+/* $Id$
+
+This file is part of libmspstrings
+Copyright © 2006-2007 Mikko Rasa
+Distributed under the LGPL
+*/
+
#ifndef MSP_STRINGS_CODEC_H_
#define MSP_STRINGS_CODEC_H_
#include <string>
-#include <msp/error.h>
+#include <msp/core/except.h>
namespace Msp {
+namespace Codecs {
+
+typedef int UnicodeChar;
+
+typedef std::basic_string<UnicodeChar> ustring;
+
+enum ErrorMode
+{
+ THROW_ON_ERROR,
+ IGNORE_ERRORS,
+ TRANSLITERATE
+};
/**
An exception thrown for all kinds of problems encountered while encoding or
Base class for string codecs. Use one of the derived classes or the function
create_codec to create a specific codec.
-For the purposes of this class, an std::wstring is considered to contain
-Unicode characters and an std::string is considered to be an encoded sequence
-of bytes. A codec is able to determine if an encoded string could be decoded
-with it.
+Unicode strings are represented as ustrings. An std::string is considered to
+be an encoded sequence of bytes. A codec is able to determine if an encoded
+string could be decoded with it.
*/
-class StringCodec
+class Codec
{
public:
/**
- Base class for string encoder. Each codec class should contain an Encoder
- class derived from this.
+ Base class for string encoder.
+
+ Each codec class should contain an Encoder class derived from this. The
+ encode_char and transliterate functions must be overloaded. Some encoders
+ may find it useful or necessary to implement some other functions too
+ (particularly sync and reset for stateful codecs).
*/
class Encoder
{
- public:
- /**
- Encodes a single character. Derived classes should use the append
- function to put the result into the internal buffer.
- */
- virtual void encode_char(wchar_t) =0;
-
- /**
- Encodes a string.
- */
- virtual void encode(const std::wstring &s)
- { for(std::wstring::const_iterator i=s.begin(); i!=s.end(); ++i) encode_char(*i); }
-
- /**
- Brings the encoder back to its initial state. This allows the encoded
- sequence to be extracted or flushed without loss of integrity.
- */
- virtual void sync() { }
-
- /**
- Returns a reference to the encoded sequence. Call sync() first to make
- sure it's a valid encoded string by itself.
- */
- const std::string &get() const { return buffer_; }
-
- /**
- Returns the number of bytes in the output buffer.
- */
- unsigned size() const { return buffer_.size(); }
-
- /**
- Clears the encoded sequence. Encoder state is left intact.
- */
- void flush() { buffer_.clear(); }
+ protected:
+ ErrorMode err_mode;
+ Encoder(ErrorMode em): err_mode(em) { }
+ public:
virtual ~Encoder() { }
+
+ /** Encodes a single unicode character. If the character can't be
+ represented in this encoding, error() should be called. */
+ virtual void encode_char(UnicodeChar ch, std::string &buf) = 0;
+
+ /** Encodes a unicode string. This is equivalent to calling encode_char
+ for each character in the string with the same buffer. */
+ virtual void encode(const ustring &str, std::string &buf);
+
+ std::string encode(const ustring &);
+
+ /** Procuces a sequence of bytes that will bring the encoder back to the
+ initial state. */
+ virtual void sync(std::string &buf) { (void)buf; }
+
+ /** Resets the encoder to the initial state without producing output. */
+ virtual void reset() { }
+
protected:
- Encoder() { }
- void append(char c) { buffer_+=c; }
- void append(const char *s, unsigned l) { buffer_.append(s, l); }
- void append(const std::string &s) { buffer_+=s; }
- private:
- std::string buffer_;
+ /** Handles an error depending on the error mode.
+
+ THROW_ON_ERROR: throws CodecError(msg)
+ IGNORE_ERRORS: does nothing
+ TRANSLITERATE: calls transliterate(ch, buf) */
+ void error(UnicodeChar ch, std::string &buf, const std::string &msg);
+
+ /** Attempts to produce an alternative encoding for a unicode character.
+ Typically this includes dropping accent marks or romanizing letters. */
+ virtual void transliterate(UnicodeChar ch, std::string &buf) = 0;
};
/**
- Base class for string decoder. Each codec class should contain an Decoder
- class derived from this.
+ Base class for string decoder.
+
+ Each codec class should contain an Decoder class derived from this.
*/
class Decoder
{
+ protected:
+ ErrorMode err_mode;
+
+ Decoder(ErrorMode em): err_mode(em) { }
public:
- virtual void decode_char(const std::string &, std::string::const_iterator &) =0;
- virtual void decode(const std::string &s)
- { for(std::string::const_iterator i=s.begin(); i!=s.end(); ) decode_char(s, i); }
-
- /**
- Ensures that all input has been processed. If this is not the case any
- buffers are cleared and an exception is thrown,
- */
- virtual void sync() { }
-
- /**
- Resets the decoder, clearing a possibly erroneus state. Does not flush
- the internal buffer.
- */
+ virtual ~Decoder() { }
+
+ /** Decodes a single character from a string. The iterator is advanced
+ to the next character. For stateful codecs, -1 may be returned if a
+ state change sequence was decoded but no character followed it. If
+ invalid input is encountered, the error() function should be called and
+ the iterator advanced only if it doesn't throw. */
+ virtual UnicodeChar decode_char(const std::string &str, std::string::const_iterator &i) = 0;
+
+ /** Decodes a string. */
+ virtual void decode(const std::string &str, ustring &buf);
+
+ ustring decode(const std::string &);
+
+ /** Resets the decoder to the initial state. */
virtual void reset() { }
- const std::wstring &get() const { return buffer_; }
- unsigned size() const { return buffer_.size(); }
- void flush() { buffer_.clear(); }
- virtual ~Decoder() { }
protected:
- Decoder() { }
- void append(wchar_t c) { buffer_+=c; }
- void append(const std::wstring &s) { buffer_+=s; }
- private:
- std::wstring buffer_;
+ /** Handles an error depending on the error mode. The return value is
+ suitable for returning from decode_char.
+
+ THROW_ON_ERROR: throws CodecError(msg)
+ IGNORE_ERRORS: returns -1
+ TRANSLITERATE: return 0xFFFE */
+ UnicodeChar error(const std::string &msg);
};
- virtual Encoder *create_encoder() const =0;
- virtual Decoder *create_decoder() const =0;
- virtual bool detect(const std::string &) const;
- virtual ~StringCodec() { }
protected:
- StringCodec() { }
+ Codec() { }
+public:
+ virtual ~Codec() { }
+
+ /** Returns the name of the encoding handled by this codec. */
+ virtual const char *get_name() const = 0;
+
+ /** Creates an encoder for this codec. */
+ virtual Encoder *create_encoder(ErrorMode err_mode = THROW_ON_ERROR) const = 0;
+
+ /** Creates a decoder for this codec. */
+ virtual Decoder *create_decoder(ErrorMode err_mode = THROW_ON_ERROR) const = 0;
+
+ /** Determines whether the given string can be successfully decoded with
+ this codec. Note that this function returning true does not guarantee that
+ the string was actually encoded with this codec. In particular, many 8-bit
+ encodings are indistinguishable. */
+ virtual bool detect(const std::string &) const;
};
-/**
-Convenience function that decodes a string using the given codec.
-*/
+typedef Codec::Encoder Encoder;
+typedef Codec::Decoder Decoder;
+
+/** Convenience function that decodes a string. */
template<class C>
-std::wstring decode(const std::string &s)
+ustring decode(const std::string &s)
{
typename C::Decoder dec;
- dec.decode(s);
- dec.sync();
- return dec.get();
+ ustring result;
+ dec.decode(s, result);
+ return result;
}
+/** Convenience function that encodes a string. */
template<class C>
-std::string encode(const std::wstring &s)
+std::string encode(const ustring &s)
{
typename C::Encoder enc;
- enc.encode(s);
- enc.sync();
- return enc.get();
+ std::string result;
+ enc.encode(s, result);
+ enc.sync(result);
+ return result;
}
+/** Convenience function that transcodes a string from one codec to another. */
template<class F, class T>
std::string transcode(const std::string &s)
{
typename F::Decoder from;
typename T::Encoder to;
- from.decode(s);
- from.sync();
- to.encode(from.get());
- to.sync();
- return to.get();
+ ustring temp;
+ from.decode(s, temp);
+ std::string result;
+ to.encode(temp, result);
+ to.sync(result);
+ return result;
}
-StringCodec *create_codec(const std::string &);
+/** Creates a codec for an encoding by name. The caller is responsible for
+deleting the codec when it's no longer needed. */
+Codec *create_codec(const std::string &);
+
+/** Automatically detects the encoding of a string and creates a codec for it.
+The codec must be deleted when it's no longer needed. */
+Codec *detect_codec(const std::string &);
+} // namespace Codecs
} // namespace Msp
#endif