fix bug introduced by the last change: we could overwrite the buffer if the file...

[wxWidgets.git] / src / common / textfile.cpp

diff --git a/src/common/textfile.cpp b/src/common/textfile.cpp
index 4347c0618b50c703c6df9121b223c1fd51f0bd43..79d837f6319ccc8e87ee574410c795b78c29b8cc 100644 (file)
--- a/src/common/textfile.cpp
+++ b/src/common/textfile.cpp
@@ -1,233 +1,316 @@
 ///////////////////////////////////////////////////////////////////////////////
 ///////////////////////////////////////////////////////////////////////////////

-// Name:        textfile.cpp
+// Name:        src/common/textfile.cpp

 // Purpose:     implementation of wxTextFile class
 // Author:      Vadim Zeitlin
 // Purpose:     implementation of wxTextFile class
 // Author:      Vadim Zeitlin

-// Modified by: 
+// Modified by:

 // Created:     03.04.98
 // RCS-ID:      $Id$
 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
 // Created:     03.04.98
 // RCS-ID:      $Id$
 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>

-// Licence:     wxWindows license
+// Licence:     wxWindows licence

 ///////////////////////////////////////////////////////////////////////////////
 
 // ============================================================================
 // headers
 // ============================================================================
 
 ///////////////////////////////////////////////////////////////////////////////
 
 // ============================================================================
 // headers
 // ============================================================================
 

-#ifdef __GNUG__
-#pragma implementation "textfile.h"
-#endif
-

 #include  "wx/wxprec.h"
 
 #ifdef    __BORLANDC__
 #include  "wx/wxprec.h"
 
 #ifdef    __BORLANDC__

-  #pragma hdrstop
+    #pragma hdrstop

 #endif  //__BORLANDC__
 
 #endif  //__BORLANDC__
 

-#include  <wx/string.h>
-#include  <wx/intl.h>
-#include  <wx/file.h>
-#include  <wx/log.h>
-#include  <wx/textfile.h>
+#if !wxUSE_FILE || !wxUSE_TEXTBUFFER
+    #undef wxUSE_TEXTFILE
+    #define wxUSE_TEXTFILE 0
+#endif // wxUSE_FILE

 
 

-// ============================================================================
-// wxTextFile class implementation
-// ============================================================================
+#if wxUSE_TEXTFILE

 
 

-// ----------------------------------------------------------------------------
-// static variables
-// ----------------------------------------------------------------------------
-  
-// default type is the native one
-const wxTextFile::Type wxTextFile::typeDefault = wxTextFile::
-#if   defined(__WXMSW__)
-  Type_Dos;
-#elif defined(__UNIX__)
-  Type_Unix;
-#elif defined(__MAC__)
-  Type_Mac;
-  // if you feel brave, remove the next line
-  #error  "wxTextFile: code for Mac files is untested."
-#else
-  Type_None;
-  #error  "wxTextFile: unsupported platform."
+#ifndef WX_PRECOMP
+    #include "wx/string.h"
+    #include "wx/intl.h"
+    #include "wx/file.h"
+    #include "wx/log.h"

 #endif
 
 #endif
 

+#include "wx/textfile.h"
+#include "wx/filename.h"
+#include "wx/buffer.h"

 
 

-// ----------------------------------------------------------------------------
-// ctors & dtor
-// ----------------------------------------------------------------------------
+// ============================================================================
+// wxTextFile class implementation
+// ============================================================================

 
 

-wxTextFile::wxTextFile(const wxString& strFile) : m_strFile(strFile)
+wxTextFile::wxTextFile(const wxString& strFileName)
+          : wxTextBuffer(strFileName)

 {
 }
 
 {
 }
 

-wxTextFile::~wxTextFile()
-{
-  // m_file dtor called automatically
-}

 
 // ----------------------------------------------------------------------------
 // file operations
 // ----------------------------------------------------------------------------
 
 
 // ----------------------------------------------------------------------------
 // file operations
 // ----------------------------------------------------------------------------
 

-bool wxTextFile::Open(const wxString& strFile)
+bool wxTextFile::OnExists() const

 {
 {

-  m_strFile = strFile;
-  return Open();
+    return wxFile::Exists(m_strBufferName);

 }
 
 }
 

-bool wxTextFile::Open()
-{
-  // file name must be either given in ctor or in Open(const wxString&)
-  wxASSERT( !m_strFile.IsEmpty() );
-
-  // open file in read-only mode
-  if ( !m_file.Open(m_strFile) )
-    return FALSE;

 
 

-  // read file into memory
-  bool bRet = Read();
+bool wxTextFile::OnOpen(const wxString &strBufferName, wxTextBufferOpenMode OpenMode)
+{
+    wxFile::OpenMode FileOpenMode;

 
 

-  m_file.Close();
+    switch ( OpenMode )
+    {
+        default:
+            wxFAIL_MSG( _T("unknown open mode in wxTextFile::Open") );
+            // fall through

 
 

-  return bRet;
-}
+        case ReadAccess :
+            FileOpenMode = wxFile::read;
+            break;

 
 

-// analyse some lines of the file trying to guess it's type.
-// if it fails, it assumes the native type for our platform.
-wxTextFile::Type wxTextFile::GuessType() const
-{
-  // file should be opened and we must be in it's beginning
-  wxASSERT( m_file.IsOpened() && m_file.Tell() == 0 );
-
-  // scan the file lines
-  uint nUnix = 0,     // number of '\n's alone
-       nDos  = 0,     // number of '\r\n'
-       nMac  = 0;     // number of '\r's
-
-  // we take MAX_LINES_SCAN in the beginning, middle and the end of file
-  #define MAX_LINES_SCAN    (10)
-  uint nCount = m_aLines.Count() / 3,
-       nScan =  nCount > 3*MAX_LINES_SCAN ? MAX_LINES_SCAN : nCount / 3;
-
-  #define   AnalyseLine(n)              \
-    switch ( m_aTypes[n] ) {            \
-      case Type_Unix: nUnix++; break;   \
-      case Type_Dos:  nDos++;  break;   \
-      case Type_Mac:  nMac++;  break;   \
+        case WriteAccess :
+            FileOpenMode = wxFile::write;
+            break;

     }
 
     }
 

-  uint n;
-  for ( n = 0; n < nScan; n++ )     // the beginning
-    AnalyseLine(n);
-  for ( n = (nCount - nScan)/2; n < (nCount + nScan)/2; n++ )
-    AnalyseLine(n);
-  for ( n = nCount - nScan; n < nCount; n++ )
-    AnalyseLine(n);
-
-  #undef   AnalyseLine
-
-  // interpret the results (@@ far from being even 50% fool proof)
-  if ( nDos + nUnix + nMac == 0 ) {
-    // no newlines at all
-    wxLogWarning("'%s' is probably a binary file.", m_strFile.c_str());
-  }
-  else {
-    #define   GREATER_OF(t1, t2) n##t1 == n##t2 ? typeDefault               \
-                                                : n##t1 > n##t2 ? Type_##t1 \
-                                                                : Type_##t2
-
-    if ( nDos > nUnix )
-      return GREATER_OF(Dos, Mac);
-    else if ( nDos < nUnix )
-      return GREATER_OF(Unix, Mac);
-    else {
-      // nDos == nUnix
-      return nMac > nDos ? Type_Mac : typeDefault;
-    }
+    return m_file.Open(strBufferName.c_str(), FileOpenMode);
+}

 
 

-    #undef    GREATER_OF
-  }

 
 

-  return typeDefault;
+bool wxTextFile::OnClose()
+{
+    return m_file.Close();

 }
 
 }
 

-bool wxTextFile::Read()
+
+bool wxTextFile::OnRead(const wxMBConv& conv)

 {
 {

-  // file should be opened and we must be in it's beginning
-  wxASSERT( m_file.IsOpened() && m_file.Tell() == 0 );
-
-  wxString str;
-  char ch, chLast = '\0';
-  while ( !m_file.Eof() ) {
-    // @@ should really use a buffer for efficiency
-    if ( m_file.Read(&ch, sizeof(ch)) == ofsInvalid ) {
-      // read error
-      m_file.Close();
-      return FALSE;
+    // file should be opened
+    wxASSERT_MSG( m_file.IsOpened(), _T("can't read closed file") );
+
+    // read the entire file in memory: this is not the most efficient thing to
+    // do it but there is no good way to avoid it in Unicode build because if
+    // we read the file block by block we can't convert each block to Unicode
+    // separately (the last multibyte char in the block might be only partially
+    // read and so the conversion would fail) and, as the file contents is kept
+    // in memory by wxTextFile anyhow, it shouldn't be a big problem to read
+    // the file entirely
+    size_t bufSize = 0;
+
+    // number of bytes to (try to) read from disk at once
+    static const size_t BLOCK_SIZE = 4096;
+
+    wxCharBuffer buf;
+
+    // first determine if the file is seekable or not and so whether we can
+    // determine its length in advance
+    wxFileOffset fileLength;
+    {
+        wxLogNull logNull;
+        fileLength = m_file.Length();

     }
 
     }
 

-    #ifdef  __MAC__
-      #pragma message("wxTextFile::Read() hasn't been tested with Mac files.")
-    #endif
-
-    switch ( ch ) {
-      case '\n':
-        // Dos/Unix line termination
-        m_aLines.Add(str);
-        m_aTypes.Add(chLast == '\r' ? Type_Dos : Type_Unix);
-        str.Empty();
-        chLast = '\n';
-        break;
-
-      case '\r':
-        if ( chLast == '\r' ) {
-          // Mac empty line
-          m_aLines.Add("");
-          m_aTypes.Add(Type_Mac);
+    // some non-seekable files under /proc under Linux pretend that they're
+    // seekable but always return 0; others do return an error
+    const bool seekable = fileLength != wxInvalidOffset && fileLength != 0;
+    if ( seekable )
+    {
+        // we know the required length, so set the buffer size in advance
+        bufSize = fileLength;
+        if ( !buf.extend(bufSize) )
+            return false;
+
+        // if the file is seekable, also check that we're at its beginning
+        wxASSERT_MSG( m_file.Tell() == 0, _T("should be at start of file") );
+
+        char *dst = buf.data();
+        for ( size_t nRemaining = bufSize; nRemaining > 0; )
+        {
+            size_t nToRead = BLOCK_SIZE;
+
+            // the file size could have changed, avoid overflowing the buffer
+            // even if it did
+            if ( nToRead > nRemaining )
+                nToRead = nRemaining;
+
+            ssize_t nRead = m_file.Read(dst, nToRead);
+
+            if ( nRead == wxInvalidOffset )
+            {
+                // read error (error message already given in wxFile::Read)
+                return false;
+            }
+
+            if ( nRead == 0 )
+            {
+                // this file can't be empty because we checked for this above
+                // so this must be the end of file
+                break;
+            }
+
+            dst += nRead;
+            nRemaining -= nRead;

         }
         }

-        else
-          chLast = '\r';
-        break;
-
-      default:
-        if ( chLast == '\r' ) {
-          // Mac line termination
-          m_aLines.Add(str);
-          m_aTypes.Add(Type_Mac);
-          str = ch;
+
+        wxASSERT_MSG( dst - buf.data() == (wxFileOffset)bufSize,
+                      _T("logic error") );
+    }
+    else // file is not seekable
+    {
+        char block[BLOCK_SIZE];
+        for ( ;; )
+        {
+            ssize_t nRead = m_file.Read(block, WXSIZEOF(block));
+
+            if ( nRead == wxInvalidOffset )
+            {
+                // read error (error message already given in wxFile::Read)
+                return false;
+            }
+
+            if ( nRead == 0 )
+            {
+                // if no bytes have been read, presumably this is a
+                // valid-but-empty file
+                if ( bufSize == 0 )
+                    return true;
+
+                // otherwise we've finished reading the file
+                break;
+            }
+
+            // extend the buffer for new data
+            if ( !buf.extend(bufSize + nRead) )
+                return false;
+
+            // and append it to the buffer
+            memcpy(buf.data() + bufSize, block, nRead);
+            bufSize += nRead;

         }
         }

-        else {
-          // add to the current line
-          str += ch;
+    }
+
+    const wxString str(buf, conv, bufSize);
+
+    // there's no risk of this happening in ANSI build
+#if wxUSE_UNICODE
+    if ( bufSize > 4 && str.empty() )
+    {
+        wxLogError(_("Failed to convert file \"%s\" to Unicode."), GetName());
+        return false;
+    }
+#endif // wxUSE_UNICODE
+
+    // we don't need this memory any more
+    buf.reset();
+
+
+    // now break the buffer in lines
+
+    // last processed character, we need to know if it was a CR or not
+    wxChar chLast = '\0';
+
+    // the beginning of the current line, changes inside the loop
+    wxString::const_iterator lineStart = str.begin();
+    const wxString::const_iterator end = str.end();
+    for ( wxString::const_iterator p = lineStart; p != end; p++ )
+    {
+        const wxChar ch = *p;
+        switch ( ch )
+        {
+            case '\n':
+                // could be a DOS or Unix EOL
+                if ( chLast == '\r' )
+                {
+                    if ( p - 1 >= lineStart )
+                    {
+                        AddLine(wxString(lineStart, p - 1), wxTextFileType_Dos);
+                    }
+                    else
+                    {
+                        // there were two line endings, so add an empty line:
+                        AddLine(wxEmptyString, wxTextFileType_Dos);
+                    }
+                }
+                else // bare '\n', Unix style
+                {
+                    AddLine(wxString(lineStart, p), wxTextFileType_Unix);
+                }
+
+                lineStart = p + 1;
+                break;
+
+            case '\r':
+                if ( chLast == '\r' )
+                {
+                    // Mac empty line
+                    AddLine(wxEmptyString, wxTextFileType_Mac);
+                    lineStart = p + 1;
+                }
+                //else: we don't know what this is yet -- could be a Mac EOL or
+                //      start of DOS EOL so wait for next char
+                break;
+
+            default:
+                if ( chLast == '\r' )
+                {
+                    // Mac line termination
+                    if ( p - 1 >= lineStart )
+                    {
+                        AddLine(wxString(lineStart, p - 1), wxTextFileType_Mac);
+                    }
+                    else
+                    {
+                        // there were two line endings, so add an empty line:
+                        AddLine(wxEmptyString, wxTextFileType_Mac);
+                    }
+                    lineStart = p;
+                }

         }
         }

+
+        chLast = ch;

     }
     }

-  }

 
 

-  // anything in the last line?
-  if ( !str.IsEmpty() ) {
-    m_aTypes.Add(Type_None);  // no line terminator
-    m_aLines.Add(str);
-  }
+    // anything in the last line?
+    if ( lineStart != end )
+    {
+        // add unterminated last line
+        AddLine(wxString(lineStart, end), wxTextFileType_None);
+    }

 
 

-  return TRUE;
+    return true;

 }
 
 }
 

-bool wxTextFile::Write(Type typeNew)
+
+bool wxTextFile::OnWrite(wxTextFileType typeNew, const wxMBConv& conv)

 {
 {

-  wxTempFile fileTmp(m_strFile);
-
-  if ( !fileTmp.IsOpened() ) {
-    wxLogError("can't write file '%s' to disk.", m_strFile.c_str());
-    return FALSE;
-  }
-
-  uint nCount = m_aLines.Count();
-  for ( uint n = 0; n < nCount; n++ ) {
-    fileTmp.Write(m_aLines[n] + 
-                  GetEOL(typeNew == Type_None ? m_aTypes[n] : typeNew));
-  }
-
-  // replace the old file with this one
-  return fileTmp.Commit();
-}
\ No newline at end of file
+    wxFileName fn = m_strBufferName;
+
+    // We do NOT want wxPATH_NORM_CASE here, or the case will not
+    // be preserved.
+    if ( !fn.IsAbsolute() )
+        fn.Normalize(wxPATH_NORM_ENV_VARS | wxPATH_NORM_DOTS | wxPATH_NORM_TILDE |
+                     wxPATH_NORM_ABSOLUTE | wxPATH_NORM_LONG);
+
+    wxTempFile fileTmp(fn.GetFullPath());
+
+    if ( !fileTmp.IsOpened() ) {
+        wxLogError(_("can't write buffer '%s' to disk."), m_strBufferName.c_str());
+        return false;
+    }
+
+    size_t nCount = GetLineCount();
+    for ( size_t n = 0; n < nCount; n++ ) {
+        fileTmp.Write(GetLine(n) +
+                      GetEOL(typeNew == wxTextFileType_None ? GetLineType(n)
+                                                            : typeNew),
+                      conv);
+    }
+
+    // replace the old file with this one
+    return fileTmp.Commit();
+}
+
+#endif // wxUSE_TEXTFILE