[PATCH v3] Fixes CJK wide character display

[David Herrmann]

Hi

On Wed, Aug 14, 2013 at 2:30 AM, Kristian Høgsberg <hoegsberg at gmail.com> wrote:
> On Thu, Aug 01, 2013 at 01:49:03PM +0800, Peng Wu wrote:
>> By jumping two columns when wide character prints,
>> and draw wide cursor under wide character.
>
> Hi Peng,
>
> I do want this feature in weston-terminal, but there's still a few
> issues to address - see below.
>
>> ---
>>  clients/Makefile.am |  2 +-
>>  clients/terminal.c  | 26 ++++++++++++++++++++++++--
>>  2 files changed, 25 insertions(+), 3 deletions(-)
>>
>> diff --git a/clients/Makefile.am b/clients/Makefile.am
>> index 09963cc..e7c0ece 100644
>> --- a/clients/Makefile.am
>> +++ b/clients/Makefile.am
>> @@ -113,7 +113,7 @@ weston_screenshooter_SOURCES =                    \
>>  weston_screenshooter_LDADD = libtoytoolkit.la
>>
>>  weston_terminal_SOURCES = terminal.c
>> -weston_terminal_LDADD = libtoytoolkit.la -lutil
>> +weston_terminal_LDADD = libtoytoolkit.la -lutil $(PANGO_LIBS)
>
> Again, I don't want to link to pango or glib... remember, the only
> reason for weston-terminal to exist is that it doesn't have any
> dependencies except cairo and the wayland libraries.
>
>>  weston_image_SOURCES = image.c
>>  weston_image_LDADD = libtoytoolkit.la
>> diff --git a/clients/terminal.c b/clients/terminal.c
>> index 0d4f726..dc56745 100644
>> --- a/clients/terminal.c
>> +++ b/clients/terminal.c
>> @@ -33,6 +33,10 @@
>>  #include <ctype.h>
>>  #include <cairo.h>
>>  #include <sys/epoll.h>
>> +#include <glib.h>
>> +#define __USE_XOPEN
>> +#include <wchar.h>
>> +#include <locale.h>
>>
>>  #include <wayland-client.h>
>>
>> @@ -93,6 +97,11 @@ union utf8_char {
>>       uint32_t ch;
>>  };
>>
>> +static bool is_wide(union utf8_char utf8){
>> +     gunichar unichar = g_utf8_get_char((const char*) utf8.byte);
>> +     return wcwidth(unichar) > 1;
>> +}
>
> ... but I just committed a patch to make the utf-8 state machine
> assemble the unicode value from the utf-8, so you can now use
> get_unicode() instead of g_utf8_get_char().
>
> Please follow the coding conventions and leave 'static int' on it's
> own line and put the opening '{' of the function on it's own line.
>
>>  enum utf8_state {
>>       utf8state_start,
>>       utf8state_accept,
>> @@ -942,6 +951,7 @@ redraw_handler(struct widget *widget, void *data)
>>       struct glyph_run run;
>>       cairo_font_extents_t extents;
>>       double average_width;
>> +     double unichar_width;
>>
>>       surface = window_get_surface(terminal->window);
>>       widget_get_allocation(terminal->widget, &allocation);
>> @@ -967,6 +977,7 @@ redraw_handler(struct widget *widget, void *data)
>>                       allocation.y + top_margin);
>>       /* paint the background */
>>       for (row = 0; row < terminal->height; row++) {
>> +             p_row = terminal_get_row(terminal, row);
>>               for (col = 0; col < terminal->width; col++) {
>>                       /* get the attributes for this character cell */
>>                       terminal_decode_attr(terminal, row, col, &attr);
>> @@ -974,12 +985,17 @@ redraw_handler(struct widget *widget, void *data)
>>                       if (attr.attr.bg == terminal->color_scheme->border)
>>                               continue;
>>
>> +                     if(is_wide(p_row[col]))
>> +                             unichar_width = 2 * average_width;
>> +                     else
>> +                             unichar_width = average_width;
>> +
>>                       terminal_set_color(terminal, cr, attr.attr.bg);
>>                       cairo_move_to(cr, col * average_width,
>>                                     row * extents.height);
>> -                     cairo_rel_line_to(cr, average_width, 0);
>> +                     cairo_rel_line_to(cr, unichar_width, 0);
>>                       cairo_rel_line_to(cr, 0, extents.height);
>> -                     cairo_rel_line_to(cr, -average_width, 0);
>> +                     cairo_rel_line_to(cr, -unichar_width, 0);
>>                       cairo_close_path(cr);
>>                       cairo_fill(cr);
>>               }
>> @@ -1871,6 +1887,10 @@ handle_char(struct terminal *terminal, union utf8_char utf8)
>>       row[terminal->column] = utf8;
>>       attr_row[terminal->column++] = terminal->curr_attr;
>>
>> +     /* cursor jump for wide character. */
>> +     if (is_wide(utf8))
>> +             row[terminal->column++].ch = 0;
>
> Inserting this 0 here confuses the selection logic which thinks 0
> means "end of line".  I think we can use an invalid utf-8 character
> instead to indicate that this cell is a wide-char filler.  We then
> also need to handle that in terminal_send_selection(), where we must
> ignore those filler cells when writing the selection to the fd.
>
>
>>       if (utf8.ch != terminal->last_char.ch)
>>               terminal->last_char = utf8;
>>  }
>> @@ -2711,6 +2731,8 @@ int main(int argc, char *argv[])
>>       struct terminal *terminal;
>>       int config_fd;
>>
>> +     setlocale(LC_ALL, "");
>
> What is this for?

This is to work around the horrible glibc API. wcwidth() depends on
LC_CTYPE, that is, LC_ALL. I avoided forcing any locale and added my
own wcwidth() implementation to kmscon:
  https://github.com/dvdhrm/kmscon/blob/master/external/wcwidth.c
Sadly, there is no way to get access to the underlying ucs4_wcwidth()
in glibc (who designs such ****?). So either force a locale or feel
free to copy wcwidth.c.

Regards
David