From: Mikko Rasa <tdb@tdb.fi>
Date: Thu, 25 Dec 2008 08:57:28 +0000 (+0000)
Subject: Make codecs able to tell their name
X-Git-Tag: strings-1.1~7
X-Git-Url: http://git.tdb.fi/?a=commitdiff_plain;h=271ffa9434c8d9397bb5170cf1ee670c5265ec60;p=libs%2Fcore.git

Make codecs able to tell their name
Support autodetecting the correct codec for a string
---

diff --git a/source/ascii.h b/source/ascii.h
index 7dc8762..ecff0db 100644
--- a/source/ascii.h
+++ b/source/ascii.h
@@ -32,8 +32,10 @@ public:
 		virtual UnicodeChar decode_char(const std::string &, std::string::const_iterator &);
 	};
 
-	Encoder *create_encoder(ErrorMode em=THROW_ON_ERROR) const { return new Encoder(em); }
-	Decoder *create_decoder(ErrorMode em=THROW_ON_ERROR) const { return new Decoder(em); }
+	virtual const char *get_name() const { return "ASCII"; }
+
+	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); }
 };
 
 } // namespace Codecs
diff --git a/source/codec.cpp b/source/codec.cpp
index 865176e..521dfea 100644
--- a/source/codec.cpp
+++ b/source/codec.cpp
@@ -102,9 +102,53 @@ Codec *create_codec(const string &n)
 	if(name=="jisx0201") return new JisX0201;
 	if(name=="jisx0208") return new JisX0208;
 	if(name=="utf8") return new Utf8;
-	if(name=="windows1252") return new Windows1252;
+	if(name=="windows1252" || name=="cp1252") return new Windows1252;
 	throw InvalidParameterValue("Unknown string codec");
 }
 
+Codec *detect_codec(const string &str)
+{
+	bool is_utf8=true;
+	bool is_ascii=true;
+	bool is_latin1=true;
+	unsigned utf8_mb=0;
+
+	for(string::const_iterator i=str.begin(); i!=str.end(); ++i)
+	{
+		unsigned char c=*i;
+		if(c&0x80)
+		{
+			is_ascii=false;
+			if((c&0xC0)==0x80)
+			{
+				if((c&0xE0)==0x80)
+					is_latin1=false;
+				if(utf8_mb)
+					--utf8_mb;
+				else
+					is_utf8=false;
+			}
+			else if((c&0xC0)==0xC0)
+			{
+				if(utf8_mb)
+					is_utf8=false;
+				else
+				{
+					for(utf8_mb=1; (c>>(6-utf8_mb))&1; ++utf8_mb) ;
+				}
+			}
+		}
+	}
+
+	if(is_ascii)
+		return new Ascii;
+	else if(is_utf8)
+		return new Utf8;
+	else if(is_latin1)
+		return new Iso88591;
+	else
+		return new Windows1252;
+}
+
 } // namespace Codecs
 } // namespace Msp
diff --git a/source/codec.h b/source/codec.h
index bb35b0b..d373740 100644
--- a/source/codec.h
+++ b/source/codec.h
@@ -160,6 +160,11 @@ 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.
 	*/
@@ -231,6 +236,12 @@ 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
 
diff --git a/source/iso2022jp.h b/source/iso2022jp.h
index f3c2736..4a452d4 100644
--- a/source/iso2022jp.h
+++ b/source/iso2022jp.h
@@ -50,8 +50,10 @@ public:
 		virtual void switch_mode(Mode);
 	};
 
-	Encoder *create_encoder(ErrorMode em=THROW_ON_ERROR) const { return new Encoder(em); }
-	Decoder *create_decoder(ErrorMode em=THROW_ON_ERROR) const { return new Decoder(em); }
+	virtual const char *get_name() const { return "ISO-2022-JP"; }
+
+	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); }
 };
 
 } // namespace Codecs
diff --git a/source/iso646fi.h b/source/iso646fi.h
index 8f38a29..3047edf 100644
--- a/source/iso646fi.h
+++ b/source/iso646fi.h
@@ -32,8 +32,10 @@ public:
 		virtual UnicodeChar decode_char(const std::string &, std::string::const_iterator &);
 	};
 
-	Encoder *create_encoder(ErrorMode em=THROW_ON_ERROR) const { return new Encoder(em); }
-	Decoder *create_decoder(ErrorMode em=THROW_ON_ERROR) const { return new Decoder(em); }
+	virtual const char *get_name() const { return "ISO-646-FI"; }
+
+	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); }
 };
 
 } // namespace Codecs
diff --git a/source/iso88591.h b/source/iso88591.h
index 1adf605..80c706b 100644
--- a/source/iso88591.h
+++ b/source/iso88591.h
@@ -32,8 +32,10 @@ public:
 		virtual UnicodeChar decode_char(const std::string &, std::string::const_iterator &);
 	};
 
-	Encoder *create_encoder(ErrorMode em=THROW_ON_ERROR) const { return new Encoder(em); }
-	Decoder *create_decoder(ErrorMode em=THROW_ON_ERROR) const { return new Decoder(em); }
+	virtual const char *get_name() const { return "ISO-8859-1"; }
+
+	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); }
 };
 
 } // namespace Codecs
diff --git a/source/iso885915.h b/source/iso885915.h
index da26922..c42ad19 100644
--- a/source/iso885915.h
+++ b/source/iso885915.h
@@ -32,8 +32,10 @@ public:
 		virtual UnicodeChar decode_char(const std::string &, std::string::const_iterator &);
 	};
 
-	Encoder *create_encoder(ErrorMode em=THROW_ON_ERROR) const { return new Encoder(em); }
-	Decoder *create_decoder(ErrorMode em=THROW_ON_ERROR) const { return new Decoder(em); }
+	virtual const char *get_name() const { return "ISO-8859-15"; }
+
+	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); }
 };
 
 } // namespace Codecs
diff --git a/source/jisx0201.h b/source/jisx0201.h
index 1c502f9..180f710 100644
--- a/source/jisx0201.h
+++ b/source/jisx0201.h
@@ -32,8 +32,10 @@ public:
 		virtual UnicodeChar decode_char(const std::string &, std::string::const_iterator &);
 	};
 
-	Encoder *create_encoder(ErrorMode em=THROW_ON_ERROR) const { return new Encoder(em); }
-	Decoder *create_decoder(ErrorMode em=THROW_ON_ERROR) const { return new Decoder(em); }
+	virtual const char *get_name() const { return "JIS X 0201"; }
+
+	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); }
 };
 
 } // namespace Codecs
diff --git a/source/jisx0208.h b/source/jisx0208.h
index 9e9ff9b..27609e6 100644
--- a/source/jisx0208.h
+++ b/source/jisx0208.h
@@ -37,8 +37,10 @@ public:
 		virtual UnicodeChar decode_char(const std::string &, std::string::const_iterator &);
 	};
 
-	Encoder *create_encoder(ErrorMode em=THROW_ON_ERROR) const { return new Encoder(em); }
-	Decoder *create_decoder(ErrorMode em=THROW_ON_ERROR) const { return new Decoder(em); }
+	virtual const char *get_name() const { return "JIS X 0208"; }
+
+	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); }
 };
 
 struct Kuten
diff --git a/source/utf8.h b/source/utf8.h
index 70d44b0..0e5e068 100644
--- a/source/utf8.h
+++ b/source/utf8.h
@@ -32,8 +32,10 @@ public:
 		virtual UnicodeChar decode_char(const std::string &, std::string::const_iterator &);
 	};
 
-	Encoder *create_encoder(ErrorMode em=THROW_ON_ERROR) const { return new Encoder(em); }
-	Decoder *create_decoder(ErrorMode em=THROW_ON_ERROR) const { return new Decoder(em); }
+	virtual const char *get_name() const { return "UTF-8"; }
+
+	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); }
 };
 
 } // namespace Codecs
diff --git a/source/windows1252.h b/source/windows1252.h
index 9f0fe0b..f817a0d 100644
--- a/source/windows1252.h
+++ b/source/windows1252.h
@@ -32,8 +32,10 @@ public:
 		virtual UnicodeChar decode_char(const std::string &, std::string::const_iterator &);
 	};
 
-	Encoder *create_encoder(ErrorMode em=THROW_ON_ERROR) const { return new Encoder(em); }
-	Decoder *create_decoder(ErrorMode em=THROW_ON_ERROR) const { return new Decoder(em); }
+	virtual const char *get_name() const { return "Windows-1252"; }
+
+	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); }
 };
 
 } // namespace Codecs