1 /*-
2 * Copyright (c) 1992 Diomidis Spinellis.
3 * Copyright (c) 1992, 1993
4 * The Regents of the University of California. All rights reserved.
5 *
6 * This code is derived from software contributed to Berkeley by
7 * Diomidis Spinellis of Imperial College, University of London.
8 *
9 * %sccs.include.redist.c%
10 */
11
12 #ifndef lint
13 static char sccsid[] = "@(#)compile.c 8.2 (Berkeley) 04/28/95";
14 #endif /* not lint */
15
16 #include <sys/types.h>
17 #include <sys/stat.h>
18
19 #include <ctype.h>
20 #include <errno.h>
21 #include <fcntl.h>
22 #include <limits.h>
23 #include <regex.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27
28 #include "defs.h"
29 #include "extern.h"
30
31 #define LHSZ 128
32 #define LHMASK (LHSZ - 1)
33 static struct labhash {
34 struct labhash *lh_next;
35 u_int lh_hash;
36 struct s_command *lh_cmd;
37 int lh_ref;
38 } *labels[LHSZ];
39
40 static char *compile_addr __P((char *, struct s_addr *));
41 static char *compile_delimited __P((char *, char *));
42 static char *compile_flags __P((char *, struct s_subst *));
43 static char *compile_re __P((char *, regex_t **));
44 static char *compile_subst __P((char *, struct s_subst *));
45 static char *compile_text __P((void));
46 static char *compile_tr __P((char *, char **));
47 static struct s_command
48 **compile_stream __P((char *, struct s_command **, char *));
49 static char *duptoeol __P((char *, char *));
50 static void enterlabel __P((struct s_command *));
51 static struct s_command
52 *findlabel __P((char *));
53 static void fixuplabel __P((struct s_command *, struct s_command *));
54 static void uselabel __P((void));
55
56 /*
57 * Command specification. This is used to drive the command parser.
58 */
59 struct s_format {
60 char code; /* Command code */
61 int naddr; /* Number of address args */
62 enum e_args args; /* Argument type */
63 };
64
65 static struct s_format cmd_fmts[] = {
66 {'{', 2, GROUP},
67 {'a', 1, TEXT},
68 {'b', 2, BRANCH},
69 {'c', 2, TEXT},
70 {'d', 2, EMPTY},
71 {'D', 2, EMPTY},
72 {'g', 2, EMPTY},
73 {'G', 2, EMPTY},
74 {'h', 2, EMPTY},
75 {'H', 2, EMPTY},
76 {'i', 1, TEXT},
77 {'l', 2, EMPTY},
78 {'n', 2, EMPTY},
79 {'N', 2, EMPTY},
80 {'p', 2, EMPTY},
81 {'P', 2, EMPTY},
82 {'q', 1, EMPTY},
83 {'r', 1, RFILE},
84 {'s', 2, SUBST},
85 {'t', 2, BRANCH},
86 {'w', 2, WFILE},
87 {'x', 2, EMPTY},
88 {'y', 2, TR},
89 {'!', 2, NONSEL},
90 {':', 0, LABEL},
91 {'#', 0, COMMENT},
92 {'=', 1, EMPTY},
93 {'\0', 0, COMMENT},
94 };
95
96 /* The compiled program. */
97 struct s_command *prog;
98
99 /*
100 * Compile the program into prog.
101 * Initialise appends.
102 */
103 void
compile()104 compile()
105 {
106 *compile_stream(NULL, &prog, NULL) = NULL;
107 fixuplabel(prog, NULL);
108 uselabel();
109 appends = xmalloc(sizeof(struct s_appends) * appendnum);
110 match = xmalloc((maxnsub + 1) * sizeof(regmatch_t));
111 }
112
113 #define EATSPACE() do { \
114 if (p) \
115 while (*p && isascii(*p) && isspace(*p)) \
116 p++; \
117 } while (0)
118
119 static struct s_command **
compile_stream(terminator,link,p)120 compile_stream(terminator, link, p)
121 char *terminator;
122 struct s_command **link;
123 register char *p;
124 {
125 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */
126 struct s_command *cmd, *cmd2;
127 struct s_format *fp;
128 int naddr; /* Number of addresses */
129
130 if (p != NULL)
131 goto semicolon;
132 for (;;) {
133 if ((p = cu_fgets(lbuf, sizeof(lbuf))) == NULL) {
134 if (terminator != NULL)
135 err(COMPILE, "unexpected EOF (pending }'s)");
136 return (link);
137 }
138
139 semicolon: EATSPACE();
140 if (p && (*p == '#' || *p == '\0'))
141 continue;
142 if (*p == '}') {
143 if (terminator == NULL)
144 err(COMPILE, "unexpected }");
145 return (link);
146 }
147 *link = cmd = xmalloc(sizeof(struct s_command));
148 link = &cmd->next;
149 cmd->nonsel = cmd->inrange = 0;
150 /* First parse the addresses */
151 naddr = 0;
152
153 /* Valid characters to start an address */
154 #define addrchar(c) (strchr("0123456789/\\$", (c)))
155 if (addrchar(*p)) {
156 naddr++;
157 cmd->a1 = xmalloc(sizeof(struct s_addr));
158 p = compile_addr(p, cmd->a1);
159 EATSPACE(); /* EXTENSION */
160 if (*p == ',') {
161 p++;
162 EATSPACE(); /* EXTENSION */
163 naddr++;
164 cmd->a2 = xmalloc(sizeof(struct s_addr));
165 p = compile_addr(p, cmd->a2);
166 EATSPACE();
167 } else
168 cmd->a2 = 0;
169 } else
170 cmd->a1 = cmd->a2 = 0;
171
172 nonsel: /* Now parse the command */
173 if (!*p)
174 err(COMPILE, "command expected");
175 cmd->code = *p;
176 for (fp = cmd_fmts; fp->code; fp++)
177 if (fp->code == *p)
178 break;
179 if (!fp->code)
180 err(COMPILE, "invalid command code %c", *p);
181 if (naddr > fp->naddr)
182 err(COMPILE,
183 "command %c expects up to %d address(es), found %d", *p, fp->naddr, naddr);
184 switch (fp->args) {
185 case NONSEL: /* ! */
186 p++;
187 EATSPACE();
188 cmd->nonsel = ! cmd->nonsel;
189 goto nonsel;
190 case GROUP: /* { */
191 p++;
192 EATSPACE();
193 if (!*p)
194 p = NULL;
195 cmd2 = xmalloc(sizeof(struct s_command));
196 cmd2->code = '}';
197 *compile_stream("}", &cmd->u.c, p) = cmd2;
198 cmd->next = cmd2;
199 link = &cmd2->next;
200 /*
201 * Short-circuit command processing, since end of
202 * group is really just a noop.
203 */
204 cmd2->nonsel = 1;
205 cmd2->a1 = cmd2->a2 = 0;
206 break;
207 case EMPTY: /* d D g G h H l n N p P q x = \0 */
208 p++;
209 EATSPACE();
210 if (*p == ';') {
211 p++;
212 link = &cmd->next;
213 goto semicolon;
214 }
215 if (*p)
216 err(COMPILE,
217 "extra characters at the end of %c command", cmd->code);
218 break;
219 case TEXT: /* a c i */
220 p++;
221 EATSPACE();
222 if (*p != '\\')
223 err(COMPILE,
224 "command %c expects \\ followed by text", cmd->code);
225 p++;
226 EATSPACE();
227 if (*p)
228 err(COMPILE,
229 "extra characters after \\ at the end of %c command", cmd->code);
230 cmd->t = compile_text();
231 break;
232 case COMMENT: /* \0 # */
233 break;
234 case WFILE: /* w */
235 p++;
236 EATSPACE();
237 if (*p == '\0')
238 err(COMPILE, "filename expected");
239 cmd->t = duptoeol(p, "w command");
240 if (aflag)
241 cmd->u.fd = -1;
242 else if ((cmd->u.fd = open(p,
243 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
244 DEFFILEMODE)) == -1)
245 err(FATAL, "%s: %s\n", p, strerror(errno));
246 break;
247 case RFILE: /* r */
248 p++;
249 EATSPACE();
250 if (*p == '\0')
251 err(COMPILE, "filename expected");
252 else
253 cmd->t = duptoeol(p, "read command");
254 break;
255 case BRANCH: /* b t */
256 p++;
257 EATSPACE();
258 if (*p == '\0')
259 cmd->t = NULL;
260 else
261 cmd->t = duptoeol(p, "branch");
262 break;
263 case LABEL: /* : */
264 p++;
265 EATSPACE();
266 cmd->t = duptoeol(p, "label");
267 if (strlen(p) == 0)
268 err(COMPILE, "empty label");
269 enterlabel(cmd);
270 break;
271 case SUBST: /* s */
272 p++;
273 if (*p == '\0' || *p == '\\')
274 err(COMPILE,
275 "substitute pattern can not be delimited by newline or backslash");
276 cmd->u.s = xmalloc(sizeof(struct s_subst));
277 p = compile_re(p, &cmd->u.s->re);
278 if (p == NULL)
279 err(COMPILE, "unterminated substitute pattern");
280 --p;
281 p = compile_subst(p, cmd->u.s);
282 p = compile_flags(p, cmd->u.s);
283 EATSPACE();
284 if (*p == ';') {
285 p++;
286 link = &cmd->next;
287 goto semicolon;
288 }
289 break;
290 case TR: /* y */
291 p++;
292 p = compile_tr(p, (char **)&cmd->u.y);
293 EATSPACE();
294 if (*p == ';') {
295 p++;
296 link = &cmd->next;
297 goto semicolon;
298 }
299 if (*p)
300 err(COMPILE,
301 "extra text at the end of a transform command");
302 break;
303 }
304 }
305 }
306
307 /*
308 * Get a delimited string. P points to the delimeter of the string; d points
309 * to a buffer area. Newline and delimiter escapes are processed; other
310 * escapes are ignored.
311 *
312 * Returns a pointer to the first character after the final delimiter or NULL
313 * in the case of a non-terminated string. The character array d is filled
314 * with the processed string.
315 */
316 static char *
compile_delimited(p,d)317 compile_delimited(p, d)
318 char *p, *d;
319 {
320 char c;
321
322 c = *p++;
323 if (c == '\0')
324 return (NULL);
325 else if (c == '\\')
326 err(COMPILE, "\\ can not be used as a string delimiter");
327 else if (c == '\n')
328 err(COMPILE, "newline can not be used as a string delimiter");
329 while (*p) {
330 if (*p == '\\' && p[1] == c)
331 p++;
332 else if (*p == '\\' && p[1] == 'n') {
333 *d++ = '\n';
334 p += 2;
335 continue;
336 } else if (*p == '\\' && p[1] == '\\')
337 *d++ = *p++;
338 else if (*p == c) {
339 *d = '\0';
340 return (p + 1);
341 }
342 *d++ = *p++;
343 }
344 return (NULL);
345 }
346
347 /*
348 * Get a regular expression. P points to the delimiter of the regular
349 * expression; repp points to the address of a regexp pointer. Newline
350 * and delimiter escapes are processed; other escapes are ignored.
351 * Returns a pointer to the first character after the final delimiter
352 * or NULL in the case of a non terminated regular expression. The regexp
353 * pointer is set to the compiled regular expression.
354 * Cflags are passed to regcomp.
355 */
356 static char *
compile_re(p,repp)357 compile_re(p, repp)
358 char *p;
359 regex_t **repp;
360 {
361 int eval;
362 char re[_POSIX2_LINE_MAX + 1];
363
364 p = compile_delimited(p, re);
365 if (p && strlen(re) == 0) {
366 *repp = NULL;
367 return (p);
368 }
369 *repp = xmalloc(sizeof(regex_t));
370 if (p && (eval = regcomp(*repp, re, 0)) != 0)
371 err(COMPILE, "RE error: %s", strregerror(eval, *repp));
372 if (maxnsub < (*repp)->re_nsub)
373 maxnsub = (*repp)->re_nsub;
374 return (p);
375 }
376
377 /*
378 * Compile the substitution string of a regular expression and set res to
379 * point to a saved copy of it. Nsub is the number of parenthesized regular
380 * expressions.
381 */
382 static char *
compile_subst(p,s)383 compile_subst(p, s)
384 char *p;
385 struct s_subst *s;
386 {
387 static char lbuf[_POSIX2_LINE_MAX + 1];
388 int asize, ref, size;
389 char c, *text, *op, *sp;
390
391 c = *p++; /* Terminator character */
392 if (c == '\0')
393 return (NULL);
394
395 s->maxbref = 0;
396 s->linenum = linenum;
397 asize = 2 * _POSIX2_LINE_MAX + 1;
398 text = xmalloc(asize);
399 size = 0;
400 do {
401 op = sp = text + size;
402 for (; *p; p++) {
403 if (*p == '\\') {
404 p++;
405 if (strchr("123456789", *p) != NULL) {
406 *sp++ = '\\';
407 ref = *p - '0';
408 if (s->re != NULL &&
409 ref > s->re->re_nsub)
410 err(COMPILE,
411 "\\%c not defined in the RE", *p);
412 if (s->maxbref < ref)
413 s->maxbref = ref;
414 } else if (*p == '&' || *p == '\\')
415 *sp++ = '\\';
416 } else if (*p == c) {
417 p++;
418 *sp++ = '\0';
419 size += sp - op;
420 s->new = xrealloc(text, size);
421 return (p);
422 } else if (*p == '\n') {
423 err(COMPILE,
424 "unescaped newline inside substitute pattern");
425 /* NOTREACHED */
426 }
427 *sp++ = *p;
428 }
429 size += sp - op;
430 if (asize - size < _POSIX2_LINE_MAX + 1) {
431 asize *= 2;
432 text = xmalloc(asize);
433 }
434 } while (cu_fgets(p = lbuf, sizeof(lbuf)));
435 err(COMPILE, "unterminated substitute in regular expression");
436 /* NOTREACHED */
437 }
438
439 /*
440 * Compile the flags of the s command
441 */
442 static char *
compile_flags(p,s)443 compile_flags(p, s)
444 char *p;
445 struct s_subst *s;
446 {
447 int gn; /* True if we have seen g or n */
448 char wfile[_POSIX2_LINE_MAX + 1], *q;
449
450 s->n = 1; /* Default */
451 s->p = 0;
452 s->wfile = NULL;
453 s->wfd = -1;
454 for (gn = 0;;) {
455 EATSPACE(); /* EXTENSION */
456 switch (*p) {
457 case 'g':
458 if (gn)
459 err(COMPILE,
460 "more than one number or 'g' in substitute flags");
461 gn = 1;
462 s->n = 0;
463 break;
464 case '\0':
465 case '\n':
466 case ';':
467 return (p);
468 case 'p':
469 s->p = 1;
470 break;
471 case '1': case '2': case '3':
472 case '4': case '5': case '6':
473 case '7': case '8': case '9':
474 if (gn)
475 err(COMPILE,
476 "more than one number or 'g' in substitute flags");
477 gn = 1;
478 /* XXX Check for overflow */
479 s->n = (int)strtol(p, &p, 10);
480 break;
481 case 'w':
482 p++;
483 #ifdef HISTORIC_PRACTICE
484 if (*p != ' ') {
485 err(WARNING, "space missing before w wfile");
486 return (p);
487 }
488 #endif
489 EATSPACE();
490 q = wfile;
491 while (*p) {
492 if (*p == '\n')
493 break;
494 *q++ = *p++;
495 }
496 *q = '\0';
497 if (q == wfile)
498 err(COMPILE, "no wfile specified");
499 s->wfile = strdup(wfile);
500 if (!aflag && (s->wfd = open(wfile,
501 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
502 DEFFILEMODE)) == -1)
503 err(FATAL, "%s: %s\n", wfile, strerror(errno));
504 return (p);
505 default:
506 err(COMPILE,
507 "bad flag in substitute command: '%c'", *p);
508 break;
509 }
510 p++;
511 }
512 }
513
514 /*
515 * Compile a translation set of strings into a lookup table.
516 */
517 static char *
compile_tr(p,transtab)518 compile_tr(p, transtab)
519 char *p;
520 char **transtab;
521 {
522 int i;
523 char *lt, *op, *np;
524 char old[_POSIX2_LINE_MAX + 1];
525 char new[_POSIX2_LINE_MAX + 1];
526
527 if (*p == '\0' || *p == '\\')
528 err(COMPILE,
529 "transform pattern can not be delimited by newline or backslash");
530 p = compile_delimited(p, old);
531 if (p == NULL) {
532 err(COMPILE, "unterminated transform source string");
533 return (NULL);
534 }
535 p = compile_delimited(--p, new);
536 if (p == NULL) {
537 err(COMPILE, "unterminated transform target string");
538 return (NULL);
539 }
540 EATSPACE();
541 if (strlen(new) != strlen(old)) {
542 err(COMPILE, "transform strings are not the same length");
543 return (NULL);
544 }
545 /* We assume characters are 8 bits */
546 lt = xmalloc(UCHAR_MAX);
547 for (i = 0; i <= UCHAR_MAX; i++)
548 lt[i] = (char)i;
549 for (op = old, np = new; *op; op++, np++)
550 lt[(u_char)*op] = *np;
551 *transtab = lt;
552 return (p);
553 }
554
555 /*
556 * Compile the text following an a or i command.
557 */
558 static char *
compile_text()559 compile_text()
560 {
561 int asize, size;
562 char *text, *p, *op, *s;
563 char lbuf[_POSIX2_LINE_MAX + 1];
564
565 asize = 2 * _POSIX2_LINE_MAX + 1;
566 text = xmalloc(asize);
567 size = 0;
568 while (cu_fgets(lbuf, sizeof(lbuf))) {
569 op = s = text + size;
570 p = lbuf;
571 EATSPACE();
572 for (; *p; p++) {
573 if (*p == '\\')
574 p++;
575 *s++ = *p;
576 }
577 size += s - op;
578 if (p[-2] != '\\') {
579 *s = '\0';
580 break;
581 }
582 if (asize - size < _POSIX2_LINE_MAX + 1) {
583 asize *= 2;
584 text = xmalloc(asize);
585 }
586 }
587 return (xrealloc(text, size + 1));
588 }
589
590 /*
591 * Get an address and return a pointer to the first character after
592 * it. Fill the structure pointed to according to the address.
593 */
594 static char *
compile_addr(p,a)595 compile_addr(p, a)
596 char *p;
597 struct s_addr *a;
598 {
599 char *end;
600
601 switch (*p) {
602 case '\\': /* Context address */
603 ++p;
604 /* FALLTHROUGH */
605 case '/': /* Context address */
606 p = compile_re(p, &a->u.r);
607 if (p == NULL)
608 err(COMPILE, "unterminated regular expression");
609 a->type = AT_RE;
610 return (p);
611
612 case '$': /* Last line */
613 a->type = AT_LAST;
614 return (p + 1);
615 /* Line number */
616 case '0': case '1': case '2': case '3': case '4':
617 case '5': case '6': case '7': case '8': case '9':
618 a->type = AT_LINE;
619 a->u.l = strtol(p, &end, 10);
620 return (end);
621 default:
622 err(COMPILE, "expected context address");
623 return (NULL);
624 }
625 }
626
627 /*
628 * duptoeol --
629 * Return a copy of all the characters up to \n or \0.
630 */
631 static char *
duptoeol(s,ctype)632 duptoeol(s, ctype)
633 register char *s;
634 char *ctype;
635 {
636 size_t len;
637 int ws;
638 char *start;
639
640 ws = 0;
641 for (start = s; *s != '\0' && *s != '\n'; ++s)
642 ws = isspace(*s);
643 *s = '\0';
644 if (ws)
645 err(WARNING, "whitespace after %s", ctype);
646 len = s - start + 1;
647 return (memmove(xmalloc(len), start, len));
648 }
649
650 /*
651 * Convert goto label names to addresses, and count a and r commands, in
652 * the given subset of the script. Free the memory used by labels in b
653 * and t commands (but not by :).
654 *
655 * TODO: Remove } nodes
656 */
657 static void
fixuplabel(cp,end)658 fixuplabel(cp, end)
659 struct s_command *cp, *end;
660 {
661
662 for (; cp != end; cp = cp->next)
663 switch (cp->code) {
664 case 'a':
665 case 'r':
666 appendnum++;
667 break;
668 case 'b':
669 case 't':
670 /* Resolve branch target. */
671 if (cp->t == NULL) {
672 cp->u.c = NULL;
673 break;
674 }
675 if ((cp->u.c = findlabel(cp->t)) == NULL)
676 err(COMPILE2, "undefined label '%s'", cp->t);
677 free(cp->t);
678 break;
679 case '{':
680 /* Do interior commands. */
681 fixuplabel(cp->u.c, cp->next);
682 break;
683 }
684 }
685
686 /*
687 * Associate the given command label for later lookup.
688 */
689 static void
enterlabel(cp)690 enterlabel(cp)
691 struct s_command *cp;
692 {
693 register struct labhash **lhp, *lh;
694 register u_char *p;
695 register u_int h, c;
696
697 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
698 h = (h << 5) + h + c;
699 lhp = &labels[h & LHMASK];
700 for (lh = *lhp; lh != NULL; lh = lh->lh_next)
701 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
702 err(COMPILE2, "duplicate label '%s'", cp->t);
703 lh = xmalloc(sizeof *lh);
704 lh->lh_next = *lhp;
705 lh->lh_hash = h;
706 lh->lh_cmd = cp;
707 lh->lh_ref = 0;
708 *lhp = lh;
709 }
710
711 /*
712 * Find the label contained in the command l in the command linked
713 * list cp. L is excluded from the search. Return NULL if not found.
714 */
715 static struct s_command *
findlabel(name)716 findlabel(name)
717 char *name;
718 {
719 register struct labhash *lh;
720 register u_char *p;
721 register u_int h, c;
722
723 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
724 h = (h << 5) + h + c;
725 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
726 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
727 lh->lh_ref = 1;
728 return (lh->lh_cmd);
729 }
730 }
731 return (NULL);
732 }
733
734 /*
735 * Warn about any unused labels. As a side effect, release the label hash
736 * table space.
737 */
738 static void
uselabel()739 uselabel()
740 {
741 register struct labhash *lh, *next;
742 register int i;
743
744 for (i = 0; i < LHSZ; i++) {
745 for (lh = labels[i]; lh != NULL; lh = next) {
746 next = lh->lh_next;
747 if (!lh->lh_ref)
748 err(WARNING, "unused label '%s'",
749 lh->lh_cmd->t);
750 free(lh);
751 }
752 }
753 }
754