继续修改编码兼容性

Author: wangdaye078

src/QArscString.cpp

@@ -1,6 +1,6 @@
-#include "QArscString.h"
-#include "common/QUtf8.h"
 #include "common/basicDefine.h"
+#include "common/QUtf8.h"
+#include "QArscString.h"
 
 uint32_t writeStringLen_16(char* _pBuff, const QString& _str)
 {
@@ -73,7 +73,6 @@ QString readString(const char* _pBuff, bool _isUTF8)
 	{
 		uint t_len = readStringLen_8(_pBuff);
 		t_str = utf8_to_QString(_pBuff, t_len);
-		QByteArray t_arr = t_str.toUtf8();
 		Q_ASSERT(_pBuff[t_len] == 0);
 	}
 	return t_str;

src/QStringPool.cpp

@@ -103,7 +103,7 @@ void QStringPool::readBuff(const char* _buff)
 	{
 		m_strings.insert(ArscRichStringMap::value_type(*i, (*i)->guid));
 		m_guid_to_string.insert((*i)->guid, *i);
-		//理论上，同样的字符串不应该重复出现。
+		//理论上，同样的字符串不应该重复出现。但是结果某些工具（比如AntiSplit-M）处理的，可能出现这个问题。
 		Q_ASSERT(m_string_to_guid.find(*i) == m_string_to_guid.end());
 		m_string_to_guid.insert(ArscRichStringMap::value_type(*i, (*i)->guid));
 	}

src/common/QUtf8.cpp

@@ -4,15 +4,14 @@
 
 QString utf8_to_QString(const char* _pBuff, quint32 _len)
 {
-	QByteArray t_byteArr(_pBuff, _len);
 	int t_needlen = (int)u8decodelen(_pBuff) + 1;
 	QVector<ucs4_t> t_ucsBuff(t_needlen, 0);
 	int t_decodelen = (int)u8decode(_pBuff, t_ucsBuff.data(), t_needlen, NULL);
 	QVector<ushort> t_utf16;
 	for (int i = 0, j = 0; i < t_decodelen; ++i)
 	{
 		uint32_t t_ucs4 = t_ucsBuff[i];
-		if (t_ucs4 > 0x10FFFF || (0xD800 <= t_ucs4 && t_ucs4 <= 0xDFFF))
+		if (t_ucs4 > 0x10FFFF)// || (0xD800 <= t_ucs4 && t_ucs4 <= 0xDFFF))
 		{
 			Q_ASSERT(false);
 		}
@@ -22,6 +21,11 @@ QString utf8_to_QString(const char* _pBuff, quint32 _len)
 			t_utf16.push_back((ushort)((t_ucs >> 10) + 0xD800));
 			t_utf16.push_back((ushort)((t_ucs & 0x3FF) + 0xDC00));
 		}
+		else if (0xD800 <= t_ucs4 && t_ucs4 <= 0xDFFF)
+		{
+			//utf-16的代理项范围，规范情况下是不允许直接使用的，但是某些不规范编码的情况下，也会出现这种情况
+			t_utf16.push_back((ushort)t_ucs4);
+		}
 		else
 			t_utf16.push_back((ushort)t_ucs4);
 	}
@@ -56,6 +60,7 @@ QByteArray QString_to_utf8(const QString& _str)
 		}
 		t_ucsBuff.push_back(t_ucs4);
 	}
+	t_ucsBuff.push_back(0);
 	int t_needlen2 = (int)u8encodelen(t_ucsBuff.data()) + 1;
 	QByteArray t_arr(t_needlen2, char('\0'));
 	/*int t_encodelen = (int)*/u8encode(t_ucsBuff.data(), t_arr.data(), t_needlen2, NULL);