xref: /freebsd/contrib/byacc/main.c (revision 2008043f386721d58158e37e0d7e50df8095942d)
1 /* $Id: main.c,v 1.73 2021/08/08 20:39:34 tom Exp $ */
2 
3 #include <signal.h>
4 #if !defined(_WIN32) || defined(__MINGW32__)
5 #include <unistd.h>		/* for _exit() */
6 #else
7 #include <stdlib.h>		/* for _exit() */
8 #endif
9 
10 #include "defs.h"
11 
12 #ifdef HAVE_MKSTEMP
13 # define USE_MKSTEMP 1
14 #elif defined(HAVE_FCNTL_H)
15 # define USE_MKSTEMP 1
16 # include <fcntl.h>		/* for open(), O_EXCL, etc. */
17 #else
18 # define USE_MKSTEMP 0
19 #endif
20 
21 #if USE_MKSTEMP
22 #include <sys/types.h>
23 #include <sys/stat.h>
24 
25 typedef struct _my_tmpfiles
26 {
27     struct _my_tmpfiles *next;
28     char *name;
29 }
30 MY_TMPFILES;
31 
32 static MY_TMPFILES *my_tmpfiles;
33 #endif /* USE_MKSTEMP */
34 
35 char dflag;
36 char dflag2;
37 char gflag;
38 char iflag;
39 char lflag;
40 static char oflag;
41 char rflag;
42 char sflag;
43 char tflag;
44 char vflag;
45 
46 const char *symbol_prefix;
47 const char *myname = "yacc";
48 
49 int lineno;
50 int outline;
51 
52 static char default_file_prefix[] = "y";
53 
54 static char *file_prefix = default_file_prefix;
55 
56 char *code_file_name;
57 char *input_file_name;
58 size_t input_file_name_len = 0;
59 char *defines_file_name;
60 char *externs_file_name;
61 
62 static char *graph_file_name;
63 static char *output_file_name;
64 static char *verbose_file_name;
65 
66 FILE *action_file;	/*  a temp file, used to save actions associated    */
67 			/*  with rules until the parser is written          */
68 FILE *code_file;	/*  y.code.c (used when the -r option is specified) */
69 FILE *defines_file;	/*  y.tab.h                                         */
70 FILE *externs_file;	/*  y.tab.i                                         */
71 FILE *input_file;	/*  the input file                                  */
72 FILE *output_file;	/*  y.tab.c                                         */
73 FILE *text_file;	/*  a temp file, used to save text until all        */
74 			/*  symbols have been defined                       */
75 FILE *union_file;	/*  a temp file, used to save the union             */
76 			/*  definition until all symbol have been           */
77 			/*  defined                                         */
78 FILE *verbose_file;	/*  y.output                                        */
79 FILE *graph_file;	/*  y.dot                                           */
80 
81 Value_t nitems;
82 Value_t nrules;
83 Value_t nsyms;
84 Value_t ntokens;
85 Value_t nvars;
86 
87 Value_t start_symbol;
88 char **symbol_name;
89 char **symbol_pname;
90 Value_t *symbol_value;
91 Value_t *symbol_prec;
92 char *symbol_assoc;
93 
94 int pure_parser;
95 int token_table;
96 int error_verbose;
97 
98 #if defined(YYBTYACC)
99 Value_t *symbol_pval;
100 char **symbol_destructor;
101 char **symbol_type_tag;
102 int locations = 0;	/* default to no position processing */
103 int backtrack = 0;	/* default is no backtracking */
104 char *initial_action = NULL;
105 #endif
106 
107 int exit_code;
108 
109 Value_t *ritem;
110 Value_t *rlhs;
111 Value_t *rrhs;
112 Value_t *rprec;
113 Assoc_t *rassoc;
114 Value_t **derives;
115 char *nullable;
116 
117 /*
118  * Since fclose() is called via the signal handler, it might die.  Don't loop
119  * if there is a problem closing a file.
120  */
121 #define DO_CLOSE(fp) \
122 	if (fp != 0) { \
123 	    FILE *use = fp; \
124 	    fp = 0; \
125 	    fclose(use); \
126 	}
127 
128 static int got_intr = 0;
129 
130 void
131 done(int k)
132 {
133     DO_CLOSE(input_file);
134     DO_CLOSE(output_file);
135     if (iflag)
136 	DO_CLOSE(externs_file);
137     if (rflag)
138 	DO_CLOSE(code_file);
139 
140     DO_CLOSE(action_file);
141     DO_CLOSE(defines_file);
142     DO_CLOSE(graph_file);
143     DO_CLOSE(text_file);
144     DO_CLOSE(union_file);
145     DO_CLOSE(verbose_file);
146 
147     if (got_intr)
148 	_exit(EXIT_FAILURE);
149 
150 #ifdef NO_LEAKS
151     DO_FREE(input_file_name);
152 
153     if (rflag)
154 	DO_FREE(code_file_name);
155 
156     if (dflag && !dflag2)
157 	DO_FREE(defines_file_name);
158 
159     if (iflag)
160 	DO_FREE(externs_file_name);
161 
162     if (oflag)
163 	DO_FREE(output_file_name);
164 
165     if (vflag)
166 	DO_FREE(verbose_file_name);
167 
168     if (gflag)
169 	DO_FREE(graph_file_name);
170 
171     lr0_leaks();
172     lalr_leaks();
173     mkpar_leaks();
174     mstring_leaks();
175     output_leaks();
176     reader_leaks();
177 #endif
178 
179     exit(k);
180 }
181 
182 static void
183 onintr(int sig GCC_UNUSED)
184 {
185     got_intr = 1;
186     done(EXIT_FAILURE);
187 }
188 
189 static void
190 set_signals(void)
191 {
192 #ifdef SIGINT
193     if (signal(SIGINT, SIG_IGN) != SIG_IGN)
194 	signal(SIGINT, onintr);
195 #endif
196 #ifdef SIGTERM
197     if (signal(SIGTERM, SIG_IGN) != SIG_IGN)
198 	signal(SIGTERM, onintr);
199 #endif
200 #ifdef SIGHUP
201     if (signal(SIGHUP, SIG_IGN) != SIG_IGN)
202 	signal(SIGHUP, onintr);
203 #endif
204 }
205 
206 #define SIZEOF(v) (sizeof(v) / sizeof((v)[0]))
207 
208 /*
209  * Long options are provided only as a compatibility aid for scripters.
210  */
211 /* *INDENT-OFF* */
212 static const struct {
213     const char long_opt[16];
214     const char yacc_arg;
215     const char yacc_opt;
216 } long_opts[] = {
217     { "defines",     1, 'H' },
218     { "file-prefix", 1, 'b' },
219     { "graph",       0, 'g' },
220     { "help",        0, 'h' },
221     { "name-prefix", 1, 'p' },
222     { "no-lines",    0, 'l' },
223     { "output",      1, 'o' },
224     { "version",     0, 'V' }
225 };
226 /* *INDENT-ON* */
227 
228 /*
229  * Usage-message is designed for 80 columns, with some unknowns.  Account for
230  * those in the maximum width so that the usage message uses no relocatable
231  * pointers.
232  */
233 #define USAGE_COLS (80 + sizeof(DEFINES_SUFFIX) + sizeof(OUTPUT_SUFFIX))
234 
235 static void
236 usage(void)
237 {
238     /* *INDENT-OFF* */
239     static const char msg[][USAGE_COLS] =
240     {
241 	{ "  -b file_prefix        set filename prefix (default \"y.\")" },
242 	{ "  -B                    create a backtracking parser" },
243 	{ "  -d                    write definitions (" DEFINES_SUFFIX ")" },
244 	{ "  -h                    print this help-message" },
245 	{ "  -H defines_file       write definitions to defines_file" },
246 	{ "  -i                    write interface (y.tab.i)" },
247 	{ "  -g                    write a graphical description" },
248 	{ "  -l                    suppress #line directives" },
249 	{ "  -L                    enable position processing, e.g., \"%locations\"" },
250 	{ "  -o output_file        (default \"" OUTPUT_SUFFIX "\")" },
251 	{ "  -p symbol_prefix      set symbol prefix (default \"yy\")" },
252 	{ "  -P                    create a reentrant parser, e.g., \"%pure-parser\"" },
253 	{ "  -r                    produce separate code and table files (y.code.c)" },
254 	{ "  -s                    suppress #define's for quoted names in %token lines" },
255 	{ "  -t                    add debugging support" },
256 	{ "  -v                    write description (y.output)" },
257 	{ "  -V                    show version information and exit" },
258     };
259     /* *INDENT-ON* */
260     unsigned n;
261 
262     fflush(stdout);
263     fprintf(stderr, "Usage: %s [options] filename\n", myname);
264 
265     fprintf(stderr, "\nOptions:\n");
266     for (n = 0; n < SIZEOF(msg); ++n)
267     {
268 	fprintf(stderr, "%s\n", msg[n]);
269     }
270 
271     fprintf(stderr, "\nLong options:\n");
272     for (n = 0; n < SIZEOF(long_opts); ++n)
273     {
274 	fprintf(stderr, "  --%-20s-%c\n",
275 		long_opts[n].long_opt,
276 		long_opts[n].yacc_opt);
277     }
278 
279     exit(EXIT_FAILURE);
280 }
281 
282 static void
283 invalid_option(const char *option)
284 {
285     fprintf(stderr, "invalid option: %s\n", option);
286     usage();
287 }
288 
289 static void
290 setflag(int ch)
291 {
292     switch (ch)
293     {
294     case 'B':
295 #if defined(YYBTYACC)
296 	backtrack = 1;
297 #else
298 	unsupported_flag_warning("-B", "reconfigure with --enable-btyacc");
299 #endif
300 	break;
301 
302     case 'd':
303 	dflag = 1;
304 	dflag2 = 0;
305 	break;
306 
307     case 'g':
308 	gflag = 1;
309 	break;
310 
311     case 'i':
312 	iflag = 1;
313 	break;
314 
315     case 'l':
316 	lflag = 1;
317 	break;
318 
319     case 'L':
320 #if defined(YYBTYACC)
321 	locations = 1;
322 #else
323 	unsupported_flag_warning("-L", "reconfigure with --enable-btyacc");
324 #endif
325 	break;
326 
327     case 'P':
328 	pure_parser = 1;
329 	break;
330 
331     case 'r':
332 	rflag = 1;
333 	break;
334 
335     case 's':
336 	sflag = 1;
337 	break;
338 
339     case 't':
340 	tflag = 1;
341 	break;
342 
343     case 'v':
344 	vflag = 1;
345 	break;
346 
347     case 'V':
348 	printf("%s - %s\n", myname, VERSION);
349 	exit(EXIT_SUCCESS);
350 
351     case 'y':
352 	/* noop for bison compatibility. byacc is already designed to be posix
353 	 * yacc compatible. */
354 	break;
355 
356     default:
357 	usage();
358     }
359 }
360 
361 static void
362 getargs(int argc, char *argv[])
363 {
364     int i;
365 #ifdef HAVE_GETOPT
366     int ch;
367 #endif
368 
369     /*
370      * Map bison's long-options into yacc short options.
371      */
372     for (i = 1; i < argc; ++i)
373     {
374 	char *a = argv[i];
375 
376 	if (!strncmp(a, "--", 2))
377 	{
378 	    char *eqls;
379 	    size_t lc;
380 	    size_t len;
381 
382 	    if ((len = strlen(a)) == 2)
383 		break;
384 
385 	    if ((eqls = strchr(a, '=')) != NULL)
386 	    {
387 		len = (size_t)(eqls - a);
388 		if (len == 0 || eqls[1] == '\0')
389 		    invalid_option(a);
390 	    }
391 
392 	    for (lc = 0; lc < SIZEOF(long_opts); ++lc)
393 	    {
394 		if (!strncmp(long_opts[lc].long_opt, a + 2, len - 2))
395 		{
396 		    if (eqls != NULL && !long_opts[lc].yacc_arg)
397 			invalid_option(a);
398 		    *a++ = '-';
399 		    *a++ = long_opts[lc].yacc_opt;
400 		    *a = '\0';
401 		    if (eqls)
402 		    {
403 			while ((*a++ = *++eqls) != '\0') /* empty */ ;
404 		    }
405 		    break;
406 		}
407 	    }
408 	    if (!strncmp(a, "--", 2))
409 		invalid_option(a);
410 	}
411     }
412 
413 #ifdef HAVE_GETOPT
414     if (argc > 0)
415 	myname = argv[0];
416 
417     while ((ch = getopt(argc, argv, "Bb:dghH:ilLo:Pp:rstVvy")) != -1)
418     {
419 	switch (ch)
420 	{
421 	case 'b':
422 	    file_prefix = optarg;
423 	    break;
424 	case 'h':
425 	    usage();
426 	    break;
427 	case 'H':
428 	    dflag = dflag2 = 1;
429 	    defines_file_name = optarg;
430 	    break;
431 	case 'o':
432 	    output_file_name = optarg;
433 	    break;
434 	case 'p':
435 	    symbol_prefix = optarg;
436 	    break;
437 	default:
438 	    setflag(ch);
439 	    break;
440 	}
441     }
442     if ((i = optind) < argc)
443     {
444 	/* getopt handles "--" specially, while we handle "-" specially */
445 	if (!strcmp(argv[i], "-"))
446 	{
447 	    if ((i + 1) < argc)
448 		usage();
449 	    input_file = stdin;
450 	    return;
451 	}
452     }
453 #else
454     char *s;
455     int ch;
456 
457     if (argc > 0)
458 	myname = argv[0];
459 
460     for (i = 1; i < argc; ++i)
461     {
462 	s = argv[i];
463 	if (*s != '-')
464 	    break;
465 	switch (ch = *++s)
466 	{
467 	case '\0':
468 	    input_file = stdin;
469 	    if (i + 1 < argc)
470 		usage();
471 	    return;
472 
473 	case '-':
474 	    ++i;
475 	    goto no_more_options;
476 
477 	case 'b':
478 	    if (*++s)
479 		file_prefix = s;
480 	    else if (++i < argc)
481 		file_prefix = argv[i];
482 	    else
483 		usage();
484 	    continue;
485 
486 	case 'H':
487 	    dflag = dflag2 = 1;
488 	    if (*++s)
489 		defines_file_name = s;
490 	    else if (++i < argc)
491 		defines_file_name = argv[i];
492 	    else
493 		usage();
494 	    continue;
495 
496 	case 'o':
497 	    if (*++s)
498 		output_file_name = s;
499 	    else if (++i < argc)
500 		output_file_name = argv[i];
501 	    else
502 		usage();
503 	    continue;
504 
505 	case 'p':
506 	    if (*++s)
507 		symbol_prefix = s;
508 	    else if (++i < argc)
509 		symbol_prefix = argv[i];
510 	    else
511 		usage();
512 	    continue;
513 
514 	default:
515 	    setflag(ch);
516 	    break;
517 	}
518 
519 	for (;;)
520 	{
521 	    switch (ch = *++s)
522 	    {
523 	    case '\0':
524 		goto end_of_option;
525 
526 	    default:
527 		setflag(ch);
528 		break;
529 	    }
530 	}
531       end_of_option:;
532     }
533 
534   no_more_options:
535 
536 #endif /* HAVE_GETOPT */
537     if (i + 1 != argc)
538 	usage();
539     input_file_name_len = strlen(argv[i]);
540     input_file_name = TMALLOC(char, input_file_name_len + 1);
541     NO_SPACE(input_file_name);
542     strcpy(input_file_name, argv[i]);
543 }
544 
545 void *
546 allocate(size_t n)
547 {
548     void *p;
549 
550     p = NULL;
551     if (n)
552     {
553 	p = CALLOC(1, n);
554 	NO_SPACE(p);
555     }
556     return (p);
557 }
558 
559 #define CREATE_FILE_NAME(dest, suffix) \
560 	dest = alloc_file_name(len, suffix)
561 
562 static char *
563 alloc_file_name(size_t len, const char *suffix)
564 {
565     char *result = TMALLOC(char, len + strlen(suffix) + 1);
566     if (result == 0)
567 	no_space();
568     strcpy(result, file_prefix);
569     strcpy(result + len, suffix);
570     return result;
571 }
572 
573 static char *
574 find_suffix(char *name, const char *suffix)
575 {
576     size_t len = strlen(name);
577     size_t slen = strlen(suffix);
578     if (len >= slen)
579     {
580 	name += len - slen;
581 	if (strcmp(name, suffix) == 0)
582 	    return name;
583     }
584     return NULL;
585 }
586 
587 static void
588 create_file_names(void)
589 {
590     size_t len;
591     const char *defines_suffix;
592     const char *externs_suffix;
593     char *suffix;
594 
595     suffix = NULL;
596     defines_suffix = DEFINES_SUFFIX;
597     externs_suffix = EXTERNS_SUFFIX;
598 
599     /* compute the file_prefix from the user provided output_file_name */
600     if (output_file_name != 0)
601     {
602 	if (!(suffix = find_suffix(output_file_name, OUTPUT_SUFFIX))
603 	    && (suffix = find_suffix(output_file_name, ".c")))
604 	{
605 	    defines_suffix = ".h";
606 	    externs_suffix = ".i";
607 	}
608     }
609 
610     if (suffix != NULL)
611     {
612 	len = (size_t)(suffix - output_file_name);
613 	file_prefix = TMALLOC(char, len + 1);
614 	NO_SPACE(file_prefix);
615 	strncpy(file_prefix, output_file_name, len)[len] = 0;
616     }
617     else
618 	len = strlen(file_prefix);
619 
620     /* if "-o filename" was not given */
621     if (output_file_name == 0)
622     {
623 	oflag = 1;
624 	CREATE_FILE_NAME(output_file_name, OUTPUT_SUFFIX);
625     }
626 
627     if (rflag)
628     {
629 	CREATE_FILE_NAME(code_file_name, CODE_SUFFIX);
630     }
631     else
632 	code_file_name = output_file_name;
633 
634     if (dflag && !dflag2)
635     {
636 	CREATE_FILE_NAME(defines_file_name, defines_suffix);
637     }
638 
639     if (iflag)
640     {
641 	CREATE_FILE_NAME(externs_file_name, externs_suffix);
642     }
643 
644     if (vflag)
645     {
646 	CREATE_FILE_NAME(verbose_file_name, VERBOSE_SUFFIX);
647     }
648 
649     if (gflag)
650     {
651 	CREATE_FILE_NAME(graph_file_name, GRAPH_SUFFIX);
652     }
653 
654     if (suffix != NULL)
655     {
656 	FREE(file_prefix);
657     }
658 }
659 
660 #if USE_MKSTEMP
661 static void
662 close_tmpfiles(void)
663 {
664     while (my_tmpfiles != 0)
665     {
666 	MY_TMPFILES *next = my_tmpfiles->next;
667 
668 	(void)chmod(my_tmpfiles->name, 0644);
669 	(void)unlink(my_tmpfiles->name);
670 
671 	free(my_tmpfiles->name);
672 	free(my_tmpfiles);
673 
674 	my_tmpfiles = next;
675     }
676 }
677 
678 #ifndef HAVE_MKSTEMP
679 static int
680 my_mkstemp(char *temp)
681 {
682     int fd;
683     char *dname;
684     char *fname;
685     char *name;
686 
687     /*
688      * Split-up to use tempnam, rather than tmpnam; the latter (like
689      * mkstemp) is unusable on Windows.
690      */
691     if ((fname = strrchr(temp, '/')) != 0)
692     {
693 	dname = strdup(temp);
694 	dname[++fname - temp] = '\0';
695     }
696     else
697     {
698 	dname = 0;
699 	fname = temp;
700     }
701     if ((name = tempnam(dname, fname)) != 0)
702     {
703 	fd = open(name, O_CREAT | O_EXCL | O_RDWR);
704 	strcpy(temp, name);
705     }
706     else
707     {
708 	fd = -1;
709     }
710 
711     if (dname != 0)
712 	free(dname);
713 
714     return fd;
715 }
716 #define mkstemp(s) my_mkstemp(s)
717 #endif
718 
719 #endif
720 
721 /*
722  * tmpfile() should be adequate, except that it may require special privileges
723  * to use, e.g., MinGW and Windows 7 where it tries to use the root directory.
724  */
725 static FILE *
726 open_tmpfile(const char *label)
727 {
728 #define MY_FMT "%s/%.*sXXXXXX"
729     FILE *result;
730 #if USE_MKSTEMP
731     const char *tmpdir;
732     char *name;
733 
734     if (((tmpdir = getenv("TMPDIR")) == 0 || access(tmpdir, W_OK) != 0) ||
735 	((tmpdir = getenv("TEMP")) == 0 || access(tmpdir, W_OK) != 0))
736     {
737 #ifdef P_tmpdir
738 	tmpdir = P_tmpdir;
739 #else
740 	tmpdir = "/tmp";
741 #endif
742 	if (access(tmpdir, W_OK) != 0)
743 	    tmpdir = ".";
744     }
745 
746     /* The size of the format is guaranteed to be longer than the result from
747      * printing empty strings with it; this calculation accounts for the
748      * string-lengths as well.
749      */
750     name = malloc(strlen(tmpdir) + sizeof(MY_FMT) + strlen(label));
751 
752     result = 0;
753     if (name != 0)
754     {
755 	int fd;
756 	const char *mark;
757 
758 	mode_t save_umask = umask(0177);
759 
760 	if ((mark = strrchr(label, '_')) == 0)
761 	    mark = label + strlen(label);
762 
763 	sprintf(name, MY_FMT, tmpdir, (int)(mark - label), label);
764 	fd = mkstemp(name);
765 	if (fd >= 0
766 	    && (result = fdopen(fd, "w+")) != 0)
767 	{
768 	    MY_TMPFILES *item;
769 
770 	    if (my_tmpfiles == 0)
771 	    {
772 		atexit(close_tmpfiles);
773 	    }
774 
775 	    item = NEW(MY_TMPFILES);
776 	    NO_SPACE(item);
777 
778 	    item->name = name;
779 	    NO_SPACE(item->name);
780 
781 	    item->next = my_tmpfiles;
782 	    my_tmpfiles = item;
783 	}
784 	else
785 	{
786 	    FREE(name);
787 	}
788 	(void)umask(save_umask);
789     }
790 #else
791     result = tmpfile();
792 #endif
793 
794     if (result == 0)
795 	open_error(label);
796     return result;
797 #undef MY_FMT
798 }
799 
800 static void
801 open_files(void)
802 {
803     create_file_names();
804 
805     if (input_file == 0)
806     {
807 	input_file = fopen(input_file_name, "r");
808 	if (input_file == 0)
809 	    open_error(input_file_name);
810     }
811 
812     action_file = open_tmpfile("action_file");
813     text_file = open_tmpfile("text_file");
814 
815     if (vflag)
816     {
817 	verbose_file = fopen(verbose_file_name, "w");
818 	if (verbose_file == 0)
819 	    open_error(verbose_file_name);
820     }
821 
822     if (gflag)
823     {
824 	graph_file = fopen(graph_file_name, "w");
825 	if (graph_file == 0)
826 	    open_error(graph_file_name);
827 	fprintf(graph_file, "digraph %s {\n", file_prefix);
828 	fprintf(graph_file, "\tedge [fontsize=10];\n");
829 	fprintf(graph_file, "\tnode [shape=box,fontsize=10];\n");
830 	fprintf(graph_file, "\torientation=landscape;\n");
831 	fprintf(graph_file, "\trankdir=LR;\n");
832 	fprintf(graph_file, "\t/*\n");
833 	fprintf(graph_file, "\tmargin=0.2;\n");
834 	fprintf(graph_file, "\tpage=\"8.27,11.69\"; // for A4 printing\n");
835 	fprintf(graph_file, "\tratio=auto;\n");
836 	fprintf(graph_file, "\t*/\n");
837     }
838 
839     if (dflag || dflag2)
840     {
841 	defines_file = fopen(defines_file_name, "w");
842 	if (defines_file == 0)
843 	    open_error(defines_file_name);
844 	union_file = open_tmpfile("union_file");
845     }
846 
847     if (iflag)
848     {
849 	externs_file = fopen(externs_file_name, "w");
850 	if (externs_file == 0)
851 	    open_error(externs_file_name);
852     }
853 
854     output_file = fopen(output_file_name, "w");
855     if (output_file == 0)
856 	open_error(output_file_name);
857 
858     if (rflag)
859     {
860 	code_file = fopen(code_file_name, "w");
861 	if (code_file == 0)
862 	    open_error(code_file_name);
863     }
864     else
865 	code_file = output_file;
866 }
867 
868 int
869 main(int argc, char *argv[])
870 {
871     SRexpect = -1;
872     RRexpect = -1;
873     exit_code = EXIT_SUCCESS;
874 
875     set_signals();
876     getargs(argc, argv);
877     open_files();
878     reader();
879     lr0();
880     lalr();
881     make_parser();
882     graph();
883     finalize_closure();
884     verbose();
885     output();
886     done(exit_code);
887     /*NOTREACHED */
888 }
889