Re: hu.po ngettext fixes

[Paul Eggert]

In the thread rooted at
<http://lists.gnu.org/archive/html/bug-coreutils/2006-08/msg00082.html>
Koblinger Egmont <address@hidden> writes:

> What is the correct approach if the number isn't presented in the string, as
> in his "This item" vs. "These items" case? Should ngettext be used in this
> case? Or this piece of code a good programming approach:
> if (n == 1) printf(_("This item...")); else printf(_("These items..."));  ?

That's a tricky question, so I'll forward this email to the gettext
maintainer.

I just now reviewed the gettext manual and led me to some other
translation glitches in coreutils.  I installed the following patch to
coreutils.  This sort of thing should be documented in the gettext
manual, and I'll follow up with another message about that.

2006-08-16  Paul Eggert  <address@hidden>

        Fix bugs when printing plurals of numbers that are not
        unsigned long int values.
        * src/system.h (select_plural): New function.
        * src/md5sum.c (digest_check): Use select_plural to avoid bug.
        * src/uptime.c (print_uptime): Likewise.
        * src/dd.c (print_stats): Likewise.  Also, don't use ngettext to
        print a floating point number, as reducing to 0 or 1 doesn't work
        for some languages.  Instead, just use "s" for seconds since it
        doesn't need a plural form.

Index: src/system.h
===================================================================
RCS file: /fetish/cu/src/system.h,v
retrieving revision 1.150
diff -p -u -r1.150 system.h
--- src/system.h        11 Jul 2006 17:22:15 -0000      1.150
+++ src/system.h        16 Aug 2006 19:34:55 -0000
@@ -383,6 +383,14 @@ static inline unsigned char to_uchar (ch
 #define _(msgid) gettext (msgid)
 #define N_(msgid) msgid
 
+/* Return a value that pluralizes the same way that N does, in all
+   languages we know of.  */
+static inline unsigned long int
+select_plural (uintmax_t n)
+{
+  return (n <= ULONG_MAX ? n : n % 1000 + 1000);
+}
+
 #define STREQ(a, b) (strcmp ((a), (b)) == 0)
 
 #if !HAVE_DECL_FREE
Index: src/md5sum.c
===================================================================
RCS file: /fetish/cu/src/md5sum.c,v
retrieving revision 1.146
diff -p -u -r1.146 md5sum.c
--- src/md5sum.c        9 Jul 2006 17:08:55 -0000       1.146
+++ src/md5sum.c        16 Aug 2006 19:34:55 -0000
@@ -563,7 +563,7 @@ digest_check (const char *checkfile_name
                             " listed file could not be read",
                             "WARNING: %" PRIuMAX " of %" PRIuMAX
                             " listed files could not be read",
-                            n_properly_formatted_lines),
+                            select_plural (n_properly_formatted_lines)),
                   n_open_or_read_failures, n_properly_formatted_lines);
 
          if (n_mismatched_checksums != 0)
@@ -575,7 +575,7 @@ digest_check (const char *checkfile_name
                               " computed checksum did NOT match",
                               "WARNING: %" PRIuMAX " of %" PRIuMAX
                               " computed checksums did NOT match",
-                              n_computed_checksums),
+                              select_plural (n_computed_checksums)),
                     n_mismatched_checksums, n_computed_checksums);
            }
        }
Index: src/uptime.c
===================================================================
RCS file: /fetish/cu/src/uptime.c,v
retrieving revision 1.54
diff -p -u -r1.54 uptime.c
--- src/uptime.c        3 Dec 2005 23:24:46 -0000       1.54
+++ src/uptime.c        16 Aug 2006 19:34:55 -0000
@@ -125,7 +125,8 @@ print_uptime (size_t n, const STRUCT_UTM
   else
     {
       if (0 < updays)
-       printf (ngettext ("%ld day", "%ld days", updays), updays);
+       printf (ngettext ("%ld day", "%ld days", select_plural (updays)),
+               updays);
       printf (" %2d:%02d,  ", uphours, upmins);
     }
   printf (ngettext ("%lu user", "%lu users", entries),
Index: src/dd.c
===================================================================
RCS file: /fetish/cu/src/dd.c,v
retrieving revision 1.197
diff -p -u -r1.197 dd.c
--- src/dd.c    15 Aug 2006 20:50:22 -0000      1.197
+++ src/dd.c    16 Aug 2006 19:34:55 -0000
@@ -550,7 +550,7 @@ print_stats (void)
     fprintf (stderr,
             ngettext ("%"PRIuMAX" truncated record\n",
                       "%"PRIuMAX" truncated records\n",
-                      MIN (r_truncate, ULONG_MAX)),
+                      select_plural (r_truncate)),
             r_truncate);
 
   if (status_flags & STATUS_NOXFER)
@@ -562,7 +562,7 @@ print_stats (void)
   fprintf (stderr,
           ngettext ("%"PRIuMAX" byte (%s) copied",
                     "%"PRIuMAX" bytes (%s) copied",
-                    MIN (w_bytes, ULONG_MAX)),
+                    select_plural (w_bytes)),
           w_bytes,
           human_readable (w_bytes, hbuf, human_opts, 1, 1));
 
@@ -581,10 +581,17 @@ print_stats (void)
       bytes_per_second = _("Infinity B");
     }
 
-  fprintf (stderr,
-          ngettext (", %g second, %s/s\n",
-                    ", %g seconds, %s/s\n", delta_s == 1),
-          delta_s, bytes_per_second);
+  /* TRANSLATORS: The two instances of "s" in this string are the SI
+     symbol "s" (meaning second), and should not be translated.
+
+     This format used to be:
+
+     ngettext (", %g second, %s/s\n", ", %g seconds, %s/s\n", delta_s == 1)
+
+     but that was incorrect for languages like Polish.  To fix this
+     bug we now use SI symbols even though they're a bit more
+     confusing in English.  */
+  fprintf (stderr, _(", %g s, %s/s\n"), delta_s, bytes_per_second);
 }
 
 static void