Allow error mode to be specified when creating the codec object

diff --git a/source/stringcodec/ascii.h b/source/stringcodec/ascii.h
index 163448b60875d214c9c8d936e81b779f2d3fab96..b052727b4cd33c18ef816fbff53519645b869acc 100644 (file)
--- a/source/stringcodec/ascii.h
+++ b/source/stringcodec/ascii.h
@@ -6,7 +6,7 @@
 namespace Msp {
 namespace StringCodec {
 
-class Ascii: public Codec
+class Ascii: public StandardCodec<Ascii>
 {
 public:
        class Encoder: public Codec::Encoder
@@ -27,10 +27,9 @@ public:
                virtual unichar decode_char(const std::string &, std::string::const_iterator &);
        };
 
-       virtual const char *get_name() const { return "ASCII"; }
+       Ascii(ErrorMode em = THROW_ON_ERROR): StandardCodec<Ascii>(em) { }
 
-       virtual Encoder *create_encoder(ErrorMode em = THROW_ON_ERROR) const { return new Encoder(em); }
-       virtual Decoder *create_decoder(ErrorMode em = THROW_ON_ERROR) const { return new Decoder(em); }
+       virtual const char *get_name() const { return "ASCII"; }
 };
 
 } // namespace StringCodec

diff --git a/source/stringcodec/codec.cpp b/source/stringcodec/codec.cpp
index b1c05c50292295e573f307497a175ac78797a08a..d1fc1dfb0730adbfc94e5c2c7d50335172596cf8 100644 (file)
--- a/source/stringcodec/codec.cpp
+++ b/source/stringcodec/codec.cpp
@@ -63,23 +63,40 @@ ustring Codec::Decoder::decode(const string &str)
 Codec *create_codec(const string &n)
 {
        string name;
-       for(string::const_iterator i=n.begin(); i!=n.end(); ++i)
+       string::const_iterator i;
+       for(i=n.begin(); i!=n.end(); ++i)
        {
-               if(isupper(*i))
+               if(*i==':')
+                       break;
+               else if(isupper(*i))
                        name += tolower(*i);
                else if(islower(*i) || isdigit(*i))
                        name += *i;
        }
 
-       if(name=="ascii") return new Ascii;
-       if(name=="iso2022jp") return new Iso2022Jp;
-       if(name=="iso646fi") return new Iso646Fi;
-       if(name=="iso88591" || name=="latin1") return new Iso88591;
-       if(name=="iso885915" || name=="latin9") return new Iso885915;
-       if(name=="jisx0201") return new JisX0201;
-       if(name=="jisx0208") return new JisX0208;
-       if(name=="utf8") return new Utf8;
-       if(name=="windows1252" || name=="cp1252") return new Windows1252;
+       ErrorMode em = THROW_ON_ERROR;
+       if(i!=n.end() && *i==':')
+       {
+               string em_str(i+1, n.end());
+               if(em_str=="throw")
+                       em = THROW_ON_ERROR;
+               else if(em_str=="ignore")
+                       em = IGNORE_ERRORS;
+               else if(em_str=="trans" || em_str=="transliterate")
+                       em = TRANSLITERATE;
+               else
+                       throw invalid_argument("invalid error mode");
+       }
+
+       if(name=="ascii") return new Ascii(em);
+       if(name=="iso2022jp") return new Iso2022Jp(em);
+       if(name=="iso646fi") return new Iso646Fi(em);
+       if(name=="iso88591" || name=="latin1") return new Iso88591(em);
+       if(name=="iso885915" || name=="latin9") return new Iso885915(em);
+       if(name=="jisx0201") return new JisX0201(em);
+       if(name=="jisx0208") return new JisX0208(em);
+       if(name=="utf8") return new Utf8(em);
+       if(name=="windows1252" || name=="cp1252") return new Windows1252(em);
        throw invalid_argument("unknown string codec");
 }
 

diff --git a/source/stringcodec/codec.h b/source/stringcodec/codec.h
index 00f03b0cc4b4c4f2f4fcd133f2d3fa67e9fac5fb..c70e052dd1b35e819a591327777f8eec4dbdc9bd 100644 (file)
--- a/source/stringcodec/codec.h
+++ b/source/stringcodec/codec.h
@@ -10,6 +10,7 @@ namespace StringCodec {
 
 enum ErrorMode
 {
+       DEFAULT,
        THROW_ON_ERROR,
        IGNORE_ERRORS,
        TRANSLITERATE
@@ -138,10 +139,10 @@ public:
        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;
+       virtual Encoder *create_encoder(ErrorMode err_mode = DEFAULT) const = 0;
 
        /** Creates a decoder for this codec. */
-       virtual Decoder *create_decoder(ErrorMode err_mode = THROW_ON_ERROR) const = 0;
+       virtual Decoder *create_decoder(ErrorMode err_mode = DEFAULT) const = 0;
 
        /** Determines whether the given string can be successfully decoded with
        this codec.  Note that this function returning true does not guarantee that
@@ -153,6 +154,28 @@ public:
 typedef Codec::Encoder Encoder;
 typedef Codec::Decoder Decoder;
 
+
+/**
+A helper class to provide some common functionality.
+*/
+template<typename C>
+class StandardCodec: public Codec
+{
+private:
+       ErrorMode err_mode;
+
+protected:
+       StandardCodec(ErrorMode em): err_mode(em==DEFAULT ? THROW_ON_ERROR : em) { }
+
+public:
+       virtual Encoder *create_encoder(ErrorMode em = DEFAULT) const
+       { return new typename C::Encoder(em==DEFAULT ? err_mode : em); }
+
+       virtual Decoder *create_decoder(ErrorMode em = DEFAULT) const
+       { return new typename C::Decoder(em==DEFAULT ? err_mode : em); }
+};
+
+
 /** Convenience function that decodes a string. */
 template<class C>
 ustring decode(const std::string &s)

diff --git a/source/stringcodec/iso2022jp.h b/source/stringcodec/iso2022jp.h
index 91f0166e19a5e0e6f533304c46dba6a5ee679e78..a3f425a95d9c9dd1ddf4a215cdd7454aa3d838a9 100644 (file)
--- a/source/stringcodec/iso2022jp.h
+++ b/source/stringcodec/iso2022jp.h
@@ -6,7 +6,7 @@
 namespace Msp {
 namespace StringCodec {
 
-class Iso2022Jp: public Codec
+class Iso2022Jp: public StandardCodec<Iso2022Jp>
 {
 public:
        enum Mode
@@ -47,10 +47,9 @@ public:
                void switch_mode(Mode);
        };
 
-       virtual const char *get_name() const { return "ISO-2022-JP"; }
+       Iso2022Jp(ErrorMode em = THROW_ON_ERROR): StandardCodec<Iso2022Jp>(em) { }
 
-       virtual Encoder *create_encoder(ErrorMode em = THROW_ON_ERROR) const { return new Encoder(em); }
-       virtual Decoder *create_decoder(ErrorMode em = THROW_ON_ERROR) const { return new Decoder(em); }
+       virtual const char *get_name() const { return "ISO-2022-JP"; }
 };
 
 } // namespace StringCodec

diff --git a/source/stringcodec/iso646fi.h b/source/stringcodec/iso646fi.h
index f61a49febc72bb8d6b36a7ce265e763a6cf574f1..7ded23cc734f2d89cedce5f0259fbeca7fca109d 100644 (file)
--- a/source/stringcodec/iso646fi.h
+++ b/source/stringcodec/iso646fi.h
@@ -6,7 +6,7 @@
 namespace Msp {
 namespace StringCodec {
 
-class Iso646Fi: public Codec
+class Iso646Fi: public StandardCodec<Iso646Fi>
 {
 public:
        class Encoder: public Codec::Encoder
@@ -27,10 +27,9 @@ public:
                virtual unichar decode_char(const std::string &, std::string::const_iterator &);
        };
 
-       virtual const char *get_name() const { return "ISO-646-FI"; }
+       Iso646Fi(ErrorMode em = THROW_ON_ERROR): StandardCodec<Iso646Fi>(em) { }
 
-       virtual Encoder *create_encoder(ErrorMode em = THROW_ON_ERROR) const { return new Encoder(em); }
-       virtual Decoder *create_decoder(ErrorMode em = THROW_ON_ERROR) const { return new Decoder(em); }
+       virtual const char *get_name() const { return "ISO-646-FI"; }
 };
 
 } // namespace StringCodec

diff --git a/source/stringcodec/iso88591.h b/source/stringcodec/iso88591.h
index 806bdb1de7d1420c701d5e99dd6ae4d08447e0c1..5c59fae5f9eccbdce443765e620ea345271854fd 100644 (file)
--- a/source/stringcodec/iso88591.h
+++ b/source/stringcodec/iso88591.h
@@ -6,7 +6,7 @@
 namespace Msp {
 namespace StringCodec {
 
-class Iso88591: public Codec
+class Iso88591: public StandardCodec<Iso88591>
 {
 public:
        class Encoder: public Codec::Encoder
@@ -27,10 +27,9 @@ public:
                virtual unichar decode_char(const std::string &, std::string::const_iterator &);
        };
 
-       virtual const char *get_name() const { return "ISO-8859-1"; }
+       Iso88591(ErrorMode em = THROW_ON_ERROR): StandardCodec<Iso88591>(em) { }
 
-       virtual Encoder *create_encoder(ErrorMode em = THROW_ON_ERROR) const { return new Encoder(em); }
-       virtual Decoder *create_decoder(ErrorMode em = THROW_ON_ERROR) const { return new Decoder(em); }
+       virtual const char *get_name() const { return "ISO-8859-1"; }
 };
 
 } // namespace StringCodec

diff --git a/source/stringcodec/iso885915.h b/source/stringcodec/iso885915.h
index f52fdac9e802a8c9e5d386d43316086e59751b36..5d6573043ce61a3950c0eaeeda546c7dd632c30a 100644 (file)
--- a/source/stringcodec/iso885915.h
+++ b/source/stringcodec/iso885915.h
@@ -6,7 +6,7 @@
 namespace Msp {
 namespace StringCodec {
 
-class Iso885915: public Codec
+class Iso885915: public StandardCodec<Iso885915>
 {
 public:
        class Encoder: public Codec::Encoder
@@ -27,10 +27,9 @@ public:
                virtual unichar decode_char(const std::string &, std::string::const_iterator &);
        };
 
-       virtual const char *get_name() const { return "ISO-8859-15"; }
+       Iso885915(ErrorMode em = THROW_ON_ERROR): StandardCodec<Iso885915>(em) { }
 
-       virtual Encoder *create_encoder(ErrorMode em = THROW_ON_ERROR) const { return new Encoder(em); }
-       virtual Decoder *create_decoder(ErrorMode em = THROW_ON_ERROR) const { return new Decoder(em); }
+       virtual const char *get_name() const { return "ISO-8859-15"; }
 };
 
 } // namespace StringCodec

diff --git a/source/stringcodec/jisx0201.h b/source/stringcodec/jisx0201.h
index 7a7223ee7cd3aed7f1b9271a45f74517899dcab4..5521d01e8b002bf6a8303900b7a0b0eac8eb9136 100644 (file)
--- a/source/stringcodec/jisx0201.h
+++ b/source/stringcodec/jisx0201.h
@@ -6,7 +6,7 @@
 namespace Msp {
 namespace StringCodec {
 
-class JisX0201: public Codec
+class JisX0201: public StandardCodec<JisX0201>
 {
 public:
        class Encoder: public Codec::Encoder
@@ -27,10 +27,9 @@ public:
                virtual unichar decode_char(const std::string &, std::string::const_iterator &);
        };
 
-       virtual const char *get_name() const { return "JIS X 0201"; }
+       JisX0201(ErrorMode em = THROW_ON_ERROR): StandardCodec<JisX0201>(em) { }
 
-       virtual Encoder *create_encoder(ErrorMode em = THROW_ON_ERROR) const { return new Encoder(em); }
-       virtual Decoder *create_decoder(ErrorMode em = THROW_ON_ERROR) const { return new Decoder(em); }
+       virtual const char *get_name() const { return "JIS X 0201"; }
 };
 
 } // namespace StringCodec

diff --git a/source/stringcodec/jisx0208.h b/source/stringcodec/jisx0208.h
index 81b39f2104ca5c638fd967b1baaa91abdb03a8b0..c59e44f5481ec32e234f41a6b12b306e248b0bf8 100644 (file)
--- a/source/stringcodec/jisx0208.h
+++ b/source/stringcodec/jisx0208.h
@@ -11,7 +11,7 @@ Codec for the JIS X 0208 encoding.  This is not particularly useful as a
 stand-alone codec, due to lack of a linefeed character among other things,
 but is included as part of some other encodings.
 */
-class JisX0208: public Codec
+class JisX0208: public StandardCodec<JisX0208>
 {
 public:
        class Encoder: public Codec::Encoder
@@ -32,10 +32,9 @@ public:
                virtual unichar decode_char(const std::string &, std::string::const_iterator &);
        };
 
-       virtual const char *get_name() const { return "JIS X 0208"; }
+       JisX0208(ErrorMode em = THROW_ON_ERROR): StandardCodec<JisX0208>(em) { }
 
-       virtual Encoder *create_encoder(ErrorMode em = THROW_ON_ERROR) const { return new Encoder(em); }
-       virtual Decoder *create_decoder(ErrorMode em = THROW_ON_ERROR) const { return new Decoder(em); }
+       virtual const char *get_name() const { return "JIS X 0208"; }
 };
 
 

diff --git a/source/stringcodec/utf8.h b/source/stringcodec/utf8.h
index cb4924aeeed36b0960668e08f315c70bbc86663e..fcf23eff35e55e0a03532f9e9b541ea2ac4a6414 100644 (file)
--- a/source/stringcodec/utf8.h
+++ b/source/stringcodec/utf8.h
@@ -6,7 +6,7 @@
 namespace Msp {
 namespace StringCodec {
 
-class Utf8: public Codec
+class Utf8: public StandardCodec<Utf8>
 {
 public:
        class Encoder: public Codec::Encoder
@@ -27,10 +27,9 @@ public:
                virtual unichar decode_char(const std::string &, std::string::const_iterator &);
        };
 
-       virtual const char *get_name() const { return "UTF-8"; }
+       Utf8(ErrorMode em = THROW_ON_ERROR): StandardCodec<Utf8>(em) { }
 
-       virtual Encoder *create_encoder(ErrorMode em = THROW_ON_ERROR) const { return new Encoder(em); }
-       virtual Decoder *create_decoder(ErrorMode em = THROW_ON_ERROR) const { return new Decoder(em); }
+       virtual const char *get_name() const { return "UTF-8"; }
 };
 
 } // namespace StringCodec

diff --git a/source/stringcodec/windows1252.h b/source/stringcodec/windows1252.h
index b10071cad07dcbf703b6fc564b6aae239336e0c8..44d97c9f14931972c0dec03ba91706d9d7d3d542 100644 (file)
--- a/source/stringcodec/windows1252.h
+++ b/source/stringcodec/windows1252.h
@@ -6,7 +6,7 @@
 namespace Msp {
 namespace StringCodec {
 
-class Windows1252: public Codec
+class Windows1252: public StandardCodec<Windows1252>
 {
 public:
        class Encoder: public Codec::Encoder
@@ -27,10 +27,9 @@ public:
                virtual unichar decode_char(const std::string &, std::string::const_iterator &);
        };
 
-       virtual const char *get_name() const { return "Windows-1252"; }
+       Windows1252(ErrorMode em = THROW_ON_ERROR): StandardCodec<Windows1252>(em) { }
 
-       virtual Encoder *create_encoder(ErrorMode em = THROW_ON_ERROR) const { return new Encoder(em); }
-       virtual Decoder *create_decoder(ErrorMode em = THROW_ON_ERROR) const { return new Decoder(em); }
+       virtual const char *get_name() const { return "Windows-1252"; }
 };
 
 } // namespace StringCodec

diff --git a/transcode.cpp b/transcode.cpp
index e8ba3eb5e05ca2b9a322288156f69fc19813ca4b..47a21848991d939637209a7c194e9c3ad4fa7460 100644 (file)
--- a/transcode.cpp
+++ b/transcode.cpp
@@ -17,8 +17,8 @@ int main(int argc, char **argv)
        StringCodec::Codec *from = StringCodec::create_codec(argv[1]);
        StringCodec::Codec *to = StringCodec::create_codec(argv[2]);
 
-       StringCodec::Decoder *from_dec = from->create_decoder(StringCodec::TRANSLITERATE);
-       StringCodec::Encoder *to_enc = to->create_encoder(StringCodec::TRANSLITERATE);
+       StringCodec::Decoder *from_dec = from->create_decoder();
+       StringCodec::Encoder *to_enc = to->create_encoder();
 
        string line;
        while(getline(cin, line))