2012-06-21 18:19:19 +02:00
|
|
|
//
|
|
|
|
// UTF32Encoding.h
|
|
|
|
//
|
|
|
|
// Library: Foundation
|
|
|
|
// Package: Text
|
|
|
|
// Module: UTF32Encoding
|
|
|
|
//
|
|
|
|
// Definition of the UTF32Encoding class.
|
|
|
|
//
|
|
|
|
// Copyright (c) 2004-2007, Applied Informatics Software Engineering GmbH.
|
|
|
|
// and Contributors.
|
|
|
|
//
|
2014-05-04 21:02:42 +02:00
|
|
|
// SPDX-License-Identifier: BSL-1.0
|
2012-06-21 18:19:19 +02:00
|
|
|
//
|
|
|
|
|
|
|
|
|
|
|
|
#ifndef Foundation_UTF32Encoding_INCLUDED
|
|
|
|
#define Foundation_UTF32Encoding_INCLUDED
|
|
|
|
|
|
|
|
|
|
|
|
#include "Poco/Foundation.h"
|
|
|
|
#include "Poco/TextEncoding.h"
|
|
|
|
|
|
|
|
|
|
|
|
namespace Poco {
|
|
|
|
|
|
|
|
|
|
|
|
class Foundation_API UTF32Encoding: public TextEncoding
|
2012-06-21 18:28:58 +02:00
|
|
|
/// UTF-32 text encoding, as defined in RFC 2781.
|
2012-06-21 18:19:19 +02:00
|
|
|
///
|
2012-06-21 18:28:58 +02:00
|
|
|
/// When converting from UTF-32 to Unicode, surrogates are
|
2012-06-21 18:19:19 +02:00
|
|
|
/// reported as they are - in other words, surrogate pairs
|
2022-07-07 11:18:20 +02:00
|
|
|
/// are not combined into one Unicode character.
|
2012-06-21 18:19:19 +02:00
|
|
|
{
|
|
|
|
public:
|
|
|
|
enum ByteOrderType
|
|
|
|
{
|
|
|
|
BIG_ENDIAN_BYTE_ORDER,
|
|
|
|
LITTLE_ENDIAN_BYTE_ORDER,
|
|
|
|
NATIVE_BYTE_ORDER
|
|
|
|
};
|
2022-07-07 11:18:20 +02:00
|
|
|
|
2012-06-21 18:19:19 +02:00
|
|
|
UTF32Encoding(ByteOrderType byteOrder = NATIVE_BYTE_ORDER);
|
|
|
|
/// Creates and initializes the encoding for the given byte order.
|
2022-07-07 11:18:20 +02:00
|
|
|
|
2012-06-21 18:19:19 +02:00
|
|
|
UTF32Encoding(int byteOrderMark);
|
|
|
|
/// Creates and initializes the encoding for the byte-order
|
|
|
|
/// indicated by the given byte-order mark, which is the Unicode
|
|
|
|
/// character 0xFEFF.
|
2022-07-07 11:18:20 +02:00
|
|
|
|
2012-06-21 18:19:19 +02:00
|
|
|
~UTF32Encoding();
|
2022-07-07 11:18:20 +02:00
|
|
|
|
2012-06-21 18:19:19 +02:00
|
|
|
ByteOrderType getByteOrder() const;
|
|
|
|
/// Returns the byte-order currently in use.
|
2022-07-07 11:18:20 +02:00
|
|
|
|
2012-06-21 18:19:19 +02:00
|
|
|
void setByteOrder(ByteOrderType byteOrder);
|
|
|
|
/// Sets the byte order.
|
2022-07-07 11:18:20 +02:00
|
|
|
|
2012-06-21 18:19:19 +02:00
|
|
|
void setByteOrder(int byteOrderMark);
|
|
|
|
/// Sets the byte order according to the given
|
|
|
|
/// byte order mark, which is the Unicode
|
|
|
|
/// character 0xFEFF.
|
2022-07-07 11:18:20 +02:00
|
|
|
|
2012-06-21 18:19:19 +02:00
|
|
|
const char* canonicalName() const;
|
|
|
|
bool isA(const std::string& encodingName) const;
|
|
|
|
const CharacterMap& characterMap() const;
|
|
|
|
int convert(const unsigned char* bytes) const;
|
|
|
|
int convert(int ch, unsigned char* bytes, int length) const;
|
|
|
|
int queryConvert(const unsigned char* bytes, int length) const;
|
|
|
|
int sequenceLength(const unsigned char* bytes, int length) const;
|
2022-07-07 11:18:20 +02:00
|
|
|
|
2023-12-04 15:27:06 +01:00
|
|
|
protected:
|
|
|
|
static int safeToInt(Poco::UInt32 value)
|
|
|
|
{
|
|
|
|
if (value <= 0x10FFFF)
|
|
|
|
return static_cast<int>(value);
|
|
|
|
else
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2012-06-21 18:19:19 +02:00
|
|
|
private:
|
|
|
|
bool _flipBytes;
|
|
|
|
static const char* _names[];
|
|
|
|
static const CharacterMap _charMap;
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
} // namespace Poco
|
|
|
|
|
|
|
|
|
|
|
|
#endif // Foundation_UTF32Encoding_INCLUDED
|