsigned int DigestSize() const =0;

    /// Provides the tag size of the hash
    /// \return the tag size of the hash.
    /// \details Same as DigestSize().
    unsigned int TagSize() const {return DigestSize();}

    /// \brief Provides the block size of the compression function
    /// \return block size of the compression function, in bytes
    /// \details BlockSize() will return 0 if the hash is not block based
    ///  or does not have an equivalent block size. For example, Keccak
    ///  and SHA-3 do not have a block size, but they do have an equivalent
    ///  block size called rate expressed as <tt>r</tt>.
    virtual unsigned int BlockSize() const {return 0;}

    /// \brief Provides the input block size most efficient for this hash.
    /// \return The input block size that is most efficient for the cipher
    /// \details The base class implementation returns MandatoryBlockSize().
    /// \details Optimal input length is
    ///  <tt>n * OptimalBlockSize() - GetOptimalBlockSizeUsed()</tt> for any <tt>n > 0</tt>.
    virtual unsigned int OptimalBlockSize() const {return 1;}

    /// \brief Provides input and output data alignment for optimal performance
    /// \return the input data alignment that provides optimal performance
    /// \sa GetAlignment() and OptimalBlockSize()
    virtual unsigned int OptimalDataAlignment() const;

    /// \brief Updates the hash with additional input and computes the hash of the current message
    /// \param digest a pointer to the buffer to receive the hash
    /// \param input the additional input as a buffer
    /// \param length the size of the buffer, in bytes
    /// \details Use this if your input is in one piece and you don't want to call Update()
    ///  and Final() separately
    /// \details CalculateDigest() restarts the hash for the next message.
    /// \pre <tt>COUNTOF(digest) == DigestSize()</tt> or <tt>COUNTOF(digest) == HASH::DIGESTSIZE</tt> ensures
    ///  the output byte buffer is a valid size.
    virtual void CalculateDigest(byte *digest, const byte *input, size_t length)
        {Update(input, length); Final(digest);}

    /// \brief Verifies the hash of the current message
    /// \param digest a pointer to the buffer of an \a existing hash
    /// \return \p true if the existing hash matches the computed hash, \p false otherwise
    /// \throw InvalidArgument() if the existing hash's size exceeds DigestSize()
    /// \details Verify() performs a bitwise compare on the buffers using VerifyBufsEqual(), which is
    ///  a constant time comparison function. digestLength cannot exceed DigestSize().
    /// \details Verify() restarts the hash for the next message.
    /// \pre <tt>COUNTOF(digest) == DigestSize()</tt> or <tt>COUNTOF(digest) == HASH::DIGESTSIZE</tt> ensures
    ///  the input byte buffer is a valid size.
    virtual bool Verify(const byte *digest)
        {return TruncatedVerify(digest, DigestSize());}

    /// \brief Updates the hash with additional input and verifies the hash of the current message
    /// \param digest a pointer to the buffer of an \a existing hash
    /// \param input the additional input as a buffer
    /// \param length the size of the buffer, in bytes
    /// \return \p true if the existing hash matches the computed hash, \p false otherwise
    /// \throw InvalidArgument() if the existing hash's size exceeds DigestSize()
    /// \details Use this if your input is in one piece and you don't want to call Update()
    ///  and Verify() separately
    /// \details VerifyDigest() performs a bitwise compare on the buffers using VerifyBufsEqual(),
    ///  which is a constant time comparison function.
    /// \details VerifyDigest() restarts the hash for the next message.
    /// \pre <tt>COUNTOF(digest) == DigestSize()</tt> or <tt>COUNTOF(digest) == HASH::DIGESTSIZE</tt> ensures
    ///  the output byte buffer is a valid size.
    virtual bool VerifyDigest(const byte *digest, const byte *input, size_t length)
        {Update(input, length); return Verify(digest);}

    /// \brief Computes the hash of the current message
    /// \param digest a pointer to the buffer to receive the hash
    /// \param digestSize the size of the truncated digest, in bytes
    /// \details TruncatedFinal() calls Final() and then copies digestSize bytes to digest.
    ///  The hash is restarted the hash for the next message.
    /// \pre <tt>COUNTOF(digest) <= DigestSize()</tt> or <tt>COUNTOF(digest) <= HASH::DIGESTSIZE</tt> ensures
    ///  the output byte buffer is a valid size.
    virtual void TruncatedFinal(byte *digest, size_t digestSize) =0;

    /// \brief Updates the hash with additional input and computes the hash of the current message
    /// \param digest a pointer to the buffer to receive the hash
    /// \param digestSize the length of the truncated hash, in bytes
    /// \param input the additional input as a buffer
    /// \param length the size of the buffer, in bytes
    /// \details Use this if your input is in one piece and you don't want to call Update()
    ///  and CalculateDigest() separately.
    /// \details CalculateTruncatedDigest() restarts the hash for the next message.
    /// \pre <tt>digestSize <= DigestSize()</tt> or <tt>digestSize <= HASH::DIGESTSIZE</tt> ensures
    ///  the output byte buffer is a valid size.
    virtual void CalculateTruncatedDigest(byte *digest, size_t digestSize, const byte *input, size_t length)
        {Update(input, length); TruncatedFinal(digest, digestSize);}

    /// \brief Verifies the hash of the current message
    /// \param digest a pointer to the buffer of an \a existing hash
    /// \param digestLength the size of the truncated hash, in bytes
    /// \return \p true if the existing hash matches the computed hash, \p false otherwise
    /// \throw InvalidArgument() if digestLength exceeds DigestSize()
    /// \details TruncatedVerify() is a truncated version of Verify(). It can operate on a
    ///  buffer smaller than DigestSize(). However, digestLength cannot exceed DigestSize().
    /// \details Verify() performs a bitwise compare on the buffers using VerifyBufsEqual(), which is
    ///  a constant time comparison function. digestLength cannot exceed DigestSize().
    /// \details TruncatedVerify() restarts the hash for the next message.
    /// \pre <tt>digestLength <= DigestSize()</tt> or <tt>digestLength <= HASH::DIGESTSIZE</tt> ensures
    ///  the input byte buffer is a valid size.
    virtual bool TruncatedVerify(const byte *digest, size_t digestLength);

    /// \brief Updates the hash with additional input and verifies the hash of the current message
    /// \param digest a pointer to the buffer of an \a existing hash
    /// \param digestLength the size of the truncated hash, in bytes
    /// \param input the additional input as a buffer
    /// \param length the size of the buffer, in bytes
    /// \return \p true if the existing hash matches the computed hash, \p false otherwise
    /// \throw InvalidArgument() if digestLength exceeds DigestSize()
    /// \details Use this if your input is in one piece and you don't want to call Update()
    ///  and TruncatedVerify() separately.
    /// \details VerifyTruncatedDigest() is a truncated version of VerifyDigest(). It can operate
    ///  on a buffer smaller than DigestSize(). However, digestLength cannot exceed DigestSize().
    /// \details VerifyTruncatedDigest() restarts the hash for the next message.
    /// \pre <tt>digestLength <= DigestSize()</tt> or <tt>digestLength <= HASH::DIGESTSIZE</tt> ensures
    ///  the input byte buffer is a valid size.
    virtual bool VerifyTruncatedDigest(const byte *digest, size_t digestLength, const byte *input, size_t length)
        {Update(input, length); return TruncatedVerify(digest, digestLength);}

protected:
    /// \brief Validates a truncated digest size
    /// \param size the requested digest size
    /// \throw InvalidArgument if the algorithm's digest size cannot be truncated to the requested size
    /// \details Throws an exception when the truncated digest size is greater than DigestSize()
    void ThrowIfInvalidTruncatedSize(size_t size) const;
};

/// \brief Interface for one direction (encryption or decryption) of a block cipher
/// \details These objects usually should not be used directly. See BlockTransformation for more details.
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE BlockCipher : public SimpleKeyingInterface, public BlockTransformation
{
protected:
    const Algorithm & GetAlgorithm() const {return *this;}
};

/// \brief Interface for one direction (encryption or decryption) of a stream cipher or cipher mode
/// \details These objects usually should not be used directly. See StreamTransformation for more details.
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE SymmetricCipher : public SimpleKeyingInterface, public StreamTransformation
{
protected:
    const Algorithm & GetAlgorithm() const {return *this;}
};

/// \brief Interface for message authentication codes
/// \details These objects usually should not be used directly. See HashTransformation for more details.
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE MessageAuthenticationCode : public SimpleKeyingInterface, public HashTransformation
{
protected:
    const Algorithm & GetAlgorithm() const {return *this;}
};

/// \brief Interface for authenticated encryption modes of operation
/// \details AuthenticatedSymmetricCipher() provides the interface for one direction
///  (encryption or decryption) of a stream cipher or block cipher mode with authentication. The
///  StreamTransformation() part of this interface is used to encrypt or decrypt the data. The
///  MessageAuthenticationCode() part of the interface is used to input additional authenticated
///  data (AAD), which is MAC'ed but not encrypted. The MessageAuthenticationCode() part is also
///  used to generate and verify the MAC.
/// \details Crypto++ provides four authenticated encryption modes of operation - CCM, EAX, GCM
///  and OCB mode. All modes implement AuthenticatedSymmetricCipher() and the motivation for
///  the API, like calling AAD a &quot;header&quot;, can be found in Bellare, Rogaway and
///  Wagner's <A HREF="http://web.cs.ucdavis.edu/~rogaway/papers/eax.pdf">The EAX Mode of
///  Operation</A>. The EAX paper suggested a basic API to help standardize AEAD schemes in
///  software and promote adoption of the modes.
/// \sa <A HREF="http://www.cryptopp.com/wiki/Authenticated_Encryption">Authenticated
///  Encryption</A> on the Crypto++ wiki.
/// \since Crypto++ 5.6.0
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE AuthenticatedSymmetricCipher : public MessageAuthenticationCode, public StreamTransformation
{
public:
    virtual ~AuthenticatedSymmetricCipher() {}

    /// \brief Exception thrown when the object is in the wrong state for the operation
    /// \details this indicates that a member function was called in the wrong state, for example trying to encrypt
    ///  a message before having set the key or IV
    class BadState : public Exception
    {
    public:
        explicit BadState(const std::string &name, const char *message) : Exception(OTHER_ERROR, name + ": " + message) {}
        explicit BadState(const std::string &name, const char *function, const char *state) : Exception(OTHER_ERROR, name + ": " + function + " was called before " + state) {}
    };

    /// \brief Provides the maximum length of AAD that can be input
    /// \return the maximum length of AAD that can be input before the encrypted data
    virtual lword MaxHeaderLength() const =0;

    /// \brief Provides the maximum length of encrypted data
    /// \return the maximum length of encrypted data
    virtual lword MaxMessageLength() const =0;

    /// \brief Provides the maximum length of AAD
    /// \return the maximum length of AAD that can be input after the encrypted data
    virtual lword MaxFooterLength() const {return 0;}

    /// \brief Determines if data lengths must be specified prior to inputting data
    /// \return true if the data lengths are required before inputting data, false otherwise
    /// \details if this function returns true, SpecifyDataLengths() must be called before attempting to input data.
    ///  This is the case for some schemes, such as CCM.
    /// \sa SpecifyDataLengths()
    virtual bool NeedsPrespecifiedDataLengths() const {return false;}

    /// \brief Prescribes the data lengths
    /// \param headerLength size of data before message is input, in bytes
    /// \param messageLength size of the message, in bytes
    /// \param footerLength size of data after message is input, in bytes
    /// \details SpecifyDataLengths() only needs to be called if NeedsPrespecifiedDataLengths() returns <tt>true</tt>.
    ///  If <tt>true</tt>, then <tt>headerLength</tt> will be validated against <tt>MaxHeaderLength()</tt>,
    ///  <tt>messageLength</tt> will be validated against <tt>MaxMessageLength()</tt>, and
    ///  <tt>footerLength</tt> will be validated against <tt>MaxFooterLength()</tt>.
    /// \sa NeedsPrespecifiedDataLengths()
    void SpecifyDataLengths(lword headerLength, lword messageLength, lword footerLength=0);

    /// \brief Encrypts and calculates a MAC in one call
    /// \param ciphertext the encryption buffer
    /// \param mac the mac buffer
    /// \param macSize the size of the MAC buffer, in bytes
    /// \param iv the iv buffer
    /// \param ivLength the size of the IV buffer, in bytes
    /// \param header the AAD buffer
    /// \param headerLength the size of the AAD buffer, in bytes
    /// \param message the message buffer
    /// \param messageLength the size of the messagetext buffer, in bytes
    /// \details EncryptAndAuthenticate() encrypts and generates the MAC in one call. The function
    ///  truncates the MAC if <tt>macSize < TagSize()</tt>.
    virtual void EncryptAndAuthenticate(byte *ciphertext, byte *mac, size_t macSize, const byte *iv, int ivLength, const byte *header, size_t headerLength, const byte *message, size_t messageLength);

    /// \brief Decrypts and verifies a MAC in one call
    /// \param message the decryption buffer
    /// \param mac the mac buffer
    /// \param macSize the size of the MAC buffer, in bytes
    /// \param iv the iv buffer
    /// \param ivLength the size of the IV buffer, in bytes
    /// \param header the AAD buffer
    /// \param headerLength the size of the AAD buffer, in bytes
    /// \param ciphertext the ciphertext buffer
    /// \param ciphertextLength the size of the ciphertext buffer, in bytes
    /// \return true if the MAC is valid and the decoding succeeded, false otherwise
    /// \details DecryptAndVerify() decrypts and verifies the MAC in one call.
    /// <tt>message</tt> is a decryption buffer and should be at least as large as the ciphertext buffer.
    /// \details The function returns true iff MAC is valid. DecryptAndVerify() assumes the MAC
    ///  is truncated if <tt>macLength < TagSize()</tt>.
    virtual bool DecryptAndVerify(byte *message, const byte *mac, size_t macSize, const byte *iv, int ivLength, const byte *header, size_t headerLength, const byte *ciphertext, size_t ciphertextLength);

    /// \brief Provides the name of this algorithm
    /// \return the standard algorithm name
    /// \details The standard algorithm name can be a name like \a AES or \a AES/GCM. Some algorithms
    ///  do not have standard names yet. For example, there is no standard algorithm name for
    ///  Shoup's ECIES.
    virtual std::string AlgorithmName() const;

    /// \brief Retrieve the provider of this algorithm
    /// \return the algorithm provider
    /// \details The algorithm provider can be a name like "C++", "SSE", "NEON", "AESNI",
    ///  "ARMv8" and "Power8". C++ is standard C++ code. Other labels, like SSE,
    ///  usually indicate a specialized implementation using instructions from a higher
    ///  instruction set architecture (ISA). Future labels may include external hardware
    ///  like a hardware security module (HSM).
    /// \details Generally speaking Wei Dai's original IA-32 ASM code falls under "SSE2".
    ///  Labels like "SSSE3" and "SSE4.1" follow after Wei's code and use intrinsics
    ///  instead of ASM.
    /// \details Algorithms which combine different instructions or ISAs provide the
    ///  dominant one. For example on x86 <tt>AES/GCM</tt> returns "AESNI" rather than
    ///  "CLMUL" or "AES+SSE4.1" or "AES+CLMUL" or "AES+SSE4.1+CLMUL".
    /// \note Provider is not universally implemented yet.
    /// \since Crypto++ 8.0
    virtual std::string AlgorithmProvider() const {return "C++";}

protected:
    const Algorithm & GetAlgorithm() const
        {return *static_cast<const MessageAuthenticationCode *>(this);}
    virtual void UncheckedSpecifyDataLengths(lword headerLength, lword messageLength, lword footerLength)
        {CRYPTOPP_UNUSED(headerLength); CRYPTOPP_UNUSED(messageLength); CRYPTOPP_UNUSED(footerLength);}
};

/// \brief Interface for random number generators
/// \details The library provides a number of random number generators, from software based
///  to hardware based generators.
/// \details All generated values are uniformly distributed over the range specified.
/// \since Crypto++ 3.1
/// \sa <A HREF="https://www.cryptopp.com/wiki/RandomNumberGenerator">RandomNumberGenerator</A>
///  on the Crypto++ wiki
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE RandomNumberGenerator : public Algorithm
{
public:
    virtual ~RandomNumberGenerator() {}

    /// \brief Update RNG state with additional unpredictable values
    /// \param input the entropy to add to the generator
    /// \param length the size of the input buffer
    /// \throw NotImplemented
    /// \details A generator may or may not accept additional entropy. Call CanIncorporateEntropy()
    ///  to test for the ability to use additional entropy.
    /// \details If a derived class does not override IncorporateEntropy(), then the base class
    ///  throws NotImplemented.
    virtual void IncorporateEntropy(const byte *input, size_t length)
    {
        CRYPTOPP_UNUSED(input); CRYPTOPP_UNUSED(length);
        throw NotImplemented("RandomNumberGenerator: IncorporateEntropy not implemented");
    }

    /// \brief Determines if a generator can accept additional entropy
    /// \return true if IncorporateEntropy() is implemented
    virtual bool CanIncorporateEntropy() const {return false;}

    /// \brief Generate new random byte and return it
    /// \return a random 8-bit byte
    /// \details Default implementation calls GenerateBlock() with one byte.
    /// \details All generated values are uniformly distributed over the range specified within the
    ///  the constraints of a particular generator.
    virtual byte GenerateByte();

    /// \brief Generate new random bit and return it
    /// \return a random bit
    /// \details The default implementation calls GenerateByte() and return its lowest bit.
    /// \details All generated values are uniformly distributed over the range specified within the
    ///  the constraints of a particular generator.
    virtual unsigned int GenerateBit();

    /// \brief Generate a random 32 bit word in the range min to max, inclusive
    /// \param min the lower bound of the range
    /// \param max the upper bound of the range
    /// \return a random 32-bit word
    /// \details The default implementation calls Crop() on the difference between max and
    ///  min, and then returns the result added to min.
    /// \details All generated values are uniformly distributed over the range specified within the
    ///  the constraints of a particular generator.
    virtual word32 GenerateWord32(word32 min=0, word32 max=0xffffffffUL);

    /// \brief Generate random array of bytes
    /// \param output the byte buffer
    /// \param size the length of the buffer, in bytes
    /// \details All generated values are uniformly distributed over the range specified within the
    ///  the constraints of a particular generator.
    /// \note A derived generator \a must override either GenerateBlock() or
    ///  GenerateIntoBufferedTransformation(). They can override both, or have one call the other.
    virtual void GenerateBlock(byte *output, size_t size);

    /// \brief Generate random bytes into a BufferedTransformation
    /// \param target the BufferedTransformation object which receives the bytes
    /// \param channel the channel on which the bytes should be pumped
    /// \param length the number of bytes to generate
    /// \details The default implementation calls GenerateBlock() and pumps the result into
    ///  the DEFAULT_CHANNEL of the target.
    /// \details All generated values are uniformly distributed over the range specified within the
    ///  the constraints of a particular generator.
    /// \note A derived generator \a must override either GenerateBlock() or
    ///  GenerateIntoBufferedTransformation(). They can override both, or have one call the other.
    virtual void GenerateIntoBufferedTransformation(BufferedTransformation &target, const std::string &channel, lword length);

    /// \brief Generate and discard n bytes
    /// \param n the number of bytes to generate and discard
    virtual void DiscardBytes(size_t n);

    /// \brief Randomly shuffle the specified array
    /// \param begin an iterator to the first element in the array
    /// \param end an iterator beyond the last element in the array
    /// \details The resulting permutation is uniformly distributed.
    template <class IT> void Shuffle(IT begin, IT end)
    {
        // TODO: What happens if there are more than 2^32 elements?
        for (; begin != end; ++begin)
            std::iter_swap(begin, begin + GenerateWord32(0, static_cast<word32>(end-begin-1)));
    }
};

/// \brief Interface for key derivation functions
/// \since Crypto++ 7.0
/// \sa <A HREF="https://www.cryptopp.com/wiki/KeyDerivationFunction">KeyDerivationFunction</A>
///  on the Crypto++ wiki
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE KeyDerivationFunction : public Algorithm
{
public:
    virtual ~KeyDerivationFunction() {}

    /// \brief Provides the name of this algorithm
    /// \return the standard algorithm name
    virtual std::string AlgorithmName() const =0;

    /// \brief Determine minimum number of bytes
    /// \return Minimum number of bytes which can be derived
    virtual size_t MinDerivedKeyLength() const;

    /// \brief Determine maximum number of bytes
    /// \return Maximum number of bytes which can be derived
    virtual size_t MaxDerivedKeyLength() const;

    /// \brief Returns a valid key length for the derivation function
    /// \param keylength the size of the derived key, in bytes
    /// \return the valid key length, in bytes
    virtual size_t GetValidDerivedLength(size_t keylength) const =0;

    /// \brief Returns whether keylength is a valid key length
    /// \param keylength the requested keylength
    /// \return true if the derived keylength is valid, false otherwise
    /// \details Internally the function calls GetValidKeyLength()
    virtual bool IsValidDerivedLength(size_t keylength) const {
        return keylength == GetValidDerivedLength(keylength);
    }

    /// \brief Derive a key from a seed
    /// \param derived the derived output buffer
    /// \param derivedLen the size of the derived buffer, in bytes
    /// \param secret the seed input buffer
    /// \param secretLen the size of the secret buffer, in bytes
    /// \param params additional initialization parameters to configure this object
    /// \return the number of iterations performed
    /// \throw InvalidDerivedKeyLength if <tt>derivedLen</tt> is invalid for the scheme
    /// \details DeriveKey() provides a standard interface to derive a key from
    ///  a secret seed and other parameters. Each class that derives from KeyDerivationFunction
    ///  provides an overload that accepts most parameters used by the derivation function.
    /// \details the number of iterations performed by DeriveKey() may be 1. For example, a
    ///  scheme like HKDF does not use the iteration count so it returns 1.
    virtual size_t DeriveKey(byte *derived, size_t derivedLen, const byte *secret, size_t secretLen, const NameValuePairs& params = g_nullNameValuePairs) const =0;

    /// \brief Set or change parameters
    /// \param params additional initialization parameters to configure this object
    /// \details SetParameters() is useful for setting common parameters when an object is
    ///  reused. Some derivation function classes may choose to implement it.
    virtual void SetParameters(const NameValuePairs& params);

protected:
    /// \brief Returns the base class Algorithm
    /// \return the base class Algorithm
    virtual const Algorithm & GetAlgorithm() const =0;

    /// \brief Validates the derived key length
    /// \param length the size of the derived key material, in bytes
    /// \throw InvalidKeyLength if the key length is invalid
    void ThrowIfInvalidDerivedKeyLength(size_t length) const;
};

/// \brief Interface for password based key derivation functions
/// \since Crypto++ 7.0
struct PasswordBasedKeyDerivationFunction : public KeyDerivationFunction
{
};

/// \brief Random Number Generator that does not produce random numbers
/// \return reference that can be passed to functions that require a RandomNumberGenerator
/// \details NullRNG() returns a reference that can be passed to functions that require a
///  RandomNumberGenerator but don't actually use it. The NullRNG() throws NotImplemented
///  when a generation function is called.
/// \sa ClassNullRNG, PK_SignatureScheme::IsProbabilistic()
CRYPTOPP_DLL RandomNumberGenerator & CRYPTOPP_API NullRNG();

class WaitObjectContainer;
class CallStack;

/// \brief Interface for objects that can be waited on.
class CRYPTOPP_NO_VTABLE Waitable
{
public:
    virtual ~Waitable() {}

    /// \brief Maximum number of wait objects that this object can return
    /// \return the maximum number of wait objects
    virtual unsigned int GetMaxWaitObjectCount() const =0;

    /// \brief Retrieves waitable objects
    /// \param container the wait container to receive the references to the objects.
    /// \param callStack CallStack() object used to select waitable objects
    /// \details GetWaitObjects() is usually called in one of two ways. First, it can
    ///  be called like <tt>something.GetWaitObjects(c, CallStack("my func after X", 0));</tt>.
    ///  Second, if in an outer GetWaitObjects() method that itself takes a callStack
    ///  parameter, it can be called like
    ///  <tt>innerThing.GetWaitObjects(c, CallStack("MyClass::GetWaitObjects at X", &callStack));</tt>.
    virtual void GetWaitObjects(WaitObjectContainer &container, CallStack const& callStack) =0;

    /// \brief Wait on this object
    /// \return true if the wait succeeded, false otherwise
    /// \details Wait() is the same as creating an empty container, calling GetWaitObjects(), and then calling
    ///  Wait() on the container.
    bool Wait(unsigned long milliseconds, CallStack const& callStack);
};

/// \brief Interface for buffered transformations
/// \details BufferedTransformation is a generalization of BlockTransformation,
///  StreamTransformation and HashTransformation.
/// \details A buffered transformation is an object that takes a stream of bytes as input (this may
///  be done in stages), does some computation on them, and then places the result into an internal
///  buffer for later retrieval. Any partial result already in the output buffer is not modified
///  by further input.
/// \details If a method takes a "blocking" parameter, and you pass false for it, then the method
///  will return before all input has been processed if the input cannot be processed without waiting
///  (for network buffers to become available, for example). In this case the method will return true
///  or a non-zero integer value. When this happens you must continue to call the method with the same
///  parameters until it returns false or zero, before calling any other method on it or attached
///  BufferedTransformation. The integer return value in this case is approximately
///  the number of bytes left to be processed, and can be used to implement a progress bar.
/// \details For functions that take a "propagation" parameter, <tt>propagation != 0</tt> means pass on
///  the signal to attached BufferedTransformation objects, with propagation decremented at each
///  step until it reaches <tt>0</tt>. <tt>-1</tt> means unlimited propagation.
/// \details \a All of the retrieval functions, like Get() and GetWord32(), return the actual
///  number of bytes retrieved, which is the lesser of the request number and MaxRetrievable().
/// \details \a Most of the input functions, like Put() and PutWord32(), return the number of
///  bytes remaining to be processed. A 0 value means all bytes were processed, and a non-0 value
///  means bytes remain to be processed.
/// \nosubgrouping
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE BufferedTransformation : public Algorithm, public Waitable
{
public:
    virtual ~BufferedTransformation() {}

    /// \brief Construct a BufferedTransformation
    BufferedTransformation() : Algorithm(false) {}

    /// \brief Provides a reference to this object
    /// \return A reference to this object
    /// \details Useful for passing a temporary object to a function that takes a non-const reference
    BufferedTransformation& Ref() {return *this;}

    /// \name INPUT
    //@{

        /// \brief Input a byte for processing
        /// \param inByte the 8-bit byte (octet) to be processed.
        /// \param blocking specifies whether the object should block when processing input.
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed).
        ///  0 indicates all bytes were processed.
        /// \details <tt>Put(byte)</tt> calls <tt>Put(byte*, size_t)</tt>.
        size_t Put(byte inByte, bool blocking=true)
            {return Put(&inByte, 1, blocking);}

        /// \brief Input a byte buffer for processing
        /// \param inString the byte buffer to process
        /// \param length the size of the string, in bytes
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed).
        ///  0 indicates all bytes were processed.
        /// \details Internally, Put() calls Put2().
        size_t Put(const byte *inString, size_t length, bool blocking=true)
            {return Put2(inString, length, 0, blocking);}

        /// Input a 16-bit word for processing.
        /// \param value the 16-bit value to be processed
        /// \param order the ByteOrder of the value to be processed.
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed).
        ///  0 indicates all bytes were processed.
        size_t PutWord16(word16 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);

        /// Input a 32-bit word for processing.
        /// \param value the 32-bit value to be processed.
        /// \param order the ByteOrder of the value to be processed.
        /// \param blocking specifies whether the object should block when processing input.
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed).
        ///  0 indicates all bytes were processed.
        size_t PutWord32(word32 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);

        /// Input a 64-bit word for processing.
        /// \param value the 64-bit value to be processed.
        /// \param order the ByteOrder of the value to be processed.
        /// \param blocking specifies whether the object should block when processing input.
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed).
        ///  0 indicates all bytes were processed.
        size_t PutWord64(word64 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);

        /// \brief Request space which can be written into by the caller
        /// \param size the requested size of the buffer
        /// \return byte pointer to the space to input data
        /// \details The purpose of this method is to help avoid extra memory allocations.
        /// \details size is an \a IN and \a OUT parameter and used as a hint. When the call is made,
        ///  size is the requested size of the buffer. When the call returns, size is the size of
        ///  the array returned to the caller.
        /// \details The base class implementation sets size to 0 and returns NULL.
        /// \note Some objects, like ArraySink, cannot create a space because its fixed. In the case of
        /// an ArraySink, the pointer to the array is returned and the size is remaining size.
        virtual byte * CreatePutSpace(size_t &size)
            {size=0; return NULLPTR;}

        /// \brief Determines whether input can be modified by the callee
        /// \return true if input can be modified, false otherwise
        /// \details The base class implementation returns false.
        virtual bool CanModifyInput() const
            {return false;}

        /// \brief Input multiple bytes that may be modified by callee.
        /// \param inString the byte buffer to process
        /// \param length the size of the string, in bytes
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed). 0 indicates all
        ///  bytes were processed.
        size_t PutModifiable(byte *inString, size_t length, bool blocking=true)
            {return PutModifiable2(inString, length, 0, blocking);}

        /// \brief Signals the end of messages to the object
        /// \param propagation the number of attached transformations the MessageEnd() signal should be passed
        /// \param blocking specifies whether the object should block when processing input
        /// \details propagation count includes this object. Setting propagation to <tt>1</tt> means this
        ///  object only. Setting propagation to <tt>-1</tt> means unlimited propagation.
        bool MessageEnd(int propagation=-1, bool blocking=true)
            {return !!Put2(NULLPTR, 0, propagation < 0 ? -1 : propagation+1, blocking);}

        /// \brief Input multiple bytes for processing and signal the end of a message
        /// \param inString the byte buffer to process
        /// \param length the size of the string, in bytes
        /// \param propagation the number of attached transformations the MessageEnd() signal should be passed
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed). 0 indicates all
        ///  bytes were processed.
        /// \details Internally, PutMessageEnd() calls Put2() with a modified propagation to
        ///  ensure all attached transformations finish processing the message.
        /// \details propagation count includes this object. Setting propagation to <tt>1</tt> means this
        ///  object only. Setting propagation to <tt>-1</tt> means unlimited propagation.
        size_t PutMessageEnd(const byte *inString, size_t length, int propagation=-1, bool blocking=true)
            {return Put2(inString, length, propagation < 0 ? -1 : propagation+1, blocking);}

        /// \brief Input multiple bytes for processing
        /// \param inString the byte buffer to process
        /// \param length the size of the string, in bytes
        /// \param messageEnd means how many filters to signal MessageEnd() to, including this one
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed). 0 indicates all
        ///  bytes were processed.
        /// \details Derived classes must implement Put2().
        virtual size_t Put2(const byte *inString, size_t length, int messageEnd, bool blocking) =0;

        /// \brief Input multiple bytes that may be modified by callee.
        /// \param inString the byte buffer to process.
        /// \param length the size of the string, in bytes.
        /// \param messageEnd means how many filters to signal MessageEnd() to, including this one.
        /// \param blocking specifies whether the object should block when processing input.
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed). 0 indicates all
        ///  bytes were processed.
        /// \details Internally, PutModifiable2() calls Put2().
        virtual size_t PutModifiable2(byte *inString, size_t length, int messageEnd, bool blocking)
            {return Put2(inString, length, messageEnd, blocking);}

        /// \brief Exception thrown by objects that have \a not implemented nonblocking input processing
        /// \details BlockingInputOnly inherits from NotImplemented
        struct BlockingInputOnly : public NotImplemented
            {BlockingInputOnly(const std::string &s) : NotImplemented(s + ": Nonblocking input is not implemented by this object.") {}};
    //@}

    /// \name WAITING
    //@{
        /// \brief Retrieves the maximum number of waitable objects
        unsigned int GetMaxWaitObjectCount() const;

        /// \brief Retrieves waitable objects
        /// \param container the wait container to receive the references to the objects
        /// \param callStack CallStack() object used to select waitable objects
        /// \details GetWaitObjects is usually called in one of two ways. First, it can
        ///  be called like <tt>something.GetWaitObjects(c, CallStack("my func after X", 0));</tt>.
        ///  Second, if in an outer GetWaitObjects() method that itself takes a callStack
        ///  parameter, it can be called like
        ///  <tt>innerThing.GetWaitObjects(c, CallStack("MyClass::GetWaitObjects at X", &callStack));</tt>.
        void GetWaitObjects(WaitObjectContainer &container, CallStack const& callStack);
    //@} // WAITING

    /// \name SIGNALS
    //@{

        /// \brief Initialize or reinitialize this object, without signal propagation
        /// \param parameters a set of NameValuePairs to initialize this object
        /// \throw NotImplemented
        /// \details IsolatedInitialize() is used to initialize or reinitialize an object using a variable
        ///  number of arbitrarily typed arguments. The function avoids the need for multiple constructors providing
        ///  all possible combintations of configurable parameters.
        /// \details IsolatedInitialize() does not call Initialize() on attached transformations. If initialization
        ///  should be propagated, then use the Initialize() function.
        /// \details If a derived class does not override IsolatedInitialize(), then the base class throws
        ///  NotImplemented.
        virtual void IsolatedInitialize(const NameValuePairs &parameters) {
            CRYPTOPP_UNUSED(parameters);
            throw NotImplemented("BufferedTransformation: this object can't be reinitialized");
        }

        /// \brief Flushes data buffered by this object, without signal propagation
        /// \param hardFlush indicates whether all data should be flushed
        /// \param blocking specifies whether the object should block when processing input
        /// \return true if the flush was successful, false otherwise
        /// \note hardFlush must be used with care
        virtual bool IsolatedFlush(bool hardFlush, bool blocking) =0;

        /// \brief Marks the end of a series of messages, without signal propagation
        /// \param blocking specifies whether the object should block when completing the processing on
        ///  the current series of messages
        /// \return true if the message was successful, false otherwise
        virtual bool IsolatedMessageSeriesEnd(bool blocking)
            {CRYPTOPP_UNUSED(blocking); return false;}

        /// \brief Initialize or reinitialize this object, with signal propagation
        /// \param parameters a set of NameValuePairs to initialize or reinitialize this object
        /// \param propagation the number of attached transformations the Initialize() signal should be passed
        /// \details Initialize() is used to initialize or reinitialize an object using a variable number of
        ///  arbitrarily typed arguments. The function avoids the need for multiple constructors providing
        ///  all possible combintations of configurable parameters.
        /// \details propagation count includes this object. Setting propagation to <tt>1</tt> means this
        ///  object only. Setting propagation to <tt>-1</tt> means unlimited propagation.
        virtual void Initialize(const NameValuePairs &parameters=g_nullNameValuePairs, int propagation=-1);

        /// \brief Flush buffered input and/or output, with signal propagation
        /// \param hardFlush is used to indicate whether all data should be flushed
        /// \param propagation the number of attached transformations the Flush()
        ///  signal should be passed
        /// \param blocking specifies whether the object should block when processing input
        /// \return true if the flush was successful, false otherwise
        /// \details propagation count includes this object. Setting propagation to
        ///  <tt>1</tt> means this object only. Setting propagation to <tt>-1</tt>
        ///  means unlimited propagation.
        /// \note Hard flushes must be used with care. It means try to process and
        ///  output everything, even if there may not be enough data to complete the
        ///  action. For example, hard flushing a HexDecoder would cause an error if
        ///  you do it after inputing an odd number of hex encoded characters.
        /// \note For some types of filters, like  ZlibDecompressor, hard flushes can
        ///  only be done at "synchronization points". These synchronization points
        ///  are positions in the data stream that are created by hard flushes on the
        ///  corresponding reverse filters, in this example ZlibCompressor. This is
        ///  useful when zlib compressed data is moved across a network in packets
        ///  and compression state is preserved across packets, as in the SSH2 protocol.
        virtual bool Flush(bool hardFlush, int propagation=-1, bool blocking=true);

        /// \brief Marks the end of a series of messages, with signal propagation
        /// \param propagation the number of attached transformations the MessageSeriesEnd() signal should be passed
        /// \param blocking specifies whether the object should block when processing input
        /// \return true if the message was successful, false otherwise
        /// \details Each object that receives the signal will perform its processing, decrement
        ///  propagation, and then pass the signal on to attached transformations if the value is not 0.
        /// \details propagation count includes this object. Setting propagation to <tt>1</tt> means this
        ///  object only. Setting propagation to <tt>-1</tt> means unlimited propagation.
        /// \note There should be a MessageEnd() immediately before MessageSeriesEnd().
        virtual bool MessageSeriesEnd(int propagation=-1, bool blocking=true);

        /// \brief Set propagation of automatically generated and transferred signals
        /// \param propagation then new value
        /// \details Setting propagation to <tt>0</tt> means do not automatically generate signals. Setting
        ///  propagation to <tt>-1</tt> means unlimited propagation.
        virtual void SetAutoSignalPropagation(int propagation)
            {CRYPTOPP_UNUSED(propagation);}

        /// \brief Retrieve automatic signal propagation value
        /// \return the number of attached transformations the signal is propagated to. 0 indicates
        ///  the signal is only witnessed by this object
        virtual int GetAutoSignalPropagation() const {return 0;}
public:

    /// \name RETRIEVAL OF ONE MESSAGE
    //@{

        /// \brief Provides the number of bytes ready for retrieval
        /// \return the number of bytes ready for retrieval
        /// \details The number of bytes available are dependent on the source. If an exact value is
        ///  available, then the exact value is returned. The exact value can include 0 if the source
        ///  is exhausted.
        /// \details Some stream-based sources do not allow seeking() on the underlying stream, such
        ///  as some FileSource(). If the stream does not allow seeking() then MaxRetrievable()
        ///  returns LWORD_MAX to indicate there are still bytes to be retrieved.
        virtual lword MaxRetrievable() const;

        /// \brief Determines whether bytes are ready for retrieval
        /// \return true if bytes are available for retrieval, false otherwise
        virtual bool AnyRetrievable() const;

        /// \brief Retrieve a 8-bit byte
        /// \param outByte the 8-bit value to be retrieved
        /// \return the number of bytes consumed during the call.
        /// \details Use the return value of Get to detect short reads.
        virtual size_t Get(byte &outByte);

        /// \brief Retrieve a block of bytes
        /// \param outString a block of bytes
        /// \param getMax the number of bytes to Get
        /// \return the number of bytes consumed during the call.
        /// \details Use the return value of Get to detect short reads.
        virtual size_t Get(byte *outString, size_t getMax);

        /// \brief Peek a 8-bit byte
        /// \param outByte the 8-bit value to be retrieved
        /// \return the number of bytes read during the call.
        /// \details Peek does not remove bytes from the object. Use the return value of
        ///  Get() to detect short reads.
        virtual size_t Peek(byte &outByte) const;

        /// \brief Peek a block of bytes
        /// \param outString a block of bytes
        /// \param peekMax the number of bytes to Peek
        /// \return the number of bytes read during the call.
        /// \details Peek does not remove bytes from the object. Use the return value of
        ///  Peek() to detect short reads.
        virtual size_t Peek(byte *outString, size_t peekMax) const;

        /// \brief Retrieve a 16-bit word
        /// \param value the 16-bit value to be retrieved
        /// \param order the ByteOrder of the value to be processed.
        /// \return the number of bytes consumed during the call.
        /// \details Use the return value of GetWord16() to detect short reads.
        size_t GetWord16(word16 &value, ByteOrder order=BIG_ENDIAN_ORDER);

        /// \brief Retrieve a 32-bit word
        /// \param value the 32-bit value to be retrieved
        /// \param order the ByteOrder of the value to be processed.
        /// \return the number of bytes consumed during the call.
        /// \details Use the return value of GetWord32() to detect short reads.
        size_t GetWord32(word32 &value, ByteOrder order=BIG_ENDIAN_ORDER);

        /// \brief Retrieve a 64-bit word
        /// \param value the 64-bit value to be retrieved
        /// \param order the ByteOrder of the value to be processed.
        /// \return the number of bytes consumed during the call.
        /// \details Use the return value of GetWord64() to detect short reads.
        /// \since Crypto++ 8.3
        size_t GetWord64(word64 &value, ByteOrder order=BIG_ENDIAN_ORDER);

        /// \brief Peek a 16-bit word
        /// \param value the 16-bit value to be retrieved
        /// \param order the ByteOrder of the value to be processed.
        /// \return the number of bytes consumed during the call.
        /// \details Peek does not consume bytes in the stream. Use the return value
        ///  of PeekWord16() to detect short reads.
        size_t PeekWord16(word16 &value, ByteOrder order=BIG_ENDIAN_ORDER) const;

        /// \brief Peek a 32-bit word
        /// \param value the 32-bit value to be retrieved
        /// \param order the ByteOrder of the value to be processed.
        /// \return the number of bytes consumed during the call.
        /// \details Peek does not consume bytes in the stream. Use the return value
        ///  of PeekWord32() to detect short reads.
        size_t PeekWord32(word32 &value, ByteOrder order=BIG_ENDIAN_ORDER) const;

        /// \brief Peek a 64-bit word
        /// \param value the 64-bit value to be retrieved
        /// \param order the ByteOrder of the value to be processed.
        /// \return the number of bytes consumed during the call.
        /// \details Peek does not consume bytes in the stream. Use the return value
        ///  of PeekWord64() to detect short reads.
        /// \since Crypto++ 8.3
        size_t PeekWord64(word64 &value, ByteOrder order=BIG_ENDIAN_ORDER) const;

        /// move transferMax bytes of the buffered output to target as input

        /// \brief Transfer bytes from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param transferMax the number of bytes to transfer
        /// \param channel the channel on which the transfer should occur
        /// \return the number of bytes transferred during the call.
        /// \details TransferTo removes bytes from this object and moves them to the destination.
        /// \details The function always returns transferMax. If an accurate count is needed, then use TransferTo2().
        lword TransferTo(BufferedTransformation &target, lword transferMax=LWORD_MAX, const std::string &channel=DEFAULT_CHANNEL)
            {TransferTo2(target, transferMax, channel); return transferMax;}

        /// \brief Discard skipMax bytes from the output buffer
        /// \param skipMax the number of bytes to discard
        /// \details Skip() discards bytes from the output buffer, which is the AttachedTransformation(), if present.
        ///  The function always returns the parameter <tt>skipMax</tt>.
        /// \details If you want to skip bytes from a Source, then perform the following.
        /// <pre>  StringSource ss(str, false, new Redirector(TheBitBucket()));
        ///   ss.Pump(10);    // Skip 10 bytes from Source
        ///   ss.Detach(new FilterChain(...));
        ///   ss.PumpAll();
        /// </pre>
        virtual lword Skip(lword skipMax=LWORD_MAX);

        /// \brief Copy bytes from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param copyMax the number of bytes to copy
        /// \param channel the channel on which the transfer should occur
        /// \return the number of bytes copied during the call.
        /// \details CopyTo copies bytes from this object to the destination. The bytes are not removed from this object.
        /// \details The function always returns copyMax. If an accurate count is needed, then use CopyRangeTo2().
        lword CopyTo(BufferedTransformation &target, lword copyMax=LWORD_MAX, const std::string &channel=DEFAULT_CHANNEL) const
            {return CopyRangeTo(target, 0, copyMax, channel);}

        /// \brief Copy bytes from this object using an index to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param position the 0-based index of the byte stream to begin the copying
        /// \param copyMax the number of bytes to copy
        /// \param channel the channel on which the transfer should occur
        /// \return the number of bytes copied during the call.
        /// \details CopyTo copies bytes from this object to the destination. The bytes remain in this
        ///  object. Copying begins at the index position in the current stream, and not from an absolute
        ///  position in the stream.
        /// \details The function returns the new position in the stream after transferring the bytes starting at the index.
        lword CopyRangeTo(BufferedTransformation &target, lword position, lword copyMax=LWORD_MAX, const std::string &channel=DEFAULT_CHANNEL) const
            {lword i = position; CopyRangeTo2(target, i, i+copyMax, channel); return i-position;}
    //@}

    /// \name RETRIEVAL OF MULTIPLE MESSAGES
    //@{

        /// \brief Provides the number of bytes ready for retrieval
        /// \return the number of bytes ready for retrieval
        virtual lword TotalBytesRetrievable() const;

        /// \brief Provides the number of meesages processed by this object
        /// \return the number of meesages processed by this object
        /// \details NumberOfMessages returns number of times MessageEnd() has been
        ///  received minus messages retrieved or skipped
        virtual unsigned int NumberOfMessages() const;

        /// \brief Determines if any messages are available for retrieval
        /// \return true if <tt>NumberOfMessages() &gt; 0</tt>, false otherwise
        /// \details AnyMessages returns true if <tt>NumberOfMessages() &gt; 0</tt>
        virtual bool AnyMessages() const;

        /// \brief Start retrieving the next message
        /// \return true if a message is ready for retrieval
        /// \details GetNextMessage() returns true if a message is ready for retrieval; false
        ///  if no more messages exist or this message is not completely retrieved.
        virtual bool GetNextMessage();

        /// \brief Skip a number of meessages
        /// \param count number of messages to skip
        /// \return 0 if the requested number of messages was skipped, non-0 otherwise
        /// \details SkipMessages() skips count number of messages. If there is an AttachedTransformation()
        ///  then SkipMessages() is called on the attached transformation. If there is no attached
        ///  transformation, then count number of messages are sent to TheBitBucket() using TransferMessagesTo().
        virtual unsigned int SkipMessages(unsigned int count=UINT_MAX);

        /// \brief Transfer messages from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param count the number of messages to transfer
        /// \param channel the channel on which the transfer should occur
        /// \return the number of bytes that remain in the current transfer block (i.e., bytes not transferred)
        /// \details TransferMessagesTo2() removes messages from this object and moves them to the destination.
        ///  If all bytes are not transferred for a message, then processing stops and the number of remaining
        ///  bytes is returned. TransferMessagesTo() does not proceed to the next message.
        /// \details A return value of 0 indicates all messages were successfully transferred.
        unsigned int TransferMessagesTo(BufferedTransformation &target, unsigned int count=UINT_MAX, const std::string &channel=DEFAULT_CHANNEL)
            {TransferMessagesTo2(target, count, channel); return count;}

        /// \brief Copy messages from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param count the number of messages to copy
        /// \param channel the channel on which the copy should occur
        /// \return the number of bytes that remain in the current transfer block (i.e., bytes not transferred)
        /// \details CopyMessagesTo copies messages from this object to the destination.
        ///  If all bytes are not transferred for a message, then processing stops and the number of remaining
        ///  bytes is returned. CopyMessagesTo() does not proceed to the next message.
        /// \details A return value of 0 indicates all messages were successfully copied.
        unsigned int CopyMessagesTo(BufferedTransformation &target, unsigned int count=UINT_MAX, const std::string &channel=DEFAULT_CHANNEL) const;

        /// \brief Skip all messages in the series
        virtual void SkipAll();

        /// \brief Transfer all bytes from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param channel the channel on which the transfer should occur
        /// \details TransferMessagesTo2() removes messages from this object and moves them to the destination.
        ///  Internally TransferAllTo() calls TransferAllTo2().
        void TransferAllTo(BufferedTransformation &target, const std::string &channel=DEFAULT_CHANNEL)
            {TransferAllTo2(target, channel);}

        /// \brief Copy messages from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param channel the channel on which the transfer should occur
        /// \details CopyAllTo copies messages from this object and copies them to the destination.
        void CopyAllTo(BufferedTransformation &target, const std::string &channel=DEFAULT_CHANNEL) const;

        /// \brief Retrieve the next message in a series
        /// \return true if a message was retrieved, false otherwise
        /// \details Internally, the base class implementation returns false.
        virtual bool GetNextMessageSeries() {return false;}
        /// \brief Provides the number of messages in a series
        /// \return the number of messages in this series
        virtual unsigned int NumberOfMessagesInThisSeries() const {return NumberOfMessages();}
        /// \brief Provides the number of messages in a series
        /// \return the number of messages in this series
        virtual unsigned int NumberOfMessageSeries() const {return 0;}
    //@}

    /// \name NON-BLOCKING TRANSFER OF OUTPUT
    //@{

        // upon return, byteCount contains number of bytes that have finished being transferred,
        // and returns the number of bytes left in the current transfer block

        /// \brief Transfer bytes from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param byteCount the number of bytes to transfer
        /// \param channel the channel on which the transfer should occur
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain in the transfer block (i.e., bytes not transferred)
        /// \details TransferTo2() removes bytes from this object and moves them to the destination.
        ///  Transfer begins at the index position in the current stream, and not from an absolute
        ///  position in the stream.
        /// \details byteCount is an \a IN and \a OUT parameter. When the call is made,
        ///  byteCount is the requested size of the transfer. When the call returns, byteCount is
        ///  the number of bytes that were transferred.
        virtual size_t TransferTo2(BufferedTransformation &target, lword &byteCount, const std::string &channel=DEFAULT_CHANNEL, bool blocking=true) =0;

        // upon return, begin contains the start position of data yet to be finished copying,
        // and returns the number of bytes left in the current transfer block

        /// \brief Copy bytes from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param begin the 0-based index of the first byte to copy in the stream
        /// \param end the 0-based index of the last byte to copy in the stream
        /// \param channel the channel on which the transfer should occur
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain in the copy block (i.e., bytes not copied)
        /// \details CopyRangeTo2 copies bytes from this object to the destination. The bytes are not
        ///  removed from this object. Copying begins at the index position in the current stream, and
        ///  not from an absolute position in the stream.
        /// \details begin is an \a IN and \a OUT parameter. When the call is made, begin is the
        ///  starting position of the copy. When the call returns, begin is the position of the first
        ///  byte that was \a not copied (which may be different than end). begin can be used for
        ///  subsequent calls to CopyRangeTo2().
        virtual size_t CopyRangeTo2(BufferedTransformation &target, lword &begin, lword end=LWORD_MAX, const std::string &channel=DEFAULT_CHANNEL, bool blocking=true) const =0;

        // upon return, messageCount contains number of messages that have finished being transferred,
        // and returns the number of bytes left in the current transfer block

        /// \brief Transfer messages from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param messageCount the number of messages to transfer
        /// \param channel the channel on which the transfer should occur
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain in the current transfer block (i.e., bytes not transferred)
        /// \details TransferMessagesTo2() removes messages from this object and moves them to the destination.
        /// \details messageCount is an \a IN and \a OUT parameter. When the call is made, messageCount is the
        ///  the number of messages requested to be transferred. When the call returns, messageCount is the
        ///  number of messages actually transferred.
        size_t TransferMessagesTo2(BufferedTransformation &target, unsigned int &messageCount, const std::string &channel=DEFAULT_CHANNEL, bool blocking=true);

        // returns the number of bytes left in the current transfer block

        /// \brief Transfer all bytes from this object to another BufferedTransformation
        /// \param target the destination BufferedTransformation
        /// \param channel the channel on which the transfer should occur
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain in the current transfer block (i.e., bytes not transferred)
        /// \details TransferMessagesTo2() removes messages from this object and moves them to the destination.
        size_t TransferAllTo2(BufferedTransformation &target, const std::string &channel=DEFAULT_CHANNEL, bool blocking=true);
    //@}

    /// \name CHANNELS
    //@{
        /// \brief Exception thrown when a filter does not support named channels
        struct NoChannelSupport : public NotImplemented
            {NoChannelSupport(const std::string &name) : NotImplemented(name + ": this object doesn't support multiple channels") {}};
        /// \brief Exception thrown when a filter does not recognize a named channel
        struct InvalidChannelName : public InvalidArgument
            {InvalidChannelName(const std::string &name, const std::string &channel) : InvalidArgument(name + ": unexpected channel name \"" + channel + "\"") {}};

        /// \brief Input a byte for processing on a channel
        /// \param channel the channel to process the data.
        /// \param inByte the 8-bit byte (octet) to be processed.
        /// \param blocking specifies whether the object should block when processing input.
        /// \return 0 indicates all bytes were processed during the call. Non-0 indicates the
        ///  number of bytes that were not processed.
        size_t ChannelPut(const std::string &channel, byte inByte, bool blocking=true)
            {return ChannelPut(channel, &inByte, 1, blocking);}

        /// \brief Input a byte buffer for processing on a channel
        /// \param channel the channel to process the data
        /// \param inString the byte buffer to process
        /// \param length the size of the string, in bytes
        /// \param blocking specifies whether the object should block when processing input
        /// \return 0 indicates all bytes were processed during the call. Non-0 indicates the
        ///  number of bytes that were not processed.
        size_t ChannelPut(const std::string &channel, const byte *inString, size_t length, bool blocking=true)
            {return ChannelPut2(channel, inString, length, 0, blocking);}

        /// \brief Input multiple bytes that may be modified by callee on a channel
        /// \param channel the channel to process the data.
        /// \param inString the byte buffer to process
        /// \param length the size of the string, in bytes
        /// \param blocking specifies whether the object should block when processing input
        /// \return 0 indicates all bytes were processed during the call. Non-0 indicates the
        ///  number of bytes that were not processed.
        size_t ChannelPutModifiable(const std::string &channel, byte *inString, size_t length, bool blocking=true)
            {return ChannelPutModifiable2(channel, inString, length, 0, blocking);}

        /// \brief Input a 16-bit word for processing on a channel.
        /// \param channel the channel to process the data.
        /// \param value the 16-bit value to be processed.
        /// \param order the ByteOrder of the value to be processed.
        /// \param blocking specifies whether the object should block when processing input.
        /// \return 0 indicates all bytes were processed during the call. Non-0 indicates the
        ///  number of bytes that were not processed.
        size_t ChannelPutWord16(const std::string &channel, word16 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);

        /// \brief Input a 32-bit word for processing on a channel.
        /// \param channel the channel to process the data.
        /// \param value the 32-bit value to be processed.
        /// \param order the ByteOrder of the value to be processed.
        /// \param blocking specifies whether the object should block when processing input.
        /// \return 0 indicates all bytes were processed during the call. Non-0 indicates the
        ///  number of bytes that were not processed.
        size_t ChannelPutWord32(const std::string &channel, word32 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);

        /// \brief Input a 64-bit word for processing on a channel.
        /// \param channel the channel to process the data.
        /// \param value the 64-bit value to be processed.
        /// \param order the ByteOrder of the value to be processed.
        /// \param blocking specifies whether the object should block when processing input.
        /// \return 0 indicates all bytes were processed during the call. Non-0 indicates the
        ///  number of bytes that were not processed.
        size_t ChannelPutWord64(const std::string &channel, word64 value, ByteOrder order=BIG_ENDIAN_ORDER, bool blocking=true);

        /// \brief Signal the end of a message
        /// \param channel the channel to process the data.
        /// \param propagation the number of attached transformations the ChannelMessageEnd() signal should be passed
        /// \param blocking specifies whether the object should block when processing input
        /// \return 0 indicates all bytes were processed during the call. Non-0 indicates the
        ///  number of bytes that were not processed.
        /// \details propagation count includes this object. Setting propagation to <tt>1</tt> means this
        ///  object only. Setting propagation to <tt>-1</tt> means unlimited propagation.
        bool ChannelMessageEnd(const std::string &channel, int propagation=-1, bool blocking=true)
            {return !!ChannelPut2(channel, NULLPTR, 0, propagation < 0 ? -1 : propagation+1, blocking);}

        /// \brief Input multiple bytes for processing and signal the end of a message
        /// \param channel the channel to process the data.
        /// \param inString the byte buffer to process
        /// \param length the size of the string, in bytes
        /// \param propagation the number of attached transformations the ChannelPutMessageEnd() signal should be passed
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed)
        /// \details propagation count includes this object. Setting propagation to <tt>1</tt> means this
        ///  object only. Setting propagation to <tt>-1</tt> means unlimited propagation.
        size_t ChannelPutMessageEnd(const std::string &channel, const byte *inString, size_t length, int propagation=-1, bool blocking=true)
            {return ChannelPut2(channel, inString, length, propagation < 0 ? -1 : propagation+1, blocking);}

        /// \brief Request space which can be written into by the caller
        /// \param channel the channel to process the data
        /// \param size the requested size of the buffer
        /// \return a pointer to a memory block with length size
        /// \details The purpose of this method is to help avoid extra memory allocations.
        /// \details size is an \a IN and \a OUT parameter and used as a hint. When the call is made,
        ///  size is the requested size of the buffer. When the call returns, size is the size of
        ///  the array returned to the caller.
        /// \details The base class implementation sets size to 0 and returns NULL.
        /// \note Some objects, like ArraySink(), cannot create a space because its fixed. In the case of
        ///  an ArraySink(), the pointer to the array is returned and the size is remaining size.
        virtual byte * ChannelCreatePutSpace(const std::string &channel, size_t &size);

        /// \brief Input multiple bytes for processing on a channel.
        /// \param channel the channel to process the data.
        /// \param inString the byte buffer to process.
        /// \param length the size of the string, in bytes.
        /// \param messageEnd means how many filters to signal MessageEnd() to, including this one.
        /// \param blocking specifies whether the object should block when processing input.
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed)
        virtual size_t ChannelPut2(const std::string &channel, const byte *inString, size_t length, int messageEnd, bool blocking);

        /// \brief Input multiple bytes that may be modified by callee on a channel
        /// \param channel the channel to process the data
        /// \param inString the byte buffer to process
        /// \param length the size of the string, in bytes
        /// \param messageEnd means how many filters to signal MessageEnd() to, including this one
        /// \param blocking specifies whether the object should block when processing input
        /// \return the number of bytes that remain to be processed (i.e., bytes not processed)
        virtual size_t ChannelPutModifiable2(const std::string &channel, byte *inString, size_t length, int messageEnd, bool blocking);

        /// \brief Flush buffered input and/or output on a channel
        /// \param channel the channel to flush the data
        /// \param hardFlush is used to indicate whether all data should be flushed
        /// \param propagation the number of attached transformations the ChannelFlush() signal should be passed
        /// \param blocking specifies whether the object should block when processing input
        /// \return true of the Flush was successful
        /// \details propagation count includes this object. Setting propagation to <tt>1</tt> means this
        ///  object only. Setting propagation to <tt>-1</tt> means unlimited propagation.
        virtual bool ChannelFlush(const std::string &channel, bool hardFlush, int propagation=-1, bool blocking=true);

        /// \brief Marks the end of a series of messages on a channel
        /// \param channel the channel to signal the end of a series of messages
        /// \param propagation the number of attached transformations the ChannelMessageSeriesEnd() signal should be passed
        /// \param blocking specifies whether the object should block when processing input
        /// \return true if the message was successful, false otherwise
        /// \details Each object that receives the signal will perform its processing, decrement
        ///  propagation, and then pass the signal on to attached transformations if the value is not 0.
        /// \details propagation count includes this object. Setting propagation to <tt>1</tt> means this
        ///  object only. Setting propagation to <tt>-1</tt> means unlimited propagation.
        /// \note There should be a MessageEnd() immediately before MessageSeriesEnd().
        virtual bool ChannelMessageSeriesEnd(const std::string &channel, int propagation=-1, bool blocking=true);

        /// \brief Sets the default retrieval channel
        /// \param channel the channel to signal the end of a series of messages
        /// \note this function may not be implemented in all objects that should support it.
        virtual void SetRetrievalChannel(const std::string &channel);
    //@}

    /// \name ATTACHMENT
    /// \details Some BufferedTransformation objects (e.g. Filter objects) allow other BufferedTransformation objects to be
    ///  attached. When this is done, the first object instead of buffering its output, sends that output to the attached
    ///  object as input. The entire attachment chain is deleted when the anchor object is destructed.

    //@{
        /// \brief Determines whether the object allows attachment
        /// \return true if the object allows an attachment, false otherwise
        /// \details Sources and Filters will returns true, while Sinks and other objects will return false.
        virtual bool Attachable() {return false;}

        /// \brief Returns the object immediately attached to this object
        /// \return the attached transformation
        /// \details AttachedTransformation() returns NULL if there is no attachment. The non-const
        ///  version of AttachedTransformation() always returns NULL.
        virtual BufferedTransformation *AttachedTransformation() {CRYPTOPP_ASSERT(!Attachable()); return NULLPTR;}

        /// \brief Returns the object immediately attached to this object
        /// \return the attached transformation
        /// \details AttachedTransformation() returns NULL if there is no attachment. The non-const
        ///  version of AttachedTransformation() always returns NULL.
        virtual const BufferedTransformation *AttachedTransformation() const
            {return const_cast<BufferedTransformation *>(this)->AttachedTransformation();}

        /// \brief Delete the current attachment chain and attach a new one
        /// \param newAttachment the new BufferedTransformation to attach
        /// \throw NotImplemented
        /// \details Detach() deletes the current attachment chain and replace it with an optional newAttachment
        /// \details If a derived class does not override Detach(), then the base class throws
        ///  NotImplemented.
        virtual void Detach(BufferedTransformation *newAttachment = NULLPTR) {
            CRYPTOPP_UNUSED(newAttachment); CRYPTOPP_ASSERT(!Attachable());
            throw NotImplemented("BufferedTransformation: this object is not attachable");
        }

        /// \brief Add newAttachment to the end of attachment chain
        /// \param newAttachment the attachment to add to the end of the chain
        virtual void Attach(BufferedTransformation *newAttachment);
    //@}

protected:
    /// \brief Decrements the propagation count while clamping at 0
    /// \return the decremented propagation or 0
    static int DecrementPropagation(int propagation)
        {return propagation != 0 ? propagation - 1 : 0;}

private:
    // for ChannelPutWord16, ChannelPutWord32 and ChannelPutWord64,
    // to ensure the buffer isn't deallocated before non-blocking
    // operation completes
    byte m_buf[8];
};

/// \brief An input discarding BufferedTransformation
/// \return a reference to a BufferedTransformation object that discards all input
CRYPTOPP_DLL BufferedTransformation & TheBitBucket();

/// \brief Interface for crypto material
/// \details CryptoMaterial() is an interface for crypto material, such as
///  public keys, private keys and crypto parameters. Derived classes generally
///  do not offer public methods such as GenerateRandom() and
///  GenerateRandomWithKeySize().
/// \sa GeneratableCryptoMaterial()
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE CryptoMaterial : public NameValuePairs
{
public:
    /// Exception thrown when invalid crypto material is detected
    class CRYPTOPP_DLL InvalidMaterial : public InvalidDataFormat
    {
    public:
        explicit InvalidMaterial(const std::string &s) : InvalidDataFormat(s) {}
    };

    virtual ~CryptoMaterial() {}

    /// \brief Assign values to this object
    /// \details This function can be used to create a public key from a private key.
    virtual void AssignFrom(const NameValuePairs &source) =0;

    /// \brief Check this object for errors
    /// \param rng a RandomNumberGenerator for objects which use randomized testing
    /// \param level the level of thoroughness
    /// \return true if the tests succeed, false otherwise
    /// \details There are four levels of thoroughness:
    ///   <ul>
    ///   <li>0 - using this object won't cause a crash or exception
    ///   <li>1 - this object will probably function, and encrypt, sign, other operations correctly
    ///   <li>2 - ensure this object will function correctly, and perform reasonable security checks
    ///   <li>3 - perform reasonable security checks, and do checks that may take a long time
    ///   </ul>
    /// \details Level 0 does not require a RandomNumberGenerator. A NullRNG() can be used for level 0.
    ///  Level 1 may not check for weak keys and such. Levels 2 and 3 are recommended.
    /// \sa ThrowIfInvalid()
    virtual bool Validate(RandomNumberGenerator &rng, unsigned int level) const =0;

    /// \brief Check this object for errors
    /// \param rng a RandomNumberGenerator for objects which use randomized testing
    /// \param level the level of thoroughness
    /// \throw InvalidMaterial
    /// \details Internally, ThrowIfInvalid() calls Validate() and throws InvalidMaterial() if validation fails.
    /// \sa Validate()
    virtual void ThrowIfInvalid(RandomNumberGenerator &rng, unsigned int level) const
        {if (!Validate(rng, level)) throw InvalidMaterial("CryptoMaterial: this object contains invalid values");}

    /// \brief Saves a key to a BufferedTransformation
    /// \param bt the destination BufferedTransformation
    /// \throw NotImplemented
    /// \details Save() writes the material to a BufferedTransformation.
    /// \details If the material is a key, then the key is written with ASN.1 DER encoding. The key
    ///  includes an object identifier with an algorithm id, like a subjectPublicKeyInfo.
    /// \details A "raw" key without the "key info" can be saved using a key's DEREncode() method.
    /// \details If a derived class does not override Save(), then the base class throws
    ///  NotImplemented().
    virtual void Save(BufferedTransformation &bt) const
        {CRYPTOPP_UNUSED(bt); throw NotImplemented("CryptoMaterial: this object does not support saving");}

    /// \brief Loads a key from a BufferedTransformation
    /// \param bt the source BufferedTransformation
    /// \throw KeyingErr
    /// \details Load() attempts to read material from a BufferedTransformation. If the
    ///  material is a key that was generated outside the library, then the following
    ///  usually applies:
    ///   <ul>
    ///   <li>the key should be ASN.1 BER encoded
    ///   <li>the key should be a "key info"
    ///   </ul>
    /// \details "key info" means the key should have an object identifier with an algorithm id,
    ///  like a subjectPublicKeyInfo.
    /// \details To read a "raw" key without the "key info", then call the key's BERDecode() method.
    /// \note Load() generally does not check that the key is valid. Call Validate(), if needed.
    virtual void Load(BufferedTransformation &bt)
        {CRYPTOPP_UNUSED(bt); throw NotImplemented("CryptoMaterial: this object does not support loading");}

    /// \brief Determines whether the object supports precomputation
    /// \return true if the object supports precomputation, false otherwise
    /// \sa Precompute()
    virtual bool SupportsPrecomputation() const {return false;}

    /// \brief Perform precomputation
    /// \param precomputationStorage the suggested number of objects for the precompute table
    /// \throw NotImplemented
    /// \details The exact semantics of Precompute() varies, but it typically means calculate
    ///  a table of n objects that can be used later to speed up computation.
    /// \details If a derived class does not override Precompute(), then the base class throws
    ///  NotImplemented.
    /// \sa SupportsPrecomputation(), LoadPrecomputation(), SavePrecomputation()
    virtual void Precompute(unsigned int precomputationStorage) {
        CRYPTOPP_UNUSED(precomputationStorage); CRYPTOPP_ASSERT(!SupportsPrecomputation());
        throw NotImplemented("CryptoMaterial: this object does not support precomputation");
    }

    /// \brief Retrieve previously saved precomputation
    /// \param storedPrecomputation BufferedTransformation with the saved precomputation
    /// \throw NotImplemented
    /// \sa SupportsPrecomputation(), Precompute()
    virtual void LoadPrecomputation(BufferedTransformation &storedPrecomputation)
        {CRYPTOPP_UNUSED(storedPrecomputation); CRYPTOPP_ASSERT(!SupportsPrecomputation()); throw NotImplemented("CryptoMaterial: this object does not support precomputation");}

    /// \brief Save precomputation for later use
    /// \param storedPrecomputation BufferedTransformation to write the precomputation
    /// \throw NotImplemented
    /// \sa SupportsPrecomputation(), Precompute()
    virtual void SavePrecomputation(BufferedTransformation &storedPrecomputation) const
        {CRYPTOPP_UNUSED(storedPrecomputation); CRYPTOPP_ASSERT(!SupportsPrecomputation()); throw NotImplemented("CryptoMaterial: this object does not support precomputation");}

    /// \brief Perform a quick sanity check
    /// \details DoQuickSanityCheck() is for internal library use, and it should not be called by library users.
    void DoQuickSanityCheck() const {ThrowIfInvalid(NullRNG(), 0);}

#if defined(__SUNPRO_CC)
    // Sun Studio 11/CC 5.8 workaround: it generates incorrect code
    // when casting to an empty virtual base class. JW, 2018: It is
    // still a problem in Sun Studio 12.6/CC 5.15 on i386. Just enable
    // it everywhere in case it affects SPARC (which we don't test).
    char m_sunCCworkaround;
#endif
};

/// \brief Interface for crypto material
/// \details GeneratableCryptoMaterial() is an interface for crypto material,
///  such as private keys and crypto parameters. Derived classes offer public
///  methods such as GenerateRandom() and GenerateRandomWithKeySize().
/// \sa CryptoMaterial()
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE GeneratableCryptoMaterial : virtual public CryptoMaterial
{
public:
    virtual ~GeneratableCryptoMaterial() {}

    /// \brief Generate a random key or crypto parameters
    /// \param rng a RandomNumberGenerator to produce keying material
    /// \param params additional initialization parameters
    /// \throw KeyingErr if a key can't be generated or algorithm parameters are invalid
    /// \details If a derived class does not override GenerateRandom(), then the base class throws
    ///  NotImplemented.
    virtual void GenerateRandom(RandomNumberGenerator &rng, const NameValuePairs &params = g_nullNameValuePairs) {
        CRYPTOPP_UNUSED(rng); CRYPTOPP_UNUSED(params);
        throw NotImplemented("GeneratableCryptoMaterial: this object does not support key/parameter generation");
    }

    /// \brief Generate a random key or crypto parameters
    /// \param rng a RandomNumberGenerator to produce keying material
    /// \param keySize the size of the key, in bits
    /// \throw KeyingErr if a key can't be generated or algorithm parameters are invalid
    /// \details GenerateRandomWithKeySize calls GenerateRandom() with a NameValuePairs
    ///  object with only "KeySize"
    void GenerateRandomWithKeySize(RandomNumberGenerator &rng, unsigned int keySize);
};

/// \brief Interface for public keys
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PublicKey : virtual public CryptoMaterial
{
};

/// \brief Interface for private keys
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PrivateKey : public GeneratableCryptoMaterial
{
};

/// \brief Interface for crypto parameters
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE CryptoParameters : public GeneratableCryptoMaterial
{
};

/// \brief Interface for certificates
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE Certificate : virtual public CryptoMaterial
{
};

/// \brief Interface for asymmetric algorithms
/// \details BERDecode() and DEREncode() were removed under Issue 569
///  and Commit 9b174e84de7a. Programs should use <tt>AccessMaterial().Load(bt)</tt>
///  or <tt>GetMaterial().Save(bt)</tt> instead.
/// \sa <A HREF="https://github.com/weidai11/cryptopp/issues/569">Issue 569</A>
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE AsymmetricAlgorithm : public Algorithm
{
public:
    virtual ~AsymmetricAlgorithm() {}

    /// \brief Retrieves a reference to CryptoMaterial
    /// \return a reference to the crypto material
    virtual CryptoMaterial & AccessMaterial() =0;

    /// \brief Retrieves a reference to CryptoMaterial
    /// \return a const reference to the crypto material
    virtual const CryptoMaterial & GetMaterial() const =0;

#if 0
    /// \brief Loads this object from a BufferedTransformation
    /// \param bt a BufferedTransformation object
    /// \details Use of BERDecode() changed to Load() at Issue 569.
    /// \deprecated for backwards compatibility, calls <tt>AccessMaterial().Load(bt)</tt>
    void BERDecode(BufferedTransformation &bt)
        {AccessMaterial().Load(bt);}

    /// \brief Saves this object to a BufferedTransformation
    /// \param bt a BufferedTransformation object
    /// \details Use of DEREncode() changed to Save() at Issue 569.
    /// \deprecated for backwards compatibility, calls GetMaterial().Save(bt)
    void DEREncode(BufferedTransformation &bt) const
        {GetMaterial().Save(bt);}
#endif
};

/// \brief Interface for asymmetric algorithms using public keys
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PublicKeyAlgorithm : public AsymmetricAlgorithm
{
public:
    virtual ~PublicKeyAlgorithm() {}

    // VC60 workaround: no co-variant return type

    /// \brief Retrieves a reference to a Public Key
    /// \return a reference to the public key
    CryptoMaterial & AccessMaterial()
        {return AccessPublicKey();}
    /// \brief Retrieves a reference to a Public Key
    /// \return a const reference the public key
    const CryptoMaterial & GetMaterial() const
        {return GetPublicKey();}

    /// \brief Retrieves a reference to a Public Key
    /// \return a reference to the public key
    virtual PublicKey & AccessPublicKey() =0;
    /// \brief Retrieves a reference to a Public Key
    /// \return a const reference the public key
    virtual const PublicKey & GetPublicKey() const
        {return const_cast<PublicKeyAlgorithm *>(this)->AccessPublicKey();}
};

/// \brief Interface for asymmetric algorithms using private keys
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PrivateKeyAlgorithm : public AsymmetricAlgorithm
{
public:
    virtual ~PrivateKeyAlgorithm() {}

    /// \brief Retrieves a reference to a Private Key
    /// \return a reference the private key
    CryptoMaterial & AccessMaterial() {return AccessPrivateKey();}
    /// \brief Retrieves a reference to a Private Key
    /// \return a const reference the private key
    const CryptoMaterial & GetMaterial() const {return GetPrivateKey();}

    /// \brief Retrieves a reference to a Private Key
    /// \return a reference the private key
    virtual PrivateKey & AccessPrivateKey() =0;
    /// \brief Retrieves a reference to a Private Key
    /// \return a const reference the private key
    virtual const PrivateKey & GetPrivateKey() const {return const_cast<PrivateKeyAlgorithm *>(this)->AccessPrivateKey();}
};

/// \brief Interface for key agreement algorithms
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE KeyAgreementAlgorithm : public AsymmetricAlgorithm
{
public:
    virtual ~KeyAgreementAlgorithm() {}

    /// \brief Retrieves a reference to Crypto Parameters
    /// \return a reference the crypto parameters
    CryptoMaterial & AccessMaterial() {return AccessCryptoParameters();}
    /// \brief Retrieves a reference to Crypto Parameters
    /// \return a const reference the crypto parameters
    const CryptoMaterial & GetMaterial() const {return GetCryptoParameters();}

    /// \brief Retrieves a reference to Crypto Parameters
    /// \return a reference the crypto parameters
    virtual CryptoParameters & AccessCryptoParameters() =0;
    /// \brief Retrieves a reference to Crypto Parameters
    /// \return a const reference the crypto parameters
    virtual const CryptoParameters & GetCryptoParameters() const {return const_cast<KeyAgreementAlgorithm *>(this)->AccessCryptoParameters();}
};

/// \brief Interface for public-key encryptors and decryptors
/// \details This class provides an interface common to encryptors and decryptors
///  for querying their plaintext and ciphertext lengths.
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PK_CryptoSystem
{
public:
    virtual ~PK_CryptoSystem() {}

    /// \brief Provides the maximum length of plaintext for a given ciphertext length
    /// \return the maximum size of the plaintext, in bytes
    /// \details This function returns 0 if ciphertextLength is not valid (too long or too short).
    virtual size_t MaxPlaintextLength(size_t ciphertextLength) const =0;

    /// \brief Calculate the length of ciphertext given length of plaintext
    /// \return the maximum size of the ciphertext, in bytes
    /// \details This function returns 0 if plaintextLength is not valid (too long).
    virtual size_t CiphertextLength(size_t plaintextLength) const =0;

    /// \brief Determines whether this object supports the use of a named parameter
    /// \param name the name of the parameter
    /// \return true if the parameter name is supported, false otherwise
    /// \details Some possible parameter names: EncodingParameters(), KeyDerivationParameters()
    ///  and others Parameters listed in argnames.h
    virtual bool ParameterSupported(const char *name) const =0;

    /// \brief Provides the fixed ciphertext length, if one exists
    /// \return the fixed ciphertext length if one exists, otherwise 0
    /// \details "Fixed" here means length of ciphertext does not depend on length of plaintext.
    ///  In this case, it usually does depend on the key length.
    virtual size_t FixedCiphertextLength() const {return 0;}

    /// \brief Provides the maximum plaintext length given a fixed ciphertext length
    /// \return maximum plaintext length given the fixed ciphertext length, if one exists,
    ///  otherwise return 0.
    /// \details FixedMaxPlaintextLength(0 returns the maximum plaintext length given the fixed ciphertext
    ///  length, if one exists, otherwise return 0.
    virtual size_t FixedMaxPlaintextLength() const {return 0;}
};

/// \brief Interface for public-key encryptors
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PK_Encryptor : public PK_CryptoSystem, public PublicKeyAlgorithm
{
public:
    /// \brief Exception thrown when trying to encrypt plaintext of invalid length
    class CRYPTOPP_DLL InvalidPlaintextLength : public Exception
    {
    public:
        InvalidPlaintextLength() : Exception(OTHER_ERROR, "PK_Encryptor: invalid plaintext length") {}
    };

    /// \brief Encrypt a byte string
    /// \param rng a RandomNumberGenerator derived class
    /// \param plaintext the plaintext byte buffer
    /// \param plaintextLength the size of the plaintext byte buffer
    /// \param ciphertext a byte buffer to hold the encrypted string
    /// \param parameters a set of NameValuePairs to initialize this object
    /// \pre <tt>CiphertextLength(plaintextLength) != 0</tt> ensures the plaintext isn't too large
    /// \pre <tt>COUNTOF(ciphertext) == CiphertextLength(plaintextLength)</tt> ensures the output
    ///  byte buffer is large enough.
    /// \sa PK_Decryptor
    virtual void Encrypt(RandomNumberGenerator &rng,
        const byte *plaintext, size_t plaintextLength,
        byte *ciphertext, const NameValuePairs &parameters = g_nullNameValuePairs) const =0;

    /// \brief Create a new encryption filter
    /// \param rng a RandomNumberGenerator derived class
    /// \param attachment an attached transformation
    /// \param parameters a set of NameValuePairs to initialize this object
    /// \details \p attachment can be \p NULL. The caller is responsible for deleting the returned pointer.
    ///  Encoding parameters should be passed in the "EP" channel.
    virtual BufferedTransformation * CreateEncryptionFilter(RandomNumberGenerator &rng,
        BufferedTransformation *attachment=NULLPTR, const NameValuePairs &parameters = g_nullNameValuePairs) const;
};

/// \brief Interface for public-key decryptors
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PK_Decryptor : public PK_CryptoSystem, public PrivateKeyAlgorithm
{
public:
    virtual ~PK_Decryptor() {}

    /// \brief Decrypt a byte string
    /// \param rng a RandomNumberGenerator derived class
    /// \param ciphertext the encrypted byte buffer
    /// \param ciphertextLength the size of the encrypted byte buffer
    /// \param plaintext a byte buffer to hold the decrypted string
    /// \param parameters a set of NameValuePairs to initialize this object
    /// \return the result of the decryption operation
    /// \details If DecodingResult::isValidCoding is true, then DecodingResult::messageLength
    ///  is valid and holds the actual length of the plaintext recovered. The result is undefined
    ///  if decryption failed. If DecodingResult::isValidCoding is false, then DecodingResult::messageLength
    ///  is undefined.
    /// \pre <tt>COUNTOF(plaintext) == MaxPlaintextLength(ciphertextLength)</tt> ensures the output
    ///  byte buffer is large enough
    /// \sa PK_Encryptor
    virtual DecodingResult Decrypt(RandomNumberGenerator &rng,
        const byte *ciphertext, size_t ciphertextLength,
        byte *plaintext, const NameValuePairs &parameters = g_nullNameValuePairs) const =0;

    /// \brief Create a new decryption filter
    /// \param rng a RandomNumberGenerator derived class
    /// \param attachment an attached transformation
    /// \param parameters a set of NameValuePairs to initialize this object
    /// \return the newly created decryption filter
    /// \note the caller is responsible for deleting the returned pointer
    virtual BufferedTransformation * CreateDecryptionFilter(RandomNumberGenerator &rng,
        BufferedTransformation *attachment=NULLPTR, const NameValuePairs &parameters = g_nullNameValuePairs) const;

    /// \brief Decrypt a fixed size ciphertext
    /// \param rng a RandomNumberGenerator derived class
    /// \param ciphertext the encrypted byte buffer
    /// \param plaintext a byte buffer to hold the decrypted string
    /// \param parameters a set of NameValuePairs to initialize this object
    /// \return the result of the decryption operation
    /// \details If DecodingResult::isValidCoding is true, then DecodingResult::messageLength
    ///  is valid and holds the actual length of the plaintext recovered. The result is undefined
    ///  if decryption failed. If DecodingResult::isValidCoding is false, then DecodingResult::messageLength
    ///  is undefined.
    /// \pre <tt>COUNTOF(plaintext) == MaxPlaintextLength(ciphertextLength)</tt> ensures the output
    ///  byte buffer is large enough
    /// \sa PK_Encryptor
    DecodingResult FixedLengthDecrypt(RandomNumberGenerator &rng, const byte *ciphertext, byte *plaintext, const NameValuePairs &parameters = g_nullNameValuePairs) const
        {return Decrypt(rng, ciphertext, FixedCiphertextLength(), plaintext, parameters);}
};

/// \brief Interface for public-key signers and verifiers
/// \details This class provides an interface common to signers and verifiers for querying scheme properties
/// \sa DL_SignatureSchemeBase, TF_SignatureSchemeBase, DL_SignerBase, TF_SignerBase
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PK_SignatureScheme
{
public:
    /// \brief Exception throw when the private or public key has a length that can't be used
    /// \details InvalidKeyLength() may be thrown by any function in this class if the private
    ///  or public key has a length that can't be used
    class CRYPTOPP_DLL InvalidKeyLength : public Exception
    {
    public:
        InvalidKeyLength(const std::string &message) : Exception(OTHER_ERROR, message) {}
    };

    /// \brief Exception throw when the private or public key is too short to sign or verify
    /// \details KeyTooShort() may be thrown by any function in this class if the private or public
    ///  key is too short to sign or verify anything
    class CRYPTOPP_DLL KeyTooShort : public InvalidKeyLength
    {
    public:
        KeyTooShort() : InvalidKeyLength("PK_Signer: key too short for this signature scheme") {}
    };

    virtual ~PK_SignatureScheme() {}

    /// \brief Provides the signature length if it only depends on the key
    /// \return the signature length if it only depends on the key, in bytes
    /// \details SignatureLength() returns the signature length if it only depends on the key, otherwise 0.
    virtual size_t SignatureLength() const =0;

    /// \brief Provides the maximum signature length produced given the length of the recoverable message part
    /// \param recoverablePartLength the length of the recoverable message part, in bytes
    /// \return the maximum signature length produced for a given length of recoverable message part, in bytes
    /// \details MaxSignatureLength() returns the maximum signature length produced given the length of the
    ///  recoverable message part.
    virtual size_t MaxSignatureLength(size_t recoverablePartLength = 0) const
        {CRYPTOPP_UNUSED(recoverablePartLength); return SignatureLength();}

    /// \brief Provides the length of longest message that can be recovered
    /// \return the length of longest message that can be recovered, in bytes
    /// \details MaxRecoverableLength() returns the length of longest message that can be recovered, or 0 if
    ///  this signature scheme does not support message recovery.
    virtual size_t MaxRecoverableLength() const =0;

    /// \brief Provides the length of longest message that can be recovered from a signature of given length
    /// \param signatureLength the length of the signature, in bytes
    /// \return the length of longest message that can be recovered from a signature of given length, in bytes
    /// \details MaxRecoverableLengthFromSignatureLength() returns the length of longest message that can be
    ///  recovered from a signature of given length, or 0 if this signature scheme does not support message
    ///  recovery.
    virtual size_t MaxRecoverableLengthFromSignatureLength(size_t signatureLength) const =0;

    /// \brief Determines whether a signature scheme requires a random number generator
    /// \return true if the signature scheme requires a RandomNumberGenerator() to sign
    /// \details if IsProbabilistic() returns false, then NullRNG() can be passed to functions that take
    ///  RandomNumberGenerator().
    virtual bool IsProbabilistic() const =0;

    /// \brief Determines whether the non-recoverable message part can be signed
    /// \return true if the non-recoverable message part can be signed
    virtual bool AllowNonrecoverablePart() const =0;

    /// \brief Determines whether the signature must be input before the message
    /// \return true if the signature must be input before the message during verifcation
    /// \details if SignatureUpfront() returns true, then you must input the signature before the message
    ///  during verification. Otherwise you can input the signature at anytime.
    virtual bool SignatureUpfront() const {return false;}

    /// \brief Determines whether the recoverable part must be input before the non-recoverable part
    /// \return true if the recoverable part must be input before the non-recoverable part during signing
    /// \details RecoverablePartFirst() determines whether you must input the recoverable part before the
    ///  non-recoverable part during signing
    virtual bool RecoverablePartFirst() const =0;
};

/// \brief Interface for accumulating messages to be signed or verified
/// \details Only Update() should be called from the PK_MessageAccumulator() class. No other functions
///  inherited from HashTransformation, like DigestSize() and TruncatedFinal(), should be called.
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PK_MessageAccumulator : public HashTransformation
{
public:
    /// \warning DigestSize() should not be called on PK_MessageAccumulator
    unsigned int DigestSize() const
        {throw NotImplemented("PK_MessageAccumulator: DigestSize() should not be called");}

    /// \warning TruncatedFinal() should not be called on PK_MessageAccumulator
    void TruncatedFinal(byte *digest, size_t digestSize)
    {
        CRYPTOPP_UNUSED(digest); CRYPTOPP_UNUSED(digestSize);
        throw NotImplemented("PK_MessageAccumulator: TruncatedFinal() should not be called");
    }
};

/// \brief Interface for public-key signers
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PK_Signer : public PK_SignatureScheme, public PrivateKeyAlgorithm
{
public:
    virtual ~PK_Signer() {}

    /// \brief Create a new HashTransformation to accumulate the message to be signed
    /// \param rng a RandomNumberGenerator derived class
    /// \return a pointer to a PK_MessageAccumulator
    /// \details NewSignatureAccumulator() can be used with all signing methods. Sign() will automatically delete the
    ///  accumulator pointer. The caller is responsible for deletion if a method is called that takes a reference.
    virtual PK_MessageAccumulator * NewSignatureAccumulator(RandomNumberGenerator &rng) const =0;

    /// \brief Input a recoverable message to an accumulator
    /// \param messageAccumulator a reference to a PK_MessageAccumulator
    /// \param recoverableMessage a pointer to the recoverable message part to be signed
    /// \param recoverableMessageLength the size of the recoverable message part
    virtual void InputRecoverableMessage(PK_MessageAccumulator &messageAccumulator, const byte *recoverableMessage, size_t recoverableMessageLength) const =0;

    /// \brief Sign and delete the messageAccumulator
    /// \param rng a RandomNumberGenerator derived class
    /// \param messageAccumulator a pointer to a PK_MessageAccumulator derived class
    /// \param signature a block of bytes for the signature
    /// \return actual signature length
    /// \details Sign() deletes the messageAccumulator, even if an exception is thrown.
    /// \pre <tt>COUNTOF(signature) == MaxSignatureLength()</tt>
    virtual size_t Sign(RandomNumberGenerator &rng, PK_MessageAccumulator *messageAccumulator, byte *signature) const;

    /// \brief Sign and restart messageAccumulator
    /// \param rng a RandomNumberGenerator derived class
    /// \param messageAccumulator a pointer to a PK_MessageAccumulator derived class
    /// \param signature a block of bytes for the signature
    /// \param restart flag indicating whether the messageAccumulator should be restarted
    /// \return actual signature length
    /// \pre <tt>COUNTOF(signature) == MaxSignatureLength()</tt>
    virtual size_t SignAndRestart(RandomNumberGenerator &rng, PK_MessageAccumulator &messageAccumulator, byte *signature, bool restart=true) const =0;

    /// \brief Sign a message
    /// \param rng a RandomNumberGenerator derived class
    /// \param message a pointer to the message
    /// \param messageLen the size of the message to be signed
    /// \param signature a block of bytes for the signature
    /// \return actual signature length
    /// \pre <tt>COUNTOF(signature) == MaxSignatureLength()</tt>
    virtual size_t SignMessage(RandomNumberGenerator &rng, const byte *message, size_t messageLen, byte *signature) const;

    /// \brief Sign a recoverable message
    /// \param rng a RandomNumberGenerator derived class
    /// \param recoverableMessage a pointer to the recoverable message part to be signed
    /// \param recoverableMessageLength the size of the recoverable message part
    /// \param nonrecoverableMessage a pointer to the non-recoverable message part to be signed
    /// \param nonrecoverableMessageLength the size of the non-recoverable message part
    /// \param signature a block of bytes for the signature
    /// \return actual signature length
    /// \pre <tt>COUNTOF(signature) == MaxSignatureLength(recoverableMessageLength)</tt>
    virtual size_t SignMessageWithRecovery(RandomNumberGenerator &rng, const byte *recoverableMessage, size_t recoverableMessageLength,
        const byte *nonrecoverableMessage, size_t nonrecoverableMessageLength, byte *signature) const;
};

/// \brief Interface for public-key signature verifiers
/// \details The Recover* functions throw NotImplemented if the signature scheme does not support
///  message recovery.
/// \details The Verify* functions throw InvalidDataFormat if the scheme does support message
///  recovery and the signature contains a non-empty recoverable message part. The
///  Recover* functions should be used in that case.
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE PK_Verifier : public PK_SignatureScheme, public PublicKeyAlgorithm
{
public:
    virtual ~PK_Verifier() {}

    /// \brief Create a new HashTransformation to accumulate the message to be verified
    /// \return a pointer to a PK_MessageAccumulator
    /// \details NewVerificationAccumulator() can be used with all verification methods. Verify() will automatically delete
    ///  the accumulator pointer. The caller is responsible for deletion if a method is called that takes a reference.
    virtual PK_MessageAccumulator * NewVerificationAccumulator() const =0;

    /// \brief Input signature into a message accumulator
    /// \param messageAccumulator a pointer to a PK_MessageAccumulator derived class
    /// \param signature the signature on the message
    /// \param signatureLength the size of the signature
    virtual void InputSignature(PK_MessageAccumulator &messageAccumulator, const byte *signature, size_t signatureLength) const =0;

    /// \brief Check whether messageAccumulator contains a valid signature and message
    /// \param messageAccumulator a pointer to a PK_MessageAccumulator derived class
    /// \return true if the signature is valid, false otherwise
    /// \details Verify() deletes the messageAccumulator, even if an exception is thrown.
    virtual bool Verify(PK_MessageAccumulator *messageAccumulator) const;

    /// \brief Check whether messageAccumulator contains a valid signature and message, and restart messageAccumulator
    /// \param messageAccumulator a reference to a PK_MessageAccumulator derived class
    /// \return true if the signature is valid, false otherwise
    /// \details VerifyAndRestart() restarts the messageAccumulator
    virtual bool VerifyAndRestart(PK_MessageAccumulator &messageAccumulator) const =0;

    /// \brief Check whether input signature is a valid signature for input message
    /// \param message a pointer to the message to be verified
    /// \param messageLen the size of the message
    /// \param signature a pointer to the signature over the message
    /// \param signatureLen the size of the signature
    /// \return true if the signature is valid, false otherwise
    virtual bool VerifyMessage(const byte *message, size_t messageLen,
        const byte *signature, size_t signatureLen) const;

    /// \brief Recover a message from its signature
    /// \param recoveredMessage a pointer to the recoverable message part to be verified
    /// \param messageAccumulator a pointer to a PK_MessageAccumulator derived class
    /// \return the result of the verification operation
    /// \details Recover() deletes the messageAccumulator, even if an exception is thrown.
    /// \pre <tt>COUNTOF(recoveredMessage) == MaxRecoverableLengthFromSignatureLength(signatureLength)</tt>
    virtual DecodingResult Recover(byte *recoveredMessage, PK_MessageAccumulator *messageAccumulator) const;

    /// \brief Recover a message from its signature
    /// \param recoveredMessage a pointer to the recoverable message part to be verified
    /// \param messageAccumulator a pointer to a PK_MessageAccumulator derived class
    /// \return the result of the verification operation
    /// \details RecoverAndRestart() restarts the messageAccumulator
    /// \pre <tt>COUNTOF(recoveredMessage) == MaxRecoverableLengthFromSignatureLength(signatureLength)</tt>
    virtual DecodingResult RecoverAndRestart(byte *recoveredMessage, PK_MessageAccumulator &messageAccumulator) const =0;

    /// \brief Recover a message from its signature
    /// \param recoveredMessage a pointer for the recovered message
    /// \param nonrecoverableMessage a pointer to the non-recoverable message part to be signed
    /// \param nonrecoverableMessageLength the size of the non-recoverable message part
    /// \param signature the signature on the message
    /// \param signatureLength the size of the signature
    /// \return the result of the verification operation
    /// \pre <tt>COUNTOF(recoveredMessage) == MaxRecoverableLengthFromSignatureLength(signatureLength)</tt>
    virtual DecodingResult RecoverMessage(byte *recoveredMessage,
        const byte *nonrecoverableMessage, size_t nonrecoverableMessageLength,
        const byte *signature, size_t signatureLength) const;
};

/// \brief Interface for domains of simple key agreement protocols
/// \details A key agreement domain is a set of parameters that must be shared
///  by two parties in a key agreement protocol, along with the algorithms
///  for generating key pairs and deriving agreed values.
/// \since Crypto++ 3.0
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE SimpleKeyAgreementDomain : public KeyAgreementAlgorithm
{
public:
    virtual ~SimpleKeyAgreementDomain() {}

    /// \brief Provides the size of the agreed value
    /// \return size of agreed value produced in this domain
    virtual unsigned int AgreedValueLength() const =0;

    /// \brief Provides the size of the private key
    /// \return size of private keys in this domain
    virtual unsigned int PrivateKeyLength() const =0;

    /// \brief Provides the size of the public key
    /// \return size of public keys in this domain
    virtual unsigned int PublicKeyLength() const =0;

    /// \brief Generate private key in this domain
    /// \param rng a RandomNumberGenerator derived class
    /// \param privateKey a byte buffer for the generated private key in this domain
    /// \pre <tt>COUNTOF(privateKey) == PrivateKeyLength()</tt>
    virtual void GeneratePrivateKey(RandomNumberGenerator &rng, byte *privateKey) const =0;

    /// \brief Generate a public key from a private key in this domain
    /// \param rng a RandomNumberGenerator derived class
    /// \param privateKey a byte buffer with the previously generated private key
    /// \param publicKey a byte buffer for the generated public key in this domain
    /// \pre <tt>COUNTOF(publicKey) == PublicKeyLength()</tt>
    virtual void GeneratePublicKey(RandomNumberGenerator &rng, const byte *privateKey, byte *publicKey) const =0;

    /// \brief Generate a private/public key pair
    /// \param rng a RandomNumberGenerator derived class
    /// \param privateKey a byte buffer for the generated private key in this domain
    /// \param publicKey a byte buffer for the generated public key in this domain
    /// \details GenerateKeyPair() is equivalent to calling GeneratePrivateKey() and then GeneratePublicKey().
    /// \pre <tt>COUNTOF(privateKey) == PrivateKeyLength()</tt>
    /// \pre <tt>COUNTOF(publicKey) == PublicKeyLength()</tt>
    virtual void GenerateKeyPair(RandomNumberGenerator &rng, byte *privateKey, byte *publicKey) const;

    /// \brief Derive agreed value
    /// \param agreedValue a byte buffer for the shared secret
    /// \param privateKey a byte buffer with your private key in this domain
    /// \param otherPublicKey a byte buffer with the other party's public key in this domain
    /// \param validateOtherPublicKey a flag indicating if the other party's public key should be validated
    /// \return true upon success, false in case of failure
    /// \details Agree() derives an agreed value from your private keys and couterparty's public keys.
    /// \details The other party's public key is validated by default. If you have previously validated the
    ///  static public key, use <tt>validateStaticOtherPublicKey=false</tt> to save time.
    /// \pre <tt>COUNTOF(agreedValue) == AgreedValueLength()</tt>
    /// \pre <tt>COUNTOF(privateKey) == PrivateKeyLength()</tt>
    /// \pre <tt>COUNTOF(otherPublicKey) == PublicKeyLength()</tt>
    virtual bool Agree(byte *agreedValue, const byte *privateKey, const byte *otherPublicKey, bool validateOtherPublicKey=true) const =0;
};

/// \brief Interface for domains of authenticated key agreement protocols
/// \details In an authenticated key agreement protocol, each party has two
///  key pairs. The long-lived key pair is called the static key pair,
///  and the short-lived key pair is called the ephemeral key pair.
/// \since Crypto++ 3.0
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE AuthenticatedKeyAgreementDomain : public KeyAgreementAlgorithm
{
public:
    virtual ~AuthenticatedKeyAgreementDomain() {}

    /// \brief Provides the size of the agreed value
    /// \return size of agreed value produced in this domain
    virtual unsigned int AgreedValueLength() const =0;

    /// \brief Provides the size of the static private key
    /// \return size of static private keys in this domain
    virtual unsigned int StaticPrivateKeyLength() const =0;

    /// \brief Provides the size of the static public key
    /// \return size of static public keys in this domain
    virtual unsigned int StaticPublicKeyLength() const =0;

    /// \brief Generate static private key in this domain
    /// \param rng a RandomNumberGenerator derived class
    /// \param privateKey a byte buffer for the generated private key in this domain
    /// \pre <tt>COUNTOF(privateKey) == PrivateStaticKeyLength()</tt>
    virtual void GenerateStaticPrivateKey(RandomNumberGenerator &rng, byte *privateKey) const =0;

    /// \brief Generate a static public key from a private key in this domain
    /// \param rng a RandomNumberGenerator derived class
    /// \param privateKey a byte buffer with the previously generated private key
    /// \param publicKey a byte buffer for the generated public key in this domain
    /// \pre <tt>COUNTOF(publicKey) == PublicStaticKeyLength()</tt>
    virtual void GenerateStaticPublicKey(RandomNumberGenerator &rng, const byte *privateKey, byte *publicKey) const =0;

    /// \brief Generate a static private/public key pair
    /// \param rng a RandomNumberGenerator derived class
    /// \param privateKey a byte buffer for the generated private key in this domain
    /// \param publicKey a byte buffer for the generated public key in this domain
    /// \details GenerateStaticKeyPair() is equivalent to calling GenerateStaticPrivateKey() and then GenerateStaticPublicKey().
    /// \pre <tt>COUNTOF(privateKey) == PrivateStaticKeyLength()</tt>
    /// \pre <tt>COUNTOF(publicKey) == PublicStaticKeyLength()</tt>
    virtual void GenerateStaticKeyPair(RandomNumberGenerator &rng, byte *privateKey, byte *publicKey) const;

    /// \brief Provides the size of ephemeral private key
    /// \return the size of ephemeral private key in this domain
    virtual unsigned int EphemeralPrivateKeyLength() const =0;

    /// \brief Provides the size of ephemeral public key
    /// \return the size of ephemeral public key in this domain
    virtual unsigned int EphemeralPublicKeyLength() const =0;

    /// \brief Generate ephemeral private key
    /// \param rng a RandomNumberGenerator derived class
    /// \param privateKey a byte buffer for the generated private key in this domain
    /// \pre <tt>COUNTOF(privateKey) == PrivateEphemeralKeyLength()</tt>
    virtual void GenerateEphemeralPrivateKey(RandomNumberGenerator &rng, byte *privateKey) const =0;

    /// \brief Generate ephemeral public key
    /// \param rng a RandomNumberGenerator derived class
    /// \param privateKey a byte buffer for the generated private key in this domain
    /// \param publicKey a byte buffer for the generated public key in this domain
    /// \pre <tt>COUNTOF(publicKey) == PublicEphemeralKeyLength()</tt>
    virtual void GenerateEphemeralPublicKey(RandomNumberGenerator &rng, const byte *privateKey, byte *publicKey) const =0;

    /// \brief Generate private/public key pair
    /// \param rng a RandomNumberGenerator derived class
    /// \param privateKey a byte buffer for the generated private key in this domain
    /// \param publicKey a byte buffer for the generated public key in this domain
    /// \details GenerateEphemeralKeyPair() is equivalent to calling GenerateEphemeralPrivateKey() and then GenerateEphemeralPublicKey()
    virtual void GenerateEphemeralKeyPair(RandomNumberGenerator &rng, byte *privateKey, byte *publicKey) const;

    /// \brief Derive agreed value
    /// \param agreedValue a byte buffer for the shared secret
    /// \param staticPrivateKey a byte buffer with your static private key in this domain
    /// \param ephemeralPrivateKey a byte buffer with your ephemeral private key in this domain
    /// \param staticOtherPublicKey a byte buffer with the other party's static public key in this domain
    /// \param ephemeralOtherPublicKey a byte buffer with the other party's ephemeral public key in this domain
    /// \param validateStaticOtherPublicKey a flag indicating if the other party's public key should be validated
    /// \return true upon success, false in case of failure
    /// \details Agree() derives an agreed value from your private keys and couterparty's public keys.
    /// \details The other party's ephemeral public key is validated by default. If you have previously validated
    ///  the static public key, use <tt>validateStaticOtherPublicKey=false</tt> to save time.
    /// \pre <tt>COUNTOF(agreedValue) == AgreedValueLength()</tt>
    /// \pre <tt>COUNTOF(staticPrivateKey) == StaticPrivateKeyLength()</tt>
    /// \pre <tt>COUNTOF(ephemeralPrivateKey) == EphemeralPrivateKeyLength()</tt>
    /// \pre <tt>COUNTOF(staticOtherPublicKey) == StaticPublicKeyLength()</tt>
    /// \pre <tt>COUNTOF(ephemeralOtherPublicKey) == EphemeralPublicKeyLength()</tt>
    virtual bool Agree(byte *agreedValue,
        const byte *staticPrivateKey, const byte *ephemeralPrivateKey,
        const byte *staticOtherPublicKey, const byte *ephemeralOtherPublicKey,
        bool validateStaticOtherPublicKey=true) const =0;
};

// interface for password authenticated key agreement protocols, not implemented yet
#if 0
/// \brief Interface for protocol sessions
/*! The methods should be called in the following order:

    InitializeSession(rng, parameters); // or call initialize method in derived class
    while (true)
    {
        if (OutgoingMessageAvailable())
        {
            length = GetOutgoingMessageLength();
            GetOutgoingMessage(message);
            ; // send outgoing message
        }

        if (LastMessageProcessed())
            break;

        ; // receive incoming message
        ProcessIncomingMessage(message);
    }
    ; // call methods in derived class to obtain result of protocol session
*/
class ProtocolSession
{
public:
    /// Exception thrown when an invalid protocol message is processed
    class ProtocolError : public Exception
    {
    public:
        ProtocolError(ErrorType errorType, const std::string &s) : Exception(errorType, s) {}
    };

    /// Exception thrown when a function is called unexpectedly
    /*! for example calling ProcessIncomingMessage() when ProcessedLastMessage() == true */
    class UnexpectedMethodCall : public Exception
    {
    public:
        UnexpectedMethodCall(const std::string &s) : Exception(OTHER_ERROR, s) {}
    };

    virtual ~ProtocolSession() {}

    ProtocolSession() : m_rng(NULLPTR), m_throwOnProtocolError(true), m_validState(false) {}

    virtual void InitializeSession(RandomNumberGenerator &rng, const NameValuePairs &parameters) =0;

    bool GetThrowOnProtocolError() const {return m_throwOnProtocolError;}
    void SetThrowOnProtocolError(bool throwOnProtocolError) {m_throwOnProtocolError = throwOnProtocolError;}

    bool HasValidState() const {return m_validState;}

    virtual bool OutgoingMessageAvailable() const =0;
    virtual unsigned int GetOutgoingMessageLength() const =0;
    virtual void GetOutgoingMessage(byte *message) =0;

    virtual bool LastMessageProcessed() const =0;
    virtual void ProcessIncomingMessage(const byte *message, unsigned int messageLength) =0;

protected:
    void HandleProtocolError(Exception::ErrorType errorType, const std::string &s) const;
    void CheckAndHandleInvalidState() const;
    void SetValidState(bool valid) {m_validState = valid;}

    RandomNumberGenerator *m_rng;

private:
    bool m_throwOnProtocolError, m_validState;
};

class KeyAgreementSession : public ProtocolSession
{
public:
    virtual ~KeyAgreementSession() {}

    virtual unsigned int GetAgreedValueLength() const =0;
    virtual void GetAgreedValue(byte *agreedValue) const =0;
};

class PasswordAuthenticatedKeyAgreementSession : public KeyAgreementSession
{
public:
    virtual ~PasswordAuthenticatedKeyAgreementSession() {}

    void InitializePasswordAuthenticatedKeyAgreementSession(RandomNumberGenerator &rng,
        const byte *myId, unsigned int myIdLength,
        const byte *counterPartyId, unsigned int counterPartyIdLength,
        const byte *passwordOrVerifier, unsigned int passwordOrVerifierLength);
};

/// \brief Password based key agreement domain
/// \since Crypto++ 3.0
class PasswordAuthenticatedKeyAgreementDomain : public KeyAgreementAlgorithm
{
public:
    virtual ~PasswordAuthenticatedKeyAgreementDomain() {}

    /// return whether the domain parameters stored in this object are valid
    virtual bool ValidateDomainParameters(RandomNumberGenerator &rng) const
        {return GetCryptoParameters().Validate(rng, 2);}

    virtual unsigned int GetPasswordVerifierLength(const byte *password, unsigned int passwordLength) const =0;
    virtual void GeneratePasswordVerifier(RandomNumberGenerator &rng, const byte *userId, unsigned int userIdLength, const byte *password, unsigned int passwordLength, byte *verifier) const =0;

    enum RoleFlags {CLIENT=1, SERVER=2, INITIATOR=4, RESPONDER=8};

    virtual bool IsValidRole(unsigned int role) =0;
    virtual PasswordAuthenticatedKeyAgreementSession * CreateProtocolSession(unsigned int role) const =0;
};
#endif

/// \brief Exception thrown when an ASN.1 BER decoing error is encountered
class CRYPTOPP_DLL BERDecodeErr : public InvalidArgument
{
public:
    BERDecodeErr() : InvalidArgument("BER decode error") {}
    BERDecodeErr(const std::string &s) : InvalidArgument(s) {}
};

/// \brief Interface for encoding and decoding ASN1 objects
/// \details Each class that derives from ASN1Object should provide a serialization format
///  that controls subobject layout. Most of the time the serialization format is
///  taken from a standard, like P1363 or an RFC.
class CRYPTOPP_DLL CRYPTOPP_NO_VTABLE ASN1Object
{
public:
    virtual ~ASN1Object() {}

    /// \brief Decode this object from a BufferedTransformation
    /// \param bt BufferedTransformation object
    /// \details Uses Basic Encoding Rules (BER)
    virtual void BERDecode(BufferedTransformation &bt) =0;

    /// \brief Encode this object into a BufferedTransformation
    /// \param bt BufferedTransformation object
    /// \details Uses Distinguished Encoding Rules (DER)
    virtual void DEREncode(BufferedTransformation &bt) const =0;

    /// \brief Encode this object into a BufferedTransformation
    /// \param bt BufferedTransformation object
    /// \details Uses Basic Encoding Rules (BER).
    /// \details This may be useful if DEREncode() would be too inefficient.
    virtual void BEREncode(BufferedTransformation &bt) const {DEREncode(bt);}
};

/// \brief Specifies the build-time version of the library
/// \return integer representing the build-time version
/// \details LibraryVersion can help detect inadvertent mixing and matching of library
///  versions. When using Crypto++ distributed by a third party, LibraryVersion()
///  records the version of the shared object that was built by the third party.
///  The LibraryVersion() record resides in <tt>cryptlib.o</tt> on Unix compatibles
///  and <tt>cryptlib.obj</tt> on Windows. It does not change when an app links
///  to the library.
/// \details LibraryVersion() is declared with C linkage (<tt>extern "C"</tt>) within the
///  CryptoPP namespace to help programs locate the symbol. If the symbol is present, then
///  the library version is 5.7 or above. If it is missing, then the library version is
///  5.6.5 or below.
/// \details The function could be used as shown below.
/// <pre>  if (LibraryVersion() != HeaderVersion())
///   {
///      cout << "Potential version mismatch" << endl;
///
///      const int lmaj = (LibraryVersion() / 100U) % 10;
///      const int lmin = (LibraryVersion() / 10U) % 10;
///      const int hmaj = (HeaderVersion() / 100U) % 10;
///      const int hmin = (HeaderVersion() / 10U) % 10;
///
///      if(lmaj != hmaj)
///         cout << "Major version mismatch" << endl;
///      else if(lmin != hmin)
///         cout << "Minor version mismatch" << endl;
///   }
/// </pre>
/// \sa HeaderVersion(), <A HREF="http://github.com/weidai11/cryptopp/issues/371">GitHub Issue 371</A>.
/// \since Crypto++ 6.0
extern "C" {
    int LibraryVersion(CRYPTOPP_NOINLINE_DOTDOTDOT);
} // C linkage

/// \brief Specifies the runtime version of the library
/// \return integer representing the runtime version
/// \details HeaderVersion() can help detect inadvertent mixing and matching of library
///  versions. When using Crypto++ distributed by a third party, HeaderVersion()
///  records the version of the headers used by the app when the app is compiled.
/// \details HeaderVersion() is declared with C linkage (<tt>extern "C"</tt>) within the
///  CryptoPP namespace to help programs locate the symbol. If the symbol is present, then
///  the library version is 5.7 or above. If it is missing, then the library version is
///  5.6.5 or below.
/// \details The function could be used as shown below.
/// <pre>  if (LibraryVersion() != HeaderVersion())
///   {
///      cout << "Potential version mismatch" << endl;
///
///      const int lmaj = (LibraryVersion() / 100U) % 10;
///      const int lmin = (LibraryVersion() / 10U) % 10;
///      const int hmaj = (HeaderVersion() / 100U) % 10;
///      const int hmin = (HeaderVersion() / 10U) % 10;
///
///      if(lmaj != hmaj)
///         cout << "Major version mismatch" << endl;
///      else if(lmin != hmin)
///         cout << "Minor version mismatch" << endl;
///   }
/// </pre>
/// \sa LibraryVersion(), <A HREF="http://github.com/weidai11/cryptopp/issues/371">GitHub Issue 371</A>.
/// \since Crypto++ 6.0
extern "C" {
inline int HeaderVersion()
{
    return CRYPTOPP_VERSION;
}
} // C linkage

NAMESPACE_END

#if CRYPTOPP_MSC_VERSION
# pragma warning(pop)
#endif

#endif