1/* finddomain.c -- handle list of needed message catalogs
2   Copyright (C) 1995 Software Foundation, Inc.
3   Written by Ulrich Drepper <drepper@gnu.ai.mit.edu>, 1995.
4
5This program is free software; you can redistribute it and/or modify
6it under the terms of the GNU General Public License as published by
7the Free Software Foundation; either version 2, or (at your option)
8any later version.
9
10This program is distributed in the hope that it will be useful,
11but WITHOUT ANY WARRANTY; without even the implied warranty of
12MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
13GNU General Public License for more details.
14
15You should have received a copy of the GNU General Public License
16along with this program; if not, write to the Free Software
17Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
18
19#ifdef HAVE_CONFIG_H
20# include <config.h>
21#endif
22
23#include <errno.h>
24#include <stdio.h>
25#include <sys/types.h>
26
27#if defined STDC_HEADERS || defined _LIBC
28# include <stdlib.h>
29#else
30# ifdef HAVE_MALLOC_H
31#  include <malloc.h>
32# else
33void free ();
34# endif
35#endif
36
37#if defined HAVE_STRING_H || defined _LIBC
38# include <string.h>
39#else
40# include <strings.h>
41#endif
42#if !HAVE_STRCHR && !defined _LIBC
43# ifndef strchr
44#  define strchr index
45# endif
46#endif
47
48#if defined HAVE_UNISTD_H || defined _LIBC
49# include <unistd.h>
50#endif
51
52#include "gettext.h"
53#include "gettextP.h"
54#ifdef _LIBC
55# include <libintl.h>
56#else
57# include "libgettext.h"
58#endif
59
60/* @@ end of prolog @@ */
61
62#ifdef _LIBC
63/* Rename the non ANSI C functions.  This is required by the standard
64   because some ANSI C functions will require linking with this object
65   file and the name space must not be polluted.  */
66# define stpcpy __stpcpy
67#endif
68
69/* Encoding of locale name parts.  */
70#define CEN_REVISION	1
71#define CEN_SPONSOR	2
72#define CEN_SPECIAL	4
73#define XPG_CODESET	8
74#define TERRITORY	16
75#define CEN_AUDIENCE	32
76#define XPG_MODIFIER	64
77
78#define CEN_SPECIFIC	(CEN_REVISION|CEN_SPONSOR|CEN_SPECIAL|CEN_AUDIENCE)
79#define XPG_SPECIFIC	(XPG_CODESET|XPG_MODIFIER)
80
81
82/* List of already loaded domains.  */
83static struct loaded_domain *_nl_loaded_domains;
84
85/* Prototypes for local functions.  */
86static struct loaded_domain *make_entry_rec __P ((const char *dirname,
87						  int mask,
88						  const char *language,
89						  const char *territory,
90						  const char *codeset,
91						  const char *modifier,
92						  const char *special,
93						  const char *sponsor,
94						  const char *revision,
95						  const char *domainname,
96						  int do_allocate));
97
98/* Substitution for systems lacking this function in their C library.  */
99#if !_LIBC && !HAVE_STPCPY
100static char *stpcpy __P ((char *dest, const char *src));
101#endif
102
103
104/* Return a data structure describing the message catalog described by
105   the DOMAINNAME and CATEGORY parameters with respect to the currently
106   established bindings.  */
107struct loaded_domain *
108_nl_find_domain (dirname, locale, domainname)
109     const char *dirname;
110     char *locale;
111     const char *domainname;
112{
113  enum { undecided, xpg, cen } syntax;
114  struct loaded_domain *retval;
115  const char *language;
116  const char *modifier = NULL;
117  const char *territory = NULL;
118  const char *codeset = NULL;
119  const char *special = NULL;
120  const char *sponsor = NULL;
121  const char *revision = NULL;
122  const char *alias_value = NULL;
123  char *cp;
124  int mask;
125
126  /* CATEGORYVALUE now possibly contains a colon separated list of
127     locales.  Each single locale can consist of up to four recognized
128     parts for the XPG syntax:
129
130		language[_territory[.codeset]][@modifier]
131
132     and six parts for the CEN syntax:
133
134	language[_territory][+audience][+special][,sponsor][_revision]
135
136     Beside the first all of them are allowed to be missing.  If the
137     full specified locale is not found, the less specific one are
138     looked for.  The various part will be stripped of according to
139     the following order:
140		(1) revision
141		(2) sponsor
142		(3) special
143		(4) codeset
144		(5) territory
145		(6) audience/modifier
146   */
147
148  /* If we have already tested for this locale entry there has to
149     be one data set in the list of loaded domains.  */
150  retval = make_entry_rec (dirname, 0, locale, NULL, NULL, NULL,
151			   NULL, NULL, NULL, domainname, 0);
152  if (retval != NULL)
153    {
154      /* We know something about this locale.  */
155      int cnt;
156
157      if (retval->decided == 0)
158	_nl_load_domain (retval); /* @@@ */
159
160      if (retval->data != NULL)
161	return retval;
162
163      for (cnt = 0; retval->successor[cnt] != NULL; ++cnt)
164	{
165	  if (retval->successor[cnt]->decided == 0)
166	    _nl_load_domain (retval->successor[cnt]);
167
168	  if (retval->successor[cnt]->data != NULL)
169	    break;
170	}
171
172      /* We really found some usable information.  */
173      return cnt >= 0 ? retval : NULL;
174      /* NOTREACHED */
175    }
176
177  /* See whether the locale value is an alias.  If yes its value
178     *overwrites* the alias name.  No test for the original value is
179     done.  */
180  alias_value = _nl_expand_alias (locale);
181  if (alias_value != NULL)
182    {
183      size_t len = strlen (alias_value) + 1;
184      locale = (char *) malloc (len);
185      if (locale == NULL)
186	return NULL;
187
188      memcpy (locale, alias_value, len);
189    }
190
191  /* Now we determine the single parts of the locale name.  First
192     look for the language.  Termination symbols are `_' and `@' if
193     we use XPG4 style, and `_', `+', and `,' if we use CEN syntax.  */
194  mask = 0;
195  syntax = undecided;
196  language = cp = locale;
197  while (cp[0] != '\0' && cp[0] != '_' && cp[0] != '@'
198	 && cp[0] != '+' && cp[0] != ',')
199    ++cp;
200
201  if (language == cp)
202    /* This does not make sense: language has to be specified.  Use
203       this entry as it is without exploding.  Perhaps it is an alias.  */
204    cp = strchr (language, '\0');
205  else if (cp[0] == '_')
206    {
207      /* Next is the territory.  */
208      cp[0] = '\0';
209      territory = ++cp;
210
211      while (cp[0] != '\0' && cp[0] != '.' && cp[0] != '@'
212	     && cp[0] != '+' && cp[0] != ',' && cp[0] != '_')
213	++cp;
214
215      mask |= TERRITORY;
216
217      if (cp[0] == '.')
218	{
219	  /* Next is the codeset.  */
220	  syntax = xpg;
221	  cp[0] = '\0';
222	  codeset = ++cp;
223
224	  while (cp[0] != '\0' && cp[0] != '@')
225	    ++cp;
226
227	  mask |= XPG_CODESET;
228	}
229    }
230
231  if (cp[0] == '@' || (syntax != xpg && cp[0] == '+'))
232    {
233      /* Next is the modifier.  */
234      syntax = cp[0] == '@' ? xpg : cen;
235      cp[0] = '\0';
236      modifier = ++cp;
237
238      while (syntax == cen && cp[0] != '\0' && cp[0] != '+'
239	     && cp[0] != ',' && cp[0] != '_')
240	++cp;
241
242      mask |= XPG_MODIFIER | CEN_AUDIENCE;
243    }
244
245  if (syntax != xpg && (cp[0] == '+' || cp[0] == ',' || cp[0] == '_'))
246    {
247      syntax = cen;
248
249      if (cp[0] == '+')
250	{
251 	  /* Next is special application (CEN syntax).  */
252	  cp[0] = '\0';
253	  special = ++cp;
254
255	  while (cp[0] != '\0' && cp[0] != ',' && cp[0] != '_')
256	    ++cp;
257
258	  mask |= CEN_SPECIAL;
259	}
260
261      if (cp[0] == ',')
262	{
263 	  /* Next is sponsor (CEN syntax).  */
264	  cp[0] = '\0';
265	  sponsor = ++cp;
266
267	  while (cp[0] != '\0' && cp[0] != '_')
268	    ++cp;
269
270	  mask |= CEN_SPONSOR;
271	}
272
273      if (cp[0] == '_')
274	{
275 	  /* Next is revision (CEN syntax).  */
276	  cp[0] = '\0';
277	  revision = ++cp;
278
279	  mask |= CEN_REVISION;
280	}
281    }
282
283  /* For CEN sytnax values it might be important to have the
284     separator character in the file name, not for XPG syntax.  */
285  if (syntax == xpg)
286    {
287      if (territory != NULL && territory[0] == '\0')
288	mask &= ~TERRITORY;
289
290      if (codeset != NULL && codeset[0] == '\0')
291	mask &= ~XPG_CODESET;
292
293      if (modifier != NULL && modifier[0] == '\0')
294	mask &= ~XPG_MODIFIER;
295    }
296
297  /* Create all possible locale entries which might be interested in
298     generalzation.  */
299  retval = make_entry_rec (dirname, mask, language, territory, codeset,
300			   modifier, special, sponsor, revision,
301			   domainname, 1);
302  if (retval == NULL)
303    /* This means we are out of core.  */
304    return NULL;
305
306  if (retval->decided == 0)
307    _nl_load_domain (retval);
308  if (retval->data == NULL)
309    {
310      int cnt;
311      for (cnt = 0; retval->successor[cnt] != NULL; ++cnt)
312	{
313	  if (retval->successor[cnt]->decided == 0)
314	    _nl_load_domain (retval->successor[cnt]);
315	  if (retval->successor[cnt]->data != NULL)
316	    break;
317
318	  /* Signal that locale is not available.  */
319	  retval->successor[cnt] = NULL;
320	}
321      if (retval->successor[cnt] == NULL)
322	retval = NULL;
323    }
324
325  /* The room for an alias was dynamically allocated.  Free it now.  */
326  if (alias_value != NULL)
327    free (locale);
328
329  return retval;
330}
331
332
333static struct loaded_domain *
334make_entry_rec (dirname, mask, language, territory, codeset, modifier,
335		special, sponsor, revision, domain, do_allocate)
336     const char *dirname;
337     int mask;
338     const char *language;
339     const char *territory;
340     const char *codeset;
341     const char *modifier;
342     const char *special;
343     const char *sponsor;
344     const char *revision;
345     const char *domain;
346     int do_allocate;
347{
348  char *filename = NULL;
349  struct loaded_domain *last = NULL;
350  struct loaded_domain *retval;
351  char *cp;
352  size_t entries;
353  int cnt;
354
355
356  /* Process the current entry described by the MASK only when it is
357     valid.  Because the mask can have in the first call bits from
358     both syntaces set this is necessary to prevent constructing
359     illegal local names.  */
360  /* FIXME: Rewrite because test is necessary only in first round.  */
361  if ((mask & CEN_SPECIFIC) == 0 || (mask & XPG_SPECIFIC) == 0)
362    {
363      /* Allocate room for the full file name.  */
364      filename = (char *) malloc (strlen (dirname) + 1
365				  + strlen (language)
366				  + ((mask & TERRITORY) != 0
367				     ? strlen (territory) : 0)
368				  + ((mask & XPG_CODESET) != 0
369				     ? strlen (codeset) : 0)
370				  + ((mask & XPG_MODIFIER) != 0 ?
371				     strlen (modifier) : 0)
372				  + ((mask & CEN_SPECIAL) != 0
373				     ? strlen (special) : 0)
374				  + ((mask & CEN_SPONSOR) != 0
375				     ? strlen (sponsor) : 0)
376				  + ((mask & CEN_REVISION) != 0
377				     ? strlen (revision) : 0) + 1
378				  + strlen (domain) + 1);
379
380      if (filename == NULL)
381	return NULL;
382
383      retval = NULL;
384      last = NULL;
385
386      /* Construct file name.  */
387      cp = stpcpy (filename, dirname);
388      *cp++ = '/';
389      cp = stpcpy (cp, language);
390
391      if ((mask & TERRITORY) != 0)
392	{
393	  *cp++ = '_';
394	  cp = stpcpy (cp, territory);
395	}
396      if ((mask & XPG_CODESET) != 0)
397	{
398	  *cp++ = '.';
399      cp = stpcpy (cp, codeset);
400	}
401      if ((mask & (XPG_MODIFIER | CEN_AUDIENCE)) != 0)
402	{
403	  /* This component can be part of both syntaces but has different
404	     leading characters.  For CEN we use `+', else `@'.  */
405	  *cp++ = (mask & CEN_AUDIENCE) != 0 ? '+' : '@';
406	  cp = stpcpy (cp, modifier);
407	}
408      if ((mask & CEN_SPECIAL) != 0)
409	{
410	  *cp++ = '+';
411	  cp = stpcpy (cp, special);
412	}
413      if ((mask & CEN_SPONSOR) != 0)
414	{
415	  *cp++ = ',';
416	  cp = stpcpy (cp, sponsor);
417	}
418      if ((mask & CEN_REVISION) != 0)
419	{
420	  *cp++ = '_';
421	  cp = stpcpy (cp, revision);
422	}
423
424      *cp++ = '/';
425      stpcpy (cp, domain);
426
427      /* Look in list of already loaded domains whether it is already
428	 available.  */
429      last = NULL;
430      for (retval = _nl_loaded_domains; retval != NULL; retval = retval->next)
431	if (retval->filename != NULL)
432	  {
433	    int compare = strcmp (retval->filename, filename);
434	    if (compare == 0)
435	      /* We found it!  */
436	      break;
437	    if (compare < 0)
438	      {
439		/* It's not in the list.  */
440		retval = NULL;
441		break;
442	      }
443
444	    last = retval;
445	  }
446
447      if (retval != NULL || do_allocate == 0)
448	{
449	  free (filename);
450	  return retval;
451	}
452    }
453
454  retval = (struct loaded_domain *) malloc (sizeof (*retval));
455  if (retval == NULL)
456    return NULL;
457
458  retval->filename = filename;
459  retval->decided = 0;
460
461  if (last == NULL)
462    {
463      retval->next = _nl_loaded_domains;
464      _nl_loaded_domains = retval;
465    }
466  else
467    {
468      retval->next = last->next;
469      last->next = retval;
470    }
471
472  entries = 0;
473  for (cnt = 126; cnt >= 0; --cnt)
474    if (cnt < mask && (cnt & ~mask) == 0
475	&& ((cnt & CEN_SPECIFIC) == 0 || (cnt & XPG_SPECIFIC) == 0))
476      retval->successor[entries++] = make_entry_rec (dirname, cnt,
477						     language, territory,
478						     codeset, modifier,
479						     special, sponsor,
480						     revision, domain, 1);
481  retval->successor[entries] = NULL;
482
483  return retval;
484}
485
486
487/* @@ begin of epilog @@ */
488
489/* We don't want libintl.a to depend on any other library.  So we
490   avoid the non-standard function stpcpy.  In GNU C Library this
491   function is available, though.  Also allow the symbol HAVE_STPCPY
492   to be defined.  */
493#if !_LIBC && !HAVE_STPCPY
494static char *
495stpcpy (dest, src)
496     char *dest;
497     const char *src;
498{
499  while ((*dest++ = *src++) != '\0')
500    /* Do nothing. */ ;
501  return dest - 1;
502}
503#endif
504