Re: [bug-gawk] _ function (gettext) broken in gawk 4.2.0

[Arnold Robbins]

Hi.

I have put this test into the gawk test suite.

Thanks,

Arnold

> From: Bruno Haible <address@hidden>
> To: address@hidden
> Date: Sat, 06 Jan 2018 08:36:50 +0100
> Subject: [bug-gawk] _ function (gettext) broken in gawk 4.2.0
>
> Hi,
>
> The following test case (from GNU gettext) works fine with
> gawk versions 3.1.6 to 4.1.4, but is now broken in some
> circumstances in gawk 4.2.0.
>
>
> How to reproduce:
>
> 1) Make sure you have a French locale installed.
> $ locale -a | grep fr_FR
> If not, create one:
> $ sudo localedef -i fr_FR -f UTF-8 fr_FR.UTF-8
>
> 2) Save the follwing files:
> ============================ prog.awk ============================
> BEGIN {
>   TEXTDOMAIN = "prog"
>   bindtextdomain ("./")
>
>   printf dcngettext ("a piece of cake", "%d pieces of cake", n) "\n", n
>
>   print _"%s is replaced by %s."
>   print _"%s is replaced by %s." "\n"
>   printf _"%s is replaced by %s." "\n", "FF", "EUR"
> }
> ==================================================================
> ============================= fr.po ============================
> msgid ""
> msgstr ""
> "Content-Type: text/plain; charset=ISO-8859-1\n"
> "Plural-Forms: nplurals=2; plural=(n > 1);\n"
>
> # Les gateaux allemands sont les meilleurs du monde.
> #, awk-format
> msgid "a piece of cake"
> msgid_plural "%d pieces of cake"
> msgstr[0] "un morceau de gateau"
> msgstr[1] "%d morceaux de gateau"
>
> # Reverse the arguments.
> #, awk-format
> msgid "%s is replaced by %s."
> msgstr "%2$s remplace %1$s."
> ================================================================
>
> 3) Compile the PO file:
> $ mkdir -p fr/LC_MESSAGES
> $ msgfmt -o fr/LC_MESSAGES/prog.mo fr.po
>
> 4) Verify that gawk is internationalized:
> $ LANGUAGE= LC_ALL=fr_FR.UTF-8 gawk --version | grep logiciel
>
> 5) Run the program:
>
> $ LANGUAGE= LC_ALL=fr_FR.UTF-8 gawk -v n=2 -f prog.awk
>
> Expected output (seen with versions 4.1.3, 4.1.4):
> --------------------------
> 2 morceaux de gateau
> %2$s remplace %1$s.
> %2$s remplace %1$s.
>
> EUR remplace FF.
> --------------------------
>
> Actual output with version 4.2.0 (on Linux and Mac OS X):
> --------------------------
> 2 morceaux de gateau
> %2$s remplace %1$s.
> %s is replaced by %s.
>
> FF is replaced by EUR.
> --------------------------
>
>
> The NEWS file does not mention a related intended change
> regarding the _ (gettext) functions between 4.1.4 and 4.2.0.
>
>
> Among the ChangeLogs between versions 4.1.4 and 4.2.0, I can see these:
>
> 2017-04-12         Arnold D. Robbins     <address@hidden>
>
>       * builtin.c (do_dcgettext): Move declaration of reslen to
>       outside the ifdefs. Thanks to Hermann Peifer for the report.
>
> 2017-01-27         Andrew J. Schorr     <address@hidden>
>
>       * builtin.c (do_strftime): Use str_terminate and str_restore.
>       (do_dcgettext): Ditto, and remove saved_end flag since equivalent
>       to testing (t2 != NULL). Fix overrun bug in calculating result
>       length when !ENABLE_NLS.
>       (do_dcngettext, do_bindtextdomain): Use str_terminate and str_restore.
>
> 2017-01-26         Andrew J. Schorr     <address@hidden>
>
>       * builtin.c (do_dcgettext): First argument also needs protection
>       from string overrun.
>       (do_dcngettext): Need to terminate string1 and string2 also,
>       and replace strlen(the_result), which could overrun.
>
> 2016-11-17         Arnold D. Robbins     <address@hidden>
>
>       General cleanup for zero termination of strings.
>
>       * builtin.c (do_fflush): Use %.*s.
>       (locale_category_from_argument, do_dcgettext, do_dcngettext,
>       do_bindtextdomain): Save and restore character after end.
>
> Probably some of these is relevant.
>
>
> Best regards,
>
>           Bruno
>
>