1 /* Implementation of the internal dcigettext function.
2 Copyright (C) 1995-2017 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU Lesser General Public License as published by
6 the Free Software Foundation; either version 2.1 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU Lesser General Public License for more details.
14 You should have received a copy of the GNU Lesser General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
17 /* Tell glibc's <string.h> to provide a prototype for mempcpy().
18 This must come before <config.h> because <config.h> may include
19 <features.h>, and once <features.h> has been included, it's too late. */
21 # define _GNU_SOURCE 1
28 #include <sys/types.h>
31 # define alloca __builtin_alloca
32 # define HAVE_ALLOCA 1
36 # define alloca _alloca
38 # if defined HAVE_ALLOCA_H || defined _LIBC
57 # define __set_errno(val) errno = (val)
65 #if defined HAVE_UNISTD_H || defined _LIBC
72 /* Guess whether integer division by zero raises signal SIGFPE.
73 Set to 1 only if you know for sure. In case of doubt, set to 0. */
74 # if defined __alpha__ || defined __arm__ || defined __i386__ \
75 || defined __m68k__ || defined __s390__
76 # define INTDIV0_RAISES_SIGFPE 1
78 # define INTDIV0_RAISES_SIGFPE 0
81 #if !INTDIV0_RAISES_SIGFPE
85 #if defined HAVE_SYS_PARAM_H || defined _LIBC
86 # include <sys/param.h>
90 # include "localcharset.h"
94 #include "plural-exp.h"
101 # include "libgnuintl.h"
103 #include "hash-string.h"
105 /* Handle multi-threaded applications. */
107 # include <libc-lock.h>
108 # define gl_rwlock_define_initialized __libc_rwlock_define_initialized
109 # define gl_rwlock_rdlock __libc_rwlock_rdlock
110 # define gl_rwlock_wrlock __libc_rwlock_wrlock
111 # define gl_rwlock_unlock __libc_rwlock_unlock
116 /* Alignment of types. */
117 #if defined __GNUC__ && __GNUC__ >= 2
118 # define alignof(TYPE) __alignof__ (TYPE)
120 # define alignof(TYPE) \
121 ((int) &((struct { char dummy1; TYPE dummy2; } *) 0)->dummy2)
124 /* Some compilers, like SunOS4 cc, don't have offsetof in <stddef.h>. */
126 # define offsetof(type,ident) ((size_t)&(((type*)0)->ident))
129 /* @@ end of prolog @@ */
132 /* Rename the non ANSI C functions. This is required by the standard
133 because some ANSI C functions will require linking with this object
134 file and the name space must not be polluted. */
135 # define strdup __strdup
136 # define getcwd __getcwd
138 # define stpcpy __stpcpy
140 # define tfind __tfind
142 # if !defined HAVE_GETCWD
144 # define getcwd(buf, max) getwd (buf)
147 # define getcwd(buf, max) (getcwd) (buf, max, 0)
153 static char *stpcpy (char *dest
, const char *src
);
155 # ifndef HAVE_MEMPCPY
156 static void *mempcpy (void *dest
, const void *src
, size_t n
);
160 /* Use a replacement if the system does not provide the `tsearch' function
162 #if defined HAVE_TSEARCH || defined _LIBC
165 # define tsearch libintl_tsearch
166 # define tfind libintl_tfind
167 # define tdelete libintl_tdelete
168 # define twalk libintl_twalk
169 # include "tsearch.h"
173 # define tsearch __tsearch
176 /* Amount to increase buffer size by in each try. */
179 /* The following is from pathmax.h. */
180 /* Non-POSIX BSD systems might have gcc's limits.h, which doesn't define
181 PATH_MAX but might cause redefinition warnings when sys/param.h is
182 later included (as on MORE/BSD 4.3). */
183 #if defined _POSIX_VERSION || (defined HAVE_LIMITS_H && !defined __GNUC__)
187 #ifndef _POSIX_PATH_MAX
188 # define _POSIX_PATH_MAX 255
191 #if !defined PATH_MAX && defined _PC_PATH_MAX
192 # define PATH_MAX (__pathconf ("/", _PC_PATH_MAX) < 1 ? 1024 : __pathconf ("/", _PC_PATH_MAX))
195 /* Don't include sys/param.h if it already has been. */
196 #if defined HAVE_SYS_PARAM_H && !defined PATH_MAX && !defined MAXPATHLEN
197 # include <sys/param.h>
200 #if !defined PATH_MAX && defined MAXPATHLEN
201 # define PATH_MAX MAXPATHLEN
205 # define PATH_MAX _POSIX_PATH_MAX
209 ISSLASH(C) tests whether C is a directory separator character.
210 IS_ABSOLUTE_PATH(P) tests whether P is an absolute path. If it is not,
211 it may be concatenated to a directory pathname.
212 IS_PATH_WITH_DIR(P) tests whether P contains a directory specification.
214 #if defined _WIN32 || defined __WIN32__ || defined __CYGWIN__ || defined __EMX__ || defined __DJGPP__
215 /* Win32, Cygwin, OS/2, DOS */
216 # define ISSLASH(C) ((C) == '/' || (C) == '\\')
217 # define HAS_DEVICE(P) \
218 ((((P)[0] >= 'A' && (P)[0] <= 'Z') || ((P)[0] >= 'a' && (P)[0] <= 'z')) \
220 # define IS_ABSOLUTE_PATH(P) (ISSLASH ((P)[0]) || HAS_DEVICE (P))
221 # define IS_PATH_WITH_DIR(P) \
222 (strchr (P, '/') != NULL || strchr (P, '\\') != NULL || HAS_DEVICE (P))
225 # define ISSLASH(C) ((C) == '/')
226 # define IS_ABSOLUTE_PATH(P) ISSLASH ((P)[0])
227 # define IS_PATH_WITH_DIR(P) (strchr (P, '/') != NULL)
230 /* Whether to support different locales in different threads. */
231 #if defined _LIBC || HAVE_USELOCALE || defined IN_LIBGLOCALE
232 # define HAVE_PER_THREAD_LOCALE
235 /* This is the type used for the search tree where known translations
237 struct known_translation_t
239 /* Domain in which to search. */
240 const char *domainname
;
245 #ifdef HAVE_PER_THREAD_LOCALE
246 /* Name of the relevant locale category, or "" for the global locale. */
247 const char *localename
;
251 /* The character encoding. */
252 const char *encoding
;
255 /* State of the catalog counter at the point the string was found. */
258 /* Catalog where the string was found. */
259 struct loaded_l10nfile
*domain
;
261 /* And finally the translation. */
262 const char *translation
;
263 size_t translation_length
;
265 /* Pointer to the string in question. */
268 char appended
[ZERO
]; /* used if domain != NULL */
269 const char *ptr
; /* used if domain == NULL */
274 gl_rwlock_define_initialized (static, tree_lock
)
276 /* Root of the search tree with known translations. */
279 /* Function to compare two entries in the table of known translations. */
281 transcmp (const void *p1
, const void *p2
)
283 const struct known_translation_t
*s1
;
284 const struct known_translation_t
*s2
;
287 s1
= (const struct known_translation_t
*) p1
;
288 s2
= (const struct known_translation_t
*) p2
;
290 result
= strcmp (s1
->domain
!= NULL
? s1
->msgid
.appended
: s1
->msgid
.ptr
,
291 s2
->domain
!= NULL
? s2
->msgid
.appended
: s2
->msgid
.ptr
);
294 result
= strcmp (s1
->domainname
, s2
->domainname
);
297 #ifdef HAVE_PER_THREAD_LOCALE
298 result
= strcmp (s1
->localename
, s2
->localename
);
303 result
= strcmp (s1
->encoding
, s2
->encoding
);
306 /* We compare the category last (though this is the cheapest
307 operation) since it is hopefully always the same (namely
309 result
= s1
->category
- s2
->category
;
317 /* Name of the default domain used for gettext(3) prior any call to
318 textdomain(3). The default value for this is "messages". */
319 const char _nl_default_default_domain
[] attribute_hidden
= "messages";
321 #ifndef IN_LIBGLOCALE
322 /* Value used as the default domain for gettext(3). */
323 const char *_nl_current_default_domain attribute_hidden
324 = _nl_default_default_domain
;
327 /* Contains the default location of the message catalogs. */
329 extern const char _nl_default_dirname
[];
332 extern const char _nl_default_dirname
[];
333 libc_hidden_proto (_nl_default_dirname
)
335 const char _nl_default_dirname
[] = LOCALEDIR
;
337 libc_hidden_data_def (_nl_default_dirname
)
341 #ifndef IN_LIBGLOCALE
342 /* List with bindings of specific domains created by bindtextdomain()
344 struct binding
*_nl_domain_bindings
;
347 /* Prototypes for local functions. */
348 static char *plural_lookup (struct loaded_l10nfile
*domain
,
350 const char *translation
, size_t translation_len
)
354 static const char *guess_category_value (int category
,
355 const char *categoryname
,
356 const char *localename
)
359 static const char *guess_category_value (int category
,
360 const char *categoryname
)
365 # include "../locale/localeinfo.h"
366 # define category_to_name(category) \
367 _nl_category_names.str + _nl_category_name_idxs[category]
369 static const char *category_to_name (int category
) internal_function
;
371 #if (defined _LIBC || HAVE_ICONV) && !defined IN_LIBGLOCALE
372 static const char *get_output_charset (struct binding
*domainbinding
)
377 /* For those losing systems which don't have `alloca' we have to add
378 some additional code emulating it. */
380 /* Nothing has to be done. */
381 # define freea(p) /* nothing */
382 # define ADD_BLOCK(list, address) /* nothing */
383 # define FREE_BLOCKS(list) /* nothing */
388 struct block_list
*next
;
390 # define ADD_BLOCK(list, addr) \
392 struct block_list *newp = (struct block_list *) malloc (sizeof (*newp)); \
393 /* If we cannot get a free block we cannot add the new element to \
395 if (newp != NULL) { \
396 newp->address = (addr); \
397 newp->next = (list); \
401 # define FREE_BLOCKS(list) \
403 while (list != NULL) { \
404 struct block_list *old = list; \
406 free (old->address); \
411 # define alloca(size) (malloc (size))
412 # define freea(p) free (p)
413 #endif /* have alloca */
417 /* List of blocks allocated for translations. */
418 typedef struct transmem_list
420 struct transmem_list
*next
;
423 static struct transmem_list
*transmem_list
;
425 typedef unsigned char transmem_block_t
;
429 /* Names for the libintl functions are a problem. They must not clash
430 with existing names and they should follow ANSI C. But this source
431 code is also used in GNU C Library where the names have a __
432 prefix. So we have to make a difference here. */
434 # define DCIGETTEXT __dcigettext
436 # define DCIGETTEXT libintl_dcigettext
439 /* Lock variable to protect the global data in the gettext implementation. */
440 gl_rwlock_define_initialized (, _nl_state_lock attribute_hidden
)
442 /* Checking whether the binaries runs SUID must be done and glibc provides
443 easier methods therefore we make a difference here. */
445 # define ENABLE_SECURE __libc_enable_secure
446 # define DETERMINE_SECURE
454 # ifndef HAVE_GETEUID
455 # define geteuid() getuid()
457 # ifndef HAVE_GETEGID
458 # define getegid() getgid()
460 static int enable_secure
;
461 # define ENABLE_SECURE (enable_secure == 1)
462 # define DETERMINE_SECURE \
463 if (enable_secure == 0) \
465 if (getuid () != geteuid () || getgid () != getegid ()) \
468 enable_secure = -1; \
472 /* Get the function to evaluate the plural expression. */
473 #include "eval-plural.h"
475 /* Look up MSGID in the DOMAINNAME message catalog for the current
476 CATEGORY locale and, if PLURAL is nonzero, search over string
477 depending on the plural form determined by N. */
480 gl_dcigettext (const char *domainname
,
481 const char *msgid1
, const char *msgid2
,
482 int plural
, unsigned long int n
,
484 const char *localename
, const char *encoding
)
487 DCIGETTEXT (const char *domainname
, const char *msgid1
, const char *msgid2
,
488 int plural
, unsigned long int n
, int category
)
492 struct block_list
*block_list
= NULL
;
494 struct loaded_l10nfile
*domain
;
495 struct binding
*binding
;
496 const char *categoryname
;
497 const char *categoryvalue
;
499 char *xdirname
= NULL
;
505 struct known_translation_t search
;
506 struct known_translation_t
**foundp
= NULL
;
507 #if defined HAVE_PER_THREAD_LOCALE && !defined IN_LIBGLOCALE
508 const char *localename
;
510 size_t domainname_len
;
512 /* If no real MSGID is given return NULL. */
517 if (category
< 0 || category
>= __LC_LAST
|| category
== LC_ALL
)
521 /* Use the Germanic plural rule. */
522 : n
== 1 ? (char *) msgid1
: (char *) msgid2
);
525 /* Preserve the `errno' value. */
529 __libc_rwlock_define (extern, __libc_setlocale_lock attribute_hidden
)
530 __libc_rwlock_rdlock (__libc_setlocale_lock
);
533 gl_rwlock_rdlock (_nl_state_lock
);
535 /* If DOMAINNAME is NULL, we are interested in the default domain. If
536 CATEGORY is not LC_MESSAGES this might not make much sense but the
537 definition left this undefined. */
538 if (domainname
== NULL
)
539 domainname
= _nl_current_default_domain
;
541 /* OS/2 specific: backward compatibility with older libintl versions */
542 #ifdef LC_MESSAGES_COMPAT
543 if (category
== LC_MESSAGES_COMPAT
)
544 category
= LC_MESSAGES
;
547 /* Try to find the translation among those which we found at
549 search
.domain
= NULL
;
550 search
.msgid
.ptr
= msgid1
;
551 search
.domainname
= domainname
;
552 search
.category
= category
;
553 #ifdef HAVE_PER_THREAD_LOCALE
554 # ifndef IN_LIBGLOCALE
556 localename
= __current_locale_name (category
);
558 categoryname
= category_to_name (category
);
559 # define CATEGORYNAME_INITIALIZED
560 localename
= _nl_locale_name_thread_unsafe (category
, categoryname
);
561 if (localename
== NULL
)
565 search
.localename
= localename
;
566 # ifdef IN_LIBGLOCALE
567 search
.encoding
= encoding
;
570 /* Since tfind/tsearch manage a balanced tree, concurrent tfind and
571 tsearch calls can be fatal. */
572 gl_rwlock_rdlock (tree_lock
);
574 foundp
= (struct known_translation_t
**) tfind (&search
, &root
, transcmp
);
576 gl_rwlock_unlock (tree_lock
);
578 if (foundp
!= NULL
&& (*foundp
)->counter
== _nl_msg_cat_cntr
)
580 /* Now deal with plural. */
582 retval
= plural_lookup ((*foundp
)->domain
, n
, (*foundp
)->translation
,
583 (*foundp
)->translation_length
);
585 retval
= (char *) (*foundp
)->translation
;
587 gl_rwlock_unlock (_nl_state_lock
);
589 __libc_rwlock_unlock (__libc_setlocale_lock
);
591 __set_errno (saved_errno
);
596 /* See whether this is a SUID binary or not. */
599 /* First find matching binding. */
601 /* We can use a trivial binding, since _nl_find_msg will ignore it anyway,
602 and _nl_load_domain and _nl_find_domain just pass it through. */
604 dirname
= bindtextdomain (domainname
, NULL
);
606 for (binding
= _nl_domain_bindings
; binding
!= NULL
; binding
= binding
->next
)
608 int compare
= strcmp (domainname
, binding
->domainname
);
614 /* It is not in the list. */
621 dirname
= _nl_default_dirname
;
624 dirname
= binding
->dirname
;
626 if (!IS_ABSOLUTE_PATH (dirname
))
628 /* We have a relative path. Make it absolute now. */
629 char *cwd
= getcwd (NULL
, 0);
631 /* We cannot get the current working directory. Don't
632 signal an error but simply return the default
634 goto return_untranslated
;
635 int ret
= __asprintf (&xdirname
, "%s/%s", cwd
, dirname
);
641 #ifndef IN_LIBGLOCALE
645 /* Now determine the symbolic name of CATEGORY and its value. */
646 #ifndef CATEGORYNAME_INITIALIZED
647 categoryname
= category_to_name (category
);
650 categoryvalue
= guess_category_value (category
, categoryname
, localename
);
652 categoryvalue
= guess_category_value (category
, categoryname
);
655 domainname_len
= strlen (domainname
);
656 xdomainname
= (char *) alloca (strlen (categoryname
)
657 + domainname_len
+ 5);
658 ADD_BLOCK (block_list
, xdomainname
);
660 stpcpy ((char *) mempcpy (stpcpy (stpcpy (xdomainname
, categoryname
), "/"),
661 domainname
, domainname_len
),
664 /* Creating working area. */
665 single_locale
= (char *) alloca (strlen (categoryvalue
) + 1);
666 ADD_BLOCK (block_list
, single_locale
);
669 /* Search for the given string. This is a loop because we perhaps
670 got an ordered list of languages to consider for the translation. */
673 /* Make CATEGORYVALUE point to the next element of the list. */
674 while (categoryvalue
[0] != '\0' && categoryvalue
[0] == ':')
676 if (categoryvalue
[0] == '\0')
678 /* The whole contents of CATEGORYVALUE has been searched but
679 no valid entry has been found. We solve this situation
680 by implicitly appending a "C" entry, i.e. no translation
682 single_locale
[0] = 'C';
683 single_locale
[1] = '\0';
687 char *cp
= single_locale
;
688 while (categoryvalue
[0] != '\0' && categoryvalue
[0] != ':')
689 *cp
++ = *categoryvalue
++;
692 /* When this is a SUID binary we must not allow accessing files
693 outside the dedicated directories. */
694 if (ENABLE_SECURE
&& IS_PATH_WITH_DIR (single_locale
))
695 /* Ingore this entry. */
699 /* If the current locale value is C (or POSIX) we don't load a
700 domain. Return the MSGID. */
701 if (strcmp (single_locale
, "C") == 0
702 || strcmp (single_locale
, "POSIX") == 0)
705 /* Find structure describing the message catalog matching the
706 DOMAINNAME and CATEGORY. */
707 domain
= _nl_find_domain (dirname
, single_locale
, xdomainname
, binding
);
711 #if defined IN_LIBGLOCALE
712 retval
= _nl_find_msg (domain
, binding
, encoding
, msgid1
, &retlen
);
714 retval
= _nl_find_msg (domain
, binding
, msgid1
, 1, &retlen
);
721 for (cnt
= 0; domain
->successor
[cnt
] != NULL
; ++cnt
)
723 #if defined IN_LIBGLOCALE
724 retval
= _nl_find_msg (domain
->successor
[cnt
], binding
,
725 encoding
, msgid1
, &retlen
);
727 retval
= _nl_find_msg (domain
->successor
[cnt
], binding
,
731 /* Resource problems are not fatal, instead we return no
733 if (__builtin_expect (retval
== (char *) -1, 0))
734 goto return_untranslated
;
738 domain
= domain
->successor
[cnt
];
744 /* Returning -1 means that some resource problem exists
745 (likely memory) and that the strings could not be
746 converted. Return the original strings. */
747 if (__builtin_expect (retval
== (char *) -1, 0))
752 /* Found the translation of MSGID1 in domain DOMAIN:
753 starting at RETVAL, RETLEN bytes. */
755 FREE_BLOCKS (block_list
);
758 /* Create a new entry and add it to the search tree. */
761 struct known_translation_t
*newp
;
763 msgid_len
= strlen (msgid1
) + 1;
764 size
= offsetof (struct known_translation_t
, msgid
)
765 + msgid_len
+ domainname_len
+ 1;
766 #ifdef HAVE_PER_THREAD_LOCALE
767 size
+= strlen (localename
) + 1;
769 newp
= (struct known_translation_t
*) malloc (size
);
772 char *new_domainname
;
773 #ifdef HAVE_PER_THREAD_LOCALE
774 char *new_localename
;
778 (char *) mempcpy (newp
->msgid
.appended
, msgid1
,
780 memcpy (new_domainname
, domainname
, domainname_len
+ 1);
781 #ifdef HAVE_PER_THREAD_LOCALE
782 new_localename
= new_domainname
+ domainname_len
+ 1;
783 strcpy (new_localename
, localename
);
785 newp
->domainname
= new_domainname
;
786 newp
->category
= category
;
787 #ifdef HAVE_PER_THREAD_LOCALE
788 newp
->localename
= new_localename
;
791 newp
->encoding
= encoding
;
793 newp
->counter
= _nl_msg_cat_cntr
;
794 newp
->domain
= domain
;
795 newp
->translation
= retval
;
796 newp
->translation_length
= retlen
;
798 gl_rwlock_wrlock (tree_lock
);
800 /* Insert the entry in the search tree. */
801 foundp
= (struct known_translation_t
**)
802 tsearch (newp
, &root
, transcmp
);
804 gl_rwlock_unlock (tree_lock
);
807 || __builtin_expect (*foundp
!= newp
, 0))
808 /* The insert failed. */
814 /* We can update the existing entry. */
815 (*foundp
)->counter
= _nl_msg_cat_cntr
;
816 (*foundp
)->domain
= domain
;
817 (*foundp
)->translation
= retval
;
818 (*foundp
)->translation_length
= retlen
;
821 __set_errno (saved_errno
);
823 /* Now deal with plural. */
825 retval
= plural_lookup (domain
, n
, retval
, retlen
);
827 gl_rwlock_unlock (_nl_state_lock
);
829 __libc_rwlock_unlock (__libc_setlocale_lock
);
837 /* Return the untranslated MSGID. */
839 FREE_BLOCKS (block_list
);
840 gl_rwlock_unlock (_nl_state_lock
);
842 __libc_rwlock_unlock (__libc_setlocale_lock
);
847 extern void _nl_log_untranslated (const char *logfilename
,
848 const char *domainname
,
849 const char *msgid1
, const char *msgid2
,
851 const char *logfilename
= getenv ("GETTEXT_LOG_UNTRANSLATED");
853 if (logfilename
!= NULL
&& logfilename
[0] != '\0')
854 _nl_log_untranslated (logfilename
, domainname
, msgid1
, msgid2
, plural
);
857 __set_errno (saved_errno
);
860 /* Use the Germanic plural rule. */
861 : n
== 1 ? (char *) msgid1
: (char *) msgid2
);
865 /* Look up the translation of msgid within DOMAIN_FILE and DOMAINBINDING.
866 Return it if found. Return NULL if not found or in case of a conversion
867 failure (problem in the particular message catalog). Return (char *) -1
868 in case of a memory allocation failure during conversion (only if
869 ENCODING != NULL resp. CONVERT == true). */
873 _nl_find_msg (struct loaded_l10nfile
*domain_file
,
874 struct binding
*domainbinding
, const char *encoding
,
878 _nl_find_msg (struct loaded_l10nfile
*domain_file
,
879 struct binding
*domainbinding
,
880 const char *msgid
, int convert
,
884 struct loaded_domain
*domain
;
890 if (domain_file
->decided
<= 0)
891 _nl_load_domain (domain_file
, domainbinding
);
893 if (domain_file
->data
== NULL
)
896 domain
= (struct loaded_domain
*) domain_file
->data
;
898 nstrings
= domain
->nstrings
;
900 /* Locate the MSGID and its translation. */
901 if (domain
->hash_tab
!= NULL
)
903 /* Use the hashing table. */
904 nls_uint32 len
= strlen (msgid
);
905 nls_uint32 hash_val
= __hash_string (msgid
);
906 nls_uint32 idx
= hash_val
% domain
->hash_size
;
907 nls_uint32 incr
= 1 + (hash_val
% (domain
->hash_size
- 2));
912 W (domain
->must_swap_hash_tab
, domain
->hash_tab
[idx
]);
915 /* Hash table entry is empty. */
920 /* Compare msgid with the original string at index nstr.
921 We compare the lengths with >=, not ==, because plural entries
922 are represented by strings with an embedded NUL. */
924 ? W (domain
->must_swap
, domain
->orig_tab
[nstr
].length
) >= len
926 domain
->data
+ W (domain
->must_swap
,
927 domain
->orig_tab
[nstr
].offset
))
929 : domain
->orig_sysdep_tab
[nstr
- nstrings
].length
> len
931 domain
->orig_sysdep_tab
[nstr
- nstrings
].pointer
)
938 if (idx
>= domain
->hash_size
- incr
)
939 idx
-= domain
->hash_size
- incr
;
947 /* Try the default method: binary search in the sorted array of
957 act
= (bottom
+ top
) / 2;
958 cmp_val
= strcmp (msgid
, (domain
->data
959 + W (domain
->must_swap
,
960 domain
->orig_tab
[act
].offset
)));
963 else if (cmp_val
> 0)
968 /* No translation was found. */
973 /* The translation was found at index ACT. If we have to convert the
974 string to use a different character set, this is the time. */
978 (domain
->data
+ W (domain
->must_swap
, domain
->trans_tab
[act
].offset
));
979 resultlen
= W (domain
->must_swap
, domain
->trans_tab
[act
].length
) + 1;
983 result
= (char *) domain
->trans_sysdep_tab
[act
- nstrings
].pointer
;
984 resultlen
= domain
->trans_sysdep_tab
[act
- nstrings
].length
;
987 #if defined _LIBC || HAVE_ICONV
988 # ifdef IN_LIBGLOCALE
989 if (encoding
!= NULL
)
994 /* We are supposed to do a conversion. */
995 # ifndef IN_LIBGLOCALE
996 const char *encoding
= get_output_charset (domainbinding
);
999 struct converted_domain
*convd
;
1002 /* Protect against reallocation of the table. */
1003 gl_rwlock_rdlock (domain
->conversions_lock
);
1005 /* Search whether a table with converted translations for this
1006 encoding has already been allocated. */
1007 nconversions
= domain
->nconversions
;
1010 for (i
= nconversions
; i
> 0; )
1013 if (strcmp (domain
->conversions
[i
].encoding
, encoding
) == 0)
1015 convd
= &domain
->conversions
[i
];
1020 gl_rwlock_unlock (domain
->conversions_lock
);
1024 /* We have to allocate a new conversions table. */
1025 gl_rwlock_wrlock (domain
->conversions_lock
);
1026 nconversions
= domain
->nconversions
;
1028 /* Maybe in the meantime somebody added the translation.
1030 for (i
= nconversions
; i
> 0; )
1033 if (strcmp (domain
->conversions
[i
].encoding
, encoding
) == 0)
1035 convd
= &domain
->conversions
[i
];
1041 /* Allocate a table for the converted translations for this
1043 struct converted_domain
*new_conversions
=
1044 (struct converted_domain
*)
1045 (domain
->conversions
!= NULL
1046 ? realloc (domain
->conversions
,
1047 (nconversions
+ 1) * sizeof (struct converted_domain
))
1048 : malloc ((nconversions
+ 1) * sizeof (struct converted_domain
)));
1050 if (__builtin_expect (new_conversions
== NULL
, 0))
1052 /* Nothing we can do, no more memory. We cannot use the
1053 translation because it might be encoded incorrectly. */
1055 gl_rwlock_unlock (domain
->conversions_lock
);
1059 domain
->conversions
= new_conversions
;
1061 /* Copy the 'encoding' string to permanent storage. */
1062 encoding
= strdup (encoding
);
1063 if (__builtin_expect (encoding
== NULL
, 0))
1064 /* Nothing we can do, no more memory. We cannot use the
1065 translation because it might be encoded incorrectly. */
1068 convd
= &new_conversions
[nconversions
];
1069 convd
->encoding
= encoding
;
1071 /* Find out about the character set the file is encoded with.
1072 This can be found (in textual form) in the entry "". If this
1073 entry does not exist or if this does not contain the 'charset='
1074 information, we will assume the charset matches the one the
1075 current locale and we don't have to perform any conversion. */
1077 convd
->conv
= (__gconv_t
) -1;
1080 convd
->conv
= (iconv_t
) -1;
1085 size_t nullentrylen
;
1087 /* Get the header entry. This is a recursion, but it doesn't
1088 reallocate domain->conversions because we pass
1089 encoding = NULL or convert = 0, respectively. */
1091 # ifdef IN_LIBGLOCALE
1092 _nl_find_msg (domain_file
, domainbinding
, NULL
, "",
1095 _nl_find_msg (domain_file
, domainbinding
, "", 0, &nullentrylen
);
1098 /* Resource problems are fatal. If we continue onwards we will
1099 only attempt to calloc a new conv_tab and fail later. */
1100 if (__builtin_expect (nullentry
== (char *) -1, 0))
1103 if (nullentry
!= NULL
)
1105 const char *charsetstr
;
1107 charsetstr
= strstr (nullentry
, "charset=");
1108 if (charsetstr
!= NULL
)
1112 const char *outcharset
;
1114 charsetstr
+= strlen ("charset=");
1115 len
= strcspn (charsetstr
, " \t\n");
1117 charset
= (char *) alloca (len
+ 1);
1118 # if defined _LIBC || HAVE_MEMPCPY
1119 *((char *) mempcpy (charset
, charsetstr
, len
)) = '\0';
1121 memcpy (charset
, charsetstr
, len
);
1122 charset
[len
] = '\0';
1125 outcharset
= encoding
;
1128 /* We always want to use transliteration. */
1129 outcharset
= norm_add_slashes (outcharset
, "TRANSLIT");
1130 charset
= norm_add_slashes (charset
, "");
1131 int r
= __gconv_open (outcharset
, charset
, &convd
->conv
,
1132 GCONV_AVOID_NOCONV
);
1133 if (__builtin_expect (r
!= __GCONV_OK
, 0))
1135 /* If the output encoding is the same there is
1136 nothing to do. Otherwise do not use the
1137 translation at all. */
1138 if (__builtin_expect (r
!= __GCONV_NULCONV
, 1))
1140 gl_rwlock_unlock (domain
->conversions_lock
);
1141 free ((char *) encoding
);
1145 convd
->conv
= (__gconv_t
) -1;
1149 /* When using GNU libc >= 2.2 or GNU libiconv >= 1.5,
1150 we want to use transliteration. */
1151 # if (((__GLIBC__ == 2 && __GLIBC_MINOR__ >= 2) || __GLIBC__ > 2) \
1152 && !defined __UCLIBC__) \
1153 || _LIBICONV_VERSION >= 0x0105
1154 if (strchr (outcharset
, '/') == NULL
)
1158 len
= strlen (outcharset
);
1159 tmp
= (char *) alloca (len
+ 10 + 1);
1160 memcpy (tmp
, outcharset
, len
);
1161 memcpy (tmp
+ len
, "//TRANSLIT", 10 + 1);
1164 convd
->conv
= iconv_open (outcharset
, charset
);
1170 convd
->conv
= iconv_open (outcharset
, charset
);
1178 convd
->conv_tab
= NULL
;
1179 /* Here domain->conversions is still == new_conversions. */
1180 domain
->nconversions
++;
1184 gl_rwlock_unlock (domain
->conversions_lock
);
1189 convd
->conv
!= (__gconv_t
) -1
1192 convd
->conv
!= (iconv_t
) -1
1197 /* We are supposed to do a conversion. First allocate an
1198 appropriate table with the same structure as the table
1199 of translations in the file, where we can put the pointers
1200 to the converted strings in.
1201 There is a slight complication with plural entries. They
1202 are represented by consecutive NUL terminated strings. We
1203 handle this case by converting RESULTLEN bytes, including
1206 /* This lock primarily protects the memory management variables
1207 freemem, freemem_size. It also protects write accesses to
1208 convd->conv_tab. It's not worth using a separate lock (such
1209 as domain->conversions_lock) for this purpose, because when
1210 modifying convd->conv_tab, we also need to lock freemem,
1211 freemem_size for most of the time. */
1212 __libc_lock_define_initialized (static, lock
)
1214 if (__builtin_expect (convd
->conv_tab
== NULL
, 0))
1216 __libc_lock_lock (lock
);
1217 if (convd
->conv_tab
== NULL
)
1220 (char **) calloc (nstrings
+ domain
->n_sysdep_strings
,
1222 if (convd
->conv_tab
!= NULL
)
1223 goto not_translated_yet
;
1224 /* Mark that we didn't succeed allocating a table. */
1225 convd
->conv_tab
= (char **) -1;
1227 __libc_lock_unlock (lock
);
1230 if (__builtin_expect (convd
->conv_tab
== (char **) -1, 0))
1231 /* Nothing we can do, no more memory. We cannot use the
1232 translation because it might be encoded incorrectly. */
1235 if (convd
->conv_tab
[act
] == NULL
)
1237 /* We haven't used this string so far, so it is not
1238 translated yet. Do this now. */
1239 /* We use a bit more efficient memory handling.
1240 We allocate always larger blocks which get used over
1241 time. This is faster than many small allocations. */
1242 # define INITIAL_BLOCK_SIZE 4080
1243 static unsigned char *freemem
;
1244 static size_t freemem_size
;
1246 const unsigned char *inbuf
;
1247 unsigned char *outbuf
;
1250 transmem_block_t
*transmem_list
;
1253 __libc_lock_lock (lock
);
1256 inbuf
= (const unsigned char *) result
;
1257 outbuf
= freemem
+ sizeof (size_t);
1259 transmem_list
= NULL
;
1265 transmem_block_t
*newmem
;
1267 size_t non_reversible
;
1270 if (freemem_size
< sizeof (size_t))
1271 goto resize_freemem
;
1273 res
= __gconv (convd
->conv
,
1274 &inbuf
, inbuf
+ resultlen
,
1276 outbuf
+ freemem_size
- sizeof (size_t),
1279 if (res
== __GCONV_OK
|| res
== __GCONV_EMPTY_INPUT
)
1282 if (res
!= __GCONV_FULL_OUTPUT
)
1284 /* We should not use the translation at all, it
1285 is incorrectly encoded. */
1286 __libc_lock_unlock (lock
);
1290 inbuf
= (const unsigned char *) result
;
1293 const char *inptr
= (const char *) inbuf
;
1294 size_t inleft
= resultlen
;
1295 char *outptr
= (char *) outbuf
;
1298 if (freemem_size
< sizeof (size_t))
1299 goto resize_freemem
;
1301 outleft
= freemem_size
- sizeof (size_t);
1302 if (iconv (convd
->conv
,
1303 (ICONV_CONST
char **) &inptr
, &inleft
,
1307 outbuf
= (unsigned char *) outptr
;
1312 __libc_lock_unlock (lock
);
1319 /* We must allocate a new buffer or resize the old one. */
1320 if (malloc_count
> 0)
1323 freemem_size
= malloc_count
* INITIAL_BLOCK_SIZE
;
1324 newmem
= (transmem_block_t
*) realloc (transmem_list
,
1328 transmem_list
= newmem
;
1331 struct transmem_list
*old
= transmem_list
;
1333 transmem_list
= transmem_list
->next
;
1341 freemem_size
= INITIAL_BLOCK_SIZE
;
1342 newmem
= (transmem_block_t
*) malloc (freemem_size
);
1346 /* Add the block to the list of blocks we have to free
1348 newmem
->next
= transmem_list
;
1349 transmem_list
= newmem
;
1351 /* Fall through and return -1. */
1354 if (__builtin_expect (newmem
== NULL
, 0))
1358 __libc_lock_unlock (lock
);
1363 freemem
= (unsigned char *) newmem
->data
;
1364 freemem_size
-= offsetof (struct transmem_list
, data
);
1366 transmem_list
= newmem
;
1370 outbuf
= freemem
+ sizeof (size_t);
1373 /* We have now in our buffer a converted string. Put this
1374 into the table of conversions. */
1375 *(size_t *) freemem
= outbuf
- freemem
- sizeof (size_t);
1376 convd
->conv_tab
[act
] = (char *) freemem
;
1377 /* Shrink freemem, but keep it aligned. */
1378 freemem_size
-= outbuf
- freemem
;
1380 freemem
+= freemem_size
& (alignof (size_t) - 1);
1381 freemem_size
= freemem_size
& ~ (alignof (size_t) - 1);
1383 __libc_lock_unlock (lock
);
1386 /* Now convd->conv_tab[act] contains the translation of all
1387 the plural variants. */
1388 result
= convd
->conv_tab
[act
] + sizeof (size_t);
1389 resultlen
= *(size_t *) convd
->conv_tab
[act
];
1393 /* The result string is converted. */
1395 #endif /* _LIBC || HAVE_ICONV */
1397 *lengthp
= resultlen
;
1402 /* Look up a plural variant. */
1405 plural_lookup (struct loaded_l10nfile
*domain
, unsigned long int n
,
1406 const char *translation
, size_t translation_len
)
1408 struct loaded_domain
*domaindata
= (struct loaded_domain
*) domain
->data
;
1409 unsigned long int index
;
1412 index
= plural_eval (domaindata
->plural
, n
);
1413 if (index
>= domaindata
->nplurals
)
1414 /* This should never happen. It means the plural expression and the
1415 given maximum value do not match. */
1418 /* Skip INDEX strings at TRANSLATION. */
1423 p
= __rawmemchr (p
, '\0');
1425 p
= strchr (p
, '\0');
1427 /* And skip over the NUL byte. */
1430 if (p
>= translation
+ translation_len
)
1431 /* This should never happen. It means the plural expression
1432 evaluated to a value larger than the number of variants
1433 available for MSGID1. */
1434 return (char *) translation
;
1440 /* Return string representation of locale CATEGORY. */
1443 category_to_name (int category
)
1451 retval
= "LC_COLLATE";
1456 retval
= "LC_CTYPE";
1461 retval
= "LC_MONETARY";
1466 retval
= "LC_NUMERIC";
1476 retval
= "LC_MESSAGES";
1481 retval
= "LC_RESPONSE";
1486 /* This might not make sense but is perhaps better than any other
1492 /* If you have a better idea for a default value let me know. */
1500 /* Guess value of current locale from value of the environment variables
1501 or system-dependent defaults. */
1504 #ifdef IN_LIBGLOCALE
1505 guess_category_value (int category
, const char *categoryname
,
1509 guess_category_value (int category
, const char *categoryname
)
1512 const char *language
;
1513 #ifndef IN_LIBGLOCALE
1516 const char *language_default
;
1517 int locale_defaulted
;
1521 /* We use the settings in the following order:
1522 1. The value of the environment variable 'LANGUAGE'. This is a GNU
1523 extension. Its value can be a colon-separated list of locale names.
1524 2. The value of the environment variable 'LC_ALL', 'LC_xxx', or 'LANG'.
1525 More precisely, the first among these that is set to a non-empty value.
1526 This is how POSIX specifies it. The value is a single locale name.
1527 3. A system-dependent preference list of languages. Its value can be a
1528 colon-separated list of locale names.
1529 4. A system-dependent default locale name.
1531 - System-dependent settings can be overridden by environment variables.
1532 - If the system provides both a list of languages and a default locale,
1533 the former is used. */
1535 #ifndef IN_LIBGLOCALE
1536 /* Fetch the locale name, through the POSIX method of looking to `LC_ALL',
1537 `LC_xxx', and `LANG'. On some systems this can be done by the
1538 `setlocale' function itself. */
1540 locale
= __current_locale_name (category
);
1542 locale_defaulted
= 0;
1544 locale
= _nl_locale_name_thread_unsafe (category
, categoryname
);
1548 locale
= _nl_locale_name_posix (category
, categoryname
);
1551 locale
= _nl_locale_name_default ();
1552 locale_defaulted
= 1;
1558 /* Ignore LANGUAGE and its system-dependent analogon if the locale is set
1560 1. "C" locale usually uses the ASCII encoding, and most international
1561 messages use non-ASCII characters. These characters get displayed
1562 as question marks (if using glibc's iconv()) or as invalid 8-bit
1563 characters (because other iconv()s refuse to convert most non-ASCII
1564 characters to ASCII). In any case, the output is ugly.
1565 2. The precise output of some programs in the "C" locale is specified
1566 by POSIX and should not depend on environment variables like
1567 "LANGUAGE" or system-dependent information. We allow such programs
1568 to use gettext(). */
1569 if (strcmp (locale
, "C") == 0)
1572 /* The highest priority value is the value of the 'LANGUAGE' environment
1574 language
= getenv ("LANGUAGE");
1575 if (language
!= NULL
&& language
[0] != '\0')
1577 #if !defined IN_LIBGLOCALE && !defined _LIBC
1578 /* The next priority value is the locale name, if not defaulted. */
1579 if (locale_defaulted
)
1581 /* The next priority value is the default language preferences list. */
1582 language_default
= _nl_language_preferences_default ();
1583 if (language_default
!= NULL
)
1584 return language_default
;
1586 /* The least priority value is the locale name, if defaulted. */
1591 #if (defined _LIBC || HAVE_ICONV) && !defined IN_LIBGLOCALE
1592 /* Returns the output charset. */
1595 get_output_charset (struct binding
*domainbinding
)
1597 /* The output charset should normally be determined by the locale. But
1598 sometimes the locale is not used or not correctly set up, so we provide
1599 a possibility for the user to override this: the OUTPUT_CHARSET
1600 environment variable. Moreover, the value specified through
1601 bind_textdomain_codeset overrides both. */
1602 if (domainbinding
!= NULL
&& domainbinding
->codeset
!= NULL
)
1603 return domainbinding
->codeset
;
1606 /* For speed reasons, we look at the value of OUTPUT_CHARSET only
1607 once. This is a user variable that is not supposed to change
1608 during a program run. */
1609 static char *output_charset_cache
;
1610 static int output_charset_cached
;
1612 if (!output_charset_cached
)
1614 const char *value
= getenv ("OUTPUT_CHARSET");
1616 if (value
!= NULL
&& value
[0] != '\0')
1618 size_t len
= strlen (value
) + 1;
1619 char *value_copy
= (char *) malloc (len
);
1621 if (value_copy
!= NULL
)
1622 memcpy (value_copy
, value
, len
);
1623 output_charset_cache
= value_copy
;
1625 output_charset_cached
= 1;
1628 if (output_charset_cache
!= NULL
)
1629 return output_charset_cache
;
1633 return _NL_CURRENT (LC_CTYPE
, CODESET
);
1636 return locale_charset ();
1644 /* @@ begin of epilog @@ */
1646 /* We don't want libintl.a to depend on any other library. So we
1647 avoid the non-standard function stpcpy. In GNU C Library this
1648 function is available, though. Also allow the symbol HAVE_STPCPY
1650 #if !_LIBC && !HAVE_STPCPY
1652 stpcpy (char *dest
, const char *src
)
1654 while ((*dest
++ = *src
++) != '\0')
1660 #if !_LIBC && !HAVE_MEMPCPY
1662 mempcpy (void *dest
, const void *src
, size_t n
)
1664 return (void *) ((char *) memcpy (dest
, src
, n
) + n
);
1668 #if !_LIBC && !HAVE_TSEARCH
1669 # include "tsearch.c"
1674 /* If we want to free all resources we have to do some work at
1676 libc_freeres_fn (free_mem
)
1680 while (_nl_domain_bindings
!= NULL
)
1682 struct binding
*oldp
= _nl_domain_bindings
;
1683 _nl_domain_bindings
= _nl_domain_bindings
->next
;
1684 if (oldp
->dirname
!= _nl_default_dirname
)
1685 /* Yes, this is a pointer comparison. */
1686 free (oldp
->dirname
);
1687 free (oldp
->codeset
);
1691 if (_nl_current_default_domain
!= _nl_default_default_domain
)
1692 /* Yes, again a pointer comparison. */
1693 free ((char *) _nl_current_default_domain
);
1695 /* Remove the search tree with the known translations. */
1696 __tdestroy (root
, free
);
1699 while (transmem_list
!= NULL
)
1701 old
= transmem_list
;
1702 transmem_list
= transmem_list
->next
;