改进UTF8编码的兼容性。

Author: wangdaye078

src/common/QUtf8.cpp

@@ -1,29 +1,64 @@
 #include "QUtf8.h"
 #include "utf8.h"
 #include <QVector>
+
 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(t_decodelen + 1, 0);
-	for (int i = 0; i < t_decodelen; ++i)
+	QVector<ushort> t_utf16;
+	for (int i = 0, j = 0; i < t_decodelen; ++i)
 	{
-		Q_ASSERT(t_ucsBuff[i] < 0xFFFF);
-		t_utf16[i] = (ushort)t_ucsBuff[i];
+		uint32_t t_ucs4 = t_ucsBuff[i];
+		if (t_ucs4 > 0x10FFFF || (0xD800 <= t_ucs4 && t_ucs4 <= 0xDFFF))
+		{
+			Q_ASSERT(false);
+		}
+		else if (t_ucsBuff[i] > 0xFFFF)
+		{
+			uint32_t t_ucs = t_ucs4 - 0x10000;
+			t_utf16.push_back((ushort)((t_ucs >> 10) + 0xD800));
+			t_utf16.push_back((ushort)((t_ucs & 0x3FF) + 0xDC00));
+		}
+		else
+			t_utf16.push_back((ushort)t_ucs4);
 	}
-
-	return QString::fromUtf16(t_utf16.data(), t_decodelen);;
+	return QString::fromUtf16(t_utf16.data(), t_utf16.length());
 }
+
 QByteArray QString_to_utf8(const QString& _str)
 {
-	QVector<ucs4_t> t_ucsBuff2(_str.length() + 1, 0);
+	QVector<ucs4_t> t_ucsBuff;
+	const ushort* t_pstr = _str.utf16();
 	for (int i = 0; i < _str.length(); ++i)
-		t_ucsBuff2[i] = *(_str.utf16() + i);
-
-	int t_needlen2 = (int)u8encodelen(t_ucsBuff2.data()) + 1;
+	{
+		uint32_t t_ucs4;
+		ushort t_ch = *(t_pstr + i);
+		if (t_ch >= 0xD800 && t_ch <= 0xDBFF && (i + 1) < _str.length())
+		{
+			ushort t_ch2 = *(t_pstr + i + 1);
+			if (t_ch2 >= 0xDC00 && t_ch2 <= 0xDFFF)
+			{
+				t_ucs4 = ((uint32_t)(t_ch - 0xD800) << 10) + (uint32_t)(t_ch2 - 0xDC00) + 0x10000;
+				i++;
+			}
+			else
+			{
+				Q_ASSERT(false);
+				t_ucs4 = t_ch;
+			}
+		}
+		else
+		{
+			t_ucs4 = t_ch;
+		}
+		t_ucsBuff.push_back(t_ucs4);
+	}
+	int t_needlen2 = (int)u8encodelen(t_ucsBuff.data()) + 1;
 	QByteArray t_arr(t_needlen2, char('\0'));
-	/*int t_encodelen = (int)*/u8encode(t_ucsBuff2.data(), t_arr.data(), t_needlen2, NULL);
+	/*int t_encodelen = (int)*/u8encode(t_ucsBuff.data(), t_arr.data(), t_needlen2, NULL);
 	t_arr.resize(t_arr.size() - 1);
 	return t_arr;
-}
+}

src/common/QUtf8.h

@@ -10,5 +10,4 @@
 
 extern QString utf8_to_QString(const char* _pBuff, quint32 _len);
 extern QByteArray QString_to_utf8(const QString& _str);
-
 #endif // QUtf8_h__

src/common/utf8.cpp

@@ -8,8 +8,8 @@
  * or commercial. It's free.
  */
 
-# include <stdint.h>
 # include <stddef.h>
+# include <stdint.h>
 
 # include "utf8.h"
 
@@ -245,11 +245,11 @@ size_t u8encode(ucs4_t* us, char* des, size_t n, size_t* illegal)
 size_t u8encodelen(ucs4_t* us)
 {
 	size_t len = 0;
-	char tmp[4];
+	char tmp[5];
 	while (*us)
 	{
 		char* ptmp = tmp;
-		size_t left = 4;
+		size_t left = 5;
 		int ret = putu8c(*us, &ptmp, &left);
 		if (ret > 0)
 		{