From e443c8a405d4b10febcd0d22d7ab4b7ce34fbd6d Mon Sep 17 00:00:00 2001
From: Edouard DUPIN <yui.heero@gmail.com>
Date: Thu, 26 Sep 2013 21:12:39 +0200
Subject: [PATCH] [DEV] add capacity at unichar and corect append of string

---
 etk/UString.cpp | 52 +++++++++++++++++-----------------
 etk/UString.h   |  4 +--
 etk/UniChar.cpp | 74 ++++++++++++++++++++++++++++++++++++++++++++-----
 etk/UniChar.h   |  4 ++-
 4 files changed, 99 insertions(+), 35 deletions(-)

diff --git a/etk/UString.cpp b/etk/UString.cpp
index c2aa9f8..56963e3 100644
--- a/etk/UString.cpp
+++ b/etk/UString.cpp
@@ -54,7 +54,7 @@ etk::CCout& etk::operator <<(etk::CCout& _os, const etk::Vector<etk::UString>& _
 etk::UString::UString(void)
 {
 	//TK_INFO("new etk::UString()");
-	m_data.PushBack('\0');
+	m_data.PushBack(etk::UniChar::Null);
 }
 
 
@@ -73,8 +73,8 @@ etk::UString::UString(const char* _data, unicode::charset_te _inputCharset)
 		unicode::convertIsoToUnicode(_inputCharset, transformData, m_data);
 	}
 	if(    0 == m_data.Size()
-	    || m_data[m_data.Size()-1]!='\0') {
-		m_data.PushBack('\0');
+	    || m_data[m_data.Size()-1]!=etk::UniChar::Null) {
+		m_data.PushBack(etk::UniChar::Null);
 	}
 }
 
@@ -131,7 +131,7 @@ etk::UString::UString(const bool _inputData, etk::UString::printMode_te _mode, b
 			}
 			break;
 	}
-	m_data.PushBack('\0');
+	m_data.PushBack(etk::UniChar::Null);
 }
 
 etk::UString::UString(const etk::UString& _obj)
@@ -143,7 +143,7 @@ etk::UString::UString(const etk::UString& _obj)
 etk::UString::UString(const etk::UniChar& _inputData)
 {
 	m_data.PushBack(_inputData);
-	m_data.PushBack('\0');
+	m_data.PushBack(etk::UniChar::Null);
 }
 
 etk::UString::UString(const float _inputData)
@@ -289,9 +289,9 @@ void etk::UString::SetNumber(bool _negative, const uint64_t& _inputData, etk::US
 		//TK_ERROR ("        " << ploppp);
 	}
 	if (m_data.Size()==0) {
-		m_data.PushBack('\0');
-	} else if (m_data[m_data.Size()-1]!='\0') {
-		m_data.PushBack('\0');
+		m_data.PushBack(etk::UniChar::Null);
+	} else if (m_data[m_data.Size()-1]!=etk::UniChar::Null) {
+		m_data.PushBack(etk::UniChar::Null);
 	}
 	//TK_ERROR(" convert : " << _inputData << " in : " << *this << " len=" << m_data.Size());
 }
@@ -369,11 +369,11 @@ void etk::UString::Set(const etk::Vector<etk::UniChar>& _inputData)
 {
 	m_data = _inputData;
 	if (m_data.Size()>0) {
-		if (m_data[m_data.Size()-1] != '\0') {
-			m_data.PushBack('\0');
+		if (m_data[m_data.Size()-1] != etk::UniChar::Null) {
+			m_data.PushBack(etk::UniChar::Null);
 		}
 	} else {
-		m_data.PushBack('\0');
+		m_data.PushBack(etk::UniChar::Null);
 	}
 	//TK_DEBUG("m_dataLen="<<m_dataLen << " m_dataLenUTF8="<<m_dataLenUTF8 << " description=" << m_data);
 }
@@ -383,7 +383,7 @@ void etk::UString::Set(const char* _inputData, int32_t _len)
 	// clear all the data
 	m_data.Clear();
 	if (NULL == _inputData) {
-		m_data.PushBack('\0');
+		m_data.PushBack(etk::UniChar::Null);
 		// nothing to add ... ==> just exit
 		return;
 	}
@@ -406,9 +406,9 @@ void etk::UString::Set(const char* _inputData, int32_t _len)
 		unicode::convertUtf8ToUnicode(tmpChar, m_data);
 	}
 	if (m_data.Size()==0) {
-		m_data.PushBack('\0');
-	} else if (m_data[m_data.Size()-1]!='\0') {
-		m_data.PushBack('\0');
+		m_data.PushBack(etk::UniChar::Null);
+	} else if (m_data[m_data.Size()-1]!=etk::UniChar::Null) {
+		m_data.PushBack(etk::UniChar::Null);
 	}
 }
 
@@ -417,7 +417,7 @@ void etk::UString::Set(const etk::UniChar* _inputData, int32_t _len)
 	// clear all the data
 	m_data.Clear();
 	if (NULL == _inputData) {
-		m_data.PushBack('\0');
+		m_data.PushBack(etk::UniChar::Null);
 		// nothing to add ... ==> just exit
 		return;
 	}
@@ -430,9 +430,9 @@ void etk::UString::Set(const etk::UniChar* _inputData, int32_t _len)
 		m_data.PushBack(_inputData, _len);
 	}
 	if (m_data.Size()==0) {
-		m_data.PushBack('\0');
-	} else if (m_data[m_data.Size()-1]!='\0') {
-		m_data.PushBack('\0');
+		m_data.PushBack(etk::UniChar::Null);
+	} else if (m_data[m_data.Size()-1]!=etk::UniChar::Null) {
+		m_data.PushBack(etk::UniChar::Null);
 	}
 }
 
@@ -592,7 +592,7 @@ const etk::UString& etk::UString::operator+= (const etk::UString &_obj)
 		// This previous include the \0 in case of the 2 UString are different...
 		if( this == &_obj ) {
 			// add the removed end UString
-			m_data.PushBack('\0');
+			m_data.PushBack(etk::UniChar::Null);
 		}
 	}
 	return *this;
@@ -606,7 +606,7 @@ const etk::UString& etk::UString::operator+= (const etk::UniChar& _obj)
 	// copy the data ...
 	m_data.PushBack(_obj);
 	// add the last '\0' element
-	m_data.PushBack('\0');
+	m_data.PushBack(etk::UniChar::Null);
 	return *this;
 }
 */
@@ -670,13 +670,15 @@ void etk::UString::Add(int32_t _currentID, const etk::UniChar _inputData)
 {
 	etk::UniChar data[2];
 	data[0] = _inputData;
-	data[1] = 0;
+	data[1] = etk::UniChar::Null;
 	Add(_currentID, data);
 }
 
 void etk::UString::Append(const etk::UniChar& _inputData)
 {
+	m_data.PopBack();
 	m_data.PushBack(_inputData);
+	m_data.PushBack(etk::UniChar::Null);
 }
 
 
@@ -694,7 +696,7 @@ void etk::UString::Remove(int32_t _currentID, int32_t _len)
 void etk::UString::Clear(void)
 {
 	m_data.Clear();
-	m_data.PushBack('\0');
+	m_data.PushBack(etk::UniChar::Null);
 }
 
 int32_t etk::UString::FindForward(const etk::UniChar _element, int32_t _startPos) const
@@ -742,7 +744,7 @@ etk::UString etk::UString::Extract(int32_t _posStart, int32_t _posEnd) const
 		_posEnd = Size();
 	}
 	out.m_data = m_data.Extract(_posStart, _posEnd);
-	out.m_data.PushBack('\0');
+	out.m_data.PushBack(etk::UniChar::Null);
 	return out;
 }
 
@@ -775,7 +777,7 @@ etk::UString etk::UString::ExtractLine(int32_t _pos) const
 		stopPos = Size();
 	}
 	out.m_data = m_data.Extract(startPos, stopPos);
-	out.m_data.PushBack('\0');
+	out.m_data.PushBack(etk::UniChar::Null);
 	return out;
 }
 
diff --git a/etk/UString.h b/etk/UString.h
index 11567d1..c8aa42b 100644
--- a/etk/UString.h
+++ b/etk/UString.h
@@ -117,10 +117,10 @@ namespace etk
 			/*****************************************************
 			 *    [] operator
 			 *****************************************************/
-			const etk::UniChar& operator[] (int32_t _pos) const {
+			const etk::UniChar& operator[] (esize_t _pos) const {
 				return m_data[_pos];
 			}
-			etk::UniChar& operator[] (int32_t _pos) {
+			etk::UniChar& operator[] (esize_t _pos) {
 				return m_data[_pos];
 			}
 			
diff --git a/etk/UniChar.cpp b/etk/UniChar.cpp
index b01cfc2..8f3b641 100644
--- a/etk/UniChar.cpp
+++ b/etk/UniChar.cpp
@@ -15,6 +15,7 @@
 #include <etk/Vector.h>
 #include <etk/Char.h>
 
+const etk::UniChar etk::UniChar::Null('\0');
 const etk::UniChar etk::UniChar::Return('\n');
 const etk::UniChar etk::UniChar::CarrierReturn('\r');
 const etk::UniChar etk::UniChar::Tabulation('\t');
@@ -179,33 +180,92 @@ int8_t etk::UniChar::GetUtf8(char _output[5]) const
 		return 4;
 	}
 }
+/*
+etk::Vector<int8_t> etk::UniChar::GetUtf8(void) const
+{
+	etk::Vector<int8_t> ret;
+	uint32_t value = GetUtf8();
+	if (0xFF >= value) {
+		ret.PushBack((char)value);
+	} else if (0xFFFF >= value) {
+		ret.PushBack((char)((value>>8)  & 0x000000FF));
+		ret.PushBack((char)value);
+	} else if (0xFFFFFF >= value) {
+		ret.PushBack((char)((value>>16) & 0x000000FF));
+		ret.PushBack((char)((value>>8)  & 0x000000FF));
+		ret.PushBack((char)value);
+	} else {
+		ret.PushBack((char)((value>>24) & 0x000000FF));
+		ret.PushBack((char)((value>>16) & 0x000000FF));
+		ret.PushBack((char)((value>>8)  & 0x000000FF));
+		ret.PushBack((char)value);
+	}
+	return ret;
+}
+*/
+uint8_t SizeElement(const char* _data, int32_t _lenMax)
+{
+	uint8_t size = 0;
+	TK_ASSERT(0 <= _lenMax, "size can not be < 0 ...");
+	if (0 > _lenMax) {
+		return 0;
+	}
+	//4 case
+	if(    _lenMax >= 1
+	    && (_data[0] & 0x80) == 0x00 ) {
+		// One Char Element
+		size = 1;
+	} else if(    _lenMax >= 2
+	           && (_data[0] & 0xE0) == 0xC0
+	           && (_data[1] & 0xC0) == 0x80) {
+		size = 2;
+	} else if(    _lenMax >= 3
+	           && (_data[0] & 0xF0) == 0xE0
+	           && (_data[1] & 0xC0) == 0x80
+	           && (_data[2] & 0xC0) == 0x80) {
+		size = 3;
+	} else if(    _lenMax >= 4
+	           && (_data[0] & 0xF8) == 0xF0
+	           && (_data[1] & 0xC0) == 0x80
+	           && (_data[2] & 0xC0) == 0x80
+	           && (_data[3] & 0xC0) == 0x80) {
+		size = 4;
+	}
+	return size;
+}
 
-void etk::UniChar::SetUtf8(const char* _input)
+
+int8_t etk::UniChar::SetUtf8(const char* _input)
 {
 	m_value = 0;
 	if (NULL == _input) {
-		return;
+		return 0;
 	}
 	int32_t len = strlen(_input);
+	len = SizeElement(_input, len);
 	switch (len) {
+		default:
+			// case 0 : An error occured...
+			m_value = _input[0];
+			return 0;
 		case 1:
 			m_value = (uint8_t)(_input[0]) & 0x7F;
-			break;
+			return 1;
 		case 2:
 			m_value  = (((uint8_t)_input[0]) & 0x1F)<< 6;
 			m_value +=  ((uint8_t)_input[1]) & 0x3F;
-			break;
+			return 2;
 		case 3:
 			m_value  = (((uint8_t)_input[0]) & 0x0F)<< 12;
 			m_value += (((uint8_t)_input[1]) & 0x3F)<< 6;
 			m_value +=  ((uint8_t)_input[2]) & 0x3F;
-			break;
-		default:
+			return 3;
+		case 4:
 			m_value  = (((uint8_t)_input[0]) & 0x07)<< 18;
 			m_value += (((uint8_t)_input[1]) & 0x3F)<< 12;
 			m_value += (((uint8_t)_input[2]) & 0x3F)<< 6;
 			m_value +=  ((uint8_t)_input[3]) & 0x3F;
-			break;
+			return 4;
 	}
 }
 
diff --git a/etk/UniChar.h b/etk/UniChar.h
index 77b1b49..32d9bbe 100644
--- a/etk/UniChar.h
+++ b/etk/UniChar.h
@@ -44,6 +44,7 @@ namespace etk
 	class UniChar
 	{
 		public: // classic unicar code :
+			static const UniChar Null; //!< '\0' 
 			static const UniChar Return; //!< '\n' 
 			static const UniChar CarrierReturn; //!< '\r' CR
 			static const UniChar Tabulation; //!< '\t' TAB
@@ -175,7 +176,8 @@ namespace etk
 			
 			uint32_t GetUtf8(void) const;
 			int8_t GetUtf8(char _output[5]) const;
-			void SetUtf8(const char* _input);
+			//etk::Vector<int8_t> GetUtf8(void) const;
+			int8_t SetUtf8(const char* _input);
 	};
 };