writing text to a file

[bkelly]

Windows XP Pro, Visual Studio 2008, C++, MFC

When writing text to a file it ignores the \n, %d, and puts spaces between all the characters.
Here is the open code:
Please excuse typos because I cannot cut and paste from my work computer to here.

m_log_handle = CreateFile(
   L"C:\\TEMP\\log_file.txt",
  FILE_WRITE_DATA,
  FILE_SHARE_WRITE,
  NULL,
  FILE_ATTRIBUTE_NORMAL,
  NULL );

An example of write code is:

CString file_string.format( L"\nport is %d", port_number );
int length = file_string.GetLength();  // Length looks good
WriteFile(  m_log_handle,
                 file_string,
                 length,
                 mp_bytes_written,
                NULL );

The debugger shows file_string contains "xport is 49000"
But replace the leading x with a little graphics square.

The port number is completely missing and the \n is displayed as a little box. No new line. I expect to see the little graphics symbol in the debugger, but not in the text file.

[VictorN]

Is your build a UNICODE?

[bkelly]

I think it is. When I look in the somewhere in project properties there is a field that says:

Character Set Use Unicode Character Set.

What would the answer be for Yes and for No.

This is a large MFC project that will never leave a closed in laboratory. Is it difficult to change to non-unicode?

Thanks for your reply.

[S_M_A]

No it's not difficult to change but since you have shown strings that's specified as L"xxx" I guess there are a lot of them and you have to find them all and change to non-unicode or _T"xxx".

How do you open the file that's created? I checked CreateFile at msdn but couldn't find anything that indicates that it's capable of handling the translation of \n into \r\n that is normal in windows text-files.

Opening it in MSVC should work since it's capable of handling files with *nix line endings (I've never tried unicode files though)

[bkelly]

The output file is written to disk as a .txt file and I just double click it to open notepad or wordpad. Is there a way to tell the write utility to not use unicode.

And what about the number. The value is 49000 so why don't I see "4 9 0 0 0" in the output? The debugger shows the characters "49000" in the string.

I would like to not use unicode, but this application interfaces with another application that I cannot change. I suspect that application uses unicode.

[Paul McKenzie]

The output file is written to disk as a .txt file and I just double click it to open notepad or wordpad. Is there a way to tell the write utility to not use unicode.

You use the proper non-Unicode version of the functions you're calling. In other words "CreateFileA" instead of CreateFile. From MSDN:

http://msdn.microsoft.com/en-us/libr...=vs.85%29.aspx

Unicode and ANSI names
CreateFileW (Unicode) and CreateFileA (ANSI)

And what about the number. The value is 49000 so why don't I see "4 9 0 0 0" in the output? The debugger shows the characters "49000" in the string.

That is because the debugger is smart and knows how to display Unicode characters. If you want the raw byte version of the string, use the Memory window to view the data.

Regards,

Paul McKenzie

[bkelly]

I think that is what I am looking for. I will code it up tomorrow.
Thank you for taking the time to post.

[VictorN]

I think it is. When I look in the somewhere in project properties there is a field that says:

Character Set Use Unicode Character Set.
...

Then your code seems to be incorrect:

CString file_string.format( L"\nport is %d", port_number );
int length = file_string.GetLength();  // Length looks good
WriteFile(  m_log_handle,
                 file_string,
                 length,
                 mp_bytes_written,
                NULL );

The third parameter of WriteFile is the number of bytes to write while you are passing in the number of wide characters in the CString which is twice less!

[krmed]

The problem is not whether the program is Unicode or not, it's just that you are writing only a line feed (\n) to the file:

CString file_string.format( L"\nport is %d", port_number );

For Windows, text files should have a carriage return (\r) and line feed (\n) at the end of each line.

Some methods allow the automatic translation when the file is written, but CreateFile does not give that option, so you have to do it yourself. Simply replace your code above with

CString file_string.format( L"\r\nport is %d", port_number );

Hope this helps.

[bkelly]

You use the proper non-Unicode version of the functions you're calling. In other words "CreateFileA" instead of CreateFile. From MSDN:

http://msdn.microsoft.com/en-us/libr...=vs.85%29.aspx

That is because the debugger is smart and knows how to display Unicode characters. If you want the raw byte version of the string, use the Memory window to view the data.

Regards,

Paul McKenzie

I switched from CreateFile to CreateFileA and did not see any difference at all.

[bkelly]

Enough! Make that too much!

I switched to the old concept:

char debug_text[ MAX_SIZE ];
sprintf( debug_text, "text %d", number );

and when the compiler fussed at me, switched again to:

sprintf_s( debug_text, MAX_SIZE, "\r\n text %d", number );

Suprisingly, the writefile function accepts the char *.