1 /* $OpenBSD: process.c,v 1.39 2024/12/10 23:49:55 millert Exp $ */
2
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/types.h>
37 #include <sys/stat.h>
38 #include <sys/uio.h>
39
40 #include <ctype.h>
41 #include <err.h>
42 #include <fcntl.h>
43 #include <limits.h>
44 #include <regex.h>
45 #include <stdio.h>
46 #include <stdlib.h>
47 #include <string.h>
48 #include <unistd.h>
49
50 #include "defs.h"
51 #include "extern.h"
52
53 static SPACE HS, PS, SS;
54 #define pd PS.deleted
55 #define ps PS.space
56 #define psl PS.len
57 #define psanl PS.append_newline
58 #define hs HS.space
59 #define hsl HS.len
60
61 static inline int applies(struct s_command *);
62 static void flush_appends(void);
63 static void lputs(char *, size_t);
64 static inline int regexec_e(regex_t *, const char *, int, int, size_t,
65 size_t);
66 static void regsub(SPACE *, char *, char *);
67 static int substitute(struct s_command *);
68
69 struct s_appends *appends; /* Array of pointers to strings to append. */
70 static size_t appendx; /* Index into appends array. */
71 size_t appendnum; /* Size of appends array. */
72
73 static int lastaddr; /* Set by applies if last address of a range. */
74 static int sdone; /* If any substitutes since last line input. */
75 /* Iov structure for 'w' commands. */
76 static regex_t *defpreg;
77 size_t maxnsub;
78 regmatch_t *match;
79
80 #define OUT() do {\
81 fwrite(ps, 1, psl, outfile);\
82 if (psanl) fputc('\n', outfile);\
83 } while (0)
84
85 void
process(void)86 process(void)
87 {
88 struct s_command *cp;
89 SPACE tspace;
90 size_t len, oldpsl;
91 char *p;
92
93 for (linenum = 0; mf_getline(&PS, REPLACE);) {
94 pd = 0;
95 top:
96 cp = prog;
97 redirect:
98 while (cp != NULL) {
99 if (!applies(cp)) {
100 cp = cp->next;
101 continue;
102 }
103 switch (cp->code) {
104 case '{':
105 cp = cp->u.c;
106 goto redirect;
107 case 'a':
108 if (appendx >= appendnum) {
109 appends = xreallocarray(appends,
110 appendnum,
111 2 * sizeof(struct s_appends));
112 appendnum *= 2;
113 }
114 appends[appendx].type = AP_STRING;
115 appends[appendx].s = cp->t;
116 appends[appendx].len = strlen(cp->t);
117 appendx++;
118 break;
119 case 'b':
120 cp = cp->u.c;
121 goto redirect;
122 case 'c':
123 pd = 1;
124 psl = 0;
125 if (cp->a2 == NULL || lastaddr || lastline())
126 (void)fprintf(outfile, "%s", cp->t);
127 goto new;
128 case 'd':
129 pd = 1;
130 goto new;
131 case 'D':
132 if (pd)
133 goto new;
134 if (psl == 0 ||
135 (p = memchr(ps, '\n', psl)) == NULL) {
136 pd = 1;
137 goto new;
138 } else {
139 psl -= (p + 1) - ps;
140 memmove(ps, p + 1, psl);
141 goto top;
142 }
143 case 'g':
144 cspace(&PS, hs, hsl, REPLACE);
145 break;
146 case 'G':
147 cspace(&PS, "\n", 1, 0);
148 cspace(&PS, hs, hsl, 0);
149 break;
150 case 'h':
151 cspace(&HS, ps, psl, REPLACE);
152 break;
153 case 'H':
154 cspace(&HS, "\n", 1, 0);
155 cspace(&HS, ps, psl, 0);
156 break;
157 case 'i':
158 (void)fprintf(outfile, "%s", cp->t);
159 break;
160 case 'l':
161 lputs(ps, psl);
162 break;
163 case 'n':
164 if (!nflag && !pd)
165 OUT();
166 flush_appends();
167 if (!mf_getline(&PS, REPLACE))
168 exit(0);
169 pd = 0;
170 break;
171 case 'N':
172 flush_appends();
173 cspace(&PS, "\n", 1, 0);
174 if (!mf_getline(&PS, 0))
175 exit(0);
176 break;
177 case 'p':
178 if (pd)
179 break;
180 OUT();
181 break;
182 case 'P':
183 if (pd)
184 break;
185 if ((p = memchr(ps, '\n', psl)) != NULL) {
186 oldpsl = psl;
187 psl = p - ps;
188 psanl = 1;
189 OUT();
190 psl = oldpsl;
191 } else {
192 OUT();
193 }
194 break;
195 case 'q':
196 if (!nflag && !pd)
197 OUT();
198 flush_appends();
199 finish_file();
200 exit(0);
201 case 'r':
202 if (appendx >= appendnum) {
203 appends = xreallocarray(appends,
204 appendnum,
205 2 * sizeof(struct s_appends));
206 appendnum *= 2;
207 }
208 appends[appendx].type = AP_FILE;
209 appends[appendx].s = cp->t;
210 appends[appendx].len = strlen(cp->t);
211 appendx++;
212 break;
213 case 's':
214 sdone |= substitute(cp);
215 break;
216 case 't':
217 if (sdone) {
218 sdone = 0;
219 cp = cp->u.c;
220 goto redirect;
221 }
222 break;
223 case 'w':
224 if (pd)
225 break;
226 if (cp->u.fd == -1 && (cp->u.fd = open(cp->t,
227 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
228 DEFFILEMODE)) == -1)
229 err(1, "%s", cp->t);
230 if ((size_t)write(cp->u.fd, ps, psl) != psl ||
231 write(cp->u.fd, "\n", 1) != 1)
232 err(1, "%s", cp->t);
233 break;
234 case 'x':
235 if (hs == NULL)
236 cspace(&HS, "", 0, REPLACE);
237 tspace = PS;
238 PS = HS;
239 psanl = tspace.append_newline;
240 HS = tspace;
241 break;
242 case 'y':
243 if (pd || psl == 0)
244 break;
245 for (p = ps, len = psl; len--; ++p)
246 *p = cp->u.y[(unsigned char)*p];
247 break;
248 case ':':
249 case '}':
250 break;
251 case '=':
252 (void)fprintf(outfile, "%lu\n", linenum);
253 }
254 cp = cp->next;
255 } /* for all cp */
256
257 new: if (!nflag && !pd)
258 OUT();
259 flush_appends();
260 } /* for all lines */
261 }
262
263 /*
264 * TRUE if the address passed matches the current program state
265 * (lastline, linenumber, ps).
266 */
267 #define MATCH(a) \
268 (a)->type == AT_RE ? regexec_e((a)->u.r, ps, 0, 1, 0, psl) : \
269 (a)->type == AT_LINE ? linenum == (a)->u.l : lastline()
270
271 /*
272 * Return TRUE if the command applies to the current line. Sets the inrange
273 * flag to process ranges. Interprets the non-select (``!'') flag.
274 */
275 static inline int
applies(struct s_command * cp)276 applies(struct s_command *cp)
277 {
278 int r;
279
280 lastaddr = 0;
281 if (cp->a1 == NULL && cp->a2 == NULL)
282 r = 1;
283 else if (cp->a2)
284 if (cp->inrange) {
285 if (MATCH(cp->a2)) {
286 cp->inrange = 0;
287 lastaddr = 1;
288 }
289 r = 1;
290 } else if (MATCH(cp->a1)) {
291 /*
292 * If the second address is a number less than or
293 * equal to the line number first selected, only
294 * one line shall be selected.
295 * -- POSIX 1003.2
296 */
297 if (cp->a2->type == AT_LINE &&
298 linenum >= cp->a2->u.l)
299 lastaddr = 1;
300 else
301 cp->inrange = 1;
302 r = 1;
303 } else
304 r = 0;
305 else
306 r = MATCH(cp->a1);
307 return (cp->nonsel ? !r : r);
308 }
309
310 /*
311 * Reset all inrange markers.
312 */
313 void
resetstate(void)314 resetstate(void)
315 {
316 struct s_command *cp;
317
318 free(HS.back);
319 memset(&HS, 0, sizeof(HS));
320
321 for (cp = prog; cp; cp = cp->code == '{' ? cp->u.c : cp->next)
322 if (cp->a2)
323 cp->inrange = 0;
324 }
325
326 /*
327 * substitute --
328 * Do substitutions in the pattern space. Currently, we build a
329 * copy of the new pattern space in the substitute space structure
330 * and then swap them.
331 */
332 static int
substitute(struct s_command * cp)333 substitute(struct s_command *cp)
334 {
335 SPACE tspace;
336 regex_t *re;
337 regoff_t slen;
338 int n, lastempty;
339 regoff_t le = 0;
340 char *s;
341
342 s = ps;
343 re = cp->u.s->re;
344 if (re == NULL) {
345 if (defpreg != NULL && cp->u.s->maxbref > defpreg->re_nsub) {
346 linenum = cp->u.s->linenum;
347 error("\\%d not defined in the RE", cp->u.s->maxbref);
348 }
349 }
350 if (!regexec_e(re, ps, 0, 0, 0, psl))
351 return (0);
352
353 SS.len = 0; /* Clean substitute space. */
354 slen = psl;
355 n = cp->u.s->n;
356 lastempty = 1;
357
358 do {
359 /* Copy the leading retained string. */
360 if (n <= 1 && (match[0].rm_so > le))
361 cspace(&SS, s, match[0].rm_so - le, APPEND);
362
363 /* Skip zero-length matches right after other matches. */
364 if (lastempty || (match[0].rm_so - le) ||
365 match[0].rm_so != match[0].rm_eo) {
366 if (n <= 1) {
367 /* Want this match: append replacement. */
368 regsub(&SS, ps, cp->u.s->new);
369 if (n == 1)
370 n = -1;
371 } else {
372 /* Want a later match: append original. */
373 if (match[0].rm_eo - le)
374 cspace(&SS, s, match[0].rm_eo - le,
375 APPEND);
376 n--;
377 }
378 }
379
380 /* Move past this match. */
381 s = ps + match[0].rm_eo;
382 slen = psl - match[0].rm_eo;
383 le = match[0].rm_eo;
384
385 /*
386 * After a zero-length match, advance one byte,
387 * and at the end of the line, terminate.
388 */
389 if (match[0].rm_so == match[0].rm_eo) {
390 if (slen > 0) {
391 cspace(&SS, s++, 1, APPEND);
392 slen--;
393 le++;
394 } else
395 slen = -1;
396 lastempty = 1;
397 } else
398 lastempty = 0;
399
400 } while (n >= 0 && slen >= 0 &&
401 regexec_e(re, ps, REG_NOTBOL, 0, le, psl));
402
403 /* Did not find the requested number of matches. */
404 if (n > 0)
405 return (0);
406
407 /* Copy the trailing retained string. */
408 if (slen > 0)
409 cspace(&SS, s, slen, APPEND);
410
411 /*
412 * Swap the substitute space and the pattern space, and make sure
413 * that any leftover pointers into stdio memory get lost.
414 */
415 tspace = PS;
416 PS = SS;
417 psanl = tspace.append_newline;
418 SS = tspace;
419 SS.space = SS.back;
420
421 /* Handle the 'p' flag. */
422 if (cp->u.s->p)
423 OUT();
424
425 /* Handle the 'w' flag. */
426 if (cp->u.s->wfile && !pd) {
427 if (cp->u.s->wfd == -1 && (cp->u.s->wfd = open(cp->u.s->wfile,
428 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, DEFFILEMODE)) == -1)
429 err(1, "%s", cp->u.s->wfile);
430 if ((size_t)write(cp->u.s->wfd, ps, psl) != psl ||
431 write(cp->u.s->wfd, "\n", 1) != 1)
432 err(1, "%s", cp->u.s->wfile);
433 }
434 return (1);
435 }
436
437 /*
438 * Flush append requests. Always called before reading a line,
439 * therefore it also resets the substitution done (sdone) flag.
440 */
441 static void
flush_appends(void)442 flush_appends(void)
443 {
444 FILE *f;
445 size_t count, idx;
446 char buf[8 * 1024];
447
448 for (idx = 0; idx < appendx; idx++)
449 switch (appends[idx].type) {
450 case AP_STRING:
451 fwrite(appends[idx].s, sizeof(char), appends[idx].len,
452 outfile);
453 break;
454 case AP_FILE:
455 /*
456 * Read files probably shouldn't be cached. Since
457 * it's not an error to read a non-existent file,
458 * it's possible that another program is interacting
459 * with the sed script through the file system. It
460 * would be truly bizarre, but possible. It's probably
461 * not that big a performance win, anyhow.
462 */
463 if ((f = fopen(appends[idx].s, "r")) == NULL)
464 break;
465 while ((count = fread(buf, sizeof(char), sizeof(buf), f)))
466 (void)fwrite(buf, sizeof(char), count, outfile);
467 (void)fclose(f);
468 break;
469 }
470 if (ferror(outfile))
471 err(1, "%s", outfname);
472 appendx = sdone = 0;
473 }
474
475 static void
lputs(char * s,size_t len)476 lputs(char *s, size_t len)
477 {
478 int count;
479 extern int termwidth;
480 const char *escapes;
481 char *p;
482
483 for (count = 0; len > 0; len--, s++) {
484 if (count >= termwidth) {
485 (void)fprintf(outfile, "\\\n");
486 count = 0;
487 }
488 if (isascii((unsigned char)*s) && isprint((unsigned char)*s)
489 && *s != '\\') {
490 (void)fputc(*s, outfile);
491 count++;
492 } else if (*s == '\n') {
493 (void)fputc('$', outfile);
494 (void)fputc('\n', outfile);
495 count = 0;
496 } else {
497 escapes = "\\\a\b\f\r\t\v";
498 (void)fputc('\\', outfile);
499 if ((p = strchr(escapes, *s)) && *s != '\0') {
500 (void)fputc("\\abfrtv"[p - escapes], outfile);
501 count += 2;
502 } else {
503 (void)fprintf(outfile, "%03o", *(u_char *)s);
504 count += 4;
505 }
506 }
507 }
508 (void)fputc('$', outfile);
509 (void)fputc('\n', outfile);
510 if (ferror(outfile))
511 err(1, "%s", outfname);
512 }
513
514 static inline int
regexec_e(regex_t * preg,const char * string,int eflags,int nomatch,size_t start,size_t stop)515 regexec_e(regex_t *preg, const char *string, int eflags,
516 int nomatch, size_t start, size_t stop)
517 {
518 int eval;
519
520 if (preg == NULL) {
521 if (defpreg == NULL)
522 errx(1, "first RE may not be empty");
523 } else
524 defpreg = preg;
525
526 /* Set anchors */
527 match[0].rm_so = start;
528 match[0].rm_eo = stop;
529
530 eval = regexec(defpreg, string,
531 nomatch ? 0 : maxnsub + 1, match, eflags | REG_STARTEND);
532 switch (eval) {
533 case 0:
534 return (1);
535 case REG_NOMATCH:
536 return (0);
537 }
538 errx(1, "RE error: %s", strregerror(eval, defpreg));
539 }
540
541 /*
542 * regsub - perform substitutions after a regexp match
543 * Based on a routine by Henry Spencer
544 */
545 static void
regsub(SPACE * sp,char * string,char * src)546 regsub(SPACE *sp, char *string, char *src)
547 {
548 int len, no;
549 char c, *dst;
550
551 #define NEEDSP(reqlen) \
552 if (sp->len + (reqlen) + 1 >= sp->blen) { \
553 size_t newlen = sp->blen + (reqlen) + 1024; \
554 sp->space = sp->back = xrealloc(sp->back, newlen); \
555 sp->blen = newlen; \
556 dst = sp->space + sp->len; \
557 }
558
559 dst = sp->space + sp->len;
560 while ((c = *src++) != '\0') {
561 if (c == '&')
562 no = 0;
563 else if (c == '\\' && isdigit((unsigned char)*src))
564 no = *src++ - '0';
565 else
566 no = -1;
567 if (no < 0) { /* Ordinary character. */
568 if (c == '\\' && (*src == '\\' || *src == '&'))
569 c = *src++;
570 NEEDSP(1);
571 *dst++ = c;
572 ++sp->len;
573 } else if (match[no].rm_so != -1 && match[no].rm_eo != -1) {
574 len = match[no].rm_eo - match[no].rm_so;
575 NEEDSP(len);
576 memmove(dst, string + match[no].rm_so, len);
577 dst += len;
578 sp->len += len;
579 }
580 }
581 NEEDSP(1);
582 *dst = '\0';
583 }
584
585 /*
586 * aspace --
587 * Append the source space to the destination space, allocating new
588 * space as necessary.
589 */
590 void
cspace(SPACE * sp,const char * p,size_t len,enum e_spflag spflag)591 cspace(SPACE *sp, const char *p, size_t len, enum e_spflag spflag)
592 {
593 size_t tlen;
594
595 /* Make sure SPACE has enough memory and ramp up quickly. */
596 tlen = sp->len + len + 1;
597 if (tlen > sp->blen) {
598 size_t newlen = tlen + 1024;
599 sp->space = sp->back = xrealloc(sp->back, newlen);
600 sp->blen = newlen;
601 }
602
603 if (spflag == REPLACE)
604 sp->len = 0;
605
606 memmove(sp->space + sp->len, p, len);
607
608 sp->space[sp->len += len] = '\0';
609 }
610
611 /*
612 * Close all cached opened files and report any errors
613 */
614 void
cfclose(struct s_command * cp,struct s_command * end)615 cfclose(struct s_command *cp, struct s_command *end)
616 {
617
618 for (; cp != end; cp = cp->next)
619 switch (cp->code) {
620 case 's':
621 if (cp->u.s->wfd != -1 && close(cp->u.s->wfd))
622 err(1, "%s", cp->u.s->wfile);
623 cp->u.s->wfd = -1;
624 break;
625 case 'w':
626 if (cp->u.fd != -1 && close(cp->u.fd))
627 err(1, "%s", cp->t);
628 cp->u.fd = -1;
629 break;
630 case '{':
631 cfclose(cp->u.c, cp->next);
632 break;
633 }
634 }
635