From aa0f95bcc6181234c4e0473089c74f77f402b275 Mon Sep 17 00:00:00 2001
From: gwdwyy <wuyanyi09@gmail.com>
Date: Mon, 22 Jul 2013 00:08:08 +0800
Subject: [PATCH] update cppcommon

---
 src/cppcommon/encoding.cpp | 111 +++++++++++++++++++++++++++++++++++++
 src/cppcommon/encoding.h   |  39 +++++++++++++
 2 files changed, 150 insertions(+)
 create mode 100644 src/cppcommon/encoding.cpp
 create mode 100644 src/cppcommon/encoding.h

diff --git a/src/cppcommon/encoding.cpp b/src/cppcommon/encoding.cpp
new file mode 100644
index 0000000..b4d4578
--- /dev/null
+++ b/src/cppcommon/encoding.cpp
@@ -0,0 +1,111 @@
+#include "encoding.h"
+
+namespace CPPCOMMON
+{
+	const string& UnicodeEncoding::UTF8ENC = "utf-8";
+	const string& UnicodeEncoding::GBKENC = "gbk";
+	
+	UnicodeEncoding::UnicodeEncoding()
+	{
+		_encVec.push_back(UTF8ENC);
+		_encVec.push_back(GBKENC);
+		_encoding = UTF8ENC;
+	}
+
+	UnicodeEncoding::~UnicodeEncoding()
+	{
+	}
+	
+	bool UnicodeEncoding::setEncoding(const string& enc)
+	{
+		if(!isInVec<string>(_encVec, enc))
+		{
+			return false;
+		}
+		_encoding = enc;
+		return true;
+	}
+	string UnicodeEncoding::encode(const string& str)
+	{
+		if(!isUniStrValid(str))
+		{
+			cout<<__FILE__<<__LINE__<<endl;
+			return "";
+		}
+		if(UTF8ENC == _encoding)
+		{
+			return unicodeToUtf8(str);
+		}
+		else if(GBKENC  == _encoding)
+		{
+			return utf8ToGbk(unicodeToUtf8(str));
+		}
+		cout<<__FILE__<<__LINE__<<endl;
+		return "";
+	}
+	string UnicodeEncoding::decode(const string& str)
+	{
+		if(str.empty())
+		{
+			cout<<__FILE__<<__LINE__<<endl;
+			return "";
+		}
+		string res;
+		if(UTF8ENC == _encoding)
+		{
+			
+			res = utf8ToUnicode(str);
+			if(isUniStrValid(res))
+			{
+				return res;
+			}
+		}
+		else if(GBKENC == _encoding)
+		{
+			res = utf8ToUnicode(gbkToUtf8(str));
+			if(isUniStrValid(res))
+			{
+				return res;
+			}
+		}
+		cout<<__FILE__<<__LINE__<<endl;
+		return "";
+	}
+}
+
+#ifdef ENCODING_UT
+using namespace CPPCOMMON;
+int main()
+{
+	UnicodeEncoding enc;
+	ifstream ifile("testdata/dict.utf8");
+	string line;
+	//enc.setEncoding(UnicodeEncoding::UTF8ENC);
+	//enc.setEncoding(UnicodeEncoding::GBKENC);
+	//while(getline(ifile, line))
+	//{
+	//	cout<<line<<endl;
+	//	cout<<enc.encode(enc.decode(line))<<endl;
+	//	cout<<enc.decode(enc.encode(line))<<endl;
+	//	cout<<enc.decode(line)<<endl;
+	//	cout<<enc.encode(line)<<endl;
+	//}
+	ifile.close();
+	ifile.open("testdata/dict.gbk");
+	enc.setEncoding(UnicodeEncoding::GBKENC);
+	while(getline(ifile, line))
+	{
+		
+		cout<<line<<endl;
+		cout<<line.size()<<endl;
+		cout<<enc.encode(enc.decode(line))<<endl;
+		cout<<enc.decode(enc.encode(line))<<endl;
+		cout<<enc.decode(line)<<endl;
+		cout<<enc.encode(line)<<endl;
+	}
+	ifile.close();
+	
+	return 0;
+}
+
+#endif
diff --git a/src/cppcommon/encoding.h b/src/cppcommon/encoding.h
new file mode 100644
index 0000000..82a3f0c
--- /dev/null
+++ b/src/cppcommon/encoding.h
@@ -0,0 +1,39 @@
+/************************************
+ * file enc : utf8
+ * author   : wuyanyi09@gmail.com
+************************************/
+
+#ifndef CPPCOMMON_ENCODING_H
+#define CPPCOMMON_ENCODING_H
+
+#include "str_functs.h"
+#include "vec_functs.h"
+
+namespace CPPCOMMON
+{
+	using namespace std;
+	class UnicodeEncoding
+	{
+		public:
+			static const string& UTF8ENC;
+			static const string& GBKENC;
+		private:
+			string _encoding;
+			vector<string> _encVec;
+		public:
+			UnicodeEncoding();
+			~UnicodeEncoding();
+		public:
+			bool setEncoding(const string& enc);
+			string encode(const string& str);
+			string decode(const string& str);
+		public:
+			bool isUniStrValid(const string& unistr)
+			{
+				return !(unistr.empty() || unistr.size() % 2);
+			}
+			
+	};
+}
+
+#endif