Update copyright notices with scripts/update-copyrights
[glibc.git] / intl / dcigettext.c
blobf3d97fb0be574eaf446267bdc68052f9f1cfe9bd
1 /* Implementation of the internal dcigettext function.
2 Copyright (C) 1995-2014 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
5 The GNU C Library is free software; you can redistribute it and/or
6 modify it under the terms of the GNU Lesser General Public
7 License as published by the Free Software Foundation; either
8 version 2.1 of the License, or (at your option) any later version.
10 The GNU C Library is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 Lesser General Public License for more details.
15 You should have received a copy of the GNU Lesser General Public
16 License along with the GNU C Library; if not, see
17 <http://www.gnu.org/licenses/>. */
19 /* Tell glibc's <string.h> to provide a prototype for mempcpy().
20 This must come before <config.h> because <config.h> may include
21 <features.h>, and once <features.h> has been included, it's too late. */
22 #ifndef _GNU_SOURCE
23 # define _GNU_SOURCE 1
24 #endif
26 #ifdef HAVE_CONFIG_H
27 # include <config.h>
28 #endif
30 #include <sys/types.h>
32 #ifdef __GNUC__
33 # define alloca __builtin_alloca
34 # define HAVE_ALLOCA 1
35 #else
36 # if defined HAVE_ALLOCA_H || defined _LIBC
37 # include <alloca.h>
38 # else
39 # ifdef _AIX
40 #pragma alloca
41 # else
42 # ifndef alloca
43 char *alloca ();
44 # endif
45 # endif
46 # endif
47 #endif
49 #include <errno.h>
50 #ifndef errno
51 extern int errno;
52 #endif
53 #ifndef __set_errno
54 # define __set_errno(val) errno = (val)
55 #endif
57 #include <stddef.h>
58 #include <stdlib.h>
59 #include <string.h>
61 #if defined HAVE_UNISTD_H || defined _LIBC
62 # include <unistd.h>
63 #endif
65 #include <locale.h>
67 #if defined HAVE_SYS_PARAM_H || defined _LIBC
68 # include <sys/param.h>
69 #endif
71 #include "gettextP.h"
72 #include "plural-exp.h"
73 #ifdef _LIBC
74 # include <libintl.h>
75 #else
76 # include "libgnuintl.h"
77 #endif
78 #include "hash-string.h"
80 /* Thread safetyness. */
81 #ifdef _LIBC
82 # include <bits/libc-lock.h>
83 #else
84 /* Provide dummy implementation if this is outside glibc. */
85 # define __libc_lock_define_initialized(CLASS, NAME)
86 # define __libc_lock_lock(NAME)
87 # define __libc_lock_unlock(NAME)
88 # define __libc_rwlock_define_initialized(CLASS, NAME)
89 # define __libc_rwlock_rdlock(NAME)
90 # define __libc_rwlock_unlock(NAME)
91 #endif
93 /* Alignment of types. */
94 #if defined __GNUC__ && __GNUC__ >= 2
95 # define alignof(TYPE) __alignof__ (TYPE)
96 #else
97 # define alignof(TYPE) \
98 ((int) &((struct { char dummy1; TYPE dummy2; } *) 0)->dummy2)
99 #endif
101 /* The internal variables in the standalone libintl.a must have different
102 names than the internal variables in GNU libc, otherwise programs
103 using libintl.a cannot be linked statically. */
104 #if !defined _LIBC
105 # define _nl_default_default_domain libintl_nl_default_default_domain
106 # define _nl_current_default_domain libintl_nl_current_default_domain
107 # define _nl_default_dirname libintl_nl_default_dirname
108 # define _nl_domain_bindings libintl_nl_domain_bindings
109 #endif
111 /* Some compilers, like SunOS4 cc, don't have offsetof in <stddef.h>. */
112 #ifndef offsetof
113 # define offsetof(type,ident) ((size_t)&(((type*)0)->ident))
114 #endif
116 /* @@ end of prolog @@ */
118 #ifdef _LIBC
119 /* Rename the non ANSI C functions. This is required by the standard
120 because some ANSI C functions will require linking with this object
121 file and the name space must not be polluted. */
122 # define getcwd __getcwd
123 # ifndef stpcpy
124 # define stpcpy __stpcpy
125 # endif
126 # define tfind __tfind
127 #else
128 # if !defined HAVE_GETCWD
129 char *getwd ();
130 # define getcwd(buf, max) getwd (buf)
131 # else
132 char *getcwd ();
133 # endif
134 # ifndef HAVE_STPCPY
135 static char *stpcpy PARAMS ((char *dest, const char *src));
136 # endif
137 # ifndef HAVE_MEMPCPY
138 static void *mempcpy PARAMS ((void *dest, const void *src, size_t n));
139 # endif
140 #endif
142 /* Amount to increase buffer size by in each try. */
143 #define PATH_INCR 32
145 /* The following is from pathmax.h. */
146 /* Non-POSIX BSD systems might have gcc's limits.h, which doesn't define
147 PATH_MAX but might cause redefinition warnings when sys/param.h is
148 later included (as on MORE/BSD 4.3). */
149 #if defined _POSIX_VERSION || (defined HAVE_LIMITS_H && !defined __GNUC__)
150 # include <limits.h>
151 #endif
153 #ifndef _POSIX_PATH_MAX
154 # define _POSIX_PATH_MAX 255
155 #endif
157 #if !defined PATH_MAX && defined _PC_PATH_MAX
158 # define PATH_MAX (pathconf ("/", _PC_PATH_MAX) < 1 ? 1024 : pathconf ("/", _PC_PATH_MAX))
159 #endif
161 /* Don't include sys/param.h if it already has been. */
162 #if defined HAVE_SYS_PARAM_H && !defined PATH_MAX && !defined MAXPATHLEN
163 # include <sys/param.h>
164 #endif
166 #if !defined PATH_MAX && defined MAXPATHLEN
167 # define PATH_MAX MAXPATHLEN
168 #endif
170 #ifndef PATH_MAX
171 # define PATH_MAX _POSIX_PATH_MAX
172 #endif
174 /* Whether to support different locales in different threads. */
175 #if defined _LIBC || HAVE_NL_LOCALE_NAME
176 # define HAVE_PER_THREAD_LOCALE
177 #endif
179 /* This is the type used for the search tree where known translations
180 are stored. */
181 struct known_translation_t
183 /* Domain in which to search. */
184 const char *domainname;
186 /* The category. */
187 int category;
189 #ifdef HAVE_PER_THREAD_LOCALE
190 /* Name of the relevant locale category, or "" for the global locale. */
191 const char *localename;
192 #endif
194 /* State of the catalog counter at the point the string was found. */
195 int counter;
197 /* Catalog where the string was found. */
198 struct loaded_l10nfile *domain;
200 /* And finally the translation. */
201 const char *translation;
202 size_t translation_length;
204 /* Pointer to the string in question. */
205 union
207 char appended[ZERO]; /* used if domain != NULL */
208 const char *ptr; /* used if domain == NULL */
210 msgid;
213 /* Root of the search tree with known translations. We can use this
214 only if the system provides the `tsearch' function family. */
215 #if defined HAVE_TSEARCH || defined _LIBC
216 # include <search.h>
218 static void *root;
220 # ifdef _LIBC
221 # define tsearch __tsearch
222 # endif
224 /* Function to compare two entries in the table of known translations. */
225 static int transcmp PARAMS ((const void *p1, const void *p2));
226 static int
227 transcmp (p1, p2)
228 const void *p1;
229 const void *p2;
231 const struct known_translation_t *s1;
232 const struct known_translation_t *s2;
233 int result;
235 s1 = (const struct known_translation_t *) p1;
236 s2 = (const struct known_translation_t *) p2;
238 result = strcmp (s1->domain != NULL ? s1->msgid.appended : s1->msgid.ptr,
239 s2->domain != NULL ? s2->msgid.appended : s2->msgid.ptr);
240 if (result == 0)
242 result = strcmp (s1->domainname, s2->domainname);
243 if (result == 0)
245 #ifdef HAVE_PER_THREAD_LOCALE
246 result = strcmp (s1->localename, s2->localename);
247 if (result == 0)
248 #endif
249 /* We compare the category last (though this is the cheapest
250 operation) since it is hopefully always the same (namely
251 LC_MESSAGES). */
252 result = s1->category - s2->category;
256 return result;
258 #endif
260 /* Name of the default domain used for gettext(3) prior any call to
261 textdomain(3). The default value for this is "messages". */
262 const char _nl_default_default_domain[] attribute_hidden = "messages";
264 /* Value used as the default domain for gettext(3). */
265 const char *_nl_current_default_domain attribute_hidden
266 = _nl_default_default_domain;
268 /* Contains the default location of the message catalogs. */
270 #ifdef _LIBC
271 extern const char _nl_default_dirname[];
272 libc_hidden_proto (_nl_default_dirname)
273 #endif
274 const char _nl_default_dirname[] = LOCALEDIR;
275 #ifdef _LIBC
276 libc_hidden_data_def (_nl_default_dirname)
277 #endif
279 /* List with bindings of specific domains created by bindtextdomain()
280 calls. */
281 struct binding *_nl_domain_bindings;
283 /* Prototypes for local functions. */
284 static char *plural_lookup PARAMS ((struct loaded_l10nfile *domain,
285 unsigned long int n,
286 const char *translation,
287 size_t translation_len))
288 internal_function;
289 static const char *guess_category_value PARAMS ((int category,
290 const char *categoryname))
291 internal_function;
292 #ifdef _LIBC
293 # include "../locale/localeinfo.h"
294 # define category_to_name(category) \
295 _nl_category_names.str + _nl_category_name_idxs[category]
296 #else
297 static const char *category_to_name PARAMS ((int category)) internal_function;
298 #endif
301 /* For those loosing systems which don't have `alloca' we have to add
302 some additional code emulating it. */
303 #ifdef HAVE_ALLOCA
304 /* Nothing has to be done. */
305 # define freea(p) /* nothing */
306 # define ADD_BLOCK(list, address) /* nothing */
307 # define FREE_BLOCKS(list) /* nothing */
308 #else
309 struct block_list
311 void *address;
312 struct block_list *next;
314 # define ADD_BLOCK(list, addr) \
315 do { \
316 struct block_list *newp = (struct block_list *) malloc (sizeof (*newp)); \
317 /* If we cannot get a free block we cannot add the new element to \
318 the list. */ \
319 if (newp != NULL) { \
320 newp->address = (addr); \
321 newp->next = (list); \
322 (list) = newp; \
324 } while (0)
325 # define FREE_BLOCKS(list) \
326 do { \
327 while (list != NULL) { \
328 struct block_list *old = list; \
329 list = list->next; \
330 free (old->address); \
331 free (old); \
333 } while (0)
334 # undef alloca
335 # define alloca(size) (malloc (size))
336 # define freea(p) free (p)
337 #endif /* have alloca */
340 #ifdef _LIBC
341 /* List of blocks allocated for translations. */
342 typedef struct transmem_list
344 struct transmem_list *next;
345 char data[ZERO];
346 } transmem_block_t;
347 static struct transmem_list *transmem_list;
348 #else
349 typedef unsigned char transmem_block_t;
350 #endif
351 #if defined _LIBC || HAVE_ICONV
352 static const char *get_output_charset PARAMS ((struct binding *domainbinding))
353 internal_function;
354 #endif
357 /* Names for the libintl functions are a problem. They must not clash
358 with existing names and they should follow ANSI C. But this source
359 code is also used in GNU C Library where the names have a __
360 prefix. So we have to make a difference here. */
361 #ifdef _LIBC
362 # define DCIGETTEXT __dcigettext
363 #else
364 # define DCIGETTEXT libintl_dcigettext
365 #endif
367 /* Lock variable to protect the global data in the gettext implementation. */
368 #ifdef _LIBC
369 __libc_rwlock_define_initialized (, _nl_state_lock attribute_hidden)
370 #endif
372 /* Checking whether the binaries runs SUID must be done and glibc provides
373 easier methods therefore we make a difference here. */
374 #ifdef _LIBC
375 # define ENABLE_SECURE __libc_enable_secure
376 # define DETERMINE_SECURE
377 #else
378 # ifndef HAVE_GETUID
379 # define getuid() 0
380 # endif
381 # ifndef HAVE_GETGID
382 # define getgid() 0
383 # endif
384 # ifndef HAVE_GETEUID
385 # define geteuid() getuid()
386 # endif
387 # ifndef HAVE_GETEGID
388 # define getegid() getgid()
389 # endif
390 static int enable_secure;
391 # define ENABLE_SECURE (enable_secure == 1)
392 # define DETERMINE_SECURE \
393 if (enable_secure == 0) \
395 if (getuid () != geteuid () || getgid () != getegid ()) \
396 enable_secure = 1; \
397 else \
398 enable_secure = -1; \
400 #endif
402 /* Get the function to evaluate the plural expression. */
403 #include "plural-eval.c"
405 /* Look up MSGID in the DOMAINNAME message catalog for the current
406 CATEGORY locale and, if PLURAL is nonzero, search over string
407 depending on the plural form determined by N. */
408 char *
409 DCIGETTEXT (domainname, msgid1, msgid2, plural, n, category)
410 const char *domainname;
411 const char *msgid1;
412 const char *msgid2;
413 int plural;
414 unsigned long int n;
415 int category;
417 #ifndef HAVE_ALLOCA
418 struct block_list *block_list = NULL;
419 #endif
420 struct loaded_l10nfile *domain;
421 struct binding *binding;
422 const char *categoryname;
423 const char *categoryvalue;
424 char *dirname, *xdomainname;
425 char *single_locale;
426 char *retval;
427 size_t retlen;
428 int saved_errno;
429 #if defined HAVE_TSEARCH || defined _LIBC
430 struct known_translation_t search;
431 struct known_translation_t **foundp = NULL;
432 # ifdef HAVE_PER_THREAD_LOCALE
433 const char *localename;
434 # endif
435 #endif
436 size_t domainname_len;
438 /* If no real MSGID is given return NULL. */
439 if (msgid1 == NULL)
440 return NULL;
442 #ifdef _LIBC
443 if (category < 0 || category >= __LC_LAST || category == LC_ALL)
444 /* Bogus. */
445 return (plural == 0
446 ? (char *) msgid1
447 /* Use the Germanic plural rule. */
448 : n == 1 ? (char *) msgid1 : (char *) msgid2);
449 #endif
451 #ifdef _LIBC
452 __libc_rwlock_define (extern, __libc_setlocale_lock attribute_hidden)
453 __libc_rwlock_rdlock (__libc_setlocale_lock);
454 #endif
456 __libc_rwlock_rdlock (_nl_state_lock);
458 /* If DOMAINNAME is NULL, we are interested in the default domain. If
459 CATEGORY is not LC_MESSAGES this might not make much sense but the
460 definition left this undefined. */
461 if (domainname == NULL)
462 domainname = _nl_current_default_domain;
464 #if defined HAVE_TSEARCH || defined _LIBC
465 /* Try to find the translation among those which we found at
466 some time. */
467 search.domain = NULL;
468 search.msgid.ptr = msgid1;
469 search.domainname = domainname;
470 search.category = category;
471 # ifdef HAVE_PER_THREAD_LOCALE
472 # ifdef _LIBC
473 localename = strdupa (__current_locale_name (category));
474 # endif
475 search.localename = localename;
476 # endif
478 /* Since tfind/tsearch manage a balanced tree, concurrent tfind and
479 tsearch calls can be fatal. */
480 __libc_rwlock_define_initialized (static, tree_lock);
481 __libc_rwlock_rdlock (tree_lock);
483 foundp = (struct known_translation_t **) tfind (&search, &root, transcmp);
485 __libc_rwlock_unlock (tree_lock);
487 if (foundp != NULL && (*foundp)->counter == _nl_msg_cat_cntr)
489 /* Now deal with plural. */
490 if (plural)
491 retval = plural_lookup ((*foundp)->domain, n, (*foundp)->translation,
492 (*foundp)->translation_length);
493 else
494 retval = (char *) (*foundp)->translation;
496 # ifdef _LIBC
497 __libc_rwlock_unlock (__libc_setlocale_lock);
498 # endif
499 __libc_rwlock_unlock (_nl_state_lock);
500 return retval;
502 #endif
504 /* Preserve the `errno' value. */
505 saved_errno = errno;
507 /* See whether this is a SUID binary or not. */
508 DETERMINE_SECURE;
510 /* First find matching binding. */
511 for (binding = _nl_domain_bindings; binding != NULL; binding = binding->next)
513 int compare = strcmp (domainname, binding->domainname);
514 if (compare == 0)
515 /* We found it! */
516 break;
517 if (compare < 0)
519 /* It is not in the list. */
520 binding = NULL;
521 break;
525 if (binding == NULL)
526 dirname = (char *) _nl_default_dirname;
527 else if (binding->dirname[0] == '/')
528 dirname = binding->dirname;
529 else
531 /* We have a relative path. Make it absolute now. */
532 size_t dirname_len = strlen (binding->dirname) + 1;
533 size_t path_max;
534 char *ret;
536 path_max = (unsigned int) PATH_MAX;
537 path_max += 2; /* The getcwd docs say to do this. */
539 for (;;)
541 dirname = (char *) alloca (path_max + dirname_len);
542 ADD_BLOCK (block_list, dirname);
544 __set_errno (0);
545 ret = getcwd (dirname, path_max);
546 if (ret != NULL || errno != ERANGE)
547 break;
549 path_max += path_max / 2;
550 path_max += PATH_INCR;
553 if (ret == NULL)
554 goto no_translation;
556 stpcpy (stpcpy (strchr (dirname, '\0'), "/"), binding->dirname);
559 /* Now determine the symbolic name of CATEGORY and its value. */
560 categoryname = category_to_name (category);
561 categoryvalue = guess_category_value (category, categoryname);
563 domainname_len = strlen (domainname);
564 xdomainname = (char *) alloca (strlen (categoryname)
565 + domainname_len + 5);
566 ADD_BLOCK (block_list, xdomainname);
568 stpcpy (mempcpy (stpcpy (stpcpy (xdomainname, categoryname), "/"),
569 domainname, domainname_len),
570 ".mo");
572 /* Creating working area. */
573 single_locale = (char *) alloca (strlen (categoryvalue) + 1);
574 ADD_BLOCK (block_list, single_locale);
577 /* Search for the given string. This is a loop because we perhaps
578 got an ordered list of languages to consider for the translation. */
579 while (1)
581 /* Make CATEGORYVALUE point to the next element of the list. */
582 while (categoryvalue[0] != '\0' && categoryvalue[0] == ':')
583 ++categoryvalue;
584 if (categoryvalue[0] == '\0')
586 /* The whole contents of CATEGORYVALUE has been searched but
587 no valid entry has been found. We solve this situation
588 by implicitly appending a "C" entry, i.e. no translation
589 will take place. */
590 single_locale[0] = 'C';
591 single_locale[1] = '\0';
593 else
595 char *cp = single_locale;
596 while (categoryvalue[0] != '\0' && categoryvalue[0] != ':')
597 *cp++ = *categoryvalue++;
598 *cp = '\0';
600 /* When this is a SUID binary we must not allow accessing files
601 outside the dedicated directories. */
602 if (ENABLE_SECURE && strchr (single_locale, '/') != NULL)
603 /* Ingore this entry. */
604 continue;
607 /* If the current locale value is C (or POSIX) we don't load a
608 domain. Return the MSGID. */
609 if (strcmp (single_locale, "C") == 0
610 || strcmp (single_locale, "POSIX") == 0)
612 no_translation:
613 FREE_BLOCKS (block_list);
614 __libc_rwlock_unlock (__libc_setlocale_lock);
615 __libc_rwlock_unlock (_nl_state_lock);
616 __set_errno (saved_errno);
617 return (plural == 0
618 ? (char *) msgid1
619 /* Use the Germanic plural rule. */
620 : n == 1 ? (char *) msgid1 : (char *) msgid2);
624 /* Find structure describing the message catalog matching the
625 DOMAINNAME and CATEGORY. */
626 domain = _nl_find_domain (dirname, single_locale, xdomainname, binding);
628 if (domain != NULL)
630 retval = _nl_find_msg (domain, binding, msgid1, 1, &retlen);
632 if (retval == NULL)
634 int cnt;
636 for (cnt = 0; domain->successor[cnt] != NULL; ++cnt)
638 retval = _nl_find_msg (domain->successor[cnt], binding,
639 msgid1, 1, &retlen);
641 /* Resource problems are not fatal, instead we return no
642 translation. */
643 if (__builtin_expect (retval == (char *) -1, 0))
644 goto no_translation;
646 if (retval != NULL)
648 domain = domain->successor[cnt];
649 break;
654 /* Returning -1 means that some resource problem exists
655 (likely memory) and that the strings could not be
656 converted. Return the original strings. */
657 if (__builtin_expect (retval == (char *) -1, 0))
658 goto no_translation;
660 if (retval != NULL)
662 /* Found the translation of MSGID1 in domain DOMAIN:
663 starting at RETVAL, RETLEN bytes. */
664 FREE_BLOCKS (block_list);
665 #if defined HAVE_TSEARCH || defined _LIBC
666 if (foundp == NULL)
668 /* Create a new entry and add it to the search tree. */
669 size_t msgid_len;
670 size_t size;
671 struct known_translation_t *newp;
673 msgid_len = strlen (msgid1) + 1;
674 size = offsetof (struct known_translation_t, msgid)
675 + msgid_len + domainname_len + 1;
676 # ifdef HAVE_PER_THREAD_LOCALE
677 size += strlen (localename) + 1;
678 # endif
679 newp = (struct known_translation_t *) malloc (size);
680 if (newp != NULL)
682 char *new_domainname;
683 # ifdef HAVE_PER_THREAD_LOCALE
684 char *new_localename;
685 # endif
687 new_domainname =
688 mempcpy (newp->msgid.appended, msgid1, msgid_len);
689 memcpy (new_domainname, domainname, domainname_len + 1);
690 # ifdef HAVE_PER_THREAD_LOCALE
691 new_localename = new_domainname + domainname_len + 1;
692 strcpy (new_localename, localename);
693 # endif
694 newp->domainname = new_domainname;
695 newp->category = category;
696 # ifdef HAVE_PER_THREAD_LOCALE
697 newp->localename = new_localename;
698 # endif
699 newp->counter = _nl_msg_cat_cntr;
700 newp->domain = domain;
701 newp->translation = retval;
702 newp->translation_length = retlen;
704 __libc_rwlock_wrlock (tree_lock);
706 /* Insert the entry in the search tree. */
707 foundp = (struct known_translation_t **)
708 tsearch (newp, &root, transcmp);
710 __libc_rwlock_unlock (tree_lock);
712 if (foundp == NULL
713 || __builtin_expect (*foundp != newp, 0))
714 /* The insert failed. */
715 free (newp);
718 else
720 /* We can update the existing entry. */
721 (*foundp)->counter = _nl_msg_cat_cntr;
722 (*foundp)->domain = domain;
723 (*foundp)->translation = retval;
724 (*foundp)->translation_length = retlen;
726 #endif
727 __set_errno (saved_errno);
729 /* Now deal with plural. */
730 if (plural)
731 retval = plural_lookup (domain, n, retval, retlen);
733 __libc_rwlock_unlock (__libc_setlocale_lock);
734 __libc_rwlock_unlock (_nl_state_lock);
735 return retval;
739 /* NOTREACHED */
743 char *
744 internal_function
745 _nl_find_msg (domain_file, domainbinding, msgid, convert, lengthp)
746 struct loaded_l10nfile *domain_file;
747 struct binding *domainbinding;
748 const char *msgid;
749 int convert;
750 size_t *lengthp;
752 struct loaded_domain *domain;
753 nls_uint32 nstrings;
754 size_t act;
755 char *result;
756 size_t resultlen;
758 if (domain_file->decided <= 0)
759 _nl_load_domain (domain_file, domainbinding);
761 if (domain_file->data == NULL)
762 return NULL;
764 domain = (struct loaded_domain *) domain_file->data;
766 nstrings = domain->nstrings;
768 /* Locate the MSGID and its translation. */
769 if (domain->hash_tab != NULL)
771 /* Use the hashing table. */
772 nls_uint32 len = strlen (msgid);
773 nls_uint32 hash_val = __hash_string (msgid);
774 nls_uint32 idx = hash_val % domain->hash_size;
775 nls_uint32 incr = 1 + (hash_val % (domain->hash_size - 2));
777 while (1)
779 nls_uint32 nstr =
780 W (domain->must_swap_hash_tab, domain->hash_tab[idx]);
782 if (nstr == 0)
783 /* Hash table entry is empty. */
784 return NULL;
786 nstr--;
788 /* Compare msgid with the original string at index nstr.
789 We compare the lengths with >=, not ==, because plural entries
790 are represented by strings with an embedded NUL. */
791 if (nstr < nstrings
792 ? W (domain->must_swap, domain->orig_tab[nstr].length) >= len
793 && (strcmp (msgid,
794 domain->data + W (domain->must_swap,
795 domain->orig_tab[nstr].offset))
796 == 0)
797 : domain->orig_sysdep_tab[nstr - nstrings].length > len
798 && (strcmp (msgid,
799 domain->orig_sysdep_tab[nstr - nstrings].pointer)
800 == 0))
802 act = nstr;
803 goto found;
806 if (idx >= domain->hash_size - incr)
807 idx -= domain->hash_size - incr;
808 else
809 idx += incr;
811 /* NOTREACHED */
813 else
815 /* Try the default method: binary search in the sorted array of
816 messages. */
817 size_t top, bottom;
819 bottom = 0;
820 top = nstrings;
821 while (bottom < top)
823 int cmp_val;
825 act = (bottom + top) / 2;
826 cmp_val = strcmp (msgid, (domain->data
827 + W (domain->must_swap,
828 domain->orig_tab[act].offset)));
829 if (cmp_val < 0)
830 top = act;
831 else if (cmp_val > 0)
832 bottom = act + 1;
833 else
834 goto found;
836 /* No translation was found. */
837 return NULL;
840 found:
841 /* The translation was found at index ACT. If we have to convert the
842 string to use a different character set, this is the time. */
843 if (act < nstrings)
845 result = (char *)
846 (domain->data + W (domain->must_swap, domain->trans_tab[act].offset));
847 resultlen = W (domain->must_swap, domain->trans_tab[act].length) + 1;
849 else
851 result = (char *) domain->trans_sysdep_tab[act - nstrings].pointer;
852 resultlen = domain->trans_sysdep_tab[act - nstrings].length;
855 #if defined _LIBC || HAVE_ICONV
856 if (convert)
858 /* We are supposed to do a conversion. */
859 const char *encoding = get_output_charset (domainbinding);
861 /* Protect against reallocation of the table. */
862 __libc_rwlock_rdlock (domain->conversions_lock);
864 /* Search whether a table with converted translations for this
865 encoding has already been allocated. */
866 size_t nconversions = domain->nconversions;
867 struct converted_domain *convd = NULL;
868 size_t i;
870 for (i = nconversions; i > 0; )
872 i--;
873 if (strcmp (domain->conversions[i].encoding, encoding) == 0)
875 convd = &domain->conversions[i];
876 break;
880 __libc_rwlock_unlock (domain->conversions_lock);
882 if (convd == NULL)
884 /* We have to allocate a new conversions table. */
885 __libc_rwlock_wrlock (domain->conversions_lock);
886 nconversions = domain->nconversions;
888 /* Maybe in the meantime somebody added the translation.
889 Recheck. */
890 for (i = nconversions; i > 0; )
892 i--;
893 if (strcmp (domain->conversions[i].encoding, encoding) == 0)
895 convd = &domain->conversions[i];
896 goto found_convd;
900 /* Allocate a table for the converted translations for this
901 encoding. */
902 struct converted_domain *new_conversions =
903 (struct converted_domain *)
904 realloc (domain->conversions,
905 (nconversions + 1) * sizeof (struct converted_domain));
907 if (__builtin_expect (new_conversions == NULL, 0))
909 /* Nothing we can do, no more memory. We cannot use the
910 translation because it might be encoded incorrectly. */
911 unlock_fail:
912 __libc_rwlock_unlock (domain->conversions_lock);
913 return (char *) -1;
916 domain->conversions = new_conversions;
918 /* Copy the 'encoding' string to permanent storage. */
919 encoding = strdup (encoding);
920 if (__builtin_expect (encoding == NULL, 0))
921 /* Nothing we can do, no more memory. We cannot use the
922 translation because it might be encoded incorrectly. */
923 goto unlock_fail;
925 convd = &new_conversions[nconversions];
926 convd->encoding = encoding;
928 /* Find out about the character set the file is encoded with.
929 This can be found (in textual form) in the entry "". If this
930 entry does not exist or if this does not contain the 'charset='
931 information, we will assume the charset matches the one the
932 current locale and we don't have to perform any conversion. */
933 # ifdef _LIBC
934 convd->conv = (__gconv_t) -1;
935 # else
936 # if HAVE_ICONV
937 convd->conv = (iconv_t) -1;
938 # endif
939 # endif
941 char *nullentry;
942 size_t nullentrylen;
944 /* Get the header entry. This is a recursion, but it doesn't
945 reallocate domain->conversions because we pass convert = 0. */
946 nullentry =
947 _nl_find_msg (domain_file, domainbinding, "", 0, &nullentrylen);
949 /* Resource problems are fatal. If we continue onwards we will
950 only attempt to calloc a new conv_tab and fail later. */
951 if (__builtin_expect (nullentry == (char *) -1, 0))
952 return (char *) -1;
954 if (nullentry != NULL)
956 const char *charsetstr;
958 charsetstr = strstr (nullentry, "charset=");
959 if (charsetstr != NULL)
961 size_t len;
962 char *charset;
963 const char *outcharset;
965 charsetstr += strlen ("charset=");
966 len = strcspn (charsetstr, " \t\n");
968 charset = (char *) alloca (len + 1);
969 # if defined _LIBC || HAVE_MEMPCPY
970 *((char *) mempcpy (charset, charsetstr, len)) = '\0';
971 # else
972 memcpy (charset, charsetstr, len);
973 charset[len] = '\0';
974 # endif
976 outcharset = encoding;
978 # ifdef _LIBC
979 /* We always want to use transliteration. */
980 outcharset = norm_add_slashes (outcharset, "TRANSLIT");
981 charset = norm_add_slashes (charset, "");
982 int r = __gconv_open (outcharset, charset, &convd->conv,
983 GCONV_AVOID_NOCONV);
984 if (__builtin_expect (r != __GCONV_OK, 0))
986 /* If the output encoding is the same there is
987 nothing to do. Otherwise do not use the
988 translation at all. */
989 if (__builtin_expect (r != __GCONV_NULCONV, 1))
991 __libc_rwlock_unlock (domain->conversions_lock);
992 free ((char *) encoding);
993 return NULL;
996 convd->conv = (__gconv_t) -1;
998 # else
999 # if HAVE_ICONV
1000 /* When using GNU libc >= 2.2 or GNU libiconv >= 1.5,
1001 we want to use transliteration. */
1002 # if (__GLIBC__ == 2 && __GLIBC_MINOR__ >= 2) || __GLIBC__ > 2 \
1003 || _LIBICONV_VERSION >= 0x0105
1004 if (strchr (outcharset, '/') == NULL)
1006 char *tmp;
1008 len = strlen (outcharset);
1009 tmp = (char *) alloca (len + 10 + 1);
1010 memcpy (tmp, outcharset, len);
1011 memcpy (tmp + len, "//TRANSLIT", 10 + 1);
1012 outcharset = tmp;
1014 convd->conv = iconv_open (outcharset, charset);
1016 freea (outcharset);
1018 else
1019 # endif
1020 convd->conv = iconv_open (outcharset, charset);
1021 # endif
1022 # endif
1024 freea (charset);
1028 convd->conv_tab = NULL;
1029 /* Here domain->conversions is still == new_conversions. */
1030 domain->nconversions++;
1032 found_convd:
1033 __libc_rwlock_unlock (domain->conversions_lock);
1036 if (
1037 # ifdef _LIBC
1038 convd->conv != (__gconv_t) -1
1039 # else
1040 # if HAVE_ICONV
1041 convd->conv != (iconv_t) -1
1042 # endif
1043 # endif
1046 __libc_lock_define_initialized (static, lock)
1047 /* We are supposed to do a conversion. First allocate an
1048 appropriate table with the same structure as the table
1049 of translations in the file, where we can put the pointers
1050 to the converted strings in.
1051 There is a slight complication with plural entries. They
1052 are represented by consecutive NUL terminated strings. We
1053 handle this case by converting RESULTLEN bytes, including
1054 NULs. */
1056 if (__builtin_expect (convd->conv_tab == NULL, 0))
1058 __libc_lock_lock (lock);
1059 if (convd->conv_tab == NULL)
1061 convd->conv_tab
1062 = calloc (nstrings + domain->n_sysdep_strings,
1063 sizeof (char *));
1064 if (convd->conv_tab != NULL)
1065 goto not_translated_yet;
1066 /* Mark that we didn't succeed allocating a table. */
1067 convd->conv_tab = (char **) -1;
1069 __libc_lock_unlock (lock);
1072 if (__builtin_expect (convd->conv_tab == (char **) -1, 0))
1073 /* Nothing we can do, no more memory. We cannot use the
1074 translation because it might be encoded incorrectly. */
1075 return (char *) -1;
1077 if (convd->conv_tab[act] == NULL)
1079 __libc_lock_lock (lock);
1080 not_translated_yet:;
1082 /* We haven't used this string so far, so it is not
1083 translated yet. Do this now. */
1084 /* We use a bit more efficient memory handling.
1085 We allocate always larger blocks which get used over
1086 time. This is faster than many small allocations. */
1087 # define INITIAL_BLOCK_SIZE 4080
1088 static unsigned char *freemem;
1089 static size_t freemem_size;
1091 const unsigned char *inbuf;
1092 unsigned char *outbuf;
1093 int malloc_count;
1094 # ifndef _LIBC
1095 transmem_block_t *transmem_list = NULL;
1096 # endif
1098 inbuf = (const unsigned char *) result;
1099 outbuf = freemem + sizeof (size_t);
1101 malloc_count = 0;
1102 while (1)
1104 transmem_block_t *newmem;
1105 # ifdef _LIBC
1106 size_t non_reversible;
1107 int res;
1109 if (freemem_size < sizeof (size_t))
1110 goto resize_freemem;
1112 res = __gconv (convd->conv,
1113 &inbuf, inbuf + resultlen,
1114 &outbuf,
1115 outbuf + freemem_size - sizeof (size_t),
1116 &non_reversible);
1118 if (res == __GCONV_OK || res == __GCONV_EMPTY_INPUT)
1119 break;
1121 if (res != __GCONV_FULL_OUTPUT)
1123 /* We should not use the translation at all, it
1124 is incorrectly encoded. */
1125 __libc_lock_unlock (lock);
1126 return NULL;
1129 inbuf = (const unsigned char *) result;
1130 # else
1131 # if HAVE_ICONV
1132 const char *inptr = (const char *) inbuf;
1133 size_t inleft = resultlen;
1134 char *outptr = (char *) outbuf;
1135 size_t outleft;
1137 if (freemem_size < sizeof (size_t))
1138 goto resize_freemem;
1140 outleft = freemem_size - sizeof (size_t);
1141 if (iconv (convd->conv,
1142 (ICONV_CONST char **) &inptr, &inleft,
1143 &outptr, &outleft)
1144 != (size_t) (-1))
1146 outbuf = (unsigned char *) outptr;
1147 break;
1149 if (errno != E2BIG)
1151 __libc_lock_unlock (lock);
1152 return NULL;
1154 # endif
1155 # endif
1157 resize_freemem:
1158 /* We must allocate a new buffer or resize the old one. */
1159 if (malloc_count > 0)
1161 ++malloc_count;
1162 freemem_size = malloc_count * INITIAL_BLOCK_SIZE;
1163 newmem = (transmem_block_t *) realloc (transmem_list,
1164 freemem_size);
1165 # ifdef _LIBC
1166 if (newmem != NULL)
1167 transmem_list = newmem;
1168 else
1170 struct transmem_list *old = transmem_list;
1172 transmem_list = transmem_list->next;
1173 free (old);
1175 # endif
1177 else
1179 malloc_count = 1;
1180 freemem_size = INITIAL_BLOCK_SIZE;
1181 newmem = (transmem_block_t *) malloc (freemem_size);
1182 # ifdef _LIBC
1183 if (newmem != NULL)
1185 /* Add the block to the list of blocks we have to free
1186 at some point. */
1187 newmem->next = transmem_list;
1188 transmem_list = newmem;
1190 /* Fall through and return -1. */
1191 # endif
1193 if (__builtin_expect (newmem == NULL, 0))
1195 freemem = NULL;
1196 freemem_size = 0;
1197 __libc_lock_unlock (lock);
1198 return (char *) -1;
1201 # ifdef _LIBC
1202 freemem = (unsigned char *) newmem->data;
1203 freemem_size -= offsetof (struct transmem_list, data);
1204 # else
1205 transmem_list = newmem;
1206 freemem = newmem;
1207 # endif
1209 outbuf = freemem + sizeof (size_t);
1212 /* We have now in our buffer a converted string. Put this
1213 into the table of conversions. */
1214 *(size_t *) freemem = outbuf - freemem - sizeof (size_t);
1215 convd->conv_tab[act] = (char *) freemem;
1216 /* Shrink freemem, but keep it aligned. */
1217 freemem_size -= outbuf - freemem;
1218 freemem = outbuf;
1219 freemem += freemem_size & (alignof (size_t) - 1);
1220 freemem_size = freemem_size & ~ (alignof (size_t) - 1);
1222 __libc_lock_unlock (lock);
1225 /* Now convd->conv_tab[act] contains the translation of all
1226 the plural variants. */
1227 result = convd->conv_tab[act] + sizeof (size_t);
1228 resultlen = *(size_t *) convd->conv_tab[act];
1232 /* The result string is converted. */
1234 #endif /* _LIBC || HAVE_ICONV */
1236 *lengthp = resultlen;
1237 return result;
1241 /* Look up a plural variant. */
1242 static char *
1243 internal_function
1244 plural_lookup (domain, n, translation, translation_len)
1245 struct loaded_l10nfile *domain;
1246 unsigned long int n;
1247 const char *translation;
1248 size_t translation_len;
1250 struct loaded_domain *domaindata = (struct loaded_domain *) domain->data;
1251 unsigned long int index;
1252 const char *p;
1254 index = plural_eval (domaindata->plural, n);
1255 if (index >= domaindata->nplurals)
1256 /* This should never happen. It means the plural expression and the
1257 given maximum value do not match. */
1258 index = 0;
1260 /* Skip INDEX strings at TRANSLATION. */
1261 p = translation;
1262 while (index-- > 0)
1264 #ifdef _LIBC
1265 p = __rawmemchr (p, '\0');
1266 #else
1267 p = strchr (p, '\0');
1268 #endif
1269 /* And skip over the NUL byte. */
1270 p++;
1272 if (p >= translation + translation_len)
1273 /* This should never happen. It means the plural expression
1274 evaluated to a value larger than the number of variants
1275 available for MSGID1. */
1276 return (char *) translation;
1278 return (char *) p;
1281 #ifndef _LIBC
1282 /* Return string representation of locale CATEGORY. */
1283 static const char *
1284 internal_function
1285 category_to_name (category)
1286 int category;
1288 const char *retval;
1290 switch (category)
1292 #ifdef LC_COLLATE
1293 case LC_COLLATE:
1294 retval = "LC_COLLATE";
1295 break;
1296 #endif
1297 #ifdef LC_CTYPE
1298 case LC_CTYPE:
1299 retval = "LC_CTYPE";
1300 break;
1301 #endif
1302 #ifdef LC_MONETARY
1303 case LC_MONETARY:
1304 retval = "LC_MONETARY";
1305 break;
1306 #endif
1307 #ifdef LC_NUMERIC
1308 case LC_NUMERIC:
1309 retval = "LC_NUMERIC";
1310 break;
1311 #endif
1312 #ifdef LC_TIME
1313 case LC_TIME:
1314 retval = "LC_TIME";
1315 break;
1316 #endif
1317 #ifdef LC_MESSAGES
1318 case LC_MESSAGES:
1319 retval = "LC_MESSAGES";
1320 break;
1321 #endif
1322 #ifdef LC_RESPONSE
1323 case LC_RESPONSE:
1324 retval = "LC_RESPONSE";
1325 break;
1326 #endif
1327 #ifdef LC_ALL
1328 case LC_ALL:
1329 /* This might not make sense but is perhaps better than any other
1330 value. */
1331 retval = "LC_ALL";
1332 break;
1333 #endif
1334 default:
1335 /* If you have a better idea for a default value let me know. */
1336 retval = "LC_XXX";
1339 return retval;
1341 #endif
1343 /* Guess value of current locale from value of the environment variables. */
1344 static const char *
1345 internal_function
1346 guess_category_value (category, categoryname)
1347 int category;
1348 const char *categoryname;
1350 const char *language;
1351 const char *retval;
1353 /* The highest priority value is the `LANGUAGE' environment
1354 variable. But we don't use the value if the currently selected
1355 locale is the C locale. This is a GNU extension. */
1356 language = getenv ("LANGUAGE");
1357 if (language != NULL && language[0] == '\0')
1358 language = NULL;
1360 /* We have to proceed with the POSIX methods of looking to `LC_ALL',
1361 `LC_xxx', and `LANG'. On some systems this can be done by the
1362 `setlocale' function itself. */
1363 #ifdef _LIBC
1364 retval = __current_locale_name (category);
1365 #else
1366 retval = _nl_locale_name (category, categoryname);
1367 #endif
1369 return language != NULL && strcmp (retval, "C") != 0 ? language : retval;
1372 #if defined _LIBC || HAVE_ICONV
1373 /* Returns the output charset. */
1374 static const char *
1375 internal_function
1376 get_output_charset (domainbinding)
1377 struct binding *domainbinding;
1379 /* The output charset should normally be determined by the locale. But
1380 sometimes the locale is not used or not correctly set up, so we provide
1381 a possibility for the user to override this: the OUTPUT_CHARSET
1382 environment variable. Moreover, the value specified through
1383 bind_textdomain_codeset overrides both. */
1384 if (domainbinding != NULL && domainbinding->codeset != NULL)
1385 return domainbinding->codeset;
1386 else
1388 /* For speed reasons, we look at the value of OUTPUT_CHARSET only
1389 once. This is a user variable that is not supposed to change
1390 during a program run. */
1391 static char *output_charset_cache;
1392 static int output_charset_cached;
1394 if (!output_charset_cached)
1396 const char *value = getenv ("OUTPUT_CHARSET");
1398 if (value != NULL && value[0] != '\0')
1400 size_t len = strlen (value) + 1;
1401 char *value_copy = (char *) malloc (len);
1403 if (value_copy != NULL)
1404 memcpy (value_copy, value, len);
1405 output_charset_cache = value_copy;
1407 output_charset_cached = 1;
1410 if (output_charset_cache != NULL)
1411 return output_charset_cache;
1412 else
1414 # ifdef _LIBC
1415 return _NL_CURRENT (LC_CTYPE, CODESET);
1416 # else
1417 # if HAVE_ICONV
1418 extern const char *locale_charset PARAMS ((void));
1419 return locale_charset ();
1420 # endif
1421 # endif
1425 #endif
1427 /* @@ begin of epilog @@ */
1429 /* We don't want libintl.a to depend on any other library. So we
1430 avoid the non-standard function stpcpy. In GNU C Library this
1431 function is available, though. Also allow the symbol HAVE_STPCPY
1432 to be defined. */
1433 #if !_LIBC && !HAVE_STPCPY
1434 static char *
1435 stpcpy (dest, src)
1436 char *dest;
1437 const char *src;
1439 while ((*dest++ = *src++) != '\0')
1440 /* Do nothing. */ ;
1441 return dest - 1;
1443 #endif
1445 #if !_LIBC && !HAVE_MEMPCPY
1446 static void *
1447 mempcpy (dest, src, n)
1448 void *dest;
1449 const void *src;
1450 size_t n;
1452 return (void *) ((char *) memcpy (dest, src, n) + n);
1454 #endif
1457 #ifdef _LIBC
1458 /* If we want to free all resources we have to do some work at
1459 program's end. */
1460 libc_freeres_fn (free_mem)
1462 void *old;
1464 while (_nl_domain_bindings != NULL)
1466 struct binding *oldp = _nl_domain_bindings;
1467 _nl_domain_bindings = _nl_domain_bindings->next;
1468 if (oldp->dirname != _nl_default_dirname)
1469 /* Yes, this is a pointer comparison. */
1470 free (oldp->dirname);
1471 free (oldp->codeset);
1472 free (oldp);
1475 if (_nl_current_default_domain != _nl_default_default_domain)
1476 /* Yes, again a pointer comparison. */
1477 free ((char *) _nl_current_default_domain);
1479 /* Remove the search tree with the known translations. */
1480 __tdestroy (root, free);
1481 root = NULL;
1483 while (transmem_list != NULL)
1485 old = transmem_list;
1486 transmem_list = transmem_list->next;
1487 free (old);
1490 #endif