drm: Fix Linux/DragonFly PAGE_MASK confusion
[dragonfly.git] / usr.bin / sed / process.c
blob340557b32fcf188b59dcf2d7431a84cadb0828a9
1 /*-
2 * Copyright (c) 1992 Diomidis Spinellis.
3 * Copyright (c) 1992, 1993, 1994
4 * The Regents of the University of California. All rights reserved.
6 * This code is derived from software contributed to Berkeley by
7 * Diomidis Spinellis of Imperial College, University of London.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
33 * @(#)process.c 8.6 (Berkeley) 4/20/94
34 * $FreeBSD: head/usr.bin/sed/process.c 277802 2015-01-27 18:58:24Z pfg $
37 #include <sys/types.h>
38 #include <sys/stat.h>
39 #include <sys/ioctl.h>
40 #include <sys/uio.h>
42 #include <ctype.h>
43 #include <err.h>
44 #include <errno.h>
45 #include <fcntl.h>
46 #include <limits.h>
47 #include <regex.h>
48 #include <stdio.h>
49 #include <stdlib.h>
50 #include <string.h>
51 #include <unistd.h>
52 #include <wchar.h>
53 #include <wctype.h>
55 #include "defs.h"
56 #include "extern.h"
58 static SPACE HS, PS, SS, YS;
59 #define pd PS.deleted
60 #define ps PS.space
61 #define psl PS.len
62 #define psanl PS.append_newline
63 #define hs HS.space
64 #define hsl HS.len
66 static inline int applies(struct s_command *);
67 static void do_tr(struct s_tr *);
68 static void flush_appends(void);
69 static void lputs(char *, size_t);
70 static int regexec_e(regex_t *, const char *, int, int, size_t);
71 static void regsub(SPACE *, char *, char *);
72 static int substitute(struct s_command *);
74 struct s_appends *appends; /* Array of pointers to strings to append. */
75 static int appendx; /* Index into appends array. */
76 int appendnum; /* Size of appends array. */
78 static int lastaddr; /* Set by applies if last address of a range. */
79 static int sdone; /* If any substitutes since last line input. */
80 /* Iov structure for 'w' commands. */
81 static regex_t *defpreg;
82 size_t maxnsub;
83 regmatch_t *match;
85 #define OUT() do { \
86 fwrite(ps, 1, psl, outfile); \
87 if (psanl) fputc('\n', outfile); \
88 } while (0)
90 void
91 process(void)
93 struct s_command *cp;
94 SPACE tspace;
95 size_t oldpsl = 0;
96 char *p;
97 int oldpsanl;
99 p = NULL;
101 for (linenum = 0; mf_fgets(&PS, REPLACE);) {
102 pd = 0;
103 top:
104 cp = prog;
105 redirect:
106 while (cp != NULL) {
107 if (!applies(cp)) {
108 cp = cp->next;
109 continue;
111 switch (cp->code) {
112 case '{':
113 cp = cp->u.c;
114 goto redirect;
115 case 'a':
116 if (appendx >= appendnum)
117 if ((appends = realloc(appends,
118 sizeof(struct s_appends) *
119 (appendnum *= 2))) == NULL)
120 err(1, "realloc");
121 appends[appendx].type = AP_STRING;
122 appends[appendx].s = cp->t;
123 appends[appendx].len = strlen(cp->t);
124 appendx++;
125 break;
126 case 'b':
127 cp = cp->u.c;
128 goto redirect;
129 case 'c':
130 pd = 1;
131 psl = 0;
132 if (cp->a2 == NULL || lastaddr || lastline())
133 (void)fprintf(outfile, "%s", cp->t);
134 break;
135 case 'd':
136 pd = 1;
137 goto new;
138 case 'D':
139 if (pd)
140 goto new;
141 if (psl == 0 ||
142 (p = memchr(ps, '\n', psl)) == NULL) {
143 pd = 1;
144 goto new;
145 } else {
146 psl -= (p + 1) - ps;
147 memmove(ps, p + 1, psl);
148 goto top;
150 case 'g':
151 cspace(&PS, hs, hsl, REPLACE);
152 break;
153 case 'G':
154 cspace(&PS, "\n", 1, APPEND);
155 cspace(&PS, hs, hsl, APPEND);
156 break;
157 case 'h':
158 cspace(&HS, ps, psl, REPLACE);
159 break;
160 case 'H':
161 cspace(&HS, "\n", 1, APPEND);
162 cspace(&HS, ps, psl, APPEND);
163 break;
164 case 'i':
165 (void)fprintf(outfile, "%s", cp->t);
166 break;
167 case 'l':
168 lputs(ps, psl);
169 break;
170 case 'n':
171 if (!nflag && !pd)
172 OUT();
173 flush_appends();
174 if (!mf_fgets(&PS, REPLACE))
175 exit(0);
176 pd = 0;
177 break;
178 case 'N':
179 flush_appends();
180 cspace(&PS, "\n", 1, APPEND);
181 if (!mf_fgets(&PS, APPEND))
182 exit(0);
183 break;
184 case 'p':
185 if (pd)
186 break;
187 OUT();
188 break;
189 case 'P':
190 if (pd)
191 break;
192 if ((p = memchr(ps, '\n', psl)) != NULL) {
193 oldpsl = psl;
194 oldpsanl = psanl;
195 psl = p - ps;
196 psanl = 1;
198 OUT();
199 if (p != NULL) {
200 psl = oldpsl;
201 psanl = oldpsanl;
203 break;
204 case 'q':
205 if (!nflag && !pd)
206 OUT();
207 flush_appends();
208 exit(0);
209 case 'r':
210 if (appendx >= appendnum)
211 if ((appends = realloc(appends,
212 sizeof(struct s_appends) *
213 (appendnum *= 2))) == NULL)
214 err(1, "realloc");
215 appends[appendx].type = AP_FILE;
216 appends[appendx].s = cp->t;
217 appends[appendx].len = strlen(cp->t);
218 appendx++;
219 break;
220 case 's':
221 sdone |= substitute(cp);
222 break;
223 case 't':
224 if (sdone) {
225 sdone = 0;
226 cp = cp->u.c;
227 goto redirect;
229 break;
230 case 'w':
231 if (pd)
232 break;
233 if (cp->u.fd == -1 && (cp->u.fd = open(cp->t,
234 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
235 DEFFILEMODE)) == -1)
236 err(1, "%s", cp->t);
237 if (write(cp->u.fd, ps, psl) != (ssize_t)psl ||
238 write(cp->u.fd, "\n", 1) != 1)
239 err(1, "%s", cp->t);
240 break;
241 case 'x':
243 * If the hold space is null, make it empty
244 * but not null. Otherwise the pattern space
245 * will become null after the swap, which is
246 * an abnormal condition.
248 if (hs == NULL)
249 cspace(&HS, "", 0, REPLACE);
250 tspace = PS;
251 PS = HS;
252 psanl = tspace.append_newline;
253 HS = tspace;
254 break;
255 case 'y':
256 if (pd || psl == 0)
257 break;
258 do_tr(cp->u.y);
259 break;
260 case ':':
261 case '}':
262 break;
263 case '=':
264 (void)fprintf(outfile, "%lu\n", linenum);
266 cp = cp->next;
267 } /* for all cp */
269 new: if (!nflag && !pd)
270 OUT();
271 flush_appends();
272 } /* for all lines */
276 * TRUE if the address passed matches the current program state
277 * (lastline, linenumber, ps).
279 #define MATCH(a) \
280 ((a)->type == AT_RE ? regexec_e((a)->u.r, ps, 0, 1, psl) : \
281 (a)->type == AT_LINE ? linenum == (a)->u.l : lastline())
284 * Return TRUE if the command applies to the current line. Sets the start
285 * line for process ranges. Interprets the non-select (``!'') flag.
287 static inline int
288 applies(struct s_command *cp)
290 int r;
292 lastaddr = 0;
293 if (cp->a1 == NULL && cp->a2 == NULL)
294 r = 1;
295 else if (cp->a2)
296 if (cp->startline > 0) {
297 switch (cp->a2->type) {
298 case AT_RELLINE:
299 if (linenum - cp->startline <= cp->a2->u.l)
300 r = 1;
301 else {
302 cp->startline = 0;
303 r = 0;
305 break;
306 default:
307 if (MATCH(cp->a2)) {
308 cp->startline = 0;
309 lastaddr = 1;
310 r = 1;
311 } else if (cp->a2->type == AT_LINE &&
312 linenum > cp->a2->u.l) {
314 * We missed the 2nd address due to a
315 * branch, so just close the range and
316 * return false.
318 cp->startline = 0;
319 r = 0;
320 } else
321 r = 1;
323 } else if (cp->a1 && MATCH(cp->a1)) {
325 * If the second address is a number less than or
326 * equal to the line number first selected, only
327 * one line shall be selected.
328 * -- POSIX 1003.2
329 * Likewise if the relative second line address is zero.
331 if ((cp->a2->type == AT_LINE &&
332 linenum >= cp->a2->u.l) ||
333 (cp->a2->type == AT_RELLINE && cp->a2->u.l == 0))
334 lastaddr = 1;
335 else {
336 cp->startline = linenum;
338 r = 1;
339 } else
340 r = 0;
341 else
342 r = MATCH(cp->a1);
343 return (cp->nonsel ? ! r : r);
347 * Reset the sed processor to its initial state.
349 void
350 resetstate(void)
352 struct s_command *cp;
355 * Reset all in-range markers.
357 for (cp = prog; cp; cp = cp->code == '{' ? cp->u.c : cp->next)
358 if (cp->a2)
359 cp->startline = 0;
362 * Clear out the hold space.
364 cspace(&HS, "", 0, REPLACE);
368 * substitute --
369 * Do substitutions in the pattern space. Currently, we build a
370 * copy of the new pattern space in the substitute space structure
371 * and then swap them.
373 static int
374 substitute(struct s_command *cp)
376 SPACE tspace;
377 regex_t *re;
378 regoff_t re_off, slen;
379 int lastempty, n;
380 char *s;
382 s = ps;
383 re = cp->u.s->re;
384 if (re == NULL) {
385 if (defpreg != NULL && cp->u.s->maxbref > defpreg->re_nsub) {
386 linenum = cp->u.s->linenum;
387 errx(1, "%lu: %s: \\%u not defined in the RE",
388 linenum, fname, cp->u.s->maxbref);
391 if (!regexec_e(re, s, 0, 0, psl))
392 return (0);
394 SS.len = 0; /* Clean substitute space. */
395 slen = psl;
396 n = cp->u.s->n;
397 lastempty = 1;
399 switch (n) {
400 case 0: /* Global */
401 do {
402 if (lastempty || match[0].rm_so != match[0].rm_eo) {
403 /* Locate start of replaced string. */
404 re_off = match[0].rm_so;
405 /* Copy leading retained string. */
406 cspace(&SS, s, re_off, APPEND);
407 /* Add in regular expression. */
408 regsub(&SS, s, cp->u.s->new);
411 /* Move past this match. */
412 if (match[0].rm_so != match[0].rm_eo) {
413 s += match[0].rm_eo;
414 slen -= match[0].rm_eo;
415 lastempty = 0;
416 } else {
417 if (match[0].rm_so < slen)
418 cspace(&SS, s + match[0].rm_so, 1,
419 APPEND);
420 s += match[0].rm_so + 1;
421 slen -= match[0].rm_so + 1;
422 lastempty = 1;
424 } while (slen >= 0 && regexec_e(re, s, REG_NOTBOL, 0, slen));
425 /* Copy trailing retained string. */
426 if (slen > 0)
427 cspace(&SS, s, slen, APPEND);
428 break;
429 default: /* Nth occurrence */
430 while (--n) {
431 if (match[0].rm_eo == match[0].rm_so)
432 match[0].rm_eo = match[0].rm_so + 1;
433 s += match[0].rm_eo;
434 slen -= match[0].rm_eo;
435 if (slen < 0)
436 return (0);
437 if (!regexec_e(re, s, REG_NOTBOL, 0, slen))
438 return (0);
440 /* FALLTHROUGH */
441 case 1: /* 1st occurrence */
442 /* Locate start of replaced string. */
443 re_off = match[0].rm_so + (s - ps);
444 /* Copy leading retained string. */
445 cspace(&SS, ps, re_off, APPEND);
446 /* Add in regular expression. */
447 regsub(&SS, s, cp->u.s->new);
448 /* Copy trailing retained string. */
449 s += match[0].rm_eo;
450 slen -= match[0].rm_eo;
451 cspace(&SS, s, slen, APPEND);
452 break;
456 * Swap the substitute space and the pattern space, and make sure
457 * that any leftover pointers into stdio memory get lost.
459 tspace = PS;
460 PS = SS;
461 psanl = tspace.append_newline;
462 SS = tspace;
463 SS.space = SS.back;
465 /* Handle the 'p' flag. */
466 if (cp->u.s->p)
467 OUT();
469 /* Handle the 'w' flag. */
470 if (cp->u.s->wfile && !pd) {
471 if (cp->u.s->wfd == -1 && (cp->u.s->wfd = open(cp->u.s->wfile,
472 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, DEFFILEMODE)) == -1)
473 err(1, "%s", cp->u.s->wfile);
474 if (write(cp->u.s->wfd, ps, psl) != (ssize_t)psl ||
475 write(cp->u.s->wfd, "\n", 1) != 1)
476 err(1, "%s", cp->u.s->wfile);
478 return (1);
482 * do_tr --
483 * Perform translation ('y' command) in the pattern space.
485 static void
486 do_tr(struct s_tr *y)
488 SPACE tmp;
489 char c, *p;
490 size_t clen, left;
491 int i;
493 if (MB_CUR_MAX == 1) {
495 * Single-byte encoding: perform in-place translation
496 * of the pattern space.
498 for (p = ps; p < &ps[psl]; p++)
499 *p = y->bytetab[(u_char)*p];
500 } else {
502 * Multi-byte encoding: perform translation into the
503 * translation space, then swap the translation and
504 * pattern spaces.
506 /* Clean translation space. */
507 YS.len = 0;
508 for (p = ps, left = psl; left > 0; p += clen, left -= clen) {
509 if ((c = y->bytetab[(u_char)*p]) != '\0') {
510 cspace(&YS, &c, 1, APPEND);
511 clen = 1;
512 continue;
514 for (i = 0; i < y->nmultis; i++)
515 if (left >= y->multis[i].fromlen &&
516 memcmp(p, y->multis[i].from,
517 y->multis[i].fromlen) == 0)
518 break;
519 if (i < y->nmultis) {
520 cspace(&YS, y->multis[i].to,
521 y->multis[i].tolen, APPEND);
522 clen = y->multis[i].fromlen;
523 } else {
524 cspace(&YS, p, 1, APPEND);
525 clen = 1;
528 /* Swap the translation space and the pattern space. */
529 tmp = PS;
530 PS = YS;
531 psanl = tmp.append_newline;
532 YS = tmp;
533 YS.space = YS.back;
538 * Flush append requests. Always called before reading a line,
539 * therefore it also resets the substitution done (sdone) flag.
541 static void
542 flush_appends(void)
544 FILE *f;
545 int count, i;
546 char buf[8 * 1024];
548 for (i = 0; i < appendx; i++)
549 switch (appends[i].type) {
550 case AP_STRING:
551 fwrite(appends[i].s, sizeof(char), appends[i].len,
552 outfile);
553 break;
554 case AP_FILE:
556 * Read files probably shouldn't be cached. Since
557 * it's not an error to read a non-existent file,
558 * it's possible that another program is interacting
559 * with the sed script through the filesystem. It
560 * would be truly bizarre, but possible. It's probably
561 * not that big a performance win, anyhow.
563 if ((f = fopen(appends[i].s, "r")) == NULL)
564 break;
565 while ((count = fread(buf, sizeof(char), sizeof(buf), f)))
566 (void)fwrite(buf, sizeof(char), count, outfile);
567 (void)fclose(f);
568 break;
570 if (ferror(outfile))
571 errx(1, "%s: %s", outfname, strerror(errno ? errno : EIO));
572 appendx = sdone = 0;
575 static void
576 lputs(char *s, size_t len)
578 static const char escapes[] = "\\\a\b\f\r\t\v";
579 int c, col, width;
580 const char *p;
581 struct winsize win;
582 static int termwidth = -1;
583 size_t clen, i;
584 wchar_t wc;
585 mbstate_t mbs;
587 if (outfile != stdout)
588 termwidth = 60;
589 if (termwidth == -1) {
590 if ((p = getenv("COLUMNS")) && *p != '\0')
591 termwidth = atoi(p);
592 else if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &win) == 0 &&
593 win.ws_col > 0)
594 termwidth = win.ws_col;
595 else
596 termwidth = 60;
598 if (termwidth <= 0)
599 termwidth = 1;
601 memset(&mbs, 0, sizeof(mbs));
602 col = 0;
603 while (len != 0) {
604 clen = mbrtowc(&wc, s, len, &mbs);
605 if (clen == 0)
606 clen = 1;
607 if (clen == (size_t)-1 || clen == (size_t)-2) {
608 wc = (unsigned char)*s;
609 clen = 1;
610 memset(&mbs, 0, sizeof(mbs));
612 if (wc == '\n') {
613 if (col + 1 >= termwidth)
614 fprintf(outfile, "\\\n");
615 fputc('$', outfile);
616 fputc('\n', outfile);
617 col = 0;
618 } else if (iswprint(wc)) {
619 width = wcwidth(wc);
620 if (col + width >= termwidth) {
621 fprintf(outfile, "\\\n");
622 col = 0;
624 fwrite(s, 1, clen, outfile);
625 col += width;
626 } else if (wc != L'\0' && (c = wctob(wc)) != EOF &&
627 (p = strchr(escapes, c)) != NULL) {
628 if (col + 2 >= termwidth) {
629 fprintf(outfile, "\\\n");
630 col = 0;
632 fprintf(outfile, "\\%c", "\\abfrtv"[p - escapes]);
633 col += 2;
634 } else {
635 if (col + 4 * clen >= (unsigned)termwidth) {
636 fprintf(outfile, "\\\n");
637 col = 0;
639 for (i = 0; i < clen; i++)
640 fprintf(outfile, "\\%03o",
641 (int)(unsigned char)s[i]);
642 col += 4 * clen;
644 s += clen;
645 len -= clen;
647 if (col + 1 >= termwidth)
648 fprintf(outfile, "\\\n");
649 (void)fputc('$', outfile);
650 (void)fputc('\n', outfile);
651 if (ferror(outfile))
652 errx(1, "%s: %s", outfname, strerror(errno ? errno : EIO));
655 static int
656 regexec_e(regex_t *preg, const char *string, int eflags, int nomatch,
657 size_t slen)
659 int eval;
661 if (preg == NULL) {
662 if (defpreg == NULL)
663 errx(1, "first RE may not be empty");
664 } else
665 defpreg = preg;
667 /* Set anchors */
668 match[0].rm_so = 0;
669 match[0].rm_eo = slen;
671 eval = regexec(defpreg, string,
672 nomatch ? 0 : maxnsub + 1, match, eflags | REG_STARTEND);
673 switch(eval) {
674 case 0:
675 return (1);
676 case REG_NOMATCH:
677 return (0);
679 errx(1, "RE error: %s", strregerror(eval, defpreg));
680 /* NOTREACHED */
684 * regsub - perform substitutions after a regexp match
685 * Based on a routine by Henry Spencer
687 static void
688 regsub(SPACE *sp, char *string, char *src)
690 int len, no;
691 char c, *dst;
693 #define NEEDSP(reqlen) \
694 /* XXX What is the +1 for? */ \
695 if (sp->len + (reqlen) + 1 >= sp->blen) { \
696 sp->blen += (reqlen) + 1024; \
697 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) \
698 == NULL) \
699 err(1, "realloc"); \
700 dst = sp->space + sp->len; \
703 dst = sp->space + sp->len;
704 while ((c = *src++) != '\0') {
705 if (c == '&')
706 no = 0;
707 else if (c == '\\' && isdigit((unsigned char)*src))
708 no = *src++ - '0';
709 else
710 no = -1;
711 if (no < 0) { /* Ordinary character. */
712 if (c == '\\' && (*src == '\\' || *src == '&'))
713 c = *src++;
714 NEEDSP(1);
715 *dst++ = c;
716 ++sp->len;
717 } else if (match[no].rm_so != -1 && match[no].rm_eo != -1) {
718 len = match[no].rm_eo - match[no].rm_so;
719 NEEDSP(len);
720 memmove(dst, string + match[no].rm_so, len);
721 dst += len;
722 sp->len += len;
725 NEEDSP(1);
726 *dst = '\0';
730 * cspace --
731 * Concatenate space: append the source space to the destination space,
732 * allocating new space as necessary.
734 void
735 cspace(SPACE *sp, const char *p, size_t len, enum e_spflag spflag)
737 size_t tlen;
739 /* Make sure SPACE has enough memory and ramp up quickly. */
740 tlen = sp->len + len + 1;
741 if (tlen > sp->blen) {
742 sp->blen = tlen + 1024;
743 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) ==
744 NULL)
745 err(1, "realloc");
748 if (spflag == REPLACE)
749 sp->len = 0;
751 memmove(sp->space + sp->len, p, len);
753 sp->space[sp->len += len] = '\0';
757 * Close all cached opened files and report any errors
759 void
760 cfclose(struct s_command *cp, struct s_command *end)
763 for (; cp != end; cp = cp->next)
764 switch(cp->code) {
765 case 's':
766 if (cp->u.s->wfd != -1 && close(cp->u.s->wfd))
767 err(1, "%s", cp->u.s->wfile);
768 cp->u.s->wfd = -1;
769 break;
770 case 'w':
771 if (cp->u.fd != -1 && close(cp->u.fd))
772 err(1, "%s", cp->t);
773 cp->u.fd = -1;
774 break;
775 case '{':
776 cfclose(cp->u.c, cp->next);
777 break;