Beefy Boxes and Bandwidth Generously Provided by pair Networks
Perl: the Markov chain saw
 
PerlMonks  

Printing Unicode on the Windows Console and the importance of of i/o layers

by nikosv (Hermit)
on Nov 17, 2010 at 07:23 UTC ( #871945=perlmeditation: print w/ replies, xml ) Need Help??

I wanted to take a look on printing Unicode on the windows console by using the Win32 api and also check how is done in other languages, rather than directly from Perl which hides a lot of details

Problems when wanting to print to the console :
1.The windows console uses an internal buffer that can mangle output
2.Invoking the console using the Unicode switch (cmd.exe /u) does not have an effect
3.Windows supports UTF-16 inherently, not utf8
4.Documentation on Unicode and the console is hard to find.MSDN library, as usual, is a labyrinth with no beginning and end where you can loose track easily

The need arose when I needed to print an old style dos box using the cp437 box drawing characters on the console using their Unicode code points rather than their ASCII representation. The output was mangled/overlapped

Take a look at this pictorial output to get a clear view of the problem

The code that generated the incorrect result is :

#unicode_box_incorrect.pl use Win32::API; binmode(STDOUT,':utf8'); #Must set the console code page to UTF8 $SetConsoleOutputCP= new Win32::API( 'kernel32.dll', 'SetConsoleOutput +CP', 'N','N' ); $SetConsoleOutputCP->Call(65001); $line1="\x{2554}".("\x{2550}"x15)."\x{2557}\n"; $line2="\x{2551}".(" "x15)."\x{2551}\n"; $line3="\x{255A}".("\x{2550}"x15)."\x{255D}"; $unicode_string=$line1.$line2.$line3; print "THIS IS THE INCORRECT EXAMPLE OUTPUT: \n"; print $unicode_string;

Since C++ has a better relationship with Windows than Perl, I did some research on how you can manipulate the console in C++ and used the underlying concepts in Perl.

Fortunately I bumped into illegalargumentexception who has a fantastic tutorial on the subject using multi-language examples. Also the blog explains various issues on Unicode. great stuff, totally recommended

So the equivalent Perl code would be:

#unicode_box_correct.pl use Win32::API; use Encode qw(from_to encode); #no need to use perlio (in this case) as we are bypassing it through t +he raw Win32API #binmode(STDOUT,':utf8'); #Set the console code page to UTF8 $SetConsoleOutputCP= new Win32::API( 'kernel32.dll', 'SetConsoleOutput +CP', 'N','N' ); $SetConsoleOutputCP->Call(65001); #Get a reference to the console STDOUT $GetStdHandle=new Win32::API( 'kernel32.dll', 'GetStdHandle', 'N', 'N' + ); $handle=$GetStdHandle->Call(-11); #Build dos window $line1="\x{2554}".("\x{2550}"x15)."\x{2557}\n"; $line2="\x{2551}".(" "x15)."\x{2551}\n"; $line3="\x{255A}".("\x{2550}"x15)."\x{255D}"; $unicode_string=$line1.$line2.$line3; print "THIS IS THE CORRECT EXAMPLE OUTPUT: \n"; #Force byte semantics because WriteFile API function needs length in b +ytes not characters $lengthx=length(Encode::encode_utf8($unicode_string)); #use WriteFile API to treat the Console as a file.WriteConsole won't d +o it $WriteFile=new Win32::API( 'kernel32.dll', 'WriteFile', 'NPNNN', 'N' ) +; $WriteFile->Call($handle,$unicode_string, $lengthx,0,0);

The trick is to use high-level console I/O (WriteFile) rather than low-level console I/O (WriteConsoleOutput) and there is no need to use the WideCharToMultiByte function since Perl uses UTF8 natively while C++ uses 16bit wide chars which need to be converted into multibytes. Note here that Windows treats the wchar as 'real' Unicode while it treats Utf8 as a multibyte encoding, the same as treating ASCII code pages.

Also note that for the example to work, the actual code page of the console does not play a role but the font must be set to Lucida console. However the Lucida Console font does not support the whole Unicode range, so it does not include all Unicode glyphs. There is only one issue, how to programmatically set the font on the users' console. This, unfortunately, can only be done on Windows vista and upwards with the SetCurrentConsoleFontEx api function

Ultimately, in pure Perl code without using any Win32 API's (although we still need it for the SetConsoleOutputCP), we turn perio buffering off by using the :unix layer, so it doesn't mess with the console buffer :

#unicode_box_correct_pure_perl.pl use Win32::API; binmode(STDOUT, ":unix:utf8"); #Must set the console code page to UTF8 $SetConsoleOutputCP= new Win32::API( 'kernel32.dll', 'SetConsoleOutput +CP', 'N','N' ); $SetConsoleOutputCP->Call(65001); $line1="\x{2554}".("\x{2550}"x15)."\x{2557}\n"; $line2="\x{2551}".(" "x15)."\x{2551}\n"; $line3="\x{255A}".("\x{2550}"x15)."\x{255D}"; $unicode_string=$line1.$line2.$line3; print "THIS IS THE CORRECT EXAMPLE OUTPUT IN PURE PERL: \n"; print $unicode_string;

Compare this little Perl example with the complexity the other languages have to go through to get to the same result and appreciate Perl's power. magic.

Comment on Printing Unicode on the Windows Console and the importance of of i/o layers
Select or Download Code
Re: Printing Unicode on the Windows Console and the importance of of i/o layers
by BrowserUk (Pope) on Nov 17, 2010 at 07:50 UTC

    Rather than messing with Win32::API (which I've never gotten to work for 64-bit), you can use Win32::Console::OutputCP(). This produces the required output:

    use Win32::Console; binmode(STDOUT, ":unix:utf8"); Win32::Console::OutputCP( 65001 ); $line1 = "\x{2554}".("\x{2550}"x15)."\x{2557}\n"; $line2 = "\x{2551}".(" "x15)."\x{2551}\n"; $line3 = "\x{255A}".("\x{2550}"x15)."\x{255D}"; $unicode_string = $line1.$line2.$line3; print "THIS IS THE CORRECT EXAMPLE OUTPUT IN PURE PERL: \n"; print $unicode_string;

    Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
    "Science is about questioning the status quo. Questioning authority".
    In the absence of evidence, opinion is indistinguishable from prejudice.
      yep,good observation
Reaped: Re: Printing Unicode on the Windows Console and the importance of of i/o layers
by NodeReaper (Curate) on Nov 22, 2010 at 07:48 UTC

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: perlmeditation [id://871945]
Approved by McDarren
Front-paged by Arunbear
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others imbibing at the Monastery: (4)
As of 2014-12-20 08:42 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    Is guessing a good strategy for surviving in the IT business?





    Results (95 votes), past polls