C++_MFC_CArchive类

1、FMD开发文集 - CArchive原理作者：冯明德MFC 提供CArchive类实现数据的缓冲区读写，同时定义了类对象的存储与读取方案。 以下对CArchvie 的内部实现作分析。 1.概述 2.内部数据 3.基本数据读写 4.缓冲区的更新 5.指定长度数据段落的读写 6.字符串的读写 7.CObject派生对象的读写 一.概述 CArchive使用了缓冲区，即一段内存空间作为临时数据存储地，对CArchive的读写都先依次排列到此缓冲区，当缓冲区满或用户要求时，将此段整理后的数据读写到指定的存储煤质。 当建立CArchive对象时，应指定其模式是用于缓冲区读，还是用于缓冲区写。 可以这样理

2、解，CArchive对象相当于铁路的货运练调度站，零散的货物被收集，当总量到达火车运量的时候，由火车装运走。 当接到火车的货物时，则货物由被分散到各自的货主。与货运不同的是，交货、取货是按时间循序执行的，而不是凭票据。因此必须保证送货的和取货的货主按同样的循序去存或取。 对于大型的货物，则是拆散成火车单位，运走，取货时，依次取各部分，组装成原物。 二.内部数据 缓冲区指针 BYTE* m_lpBufStart,指向缓冲区，这个缓冲区有可能是底层CFile(如派生类CMemFile)对象提供的，但一般是CArchive自己建立的。 缓冲区尾部指针 BYTE* m_lpBufMax; 缓冲区当前位

3、置指针 BYTE* m_lpBufCur; 初始化时，如果是读模式,当前位置在尾部，如果是写模式，当前位置在头部: m_lpBufCur = (IsLoading() ? m_lpBufMax : m_lpBufStart;三.基本数据读写 对于基本的数据类型，例如字节、双字等，可以直接使用">>"、"<<"符号进行读出、写入。 /操作符定义捕:/插入操作CArchive& operator<<(BYTE by);CArchive& operator<<(WORD w);CArchive&a

4、mp; operator<<(LONG l);CArchive& operator<<(DWORD dw);CArchive& operator<<(float f);CArchive& operator<<(double d);CArchive& operator<<(int i);CArchive& operator<<(short w);CArchive& operator<<(char ch);CArchive& operator<<(

5、unsigned u);/提取操作CArchive& operator>>(BYTE& by);CArchive& operator>>(WORD& w);CArchive& operator>>(DWORD& dw);CArchive& operator>>(LONG& l);CArchive& operator>>(float& f);CArchive& operator>>(double& d);CArchive&

6、 operator>>(int& i);CArchive& operator>>(short& w);CArchive& operator>>(char& ch);CArchive& operator>>(unsigned& u);下面以双字为例，分析原码 双字的插入(写) CArchive& CArchive:operator<<(DWORD dw)if (m_lpBufCur + sizeof(DWORD) > m_lpBufMax) /缓冲区空间不够Flush

7、(); /缓冲区内容提交到实际存储煤质。if (!(m_nMode & bNoByteSwap)_AfxByteSwap(dw, m_lpBufCur); /处理字节顺序else*(DWORD*)m_lpBufCur = dw; /添入缓冲区m_lpBufCur += sizeof(DWORD); /移动当前指针return *this;双字的提取(读) CArchive& CArchive:operator>>(DWORD& dw)if (m_lpBufCur + sizeof(DWORD) > m_lpBufMax) /缓冲区要读完了FillBuf

8、fer(sizeof(DWORD) - (UINT)(m_lpBufMax - m_lpBufCur); /重新读入内容到缓冲区dw = *(DWORD*)m_lpBufCur;/读取双字m_lpBufCur += sizeof(DWORD);/移动当前位置指针if (!(m_nMode & bNoByteSwap)_AfxByteSwap(dw, (BYTE*)&dw); /处理字节顺序return *this;四.缓冲区的更新 以上操作中，当缓冲区将插入满或缓冲区将提取空时，都将对缓冲区进行更新处理。缓冲区将插入满时调用Flush(); void CArchive:Flus

9、h()ASSERT_VALID(m_pFile);ASSERT(m_bDirectBuffer | m_lpBufStart != NULL);ASSERT(m_bDirectBuffer | m_lpBufCur != NULL);ASSERT(m_lpBufStart = NULL |AfxIsValidAddress(m_lpBufStart, m_lpBufMax - m_lpBufStart, IsStoring();ASSERT(m_lpBufCur = NULL |AfxIsValidAddress(m_lpBufCur, m_lpBufMax - m_lpBufCur, IsS

10、toring();if (IsLoading()/ unget the characters in the buffer, seek back unused amountif (m_lpBufMax != m_lpBufCur)m_pFile-> Seek(-(m_lpBufMax - m_lpBufCur), CFile:current);m_lpBufCur = m_lpBufMax; / 指向尾else /写模式if (!m_bDirectBuffer) / 内容写入到文件if (m_lpBufCur != m_lpBufStart)m_pFile-> Write(m_lpB

11、ufStart, m_lpBufCur - m_lpBufStart);else/如果是直接针对内存区域的的(例如CMemFile中) (只需移动相关指针，指向新的一块内存)if (m_lpBufCur != m_lpBufStart)m_pFile-> GetBufferPtr(CFile:bufferCommit, m_lpBufCur - m_lpBufStart);/ get next bufferVERIFY(m_pFile-> GetBufferPtr(CFile:bufferWrite, m_nBufSize,(void*)&m_lpBufStart, (vo

12、id*)&m_lpBufMax) = (UINT)m_nBufSize);ASSERT(UINT)m_nBufSize = (UINT)(m_lpBufMax - m_lpBufStart);m_lpBufCur = m_lpBufStart; /指向缓冲区首缓冲区将提取空,会调用FillBuffer。 nBytesNeeded为当前剩余部分上尚有用的字节void CArchive:FillBuffer(UINT nBytesNeeded)ASSERT_VALID(m_pFile);ASSERT(IsLoading();ASSERT(m_bDirectBuffer | m_lpBufS

13、tart != NULL);ASSERT(m_bDirectBuffer | m_lpBufCur != NULL);ASSERT(nBytesNeeded > 0);ASSERT(nBytesNeeded <= (UINT)m_nBufSize);ASSERT(m_lpBufStart = NULL |AfxIsValidAddress(m_lpBufStart, m_lpBufMax - m_lpBufStart, FALSE);ASSERT(m_lpBufCur = NULL |AfxIsValidAddress(m_lpBufCur, m_lpBufMax - m_lpBu

14、fCur, FALSE);UINT nUnused = m_lpBufMax - m_lpBufCur;ULONG nTotalNeeded = (ULONG)nBytesNeeded) + nUnused;/ 从文件中读取if (!m_bDirectBuffer)ASSERT(m_lpBufCur != NULL);ASSERT(m_lpBufStart != NULL);ASSERT(m_lpBufMax != NULL);if (m_lpBufCur > m_lpBufStart)/保留剩余的尚未处理的部分，将它们移动到头if (int)nUnused > 0)memmove

15、(m_lpBufStart, m_lpBufCur, nUnused);m_lpBufCur = m_lpBufStart;m_lpBufMax = m_lpBufStart + nUnused;/ read to satisfy nBytesNeeded or nLeft if possibleUINT nRead = nUnused;UINT nLeft = m_nBufSize-nUnused;UINT nBytes;BYTE* lpTemp = m_lpBufStart + nUnused;donBytes = m_pFile-> Read(lpTemp, nLeft);lpTe

16、mp = lpTemp + nBytes;nRead += nBytes;nLeft -= nBytes;while (nBytes > 0 && nLeft > 0 && nRead < nBytesNeeded);m_lpBufCur = m_lpBufStart;m_lpBufMax = m_lpBufStart + nRead;else/ 如果是针对内存区域(CMemFile),移动相关指针，指向新的一块内存if (nUnused != 0)m_pFile-> Seek(-(LONG)nUnused, CFile:current)

17、;UINT nActual = m_pFile-> GetBufferPtr(CFile:bufferRead, m_nBufSize,(void*)&m_lpBufStart, (void*)&m_lpBufMax);ASSERT(nActual = (UINT)(m_lpBufMax - m_lpBufStart);m_lpBufCur = m_lpBufStart;/ not enough data to fill request?if (ULONG)(m_lpBufMax - m_lpBufCur) < nTotalNeeded)AfxThrowArchiv

18、eException(CArchiveException:endOfFile);五.指定长度数据段落的读写以下分析 UINT Read(void* lpBuf, UINT nMax); 读取长度为nMax的数据 void Write(const void* lpBuf, UINT nMax); 写入指定长度nMax的数据 对于大段数据的读写，先使用当前缓冲区中的内容或空间读取或写入，若这些空间够用了，则结束。 否则，从剩余的数据中找出最大的缓冲区整数倍大小的一块数据，直接读写到存储煤质(不反复使用缓冲区)。 剩余的余数部分，再使用缓冲区读写。 (说明:缓冲区读写的主要目的是将零散的数据以缓冲区

19、大小为尺度来处理。对于大型数据，其中间的部分，不是零散的数据，使用缓冲区已经没有意思，故直接读写) 读取 UINT CArchive:Read(void* lpBuf, UINT nMax)ASSERT_VALID(m_pFile);if (nMax = 0)return 0;UINT nMaxTemp = nMax; /还需要读入的长度，读入一部分，就减相应数值，直到此数值变为零/处理当前缓冲区中剩余部分。/如果要求读入字节小于缓冲区中剩余部分,则第一部分为要求读入的字节数，/否则读入全部剩余部分UINT nTemp = min(nMaxTemp, (UINT)(m_lpBufMax - m

20、_lpBufCur); memcpy(lpBuf, m_lpBufCur, nTemp);m_lpBufCur += nTemp;lpBuf = (BYTE*)lpBuf + nTemp; /移动读出内容所在区域的指针nMaxTemp -= nTemp;/当前缓冲区中剩余部分不够要求读入的长度。/还有字节需要读，则需要根据需要执行若干次填充缓冲区，读出，直到读出指定字节。if (nMaxTemp != 0) /计算出去除尾数部分的字节大小(整数个缓冲区大小) /对于这些部分，字节从文件对象中读出，放到输出缓冲区nTemp = nMaxTemp - (nMaxTemp % m_nBufSize)

21、; UINT nRead = 0;UINT nLeft = nTemp;UINT nBytes;donBytes = m_pFile-> Read(lpBuf, nLeft); /要求读入此整数缓冲区部分大小lpBuf = (BYTE*)lpBuf + nBytes;nRead += nBytes;nLeft -= nBytes;while (nBytes > 0) && (nLeft > 0); 知道读入了预定大小，或到达文件尾nMaxTemp -= nRead;if (nRead = nTemp) /读入的字节等于读入的整数倍部分 该读最后的余数部分了/

22、建立装有此最后余数部分的内容的CArchive的工作缓冲区。if (!m_bDirectBuffer)UINT nLeft = max(nMaxTemp, (UINT)m_nBufSize);UINT nBytes;BYTE* lpTemp = m_lpBufStart;nRead = 0;donBytes = m_pFile-> Read(lpTemp, nLeft); /从文件中读入到CArchive缓冲区lpTemp = lpTemp + nBytes;nRead += nBytes;nLeft -= nBytes;while (nBytes > 0) &&

23、(nLeft > 0) && nRead < nMaxTemp);m_lpBufCur = m_lpBufStart;m_lpBufMax = m_lpBufStart + nRead;elsenRead = m_pFile-> GetBufferPtr(CFile:bufferRead, m_nBufSize,(void*)&m_lpBufStart, (void*)&m_lpBufMax);ASSERT(nRead = (UINT)(m_lpBufMax - m_lpBufStart);m_lpBufCur = m_lpBufStart;/

24、读出此剩余部分到输出nTemp = min(nMaxTemp, (UINT)(m_lpBufMax - m_lpBufCur);memcpy(lpBuf, m_lpBufCur, nTemp);m_lpBufCur += nTemp;nMaxTemp -= nTemp;return nMax - nMaxTemp;保存,写入 void CArchive:Write(const void* lpBuf, UINT nMax)if (nMax = 0)return;/读入可能的部分到缓冲区当前的剩余部分UINT nTemp = min(nMax, (UINT)(m_lpBufMax - m_lpB

25、ufCur);memcpy(m_lpBufCur, lpBuf, nTemp);m_lpBufCur += nTemp;lpBuf = (BYTE*)lpBuf + nTemp;nMax -= nTemp;if (nMax > 0) /还有未写入的部分Flush(); /将当前缓冲区写入到存储煤质/计算出整数倍缓冲区大小的字节数nTemp = nMax - (nMax % m_nBufSize);m_pFile-> Write(lpBuf, nTemp); /直接写到文件lpBuf = (BYTE*)lpBuf + nTemp;nMax -= nTemp;/剩余部分添加到缓冲区if

26、 (m_bDirectBuffer)/ sync up direct mode buffer to new file positionVERIFY(m_pFile-> GetBufferPtr(CFile:bufferWrite, m_nBufSize,(void*)&m_lpBufStart,(void*)&m_lpBufMax)=(UINT)m_nBufSize);ASSERT(UINT)m_nBufSize = (UINT)(m_lpBufMax - m_lpBufStart);m_lpBufCur = m_lpBufStart;/ copy remaining t

27、o active bufferASSERT(nMax < (UINT)m_nBufSize);ASSERT(m_lpBufCur = m_lpBufStart);memcpy(m_lpBufCur, lpBuf, nMax);m_lpBufCur += nMax;六.字符串的读写CArchive提供的WriteString和ReadString 字符串写 void CArchive:WriteString(LPCTSTR lpsz)ASSERT(AfxIsValidString(lpsz);Write(lpsz, lstrlen(lpsz) * sizeof(TCHAR); /调用Wri

28、te，将字符串对应的一段数据写入字符串读(读取一行字符串) LPTSTR CArchive:ReadString(LPTSTR lpsz, UINT nMax)/ if nMax is negative (such a large number doesn''t make sense given today''s/ 2gb address space), then assume it to mean "keep the newline".int nStop = (int)nMax < 0 ? -(int)nMax : (int)nMax

29、;ASSERT(AfxIsValidAddress(lpsz, (nStop+1) * sizeof(TCHAR);_TUCHAR ch;int nRead = 0;TRYwhile (nRead < nStop)*this >> ch; /读出一个字节/ stop and end-of-line (trailing ''n'' is ignored) 遇换行回车if (ch = ''n'' | ch = ''r'')if (ch = ''r'')*

30、this >> ch;/ store the newline when called with negative nMaxif (int)nMax != nStop)lpsznRead+ = ch;break;lpsznRead+ = ch;CATCH(CArchiveException, e)if (e-> m_cause = CArchiveException:endOfFile)DELETE_EXCEPTION(e);if (nRead = 0)return NULL;elseTHROW_LAST();END_CATCHlpsznRead = ''0&#

31、39;'return lpsz;ReadString到CString对象,可以多行字符 BOOL CArchive:ReadString(CString& rString)rString = &afxChNil; / empty string without deallocatingconst int nMaxSize = 128;LPTSTR lpsz = rString.GetBuffer(nMaxSize);LPTSTR lpszResult;int nLen;for (;)lpszResult = ReadString(lpsz, (UINT)-nMaxSize

32、); / store the newlinerString.ReleaseBuffer();/ if string is read completely or EOFif (lpszResult = NULL |(nLen = lstrlen(lpsz) < nMaxSize |lpsznLen-1 = ''n'')break;nLen = rString.GetLength();lpsz = rString.GetBuffer(nMaxSize + nLen) + nLen;/ remove ''n'' from end

33、of string if presentlpsz = rString.GetBuffer(0);nLen = rString.GetLength();if (nLen != 0 && lpsznLen-1 = ''n'')rString.GetBufferSetLength(nLen-1);return lpszResult != NULL;使用CString对象的"<<"与">>"符读写字符串 CString定义了输入输出符,可以象基本类型的数据一样使用CArchive 的操作符定

34、义 friend CArchive& AFXAPI operator<<(CArchive& ar, const CString& string);friend CArchive& AFXAPI operator>>(CArchive& ar, CString& string);/ CString serialization code/ String format:/ UNICODE strings are always prefixed by 0xff, 0xfffe/ if < 0xff chars: len:B

35、YTE, TCHAR chars/ if >= 0xff characters: 0xff, len:WORD, TCHAR chars/ if >= 0xfffe characters: 0xff, 0xffff, len:DWORD, TCHARsCArchive& AFXAPI operator<<(CArchive& ar, const CString& string)/ special signature to recognize unicode strings#ifdef _UNICODEar << (BYTE)0xff;

36、ar << (WORD)0xfffe;#endifif (string.GetData()-> nDataLength < 255)ar << (BYTE)string.GetData()-> nDataLength;else if (string.GetData()-> nDataLength < 0xfffe)ar << (BYTE)0xff;ar << (WORD)string.GetData()-> nDataLength;elsear << (BYTE)0xff;ar << (

37、WORD)0xffff;ar << (DWORD)string.GetData()-> nDataLength;ar.Write(string.m_pchData, string.GetData()-> nDataLength*sizeof(TCHAR);return ar;/ return string length or -1 if UNICODE string is found in the archiveAFX_STATIC UINT AFXAPI _AfxReadStringLength(CArchive& ar)DWORD nNewLen;/ att

38、empt BYTE length firstBYTE bLen;ar >> bLen;if (bLen < 0xff)return bLen;/ attempt WORD lengthWORD wLen;ar >> wLen;if (wLen = 0xfffe)/ UNICODE string prefix (length will follow)return (UINT)-1;else if (wLen = 0xffff)/ read DWORD of lengthar >> nNewLen;return (UINT)nNewLen;elseretu

39、rn wLen;CArchive& AFXAPI operator>>(CArchive& ar, CString& string)#ifdef _UNICODEint nConvert = 1; / if we get ANSI, convert#elseint nConvert = 0; / if we get UNICODE, convert#endifUINT nNewLen = _AfxReadStringLength(ar);if (nNewLen = (UINT)-1)nConvert = 1 - nConvert;nNewLen = _Afx

40、ReadStringLength(ar);ASSERT(nNewLen != -1);/ set length of string to new lengthUINT nByteLen = nNewLen;#ifdef _UNICODEstring.GetBufferSetLength(int)nNewLen);nByteLen += nByteLen * (1 - nConvert); / bytes to read#elsenByteLen += nByteLen * nConvert; / bytes to readif (nNewLen = 0)string.GetBufferSetL

41、ength(0);elsestring.GetBufferSetLength(int)nByteLen+nConvert);#endif/ read in the charactersif (nNewLen != 0)ASSERT(nByteLen != 0);/ read new dataif (ar.Read(string.m_pchData, nByteLen) != nByteLen)AfxThrowArchiveException(CArchiveException:endOfFile);/ convert the data if as necessaryif (nConvert !

42、= 0)#ifdef _UNICODECStringData* pOldData = string.GetData();LPSTR lpsz = (LPSTR)string.m_pchData;#elseCStringData* pOldData = string.GetData();LPWSTR lpsz = (LPWSTR)string.m_pchData;#endiflpsznNewLen = ''0'' / must be NUL terminatedstring.Init(); / don''t delete the old datas

43、tring = lpsz; / convert with operator=(LPWCSTR)CString:FreeData(pOldData);return ar;七.CObject派生对象的读写 MFC中多数类都从CObject类派生，CObject类与CArchive类有着良好的合作关系，能实现将对象序列化储存到文件或其他媒介中去，或者读取预先储存的对象，动态建立对象等功能。 CObject定义了针对CArvhive的输入输出操作符,可以向其他基本数据类型一样使用"<<"、"<<"符号 CArchive& AFXA

44、PI operator<<(CArchive& ar, const CObject* pOb) ar.WriteObject(pOb); return ar; CArchive& AFXAPI operator>>(CArchive& ar, CObject*& pOb) pOb = ar.ReadObject(NULL); return ar; 当使用这些符号时，实际上执行的是CArchive的WriteObject和ReadObject成员 WriteObject与ReadObject 在WriteObject与ReadObject中

45、先写入或读取运行时类信息(CRuntimeClas)，再调用Serialze(.),按其中的代码读写具体的对象数据。 因此，只要在CObject派生类中重载Serilize()函数，写入具体的读写过程，就可以使对象具有存储与创建能力。/将对象写入到缓冲区void CArchive:WriteObject(const CObject* pOb)DWORD nObIndex;/ make sure m_pStoreMap is initializedMapObject(NULL);if (pOb = NULL)/ save out null tag to represent NULL pointe

46、r*this << wNullTag;else if (nObIndex = (DWORD)(*m_pStoreMap)(void*)pOb) != 0)/ assumes initialized to 0 map/ save out index of already stored objectif (nObIndex < wBigObjectTag)*this << (WORD)nObIndex;else*this << wBigObjectTag;*this << nObIndex;else/ write class of object

47、 firstCRuntimeClass* pClassRef = pOb-> GetRuntimeClass();WriteClass(pClassRef); /写入运行类信息/ enter in stored object table, checking for overflowCheckCount();(*m_pStoreMap)(void*)pOb = (void*)m_nMapCount+;/ 调用CObject的Serialize成员，按其中的代码写入类中数据。(CObject*)pOb)-> Serialize(*this);CObject* CArchive:Read

48、Object(const CRuntimeClass* pClassRefRequested)/ attempt to load next stream as CRuntimeClassUINT nSchema;DWORD obTag;/先读入运行时类信息CRuntimeClass* pClassRef = ReadClass(pClassRefRequested, &nSchema, &obTag);/ check to see if tag to already loaded objectCObject* pOb;if (pClassRef = NULL)if (obTag

49、 > (DWORD)m_pLoadArray-> GetUpperBound()/ tag is too large for the number of objects read so farAfxThrowArchiveException(CArchiveException:badIndex,m_strFileName);pOb = (CObject*)m_pLoadArray-> GetAt(obTag);if (pOb != NULL && pClassRefRequested != NULL && !pOb-> IsKindOf(

50、pClassRefRequested)/ loaded an object but of the wrong classAfxThrowArchiveException(CArchiveException:badClass,m_strFileName);else/ 建立对象pOb = pClassRef-> CreateObject();if (pOb = NULL)AfxThrowMemoryException();/ Add to mapping array BEFORE de-serializingCheckCount();m_pLoadArray-> InsertAt(m_

51、nMapCount+, pOb);/ Serialize the object with the schema number set in the archiveUINT nSchemaSave = m_nObjectSchema;m_nObjectSchema = nSchema;pOb-> Serialize(*this); /调用CObject的Serialize,按其中代码读入对象数据。m_nObjectSchema = nSchemaSave;ASSERT_VALID(pOb);return pOb;运行时类信息的读写 为了避免众多重复的同类对象写入重复的类信息，CArchiv

52、e中使用CMap对象储存和检索类信息。void CArchive:WriteClass(const CRuntimeClass* pClassRef)ASSERT(pClassRef != NULL);ASSERT(IsStoring(); / proper directionif (pClassRef-> m_wSchema = 0xFFFF)TRACE1("Warning: Cannot call WriteClass/WriteObject for %hs.n",pClassRef-> m_lpszClassName);AfxThrowNotSupport

53、edException();/ make sure m_pStoreMap is initializedMapObject(NULL);/ write out class id of pOb, with high bit set to indicate/ new object follows/ ASSUME: initialized to 0 mapDWORD nClassIndex;if (nClassIndex = (DWORD)(*m_pStoreMap)(void*)pClassRef) != 0)/ previously seen class, write out the index

54、 tagged by high bitif (nClassIndex < wBigObjectTag)*this << (WORD)(wClassTag | nClassIndex);else*this << wBigObjectTag;*this << (dwBigClassTag | nClassIndex);else/ store new class*this << wNewClassTag;pClassRef-> Store(*this);/ store new class reference in map, checking for overflowCheckCount();(*m_pStoreMap)(void*)pClassRef = (void*)m_nMapCount+;CRuntimeClass* CArchive:ReadClass(const CRuntim