1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1992 Diomidis Spinellis.
5 * Copyright (c) 1992, 1993
6 * The Regents of the University of California. All rights reserved.
7 *
8 * This code is derived from software contributed to Berkeley by
9 * Diomidis Spinellis of Imperial College, University of London.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 */
35
36 #include <sys/cdefs.h>
37 __FBSDID("$FreeBSD$");
38
39 #ifndef lint
40 static const char sccsid[] = "@(#)compile.c 8.1 (Berkeley) 6/6/93";
41 #endif
42
43 #include <sys/types.h>
44 #include <sys/stat.h>
45
46 #include <ctype.h>
47 #include <err.h>
48 #include <errno.h>
49 #include <fcntl.h>
50 #include <limits.h>
51 #include <regex.h>
52 #include <stdbool.h>
53 #include <stdio.h>
54 #include <stdlib.h>
55 #include <string.h>
56 #include <wchar.h>
57
58 #include "defs.h"
59 #include "extern.h"
60
61 #define LHSZ 128
62 #define LHMASK (LHSZ - 1)
63 static struct labhash {
64 struct labhash *lh_next;
65 u_int lh_hash;
66 struct s_command *lh_cmd;
67 int lh_ref;
68 } *labels[LHSZ];
69
70 static char *compile_addr(char *, struct s_addr *);
71 static char *compile_ccl(char **, char *);
72 static char *compile_delimited(char *, char *, int);
73 static char *compile_flags(char *, struct s_subst *);
74 static regex_t *compile_re(char *, int);
75 static char *compile_subst(char *, struct s_subst *);
76 static char *compile_text(void);
77 static char *compile_tr(char *, struct s_tr **);
78 static struct s_command
79 **compile_stream(struct s_command **);
80 static char *duptoeol(char *, const char *);
81 static void enterlabel(struct s_command *);
82 static struct s_command
83 *findlabel(char *);
84 static void fixuplabel(struct s_command *, struct s_command *);
85 static void uselabel(void);
86
87 /*
88 * Command specification. This is used to drive the command parser.
89 */
90 struct s_format {
91 char code; /* Command code */
92 int naddr; /* Number of address args */
93 enum e_args args; /* Argument type */
94 };
95
96 static struct s_format cmd_fmts[] = {
97 {'{', 2, GROUP},
98 {'}', 0, ENDGROUP},
99 {'a', 1, TEXT},
100 {'b', 2, BRANCH},
101 {'c', 2, TEXT},
102 {'d', 2, EMPTY},
103 {'D', 2, EMPTY},
104 {'g', 2, EMPTY},
105 {'G', 2, EMPTY},
106 {'h', 2, EMPTY},
107 {'H', 2, EMPTY},
108 {'i', 1, TEXT},
109 {'l', 2, EMPTY},
110 {'n', 2, EMPTY},
111 {'N', 2, EMPTY},
112 {'p', 2, EMPTY},
113 {'P', 2, EMPTY},
114 {'q', 1, EMPTY},
115 {'r', 1, RFILE},
116 {'s', 2, SUBST},
117 {'t', 2, BRANCH},
118 {'w', 2, WFILE},
119 {'x', 2, EMPTY},
120 {'y', 2, TR},
121 {'!', 2, NONSEL},
122 {':', 0, LABEL},
123 {'#', 0, COMMENT},
124 {'=', 1, EMPTY},
125 {'\0', 0, COMMENT},
126 };
127
128 /* The compiled program. */
129 struct s_command *prog;
130
131 /*
132 * Compile the program into prog.
133 * Initialise appends.
134 */
135 void
compile(void)136 compile(void)
137 {
138 *compile_stream(&prog) = NULL;
139 fixuplabel(prog, NULL);
140 uselabel();
141 if (appendnum == 0)
142 appends = NULL;
143 else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) ==
144 NULL)
145 err(1, "malloc");
146 if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL)
147 err(1, "malloc");
148 }
149
150 #define EATSPACE() do { \
151 if (p) \
152 while (*p && isspace((unsigned char)*p)) \
153 p++; \
154 } while (0)
155
156 static struct s_command **
compile_stream(struct s_command ** link)157 compile_stream(struct s_command **link)
158 {
159 char *p;
160 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */
161 struct s_command *cmd, *cmd2, *stack;
162 struct s_format *fp;
163 char re[_POSIX2_LINE_MAX + 1];
164 int naddr; /* Number of addresses */
165
166 stack = NULL;
167 for (;;) {
168 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) {
169 if (stack != NULL)
170 errx(1, "%lu: %s: unexpected EOF (pending }'s)",
171 linenum, fname);
172 return (link);
173 }
174
175 semicolon: EATSPACE();
176 if (p) {
177 if (*p == '#' || *p == '\0')
178 continue;
179 else if (*p == ';') {
180 p++;
181 goto semicolon;
182 }
183 }
184 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL)
185 err(1, "malloc");
186 link = &cmd->next;
187 cmd->startline = cmd->nonsel = 0;
188 /* First parse the addresses */
189 naddr = 0;
190
191 /* Valid characters to start an address */
192 #define addrchar(c) (strchr("0123456789/\\$", (c)))
193 if (addrchar(*p)) {
194 naddr++;
195 if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL)
196 err(1, "malloc");
197 p = compile_addr(p, cmd->a1);
198 EATSPACE(); /* EXTENSION */
199 if (*p == ',') {
200 p++;
201 EATSPACE(); /* EXTENSION */
202 naddr++;
203 if ((cmd->a2 = malloc(sizeof(struct s_addr)))
204 == NULL)
205 err(1, "malloc");
206 p = compile_addr(p, cmd->a2);
207 EATSPACE();
208 } else
209 cmd->a2 = NULL;
210 } else
211 cmd->a1 = cmd->a2 = NULL;
212
213 nonsel: /* Now parse the command */
214 if (!*p)
215 errx(1, "%lu: %s: command expected", linenum, fname);
216 cmd->code = *p;
217 for (fp = cmd_fmts; fp->code; fp++)
218 if (fp->code == *p)
219 break;
220 if (!fp->code)
221 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p);
222 if (naddr > fp->naddr)
223 errx(1,
224 "%lu: %s: command %c expects up to %d address(es), found %d",
225 linenum, fname, *p, fp->naddr, naddr);
226 switch (fp->args) {
227 case NONSEL: /* ! */
228 p++;
229 EATSPACE();
230 cmd->nonsel = 1;
231 goto nonsel;
232 case GROUP: /* { */
233 p++;
234 EATSPACE();
235 cmd->next = stack;
236 stack = cmd;
237 link = &cmd->u.c;
238 if (*p)
239 goto semicolon;
240 break;
241 case ENDGROUP:
242 /*
243 * Short-circuit command processing, since end of
244 * group is really just a noop.
245 */
246 cmd->nonsel = 1;
247 if (stack == NULL)
248 errx(1, "%lu: %s: unexpected }", linenum, fname);
249 cmd2 = stack;
250 stack = cmd2->next;
251 cmd2->next = cmd;
252 /*FALLTHROUGH*/
253 case EMPTY: /* d D g G h H l n N p P q x = \0 */
254 p++;
255 EATSPACE();
256 if (*p == ';') {
257 p++;
258 link = &cmd->next;
259 goto semicolon;
260 }
261 if (*p)
262 errx(1, "%lu: %s: extra characters at the end of %c command",
263 linenum, fname, cmd->code);
264 break;
265 case TEXT: /* a c i */
266 p++;
267 EATSPACE();
268 if (*p != '\\')
269 errx(1,
270 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code);
271 p++;
272 EATSPACE();
273 if (*p)
274 errx(1,
275 "%lu: %s: extra characters after \\ at the end of %c command",
276 linenum, fname, cmd->code);
277 cmd->t = compile_text();
278 break;
279 case COMMENT: /* \0 # */
280 break;
281 case WFILE: /* w */
282 p++;
283 EATSPACE();
284 if (*p == '\0')
285 errx(1, "%lu: %s: filename expected", linenum, fname);
286 cmd->t = duptoeol(p, "w command");
287 if (aflag)
288 cmd->u.fd = -1;
289 else if ((cmd->u.fd = open(p,
290 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
291 DEFFILEMODE)) == -1)
292 err(1, "%s", p);
293 break;
294 case RFILE: /* r */
295 p++;
296 EATSPACE();
297 if (*p == '\0')
298 errx(1, "%lu: %s: filename expected", linenum, fname);
299 else
300 cmd->t = duptoeol(p, "read command");
301 break;
302 case BRANCH: /* b t */
303 p++;
304 EATSPACE();
305 if (*p == '\0')
306 cmd->t = NULL;
307 else
308 cmd->t = duptoeol(p, "branch");
309 break;
310 case LABEL: /* : */
311 p++;
312 EATSPACE();
313 cmd->t = duptoeol(p, "label");
314 if (strlen(p) == 0)
315 errx(1, "%lu: %s: empty label", linenum, fname);
316 enterlabel(cmd);
317 break;
318 case SUBST: /* s */
319 p++;
320 if (*p == '\0' || *p == '\\')
321 errx(1,
322 "%lu: %s: substitute pattern can not be delimited by newline or backslash",
323 linenum, fname);
324 if ((cmd->u.s = calloc(1, sizeof(struct s_subst))) == NULL)
325 err(1, "malloc");
326 p = compile_delimited(p, re, 0);
327 if (p == NULL)
328 errx(1,
329 "%lu: %s: unterminated substitute pattern", linenum, fname);
330
331 /* Compile RE with no case sensitivity temporarily */
332 if (*re == '\0')
333 cmd->u.s->re = NULL;
334 else
335 cmd->u.s->re = compile_re(re, 0);
336 --p;
337 p = compile_subst(p, cmd->u.s);
338 p = compile_flags(p, cmd->u.s);
339
340 /* Recompile RE with case sensitivity from "I" flag if any */
341 if (*re == '\0')
342 cmd->u.s->re = NULL;
343 else
344 cmd->u.s->re = compile_re(re, cmd->u.s->icase);
345 EATSPACE();
346 if (*p == ';') {
347 p++;
348 link = &cmd->next;
349 goto semicolon;
350 }
351 break;
352 case TR: /* y */
353 p++;
354 p = compile_tr(p, &cmd->u.y);
355 EATSPACE();
356 if (*p == ';') {
357 p++;
358 link = &cmd->next;
359 goto semicolon;
360 }
361 if (*p)
362 errx(1,
363 "%lu: %s: extra text at the end of a transform command", linenum, fname);
364 break;
365 }
366 }
367 }
368
369 static int
hex2char(const char * in,char * out,int len)370 hex2char(const char *in, char *out, int len)
371 {
372 long ord;
373 char *endptr, hexbuf[3];
374
375 hexbuf[0] = in[0];
376 hexbuf[1] = len > 1 ? in[1] : '\0';
377 hexbuf[2] = '\0';
378
379 errno = 0;
380 ord = strtol(hexbuf, &endptr, 16);
381 if (*endptr != '\0' || errno != 0)
382 return (ERANGE);
383 *out = (char)ord;
384 return (0);
385 }
386
387 static bool
hexdigit(char c)388 hexdigit(char c)
389 {
390 int lc;
391
392 lc = tolower(c);
393 return isdigit(lc) || (lc >= 'a' && lc <= 'f');
394 }
395
396 static bool
dohex(const char * in,char * out,int * len)397 dohex(const char *in, char *out, int *len)
398 {
399 int tmplen;
400
401 if (!hexdigit(in[0]))
402 return (false);
403 tmplen = 1;
404 if (hexdigit(in[1]))
405 ++tmplen;
406 if (hex2char(in, out, tmplen) == 0) {
407 *len = tmplen;
408 return (true);
409 }
410
411 return (false);
412 }
413
414 /*
415 * Get a delimited string. P points to the delimiter of the string; d points
416 * to a buffer area. Newline and delimiter escapes are processed; other
417 * escapes are ignored.
418 *
419 * Returns a pointer to the first character after the final delimiter or NULL
420 * in the case of a non-terminated string. The character array d is filled
421 * with the processed string.
422 */
423 static char *
compile_delimited(char * p,char * d,int is_tr)424 compile_delimited(char *p, char *d, int is_tr)
425 {
426 int hexlen;
427 char c;
428
429 c = *p++;
430 if (c == '\0')
431 return (NULL);
432 else if (c == '\\')
433 errx(1, "%lu: %s: \\ can not be used as a string delimiter",
434 linenum, fname);
435 else if (c == '\n')
436 errx(1, "%lu: %s: newline can not be used as a string delimiter",
437 linenum, fname);
438 while (*p) {
439 if (*p == '[' && *p != c) {
440 if (!is_tr) {
441 if ((d = compile_ccl(&p, d)) == NULL) {
442 errx(1,
443 "%lu: %s: unbalanced brackets ([])",
444 linenum, fname);
445 }
446 continue;
447 }
448 } else if (*p == '\\' && p[1] == '[') {
449 if (is_tr)
450 p++;
451 else
452 *d++ = *p++;
453 } else if (*p == '\\' && p[1] == c) {
454 p++;
455 } else if (*p == '\\' &&
456 (p[1] == 'n' || p[1] == 'r' || p[1] == 't')) {
457 switch (p[1]) {
458 case 'n':
459 *d++ = '\n';
460 break;
461 case 'r':
462 *d++ = '\r';
463 break;
464 case 't':
465 *d++ = '\t';
466 break;
467 }
468 p += 2;
469 continue;
470 } else if (*p == '\\' && p[1] == 'x') {
471 if (dohex(&p[2], d, &hexlen)) {
472 ++d;
473 p += hexlen + 2;
474 continue;
475 }
476 } else if (*p == '\\' && p[1] == '\\') {
477 if (is_tr)
478 p++;
479 else
480 *d++ = *p++;
481 } else if (*p == c) {
482 *d = '\0';
483 return (p + 1);
484 }
485 *d++ = *p++;
486 }
487 return (NULL);
488 }
489
490
491 /* compile_ccl: expand a POSIX character class */
492 static char *
compile_ccl(char ** sp,char * t)493 compile_ccl(char **sp, char *t)
494 {
495 int c, d, hexlen;
496 char *s = *sp;
497
498 *t++ = *s++;
499 if (*s == '^')
500 *t++ = *s++;
501 if (*s == ']')
502 *t++ = *s++;
503 for (; *s && (*t = *s) != ']'; s++, t++) {
504 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
505 *++t = *++s, t++, s++;
506 for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
507 if ((c = *s) == '\0')
508 return NULL;
509 } else if (*s == '\\') {
510 switch (s[1]) {
511 case 'n':
512 *t = '\n';
513 s++;
514 break;
515 case 'r':
516 *t = '\r';
517 s++;
518 break;
519 case 't':
520 *t = '\t';
521 s++;
522 break;
523 case 'x':
524 if (dohex(&s[2], t, &hexlen))
525 s += hexlen + 1;
526 break;
527 }
528 }
529 }
530 return (*s == ']') ? *sp = ++s, ++t : NULL;
531 }
532
533 /*
534 * Compiles the regular expression in RE and returns a pointer to the compiled
535 * regular expression.
536 * Cflags are passed to regcomp.
537 */
538 static regex_t *
compile_re(char * re,int case_insensitive)539 compile_re(char *re, int case_insensitive)
540 {
541 regex_t *rep;
542 int eval, flags;
543
544
545 flags = rflags;
546 if (case_insensitive)
547 flags |= REG_ICASE;
548 if ((rep = malloc(sizeof(regex_t))) == NULL)
549 err(1, "malloc");
550 if ((eval = regcomp(rep, re, flags)) != 0)
551 errx(1, "%lu: %s: RE error: %s",
552 linenum, fname, strregerror(eval, rep));
553 if (maxnsub < rep->re_nsub)
554 maxnsub = rep->re_nsub;
555 return (rep);
556 }
557
558 /*
559 * Compile the substitution string of a regular expression and set res to
560 * point to a saved copy of it. Nsub is the number of parenthesized regular
561 * expressions.
562 */
563 static char *
compile_subst(char * p,struct s_subst * s)564 compile_subst(char *p, struct s_subst *s)
565 {
566 static char lbuf[_POSIX2_LINE_MAX + 1];
567 int asize, hexlen, size;
568 u_char ref;
569 char c, *text, *op, *sp;
570 int more = 1, sawesc = 0;
571
572 c = *p++; /* Terminator character */
573 if (c == '\0')
574 return (NULL);
575
576 s->maxbref = 0;
577 s->linenum = linenum;
578 asize = 2 * _POSIX2_LINE_MAX + 1;
579 if ((text = malloc(asize)) == NULL)
580 err(1, "malloc");
581 size = 0;
582 do {
583 op = sp = text + size;
584 for (; *p; p++) {
585 if (*p == '\\' || sawesc) {
586 /*
587 * If this is a continuation from the last
588 * buffer, we won't have a character to
589 * skip over.
590 */
591 if (sawesc)
592 sawesc = 0;
593 else
594 p++;
595
596 if (*p == '\0') {
597 /*
598 * This escaped character is continued
599 * in the next part of the line. Note
600 * this fact, then cause the loop to
601 * exit w/ normal EOL case and reenter
602 * above with the new buffer.
603 */
604 sawesc = 1;
605 p--;
606 continue;
607 } else if (strchr("123456789", *p) != NULL) {
608 *sp++ = '\\';
609 ref = *p - '0';
610 if (s->re != NULL &&
611 ref > s->re->re_nsub)
612 errx(1, "%lu: %s: \\%c not defined in the RE",
613 linenum, fname, *p);
614 if (s->maxbref < ref)
615 s->maxbref = ref;
616 } else {
617 switch (*p) {
618 case '&':
619 case '\\':
620 *sp++ = '\\';
621 break;
622 case 'n':
623 *p = '\n';
624 break;
625 case 'r':
626 *p = '\r';
627 break;
628 case 't':
629 *p = '\t';
630 break;
631 case 'x':
632 #define ADVANCE_N(s, n) \
633 do { \
634 char *adv = (s); \
635 while (*(adv + (n) - 1) != '\0') { \
636 *adv = *(adv + (n)); \
637 ++adv; \
638 } \
639 *adv = '\0'; \
640 } while (0);
641 if (dohex(&p[1], p, &hexlen)) {
642 ADVANCE_N(p + 1,
643 hexlen);
644 }
645 break;
646 }
647 }
648 } else if (*p == c) {
649 if (*++p == '\0' && more) {
650 if (cu_fgets(lbuf, sizeof(lbuf), &more))
651 p = lbuf;
652 }
653 *sp++ = '\0';
654 size += sp - op;
655 if ((s->new = realloc(text, size)) == NULL)
656 err(1, "realloc");
657 return (p);
658 } else if (*p == '\n') {
659 errx(1,
660 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname);
661 /* NOTREACHED */
662 }
663 *sp++ = *p;
664 }
665 size += sp - op;
666 if (asize - size < _POSIX2_LINE_MAX + 1) {
667 asize *= 2;
668 if ((text = realloc(text, asize)) == NULL)
669 err(1, "realloc");
670 }
671 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more) != NULL);
672 errx(1, "%lu: %s: unterminated substitute in regular expression",
673 linenum, fname);
674 /* NOTREACHED */
675 }
676
677 /*
678 * Compile the flags of the s command
679 */
680 static char *
compile_flags(char * p,struct s_subst * s)681 compile_flags(char *p, struct s_subst *s)
682 {
683 int gn; /* True if we have seen g or n */
684 unsigned long nval;
685 char wfile[_POSIX2_LINE_MAX + 1], *q, *eq;
686
687 s->n = 1; /* Default */
688 s->p = 0;
689 s->wfile = NULL;
690 s->wfd = -1;
691 s->icase = 0;
692 for (gn = 0;;) {
693 EATSPACE(); /* EXTENSION */
694 switch (*p) {
695 case 'g':
696 if (gn)
697 errx(1,
698 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
699 gn = 1;
700 s->n = 0;
701 break;
702 case '\0':
703 case '\n':
704 case ';':
705 return (p);
706 case 'p':
707 s->p = 1;
708 break;
709 case 'i':
710 case 'I':
711 s->icase = 1;
712 break;
713 case '1': case '2': case '3':
714 case '4': case '5': case '6':
715 case '7': case '8': case '9':
716 if (gn)
717 errx(1,
718 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
719 gn = 1;
720 errno = 0;
721 nval = strtol(p, &p, 10);
722 if (errno == ERANGE || nval > INT_MAX)
723 errx(1,
724 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname);
725 s->n = nval;
726 p--;
727 break;
728 case 'w':
729 p++;
730 #ifdef HISTORIC_PRACTICE
731 if (*p != ' ') {
732 warnx("%lu: %s: space missing before w wfile", linenum, fname);
733 return (p);
734 }
735 #endif
736 EATSPACE();
737 q = wfile;
738 eq = wfile + sizeof(wfile) - 1;
739 while (*p) {
740 if (*p == '\n')
741 break;
742 if (q >= eq)
743 err(1, "wfile too long");
744 *q++ = *p++;
745 }
746 *q = '\0';
747 if (q == wfile)
748 errx(1, "%lu: %s: no wfile specified", linenum, fname);
749 s->wfile = strdup(wfile);
750 if (!aflag && (s->wfd = open(wfile,
751 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
752 DEFFILEMODE)) == -1)
753 err(1, "%s", wfile);
754 return (p);
755 default:
756 errx(1, "%lu: %s: bad flag in substitute command: '%c'",
757 linenum, fname, *p);
758 break;
759 }
760 p++;
761 }
762 }
763
764 /*
765 * Compile a translation set of strings into a lookup table.
766 */
767 static char *
compile_tr(char * p,struct s_tr ** py)768 compile_tr(char *p, struct s_tr **py)
769 {
770 struct s_tr *y;
771 int i;
772 const char *op, *np;
773 char old[_POSIX2_LINE_MAX + 1];
774 char new[_POSIX2_LINE_MAX + 1];
775 size_t oclen, oldlen, nclen, newlen;
776 mbstate_t mbs1, mbs2;
777
778 if ((*py = y = malloc(sizeof(*y))) == NULL)
779 err(1, NULL);
780 y->multis = NULL;
781 y->nmultis = 0;
782
783 if (*p == '\0' || *p == '\\')
784 errx(1,
785 "%lu: %s: transform pattern can not be delimited by newline or backslash",
786 linenum, fname);
787 p = compile_delimited(p, old, 1);
788 if (p == NULL)
789 errx(1, "%lu: %s: unterminated transform source string",
790 linenum, fname);
791 p = compile_delimited(p - 1, new, 1);
792 if (p == NULL)
793 errx(1, "%lu: %s: unterminated transform target string",
794 linenum, fname);
795 EATSPACE();
796 op = old;
797 oldlen = mbsrtowcs(NULL, &op, 0, NULL);
798 if (oldlen == (size_t)-1)
799 err(1, NULL);
800 np = new;
801 newlen = mbsrtowcs(NULL, &np, 0, NULL);
802 if (newlen == (size_t)-1)
803 err(1, NULL);
804 if (newlen != oldlen)
805 errx(1, "%lu: %s: transform strings are not the same length",
806 linenum, fname);
807 if (MB_CUR_MAX == 1) {
808 /*
809 * The single-byte encoding case is easy: generate a
810 * lookup table.
811 */
812 for (i = 0; i <= UCHAR_MAX; i++)
813 y->bytetab[i] = (char)i;
814 for (; *op; op++, np++)
815 y->bytetab[(u_char)*op] = *np;
816 } else {
817 /*
818 * Multi-byte encoding case: generate a lookup table as
819 * above, but only for single-byte characters. The first
820 * bytes of multi-byte characters have their lookup table
821 * entries set to 0, which causes do_tr() to search through
822 * an auxiliary vector of multi-byte mappings.
823 */
824 memset(&mbs1, 0, sizeof(mbs1));
825 memset(&mbs2, 0, sizeof(mbs2));
826 for (i = 0; i <= UCHAR_MAX; i++)
827 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0;
828 while (*op != '\0') {
829 oclen = mbrlen(op, MB_LEN_MAX, &mbs1);
830 if (oclen == (size_t)-1 || oclen == (size_t)-2)
831 errc(1, EILSEQ, NULL);
832 nclen = mbrlen(np, MB_LEN_MAX, &mbs2);
833 if (nclen == (size_t)-1 || nclen == (size_t)-2)
834 errc(1, EILSEQ, NULL);
835 if (oclen == 1 && nclen == 1)
836 y->bytetab[(u_char)*op] = *np;
837 else {
838 y->bytetab[(u_char)*op] = 0;
839 y->multis = realloc(y->multis,
840 (y->nmultis + 1) * sizeof(*y->multis));
841 if (y->multis == NULL)
842 err(1, NULL);
843 i = y->nmultis++;
844 y->multis[i].fromlen = oclen;
845 memcpy(y->multis[i].from, op, oclen);
846 y->multis[i].tolen = nclen;
847 memcpy(y->multis[i].to, np, nclen);
848 }
849 op += oclen;
850 np += nclen;
851 }
852 }
853 return (p);
854 }
855
856 /*
857 * Compile the text following an a, c, or i command.
858 */
859 static char *
compile_text(void)860 compile_text(void)
861 {
862 int asize, esc_nl, size;
863 char *text, *p, *op, *s;
864 char lbuf[_POSIX2_LINE_MAX + 1];
865
866 asize = 2 * _POSIX2_LINE_MAX + 1;
867 if ((text = malloc(asize)) == NULL)
868 err(1, "malloc");
869 size = 0;
870 while (cu_fgets(lbuf, sizeof(lbuf), NULL) != NULL) {
871 op = s = text + size;
872 p = lbuf;
873 #ifdef LEGACY_BSDSED_COMPAT
874 EATSPACE();
875 #endif
876 for (esc_nl = 0; *p != '\0'; p++) {
877 if (*p == '\\' && p[1] != '\0' && *++p == '\n')
878 esc_nl = 1;
879 *s++ = *p;
880 }
881 size += s - op;
882 if (!esc_nl) {
883 *s = '\0';
884 break;
885 }
886 if (asize - size < _POSIX2_LINE_MAX + 1) {
887 asize *= 2;
888 if ((text = realloc(text, asize)) == NULL)
889 err(1, "realloc");
890 }
891 }
892 text[size] = '\0';
893 if ((p = realloc(text, size + 1)) == NULL)
894 err(1, "realloc");
895 return (p);
896 }
897
898 /*
899 * Get an address and return a pointer to the first character after
900 * it. Fill the structure pointed to according to the address.
901 */
902 static char *
compile_addr(char * p,struct s_addr * a)903 compile_addr(char *p, struct s_addr *a)
904 {
905 char *end, re[_POSIX2_LINE_MAX + 1];
906 int icase;
907
908 icase = 0;
909
910 a->type = 0;
911 switch (*p) {
912 case '\\': /* Context address */
913 ++p;
914 /* FALLTHROUGH */
915 case '/': /* Context address */
916 p = compile_delimited(p, re, 0);
917 if (p == NULL)
918 errx(1, "%lu: %s: unterminated regular expression", linenum, fname);
919 /* Check for case insensitive regexp flag */
920 if (*p == 'I') {
921 icase = 1;
922 p++;
923 }
924 if (*re == '\0')
925 a->u.r = NULL;
926 else
927 a->u.r = compile_re(re, icase);
928 a->type = AT_RE;
929 return (p);
930
931 case '$': /* Last line */
932 a->type = AT_LAST;
933 return (p + 1);
934
935 case '+': /* Relative line number */
936 a->type = AT_RELLINE;
937 p++;
938 /* FALLTHROUGH */
939 /* Line number */
940 case '0': case '1': case '2': case '3': case '4':
941 case '5': case '6': case '7': case '8': case '9':
942 if (a->type == 0)
943 a->type = AT_LINE;
944 a->u.l = strtol(p, &end, 10);
945 return (end);
946 default:
947 errx(1, "%lu: %s: expected context address", linenum, fname);
948 return (NULL);
949 }
950 }
951
952 /*
953 * duptoeol --
954 * Return a copy of all the characters up to \n or \0.
955 */
956 static char *
duptoeol(char * s,const char * ctype)957 duptoeol(char *s, const char *ctype)
958 {
959 size_t len;
960 int ws;
961 char *p, *start;
962
963 ws = 0;
964 for (start = s; *s != '\0' && *s != '\n'; ++s)
965 ws = isspace((unsigned char)*s);
966 *s = '\0';
967 if (ws)
968 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype);
969 len = s - start + 1;
970 if ((p = malloc(len)) == NULL)
971 err(1, "malloc");
972 return (memmove(p, start, len));
973 }
974
975 /*
976 * Convert goto label names to addresses, and count a and r commands, in
977 * the given subset of the script. Free the memory used by labels in b
978 * and t commands (but not by :).
979 *
980 * TODO: Remove } nodes
981 */
982 static void
fixuplabel(struct s_command * cp,struct s_command * end)983 fixuplabel(struct s_command *cp, struct s_command *end)
984 {
985
986 for (; cp != end; cp = cp->next)
987 switch (cp->code) {
988 case 'a':
989 case 'r':
990 appendnum++;
991 break;
992 case 'b':
993 case 't':
994 /* Resolve branch target. */
995 if (cp->t == NULL) {
996 cp->u.c = NULL;
997 break;
998 }
999 if ((cp->u.c = findlabel(cp->t)) == NULL)
1000 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t);
1001 free(cp->t);
1002 break;
1003 case '{':
1004 /* Do interior commands. */
1005 fixuplabel(cp->u.c, cp->next);
1006 break;
1007 }
1008 }
1009
1010 /*
1011 * Associate the given command label for later lookup.
1012 */
1013 static void
enterlabel(struct s_command * cp)1014 enterlabel(struct s_command *cp)
1015 {
1016 struct labhash **lhp, *lh;
1017 u_char *p;
1018 u_int h, c;
1019
1020 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
1021 h = (h << 5) + h + c;
1022 lhp = &labels[h & LHMASK];
1023 for (lh = *lhp; lh != NULL; lh = lh->lh_next)
1024 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
1025 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t);
1026 if ((lh = malloc(sizeof *lh)) == NULL)
1027 err(1, "malloc");
1028 lh->lh_next = *lhp;
1029 lh->lh_hash = h;
1030 lh->lh_cmd = cp;
1031 lh->lh_ref = 0;
1032 *lhp = lh;
1033 }
1034
1035 /*
1036 * Find the label contained in the command l in the command linked
1037 * list cp. L is excluded from the search. Return NULL if not found.
1038 */
1039 static struct s_command *
findlabel(char * name)1040 findlabel(char *name)
1041 {
1042 struct labhash *lh;
1043 u_char *p;
1044 u_int h, c;
1045
1046 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
1047 h = (h << 5) + h + c;
1048 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
1049 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
1050 lh->lh_ref = 1;
1051 return (lh->lh_cmd);
1052 }
1053 }
1054 return (NULL);
1055 }
1056
1057 /*
1058 * Warn about any unused labels. As a side effect, release the label hash
1059 * table space.
1060 */
1061 static void
uselabel(void)1062 uselabel(void)
1063 {
1064 struct labhash *lh, *next;
1065 int i;
1066
1067 for (i = 0; i < LHSZ; i++) {
1068 for (lh = labels[i]; lh != NULL; lh = next) {
1069 next = lh->lh_next;
1070 if (!lh->lh_ref)
1071 warnx("%lu: %s: unused label '%s'",
1072 linenum, fname, lh->lh_cmd->t);
1073 free(lh);
1074 }
1075 }
1076 }
1077