1 /*
2 * This file and its contents are supplied under the terms of the
3 * Common Development and Distribution License ("CDDL"), version 1.0.
4 * You may only use this file in accordance with the terms of version
5 * 1.0 of the CDDL.
6 *
7 * A full copy of the text of the CDDL should have accompanied this
8 * source. A copy of the CDDL is also available via the Internet at
9 * http://www.illumos.org/license/CDDL.
10 */
11
12 /*
13 * Copyright 2010 Nexenta Systems, Inc. All rights reserved.
14 * Copyright 2013 DEY Storage Systems, Inc.
15 */
16
17 /*
18 * This file contains the "scanner", which tokenizes the input files
19 * for localedef for processing by the higher level grammar processor.
20 */
21
22 #include <stdio.h>
23 #include <stdlib.h>
24 #include <ctype.h>
25 #include <limits.h>
26 #include <string.h>
27 #include <widec.h>
28 #include <sys/types.h>
29 #include <assert.h>
30 #include "localedef.h"
31 #include "parser.tab.h"
32
33 int com_char = '#';
34 int esc_char = '\\';
35 int mb_cur_min = 1;
36 int mb_cur_max = 1;
37 int lineno = 1;
38 int warnings = 0;
39 static int nextline;
40 static FILE *input = stdin;
41 static const char *filename = "<stdin>";
42 static int instring = 0;
43 static int escaped = 0;
44
45 /*
46 * Token space ... grows on demand.
47 */
48 static char *token = NULL;
49 static int tokidx;
50 static int toksz = 0;
51 static int hadtok = 0;
52
53 /*
54 * Wide string space ... grows on demand.
55 */
56 static wchar_t *widestr = NULL;
57 static int wideidx = 0;
58 static int widesz = 0;
59
60 /*
61 * The last keyword seen. This is useful to trigger the special lexer rules
62 * for "copy" and also collating symbols and elements.
63 */
64 int last_kw = 0;
65 static int category = T_END;
66
67 static struct token {
68 int id;
69 const char *name;
70 } keywords[] = {
71 { T_COM_CHAR, "comment_char" },
72 { T_ESC_CHAR, "escape_char" },
73 { T_END, "END" },
74 { T_COPY, "copy" },
75 { T_MESSAGES, "LC_MESSAGES" },
76 { T_YESSTR, "yesstr" },
77 { T_YESEXPR, "yesexpr" },
78 { T_NOSTR, "nostr" },
79 { T_NOEXPR, "noexpr" },
80 { T_MONETARY, "LC_MONETARY" },
81 { T_INT_CURR_SYMBOL, "int_curr_symbol" },
82 { T_CURRENCY_SYMBOL, "currency_symbol" },
83 { T_MON_DECIMAL_POINT, "mon_decimal_point" },
84 { T_MON_THOUSANDS_SEP, "mon_thousands_sep" },
85 { T_POSITIVE_SIGN, "positive_sign" },
86 { T_NEGATIVE_SIGN, "negative_sign" },
87 { T_MON_GROUPING, "mon_grouping" },
88 { T_INT_FRAC_DIGITS, "int_frac_digits" },
89 { T_FRAC_DIGITS, "frac_digits" },
90 { T_P_CS_PRECEDES, "p_cs_precedes" },
91 { T_P_SEP_BY_SPACE, "p_sep_by_space" },
92 { T_N_CS_PRECEDES, "n_cs_precedes" },
93 { T_N_SEP_BY_SPACE, "n_sep_by_space" },
94 { T_P_SIGN_POSN, "p_sign_posn" },
95 { T_N_SIGN_POSN, "n_sign_posn" },
96 { T_INT_P_CS_PRECEDES, "int_p_cs_precedes" },
97 { T_INT_N_CS_PRECEDES, "int_n_cs_precedes" },
98 { T_INT_P_SEP_BY_SPACE, "int_p_sep_by_space" },
99 { T_INT_N_SEP_BY_SPACE, "int_n_sep_by_space" },
100 { T_INT_P_SIGN_POSN, "int_p_sign_posn" },
101 { T_INT_N_SIGN_POSN, "int_n_sign_posn" },
102 { T_COLLATE, "LC_COLLATE" },
103 { T_COLLATING_SYMBOL, "collating-symbol" },
104 { T_COLLATING_ELEMENT, "collating-element" },
105 { T_FROM, "from" },
106 { T_ORDER_START, "order_start" },
107 { T_ORDER_END, "order_end" },
108 { T_FORWARD, "forward" },
109 { T_BACKWARD, "backward" },
110 { T_POSITION, "position" },
111 { T_IGNORE, "IGNORE" },
112 { T_UNDEFINED, "UNDEFINED" },
113 { T_NUMERIC, "LC_NUMERIC" },
114 { T_DECIMAL_POINT, "decimal_point" },
115 { T_THOUSANDS_SEP, "thousands_sep" },
116 { T_GROUPING, "grouping" },
117 { T_TIME, "LC_TIME" },
118 { T_ABDAY, "abday" },
119 { T_DAY, "day" },
120 { T_ABMON, "abmon" },
121 { T_MON, "mon" },
122 { T_D_T_FMT, "d_t_fmt" },
123 { T_D_FMT, "d_fmt" },
124 { T_T_FMT, "t_fmt" },
125 { T_AM_PM, "am_pm" },
126 { T_T_FMT_AMPM, "t_fmt_ampm" },
127 { T_ERA, "era" },
128 { T_ERA_D_FMT, "era_d_fmt" },
129 { T_ERA_T_FMT, "era_t_fmt" },
130 { T_ERA_D_T_FMT, "era_d_t_fmt" },
131 { T_ALT_DIGITS, "alt_digits" },
132 { T_CTYPE, "LC_CTYPE" },
133 { T_ISUPPER, "upper" },
134 { T_ISLOWER, "lower" },
135 { T_ISALPHA, "alpha" },
136 { T_ISDIGIT, "digit" },
137 { T_ISPUNCT, "punct" },
138 { T_ISXDIGIT, "xdigit" },
139 { T_ISSPACE, "space" },
140 { T_ISPRINT, "print" },
141 { T_ISGRAPH, "graph" },
142 { T_ISBLANK, "blank" },
143 { T_ISCNTRL, "cntrl" },
144 /*
145 * These entries are local additions, and not specified by
146 * TOG. Note that they are not guaranteed to be accurate for
147 * all locales, and so applications should not depend on them.
148 */
149 { T_ISSPECIAL, "special" },
150 { T_ISENGLISH, "english" },
151 { T_ISPHONOGRAM, "phonogram" },
152 { T_ISIDEOGRAM, "ideogram" },
153 { T_ISNUMBER, "number" },
154 /*
155 * We have to support this in the grammar, but it would be a
156 * syntax error to define a character as one of these without
157 * also defining it as an alpha or digit. We ignore it in our
158 * parsing.
159 */
160 { T_ISALNUM, "alnum" },
161 { T_TOUPPER, "toupper" },
162 { T_TOLOWER, "tolower" },
163
164 /*
165 * These are keywords used in the charmap file. Note that
166 * Solaris originally used angle brackets to wrap some of them,
167 * but we removed that to simplify our parser. The first of these
168 * items are "global items."
169 */
170 { T_CHARMAP, "CHARMAP" },
171 { T_WIDTH, "WIDTH" },
172
173 { -1, NULL },
174 };
175
176 /*
177 * These special words are only used in a charmap file, enclosed in <>.
178 */
179 static struct token symwords[] = {
180 { T_COM_CHAR, "comment_char" },
181 { T_ESC_CHAR, "escape_char" },
182 { T_CODE_SET, "code_set_name" },
183 { T_MB_CUR_MAX, "mb_cur_max" },
184 { T_MB_CUR_MIN, "mb_cur_min" },
185 { -1, NULL },
186 };
187
188 static int categories[] = {
189 T_CHARMAP,
190 T_CTYPE,
191 T_COLLATE,
192 T_MESSAGES,
193 T_MONETARY,
194 T_NUMERIC,
195 T_TIME,
196 T_WIDTH,
197 0
198 };
199
200 void
reset_scanner(const char * fname)201 reset_scanner(const char *fname)
202 {
203 if (fname == NULL) {
204 filename = "<stdin>";
205 input = stdin;
206 } else {
207 if (input != stdin)
208 (void) fclose(input);
209 if ((input = fopen(fname, "r")) == NULL) {
210 perror("fopen");
211 exit(4);
212 }
213 filename = fname;
214 }
215 com_char = '#';
216 esc_char = '\\';
217 instring = 0;
218 escaped = 0;
219 lineno = 1;
220 nextline = 1;
221 tokidx = 0;
222 wideidx = 0;
223 }
224
225 #define hex(x) \
226 (isdigit(x) ? (x - '0') : ((islower(x) ? (x - 'a') : (x - 'A')) + 10))
227 #define isodigit(x) ((x >= '0') && (x <= '7'))
228
229 static int
scanc(void)230 scanc(void)
231 {
232 int c;
233
234 c = getc(input);
235 lineno = nextline;
236 if (c == '\n') {
237 nextline++;
238 }
239 return (c);
240 }
241
242 static void
unscanc(int c)243 unscanc(int c)
244 {
245 if (c == '\n') {
246 nextline--;
247 }
248 if (ungetc(c, input) < 0) {
249 (void) yyerror(_("ungetc failed"));
250 }
251 }
252
253 static int
scan_hex_byte(void)254 scan_hex_byte(void)
255 {
256 int c1, c2;
257 int v;
258
259 c1 = scanc();
260 if (!isxdigit(c1)) {
261 (void) yyerror(_("malformed hex digit"));
262 }
263 c2 = scanc();
264 if (!isxdigit(c2)) {
265 (void) yyerror(_("malformed hex digit"));
266 }
267 v = ((hex(c1) << 4) | hex(c2));
268 return (v);
269 }
270
271 static int
scan_dec_byte(void)272 scan_dec_byte(void)
273 {
274 int c1, c2, c3;
275 int b;
276
277 c1 = scanc();
278 if (!isdigit(c1)) {
279 (void) yyerror(_("malformed decimal digit"));
280 }
281 b = c1 - '0';
282 c2 = scanc();
283 if (!isdigit(c2)) {
284 (void) yyerror(_("malformed decimal digit"));
285 }
286 b *= 10;
287 b += (c2 - '0');
288 c3 = scanc();
289 if (!isdigit(c3)) {
290 unscanc(c3);
291 } else {
292 b *= 10;
293 b += (c3 - '0');
294 }
295 return (b);
296 }
297
298 static int
scan_oct_byte(void)299 scan_oct_byte(void)
300 {
301 int c1, c2, c3;
302 int b;
303
304 b = 0;
305
306 c1 = scanc();
307 if (!isodigit(c1)) {
308 (void) yyerror(_("malformed octal digit"));
309 }
310 b = c1 - '0';
311 c2 = scanc();
312 if (!isodigit(c2)) {
313 (void) yyerror(_("malformed octal digit"));
314 }
315 b *= 8;
316 b += (c2 - '0');
317 c3 = scanc();
318 if (!isodigit(c3)) {
319 unscanc(c3);
320 } else {
321 b *= 8;
322 b += (c3 - '0');
323 }
324 return (b);
325 }
326
327 void
add_tok(int c)328 add_tok(int c)
329 {
330 if ((tokidx + 1) >= toksz) {
331 toksz += 64;
332 if ((token = realloc(token, toksz)) == NULL) {
333 (void) yyerror(_("out of memory"));
334 }
335 }
336
337 token[tokidx++] = (char)c;
338 token[tokidx] = 0;
339 }
340 void
add_wcs(wchar_t c)341 add_wcs(wchar_t c)
342 {
343 if ((wideidx + 1) >= widesz) {
344 widesz += 64;
345 widestr = realloc(widestr, (widesz * sizeof (wchar_t)));
346 if (widestr == NULL) {
347 (void) yyerror(_("out of memory"));
348 }
349 }
350
351 widestr[wideidx++] = c;
352 widestr[wideidx] = 0;
353 }
354
355 wchar_t *
get_wcs(void)356 get_wcs(void)
357 {
358 wchar_t *ws = widestr;
359 wideidx = 0;
360 widestr = NULL;
361 widesz = 0;
362 if (ws == NULL) {
363 if ((ws = wsdup(L"")) == NULL) {
364 (void) yyerror(_("out of memory"));
365 }
366 }
367 return (ws);
368 }
369
370 static int
get_byte(void)371 get_byte(void)
372 {
373 int c;
374
375 if ((c = scanc()) != esc_char) {
376 unscanc(c);
377 return (EOF);
378 }
379 c = scanc();
380
381 switch (c) {
382 case 'd':
383 case 'D':
384 return (scan_dec_byte());
385 case 'x':
386 case 'X':
387 return (scan_hex_byte());
388 case '0':
389 case '1':
390 case '2':
391 case '3':
392 case '4':
393 case '5':
394 case '6':
395 case '7':
396 /* put the character back so we can get it */
397 unscanc(c);
398 return (scan_oct_byte());
399 default:
400 unscanc(c);
401 unscanc(esc_char);
402 return (EOF);
403 }
404 }
405
406 int
get_escaped(int c)407 get_escaped(int c)
408 {
409 switch (c) {
410 case 'n':
411 return ('\n');
412 case 'r':
413 return ('\r');
414 case 't':
415 return ('\t');
416 case 'f':
417 return ('\f');
418 case 'v':
419 return ('\v');
420 case 'b':
421 return ('\b');
422 case 'a':
423 return ('\a');
424 default:
425 return (c);
426 }
427 }
428
429 int
get_wide(void)430 get_wide(void)
431 {
432 static char mbs[MB_LEN_MAX + 1] = "";
433 static int mbi = 0;
434 int c;
435 wchar_t wc;
436
437 if (mb_cur_max >= sizeof (mbs)) {
438 (void) yyerror(_("max multibyte character size too big"));
439 }
440 for (;;) {
441 if ((mbi == mb_cur_max) || ((c = get_byte()) == EOF)) {
442 /*
443 * end of the byte sequence reached, but no
444 * valid wide decoding. fatal error.
445 */
446 mbi = 0;
447 (void) yyerror(_("not a valid character encoding"));
448 }
449 mbs[mbi++] = c;
450 mbs[mbi] = 0;
451
452 /* does it decode? */
453 if (to_wide(&wc, mbs) >= 0) {
454 break;
455 }
456 }
457
458 mbi = 0;
459 if ((category != T_CHARMAP) && (category != T_WIDTH)) {
460 if (check_charmap(wc) < 0) {
461 (void) yyerror(_("no symbolic name for character"));
462 }
463 }
464
465 yylval.wc = wc;
466 return (T_CHAR);
467 }
468
469 int
get_symbol(void)470 get_symbol(void)
471 {
472 int c;
473
474 while ((c = scanc()) != EOF) {
475 if (escaped) {
476 escaped = 0;
477 if (c == '\n')
478 continue;
479 add_tok(get_escaped(c));
480 continue;
481 }
482 if (c == esc_char) {
483 escaped = 1;
484 continue;
485 }
486 if (c == '\n') { /* well that's strange! */
487 (void) yyerror(_("unterminated symbolic name"));
488 }
489 if (c == '>') { /* end of symbol */
490
491 /*
492 * This restarts the token from the beginning
493 * the next time we scan a character. (This
494 * token is complete.)
495 */
496
497 if (token == NULL) {
498 (void) yyerror(_("missing symbolic name"));
499 }
500 tokidx = 0;
501
502 /*
503 * A few symbols are handled as keywords outside
504 * of the normal categories.
505 */
506 if (category == T_END) {
507 int i;
508 for (i = 0; symwords[i].name != 0; i++) {
509 if (strcmp(token, symwords[i].name) ==
510 0) {
511 last_kw = symwords[i].id;
512 return (last_kw);
513 }
514 }
515 }
516 /*
517 * Contextual rule: Only literal characters are
518 * permitted in CHARMAP. Anywhere else the symbolic
519 * forms are fine.
520 */
521 if ((category != T_CHARMAP) &&
522 (lookup_charmap(token, &yylval.wc)) != -1) {
523 return (T_CHAR);
524 }
525 if ((yylval.collsym = lookup_collsym(token)) != NULL) {
526 return (T_COLLSYM);
527 }
528 if ((yylval.collelem = lookup_collelem(token)) !=
529 NULL) {
530 return (T_COLLELEM);
531 }
532 /* its an undefined symbol */
533 yylval.token = strdup(token);
534 token = NULL;
535 toksz = 0;
536 tokidx = 0;
537 return (T_SYMBOL);
538 }
539 add_tok(c);
540 }
541
542 (void) yyerror(_("unterminated symbolic name"));
543 return (EOF);
544 }
545
546 int
get_category(void)547 get_category(void)
548 {
549 return (category);
550 }
551
552 static int
consume_token(void)553 consume_token(void)
554 {
555 int len = tokidx;
556 int i;
557
558 tokidx = 0;
559 if (token == NULL)
560 return (T_NULL);
561
562 /*
563 * this one is special, because we don't want it to alter the
564 * last_kw field.
565 */
566 if (strcmp(token, "...") == 0) {
567 return (T_ELLIPSIS);
568 }
569
570 /* search for reserved words first */
571 for (i = 0; keywords[i].name; i++) {
572 int j;
573 if (strcmp(keywords[i].name, token) != 0) {
574 continue;
575 }
576
577 last_kw = keywords[i].id;
578
579 /* clear the top level category if we're done with it */
580 if (last_kw == T_END) {
581 category = T_END;
582 }
583
584 /* set the top level category if we're changing */
585 for (j = 0; categories[j]; j++) {
586 if (categories[j] != last_kw)
587 continue;
588 category = last_kw;
589 }
590
591 return (keywords[i].id);
592 }
593
594 /* maybe its a numeric constant? */
595 if (isdigit(*token) || (*token == '-' && isdigit(token[1]))) {
596 char *eptr;
597 yylval.num = strtol(token, &eptr, 10);
598 if (*eptr != 0)
599 (void) yyerror(_("malformed number"));
600 return (T_NUMBER);
601 }
602
603 /*
604 * A single lone character is treated as a character literal.
605 * To avoid duplication of effort, we stick in the charmap.
606 */
607 if (len == 1) {
608 yylval.wc = token[0];
609 return (T_CHAR);
610 }
611
612 /* anything else is treated as a symbolic name */
613 yylval.token = strdup(token);
614 token = NULL;
615 toksz = 0;
616 tokidx = 0;
617 return (T_NAME);
618 }
619
620 void
scan_to_eol(void)621 scan_to_eol(void)
622 {
623 int c;
624 while ((c = scanc()) != '\n') {
625 if (c == EOF) {
626 /* end of file without newline! */
627 errf(_("missing newline"));
628 return;
629 }
630 }
631 assert(c == '\n');
632 }
633
634 int
yylex(void)635 yylex(void)
636 {
637 int c;
638
639 while ((c = scanc()) != EOF) {
640
641 /* special handling for quoted string */
642 if (instring) {
643 if (escaped) {
644 escaped = 0;
645
646 /* if newline, just eat and forget it */
647 if (c == '\n')
648 continue;
649
650 if (strchr("xXd01234567", c)) {
651 unscanc(c);
652 unscanc(esc_char);
653 return (get_wide());
654 }
655 yylval.wc = get_escaped(c);
656 return (T_CHAR);
657 }
658 if (c == esc_char) {
659 escaped = 1;
660 continue;
661 }
662 switch (c) {
663 case '<':
664 return (get_symbol());
665 case '>':
666 /* oops! should generate syntax error */
667 return (T_GT);
668 case '"':
669 instring = 0;
670 return (T_QUOTE);
671 default:
672 yylval.wc = c;
673 return (T_CHAR);
674 }
675 }
676
677 /* escaped characters first */
678 if (escaped) {
679 escaped = 0;
680 if (c == '\n') {
681 /* eat the newline */
682 continue;
683 }
684 hadtok = 1;
685 if (tokidx) {
686 /* an escape mid-token is nonsense */
687 return (T_NULL);
688 }
689
690 /* numeric escapes are treated as wide characters */
691 if (strchr("xXd01234567", c)) {
692 unscanc(c);
693 unscanc(esc_char);
694 return (get_wide());
695 }
696
697 add_tok(get_escaped(c));
698 continue;
699 }
700
701 /* if it is the escape charter itself note it */
702 if (c == esc_char) {
703 escaped = 1;
704 continue;
705 }
706
707 /* remove from the comment char to end of line */
708 if (c == com_char) {
709 while (c != '\n') {
710 if ((c = scanc()) == EOF) {
711 /* end of file without newline! */
712 return (EOF);
713 }
714 }
715 assert(c == '\n');
716 if (!hadtok) {
717 /*
718 * If there were no tokens on this line,
719 * then just pretend it didn't exist at all.
720 */
721 continue;
722 }
723 hadtok = 0;
724 return (T_NL);
725 }
726
727 if (strchr(" \t\n;()<>,\"", c) && (tokidx != 0)) {
728 /*
729 * These are all token delimiters. If there
730 * is a token already in progress, we need to
731 * process it.
732 */
733 unscanc(c);
734 return (consume_token());
735 }
736
737 switch (c) {
738 case '\n':
739 if (!hadtok) {
740 /*
741 * If the line was completely devoid of tokens,
742 * then just ignore it.
743 */
744 continue;
745 }
746 /* we're starting a new line, reset the token state */
747 hadtok = 0;
748 return (T_NL);
749 case ',':
750 hadtok = 1;
751 return (T_COMMA);
752 case ';':
753 hadtok = 1;
754 return (T_SEMI);
755 case '(':
756 hadtok = 1;
757 return (T_LPAREN);
758 case ')':
759 hadtok = 1;
760 return (T_RPAREN);
761 case '>':
762 hadtok = 1;
763 return (T_GT);
764 case '<':
765 /* symbol start! */
766 hadtok = 1;
767 return (get_symbol());
768 case ' ':
769 case '\t':
770 /* whitespace, just ignore it */
771 continue;
772 case '"':
773 hadtok = 1;
774 instring = 1;
775 return (T_QUOTE);
776 default:
777 hadtok = 1;
778 add_tok(c);
779 continue;
780 }
781 }
782 return (EOF);
783 }
784
785 int
yyerror(const char * msg)786 yyerror(const char *msg)
787 {
788 (void) fprintf(stderr, _("%s: %d: error: %s\n"),
789 filename, lineno, msg);
790 exit(4);
791 }
792
793 void
errf(const char * fmt,...)794 errf(const char *fmt, ...)
795 {
796 char *msg;
797
798 va_list va;
799 va_start(va, fmt);
800 (void) vasprintf(&msg, fmt, va);
801 va_end(va);
802
803 (void) fprintf(stderr, _("%s: %d: error: %s\n"),
804 filename, lineno, msg);
805 free(msg);
806 exit(4);
807 }
808
809 void
warn(const char * fmt,...)810 warn(const char *fmt, ...)
811 {
812 char *msg;
813
814 va_list va;
815 va_start(va, fmt);
816 (void) vasprintf(&msg, fmt, va);
817 va_end(va);
818
819 (void) fprintf(stderr, _("%s: %d: warning: %s\n"),
820 filename, lineno, msg);
821 free(msg);
822 warnings++;
823 if (!warnok)
824 exit(4);
825 }
826