1/* Copyright (C) 1996-2019 Free Software Foundation, Inc.
2 This file is part of the GNU C Library.
3 Contributed by Ulrich Drepper <drepper@gnu.org>, 1996.
4
5 This program is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published
7 by the Free Software Foundation; version 2 of the License, or
8 (at your option) any later version.
9
10 This program 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
13 GNU General Public License for more details.
14
15 You should have received a copy of the GNU General Public License
16 along with this program; if not, see <http://www.gnu.org/licenses/>. */
17
18#ifdef HAVE_CONFIG_H
19# include <config.h>
20#endif
21
22#include <ctype.h>
23#include <errno.h>
24#include <libintl.h>
25#include <limits.h>
26#include <stdio.h>
27#include <stdlib.h>
28#include <string.h>
29#include <stdint.h>
30
31#include "localedef.h"
32#include "linereader.h"
33#include "charmap.h"
34#include "charmap-dir.h"
35
36#include <assert.h>
37
38
39/* Define the lookup function. */
40#include "charmap-kw.h"
41
42
43/* Prototypes for local functions. */
44static struct charmap_t *parse_charmap (struct linereader *cmfile,
45 int verbose, int be_quiet);
46static void new_width (struct linereader *cmfile, struct charmap_t *result,
47 const char *from, const char *to,
48 unsigned long int width);
49static void charmap_new_char (struct linereader *lr, struct charmap_t *cm,
50 size_t nbytes, unsigned char *bytes,
51 const char *from, const char *to,
52 int decimal_ellipsis, int step);
53
54
55bool enc_not_ascii_compatible;
56
57
58#ifdef NEED_NULL_POINTER
59static const char *null_pointer;
60#endif
61
62static struct linereader *
63cmlr_open (const char *directory, const char *name, kw_hash_fct_t hf)
64{
65 FILE *fp;
66
67 fp = charmap_open (directory, name);
68 if (fp == NULL)
69 return NULL;
70 else
71 {
72 size_t dlen = strlen (directory);
73 int add_slash = (dlen == 0 || directory[dlen - 1] != '/');
74 size_t nlen = strlen (name);
75 char *pathname;
76 char *p;
77
78 pathname = alloca (dlen + add_slash + nlen + 1);
79 p = stpcpy (pathname, directory);
80 if (add_slash)
81 *p++ = '/';
82 stpcpy (p, name);
83
84 return lr_create (fp, pathname, hf);
85 }
86}
87
88struct charmap_t *
89charmap_read (const char *filename, int verbose, int error_not_found,
90 int be_quiet, int use_default)
91{
92 struct charmap_t *result = NULL;
93
94 if (filename != NULL)
95 {
96 struct linereader *cmfile;
97
98 /* First try the name as found in the parameter. */
99 cmfile = lr_open (filename, charmap_hash);
100 if (cmfile == NULL)
101 {
102 /* No successful. So start looking through the directories
103 in the I18NPATH if this is a simple name. */
104 if (strchr (filename, '/') == NULL)
105 {
106 char *i18npath = getenv ("I18NPATH");
107 if (i18npath != NULL && *i18npath != '\0')
108 {
109 const size_t pathlen = strlen (i18npath);
110 char i18npathbuf[pathlen + 1];
111 char path[pathlen + sizeof ("/charmaps")];
112 char *next;
113 i18npath = memcpy (i18npathbuf, i18npath, pathlen + 1);
114
115 while (cmfile == NULL
116 && (next = strsep (&i18npath, ":")) != NULL)
117 {
118 stpcpy (stpcpy (path, next), "/charmaps");
119 cmfile = cmlr_open (path, filename, charmap_hash);
120
121 if (cmfile == NULL)
122 /* Try without the "/charmaps" part. */
123 cmfile = cmlr_open (next, filename, charmap_hash);
124 }
125 }
126
127 if (cmfile == NULL)
128 /* Try the default directory. */
129 cmfile = cmlr_open (CHARMAP_PATH, filename, charmap_hash);
130 }
131 }
132
133 if (cmfile != NULL)
134 result = parse_charmap (cmfile, verbose, be_quiet);
135
136 if (result == NULL && error_not_found)
137 record_error (0, errno,
138 _("character map file `%s' not found"),
139 filename);
140 }
141
142 if (result == NULL && filename != NULL && strchr (filename, '/') == NULL)
143 {
144 /* OK, one more try. We also accept the names given to the
145 character sets in the files. Sometimes they differ from the
146 file name. */
147 CHARMAP_DIR *dir;
148
149 dir = charmap_opendir (CHARMAP_PATH);
150 if (dir != NULL)
151 {
152 const char *dirent;
153
154 while ((dirent = charmap_readdir (dir)) != NULL)
155 {
156 char **aliases;
157 char **p;
158 int found;
159
160 aliases = charmap_aliases (CHARMAP_PATH, dirent);
161 found = 0;
162 for (p = aliases; *p; p++)
163 if (strcasecmp (*p, filename) == 0)
164 {
165 found = 1;
166 break;
167 }
168 charmap_free_aliases (aliases);
169
170 if (found)
171 {
172 struct linereader *cmfile;
173
174 cmfile = cmlr_open (CHARMAP_PATH, dirent, charmap_hash);
175 if (cmfile != NULL)
176 result = parse_charmap (cmfile, verbose, be_quiet);
177
178 break;
179 }
180 }
181
182 charmap_closedir (dir);
183 }
184 }
185
186 if (result == NULL && DEFAULT_CHARMAP != NULL)
187 {
188 struct linereader *cmfile;
189
190 cmfile = cmlr_open (CHARMAP_PATH, DEFAULT_CHARMAP, charmap_hash);
191 if (cmfile != NULL)
192 result = parse_charmap (cmfile, verbose, be_quiet);
193
194 if (result == NULL)
195 record_error (4, errno,
196 _("default character map file `%s' not found"),
197 DEFAULT_CHARMAP);
198 }
199
200 if (result != NULL && result->code_set_name == NULL)
201 /* The input file does not specify a code set name. This
202 shouldn't happen but we should cope with it. */
203 result->code_set_name = basename (filename);
204
205 /* Test of ASCII compatibility of locale encoding.
206
207 Verify that the encoding to be used in a locale is ASCII compatible,
208 at least for the graphic characters, excluding the control characters,
209 '$' and '@'. This constraint comes from an ISO C 99 restriction.
210
211 ISO C 99 section 7.17.(2) (about wchar_t):
212 the null character shall have the code value zero and each member of
213 the basic character set shall have a code value equal to its value
214 when used as the lone character in an integer character constant.
215 ISO C 99 section 5.2.1.(3):
216 Both the basic source and basic execution character sets shall have
217 the following members: the 26 uppercase letters of the Latin alphabet
218 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
219 the 26 lowercase letters of the Latin alphabet
220 a b c d e f g h i j k l m n o p q r s t u v w x y z
221 the 10 decimal digits
222 0 1 2 3 4 5 6 7 8 9
223 the following 29 graphic characters
224 ! " # % & ' ( ) * + , - . / : ; < = > ? [ \ ] ^ _ { | } ~
225 the space character, and control characters representing horizontal
226 tab, vertical tab, and form feed.
227
228 Therefore, for all members of the "basic character set", the 'char' code
229 must have the same value as the 'wchar_t' code, which in glibc is the
230 same as the Unicode code, which for all of the enumerated characters
231 is identical to the ASCII code. */
232 if (result != NULL && use_default)
233 {
234 static const char basic_charset[] =
235 {
236 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
237 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z',
238 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
239 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z',
240 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9',
241 '!', '"', '#', '%', '&', '\'', '(', ')', '*', '+', ',', '-',
242 '.', '/', ':', ';', '<', '=', '>', '?', '[', '\\', ']', '^',
243 '_', '{', '|', '}', '~', ' ', '\t', '\v', '\f', '\0'
244 };
245 int failed = 0;
246 const char *p = basic_charset;
247
248 do
249 {
250 struct charseq *seq = charmap_find_symbol (result, p, 1);
251
252 if (seq == NULL || seq->ucs4 != (uint32_t) *p)
253 failed = 1;
254 }
255 while (*p++ != '\0');
256
257 if (failed)
258 {
259 /* A user may disable the ASCII compatibility warning check,
260 but we must remember that the encoding is not ASCII
261 compatible, since it may have other implications. Later
262 we will set _NL_CTYPE_MAP_TO_NONASCII from this value. */
263 if (warn_ascii)
264 record_warning (_(
265"character map `%s' is not ASCII compatible, locale not ISO C compliant "
266"[--no-warnings=ascii]"),
267 result->code_set_name);
268 enc_not_ascii_compatible = true;
269 }
270 }
271
272 return result;
273}
274
275
276static struct charmap_t *
277parse_charmap (struct linereader *cmfile, int verbose, int be_quiet)
278{
279 struct charmap_t *result;
280 int state;
281 enum token_t expected_tok = tok_error;
282 const char *expected_str = NULL;
283 char *from_name = NULL;
284 char *to_name = NULL;
285 enum token_t ellipsis = 0;
286 int step = 1;
287
288 /* We don't want symbolic names in string to be translated. */
289 cmfile->translate_strings = 0;
290
291 /* Allocate room for result. */
292 result = (struct charmap_t *) xmalloc (sizeof (struct charmap_t));
293 memset (result, '\0', sizeof (struct charmap_t));
294 /* The default DEFAULT_WIDTH is 1. */
295 result->width_default = 1;
296
297#define obstack_chunk_alloc malloc
298#define obstack_chunk_free free
299 obstack_init (&result->mem_pool);
300
301 if (init_hash (&result->char_table, 256)
302 || init_hash (&result->byte_table, 256))
303 {
304 free (result);
305 return NULL;
306 }
307
308 /* We use a state machine to describe the charmap description file
309 format. */
310 state = 1;
311 while (1)
312 {
313 /* What's on? */
314 struct token *now = lr_token (cmfile, NULL, NULL, NULL, verbose);
315 enum token_t nowtok = now->tok;
316 struct token *arg;
317
318 if (nowtok == tok_eof)
319 break;
320
321 switch (state)
322 {
323 case 1:
324 /* The beginning. We expect the special declarations, EOL or
325 `CHARMAP'. */
326 if (nowtok == tok_eol)
327 /* Ignore empty lines. */
328 continue;
329
330 if (nowtok == tok_charmap)
331 {
332 from_name = NULL;
333 to_name = NULL;
334
335 /* We have to set up the real work. Fill in some
336 default values. */
337 if (result->mb_cur_max == 0)
338 result->mb_cur_max = 1;
339 if (result->mb_cur_min == 0)
340 result->mb_cur_min = result->mb_cur_max;
341 if (result->mb_cur_min > result->mb_cur_max)
342 {
343 record_error (0, 0, _("\
344%s: <mb_cur_max> must be greater than <mb_cur_min>\n"),
345 cmfile->fname);
346
347 result->mb_cur_min = result->mb_cur_max;
348 }
349
350 lr_ignore_rest (cmfile, 1);
351
352 state = 2;
353 continue;
354 }
355
356 if (nowtok != tok_code_set_name && nowtok != tok_mb_cur_max
357 && nowtok != tok_mb_cur_min && nowtok != tok_escape_char
358 && nowtok != tok_comment_char && nowtok != tok_g0esc
359 && nowtok != tok_g1esc && nowtok != tok_g2esc
360 && nowtok != tok_g3esc && nowtok != tok_repertoiremap
361 && nowtok != tok_include)
362 {
363 lr_error (cmfile, _("syntax error in prolog: %s"),
364 _("invalid definition"));
365
366 lr_ignore_rest (cmfile, 0);
367 continue;
368 }
369
370 /* We know that we need an argument. */
371 arg = lr_token (cmfile, NULL, NULL, NULL, verbose);
372
373 switch (nowtok)
374 {
375 case tok_code_set_name:
376 case tok_repertoiremap:
377 if (arg->tok != tok_ident && arg->tok != tok_string)
378 {
379 badarg:
380 lr_error (cmfile, _("syntax error in prolog: %s"),
381 _("bad argument"));
382
383 lr_ignore_rest (cmfile, 0);
384 continue;
385 }
386
387 if (nowtok == tok_code_set_name)
388 result->code_set_name = obstack_copy0 (&result->mem_pool,
389 arg->val.str.startmb,
390 arg->val.str.lenmb);
391 else
392 result->repertoiremap = obstack_copy0 (&result->mem_pool,
393 arg->val.str.startmb,
394 arg->val.str.lenmb);
395
396 lr_ignore_rest (cmfile, 1);
397 continue;
398
399 case tok_mb_cur_max:
400 case tok_mb_cur_min:
401 if (arg->tok != tok_number)
402 goto badarg;
403
404 if ((nowtok == tok_mb_cur_max
405 && result->mb_cur_max != 0)
406 || (nowtok == tok_mb_cur_max
407 && result->mb_cur_max != 0))
408 lr_error (cmfile, _("duplicate definition of <%s>"),
409 nowtok == tok_mb_cur_min
410 ? "mb_cur_min" : "mb_cur_max");
411
412 if (arg->val.num < 1)
413 {
414 lr_error (cmfile,
415 _("value for <%s> must be 1 or greater"),
416 nowtok == tok_mb_cur_min
417 ? "mb_cur_min" : "mb_cur_max");
418
419 lr_ignore_rest (cmfile, 0);
420 continue;
421 }
422 if ((nowtok == tok_mb_cur_max && result->mb_cur_min != 0
423 && (int) arg->val.num < result->mb_cur_min)
424 || (nowtok == tok_mb_cur_min && result->mb_cur_max != 0
425 && (int) arg->val.num > result->mb_cur_max))
426 {
427 lr_error (cmfile, _("\
428value of <%s> must be greater or equal than the value of <%s>"),
429 "mb_cur_max", "mb_cur_min");
430
431 lr_ignore_rest (cmfile, 0);
432 continue;
433 }
434
435 if (nowtok == tok_mb_cur_max)
436 result->mb_cur_max = arg->val.num;
437 else
438 result->mb_cur_min = arg->val.num;
439
440 lr_ignore_rest (cmfile, 1);
441 continue;
442
443 case tok_escape_char:
444 case tok_comment_char:
445 if (arg->tok != tok_ident)
446 goto badarg;
447
448 if (arg->val.str.lenmb != 1)
449 {
450 lr_error (cmfile, _("\
451argument to <%s> must be a single character"),
452 nowtok == tok_escape_char ? "escape_char"
453 : "comment_char");
454
455 lr_ignore_rest (cmfile, 0);
456 continue;
457 }
458
459 if (nowtok == tok_escape_char)
460 cmfile->escape_char = *arg->val.str.startmb;
461 else
462 cmfile->comment_char = *arg->val.str.startmb;
463
464 lr_ignore_rest (cmfile, 1);
465 continue;
466
467 case tok_g0esc:
468 case tok_g1esc:
469 case tok_g2esc:
470 case tok_g3esc:
471 case tok_escseq:
472 lr_ignore_rest (cmfile, 0); /* XXX */
473 continue;
474
475 case tok_include:
476 lr_error (cmfile, _("\
477character sets with locking states are not supported"));
478 exit (4);
479
480 default:
481 /* Cannot happen. */
482 assert (! "Should not happen");
483 }
484 break;
485
486 case 2:
487 /* We have seen `CHARMAP' and now are in the body. Each line
488 must have the format "%s %s %s\n" or "%s...%s %s %s\n". */
489 if (nowtok == tok_eol)
490 /* Ignore empty lines. */
491 continue;
492
493 if (nowtok == tok_end)
494 {
495 expected_tok = tok_charmap;
496 expected_str = "CHARMAP";
497 state = 90;
498 continue;
499 }
500
501 if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
502 {
503 lr_error (cmfile, _("syntax error in %s definition: %s"),
504 "CHARMAP", _("no symbolic name given"));
505
506 lr_ignore_rest (cmfile, 0);
507 continue;
508 }
509
510 /* If the previous line was not completely correct free the
511 used memory. */
512 if (from_name != NULL)
513 obstack_free (&result->mem_pool, from_name);
514
515 if (nowtok == tok_bsymbol)
516 from_name = (char *) obstack_copy0 (&result->mem_pool,
517 now->val.str.startmb,
518 now->val.str.lenmb);
519 else
520 {
521 obstack_printf (&result->mem_pool, "U%08X",
522 cmfile->token.val.ucs4);
523 obstack_1grow (&result->mem_pool, '\0');
524 from_name = (char *) obstack_finish (&result->mem_pool);
525 }
526 to_name = NULL;
527
528 state = 3;
529 continue;
530
531 case 3:
532 /* We have two possibilities: We can see an ellipsis or an
533 encoding value. */
534 if (nowtok == tok_ellipsis3 || nowtok == tok_ellipsis4
535 || nowtok == tok_ellipsis2 || nowtok == tok_ellipsis4_2
536 || nowtok == tok_ellipsis2_2)
537 {
538 ellipsis = nowtok;
539 if (nowtok == tok_ellipsis4_2)
540 {
541 step = 2;
542 nowtok = tok_ellipsis4;
543 }
544 else if (nowtok == tok_ellipsis2_2)
545 {
546 step = 2;
547 nowtok = tok_ellipsis2;
548 }
549 state = 4;
550 continue;
551 }
552 /* FALLTHROUGH */
553
554 case 5:
555 if (nowtok != tok_charcode)
556 {
557 lr_error (cmfile, _("syntax error in %s definition: %s"),
558 "CHARMAP", _("invalid encoding given"));
559
560 lr_ignore_rest (cmfile, 0);
561
562 state = 2;
563 continue;
564 }
565
566 if (now->val.charcode.nbytes < result->mb_cur_min)
567 lr_error (cmfile, _("too few bytes in character encoding"));
568 else if (now->val.charcode.nbytes > result->mb_cur_max)
569 lr_error (cmfile, _("too many bytes in character encoding"));
570 else
571 charmap_new_char (cmfile, result, now->val.charcode.nbytes,
572 now->val.charcode.bytes, from_name, to_name,
573 ellipsis != tok_ellipsis2, step);
574
575 /* Ignore trailing comment silently. */
576 lr_ignore_rest (cmfile, 0);
577
578 from_name = NULL;
579 to_name = NULL;
580 ellipsis = tok_none;
581 step = 1;
582
583 state = 2;
584 continue;
585
586 case 4:
587 if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
588 {
589 lr_error (cmfile, _("syntax error in %s definition: %s"),
590 "CHARMAP",
591 _("no symbolic name given for end of range"));
592
593 lr_ignore_rest (cmfile, 0);
594 continue;
595 }
596
597 /* Copy the to-name in a safe place. */
598 if (nowtok == tok_bsymbol)
599 to_name = (char *) obstack_copy0 (&result->mem_pool,
600 cmfile->token.val.str.startmb,
601 cmfile->token.val.str.lenmb);
602 else
603 {
604 obstack_printf (&result->mem_pool, "U%08X",
605 cmfile->token.val.ucs4);
606 obstack_1grow (&result->mem_pool, '\0');
607 to_name = (char *) obstack_finish (&result->mem_pool);
608 }
609
610 state = 5;
611 continue;
612
613 case 90:
614 if (nowtok != expected_tok)
615 lr_error (cmfile, _("\
616%1$s: definition does not end with `END %1$s'"), expected_str);
617
618 lr_ignore_rest (cmfile, nowtok == expected_tok);
619 state = 91;
620 continue;
621
622 case 91:
623 /* Waiting for WIDTH... */
624 if (nowtok == tok_eol)
625 /* Ignore empty lines. */
626 continue;
627
628 if (nowtok == tok_width_default)
629 {
630 state = 92;
631 continue;
632 }
633
634 if (nowtok == tok_width)
635 {
636 lr_ignore_rest (cmfile, 1);
637 state = 93;
638 continue;
639 }
640
641 if (nowtok == tok_width_variable)
642 {
643 lr_ignore_rest (cmfile, 1);
644 state = 98;
645 continue;
646 }
647
648 lr_error (cmfile, _("\
649only WIDTH definitions are allowed to follow the CHARMAP definition"));
650
651 lr_ignore_rest (cmfile, 0);
652 continue;
653
654 case 92:
655 if (nowtok != tok_number)
656 lr_error (cmfile, _("value for %s must be an integer"),
657 "WIDTH_DEFAULT");
658 else
659 result->width_default = now->val.num;
660
661 lr_ignore_rest (cmfile, nowtok == tok_number);
662
663 state = 91;
664 continue;
665
666 case 93:
667 /* We now expect `END WIDTH' or lines of the format "%s %d\n" or
668 "%s...%s %d\n". */
669 if (nowtok == tok_eol)
670 /* ignore empty lines. */
671 continue;
672
673 if (nowtok == tok_end)
674 {
675 expected_tok = tok_width;
676 expected_str = "WIDTH";
677 state = 90;
678 continue;
679 }
680
681 if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
682 {
683 lr_error (cmfile, _("syntax error in %s definition: %s"),
684 "WIDTH", _("no symbolic name given"));
685
686 lr_ignore_rest (cmfile, 0);
687 continue;
688 }
689
690 if (from_name != NULL)
691 obstack_free (&result->mem_pool, from_name);
692
693 if (nowtok == tok_bsymbol)
694 from_name = (char *) obstack_copy0 (&result->mem_pool,
695 now->val.str.startmb,
696 now->val.str.lenmb);
697 else
698 {
699 obstack_printf (&result->mem_pool, "U%08X",
700 cmfile->token.val.ucs4);
701 obstack_1grow (&result->mem_pool, '\0');
702 from_name = (char *) obstack_finish (&result->mem_pool);
703 }
704
705 to_name = NULL;
706
707 state = 94;
708 continue;
709
710 case 94:
711 if (nowtok == tok_ellipsis3)
712 {
713 state = 95;
714 continue;
715 }
716 /* Fall through. */
717
718 case 96:
719 if (nowtok != tok_number)
720 lr_error (cmfile, _("value for %s must be an integer"),
721 "WIDTH");
722 else
723 {
724 /* Store width for chars. */
725 new_width (cmfile, result, from_name, to_name, now->val.num);
726
727 from_name = NULL;
728 to_name = NULL;
729 }
730
731 lr_ignore_rest (cmfile, nowtok == tok_number);
732
733 state = 93;
734 continue;
735
736 case 95:
737 if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
738 {
739 lr_error (cmfile, _("syntax error in %s definition: %s"),
740 "WIDTH", _("no symbolic name given for end of range"));
741
742 lr_ignore_rest (cmfile, 0);
743
744 state = 93;
745 continue;
746 }
747
748 if (nowtok == tok_bsymbol)
749 to_name = (char *) obstack_copy0 (&result->mem_pool,
750 now->val.str.startmb,
751 now->val.str.lenmb);
752 else
753 {
754 obstack_printf (&result->mem_pool, "U%08X",
755 cmfile->token.val.ucs4);
756 obstack_1grow (&result->mem_pool, '\0');
757 to_name = (char *) obstack_finish (&result->mem_pool);
758 }
759
760 state = 96;
761 continue;
762
763 case 98:
764 /* We now expect `END WIDTH_VARIABLE' or lines of the format
765 "%s\n" or "%s...%s\n". */
766 if (nowtok == tok_eol)
767 /* ignore empty lines. */
768 continue;
769
770 if (nowtok == tok_end)
771 {
772 expected_tok = tok_width_variable;
773 expected_str = "WIDTH_VARIABLE";
774 state = 90;
775 continue;
776 }
777
778 if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
779 {
780 lr_error (cmfile, _("syntax error in %s definition: %s"),
781 "WIDTH_VARIABLE", _("no symbolic name given"));
782
783 lr_ignore_rest (cmfile, 0);
784
785 continue;
786 }
787
788 if (from_name != NULL)
789 obstack_free (&result->mem_pool, from_name);
790
791 if (nowtok == tok_bsymbol)
792 from_name = (char *) obstack_copy0 (&result->mem_pool,
793 now->val.str.startmb,
794 now->val.str.lenmb);
795 else
796 {
797 obstack_printf (&result->mem_pool, "U%08X",
798 cmfile->token.val.ucs4);
799 obstack_1grow (&result->mem_pool, '\0');
800 from_name = (char *) obstack_finish (&result->mem_pool);
801 }
802 to_name = NULL;
803
804 state = 99;
805 continue;
806
807 case 99:
808 if (nowtok == tok_ellipsis3)
809 state = 100;
810
811 /* Store info. */
812 from_name = NULL;
813
814 /* Warn */
815 state = 98;
816 continue;
817
818 case 100:
819 if (nowtok != tok_bsymbol && nowtok != tok_ucs4)
820 {
821 lr_error (cmfile, _("syntax error in %s definition: %s"),
822 "WIDTH_VARIABLE",
823 _("no symbolic name given for end of range"));
824 lr_ignore_rest (cmfile, 0);
825 continue;
826 }
827
828 if (nowtok == tok_bsymbol)
829 to_name = (char *) obstack_copy0 (&result->mem_pool,
830 now->val.str.startmb,
831 now->val.str.lenmb);
832 else
833 {
834 obstack_printf (&result->mem_pool, "U%08X",
835 cmfile->token.val.ucs4);
836 obstack_1grow (&result->mem_pool, '\0');
837 to_name = (char *) obstack_finish (&result->mem_pool);
838 }
839
840 /* XXX Enter value into table. */
841
842 lr_ignore_rest (cmfile, 1);
843
844 state = 98;
845 continue;
846
847 default:
848 record_error (5, 0, _("%s: error in state machine"),
849 __FILE__);
850 /* NOTREACHED */
851 }
852 break;
853 }
854
855 if (state != 91)
856 record_error (0, 0, _("%s: premature end of file"),
857 cmfile->fname);
858
859 lr_close (cmfile);
860
861 return result;
862}
863
864
865static void
866new_width (struct linereader *cmfile, struct charmap_t *result,
867 const char *from, const char *to, unsigned long int width)
868{
869 struct charseq *from_val;
870 struct charseq *to_val;
871
872 from_val = charmap_find_value (result, from, strlen (from));
873 if (from_val == NULL)
874 {
875 lr_error (cmfile, _("unknown character `%s'"), from);
876 return;
877 }
878
879 if (to == NULL)
880 to_val = from_val;
881 else
882 {
883 to_val = charmap_find_value (result, to, strlen (to));
884 if (to_val == NULL)
885 {
886 lr_error (cmfile, _("unknown character `%s'"), to);
887 return;
888 }
889
890 /* Make sure the number of bytes for the end points of the range
891 is correct. */
892 if (from_val->nbytes != to_val->nbytes)
893 {
894 lr_error (cmfile, _("\
895number of bytes for byte sequence of beginning and end of range not the same: %d vs %d"),
896 from_val->nbytes, to_val->nbytes);
897 return;
898 }
899 }
900
901 if (result->nwidth_rules >= result->nwidth_rules_max)
902 {
903 size_t new_size = result->nwidth_rules + 32;
904 struct width_rule *new_rules =
905 (struct width_rule *) obstack_alloc (&result->mem_pool,
906 (new_size
907 * sizeof (struct width_rule)));
908
909 memcpy (new_rules, result->width_rules,
910 result->nwidth_rules_max * sizeof (struct width_rule));
911
912 result->width_rules = new_rules;
913 result->nwidth_rules_max = new_size;
914 }
915
916 result->width_rules[result->nwidth_rules].from = from_val;
917 result->width_rules[result->nwidth_rules].to = to_val;
918 result->width_rules[result->nwidth_rules].width = (unsigned int) width;
919 ++result->nwidth_rules;
920}
921
922
923struct charseq *
924charmap_find_value (const struct charmap_t *cm, const char *name, size_t len)
925{
926 void *result;
927
928 return (find_entry ((hash_table *) &cm->char_table, name, len, &result)
929 < 0 ? NULL : (struct charseq *) result);
930}
931
932
933static void
934charmap_new_char (struct linereader *lr, struct charmap_t *cm,
935 size_t nbytes, unsigned char *bytes,
936 const char *from, const char *to,
937 int decimal_ellipsis, int step)
938{
939 hash_table *ht = &cm->char_table;
940 hash_table *bt = &cm->byte_table;
941 struct obstack *ob = &cm->mem_pool;
942 char *from_end;
943 char *to_end;
944 const char *cp;
945 int prefix_len, len1, len2;
946 unsigned int from_nr, to_nr, cnt;
947 struct charseq *newp;
948
949 len1 = strlen (from);
950
951 if (to == NULL)
952 {
953 newp = (struct charseq *) obstack_alloc (ob, sizeof (*newp) + nbytes);
954 newp->nbytes = nbytes;
955 memcpy (newp->bytes, bytes, nbytes);
956 newp->name = from;
957
958 newp->ucs4 = UNINITIALIZED_CHAR_VALUE;
959 if ((from[0] == 'U' || from[0] == 'P') && (len1 == 5 || len1 == 9))
960 {
961 /* Maybe the name is of the form `Uxxxx' or `Uxxxxxxxx' where
962 xxxx and xxxxxxxx are hexadecimal numbers. In this case
963 we use the value of xxxx or xxxxxxxx as the UCS4 value of
964 this character and we don't have to consult the repertoire
965 map.
966
967 If the name is of the form `Pxxxx' or `Pxxxxxxxx' the xxxx
968 and xxxxxxxx also give the code point in UCS4 but this must
969 be in the private, i.e., unassigned, area. This should be
970 used for characters which do not (yet) have an equivalent
971 in ISO 10646 and Unicode. */
972 char *endp;
973
974 errno = 0;
975 newp->ucs4 = strtoul (from + 1, &endp, 16);
976 if (endp - from != len1
977 || (newp->ucs4 == ~((uint32_t) 0) && errno == ERANGE)
978 || newp->ucs4 >= 0x80000000)
979 /* This wasn't successful. Signal this name cannot be a
980 correct UCS value. */
981 newp->ucs4 = UNINITIALIZED_CHAR_VALUE;
982 }
983
984 insert_entry (ht, from, len1, newp);
985 insert_entry (bt, newp->bytes, nbytes, newp);
986 /* Please note that it isn't a bug if a symbol is defined more
987 than once. All later definitions are simply discarded. */
988 return;
989 }
990
991 /* We have a range: the names must have names with equal prefixes
992 and an equal number of digits, where the second number is greater
993 or equal than the first. */
994 len2 = strlen (to);
995
996 if (len1 != len2)
997 {
998 illegal_range:
999 lr_error (lr, _("invalid names for character range"));
1000 return;
1001 }
1002
1003 cp = &from[len1 - 1];
1004 if (decimal_ellipsis)
1005 while (isdigit (*cp) && cp >= from)
1006 --cp;
1007 else
1008 while (isxdigit (*cp) && cp >= from)
1009 {
1010 if (!isdigit (*cp) && !isupper (*cp))
1011 lr_error (lr, _("\
1012hexadecimal range format should use only capital characters"));
1013 --cp;
1014 }
1015
1016 prefix_len = (cp - from) + 1;
1017
1018 if (cp == &from[len1 - 1] || strncmp (from, to, prefix_len) != 0)
1019 goto illegal_range;
1020
1021 errno = 0;
1022 from_nr = strtoul (&from[prefix_len], &from_end, decimal_ellipsis ? 10 : 16);
1023 if (*from_end != '\0' || (from_nr == UINT_MAX && errno == ERANGE)
1024 || ((to_nr = strtoul (&to[prefix_len], &to_end,
1025 decimal_ellipsis ? 10 : 16)) == UINT_MAX
1026 && errno == ERANGE)
1027 || *to_end != '\0')
1028 {
1029 lr_error (lr, _("<%s> and <%s> are invalid names for range"), from, to);
1030 return;
1031 }
1032
1033 if (from_nr > to_nr)
1034 {
1035 lr_error (lr, _("upper limit in range is smaller than lower limit"));
1036 return;
1037 }
1038
1039 for (cnt = from_nr; cnt <= to_nr; cnt += step)
1040 {
1041 char *name_end;
1042 obstack_printf (ob, decimal_ellipsis ? "%.*s%0*d" : "%.*s%0*X",
1043 prefix_len, from, len1 - prefix_len, cnt);
1044 obstack_1grow (ob, '\0');
1045 name_end = obstack_finish (ob);
1046
1047 newp = (struct charseq *) obstack_alloc (ob, sizeof (*newp) + nbytes);
1048 newp->nbytes = nbytes;
1049 memcpy (newp->bytes, bytes, nbytes);
1050 newp->name = name_end;
1051
1052 newp->ucs4 = UNINITIALIZED_CHAR_VALUE;
1053 if ((name_end[0] == 'U' || name_end[0] == 'P')
1054 && (len1 == 5 || len1 == 9))
1055 {
1056 /* Maybe the name is of the form `Uxxxx' or `Uxxxxxxxx' where
1057 xxxx and xxxxxxxx are hexadecimal numbers. In this case
1058 we use the value of xxxx or xxxxxxxx as the UCS4 value of
1059 this character and we don't have to consult the repertoire
1060 map.
1061
1062 If the name is of the form `Pxxxx' or `Pxxxxxxxx' the xxxx
1063 and xxxxxxxx also give the code point in UCS4 but this must
1064 be in the private, i.e., unassigned, area. This should be
1065 used for characters which do not (yet) have an equivalent
1066 in ISO 10646 and Unicode. */
1067 char *endp;
1068
1069 errno = 0;
1070 newp->ucs4 = strtoul (name_end + 1, &endp, 16);
1071 if (endp - name_end != len1
1072 || (newp->ucs4 == ~((uint32_t) 0) && errno == ERANGE)
1073 || newp->ucs4 >= 0x80000000)
1074 /* This wasn't successful. Signal this name cannot be a
1075 correct UCS value. */
1076 newp->ucs4 = UNINITIALIZED_CHAR_VALUE;
1077 }
1078
1079 insert_entry (ht, name_end, len1, newp);
1080 insert_entry (bt, newp->bytes, nbytes, newp);
1081 /* Please note we don't examine the return value since it is no error
1082 if we have two definitions for a symbol. */
1083
1084 /* Increment the value in the byte sequence. */
1085 if (++bytes[nbytes - 1] == '\0')
1086 {
1087 int b = nbytes - 2;
1088
1089 do
1090 if (b < 0)
1091 {
1092 lr_error (lr,
1093 _("resulting bytes for range not representable."));
1094 return;
1095 }
1096 while (++bytes[b--] == 0);
1097 }
1098 }
1099}
1100
1101
1102struct charseq *
1103charmap_find_symbol (const struct charmap_t *cm, const char *bytes,
1104 size_t nbytes)
1105{
1106 void *result;
1107
1108 return (find_entry ((hash_table *) &cm->byte_table, bytes, nbytes, &result)
1109 < 0 ? NULL : (struct charseq *) result);
1110}
1111