]> git.saurik.com Git - bison.git/blame - src/reader.c
* Makefile.am (AUTOMAKE_OPTIONS): 1.5.
[bison.git] / src / reader.c
CommitLineData
1ff442ca 1/* Input parser for bison
8c7ebe49 2 Copyright 1984, 1986, 1989, 1992, 1998, 2000
a70083a3 3 Free Software Foundation, Inc.
1ff442ca 4
41aca2e0 5 This file is part of Bison, the GNU Compiler Compiler.
1ff442ca 6
41aca2e0
AD
7 Bison is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2, or (at your option)
10 any later version.
1ff442ca 11
41aca2e0
AD
12 Bison is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
1ff442ca 16
41aca2e0
AD
17 You should have received a copy of the GNU General Public License
18 along with Bison; see the file COPYING. If not, write to
19 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
20 Boston, MA 02111-1307, USA. */
1ff442ca
NF
21
22
1ff442ca 23#include "system.h"
8c7ebe49 24#include "obstack.h"
2a91a95e
AD
25#include "quotearg.h"
26#include "quote.h"
ceed8467 27#include "getargs.h"
1ff442ca 28#include "files.h"
d7913476 29#include "xalloc.h"
1ff442ca
NF
30#include "symtab.h"
31#include "lex.h"
32#include "gram.h"
a0f6b076 33#include "complain.h"
6c89f1c1 34#include "output.h"
b2ca4022 35#include "reader.h"
340ef489 36#include "conflicts.h"
b6610515 37#include "macrotab.h"
1ff442ca 38
1ff442ca 39/* Number of slots allocated (but not necessarily used yet) in `rline' */
4a120d45 40static int rline_allocated;
1ff442ca 41
a70083a3
AD
42typedef struct symbol_list
43{
44 struct symbol_list *next;
45 bucket *sym;
46 bucket *ruleprec;
47}
48symbol_list;
118fb205 49
1ff442ca 50int lineno;
1ff442ca 51char **tags;
d019d655 52short *user_toknums;
4a120d45
JT
53static symbol_list *grammar;
54static int start_flag;
55static bucket *startval;
1ff442ca
NF
56
57/* Nonzero if components of semantic values are used, implying
58 they must be unions. */
59static int value_components_used;
60
d7020c20
AD
61/* Nonzero if %union has been seen. */
62static int typed;
1ff442ca 63
d7020c20
AD
64/* Incremented for each %left, %right or %nonassoc seen */
65static int lastprec;
1ff442ca 66
1ff442ca 67static bucket *errtoken;
5b2e3c89 68static bucket *undeftoken;
0d533154 69\f
a70083a3 70
0d533154
AD
71/*===================\
72| Low level lexing. |
73\===================*/
943819bf
RS
74
75static void
118fb205 76skip_to_char (int target)
943819bf
RS
77{
78 int c;
79 if (target == '\n')
a0f6b076 80 complain (_(" Skipping to next \\n"));
943819bf 81 else
a0f6b076 82 complain (_(" Skipping to next %c"), target);
943819bf
RS
83
84 do
0d533154 85 c = skip_white_space ();
943819bf 86 while (c != target && c != EOF);
a083fbbf 87 if (c != EOF)
0d533154 88 ungetc (c, finput);
943819bf
RS
89}
90
91
0d533154
AD
92/*---------------------------------------------------------.
93| Read a signed integer from STREAM and return its value. |
94`---------------------------------------------------------*/
95
96static inline int
97read_signed_integer (FILE *stream)
98{
a70083a3
AD
99 int c = getc (stream);
100 int sign = 1;
101 int n = 0;
0d533154
AD
102
103 if (c == '-')
104 {
105 c = getc (stream);
106 sign = -1;
107 }
108
109 while (isdigit (c))
110 {
111 n = 10 * n + (c - '0');
112 c = getc (stream);
113 }
114
115 ungetc (c, stream);
116
117 return sign * n;
118}
119\f
79282c5a
AD
120/*--------------------------------------------------------------.
121| Get the data type (alternative in the union) of the value for |
122| symbol N in rule RULE. |
123`--------------------------------------------------------------*/
124
125static char *
126get_type_name (int n, symbol_list * rule)
127{
128 int i;
129 symbol_list *rp;
130
131 if (n < 0)
132 {
133 complain (_("invalid $ value"));
134 return NULL;
135 }
136
137 rp = rule;
138 i = 0;
139
140 while (i < n)
141 {
142 rp = rp->next;
143 if (rp == NULL || rp->sym == NULL)
144 {
145 complain (_("invalid $ value"));
146 return NULL;
147 }
148 i++;
149 }
150
151 return rp->sym->type_name;
152}
153\f
337bab46
AD
154/*------------------------------------------------------------.
155| Dump the string from FIN to OOUT if non null. MATCH is the |
156| delimiter of the string (either ' or "). |
157`------------------------------------------------------------*/
ae3c3164
AD
158
159static inline void
b6610515 160copy_string2 (FILE *fin, struct obstack *oout, int match, int store)
ae3c3164
AD
161{
162 int c;
163
b6610515
RA
164 if (store)
165 obstack_1grow (oout, match);
8c7ebe49 166
4a120d45 167 c = getc (fin);
ae3c3164
AD
168
169 while (c != match)
170 {
171 if (c == EOF)
172 fatal (_("unterminated string at end of file"));
173 if (c == '\n')
174 {
a0f6b076 175 complain (_("unterminated string"));
4a120d45 176 ungetc (c, fin);
ae3c3164
AD
177 c = match; /* invent terminator */
178 continue;
179 }
180
337bab46 181 obstack_1grow (oout, c);
ae3c3164
AD
182
183 if (c == '\\')
184 {
4a120d45 185 c = getc (fin);
ae3c3164
AD
186 if (c == EOF)
187 fatal (_("unterminated string at end of file"));
337bab46 188 obstack_1grow (oout, c);
8c7ebe49 189
ae3c3164
AD
190 if (c == '\n')
191 lineno++;
192 }
193
a70083a3 194 c = getc (fin);
ae3c3164
AD
195 }
196
b6610515
RA
197 if (store)
198 obstack_1grow (oout, c);
199}
200
201/* FIXME. */
202
203static inline void
204copy_string (FILE *fin, struct obstack *oout, int match)
205{
206 copy_string2 (fin, oout, match, 1);
ae3c3164
AD
207}
208
b6610515
RA
209/* FIXME. */
210
211static inline void
212copy_identifier (FILE *fin, struct obstack *oout)
213{
214 int c;
215
216 while (isalnum (c = getc (fin)) || c == '_')
217 obstack_1grow (oout, c);
218
219 ungetc (c, fin);
220}
ae3c3164 221
337bab46
AD
222/*-----------------------------------------------------------------.
223| Dump the wannabee comment from IN to OUT1 and OUT2 (which can be |
224| NULL). In fact we just saw a `/', which might or might not be a |
225| comment. In any case, copy what we saw. |
226| |
227| OUT2 might be NULL. |
228`-----------------------------------------------------------------*/
ae3c3164
AD
229
230static inline void
337bab46 231copy_comment2 (FILE *fin, struct obstack *oout1, struct obstack *oout2)
ae3c3164
AD
232{
233 int cplus_comment;
a70083a3 234 int ended;
550a72a3
AD
235 int c;
236
237 /* We read a `/', output it. */
337bab46 238 obstack_1grow (oout1, '/');
896fe5c1
AD
239 if (oout2)
240 obstack_1grow (oout2, '/');
550a72a3
AD
241
242 switch ((c = getc (fin)))
243 {
244 case '/':
245 cplus_comment = 1;
246 break;
247 case '*':
248 cplus_comment = 0;
249 break;
250 default:
251 ungetc (c, fin);
252 return;
253 }
ae3c3164 254
337bab46 255 obstack_1grow (oout1, c);
896fe5c1
AD
256 if (oout2)
257 obstack_1grow (oout2, c);
550a72a3 258 c = getc (fin);
ae3c3164
AD
259
260 ended = 0;
261 while (!ended)
262 {
263 if (!cplus_comment && c == '*')
264 {
265 while (c == '*')
266 {
337bab46 267 obstack_1grow (oout1, c);
896fe5c1
AD
268 if (oout2)
269 obstack_1grow (oout2, c);
550a72a3 270 c = getc (fin);
ae3c3164
AD
271 }
272
273 if (c == '/')
274 {
337bab46 275 obstack_1grow (oout1, c);
896fe5c1
AD
276 if (oout2)
277 obstack_1grow (oout2, c);
ae3c3164
AD
278 ended = 1;
279 }
280 }
281 else if (c == '\n')
282 {
283 lineno++;
337bab46 284 obstack_1grow (oout1, c);
896fe5c1
AD
285 if (oout2)
286 obstack_1grow (oout2, c);
ae3c3164
AD
287 if (cplus_comment)
288 ended = 1;
289 else
550a72a3 290 c = getc (fin);
ae3c3164
AD
291 }
292 else if (c == EOF)
293 fatal (_("unterminated comment"));
294 else
295 {
337bab46 296 obstack_1grow (oout1, c);
896fe5c1
AD
297 if (oout2)
298 obstack_1grow (oout2, c);
550a72a3 299 c = getc (fin);
ae3c3164
AD
300 }
301 }
302}
303
304
550a72a3
AD
305/*-------------------------------------------------------------------.
306| Dump the comment (actually the current string starting with a `/') |
337bab46 307| from FIN to OOUT. |
550a72a3 308`-------------------------------------------------------------------*/
27821bff
AD
309
310static inline void
337bab46 311copy_comment (FILE *fin, struct obstack *oout)
27821bff 312{
337bab46 313 copy_comment2 (fin, oout, NULL);
27821bff
AD
314}
315
316
a70083a3 317/*-----------------------------------------------------------------.
337bab46 318| FIN is pointing to a location (i.e., a `@'). Output to OOUT a |
a70083a3
AD
319| reference to this location. STACK_OFFSET is the number of values |
320| in the current rule so far, which says where to find `$0' with |
321| respect to the top of the stack. |
322`-----------------------------------------------------------------*/
1ff442ca 323
a70083a3 324static inline void
337bab46 325copy_at (FILE *fin, struct obstack *oout, int stack_offset)
1ff442ca 326{
a70083a3 327 int c;
1ff442ca 328
a70083a3
AD
329 c = getc (fin);
330 if (c == '$')
1ff442ca 331 {
ff4423cc 332 obstack_sgrow (oout, "yyloc");
89cab50d 333 locations_flag = 1;
a70083a3
AD
334 }
335 else if (isdigit (c) || c == '-')
336 {
337 int n;
1ff442ca 338
a70083a3
AD
339 ungetc (c, fin);
340 n = read_signed_integer (fin);
943819bf 341
337bab46 342 obstack_fgrow1 (oout, "yylsp[%d]", n - stack_offset);
89cab50d 343 locations_flag = 1;
1ff442ca 344 }
a70083a3 345 else
ff4a34be
AD
346 {
347 char buf[] = "@c";
348 buf[1] = c;
349 complain (_("%s is invalid"), quote (buf));
350 }
1ff442ca 351}
79282c5a
AD
352
353
354/*-------------------------------------------------------------------.
355| FIN is pointing to a wannabee semantic value (i.e., a `$'). |
356| |
357| Possible inputs: $[<TYPENAME>]($|integer) |
358| |
337bab46 359| Output to OOUT a reference to this semantic value. STACK_OFFSET is |
79282c5a
AD
360| the number of values in the current rule so far, which says where |
361| to find `$0' with respect to the top of the stack. |
362`-------------------------------------------------------------------*/
363
364static inline void
337bab46 365copy_dollar (FILE *fin, struct obstack *oout,
79282c5a
AD
366 symbol_list *rule, int stack_offset)
367{
368 int c = getc (fin);
b0ce6046 369 const char *type_name = NULL;
79282c5a 370
f282676b 371 /* Get the type name if explicit. */
79282c5a
AD
372 if (c == '<')
373 {
f282676b 374 read_type_name (fin);
79282c5a
AD
375 type_name = token_buffer;
376 value_components_used = 1;
79282c5a
AD
377 c = getc (fin);
378 }
379
380 if (c == '$')
381 {
ff4423cc 382 obstack_sgrow (oout, "yyval");
8c7ebe49 383
79282c5a
AD
384 if (!type_name)
385 type_name = get_type_name (0, rule);
386 if (type_name)
337bab46 387 obstack_fgrow1 (oout, ".%s", type_name);
79282c5a
AD
388 if (!type_name && typed)
389 complain (_("$$ of `%s' has no declared type"),
390 rule->sym->tag);
391 }
392 else if (isdigit (c) || c == '-')
393 {
394 int n;
395 ungetc (c, fin);
396 n = read_signed_integer (fin);
397
398 if (!type_name && n > 0)
399 type_name = get_type_name (n, rule);
400
337bab46 401 obstack_fgrow1 (oout, "yyvsp[%d]", n - stack_offset);
8c7ebe49 402
79282c5a 403 if (type_name)
337bab46 404 obstack_fgrow1 (oout, ".%s", type_name);
79282c5a
AD
405 if (!type_name && typed)
406 complain (_("$%d of `%s' has no declared type"),
407 n, rule->sym->tag);
408 }
409 else
410 {
411 char buf[] = "$c";
412 buf[1] = c;
413 complain (_("%s is invalid"), quote (buf));
414 }
415}
a70083a3
AD
416\f
417/*-------------------------------------------------------------------.
418| Copy the contents of a `%{ ... %}' into the definitions file. The |
419| `%{' has already been read. Return after reading the `%}'. |
420`-------------------------------------------------------------------*/
1ff442ca 421
4a120d45 422static void
118fb205 423copy_definition (void)
1ff442ca 424{
a70083a3 425 int c;
ae3c3164 426 /* -1 while reading a character if prev char was %. */
a70083a3 427 int after_percent;
1ff442ca 428
b6610515 429#if 0
89cab50d 430 if (!no_lines_flag)
2a91a95e
AD
431 obstack_fgrow2 (&attrs_obstack, "#line %d %s\n",
432 lineno, quotearg_style (c_quoting_style, infile));
b6610515 433#endif
1ff442ca
NF
434
435 after_percent = 0;
436
ae3c3164 437 c = getc (finput);
1ff442ca
NF
438
439 for (;;)
440 {
441 switch (c)
442 {
443 case '\n':
dd60faec 444 obstack_1grow (&attrs_obstack, c);
1ff442ca
NF
445 lineno++;
446 break;
447
448 case '%':
a70083a3 449 after_percent = -1;
1ff442ca 450 break;
a083fbbf 451
1ff442ca
NF
452 case '\'':
453 case '"':
337bab46 454 copy_string (finput, &attrs_obstack, c);
1ff442ca
NF
455 break;
456
457 case '/':
337bab46 458 copy_comment (finput, &attrs_obstack);
1ff442ca
NF
459 break;
460
461 case EOF:
a70083a3 462 fatal ("%s", _("unterminated `%{' definition"));
1ff442ca
NF
463
464 default:
dd60faec 465 obstack_1grow (&attrs_obstack, c);
1ff442ca
NF
466 }
467
a70083a3 468 c = getc (finput);
1ff442ca
NF
469
470 if (after_percent)
471 {
472 if (c == '}')
473 return;
dd60faec 474 obstack_1grow (&attrs_obstack, '%');
1ff442ca
NF
475 }
476 after_percent = 0;
1ff442ca 477 }
1ff442ca
NF
478}
479
480
d7020c20
AD
481/*-------------------------------------------------------------------.
482| Parse what comes after %token or %nterm. For %token, WHAT_IS is |
483| token_sym and WHAT_IS_NOT is nterm_sym. For %nterm, the arguments |
484| are reversed. |
485`-------------------------------------------------------------------*/
1ff442ca 486
4a120d45 487static void
d7020c20 488parse_token_decl (symbol_class what_is, symbol_class what_is_not)
1ff442ca 489{
f17bcd1f 490 token_t token = 0;
a70083a3 491 char *typename = 0;
1ff442ca 492
1e9798d5
AD
493 /* The symbol being defined. */
494 struct bucket *symbol = NULL;
495
496 /* After `%token' and `%nterm', any number of symbols maybe be
497 defined. */
1ff442ca
NF
498 for (;;)
499 {
e6011337
JT
500 int tmp_char = ungetc (skip_white_space (), finput);
501
1e9798d5
AD
502 /* `%' (for instance from `%token', or from `%%' etc.) is the
503 only valid means to end this declaration. */
e6011337 504 if (tmp_char == '%')
1ff442ca 505 return;
e6011337 506 if (tmp_char == EOF)
a0f6b076 507 fatal (_("Premature EOF after %s"), token_buffer);
e6011337 508
a70083a3 509 token = lex ();
511e79b3 510 if (token == tok_comma)
943819bf
RS
511 {
512 symbol = NULL;
513 continue;
514 }
511e79b3 515 if (token == tok_typename)
1ff442ca 516 {
95e36146 517 typename = xstrdup (token_buffer);
1ff442ca 518 value_components_used = 1;
943819bf
RS
519 symbol = NULL;
520 }
511e79b3 521 else if (token == tok_identifier && *symval->tag == '\"' && symbol)
943819bf 522 {
8e03724b
AD
523 if (symval->alias)
524 warn (_("symbol `%s' used more than once as a literal string"),
525 symval->tag);
526 else if (symbol->alias)
527 warn (_("symbol `%s' given more than one literal string"),
528 symbol->tag);
529 else
530 {
531 symval->class = token_sym;
532 symval->type_name = typename;
533 symval->user_token_number = symbol->user_token_number;
534 symbol->user_token_number = SALIAS;
535 symval->alias = symbol;
536 symbol->alias = symval;
537 /* symbol and symval combined are only one symbol */
538 nsyms--;
539 }
943819bf 540 translations = 1;
8e03724b 541 symbol = NULL;
1ff442ca 542 }
511e79b3 543 else if (token == tok_identifier)
1ff442ca
NF
544 {
545 int oldclass = symval->class;
943819bf 546 symbol = symval;
1ff442ca 547
943819bf 548 if (symbol->class == what_is_not)
a0f6b076 549 complain (_("symbol %s redefined"), symbol->tag);
943819bf 550 symbol->class = what_is;
d7020c20 551 if (what_is == nterm_sym && oldclass != nterm_sym)
943819bf 552 symbol->value = nvars++;
1ff442ca
NF
553
554 if (typename)
555 {
943819bf
RS
556 if (symbol->type_name == NULL)
557 symbol->type_name = typename;
a70083a3 558 else if (strcmp (typename, symbol->type_name) != 0)
a0f6b076 559 complain (_("type redeclaration for %s"), symbol->tag);
1ff442ca
NF
560 }
561 }
511e79b3 562 else if (symbol && token == tok_number)
a70083a3 563 {
943819bf 564 symbol->user_token_number = numval;
1ff442ca 565 translations = 1;
a70083a3 566 }
1ff442ca 567 else
943819bf 568 {
a0f6b076 569 complain (_("`%s' is invalid in %s"),
d7020c20 570 token_buffer, (what_is == token_sym) ? "%token" : "%nterm");
a70083a3 571 skip_to_char ('%');
943819bf 572 }
1ff442ca
NF
573 }
574
575}
576
1ff442ca 577
d7020c20
AD
578/*------------------------------.
579| Parse what comes after %start |
580`------------------------------*/
1ff442ca 581
4a120d45 582static void
118fb205 583parse_start_decl (void)
1ff442ca
NF
584{
585 if (start_flag)
27821bff 586 complain (_("multiple %s declarations"), "%start");
511e79b3 587 if (lex () != tok_identifier)
27821bff 588 complain (_("invalid %s declaration"), "%start");
943819bf
RS
589 else
590 {
591 start_flag = 1;
592 startval = symval;
593 }
1ff442ca
NF
594}
595
a70083a3
AD
596/*-----------------------------------------------------------.
597| read in a %type declaration and record its information for |
598| get_type_name to access |
599`-----------------------------------------------------------*/
600
601static void
602parse_type_decl (void)
603{
a70083a3
AD
604 char *name;
605
511e79b3 606 if (lex () != tok_typename)
a70083a3
AD
607 {
608 complain ("%s", _("%type declaration has no <typename>"));
609 skip_to_char ('%');
610 return;
611 }
612
95e36146 613 name = xstrdup (token_buffer);
a70083a3
AD
614
615 for (;;)
616 {
f17bcd1f 617 token_t t;
a70083a3
AD
618 int tmp_char = ungetc (skip_white_space (), finput);
619
620 if (tmp_char == '%')
621 return;
622 if (tmp_char == EOF)
623 fatal (_("Premature EOF after %s"), token_buffer);
624
625 t = lex ();
626
627 switch (t)
1ff442ca
NF
628 {
629
511e79b3
AD
630 case tok_comma:
631 case tok_semicolon:
1ff442ca
NF
632 break;
633
511e79b3 634 case tok_identifier:
1ff442ca
NF
635 if (symval->type_name == NULL)
636 symval->type_name = name;
a70083a3 637 else if (strcmp (name, symval->type_name) != 0)
a0f6b076 638 complain (_("type redeclaration for %s"), symval->tag);
1ff442ca
NF
639
640 break;
641
642 default:
a0f6b076
AD
643 complain (_("invalid %%type declaration due to item: %s"),
644 token_buffer);
a70083a3 645 skip_to_char ('%');
1ff442ca
NF
646 }
647 }
648}
649
650
651
d7020c20
AD
652/*----------------------------------------------------------------.
653| Read in a %left, %right or %nonassoc declaration and record its |
654| information. |
655`----------------------------------------------------------------*/
1ff442ca 656
4a120d45 657static void
d7020c20 658parse_assoc_decl (associativity assoc)
1ff442ca 659{
a70083a3
AD
660 char *name = NULL;
661 int prev = 0;
1ff442ca 662
a70083a3 663 lastprec++; /* Assign a new precedence level, never 0. */
1ff442ca 664
1ff442ca
NF
665 for (;;)
666 {
f17bcd1f 667 token_t t;
e6011337 668 int tmp_char = ungetc (skip_white_space (), finput);
1ff442ca 669
e6011337 670 if (tmp_char == '%')
1ff442ca 671 return;
e6011337 672 if (tmp_char == EOF)
a0f6b076 673 fatal (_("Premature EOF after %s"), token_buffer);
1ff442ca 674
a70083a3 675 t = lex ();
1ff442ca
NF
676
677 switch (t)
678 {
511e79b3 679 case tok_typename:
95e36146 680 name = xstrdup (token_buffer);
1ff442ca
NF
681 break;
682
511e79b3 683 case tok_comma:
1ff442ca
NF
684 break;
685
511e79b3 686 case tok_identifier:
1ff442ca 687 if (symval->prec != 0)
a0f6b076 688 complain (_("redefining precedence of %s"), symval->tag);
1ff442ca
NF
689 symval->prec = lastprec;
690 symval->assoc = assoc;
d7020c20 691 if (symval->class == nterm_sym)
a0f6b076 692 complain (_("symbol %s redefined"), symval->tag);
d7020c20 693 symval->class = token_sym;
1ff442ca 694 if (name)
a70083a3 695 { /* record the type, if one is specified */
1ff442ca
NF
696 if (symval->type_name == NULL)
697 symval->type_name = name;
a70083a3 698 else if (strcmp (name, symval->type_name) != 0)
a0f6b076 699 complain (_("type redeclaration for %s"), symval->tag);
1ff442ca
NF
700 }
701 break;
702
511e79b3
AD
703 case tok_number:
704 if (prev == tok_identifier)
a70083a3 705 {
1ff442ca
NF
706 symval->user_token_number = numval;
707 translations = 1;
a70083a3
AD
708 }
709 else
710 {
711 complain (_
712 ("invalid text (%s) - number should be after identifier"),
713token_buffer);
714 skip_to_char ('%');
715 }
1ff442ca
NF
716 break;
717
511e79b3 718 case tok_semicolon:
1ff442ca
NF
719 return;
720
721 default:
a0f6b076 722 complain (_("unexpected item: %s"), token_buffer);
a70083a3 723 skip_to_char ('%');
1ff442ca
NF
724 }
725
726 prev = t;
727
728 }
729}
730
731
732
dd60faec
AD
733/*--------------------------------------------------------------.
734| Copy the union declaration into ATTRS_OBSTACK (and fdefines), |
735| where it is made into the definition of YYSTYPE, the type of |
736| elements of the parser value stack. |
737`--------------------------------------------------------------*/
1ff442ca 738
4a120d45 739static void
118fb205 740parse_union_decl (void)
1ff442ca 741{
a70083a3
AD
742 int c;
743 int count = 0;
1ff442ca
NF
744
745 if (typed)
27821bff 746 complain (_("multiple %s declarations"), "%union");
1ff442ca
NF
747
748 typed = 1;
749
89cab50d 750 if (!no_lines_flag)
2a91a95e
AD
751 obstack_fgrow2 (&attrs_obstack, "\n#line %d %s\n",
752 lineno, quotearg_style (c_quoting_style, infile));
1ff442ca 753 else
dd60faec 754 obstack_1grow (&attrs_obstack, '\n');
1ff442ca 755
ff4423cc 756 obstack_sgrow (&attrs_obstack, "typedef union");
896fe5c1 757 if (defines_flag)
ff4423cc 758 obstack_sgrow (&defines_obstack, "typedef union");
1ff442ca 759
27821bff 760 c = getc (finput);
1ff442ca
NF
761
762 while (c != EOF)
763 {
dd60faec 764 obstack_1grow (&attrs_obstack, c);
896fe5c1 765 if (defines_flag)
d7045ec6 766 obstack_1grow (&defines_obstack, c);
1ff442ca
NF
767
768 switch (c)
769 {
770 case '\n':
771 lineno++;
772 break;
773
774 case '/':
337bab46 775 copy_comment2 (finput, &defines_obstack, &attrs_obstack);
1ff442ca
NF
776 break;
777
1ff442ca
NF
778 case '{':
779 count++;
780 break;
781
782 case '}':
783 if (count == 0)
27821bff 784 complain (_("unmatched %s"), "`}'");
1ff442ca 785 count--;
943819bf 786 if (count <= 0)
1ff442ca 787 {
ff4423cc 788 obstack_sgrow (&attrs_obstack, " YYSTYPE;\n");
896fe5c1 789 if (defines_flag)
ff4423cc 790 obstack_sgrow (&defines_obstack, " YYSTYPE;\n");
1ff442ca 791 /* JF don't choke on trailing semi */
27821bff
AD
792 c = skip_white_space ();
793 if (c != ';')
a70083a3 794 ungetc (c, finput);
1ff442ca
NF
795 return;
796 }
797 }
798
27821bff 799 c = getc (finput);
1ff442ca
NF
800 }
801}
802
d7020c20
AD
803
804/*-------------------------------------------------------.
805| Parse the declaration %expect N which says to expect N |
806| shift-reduce conflicts. |
807`-------------------------------------------------------*/
1ff442ca 808
4a120d45 809static void
118fb205 810parse_expect_decl (void)
1ff442ca 811{
131e2fef 812 int c = skip_white_space ();
1ff442ca
NF
813 ungetc (c, finput);
814
131e2fef 815 if (!isdigit (c))
79282c5a 816 complain (_("argument of %%expect is not an integer"));
131e2fef
AD
817 else
818 expected_conflicts = read_signed_integer (finput);
1ff442ca
NF
819}
820
a70083a3
AD
821
822/*-------------------------------------------------------------------.
823| Parse what comes after %thong. the full syntax is |
824| |
825| %thong <type> token number literal |
826| |
827| the <type> or number may be omitted. The number specifies the |
828| user_token_number. |
829| |
830| Two symbols are entered in the table, one for the token symbol and |
831| one for the literal. Both are given the <type>, if any, from the |
832| declaration. The ->user_token_number of the first is SALIAS and |
833| the ->user_token_number of the second is set to the number, if |
834| any, from the declaration. The two symbols are linked via |
835| pointers in their ->alias fields. |
836| |
837| During OUTPUT_DEFINES_TABLE, the symbol is reported thereafter, |
838| only the literal string is retained it is the literal string that |
839| is output to yytname |
840`-------------------------------------------------------------------*/
841
842static void
843parse_thong_decl (void)
7b306f52 844{
f17bcd1f 845 token_t token;
a70083a3
AD
846 struct bucket *symbol;
847 char *typename = 0;
95e36146 848 int usrtoknum;
7b306f52 849
a70083a3
AD
850 translations = 1;
851 token = lex (); /* fetch typename or first token */
511e79b3 852 if (token == tok_typename)
7b306f52 853 {
95e36146 854 typename = xstrdup (token_buffer);
a70083a3
AD
855 value_components_used = 1;
856 token = lex (); /* fetch first token */
7b306f52 857 }
7b306f52 858
a70083a3 859 /* process first token */
7b306f52 860
511e79b3 861 if (token != tok_identifier)
a70083a3
AD
862 {
863 complain (_("unrecognized item %s, expected an identifier"),
864 token_buffer);
865 skip_to_char ('%');
866 return;
7b306f52 867 }
d7020c20 868 symval->class = token_sym;
a70083a3
AD
869 symval->type_name = typename;
870 symval->user_token_number = SALIAS;
871 symbol = symval;
7b306f52 872
a70083a3 873 token = lex (); /* get number or literal string */
1ff442ca 874
511e79b3 875 if (token == tok_number)
943819bf 876 {
a70083a3
AD
877 usrtoknum = numval;
878 token = lex (); /* okay, did number, now get literal */
943819bf 879 }
a70083a3
AD
880 else
881 usrtoknum = 0;
1ff442ca 882
a70083a3 883 /* process literal string token */
1ff442ca 884
511e79b3 885 if (token != tok_identifier || *symval->tag != '\"')
1ff442ca 886 {
a70083a3
AD
887 complain (_("expected string constant instead of %s"), token_buffer);
888 skip_to_char ('%');
889 return;
1ff442ca 890 }
d7020c20 891 symval->class = token_sym;
a70083a3
AD
892 symval->type_name = typename;
893 symval->user_token_number = usrtoknum;
1ff442ca 894
a70083a3
AD
895 symval->alias = symbol;
896 symbol->alias = symval;
1ff442ca 897
79282c5a
AD
898 /* symbol and symval combined are only one symbol. */
899 nsyms--;
a70083a3 900}
3cef001a 901
b6610515
RA
902/* FIXME. */
903
904static void
905parse_macro_decl (void)
906{
907 int ch = ungetc (skip_white_space (), finput);
908 char* macro_key;
909 char* macro_value;
910 struct obstack macro_obstack;
911
912 obstack_init (&macro_obstack);
913
914 /* Read key. */
915 if (!isalpha (ch) && ch != '_')
916 {
917 complain (_("invalid %s declaration"), "%define");
918 skip_to_char ('%');
919 return;
920 }
921 copy_identifier (finput, &macro_obstack);
922 macro_key = obstack_finish (&macro_obstack);
923
924 /* Read value. */
925 ch = skip_white_space ();
926 if (ch != '"')
927 {
928 ungetc (ch, finput);
929 if (ch != EOF)
930 {
931 complain (_("invalid %s declaration"), "%define");
932 skip_to_char ('%');
933 return;
934 }
935 else
936 fatal (_("Premature EOF after %s"), "\"");
937 }
938 copy_string (finput, &macro_obstack, '"');
939 macro_value = obstack_finish (&macro_obstack);
940
941 obstack_free (&macro_obstack, 0);
942
943 /* Store the (key, value) pair in the environment. */
944 macro_insert (macro_key, macro_value);
945}
946
7333d403
AD
947/*------------------------------------------.
948| Parse what comes after %header_extension. |
949`------------------------------------------*/
950
951static void
952parse_header_extension_decl (void)
953{
954 char buff[32];
3e3da797 955
7333d403
AD
956 if (header_extension)
957 complain (_("multiple %%header_extension declarations"));
958 fscanf (finput, "%s", buff);
3e3da797 959 header_extension = xstrdup (buff);
7333d403
AD
960}
961
962/*------------------------------------------.
963| Parse what comes after %source_extension. |
964`------------------------------------------*/
965
966static void
967parse_source_extension_decl (void)
968{
969 char buff[32];
3e3da797 970
7333d403
AD
971 if (src_extension)
972 complain (_("multiple %%source_extension declarations"));
973 fscanf (finput, "%s", buff);
3e3da797 974 src_extension = xstrdup (buff);
7333d403 975}
d7020c20 976
a70083a3
AD
977/*----------------------------------------------------------------.
978| Read from finput until `%%' is seen. Discard the `%%'. Handle |
979| any `%' declarations, and copy the contents of any `%{ ... %}' |
dd60faec 980| groups to ATTRS_OBSTACK. |
a70083a3 981`----------------------------------------------------------------*/
1ff442ca 982
4a120d45 983static void
a70083a3 984read_declarations (void)
1ff442ca 985{
a70083a3
AD
986 int c;
987 int tok;
1ff442ca 988
a70083a3 989 for (;;)
1ff442ca 990 {
a70083a3 991 c = skip_white_space ();
1ff442ca 992
a70083a3
AD
993 if (c == '%')
994 {
995 tok = parse_percent_token ();
1ff442ca 996
a70083a3 997 switch (tok)
943819bf 998 {
511e79b3 999 case tok_two_percents:
a70083a3 1000 return;
1ff442ca 1001
511e79b3 1002 case tok_percent_left_curly:
a70083a3
AD
1003 copy_definition ();
1004 break;
1ff442ca 1005
511e79b3 1006 case tok_token:
d7020c20 1007 parse_token_decl (token_sym, nterm_sym);
a70083a3 1008 break;
1ff442ca 1009
511e79b3 1010 case tok_nterm:
d7020c20 1011 parse_token_decl (nterm_sym, token_sym);
a70083a3 1012 break;
1ff442ca 1013
511e79b3 1014 case tok_type:
a70083a3
AD
1015 parse_type_decl ();
1016 break;
1ff442ca 1017
511e79b3 1018 case tok_start:
a70083a3
AD
1019 parse_start_decl ();
1020 break;
118fb205 1021
511e79b3 1022 case tok_union:
a70083a3
AD
1023 parse_union_decl ();
1024 break;
1ff442ca 1025
511e79b3 1026 case tok_expect:
a70083a3
AD
1027 parse_expect_decl ();
1028 break;
6deb4447 1029
511e79b3 1030 case tok_thong:
a70083a3
AD
1031 parse_thong_decl ();
1032 break;
d7020c20 1033
511e79b3 1034 case tok_left:
d7020c20 1035 parse_assoc_decl (left_assoc);
a70083a3 1036 break;
1ff442ca 1037
511e79b3 1038 case tok_right:
d7020c20 1039 parse_assoc_decl (right_assoc);
a70083a3 1040 break;
1ff442ca 1041
511e79b3 1042 case tok_nonassoc:
d7020c20 1043 parse_assoc_decl (non_assoc);
a70083a3 1044 break;
1ff442ca 1045
7333d403 1046 case tok_hdrext:
09a6de7e 1047 parse_header_extension_decl ();
7333d403
AD
1048 break;
1049
1050 case tok_srcext:
09a6de7e 1051 parse_source_extension_decl ();
7333d403
AD
1052 break;
1053
b6610515
RA
1054 case tok_define:
1055 parse_macro_decl ();
1056 break;
1057
511e79b3 1058 case tok_noop:
a70083a3 1059 break;
1ff442ca 1060
a70083a3
AD
1061 default:
1062 complain (_("unrecognized: %s"), token_buffer);
1063 skip_to_char ('%');
1064 }
1065 }
1066 else if (c == EOF)
1067 fatal (_("no input grammar"));
1068 else
1069 {
ff4a34be
AD
1070 char buf[] = "c";
1071 buf[0] = c;
1072 complain (_("unknown character: %s"), quote (buf));
a70083a3 1073 skip_to_char ('%');
1ff442ca 1074 }
1ff442ca 1075 }
1ff442ca 1076}
a70083a3
AD
1077\f
1078/*-------------------------------------------------------------------.
1079| Assuming that a `{' has just been seen, copy everything up to the |
1080| matching `}' into the actions file. STACK_OFFSET is the number of |
1081| values in the current rule so far, which says where to find `$0' |
1082| with respect to the top of the stack. |
1083`-------------------------------------------------------------------*/
1ff442ca 1084
4a120d45 1085static void
79282c5a 1086copy_action (symbol_list *rule, int stack_offset)
1ff442ca 1087{
a70083a3 1088 int c;
a70083a3 1089 int count;
8c7ebe49 1090 char buf[4096];
1ff442ca
NF
1091
1092 /* offset is always 0 if parser has already popped the stack pointer */
41aca2e0
AD
1093 if (semantic_parser)
1094 stack_offset = 0;
1ff442ca 1095
8c7ebe49
AD
1096 sprintf (buf, "\ncase %d:\n", nrules);
1097 obstack_grow (&action_obstack, buf, strlen (buf));
1098
89cab50d 1099 if (!no_lines_flag)
8c7ebe49 1100 {
2a91a95e
AD
1101 sprintf (buf, "#line %d %s\n",
1102 lineno, quotearg_style (c_quoting_style, infile));
8c7ebe49
AD
1103 obstack_grow (&action_obstack, buf, strlen (buf));
1104 }
1105 obstack_1grow (&action_obstack, '{');
1ff442ca
NF
1106
1107 count = 1;
a70083a3 1108 c = getc (finput);
1ff442ca
NF
1109
1110 while (count > 0)
1111 {
1112 while (c != '}')
a70083a3
AD
1113 {
1114 switch (c)
1ff442ca
NF
1115 {
1116 case '\n':
8c7ebe49 1117 obstack_1grow (&action_obstack, c);
1ff442ca
NF
1118 lineno++;
1119 break;
1120
1121 case '{':
8c7ebe49 1122 obstack_1grow (&action_obstack, c);
1ff442ca
NF
1123 count++;
1124 break;
1125
1126 case '\'':
1127 case '"':
337bab46 1128 copy_string (finput, &action_obstack, c);
1ff442ca
NF
1129 break;
1130
1131 case '/':
337bab46 1132 copy_comment (finput, &action_obstack);
1ff442ca
NF
1133 break;
1134
1135 case '$':
337bab46 1136 copy_dollar (finput, &action_obstack,
8c7ebe49 1137 rule, stack_offset);
1ff442ca
NF
1138 break;
1139
1140 case '@':
337bab46 1141 copy_at (finput, &action_obstack,
8c7ebe49 1142 stack_offset);
6666f98f 1143 break;
1ff442ca
NF
1144
1145 case EOF:
27821bff 1146 fatal (_("unmatched %s"), "`{'");
1ff442ca
NF
1147
1148 default:
8c7ebe49 1149 obstack_1grow (&action_obstack, c);
a70083a3
AD
1150 }
1151
1152 c = getc (finput);
1153 }
1154
1155 /* above loop exits when c is '}' */
1156
1157 if (--count)
1158 {
8c7ebe49 1159 obstack_1grow (&action_obstack, c);
a70083a3
AD
1160 c = getc (finput);
1161 }
1162 }
1163
ff4423cc 1164 obstack_sgrow (&action_obstack, ";\n break;}");
a70083a3
AD
1165}
1166\f
1167/*-------------------------------------------------------------------.
1168| After `%guard' is seen in the input file, copy the actual guard |
1169| into the guards file. If the guard is followed by an action, copy |
1170| that into the actions file. STACK_OFFSET is the number of values |
1171| in the current rule so far, which says where to find `$0' with |
1172| respect to the top of the stack, for the simple parser in which |
1173| the stack is not popped until after the guard is run. |
1174`-------------------------------------------------------------------*/
1175
1176static void
79282c5a 1177copy_guard (symbol_list *rule, int stack_offset)
a70083a3
AD
1178{
1179 int c;
a70083a3 1180 int count;
a70083a3
AD
1181 int brace_flag = 0;
1182
1183 /* offset is always 0 if parser has already popped the stack pointer */
1184 if (semantic_parser)
1185 stack_offset = 0;
1186
ea5607fd 1187 obstack_fgrow1 (&guard_obstack, "\ncase %d:\n", nrules);
89cab50d 1188 if (!no_lines_flag)
ea5607fd
AD
1189 obstack_fgrow2 (&guard_obstack, "#line %d %s\n",
1190 lineno, quotearg_style (c_quoting_style, infile));
1191 obstack_1grow (&guard_obstack, '{');
a70083a3
AD
1192
1193 count = 0;
1194 c = getc (finput);
1195
1196 while (brace_flag ? (count > 0) : (c != ';'))
1197 {
1198 switch (c)
1199 {
1200 case '\n':
ea5607fd 1201 obstack_1grow (&guard_obstack, c);
a70083a3
AD
1202 lineno++;
1203 break;
1204
1205 case '{':
ea5607fd 1206 obstack_1grow (&guard_obstack, c);
a70083a3
AD
1207 brace_flag = 1;
1208 count++;
1209 break;
1210
1211 case '}':
ea5607fd 1212 obstack_1grow (&guard_obstack, c);
a70083a3
AD
1213 if (count > 0)
1214 count--;
1215 else
1216 {
1217 complain (_("unmatched %s"), "`}'");
1218 c = getc (finput); /* skip it */
1219 }
1220 break;
1221
1222 case '\'':
1223 case '"':
337bab46 1224 copy_string (finput, &guard_obstack, c);
a70083a3
AD
1225 break;
1226
1227 case '/':
337bab46 1228 copy_comment (finput, &guard_obstack);
a70083a3
AD
1229 break;
1230
1231 case '$':
337bab46 1232 copy_dollar (finput, &guard_obstack, rule, stack_offset);
a70083a3 1233 break;
1ff442ca 1234
a70083a3 1235 case '@':
337bab46 1236 copy_at (finput, &guard_obstack, stack_offset);
a70083a3 1237 break;
1ff442ca 1238
a70083a3
AD
1239 case EOF:
1240 fatal ("%s", _("unterminated %guard clause"));
1ff442ca 1241
a70083a3 1242 default:
ea5607fd 1243 obstack_1grow (&guard_obstack, c);
1ff442ca 1244 }
a70083a3
AD
1245
1246 if (c != '}' || count != 0)
1247 c = getc (finput);
1ff442ca
NF
1248 }
1249
a70083a3
AD
1250 c = skip_white_space ();
1251
ff4423cc 1252 obstack_sgrow (&guard_obstack, ";\n break;}");
a70083a3
AD
1253 if (c == '{')
1254 copy_action (rule, stack_offset);
1255 else if (c == '=')
1256 {
1257 c = getc (finput); /* why not skip_white_space -wjh */
1258 if (c == '{')
1259 copy_action (rule, stack_offset);
1260 }
1261 else
1262 ungetc (c, finput);
1ff442ca 1263}
a70083a3
AD
1264\f
1265
1266static void
1267record_rule_line (void)
1268{
1269 /* Record each rule's source line number in rline table. */
1ff442ca 1270
a70083a3
AD
1271 if (nrules >= rline_allocated)
1272 {
1273 rline_allocated = nrules * 2;
d7913476 1274 rline = XREALLOC (rline, short, rline_allocated);
a70083a3
AD
1275 }
1276 rline[nrules] = lineno;
1277}
1ff442ca
NF
1278
1279
a70083a3
AD
1280/*-------------------------------------------------------------------.
1281| Generate a dummy symbol, a nonterminal, whose name cannot conflict |
1282| with the user's names. |
1283`-------------------------------------------------------------------*/
1ff442ca 1284
4a120d45 1285static bucket *
118fb205 1286gensym (void)
1ff442ca 1287{
274d42ce
AD
1288 /* Incremented for each generated symbol */
1289 static int gensym_count = 0;
1290 static char buf[256];
1291
a70083a3 1292 bucket *sym;
1ff442ca 1293
274d42ce
AD
1294 sprintf (buf, "@%d", ++gensym_count);
1295 token_buffer = buf;
a70083a3 1296 sym = getsym (token_buffer);
d7020c20 1297 sym->class = nterm_sym;
1ff442ca 1298 sym->value = nvars++;
36281465 1299 return sym;
1ff442ca
NF
1300}
1301
a70083a3
AD
1302#if 0
1303/*------------------------------------------------------------------.
1304| read in a %type declaration and record its information for |
1305| get_type_name to access. This is unused. It is only called from |
1306| the #if 0 part of readgram |
1307`------------------------------------------------------------------*/
1308
1309static int
1310get_type (void)
1311{
1312 int k;
f17bcd1f 1313 token_t token;
a70083a3
AD
1314 char *name;
1315
f17bcd1f 1316 token = lex ();
a70083a3 1317
f17bcd1f 1318 if (token != tok_typename)
a70083a3
AD
1319 {
1320 complain (_("invalid %s declaration"), "%type");
1321 return t;
1322 }
1323
95e36146 1324 name = xstrdup (token_buffer);
a70083a3
AD
1325
1326 for (;;)
1327 {
f17bcd1f 1328 token = lex ();
a70083a3 1329
f17bcd1f 1330 switch (token)
a70083a3 1331 {
511e79b3 1332 case tok_semicolon:
a70083a3
AD
1333 return lex ();
1334
511e79b3 1335 case tok_comma:
a70083a3
AD
1336 break;
1337
511e79b3 1338 case tok_identifier:
a70083a3
AD
1339 if (symval->type_name == NULL)
1340 symval->type_name = name;
1341 else if (strcmp (name, symval->type_name) != 0)
1342 complain (_("type redeclaration for %s"), symval->tag);
1343
1344 break;
1345
1346 default:
f17bcd1f 1347 return token;
a70083a3
AD
1348 }
1349 }
1350}
1ff442ca 1351
a70083a3
AD
1352#endif
1353\f
1354/*------------------------------------------------------------------.
1355| Parse the input grammar into a one symbol_list structure. Each |
1356| rule is represented by a sequence of symbols: the left hand side |
1357| followed by the contents of the right hand side, followed by a |
1358| null pointer instead of a symbol to terminate the rule. The next |
1359| symbol is the lhs of the following rule. |
1360| |
1361| All guards and actions are copied out to the appropriate files, |
1362| labelled by the rule number they apply to. |
1363`------------------------------------------------------------------*/
1ff442ca 1364
4a120d45 1365static void
118fb205 1366readgram (void)
1ff442ca 1367{
f17bcd1f 1368 token_t t;
a70083a3
AD
1369 bucket *lhs = NULL;
1370 symbol_list *p;
1371 symbol_list *p1;
1372 bucket *bp;
1ff442ca 1373
ff4a34be
AD
1374 /* Points to first symbol_list of current rule. its symbol is the
1375 lhs of the rule. */
1376 symbol_list *crule;
1377 /* Points to the symbol_list preceding crule. */
1378 symbol_list *crule1;
1ff442ca
NF
1379
1380 p1 = NULL;
1381
a70083a3 1382 t = lex ();
1ff442ca 1383
511e79b3 1384 while (t != tok_two_percents && t != tok_eof)
1ff442ca 1385 {
511e79b3 1386 if (t == tok_identifier || t == tok_bar)
1ff442ca 1387 {
89cab50d 1388 int action_flag = 0;
ff4a34be
AD
1389 /* Number of symbols in rhs of this rule so far */
1390 int rulelength = 0;
1ff442ca
NF
1391 int xactions = 0; /* JF for error checking */
1392 bucket *first_rhs = 0;
1393
511e79b3 1394 if (t == tok_identifier)
1ff442ca
NF
1395 {
1396 lhs = symval;
943819bf
RS
1397
1398 if (!start_flag)
1399 {
1400 startval = lhs;
1401 start_flag = 1;
1402 }
a083fbbf 1403
a70083a3 1404 t = lex ();
511e79b3 1405 if (t != tok_colon)
943819bf 1406 {
a0f6b076 1407 complain (_("ill-formed rule: initial symbol not followed by colon"));
a70083a3 1408 unlex (t);
943819bf 1409 }
1ff442ca
NF
1410 }
1411
511e79b3 1412 if (nrules == 0 && t == tok_bar)
1ff442ca 1413 {
a0f6b076 1414 complain (_("grammar starts with vertical bar"));
943819bf 1415 lhs = symval; /* BOGUS: use a random symval */
1ff442ca 1416 }
1ff442ca
NF
1417 /* start a new rule and record its lhs. */
1418
1419 nrules++;
1420 nitems++;
1421
1422 record_rule_line ();
1423
d7913476 1424 p = XCALLOC (symbol_list, 1);
1ff442ca
NF
1425 p->sym = lhs;
1426
1427 crule1 = p1;
1428 if (p1)
1429 p1->next = p;
1430 else
1431 grammar = p;
1432
1433 p1 = p;
1434 crule = p;
1435
1436 /* mark the rule's lhs as a nonterminal if not already so. */
1437
d7020c20 1438 if (lhs->class == unknown_sym)
1ff442ca 1439 {
d7020c20 1440 lhs->class = nterm_sym;
1ff442ca
NF
1441 lhs->value = nvars;
1442 nvars++;
1443 }
d7020c20 1444 else if (lhs->class == token_sym)
a0f6b076 1445 complain (_("rule given for %s, which is a token"), lhs->tag);
1ff442ca
NF
1446
1447 /* read the rhs of the rule. */
1448
1449 for (;;)
1450 {
a70083a3 1451 t = lex ();
511e79b3 1452 if (t == tok_prec)
943819bf 1453 {
a70083a3 1454 t = lex ();
943819bf 1455 crule->ruleprec = symval;
a70083a3 1456 t = lex ();
943819bf 1457 }
1ff442ca 1458
511e79b3 1459 if (!(t == tok_identifier || t == tok_left_curly))
a70083a3 1460 break;
1ff442ca
NF
1461
1462 /* If next token is an identifier, see if a colon follows it.
a70083a3 1463 If one does, exit this rule now. */
511e79b3 1464 if (t == tok_identifier)
1ff442ca 1465 {
a70083a3 1466 bucket *ssave;
f17bcd1f 1467 token_t t1;
1ff442ca
NF
1468
1469 ssave = symval;
a70083a3
AD
1470 t1 = lex ();
1471 unlex (t1);
1ff442ca 1472 symval = ssave;
511e79b3 1473 if (t1 == tok_colon)
a70083a3 1474 break;
1ff442ca 1475
a70083a3 1476 if (!first_rhs) /* JF */
1ff442ca
NF
1477 first_rhs = symval;
1478 /* Not followed by colon =>
1479 process as part of this rule's rhs. */
1480 }
1481
1482 /* If we just passed an action, that action was in the middle
a70083a3
AD
1483 of a rule, so make a dummy rule to reduce it to a
1484 non-terminal. */
89cab50d 1485 if (action_flag)
1ff442ca 1486 {
a70083a3 1487 bucket *sdummy;
1ff442ca 1488
f282676b
AD
1489 /* Since the action was written out with this rule's
1490 number, we must give the new rule this number by
1491 inserting the new rule before it. */
1ff442ca
NF
1492
1493 /* Make a dummy nonterminal, a gensym. */
a70083a3 1494 sdummy = gensym ();
1ff442ca
NF
1495
1496 /* Make a new rule, whose body is empty,
1497 before the current one, so that the action
1498 just read can belong to it. */
1499 nrules++;
1500 nitems++;
1501 record_rule_line ();
d7913476 1502 p = XCALLOC (symbol_list, 1);
1ff442ca
NF
1503 if (crule1)
1504 crule1->next = p;
a70083a3
AD
1505 else
1506 grammar = p;
1ff442ca 1507 p->sym = sdummy;
d7913476 1508 crule1 = XCALLOC (symbol_list, 1);
1ff442ca
NF
1509 p->next = crule1;
1510 crule1->next = crule;
1511
f282676b
AD
1512 /* Insert the dummy generated by that rule into this
1513 rule. */
1ff442ca 1514 nitems++;
d7913476 1515 p = XCALLOC (symbol_list, 1);
1ff442ca
NF
1516 p->sym = sdummy;
1517 p1->next = p;
1518 p1 = p;
1519
89cab50d 1520 action_flag = 0;
1ff442ca
NF
1521 }
1522
511e79b3 1523 if (t == tok_identifier)
1ff442ca
NF
1524 {
1525 nitems++;
d7913476 1526 p = XCALLOC (symbol_list, 1);
1ff442ca
NF
1527 p->sym = symval;
1528 p1->next = p;
1529 p1 = p;
1530 }
a70083a3 1531 else /* handle an action. */
1ff442ca 1532 {
a70083a3 1533 copy_action (crule, rulelength);
89cab50d 1534 action_flag = 1;
1ff442ca
NF
1535 xactions++; /* JF */
1536 }
1537 rulelength++;
a70083a3 1538 } /* end of read rhs of rule */
1ff442ca
NF
1539
1540 /* Put an empty link in the list to mark the end of this rule */
d7913476 1541 p = XCALLOC (symbol_list, 1);
1ff442ca
NF
1542 p1->next = p;
1543 p1 = p;
1544
511e79b3 1545 if (t == tok_prec)
1ff442ca 1546 {
a0f6b076 1547 complain (_("two @prec's in a row"));
a70083a3 1548 t = lex ();
1ff442ca 1549 crule->ruleprec = symval;
a70083a3 1550 t = lex ();
1ff442ca 1551 }
511e79b3 1552 if (t == tok_guard)
1ff442ca 1553 {
a70083a3 1554 if (!semantic_parser)
ff4a34be 1555 complain (_("%%guard present but %%semantic_parser not specified"));
1ff442ca 1556
a70083a3
AD
1557 copy_guard (crule, rulelength);
1558 t = lex ();
1ff442ca 1559 }
511e79b3 1560 else if (t == tok_left_curly)
1ff442ca 1561 {
a70083a3 1562 /* This case never occurs -wjh */
89cab50d 1563 if (action_flag)
a0f6b076 1564 complain (_("two actions at end of one rule"));
a70083a3 1565 copy_action (crule, rulelength);
89cab50d 1566 action_flag = 1;
943819bf 1567 xactions++; /* -wjh */
a70083a3 1568 t = lex ();
1ff442ca 1569 }
a0f6b076 1570 /* If $$ is being set in default way, report if any type
6666f98f
AD
1571 mismatch. */
1572 else if (!xactions
a70083a3 1573 && first_rhs && lhs->type_name != first_rhs->type_name)
1ff442ca 1574 {
6666f98f
AD
1575 if (lhs->type_name == 0
1576 || first_rhs->type_name == 0
a70083a3 1577 || strcmp (lhs->type_name, first_rhs->type_name))
a0f6b076
AD
1578 complain (_("type clash (`%s' `%s') on default action"),
1579 lhs->type_name ? lhs->type_name : "",
a70083a3 1580 first_rhs->type_name ? first_rhs->type_name : "");
1ff442ca
NF
1581 }
1582 /* Warn if there is no default for $$ but we need one. */
1583 else if (!xactions && !first_rhs && lhs->type_name != 0)
a0f6b076 1584 complain (_("empty rule for typed nonterminal, and no action"));
511e79b3 1585 if (t == tok_semicolon)
a70083a3 1586 t = lex ();
a083fbbf 1587 }
943819bf 1588#if 0
a70083a3 1589 /* these things can appear as alternatives to rules. */
943819bf
RS
1590/* NO, they cannot.
1591 a) none of the documentation allows them
1592 b) most of them scan forward until finding a next %
1593 thus they may swallow lots of intervening rules
1594*/
511e79b3 1595 else if (t == tok_token)
1ff442ca 1596 {
d7020c20 1597 parse_token_decl (token_sym, nterm_sym);
a70083a3 1598 t = lex ();
1ff442ca 1599 }
511e79b3 1600 else if (t == tok_nterm)
1ff442ca 1601 {
d7020c20 1602 parse_token_decl (nterm_sym, token_sym);
a70083a3 1603 t = lex ();
1ff442ca 1604 }
511e79b3 1605 else if (t == tok_type)
1ff442ca 1606 {
a70083a3 1607 t = get_type ();
1ff442ca 1608 }
511e79b3 1609 else if (t == tok_union)
1ff442ca 1610 {
a70083a3
AD
1611 parse_union_decl ();
1612 t = lex ();
1ff442ca 1613 }
511e79b3 1614 else if (t == tok_expect)
1ff442ca 1615 {
a70083a3
AD
1616 parse_expect_decl ();
1617 t = lex ();
1ff442ca 1618 }
511e79b3 1619 else if (t == tok_start)
1ff442ca 1620 {
a70083a3
AD
1621 parse_start_decl ();
1622 t = lex ();
1ff442ca 1623 }
943819bf
RS
1624#endif
1625
1ff442ca 1626 else
943819bf 1627 {
d01c415b 1628 complain (_("invalid input: %s"), quote (token_buffer));
a70083a3 1629 t = lex ();
943819bf 1630 }
1ff442ca
NF
1631 }
1632
943819bf
RS
1633 /* grammar has been read. Do some checking */
1634
1ff442ca 1635 if (nsyms > MAXSHORT)
a0f6b076
AD
1636 fatal (_("too many symbols (tokens plus nonterminals); maximum %d"),
1637 MAXSHORT);
1ff442ca 1638 if (nrules == 0)
a0f6b076 1639 fatal (_("no rules in the input grammar"));
1ff442ca 1640
b6610515 1641#if 0 /* This code is in the skeleton now. */
ff4a34be
AD
1642 /* JF put out same default YYSTYPE as YACC does */
1643 if (typed == 0
1ff442ca
NF
1644 && !value_components_used)
1645 {
1646 /* We used to use `unsigned long' as YYSTYPE on MSDOS,
a70083a3
AD
1647 but it seems better to be consistent.
1648 Most programs should declare their own type anyway. */
ff4423cc 1649 obstack_sgrow (&attrs_obstack,
573c1d9f 1650 "#ifndef YYSTYPE\n#define YYSTYPE int\n#endif\n");
896fe5c1 1651 if (defines_flag)
ff4423cc 1652 obstack_sgrow (&defines_obstack, "\
0b8afb77
AD
1653# ifndef YYSTYPE\n\
1654# define YYSTYPE int\n\
1655# endif\n");
1ff442ca 1656 }
b6610515 1657#endif
1ff442ca
NF
1658
1659 /* Report any undefined symbols and consider them nonterminals. */
1660
1661 for (bp = firstsymbol; bp; bp = bp->next)
d7020c20 1662 if (bp->class == unknown_sym)
1ff442ca 1663 {
a70083a3
AD
1664 complain (_
1665 ("symbol %s is used, but is not defined as a token and has no rules"),
ff4a34be 1666 bp->tag);
d7020c20 1667 bp->class = nterm_sym;
1ff442ca
NF
1668 bp->value = nvars++;
1669 }
1670
1671 ntokens = nsyms - nvars;
1672}
a70083a3
AD
1673\f
1674/*--------------------------------------------------------------.
1675| For named tokens, but not literal ones, define the name. The |
1676| value is the user token number. |
1677`--------------------------------------------------------------*/
1ff442ca 1678
4a120d45 1679static void
896fe5c1 1680output_token_defines (struct obstack *oout)
1ff442ca 1681{
a70083a3
AD
1682 bucket *bp;
1683 char *cp, *symbol;
1684 char c;
1ff442ca 1685
a70083a3 1686 for (bp = firstsymbol; bp; bp = bp->next)
1ff442ca 1687 {
a70083a3
AD
1688 symbol = bp->tag; /* get symbol */
1689
1690 if (bp->value >= ntokens)
1691 continue;
1692 if (bp->user_token_number == SALIAS)
1693 continue;
1694 if ('\'' == *symbol)
1695 continue; /* skip literal character */
1696 if (bp == errtoken)
1697 continue; /* skip error token */
1698 if ('\"' == *symbol)
1ff442ca 1699 {
a70083a3
AD
1700 /* use literal string only if given a symbol with an alias */
1701 if (bp->alias)
1702 symbol = bp->alias->tag;
1703 else
1704 continue;
1705 }
1ff442ca 1706
a70083a3
AD
1707 /* Don't #define nonliteral tokens whose names contain periods. */
1708 cp = symbol;
1709 while ((c = *cp++) && c != '.');
1710 if (c != '\0')
1711 continue;
1ff442ca 1712
0b8afb77 1713 obstack_fgrow2 (oout, "# define\t%s\t%d\n",
896fe5c1 1714 symbol,
62ab6972 1715 (translations ? bp->user_token_number : bp->value));
a70083a3 1716 if (semantic_parser)
0b8afb77 1717 obstack_fgrow2 (oout, "# define\tT%s\t%d\n", symbol, bp->value);
1ff442ca 1718 }
a70083a3 1719
b6610515 1720 /* obstack_1grow (oout, '\n'); */
1ff442ca 1721}
1ff442ca
NF
1722
1723
a70083a3
AD
1724/*------------------------------------------------------------------.
1725| Assign symbol numbers, and write definition of token names into |
b2ca4022 1726| FDEFINES. Set up vectors TAGS and SPREC of names and precedences |
a70083a3
AD
1727| of symbols. |
1728`------------------------------------------------------------------*/
1ff442ca 1729
4a120d45 1730static void
118fb205 1731packsymbols (void)
1ff442ca 1732{
a70083a3
AD
1733 bucket *bp;
1734 int tokno = 1;
1735 int i;
1736 int last_user_token_number;
4a120d45 1737 static char DOLLAR[] = "$";
1ff442ca
NF
1738
1739 /* int lossage = 0; JF set but not used */
1740
d7913476 1741 tags = XCALLOC (char *, nsyms + 1);
4a120d45 1742 tags[0] = DOLLAR;
d7913476 1743 user_toknums = XCALLOC (short, nsyms + 1);
943819bf 1744 user_toknums[0] = 0;
1ff442ca 1745
d7913476
AD
1746 sprec = XCALLOC (short, nsyms);
1747 sassoc = XCALLOC (short, nsyms);
1ff442ca
NF
1748
1749 max_user_token_number = 256;
1750 last_user_token_number = 256;
1751
1752 for (bp = firstsymbol; bp; bp = bp->next)
1753 {
d7020c20 1754 if (bp->class == nterm_sym)
1ff442ca
NF
1755 {
1756 bp->value += ntokens;
1757 }
943819bf
RS
1758 else if (bp->alias)
1759 {
0a6384c4
AD
1760 /* this symbol and its alias are a single token defn.
1761 allocate a tokno, and assign to both check agreement of
1762 ->prec and ->assoc fields and make both the same */
1763 if (bp->value == 0)
1764 bp->value = bp->alias->value = tokno++;
943819bf 1765
0a6384c4
AD
1766 if (bp->prec != bp->alias->prec)
1767 {
1768 if (bp->prec != 0 && bp->alias->prec != 0
1769 && bp->user_token_number == SALIAS)
a0f6b076
AD
1770 complain (_("conflicting precedences for %s and %s"),
1771 bp->tag, bp->alias->tag);
0a6384c4
AD
1772 if (bp->prec != 0)
1773 bp->alias->prec = bp->prec;
1774 else
1775 bp->prec = bp->alias->prec;
1776 }
943819bf 1777
0a6384c4
AD
1778 if (bp->assoc != bp->alias->assoc)
1779 {
a0f6b076
AD
1780 if (bp->assoc != 0 && bp->alias->assoc != 0
1781 && bp->user_token_number == SALIAS)
1782 complain (_("conflicting assoc values for %s and %s"),
1783 bp->tag, bp->alias->tag);
1784 if (bp->assoc != 0)
1785 bp->alias->assoc = bp->assoc;
1786 else
1787 bp->assoc = bp->alias->assoc;
1788 }
0a6384c4
AD
1789
1790 if (bp->user_token_number == SALIAS)
a70083a3 1791 continue; /* do not do processing below for SALIASs */
943819bf 1792
a70083a3 1793 }
d7020c20 1794 else /* bp->class == token_sym */
943819bf
RS
1795 {
1796 bp->value = tokno++;
1797 }
1798
d7020c20 1799 if (bp->class == token_sym)
1ff442ca
NF
1800 {
1801 if (translations && !(bp->user_token_number))
1802 bp->user_token_number = ++last_user_token_number;
1803 if (bp->user_token_number > max_user_token_number)
1804 max_user_token_number = bp->user_token_number;
1ff442ca
NF
1805 }
1806
1807 tags[bp->value] = bp->tag;
943819bf 1808 user_toknums[bp->value] = bp->user_token_number;
1ff442ca
NF
1809 sprec[bp->value] = bp->prec;
1810 sassoc[bp->value] = bp->assoc;
1811
1812 }
1813
1814 if (translations)
1815 {
a70083a3 1816 int j;
1ff442ca 1817
d7913476 1818 token_translations = XCALLOC (short, max_user_token_number + 1);
1ff442ca 1819
0a6384c4 1820 /* initialize all entries for literal tokens to 2, the internal
a70083a3
AD
1821 token number for $undefined., which represents all invalid
1822 inputs. */
4a120d45 1823 for (j = 0; j <= max_user_token_number; j++)
a70083a3 1824 token_translations[j] = 2;
1ff442ca 1825
943819bf 1826 for (bp = firstsymbol; bp; bp = bp->next)
a70083a3
AD
1827 {
1828 if (bp->value >= ntokens)
1829 continue; /* non-terminal */
1830 if (bp->user_token_number == SALIAS)
0a6384c4 1831 continue;
a70083a3 1832 if (token_translations[bp->user_token_number] != 2)
a0f6b076
AD
1833 complain (_("tokens %s and %s both assigned number %d"),
1834 tags[token_translations[bp->user_token_number]],
a70083a3
AD
1835 bp->tag, bp->user_token_number);
1836 token_translations[bp->user_token_number] = bp->value;
1837 }
1ff442ca
NF
1838 }
1839
1840 error_token_number = errtoken->value;
1841
b6610515
RA
1842 output_token_defines (&output_obstack);
1843 obstack_1grow (&output_obstack, 0);
1844 macro_insert ("tokendef", obstack_finish (&output_obstack));
1845
1846 /* if (!no_parser_flag)
1847 output_token_defines (&table_obstack); */
1ff442ca 1848
d7020c20 1849 if (startval->class == unknown_sym)
a0f6b076 1850 fatal (_("the start symbol %s is undefined"), startval->tag);
d7020c20 1851 else if (startval->class == token_sym)
a0f6b076 1852 fatal (_("the start symbol %s is a token"), startval->tag);
1ff442ca
NF
1853
1854 start_symbol = startval->value;
1855
89cab50d 1856 if (defines_flag)
1ff442ca 1857 {
896fe5c1 1858 output_token_defines (&defines_obstack);
1ff442ca
NF
1859
1860 if (!pure_parser)
1861 {
1862 if (spec_name_prefix)
896fe5c1
AD
1863 obstack_fgrow1 (&defines_obstack, "\nextern YYSTYPE %slval;\n",
1864 spec_name_prefix);
1ff442ca 1865 else
ff4423cc 1866 obstack_sgrow (&defines_obstack,
573c1d9f 1867 "\nextern YYSTYPE yylval;\n");
1ff442ca
NF
1868 }
1869
1870 if (semantic_parser)
1871 for (i = ntokens; i < nsyms; i++)
1872 {
1873 /* don't make these for dummy nonterminals made by gensym. */
1874 if (*tags[i] != '@')
896fe5c1 1875 obstack_fgrow2 (&defines_obstack,
0b8afb77 1876 "# define\tNT%s\t%d\n", tags[i], i);
1ff442ca
NF
1877 }
1878#if 0
1879 /* `fdefines' is now a temporary file, so we need to copy its
1880 contents in `done', so we can't close it here. */
a70083a3 1881 fclose (fdefines);
1ff442ca
NF
1882 fdefines = NULL;
1883#endif
1884 }
1885}
a083fbbf 1886
1ff442ca 1887
a70083a3
AD
1888/*---------------------------------------------------------------.
1889| Convert the rules into the representation using RRHS, RLHS and |
1890| RITEMS. |
1891`---------------------------------------------------------------*/
1ff442ca 1892
4a120d45 1893static void
118fb205 1894packgram (void)
1ff442ca 1895{
a70083a3
AD
1896 int itemno;
1897 int ruleno;
1898 symbol_list *p;
1ff442ca
NF
1899
1900 bucket *ruleprec;
1901
d7913476
AD
1902 ritem = XCALLOC (short, nitems + 1);
1903 rlhs = XCALLOC (short, nrules) - 1;
1904 rrhs = XCALLOC (short, nrules) - 1;
1905 rprec = XCALLOC (short, nrules) - 1;
1906 rprecsym = XCALLOC (short, nrules) - 1;
1907 rassoc = XCALLOC (short, nrules) - 1;
1ff442ca
NF
1908
1909 itemno = 0;
1910 ruleno = 1;
1911
1912 p = grammar;
1913 while (p)
1914 {
1915 rlhs[ruleno] = p->sym->value;
1916 rrhs[ruleno] = itemno;
1917 ruleprec = p->ruleprec;
1918
1919 p = p->next;
1920 while (p && p->sym)
1921 {
1922 ritem[itemno++] = p->sym->value;
1923 /* A rule gets by default the precedence and associativity
1924 of the last token in it. */
d7020c20 1925 if (p->sym->class == token_sym)
1ff442ca
NF
1926 {
1927 rprec[ruleno] = p->sym->prec;
1928 rassoc[ruleno] = p->sym->assoc;
1929 }
a70083a3
AD
1930 if (p)
1931 p = p->next;
1ff442ca
NF
1932 }
1933
1934 /* If this rule has a %prec,
a70083a3 1935 the specified symbol's precedence replaces the default. */
1ff442ca
NF
1936 if (ruleprec)
1937 {
a70083a3
AD
1938 rprec[ruleno] = ruleprec->prec;
1939 rassoc[ruleno] = ruleprec->assoc;
1ff442ca
NF
1940 rprecsym[ruleno] = ruleprec->value;
1941 }
1942
1943 ritem[itemno++] = -ruleno;
1944 ruleno++;
1945
a70083a3
AD
1946 if (p)
1947 p = p->next;
1ff442ca
NF
1948 }
1949
1950 ritem[itemno] = 0;
1951}
a70083a3
AD
1952\f
1953/*-------------------------------------------------------------------.
1954| Read in the grammar specification and record it in the format |
ea5607fd 1955| described in gram.h. All guards are copied into the GUARD_OBSTACK |
8c7ebe49
AD
1956| and all actions into ACTION_OBSTACK, in each case forming the body |
1957| of a C function (YYGUARD or YYACTION) which contains a switch |
1958| statement to decide which guard or action to execute. |
a70083a3
AD
1959`-------------------------------------------------------------------*/
1960
1961void
1962reader (void)
1963{
1964 start_flag = 0;
1965 startval = NULL; /* start symbol not specified yet. */
1966
1967#if 0
1968 /* initially assume token number translation not needed. */
1969 translations = 0;
1970#endif
1971 /* Nowadays translations is always set to 1, since we give `error' a
1972 user-token-number to satisfy the Posix demand for YYERRCODE==256.
1973 */
1974 translations = 1;
1975
1976 nsyms = 1;
1977 nvars = 0;
1978 nrules = 0;
1979 nitems = 0;
1980 rline_allocated = 10;
d7913476 1981 rline = XCALLOC (short, rline_allocated);
a70083a3
AD
1982
1983 typed = 0;
1984 lastprec = 0;
1985
a70083a3
AD
1986 semantic_parser = 0;
1987 pure_parser = 0;
a70083a3
AD
1988
1989 grammar = NULL;
1990
1991 init_lex ();
1992 lineno = 1;
1993
1994 /* Initialize the symbol table. */
1995 tabinit ();
b6610515 1996
a70083a3
AD
1997 /* Construct the error token */
1998 errtoken = getsym ("error");
d7020c20 1999 errtoken->class = token_sym;
a70083a3 2000 errtoken->user_token_number = 256; /* Value specified by POSIX. */
b6610515 2001
a70083a3
AD
2002 /* Construct a token that represents all undefined literal tokens.
2003 It is always token number 2. */
2004 undeftoken = getsym ("$undefined.");
d7020c20 2005 undeftoken->class = token_sym;
a70083a3
AD
2006 undeftoken->user_token_number = 2;
2007
896fe5c1
AD
2008 /* Read the declaration section. Copy %{ ... %} groups to
2009 TABLE_OBSTACK and FDEFINES file. Also notice any %token, %left,
2010 etc. found there. */
b6610515
RA
2011 /* obstack_1grow (&table_obstack, '\n'); */
2012
a70083a3
AD
2013 read_declarations ();
2014 /* Start writing the guard and action files, if they are needed. */
b6610515 2015 /* output_headers (); */
a70083a3
AD
2016 /* Read in the grammar, build grammar in list form. Write out
2017 guards and actions. */
2018 readgram ();
2019 /* Now we know whether we need the line-number stack. If we do,
2020 write its type into the .tab.h file. */
b6610515
RA
2021 /* if (defines_flag)
2022 reader_output_yylsp (&defines_obstack); */
a70083a3 2023 /* Write closing delimiters for actions and guards. */
b6610515
RA
2024 /* output_trailers (); */
2025 /* if (locations_flag)
2026 obstack_sgrow (&table_obstack, "#define YYLSP_NEEDED 1\n\n"); */
a70083a3
AD
2027 /* Assign the symbols their symbol numbers. Write #defines for the
2028 token symbols into FDEFINES if requested. */
2029 packsymbols ();
2030 /* Convert the grammar into the format described in gram.h. */
2031 packgram ();
2032 /* Free the symbol table data structure since symbols are now all
2033 referred to by symbol number. */
2034 free_symtab ();
2035}