Complete C++ i18n gettext() “hello world” example

Question

I am looking for a complete i18n gettext() hello world example. I have started a script based upon A tutorial on Native Language Support using GNU gettext by G. Mohanty. I am using Linux and G++.

Code:

cat >hellogt.cxx <<EOF
// hellogt.cxx
#include <libintl.h>
#include <locale.h>
#include <iostream>
#include <cstdlib>
int main (){
    char* cwd = getenv("PWD");
    std::cout << "getenv(PWD): " << (cwd?cwd:"NULL") << std::endl;
    char* l = getenv("LANG");
    std::cout << "getenv(LANG): " << (l?l:"NULL") << std::endl;
    char* s = setlocale(LC_ALL, "");
    std::cout << "setlocale(): " << (s?s:"NULL") << std::endl;
    std::cout << "bindtextdomain(): " << bindtextdomain("hellogt", cwd) << std::endl;
    std::cout << "textdomain(): " << textdomain( "hellogt") << std::endl;
    std::cout << gettext("hello, world!") << std::endl;
}
EOF
g++ -ohellogt hellogt.cxx
xgettext -d hellogt -o hellogt.pot hellogt.cxx
msginit --no-translator -l es_MX -o hellogt_spanish.po -i hellogt.pot
sed --in-place hellogt_spanish.po --expression='/#: /,$ s/""/"hola mundo"/'
sed --in-place hellogt_spanish.po --expression='s/PACKAGE VERSION/hellogt 1.0/'
mkdir -p ./es_MX/LC_MESSAGES
msgfmt -c -v -o ./es_MX/LC_MESSAGES/hellogt.mo hellogt_spanish.po
export LANG=es_MX
ls -l $PWD/es_MX/LC_MESSAGES/hellogt.mo
./hellogt
strace -e trace=open ./hellogt

The program compiles, the text is extracted, Spanish file is created, modified and binary created but hellogt still displays English. The trace shows no evidence of looking in the current working directory for es_MX nor any references to LC_MESSAGES directory.

Answer 1

cat >hellogt.cxx <<EOF
// hellogt.cxx
#include <libintl.h>
#include <locale.h>
#include <iostream>
int main (){
    setlocale(LC_ALL, "");
    bindtextdomain("hellogt", ".");
    textdomain( "hellogt");
    std::cout << gettext("hello, world!") << std::endl;
}
EOF
g++ -o hellogt hellogt.cxx
xgettext --package-name hellogt --package-version 1.2 --default-domain hellogt --output hellogt.pot hellogt.cxx
msginit --no-translator --locale es_MX --output-file hellogt_spanish.po --input hellogt.pot
sed --in-place hellogt_spanish.po --expression='/"hello, world!"/,/#: / s/""/"hola mundo"/'
mkdir --parents ./es_MX.utf8/LC_MESSAGES
msgfmt --check --verbose --output-file ./es_MX.utf8/LC_MESSAGES/hellogt.mo hellogt_spanish.po
LANGUAGE=es_MX.utf8 ./hellogt

Here is a description of the files created by the above:

hellogt.cxx         C++ source file
hellogt             Executable image
hellogt.pot         Extracted text from C++ source file (portable object template)
hellogt_spanish.po  Modified text for Spanish with translations added (using sed)
es_MX.utf8/
 LC_MESSAGES/
   hellogt.mo       Binary translated text for Spanish used at run-time

Answer 2

Your problem is that hellogt.mo is in the wrong location - your program isn't actually opening it. You can tell this by using strace to trace open syscalls:

strace -e trace=open ./hellogt
...
open("/tmp/.//es_MX/LC_MESSAGES/hellogt.mo", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/tmp/.//es/LC_MESSAGES/hellogt.mo", O_RDONLY) = -1 ENOENT (No such file or directory)

You can affect where gettext looks for message catalogs with the LOCPATH environment variable, but if you move it to where gettext is attempting to load it from your example works:

mkdir -p es/LC_MESSAGES
cp hellogt.mo es/LC_MESSAGES
./hellogt 
hola mundo

Answer 3

Here is a description of gettext from Fedora Project. It is simple to follow. But it is in C.

http://fedoraproject.org/wiki/How_to_do_I18N_through_gettext