X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/ef8d96c29c73b9da8cdaa2b8b4737862c85ccc20..5769cf0f89fa4d076042e9f8dd24a32501280f8c:/src/common/textfile.cpp?ds=sidebyside diff --git a/src/common/textfile.cpp b/src/common/textfile.cpp index 5c74b3ab57..ef27836e89 100644 --- a/src/common/textfile.cpp +++ b/src/common/textfile.cpp @@ -1,335 +1,291 @@ /////////////////////////////////////////////////////////////////////////////// -// Name: textfile.cpp +// Name: src/common/textfile.cpp // Purpose: implementation of wxTextFile class // Author: Vadim Zeitlin // Modified by: // Created: 03.04.98 // RCS-ID: $Id$ // Copyright: (c) 1998 Vadim Zeitlin -// Licence: wxWindows license +// Licence: wxWindows licence /////////////////////////////////////////////////////////////////////////////// // ============================================================================ // headers // ============================================================================ -#ifdef __GNUG__ - #pragma implementation "textfile.h" -#endif - #include "wx/wxprec.h" #ifdef __BORLANDC__ #pragma hdrstop #endif //__BORLANDC__ -#if !wxUSE_FILE +#if !wxUSE_FILE || !wxUSE_TEXTBUFFER #undef wxUSE_TEXTFILE #define wxUSE_TEXTFILE 0 #endif // wxUSE_FILE +#if wxUSE_TEXTFILE + #ifndef WX_PRECOMP - #include "wx/string.h" - #include "wx/intl.h" - #include "wx/file.h" - #include "wx/log.h" + #include "wx/string.h" + #include "wx/intl.h" + #include "wx/file.h" + #include "wx/log.h" #endif -#include "wx/textfile.h" +#include "wx/textfile.h" +#include "wx/filename.h" +#include "wx/buffer.h" // ============================================================================ // wxTextFile class implementation // ============================================================================ +wxTextFile::wxTextFile(const wxString& strFileName) + : wxTextBuffer(strFileName) +{ +} + + // ---------------------------------------------------------------------------- -// static methods (always compiled in) +// file operations // ---------------------------------------------------------------------------- -// default type is the native one -const wxTextFileType wxTextFile::typeDefault = -#if defined(__WINDOWS__) - wxTextFileType_Dos; -#elif defined(__UNIX__) - wxTextFileType_Unix; -#elif defined(__WXMAC__) - wxTextFileType_Mac; -#elif defined(__WXPM__) - wxTextFileType_Os2; -#else - wxTextFileType_None; - #error "wxTextFile: unsupported platform." -#endif - -const wxChar *wxTextFile::GetEOL(wxTextFileType type) +bool wxTextFile::OnExists() const { - switch ( type ) { - default: - wxFAIL_MSG(wxT("bad file type in wxTextFile::GetEOL.")); - // fall through nevertheless - we must return something... - - case wxTextFileType_None: return wxT(""); - case wxTextFileType_Unix: return wxT("\n"); - case wxTextFileType_Dos: return wxT("\r\n"); - case wxTextFileType_Mac: return wxT("\r"); - } + return wxFile::Exists(m_strBufferName); } -wxString wxTextFile::Translate(const wxString& text, wxTextFileType type) +bool wxTextFile::OnOpen(const wxString &strBufferName, wxTextBufferOpenMode OpenMode) { - // don't do anything if there is nothing to do - if ( type == wxTextFileType_None ) - return text; - - wxString eol = GetEOL(type), result; - - // optimization: we know that the length of the new string will be about - // the same as the length of the old one, so prealloc memory to aviod - // unnecessary relocations - result.Alloc(text.Len()); + wxFile::OpenMode FileOpenMode; - wxChar chLast = 0; - for ( const wxChar *pc = text.c_str(); *pc; pc++ ) + switch ( OpenMode ) { - wxChar ch = *pc; - switch ( ch ) { - case '\n': - // Dos/Unix line termination - result += eol; - chLast = '\n'; - break; + default: + wxFAIL_MSG( _T("unknown open mode in wxTextFile::Open") ); + // fall through - case '\r': - if ( chLast == '\r' ) { - // Mac empty line - result += eol; - } - else - chLast = '\r'; - break; + case ReadAccess : + FileOpenMode = wxFile::read; + break; - default: - if ( chLast == '\r' ) { - // Mac line termination - result += eol; - chLast = ch; - } - else { - // add to the current line - result += ch; - } - } + case WriteAccess : + FileOpenMode = wxFile::write; + break; } - return result; + return m_file.Open(strBufferName.c_str(), FileOpenMode); } -#if wxUSE_TEXTFILE - -// ---------------------------------------------------------------------------- -// ctors & dtor -// ---------------------------------------------------------------------------- - -wxTextFile::wxTextFile(const wxString& strFile) : m_strFile(strFile) -{ - m_nCurLine = 0; - m_isOpened = FALSE; -} -wxTextFile::~wxTextFile() +bool wxTextFile::OnClose() { - // m_file dtor called automatically + return m_file.Close(); } -// ---------------------------------------------------------------------------- -// file operations -// ---------------------------------------------------------------------------- -bool wxTextFile::Exists() const +bool wxTextFile::OnRead(const wxMBConv& conv) { - return wxFile::Exists(m_strFile); -} + // 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 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, + bufPos = 0; + char block[1024]; + 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(); + } -bool wxTextFile::Open(const wxString& strFile) -{ - m_strFile = strFile; + // 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 - 1 /* it adds 1 internally */) ) + return false; - return Open(); -} + // 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") ); + } -bool wxTextFile::Open() -{ - // file name must be either given in ctor or in Open(const wxString&) - wxASSERT( !m_strFile.IsEmpty() ); + for ( ;; ) + { + ssize_t nRead = m_file.Read(block, WXSIZEOF(block)); - // open file in read-only mode - if ( !m_file.Open(m_strFile) ) - return FALSE; + if ( nRead == wxInvalidOffset ) + { + // read error (error message already given in wxFile::Read) + return false; + } - // read file into memory - m_isOpened = Read(); + if ( nRead == 0 ) + { + // if no bytes have been read, presumably this is a valid-but-empty file + if ( bufPos == 0 ) + return true; - m_file.Close(); + // otherwise we've finished reading the file + break; + } - return m_isOpened; -} + if ( seekable ) + { + // this shouldn't happen but don't overwrite the buffer if it does + wxCHECK_MSG( bufPos + nRead <= bufSize, false, + _T("read more than file length?") ); + } + else // !seekable + { + // for non-seekable files we have to allocate more memory on the go + if ( !buf.extend(bufPos + nRead - 1 /* it adds 1 internally */) ) + return false; + } -// analyse some lines of the file trying to guess it's type. -// if it fails, it assumes the native type for our platform. -wxTextFileType 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 - size_t 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) - size_t nCount = m_aLines.Count() / 3, - nScan = nCount > 3*MAX_LINES_SCAN ? MAX_LINES_SCAN : nCount / 3; - - #define AnalyseLine(n) \ - switch ( m_aTypes[n] ) { \ - case wxTextFileType_Unix: nUnix++; break; \ - case wxTextFileType_Dos: nDos++; break; \ - case wxTextFileType_Mac: nMac++; break; \ - default: wxFAIL_MSG(_("unknown line terminator")); \ + // append to the buffer + memcpy(buf.data() + bufPos, block, nRead); + bufPos += nRead; } - size_t 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 (FIXME 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 \ - ? wxTextFileType_##t1 \ - : wxTextFileType_##t2 - - // Watcom C++ doesn't seem to be able to handle the macro -#if !defined(__WATCOMC__) - if ( nDos > nUnix ) - return GREATER_OF(Dos, Mac); - else if ( nDos < nUnix ) - return GREATER_OF(Unix, Mac); - else { - // nDos == nUnix - return nMac > nDos ? wxTextFileType_Mac : typeDefault; + if ( !seekable ) + { + bufSize = bufPos; } -#endif // __WATCOMC__ - #undef GREATER_OF - } + const wxString str(buf, conv, bufPos); - return typeDefault; -} - -bool wxTextFile::Read() -{ - // 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'; - char buf[1024]; - int n, nRead; - while ( !m_file.Eof() ) { - nRead = m_file.Read(buf, WXSIZEOF(buf)); - if ( nRead == wxInvalidOffset ) { - // read error (error message already given in wxFile::Read) - return FALSE; + // 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 - for ( n = 0; n < nRead; n++ ) { - ch = buf[n]; - switch ( ch ) { - case '\n': - // Dos/Unix line termination - m_aLines.Add(str); - m_aTypes.Add(chLast == '\r' ? wxTextFileType_Dos - : wxTextFileType_Unix); - str.Empty(); - chLast = '\n'; - break; - - case '\r': - if ( chLast == '\r' ) { - // Mac empty line - m_aLines.Add(""); - m_aTypes.Add(wxTextFileType_Mac); - } - else - chLast = '\r'; - break; + free(buf.release()); // we don't need this memory any more - default: - if ( chLast == '\r' ) { - // Mac line termination - m_aLines.Add(str); - m_aTypes.Add(wxTextFileType_Mac); - chLast = ch; - str = ch; - } - else { - // add to the current line - str += ch; - } - } + + // 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(wxTextFileType_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::Close() + +bool wxTextFile::OnWrite(wxTextFileType typeNew, const wxMBConv& conv) { - m_aTypes.Clear(); - m_aLines.Clear(); - m_nCurLine = 0; - m_isOpened = FALSE; + wxFileName fn = m_strBufferName; - return TRUE; -} + // 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); -bool wxTextFile::Write(wxTextFileType typeNew) -{ - wxTempFile fileTmp(m_strFile); - - if ( !fileTmp.IsOpened() ) { - wxLogError(_("can't write file '%s' to disk."), m_strFile.c_str()); - return FALSE; - } - - size_t nCount = m_aLines.Count(); - for ( size_t n = 0; n < nCount; n++ ) { - fileTmp.Write(m_aLines[n] + - GetEOL(typeNew == wxTextFileType_None ? m_aTypes[n] - : typeNew)); - } - - // replace the old file with this one - return fileTmp.Commit(); + 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 -