X-Git-Url: http://git.tdb.fi/?a=blobdiff_plain;f=source%2Fstringcodec%2Fcodec.h;h=00f03b0cc4b4c4f2f4fcd133f2d3fa67e9fac5fb;hb=94ee3a1040f67d9de2e92fc34049642b08d65b3e;hp=53775206ea1915026b79e5ed75d02b227d74dca0;hpb=967785734be5c3fc6f75da122c2d93ebbb338271;p=libs%2Fcore.git

diff --git a/source/stringcodec/codec.h b/source/stringcodec/codec.h
index 5377520..00f03b0 100644
--- a/source/stringcodec/codec.h
+++ b/source/stringcodec/codec.h
@@ -1,15 +1,12 @@
-#ifndef MSP_STRINGS_CODEC_H_
-#define MSP_STRINGS_CODEC_H_
+#ifndef MSP_STRINGCODEC_CODEC_H_
+#define MSP_STRINGCODEC_CODEC_H_
 
 #include <string>
-#include <msp/core/except.h>
+#include "except.h"
+#include "ustring.h"
 
 namespace Msp {
-namespace Codecs {
-
-typedef int UnicodeChar;
-
-typedef std::basic_string<UnicodeChar> ustring;
+namespace StringCodec {
 
 enum ErrorMode
 {
@@ -18,15 +15,6 @@ enum ErrorMode
 	TRANSLITERATE
 };
 
-/**
-An exception thrown for all kinds of problems encountered while encoding or
-decoding strings.
-*/
-class CodecError: public Exception
-{
-public:
-	CodecError(const std::string &w_): Exception(w_) { }
-};
 
 /**
 Base class for string codecs.  Use one of the derived classes or the function
@@ -58,7 +46,7 @@ public:
 
 		/** 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;
+		virtual void encode_char(unichar 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. */
@@ -76,14 +64,21 @@ public:
 	protected:
 		/** Handles an error depending on the error mode.
 
-		THROW_ON_ERROR: throws CodecError(msg)
+		THROW_ON_ERROR: throws err
 		IGNORE_ERRORS: does nothing
 		TRANSLITERATE: calls transliterate(ch, buf) */
-		void error(UnicodeChar ch, std::string &buf, const std::string &msg);
+		template<typename E>
+		void error(unichar ch, std::string &buf, const E &err)
+		{
+			if(err_mode==TRANSLITERATE)
+				transliterate(ch, buf);
+			else if(err_mode!=IGNORE_ERRORS)
+				throw err;
+		}
 
 		/** 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;
+		virtual void transliterate(unichar ch, std::string &buf) = 0;
 	};
 
 	/**
@@ -105,7 +100,7 @@ public:
 		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;
+		virtual unichar decode_char(const std::string &str, std::string::const_iterator &i) = 0;
 
 		/** Decodes a string. */
 		virtual void decode(const std::string &str, ustring &buf);
@@ -119,10 +114,19 @@ public:
 		/** Handles an error depending on the error mode.  The return value is
 		suitable for returning from decode_char.
 		
-		THROW_ON_ERROR: throws CodecError(msg)
+		THROW_ON_ERROR: throws err
 		IGNORE_ERRORS: returns -1
-		TRANSLITERATE: return 0xFFFE */
-		UnicodeChar error(const std::string &msg);
+		TRANSLITERATE: return 0xFFFD */
+		template<typename E>
+		unichar error(const E &err)
+		{
+			if(err_mode==TRANSLITERATE)
+				return 0xFFFD;
+			else if(err_mode==IGNORE_ERRORS)
+				return -1;
+			else
+				throw err;
+		}
 	};
 
 protected:
@@ -154,9 +158,7 @@ template<class C>
 ustring decode(const std::string &s)
 {
 	typename C::Decoder dec;
-	ustring result;
-	dec.decode(s, result);
-	return result;
+	return dec.decode(s);
 }
 
 /** Convenience function that encodes a string. */
@@ -164,24 +166,14 @@ template<class C>
 std::string encode(const ustring &s)
 {
 	typename C::Encoder enc;
-	std::string result;
-	enc.encode(s, result);
-	enc.sync(result);
-	return result;
+	return enc.encode(s);
 }
 
 /** 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;
-	ustring temp;
-	from.decode(s, temp);
-	std::string result;
-	to.encode(temp, result);
-	to.sync(result);
-	return result;
+	return encode<T>(decode<F>(s));
 }
 
 /** Creates a codec for an encoding by name.  The caller is responsible for
@@ -192,7 +184,7 @@ Codec *create_codec(const std::string &);
 The codec must be deleted when it's no longer needed. */
 Codec *detect_codec(const std::string &);
 
-} // namespace Codecs
+} // namespace StringCodec
 } // namespace Msp
 
 #endif