Bug 839

Created attachment 877 [details]
cal produces garbled week-day names

When locale is zh_TW.utf8 or ja_JP.utf8, cal produces garbled week-day names.
In addition, this attachment shows version of BusyBox, output of date and a portion of output of locale.

I see this:

# LANG=zh_TW.utf8 cal
      一月 2010     
日 一 二 三 四 五 六
                1  2
 3  4  5  6  7  8  9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30
31

What do you see with "standard" cal?

My machine does not have right fonts to shot those glyphs, so I have a question: do week day names take 2 char positions per hieroglyph?

The responsible code in cal.c is:

        i = 0;
        do {
                zero_tm.tm_mon = i;
                /* full month name according to locale */
                strftime(buf, sizeof(buf), "%B", &zero_tm);
                month_names[i] = xstrdup(buf);

                if (i < 7) {
                        zero_tm.tm_wday = i;
                        /* abbreviated weekday name according to locale */
                        strftime(buf, sizeof(buf), "%a", &zero_tm);
====>===BAD======>      strncpy(day_headings + i * (3+julian) + julian, buf, 2);
                }
        } while (++i < 12);

Arrow indicates code which simply remembers 2 BYTES from every weekday name. That's much.

Just converting it to take two *unicode chars* instead would be not so hard, but I need to know whether these unicode chars (in this case more like hieroglyphs) are twice as wide and ACSII chars? If yes, we'd need the code to account for this.

Dear Sir,
     I am sorry for my reply being late.
     The unicode chars are twice as wide as ASCII chars.
     By the way, there are many GPL TTFs containing Chinese chars.  For example, FireFly TTF (http://cle.linux.org.tw/fonts/FireFly/fireflysung-1.3.0.tar.gz) is one of them.

(In reply to comment #2)
> I see this:
> 
> # LANG=zh_TW.utf8 cal
>       一月 2010     
> 日 一 二 三 四 五 六
>                 1  2
>  3  4  5  6  7  8  9
> 10 11 12 13 14 15 16
> 17 18 19 20 21 22 23
> 24 25 26 27 28 29 30
> 31
> 
> What do you see with "standard" cal?
> 
> My machine does not have right fonts to shot those glyphs, so I have a
> question: do week day names take 2 char positions per hieroglyph?
> 
> The responsible code in cal.c is:
> 
>         i = 0;
>         do {
>                 zero_tm.tm_mon = i;
>                 /* full month name according to locale */
>                 strftime(buf, sizeof(buf), "%B", &zero_tm);
>                 month_names[i] = xstrdup(buf);
> 
>                 if (i < 7) {
>                         zero_tm.tm_wday = i;
>                         /* abbreviated weekday name according to locale */
>                         strftime(buf, sizeof(buf), "%a", &zero_tm);
> ====>===BAD======>      strncpy(day_headings + i * (3+julian) + julian, buf,
> 2);
>                 }
>         } while (++i < 12);
> 
> Arrow indicates code which simply remembers 2 BYTES from every weekday name.
> That's much.
> 
> Just converting it to take two *unicode chars* instead would be not so hard,
> but I need to know whether these unicode chars (in this case more like
> hieroglyphs) are twice as wide and ACSII chars? If yes, we'd need the code to
> account for this.
> 

Created attachment 973 [details]
Fix

This fix is comitted to git

1.16.0 is released and it has this fixed. Please test it and let me know if it does not work.

(Double check that you enabled unicode support in .config)

Created attachment 1015 [details]
cal produces correct week-day names when locale is ja_JP.utf8 or zh_TW.utf8

Fixed in 1.16.0