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 * @(#)compile.c 8.1 (Berkeley) 6/6/93
36 * $FreeBSD: head/usr.bin/sed/compile.c 361884 2020-06-07 04:32:38Z kevans $
37 */
38
39 #include <sys/cdefs.h>
40
41 #include <sys/types.h>
42 #include <sys/stat.h>
43
44 #include <ctype.h>
45 #include <err.h>
46 #include <errno.h>
47 #include <fcntl.h>
48 #include <limits.h>
49 #include <regex.h>
50 #include <stdbool.h>
51 #include <stdio.h>
52 #include <stdlib.h>
53 #include <string.h>
54 #include <wchar.h>
55
56 #include "defs.h"
57 #include "extern.h"
58
59 #define LHSZ 128
60 #define LHMASK (LHSZ - 1)
61 static struct labhash {
62 struct labhash *lh_next;
63 u_int lh_hash;
64 struct s_command *lh_cmd;
65 int lh_ref;
66 } *labels[LHSZ];
67
68 static char *compile_addr(char *, struct s_addr *);
69 static char *compile_ccl(char **, char *);
70 static char *compile_delimited(char *, char *, int);
71 static char *compile_flags(char *, struct s_subst *);
72 static regex_t *compile_re(char *, int);
73 static char *compile_subst(char *, struct s_subst *);
74 static char *compile_text(void);
75 static char *compile_tr(char *, struct s_tr **);
76 static struct s_command
77 **compile_stream(struct s_command **);
78 static char *duptoeol(char *, const char *);
79 static void enterlabel(struct s_command *);
80 static struct s_command
81 *findlabel(char *);
82 static void fixuplabel(struct s_command *, struct s_command *);
83 static void uselabel(void);
84
85 /*
86 * Command specification. This is used to drive the command parser.
87 */
88 struct s_format {
89 char code; /* Command code */
90 int naddr; /* Number of address args */
91 enum e_args args; /* Argument type */
92 };
93
94 static struct s_format cmd_fmts[] = {
95 {'{', 2, GROUP},
96 {'}', 0, ENDGROUP},
97 {'a', 1, TEXT},
98 {'b', 2, BRANCH},
99 {'c', 2, TEXT},
100 {'d', 2, EMPTY},
101 {'D', 2, EMPTY},
102 {'g', 2, EMPTY},
103 {'G', 2, EMPTY},
104 {'h', 2, EMPTY},
105 {'H', 2, EMPTY},
106 {'i', 1, TEXT},
107 {'l', 2, EMPTY},
108 {'n', 2, EMPTY},
109 {'N', 2, EMPTY},
110 {'p', 2, EMPTY},
111 {'P', 2, EMPTY},
112 {'q', 1, EMPTY},
113 {'r', 1, RFILE},
114 {'s', 2, SUBST},
115 {'t', 2, BRANCH},
116 {'w', 2, WFILE},
117 {'x', 2, EMPTY},
118 {'y', 2, TR},
119 {'!', 2, NONSEL},
120 {':', 0, LABEL},
121 {'#', 0, COMMENT},
122 {'=', 1, EMPTY},
123 {'\0', 0, COMMENT},
124 };
125
126 /* The compiled program. */
127 struct s_command *prog;
128
129 /*
130 * Compile the program into prog.
131 * Initialise appends.
132 */
133 void
compile(void)134 compile(void)
135 {
136 *compile_stream(&prog) = NULL;
137 fixuplabel(prog, NULL);
138 uselabel();
139 if (appendnum == 0)
140 appends = NULL;
141 else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) ==
142 NULL)
143 err(1, "malloc");
144 if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL)
145 err(1, "malloc");
146 }
147
148 #define EATSPACE() do { \
149 if (p) \
150 while (*p && isspace((unsigned char)*p)) \
151 p++; \
152 } while (0)
153
154 static struct s_command **
compile_stream(struct s_command ** link)155 compile_stream(struct s_command **link)
156 {
157 char *p;
158 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */
159 struct s_command *cmd, *cmd2, *stack;
160 struct s_format *fp;
161 char re[_POSIX2_LINE_MAX + 1];
162 int naddr; /* Number of addresses */
163
164 stack = NULL;
165 for (;;) {
166 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) {
167 if (stack != NULL)
168 errx(1, "%lu: %s: unexpected EOF (pending }'s)",
169 linenum, fname);
170 return (link);
171 }
172
173 semicolon: EATSPACE();
174 if (p) {
175 if (*p == '#' || *p == '\0')
176 continue;
177 else if (*p == ';') {
178 p++;
179 goto semicolon;
180 }
181 }
182 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL)
183 err(1, "malloc");
184 link = &cmd->next;
185 cmd->startline = cmd->nonsel = 0;
186 /* First parse the addresses */
187 naddr = 0;
188
189 /* Valid characters to start an address */
190 #define addrchar(c) (strchr("0123456789/\\$", (c)))
191 if (addrchar(*p)) {
192 naddr++;
193 if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL)
194 err(1, "malloc");
195 p = compile_addr(p, cmd->a1);
196 EATSPACE(); /* EXTENSION */
197 if (*p == ',') {
198 p++;
199 EATSPACE(); /* EXTENSION */
200 naddr++;
201 if ((cmd->a2 = malloc(sizeof(struct s_addr)))
202 == NULL)
203 err(1, "malloc");
204 p = compile_addr(p, cmd->a2);
205 EATSPACE();
206 } else
207 cmd->a2 = NULL;
208 } else
209 cmd->a1 = cmd->a2 = NULL;
210
211 nonsel: /* Now parse the command */
212 if (!*p)
213 errx(1, "%lu: %s: command expected", linenum, fname);
214 cmd->code = *p;
215 for (fp = cmd_fmts; fp->code; fp++)
216 if (fp->code == *p)
217 break;
218 if (!fp->code)
219 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p);
220 if (naddr > fp->naddr)
221 errx(1,
222 "%lu: %s: command %c expects up to %d address(es), found %d",
223 linenum, fname, *p, fp->naddr, naddr);
224 switch (fp->args) {
225 case NONSEL: /* ! */
226 p++;
227 EATSPACE();
228 cmd->nonsel = 1;
229 goto nonsel;
230 case GROUP: /* { */
231 p++;
232 EATSPACE();
233 cmd->next = stack;
234 stack = cmd;
235 link = &cmd->u.c;
236 if (*p)
237 goto semicolon;
238 break;
239 case ENDGROUP:
240 /*
241 * Short-circuit command processing, since end of
242 * group is really just a noop.
243 */
244 cmd->nonsel = 1;
245 if (stack == NULL)
246 errx(1, "%lu: %s: unexpected }", linenum, fname);
247 cmd2 = stack;
248 stack = cmd2->next;
249 cmd2->next = cmd;
250 /*FALLTHROUGH*/
251 case EMPTY: /* d D g G h H l n N p P q x = \0 */
252 p++;
253 EATSPACE();
254 if (*p == ';') {
255 p++;
256 link = &cmd->next;
257 goto semicolon;
258 }
259 if (*p)
260 errx(1, "%lu: %s: extra characters at the end of %c command",
261 linenum, fname, cmd->code);
262 break;
263 case TEXT: /* a c i */
264 p++;
265 EATSPACE();
266 if (*p != '\\')
267 errx(1,
268 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code);
269 p++;
270 EATSPACE();
271 if (*p)
272 errx(1,
273 "%lu: %s: extra characters after \\ at the end of %c command",
274 linenum, fname, cmd->code);
275 cmd->t = compile_text();
276 break;
277 case COMMENT: /* \0 # */
278 break;
279 case WFILE: /* w */
280 p++;
281 EATSPACE();
282 if (*p == '\0')
283 errx(1, "%lu: %s: filename expected", linenum, fname);
284 cmd->t = duptoeol(p, "w command");
285 if (aflag)
286 cmd->u.fd = -1;
287 else if ((cmd->u.fd = open(p,
288 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
289 DEFFILEMODE)) == -1)
290 err(1, "%s", p);
291 break;
292 case RFILE: /* r */
293 p++;
294 EATSPACE();
295 if (*p == '\0')
296 errx(1, "%lu: %s: filename expected", linenum, fname);
297 else
298 cmd->t = duptoeol(p, "read command");
299 break;
300 case BRANCH: /* b t */
301 p++;
302 EATSPACE();
303 if (*p == '\0')
304 cmd->t = NULL;
305 else
306 cmd->t = duptoeol(p, "branch");
307 break;
308 case LABEL: /* : */
309 p++;
310 EATSPACE();
311 cmd->t = duptoeol(p, "label");
312 if (strlen(p) == 0)
313 errx(1, "%lu: %s: empty label", linenum, fname);
314 enterlabel(cmd);
315 break;
316 case SUBST: /* s */
317 p++;
318 if (*p == '\0' || *p == '\\')
319 errx(1,
320 "%lu: %s: substitute pattern can not be delimited by newline or backslash",
321 linenum, fname);
322 if ((cmd->u.s = calloc(1, sizeof(struct s_subst))) == NULL)
323 err(1, "malloc");
324 p = compile_delimited(p, re, 0);
325 if (p == NULL)
326 errx(1,
327 "%lu: %s: unterminated substitute pattern", linenum, fname);
328
329 /* Compile RE with no case sensitivity temporarily */
330 if (*re == '\0')
331 cmd->u.s->re = NULL;
332 else
333 cmd->u.s->re = compile_re(re, 0);
334 --p;
335 p = compile_subst(p, cmd->u.s);
336 p = compile_flags(p, cmd->u.s);
337
338 /* Recompile RE with case sensitivity from "I" flag if any */
339 if (*re == '\0')
340 cmd->u.s->re = NULL;
341 else
342 cmd->u.s->re = compile_re(re, cmd->u.s->icase);
343 EATSPACE();
344 if (*p == ';') {
345 p++;
346 link = &cmd->next;
347 goto semicolon;
348 }
349 break;
350 case TR: /* y */
351 p++;
352 p = compile_tr(p, &cmd->u.y);
353 EATSPACE();
354 if (*p == ';') {
355 p++;
356 link = &cmd->next;
357 goto semicolon;
358 }
359 if (*p)
360 errx(1,
361 "%lu: %s: extra text at the end of a transform command", linenum, fname);
362 break;
363 }
364 }
365 }
366
367 static int
hex2char(const char * in,char * out,int len)368 hex2char(const char *in, char *out, int len)
369 {
370 long ord;
371 char *endptr, hexbuf[3];
372
373 hexbuf[0] = in[0];
374 hexbuf[1] = len > 1 ? in[1] : '\0';
375 hexbuf[2] = '\0';
376
377 errno = 0;
378 ord = strtol(hexbuf, &endptr, 16);
379 if (*endptr != '\0' || errno != 0)
380 return (ERANGE);
381 *out = (char)ord;
382 return (0);
383 }
384
385 static bool
hexdigit(char c)386 hexdigit(char c)
387 {
388 int lc;
389
390 lc = tolower(c);
391 return isdigit(lc) || (lc >= 'a' && lc <= 'f');
392 }
393
394 static bool
dohex(const char * in,char * out,int * len)395 dohex(const char *in, char *out, int *len)
396 {
397 int tmplen;
398
399 if (!hexdigit(in[0]))
400 return (false);
401 tmplen = 1;
402 if (hexdigit(in[1]))
403 ++tmplen;
404 if (hex2char(in, out, tmplen) == 0) {
405 *len = tmplen;
406 return (true);
407 }
408
409 return (false);
410 }
411
412 /*
413 * Get a delimited string. P points to the delimiter of the string; d points
414 * to a buffer area. Newline and delimiter escapes are processed; other
415 * escapes are ignored.
416 *
417 * Returns a pointer to the first character after the final delimiter or NULL
418 * in the case of a non-terminated string. The character array d is filled
419 * with the processed string.
420 */
421 static char *
compile_delimited(char * p,char * d,int is_tr)422 compile_delimited(char *p, char *d, int is_tr)
423 {
424 int hexlen;
425 char c;
426
427 c = *p++;
428 if (c == '\0')
429 return (NULL);
430 else if (c == '\\')
431 errx(1, "%lu: %s: \\ can not be used as a string delimiter",
432 linenum, fname);
433 else if (c == '\n')
434 errx(1, "%lu: %s: newline can not be used as a string delimiter",
435 linenum, fname);
436 while (*p) {
437 if (*p == '[' && *p != c) {
438 if ((d = compile_ccl(&p, d)) == NULL)
439 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname);
440 continue;
441 } else if (*p == '\\' && p[1] == '[') {
442 *d++ = *p++;
443 } else if (*p == '\\' && p[1] == c) {
444 p++;
445 } else if (*p == '\\' &&
446 (p[1] == 'n' || p[1] == 'r' || p[1] == 't')) {
447 switch (p[1]) {
448 case 'n':
449 *d++ = '\n';
450 break;
451 case 'r':
452 *d++ = '\r';
453 break;
454 case 't':
455 *d++ = '\t';
456 break;
457 }
458 p += 2;
459 continue;
460 } else if (*p == '\\' && p[1] == 'x') {
461 if (dohex(&p[2], d, &hexlen)) {
462 ++d;
463 p += hexlen + 2;
464 continue;
465 }
466 } else if (*p == '\\' && p[1] == '\\') {
467 if (is_tr)
468 p++;
469 else
470 *d++ = *p++;
471 } else if (*p == c) {
472 *d = '\0';
473 return (p + 1);
474 }
475 *d++ = *p++;
476 }
477 return (NULL);
478 }
479
480
481 /* compile_ccl: expand a POSIX character class */
482 static char *
compile_ccl(char ** sp,char * t)483 compile_ccl(char **sp, char *t)
484 {
485 int c, d, hexlen;
486 char *s = *sp;
487
488 *t++ = *s++;
489 if (*s == '^')
490 *t++ = *s++;
491 if (*s == ']')
492 *t++ = *s++;
493 for (; *s && (*t = *s) != ']'; s++, t++) {
494 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
495 *++t = *++s, t++, s++;
496 for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
497 if ((c = *s) == '\0')
498 return NULL;
499 } else if (*s == '\\') {
500 switch (s[1]) {
501 case 'n':
502 *t = '\n';
503 s++;
504 break;
505 case 'r':
506 *t = '\r';
507 s++;
508 break;
509 case 't':
510 *t = '\t';
511 s++;
512 break;
513 case 'x':
514 if (dohex(&s[2], t, &hexlen))
515 s += hexlen + 1;
516 break;
517 }
518 }
519 }
520 return (*s == ']') ? *sp = ++s, ++t : NULL;
521 }
522
523 /*
524 * Compiles the regular expression in RE and returns a pointer to the compiled
525 * regular expression.
526 * Cflags are passed to regcomp.
527 */
528 static regex_t *
compile_re(char * re,int case_insensitive)529 compile_re(char *re, int case_insensitive)
530 {
531 regex_t *rep;
532 int eval, flags;
533
534
535 flags = rflags;
536 if (case_insensitive)
537 flags |= REG_ICASE;
538 if ((rep = malloc(sizeof(regex_t))) == NULL)
539 err(1, "malloc");
540 if ((eval = regcomp(rep, re, flags)) != 0)
541 errx(1, "%lu: %s: RE error: %s",
542 linenum, fname, strregerror(eval, rep));
543 if (maxnsub < rep->re_nsub)
544 maxnsub = rep->re_nsub;
545 return (rep);
546 }
547
548 /*
549 * Compile the substitution string of a regular expression and set res to
550 * point to a saved copy of it. Nsub is the number of parenthesized regular
551 * expressions.
552 */
553 static char *
compile_subst(char * p,struct s_subst * s)554 compile_subst(char *p, struct s_subst *s)
555 {
556 static char lbuf[_POSIX2_LINE_MAX + 1];
557 int asize, hexlen, size;
558 u_char ref;
559 char c, *text, *op, *sp;
560 int more = 1, sawesc = 0;
561
562 c = *p++; /* Terminator character */
563 if (c == '\0')
564 return (NULL);
565
566 s->maxbref = 0;
567 s->linenum = linenum;
568 asize = 2 * _POSIX2_LINE_MAX + 1;
569 if ((text = malloc(asize)) == NULL)
570 err(1, "malloc");
571 size = 0;
572 do {
573 op = sp = text + size;
574 for (; *p; p++) {
575 if (*p == '\\' || sawesc) {
576 /*
577 * If this is a continuation from the last
578 * buffer, we won't have a character to
579 * skip over.
580 */
581 if (sawesc)
582 sawesc = 0;
583 else
584 p++;
585
586 if (*p == '\0') {
587 /*
588 * This escaped character is continued
589 * in the next part of the line. Note
590 * this fact, then cause the loop to
591 * exit w/ normal EOL case and reenter
592 * above with the new buffer.
593 */
594 sawesc = 1;
595 p--;
596 continue;
597 } else if (strchr("123456789", *p) != NULL) {
598 *sp++ = '\\';
599 ref = *p - '0';
600 if (s->re != NULL &&
601 ref > s->re->re_nsub)
602 errx(1, "%lu: %s: \\%c not defined in the RE",
603 linenum, fname, *p);
604 if (s->maxbref < ref)
605 s->maxbref = ref;
606 } else {
607 switch (*p) {
608 case '&':
609 case '\\':
610 *sp++ = '\\';
611 break;
612 case 'n':
613 *p = '\n';
614 break;
615 case 'r':
616 *p = '\r';
617 break;
618 case 't':
619 *p = '\t';
620 break;
621 case 'x':
622 #define ADVANCE_N(s, n) \
623 do { \
624 char *adv = (s); \
625 while (*(adv + (n) - 1) != '\0') { \
626 *adv = *(adv + (n)); \
627 ++adv; \
628 } \
629 *adv = '\0'; \
630 } while (0);
631 if (dohex(&p[1], p, &hexlen)) {
632 ADVANCE_N(p + 1,
633 hexlen);
634 }
635 break;
636 }
637 }
638 } else if (*p == c) {
639 if (*++p == '\0' && more) {
640 if (cu_fgets(lbuf, sizeof(lbuf), &more))
641 p = lbuf;
642 }
643 *sp++ = '\0';
644 size += sp - op;
645 if ((s->new = realloc(text, size)) == NULL)
646 err(1, "realloc");
647 return (p);
648 } else if (*p == '\n') {
649 errx(1,
650 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname);
651 /* NOTREACHED */
652 }
653 *sp++ = *p;
654 }
655 size += sp - op;
656 if (asize - size < _POSIX2_LINE_MAX + 1) {
657 asize *= 2;
658 if ((text = realloc(text, asize)) == NULL)
659 err(1, "realloc");
660 }
661 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more) != NULL);
662 errx(1, "%lu: %s: unterminated substitute in regular expression",
663 linenum, fname);
664 /* NOTREACHED */
665 }
666
667 /*
668 * Compile the flags of the s command
669 */
670 static char *
compile_flags(char * p,struct s_subst * s)671 compile_flags(char *p, struct s_subst *s)
672 {
673 int gn; /* True if we have seen g or n */
674 unsigned long nval;
675 char wfile[_POSIX2_LINE_MAX + 1], *q, *eq;
676
677 s->n = 1; /* Default */
678 s->p = 0;
679 s->wfile = NULL;
680 s->wfd = -1;
681 s->icase = 0;
682 for (gn = 0;;) {
683 EATSPACE(); /* EXTENSION */
684 switch (*p) {
685 case 'g':
686 if (gn)
687 errx(1,
688 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
689 gn = 1;
690 s->n = 0;
691 break;
692 case '\0':
693 case '\n':
694 case ';':
695 return (p);
696 case 'p':
697 s->p = 1;
698 break;
699 case 'i':
700 case 'I':
701 s->icase = 1;
702 break;
703 case '1': case '2': case '3':
704 case '4': case '5': case '6':
705 case '7': case '8': case '9':
706 if (gn)
707 errx(1,
708 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
709 gn = 1;
710 errno = 0;
711 nval = strtol(p, &p, 10);
712 if (errno == ERANGE || nval > INT_MAX)
713 errx(1,
714 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname);
715 s->n = nval;
716 p--;
717 break;
718 case 'w':
719 p++;
720 #ifdef HISTORIC_PRACTICE
721 if (*p != ' ') {
722 warnx("%lu: %s: space missing before w wfile", linenum, fname);
723 return (p);
724 }
725 #endif
726 EATSPACE();
727 q = wfile;
728 eq = wfile + sizeof(wfile) - 1;
729 while (*p) {
730 if (*p == '\n')
731 break;
732 if (q >= eq)
733 err(1, "wfile too long");
734 *q++ = *p++;
735 }
736 *q = '\0';
737 if (q == wfile)
738 errx(1, "%lu: %s: no wfile specified", linenum, fname);
739 s->wfile = strdup(wfile);
740 if (!aflag && (s->wfd = open(wfile,
741 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
742 DEFFILEMODE)) == -1)
743 err(1, "%s", wfile);
744 return (p);
745 default:
746 errx(1, "%lu: %s: bad flag in substitute command: '%c'",
747 linenum, fname, *p);
748 break;
749 }
750 p++;
751 }
752 }
753
754 /*
755 * Compile a translation set of strings into a lookup table.
756 */
757 static char *
compile_tr(char * p,struct s_tr ** py)758 compile_tr(char *p, struct s_tr **py)
759 {
760 struct s_tr *y;
761 int i;
762 const char *op, *np;
763 char old[_POSIX2_LINE_MAX + 1];
764 char new[_POSIX2_LINE_MAX + 1];
765 size_t oclen, oldlen, nclen, newlen;
766 mbstate_t mbs1, mbs2;
767
768 if ((*py = y = malloc(sizeof(*y))) == NULL)
769 err(1, NULL);
770 y->multis = NULL;
771 y->nmultis = 0;
772
773 if (*p == '\0' || *p == '\\')
774 errx(1,
775 "%lu: %s: transform pattern can not be delimited by newline or backslash",
776 linenum, fname);
777 p = compile_delimited(p, old, 1);
778 if (p == NULL)
779 errx(1, "%lu: %s: unterminated transform source string",
780 linenum, fname);
781 p = compile_delimited(p - 1, new, 1);
782 if (p == NULL)
783 errx(1, "%lu: %s: unterminated transform target string",
784 linenum, fname);
785 EATSPACE();
786 op = old;
787 oldlen = mbsrtowcs(NULL, &op, 0, NULL);
788 if (oldlen == (size_t)-1)
789 err(1, NULL);
790 np = new;
791 newlen = mbsrtowcs(NULL, &np, 0, NULL);
792 if (newlen == (size_t)-1)
793 err(1, NULL);
794 if (newlen != oldlen)
795 errx(1, "%lu: %s: transform strings are not the same length",
796 linenum, fname);
797 if (MB_CUR_MAX == 1) {
798 /*
799 * The single-byte encoding case is easy: generate a
800 * lookup table.
801 */
802 for (i = 0; i <= UCHAR_MAX; i++)
803 y->bytetab[i] = (char)i;
804 for (; *op; op++, np++)
805 y->bytetab[(u_char)*op] = *np;
806 } else {
807 /*
808 * Multi-byte encoding case: generate a lookup table as
809 * above, but only for single-byte characters. The first
810 * bytes of multi-byte characters have their lookup table
811 * entries set to 0, which causes do_tr() to search through
812 * an auxiliary vector of multi-byte mappings.
813 */
814 memset(&mbs1, 0, sizeof(mbs1));
815 memset(&mbs2, 0, sizeof(mbs2));
816 for (i = 0; i <= UCHAR_MAX; i++)
817 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0;
818 while (*op != '\0') {
819 oclen = mbrlen(op, MB_LEN_MAX, &mbs1);
820 if (oclen == (size_t)-1 || oclen == (size_t)-2)
821 errc(1, EILSEQ, NULL);
822 nclen = mbrlen(np, MB_LEN_MAX, &mbs2);
823 if (nclen == (size_t)-1 || nclen == (size_t)-2)
824 errc(1, EILSEQ, NULL);
825 if (oclen == 1 && nclen == 1)
826 y->bytetab[(u_char)*op] = *np;
827 else {
828 y->bytetab[(u_char)*op] = 0;
829 y->multis = realloc(y->multis,
830 (y->nmultis + 1) * sizeof(*y->multis));
831 if (y->multis == NULL)
832 err(1, NULL);
833 i = y->nmultis++;
834 y->multis[i].fromlen = oclen;
835 memcpy(y->multis[i].from, op, oclen);
836 y->multis[i].tolen = nclen;
837 memcpy(y->multis[i].to, np, nclen);
838 }
839 op += oclen;
840 np += nclen;
841 }
842 }
843 return (p);
844 }
845
846 /*
847 * Compile the text following an a, c, or i command.
848 */
849 static char *
compile_text(void)850 compile_text(void)
851 {
852 int asize, esc_nl, size;
853 char *text, *p, *op, *s;
854 char lbuf[_POSIX2_LINE_MAX + 1];
855
856 asize = 2 * _POSIX2_LINE_MAX + 1;
857 if ((text = malloc(asize)) == NULL)
858 err(1, "malloc");
859 size = 0;
860 while (cu_fgets(lbuf, sizeof(lbuf), NULL) != NULL) {
861 op = s = text + size;
862 p = lbuf;
863 #ifdef LEGACY_BSDSED_COMPAT
864 EATSPACE();
865 #endif
866 for (esc_nl = 0; *p != '\0'; p++) {
867 if (*p == '\\' && p[1] != '\0' && *++p == '\n')
868 esc_nl = 1;
869 *s++ = *p;
870 }
871 size += s - op;
872 if (!esc_nl) {
873 *s = '\0';
874 break;
875 }
876 if (asize - size < _POSIX2_LINE_MAX + 1) {
877 asize *= 2;
878 if ((text = realloc(text, asize)) == NULL)
879 err(1, "realloc");
880 }
881 }
882 text[size] = '\0';
883 if ((p = realloc(text, size + 1)) == NULL)
884 err(1, "realloc");
885 return (p);
886 }
887
888 /*
889 * Get an address and return a pointer to the first character after
890 * it. Fill the structure pointed to according to the address.
891 */
892 static char *
compile_addr(char * p,struct s_addr * a)893 compile_addr(char *p, struct s_addr *a)
894 {
895 char *end, re[_POSIX2_LINE_MAX + 1];
896 int icase;
897
898 icase = 0;
899
900 a->type = 0;
901 switch (*p) {
902 case '\\': /* Context address */
903 ++p;
904 /* FALLTHROUGH */
905 case '/': /* Context address */
906 p = compile_delimited(p, re, 0);
907 if (p == NULL)
908 errx(1, "%lu: %s: unterminated regular expression", linenum, fname);
909 /* Check for case insensitive regexp flag */
910 if (*p == 'I') {
911 icase = 1;
912 p++;
913 }
914 if (*re == '\0')
915 a->u.r = NULL;
916 else
917 a->u.r = compile_re(re, icase);
918 a->type = AT_RE;
919 return (p);
920
921 case '$': /* Last line */
922 a->type = AT_LAST;
923 return (p + 1);
924
925 case '+': /* Relative line number */
926 a->type = AT_RELLINE;
927 p++;
928 /* FALLTHROUGH */
929 /* Line number */
930 case '0': case '1': case '2': case '3': case '4':
931 case '5': case '6': case '7': case '8': case '9':
932 if (a->type == 0)
933 a->type = AT_LINE;
934 a->u.l = strtol(p, &end, 10);
935 return (end);
936 default:
937 errx(1, "%lu: %s: expected context address", linenum, fname);
938 return (NULL);
939 }
940 }
941
942 /*
943 * duptoeol --
944 * Return a copy of all the characters up to \n or \0.
945 */
946 static char *
duptoeol(char * s,const char * ctype)947 duptoeol(char *s, const char *ctype)
948 {
949 size_t len;
950 int ws;
951 char *p, *start;
952
953 ws = 0;
954 for (start = s; *s != '\0' && *s != '\n'; ++s)
955 ws = isspace((unsigned char)*s);
956 *s = '\0';
957 if (ws)
958 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype);
959 len = s - start + 1;
960 if ((p = malloc(len)) == NULL)
961 err(1, "malloc");
962 return (memmove(p, start, len));
963 }
964
965 /*
966 * Convert goto label names to addresses, and count a and r commands, in
967 * the given subset of the script. Free the memory used by labels in b
968 * and t commands (but not by :).
969 *
970 * TODO: Remove } nodes
971 */
972 static void
fixuplabel(struct s_command * cp,struct s_command * end)973 fixuplabel(struct s_command *cp, struct s_command *end)
974 {
975
976 for (; cp != end; cp = cp->next)
977 switch (cp->code) {
978 case 'a':
979 case 'r':
980 appendnum++;
981 break;
982 case 'b':
983 case 't':
984 /* Resolve branch target. */
985 if (cp->t == NULL) {
986 cp->u.c = NULL;
987 break;
988 }
989 if ((cp->u.c = findlabel(cp->t)) == NULL)
990 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t);
991 free(cp->t);
992 break;
993 case '{':
994 /* Do interior commands. */
995 fixuplabel(cp->u.c, cp->next);
996 break;
997 }
998 }
999
1000 /*
1001 * Associate the given command label for later lookup.
1002 */
1003 static void
enterlabel(struct s_command * cp)1004 enterlabel(struct s_command *cp)
1005 {
1006 struct labhash **lhp, *lh;
1007 u_char *p;
1008 u_int h, c;
1009
1010 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
1011 h = (h << 5) + h + c;
1012 lhp = &labels[h & LHMASK];
1013 for (lh = *lhp; lh != NULL; lh = lh->lh_next)
1014 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
1015 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t);
1016 if ((lh = malloc(sizeof *lh)) == NULL)
1017 err(1, "malloc");
1018 lh->lh_next = *lhp;
1019 lh->lh_hash = h;
1020 lh->lh_cmd = cp;
1021 lh->lh_ref = 0;
1022 *lhp = lh;
1023 }
1024
1025 /*
1026 * Find the label contained in the command l in the command linked
1027 * list cp. L is excluded from the search. Return NULL if not found.
1028 */
1029 static struct s_command *
findlabel(char * name)1030 findlabel(char *name)
1031 {
1032 struct labhash *lh;
1033 u_char *p;
1034 u_int h, c;
1035
1036 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
1037 h = (h << 5) + h + c;
1038 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
1039 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
1040 lh->lh_ref = 1;
1041 return (lh->lh_cmd);
1042 }
1043 }
1044 return (NULL);
1045 }
1046
1047 /*
1048 * Warn about any unused labels. As a side effect, release the label hash
1049 * table space.
1050 */
1051 static void
uselabel(void)1052 uselabel(void)
1053 {
1054 struct labhash *lh, *next;
1055 int i;
1056
1057 for (i = 0; i < LHSZ; i++) {
1058 for (lh = labels[i]; lh != NULL; lh = next) {
1059 next = lh->lh_next;
1060 if (!lh->lh_ref)
1061 warnx("%lu: %s: unused label '%s'",
1062 linenum, fname, lh->lh_cmd->t);
1063 free(lh);
1064 }
1065 }
1066 }
1067