[FFmpeg-devel] [PATCH] add colours to warnings and errors

[Michael Niedermayer]

On Sat, Apr 24, 2010 at 03:03:58AM +0200, James Darnley wrote:
> Attached is hopefully the final revision.
> It should not alter the behavior when _WIN32 is not defined at compile time.
> When it is, colour will be printed in:
>  - "plain" cmd.exe
>  - cygwin and msys cmd.exe terminals when "tty" is not in the CYGWIN
> environment variable
> 
> Other terminals in which you might run a native Windows executable
> look like they are SOL.  If they get changed so that isatty() returns
> true when you would expect it to then they should start working
> without any change to ffmpeg.
> 
> If someone wanted to try and make a hack for isatty() then I suggest
> this as a starting point:
> http://www.cygwin.com/ml/cygwin/2007-03/msg00663.html
> Some state of the stderr pipe might let you find if it is attached to
> a cygwin tty.  I tried to recreate that myself but I wasn't very
> successful.  I also think that I showed that you can't know whether it
> is rxvt or another program using the pipe, perl or python for example.

>  log.c |   69 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 69 insertions(+)
> 7b72f819a31ca8429b43432c8a57be878f1ea3ec  colours.diff
> Index: libavutil/log.c
> ===================================================================
> --- libavutil/log.c	(revision 22957)
> +++ libavutil/log.c	(working copy)
> @@ -24,6 +24,10 @@
>   * logging functions
>   */
>  
> +#ifdef _WIN32
> +#include <windows.h>
> +#include <string.h>
> +#endif
>  #include <unistd.h>
>  #include <stdlib.h>
>  #include "avutil.h"
> @@ -35,9 +39,73 @@
>  int av_log_level = AV_LOG_INFO;
>  
>  static int use_ansi_color=-1;

> +#ifdef _WIN32
> +static int use_win_color=-1;
> +static int16_t attr, attr_orig;
> +static HANDLE con;
> +#endif
>  
> +/* FIXME: On Windows isatty() returns true when ANSI color codes won't work.
> +Some hack to detect output to other terminals would be good, fixing the other
> +terminals would be better. One probable exception is when the user has
> +ANSI.SYS loaded but the Windows API should then still work. */
> +
>  #undef fprintf
>  static void colored_fputs(int color, const char *str){
> +#ifdef _WIN32

these 2 ifdefs can be merged


> +    static const int16_t win_color_conv[] = {
> +        0,
> +        FOREGROUND_RED,
> +        FOREGROUND_GREEN,
> +        FOREGROUND_RED  |FOREGROUND_GREEN,
> +        FOREGROUND_BLUE,
> +        FOREGROUND_RED  |FOREGROUND_BLUE,
> +        FOREGROUND_GREEN|FOREGROUND_BLUE,
> +        FOREGROUND_RED  |FOREGROUND_GREEN|FOREGROUND_BLUE
> +    };
> +    if (use_ansi_color<0 || use_win_color<0) {
> +        CONSOLE_SCREEN_BUFFER_INFO con_info;
> +
> +        use_ansi_color = isatty(2) && getenv("TERM") && !getenv("NO_COLOR")
> +            && (   !strncmp(getenv("TERM"), "msys",   4)
> +                || !strncmp(getenv("TERM"), "xterm",  5)
> +                || !strncmp(getenv("TERM"), "rxvt",   4)
> +                //|| !strncmp(getenv("TERM"), "cygwin", 6)
> +/* The CYGWIN environment variable makes this hard for native executables.
> +notty -- programs are directly connected to cmd so the Windows API works
> +tty -- programs are not directly connected so ANSI color codes work
> +The default is notty so leaving "cygwin" excluded doesn't cause problems. */
> +            );
> +
> +        con = GetStdHandle(STD_ERROR_HANDLE);
> +        if (con!=INVALID_HANDLE_VALUE) {
> +            GetConsoleScreenBufferInfo(con, &con_info);
> +            attr_orig = con_info.wAttributes;
> +
> +            attr = (attr_orig&0xF0);
> +            attr |= (attr_orig&BACKGROUND_INTENSITY)?0:FOREGROUND_INTENSITY;
> +
> +            use_win_color = 1;
> +        } else
> +            use_win_color = 0;
> +    }
> +
> +    if ((color&15)==9) {
> +        fputs(str, stderr);
> +        return;
> +    }

> +    if (use_ansi_color)
> +        fprintf( stderr, "\033[%d;3%dm", color>>4, color&15 );
> +    else if (use_win_color)
> +        SetConsoleTextAttribute( con, attr | win_color_conv[color&15] );
> +
> +    fputs(str, stderr);
> +
> +    if (use_ansi_color)
> +        fprintf( stderr, "\033[0m" );
> +    else if (use_win_color)
> +        SetConsoleTextAttribute( con, attr_orig );
> +#else
>      if(use_ansi_color<0){
>  #if HAVE_ISATTY && !defined(_WIN32)
>          use_ansi_color= getenv("TERM") && !getenv("NO_COLOR") && isatty(2);
> @@ -53,6 +121,7 @@
>      if(use_ansi_color){
>          fprintf(stderr, "\033[0m");
>      }
> +#endif

    if(use_win_color){
    ...
    }else
#else
    {
        if(use_ansi_color<0){
    #if HAVE_ISATTY && !defined(_WIN32)
            use_ansi_color= getenv("TERM") && !getenv("NO_COLOR") && isatty(2);
@@ -53,6 +121,7 @@
        if(use_ansi_color){
            fprintf(stderr, "\033[0m");
        }
    }
#endif

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

If a bugfix only changes things apparently unrelated to the bug with no
further explanation, that is a good sign that the bugfix is wrong.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20100424/04f589f2/attachment.pgp>