PATCH: [ 797741 ] Faster file loading

diff --git a/Src/MergeDoc.cpp b/Src/MergeDoc.cpp
index 946a4dc..f6a8f34 100644 (file)
--- a/Src/MergeDoc.cpp
+++ b/Src/MergeDoc.cpp
@@ -922,11 +922,11 @@ int CMergeDoc::CDiffTextBuffer::DetermineCRLFStyle(LPVOID lpBuf, DWORD dwLength)
 }
 
 /// Reads one line from filebuffer and inserts to textbuffer
-void CMergeDoc::CDiffTextBuffer::ReadLineFromBuffer(TCHAR *lpLineBegin, DWORD dwLineLen /* =0 */)
+void CMergeDoc::CDiffTextBuffer::ReadLineFromBuffer(TCHAR *lpLineBegin, DWORD dwLineNum, DWORD dwLineLen /* =0 */)
 {
        if (m_nSourceEncoding >= 0)
                iconvert (lpLineBegin, m_nSourceEncoding, 1, m_nSourceEncoding == 15);
-       InsertLine(lpLineBegin, dwLineLen);
+       AppendLine(dwLineNum, lpLineBegin, dwLineLen);
 }
 
 /// Sets path for temporary files
@@ -944,7 +944,8 @@ int CMergeDoc::CDiffTextBuffer::LoadFromFile(LPCTSTR pszFileName,
        MAPPEDFILEDATA fileData = {0};
        CString sExt;
        BOOL bSuccess = FALSE;
-       BOOL nRetVal = FRESULT_OK;
+       int nRetVal = FRESULT_OK;
+       DWORD dwLines = 0;
 
        // Set encoding based on extension, if we know one
        SplitFilename(pszFileName, NULL, NULL, &sExt);
@@ -960,7 +961,7 @@ int CMergeDoc::CDiffTextBuffer::LoadFromFile(LPCTSTR pszFileName,
        bSuccess = files_openFileMapped(&fileData);
 
        if (bSuccess)
-               nRetVal = files_binCheck(&fileData);
+               nRetVal = files_analyzeFile(&fileData, &dwLines);
        else
                nRetVal = FRESULT_ERROR;
        
@@ -973,12 +974,13 @@ int CMergeDoc::CDiffTextBuffer::LoadFromFile(LPCTSTR pszFileName,
                ASSERT(nCrlfStyle >= 0 && nCrlfStyle <= 2);
                SetCRLFMode(nCrlfStyle);
                
-               m_aLines.SetSize(0, 4096);
+               m_aLines.SetSize(dwLines, 4096);
                
                DWORD dwBytesRead = 0;
                TCHAR *lpChar = (TCHAR *)fileData.pMapBase;
                TCHAR *lpLineBegin = lpChar;
                int eolChars = 0;
+               DWORD nLineNum = 0;
                while (dwBytesRead < fileData.dwSize)
                {
                        TCHAR c = *lpChar;
@@ -996,7 +998,8 @@ int CMergeDoc::CDiffTextBuffer::LoadFromFile(LPCTSTR pszFileName,
                                        dwBytesRead++;
 
                                        // This is an EOL
-                                       ReadLineFromBuffer(lpLineBegin, lpChar - lpLineBegin);
+                                       ReadLineFromBuffer(lpLineBegin, nLineNum, lpChar - lpLineBegin);
+                                       nLineNum++;
                                        lpLineBegin = lpChar;
                                        continue;
                                }
@@ -1007,7 +1010,7 @@ int CMergeDoc::CDiffTextBuffer::LoadFromFile(LPCTSTR pszFileName,
                // Handle case where file ended with line without EOL
                if (lpChar > lpLineBegin)
                        // no ghost line, because that would append an EOL to the last line
-                       ReadLineFromBuffer(lpLineBegin, lpChar - lpLineBegin);
+                       ReadLineFromBuffer(lpLineBegin, nLineNum, lpChar - lpLineBegin);
                else
                        // Last line had EOL, so append succeeding ghost line
                        InsertLine(_T(""), 0);

diff --git a/Src/MergeDoc.h b/Src/MergeDoc.h
index 673bade..d09e3a9 100644 (file)
--- a/Src/MergeDoc.h
+++ b/Src/MergeDoc.h
@@ -94,7 +94,7 @@ private :
                CString m_strTempPath;
 
                int DetermineCRLFStyle(LPVOID lpBuf, DWORD dwLength);
-               void ReadLineFromBuffer(TCHAR *lpLineBegin, DWORD dwLineLen = 0);
+               void ReadLineFromBuffer(TCHAR *lpLineBegin, DWORD dwLineNum, DWORD dwLineLen = 0);
 public :
                void SetTempPath(CString path);
                virtual void AddUndoRecord (BOOL bInsert, const CPoint & ptStartPos, const CPoint & ptEndPos,

diff --git a/Src/files.cpp b/Src/files.cpp
index 5b4ab00..bdab5ad 100644 (file)
--- a/Src/files.cpp
+++ b/Src/files.cpp
@@ -151,7 +151,7 @@ BOOL files_closeFileMapped(MAPPEDFILEDATA *fileData, DWORD newSize, BOOL flush)
  * @note This function is safe: it first checks that there are unread bytes,
  * @note so that we do not read past EOF
  */
-int files_readEOL(TCHAR *lpLineEnd, DWORD bytesLeft, BOOL bEOLSensitive)
+int files_readEOL(TBYTE *lpLineEnd, DWORD bytesLeft, BOOL bEOLSensitive)
 {
        int eolBytes = 0;
        
@@ -197,16 +197,18 @@ int files_readEOL(TCHAR *lpLineEnd, DWORD bytesLeft, BOOL bEOLSensitive)
 }
 
 /**
- * @brief Checks memory-mapped file for a binary data
+ * @brief Checks Checks memory-mapped for binary data and counts lines
+ * in textfile.
  * @note This does not work for UNICODE files
- * @note as WinMerge is not compiled UNICODE enabled
+ * as WinMerge is not compiled UNICODE enabled
  */
-int files_binCheck(MAPPEDFILEDATA *fileData)
+int files_analyzeFile(MAPPEDFILEDATA *fileData, DWORD * dwLineCount)
 {
        // Use unsigned type for binary compare
        TBYTE *lpByte = (TBYTE *)fileData->pMapBase;
        BOOL bBinary = FALSE;
        DWORD dwBytesRead = 0;
+       *dwLineCount = 1;
        
        while ((dwBytesRead < fileData->dwSize - 1) && (bBinary == FALSE))
        {
@@ -214,9 +216,21 @@ int files_binCheck(MAPPEDFILEDATA *fileData)
                if (*lpByte < 0x09)
                {
                        bBinary = TRUE;
+                       lpByte++;
+                       dwBytesRead++;
+               }
+               else if (*lpByte == '\r' || *lpByte == '\n')
+               {
+                       const int nEolBytes = files_readEOL(lpByte, fileData->dwSize - dwBytesRead, TRUE);
+                       lpByte += nEolBytes;
+                       dwBytesRead += nEolBytes;       // Skip EOL chars
+                       (*dwLineCount)++;
+               }
+               else
+               {
+                       lpByte++;
+                       dwBytesRead++;
                }
-               lpByte++;
-               dwBytesRead++;
        }
 
        if (bBinary)

diff --git a/Src/files.h b/Src/files.h
index aa356f9..2b107c6 100644 (file)
--- a/Src/files.h
+++ b/Src/files.h
@@ -53,7 +53,7 @@ struct MAPPEDFILEDATA
 
 BOOL files_openFileMapped(MAPPEDFILEDATA *fileData);
 BOOL files_closeFileMapped(MAPPEDFILEDATA *fileData, DWORD newSize, BOOL flush);
-int files_readEOL(TCHAR *lpLineEnd, DWORD bytesLeft, BOOL bEOLSensitive);
-int files_binCheck(MAPPEDFILEDATA *fileData);
+int files_readEOL(TBYTE *lpLineEnd, DWORD bytesLeft, BOOL bEOLSensitive);
+int files_analyzeFile(MAPPEDFILEDATA *fileData, DWORD * dwLineCount);
 
 #endif // _FILES_H
\ No newline at end of file

diff --git a/Src/readme.txt b/Src/readme.txt
index d97cc07..9b7b810 100644 (file)
--- a/Src/readme.txt
+++ b/Src/readme.txt
@@ -1,7 +1,11 @@
+2003-08-31 Kimmo
+ PATCH: [ 797741 ] Faster file loading
+  WinMerge: files.h files.cpp MergeDoc.h MergeDoc.cpp
+
 2003-08-30 Kimmo
  PATCH: [ 797797 ] OnOptions must exchange diffOptions with the registry
   Submitted by Laoran.
-   WinMerge: MainFrm.cpp
+  WinMerge: MainFrm.cpp
 
 2003-08-30 Laoran
  PATCH: [ 784109 ] layout with two new horizontal panels