1 #ifndef MSP_STRINGCODEC_CODEC_H_
2 #define MSP_STRINGCODEC_CODEC_H_
5 #include <msp/core/except.h>
9 namespace StringCodec {
19 An exception thrown for all kinds of problems encountered while encoding or
22 class CodecError: public Exception
25 CodecError(const std::string &w_): Exception(w_) { }
29 Base class for string codecs. Use one of the derived classes or the function
30 create_codec to create a specific codec.
32 Unicode strings are represented as ustrings. An std::string is considered to
33 be an encoded sequence of bytes. A codec is able to determine if an encoded
34 string could be decoded with it.
40 Base class for string encoder.
42 Each codec class should contain an Encoder class derived from this. The
43 encode_char and transliterate functions must be overloaded. Some encoders
44 may find it useful or necessary to implement some other functions too
45 (particularly sync and reset for stateful codecs).
52 Encoder(ErrorMode em): err_mode(em) { }
54 virtual ~Encoder() { }
56 /** Encodes a single unicode character. If the character can't be
57 represented in this encoding, error() should be called. */
58 virtual void encode_char(unichar ch, std::string &buf) = 0;
60 /** Encodes a unicode string. This is equivalent to calling encode_char
61 for each character in the string with the same buffer. */
62 virtual void encode(const ustring &str, std::string &buf);
64 std::string encode(const ustring &);
66 /** Procuces a sequence of bytes that will bring the encoder back to the
68 virtual void sync(std::string &buf) { (void)buf; }
70 /** Resets the encoder to the initial state without producing output. */
71 virtual void reset() { }
74 /** Handles an error depending on the error mode.
76 THROW_ON_ERROR: throws CodecError(msg)
77 IGNORE_ERRORS: does nothing
78 TRANSLITERATE: calls transliterate(ch, buf) */
79 void error(unichar ch, std::string &buf, const std::string &msg);
81 /** Attempts to produce an alternative encoding for a unicode character.
82 Typically this includes dropping accent marks or romanizing letters. */
83 virtual void transliterate(unichar ch, std::string &buf) = 0;
87 Base class for string decoder.
89 Each codec class should contain an Decoder class derived from this.
96 Decoder(ErrorMode em): err_mode(em) { }
98 virtual ~Decoder() { }
100 /** Decodes a single character from a string. The iterator is advanced
101 to the next character. For stateful codecs, -1 may be returned if a
102 state change sequence was decoded but no character followed it. If
103 invalid input is encountered, the error() function should be called and
104 the iterator advanced only if it doesn't throw. */
105 virtual unichar decode_char(const std::string &str, std::string::const_iterator &i) = 0;
107 /** Decodes a string. */
108 virtual void decode(const std::string &str, ustring &buf);
110 ustring decode(const std::string &);
112 /** Resets the decoder to the initial state. */
113 virtual void reset() { }
116 /** Handles an error depending on the error mode. The return value is
117 suitable for returning from decode_char.
119 THROW_ON_ERROR: throws CodecError(msg)
120 IGNORE_ERRORS: returns -1
121 TRANSLITERATE: return 0xFFFD */
122 unichar error(const std::string &msg);
130 /** Returns the name of the encoding handled by this codec. */
131 virtual const char *get_name() const = 0;
133 /** Creates an encoder for this codec. */
134 virtual Encoder *create_encoder(ErrorMode err_mode = THROW_ON_ERROR) const = 0;
136 /** Creates a decoder for this codec. */
137 virtual Decoder *create_decoder(ErrorMode err_mode = THROW_ON_ERROR) const = 0;
139 /** Determines whether the given string can be successfully decoded with
140 this codec. Note that this function returning true does not guarantee that
141 the string was actually encoded with this codec. In particular, many 8-bit
142 encodings are indistinguishable. */
143 virtual bool detect(const std::string &) const;
146 typedef Codec::Encoder Encoder;
147 typedef Codec::Decoder Decoder;
149 /** Convenience function that decodes a string. */
151 ustring decode(const std::string &s)
153 typename C::Decoder dec;
154 return dec.decode(s);
157 /** Convenience function that encodes a string. */
159 std::string encode(const ustring &s)
161 typename C::Encoder enc;
162 return enc.encode(s);
165 /** Convenience function that transcodes a string from one codec to another. */
166 template<class F, class T>
167 std::string transcode(const std::string &s)
169 return encode<T>(decode<F>(s));
172 /** Creates a codec for an encoding by name. The caller is responsible for
173 deleting the codec when it's no longer needed. */
174 Codec *create_codec(const std::string &);
176 /** Automatically detects the encoding of a string and creates a codec for it.
177 The codec must be deleted when it's no longer needed. */
178 Codec *detect_codec(const std::string &);
180 } // namespace StringCodec