usr.bin/sed/process.c

   1 /*-
   2  * Copyright (c) 1992 Diomidis Spinellis.
   3  * Copyright (c) 1992, 1993, 1994
   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  * 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  * 4. 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.
  20  *
  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.
  32  */
  33
  34 #include <sys/cdefs.h>
  35 __FBSDID("$FreeBSD$");
  36
  37 #ifndef lint
  38 static const char sccsid[] = "@(#)process.c     8.6 (Berkeley) 4/20/94";
  39 #endif
  40
  41 #include <sys/types.h>
  42 #include <sys/stat.h>
  43 #include <sys/ioctl.h>
  44 #include <sys/uio.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 <stdio.h>
  53 #include <stdlib.h>
  54 #include <string.h>
  55 #include <unistd.h>
  56 #include <wchar.h>
  57 #include <wctype.h>
  58
  59 #include "defs.h"
  60 #include "extern.h"
  61
  62 static SPACE HS, PS, SS, YS;
  63 #define pd              PS.deleted
  64 #define ps              PS.space
  65 #define psl             PS.len
  66 #define psanl           PS.append_newline
  67 #define hs              HS.space
  68 #define hsl             HS.len
  69
  70 static inline int        applies(struct s_command *);
  71 static void              do_tr(struct s_tr *);
  72 static void              flush_appends(void);
  73 static void              lputs(char *, size_t);
  74 static int               regexec_e(regex_t *, const char *, int, int, size_t);
  75 static void              regsub(SPACE *, char *, char *);
  76 static int               substitute(struct s_command *);
  77
  78 struct s_appends *appends;      /* Array of pointers to strings to append. */
  79 static int appendx;             /* Index into appends array. */
  80 int appendnum;                  /* Size of appends array. */
  81
  82 static int lastaddr;            /* Set by applies if last address of a range. */
  83 static int sdone;               /* If any substitutes since last line input. */
  84                                 /* Iov structure for 'w' commands. */
  85 static regex_t *defpreg;
  86 size_t maxnsub;
  87 regmatch_t *match;
  88
  89 #define OUT() do {                                                      \
  90         fwrite(ps, 1, psl, outfile);                                    \
  91         if (psanl) fputc('\n', outfile);                                \
  92 } while (0)
  93
  94 void
  95 process(void)
  96 {
  97         struct s_command *cp;
  98         SPACE tspace;
  99         size_t oldpsl = 0;
 100         char *p;
 101         int oldpsanl;
 102
 103         p = NULL;
 104
 105         for (linenum = 0; mf_fgets(&PS, REPLACE);) {
 106                 pd = 0;
 107 top:
 108                 cp = prog;
 109 redirect:
 110                 while (cp != NULL) {
 111                         if (!applies(cp)) {
 112                                 cp = cp->next;
 113                                 continue;
 114                         }
 115                         switch (cp->code) {
 116                         case '{':
 117                                 cp = cp->u.c;
 118                                 goto redirect;
 119                         case 'a':
 120                                 if (appendx >= appendnum)
 121                                         if ((appends = realloc(appends,
 122                                             sizeof(struct s_appends) *
 123                                             (appendnum *= 2))) == NULL)
 124                                                 err(1, "realloc");
 125                                 appends[appendx].type = AP_STRING;
 126                                 appends[appendx].s = cp->t;
 127                                 appends[appendx].len = strlen(cp->t);
 128                                 appendx++;
 129                                 break;
 130                         case 'b':
 131                                 cp = cp->u.c;
 132                                 goto redirect;
 133                         case 'c':
 134                                 pd = 1;
 135                                 psl = 0;
 136                                 if (cp->a2 == NULL || lastaddr || lastline())
 137                                         (void)fprintf(outfile, "%s", cp->t);
 138                                 break;
 139                         case 'd':
 140                                 pd = 1;
 141                                 goto new;
 142                         case 'D':
 143                                 if (pd)
 144                                         goto new;
 145                                 if (psl == 0 ||
 146                                     (p = memchr(ps, '\n', psl)) == NULL) {
 147                                         pd = 1;
 148                                         goto new;
 149                                 } else {
 150                                         psl -= (p + 1) - ps;
 151                                         memmove(ps, p + 1, psl);
 152                                         goto top;
 153                                 }
 154                         case 'g':
 155                                 cspace(&PS, hs, hsl, REPLACE);
 156                                 break;
 157                         case 'G':
 158                                 cspace(&PS, "\n", 1, APPEND);
 159                                 cspace(&PS, hs, hsl, APPEND);
 160                                 break;
 161                         case 'h':
 162                                 cspace(&HS, ps, psl, REPLACE);
 163                                 break;
 164                         case 'H':
 165                                 cspace(&HS, "\n", 1, APPEND);
 166                                 cspace(&HS, ps, psl, APPEND);
 167                                 break;
 168                         case 'i':
 169                                 (void)fprintf(outfile, "%s", cp->t);
 170                                 break;
 171                         case 'l':
 172                                 lputs(ps, psl);
 173                                 break;
 174                         case 'n':
 175                                 if (!nflag && !pd)
 176                                         OUT();
 177                                 flush_appends();
 178                                 if (!mf_fgets(&PS, REPLACE))
 179                                         exit(0);
 180                                 pd = 0;
 181                                 break;
 182                         case 'N':
 183                                 flush_appends();
 184                                 cspace(&PS, "\n", 1, APPEND);
 185                                 if (!mf_fgets(&PS, APPEND))
 186                                         exit(0);
 187                                 break;
 188                         case 'p':
 189                                 if (pd)
 190                                         break;
 191                                 OUT();
 192                                 break;
 193                         case 'P':
 194                                 if (pd)
 195                                         break;
 196                                 if ((p = memchr(ps, '\n', psl)) != NULL) {
 197                                         oldpsl = psl;
 198                                         oldpsanl = psanl;
 199                                         psl = p - ps;
 200                                         psanl = 1;
 201                                 }
 202                                 OUT();
 203                                 if (p != NULL) {
 204                                         psl = oldpsl;
 205                                         psanl = oldpsanl;
 206                                 }
 207                                 break;
 208                         case 'q':
 209                                 if (!nflag && !pd)
 210                                         OUT();
 211                                 flush_appends();
 212                                 exit(0);
 213                         case 'r':
 214                                 if (appendx >= appendnum)
 215                                         if ((appends = realloc(appends,
 216                                             sizeof(struct s_appends) *
 217                                             (appendnum *= 2))) == NULL)
 218                                                 err(1, "realloc");
 219                                 appends[appendx].type = AP_FILE;
 220                                 appends[appendx].s = cp->t;
 221                                 appends[appendx].len = strlen(cp->t);
 222                                 appendx++;
 223                                 break;
 224                         case 's':
 225                                 sdone |= substitute(cp);
 226                                 break;
 227                         case 't':
 228                                 if (sdone) {
 229                                         sdone = 0;
 230                                         cp = cp->u.c;
 231                                         goto redirect;
 232                                 }
 233                                 break;
 234                         case 'w':
 235                                 if (pd)
 236                                         break;
 237                                 if (cp->u.fd == -1 && (cp->u.fd = open(cp->t,
 238                                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
 239                                     DEFFILEMODE)) == -1)
 240                                         err(1, "%s", cp->t);
 241                                 if (write(cp->u.fd, ps, psl) != (ssize_t)psl ||
 242                                     write(cp->u.fd, "\n", 1) != 1)
 243                                         err(1, "%s", cp->t);
 244                                 break;
 245                         case 'x':
 246                                 /*
 247                                  * If the hold space is null, make it empty
 248                                  * but not null.  Otherwise the pattern space
 249                                  * will become null after the swap, which is
 250                                  * an abnormal condition.
 251                                  */
 252                                 if (hs == NULL)
 253                                         cspace(&HS, "", 0, REPLACE);
 254                                 tspace = PS;
 255                                 PS = HS;
 256                                 psanl = tspace.append_newline;
 257                                 HS = tspace;
 258                                 break;
 259                         case 'y':
 260                                 if (pd || psl == 0)
 261                                         break;
 262                                 do_tr(cp->u.y);
 263                                 break;
 264                         case ':':
 265                         case '}':
 266                                 break;
 267                         case '=':
 268                                 (void)fprintf(outfile, "%lu\n", linenum);
 269                         }
 270                         cp = cp->next;
 271                 } /* for all cp */
 272
 273 new:            if (!nflag && !pd)
 274                         OUT();
 275                 flush_appends();
 276         } /* for all lines */
 277 }
 278
 279 /*
 280  * TRUE if the address passed matches the current program state
 281  * (lastline, linenumber, ps).
 282  */
 283 #define MATCH(a)                                                        \
 284         ((a)->type == AT_RE ? regexec_e((a)->u.r, ps, 0, 1, psl) :      \
 285             (a)->type == AT_LINE ? linenum == (a)->u.l : lastline())
 286
 287 /*
 288  * Return TRUE if the command applies to the current line.  Sets the start
 289  * line for process ranges.  Interprets the non-select (``!'') flag.
 290  */
 291 static inline int
 292 applies(struct s_command *cp)
 293 {
 294         int r;
 295
 296         lastaddr = 0;
 297         if (cp->a1 == NULL && cp->a2 == NULL)
 298                 r = 1;
 299         else if (cp->a2)
 300                 if (cp->startline > 0) {
 301                         switch (cp->a2->type) {
 302                         case AT_RELLINE:
 303                                 if (linenum - cp->startline <= cp->a2->u.l)
 304                                         r = 1;
 305                                 else {
 306                                         cp->startline = 0;
 307                                         r = 0;
 308                                 }
 309                                 break;
 310                         default:
 311                                 if (MATCH(cp->a2)) {
 312                                         cp->startline = 0;
 313                                         lastaddr = 1;
 314                                         r = 1;
 315                                 } else if (cp->a2->type == AT_LINE &&
 316                                             linenum > cp->a2->u.l) {
 317                                         /*
 318                                          * We missed the 2nd address due to a
 319                                          * branch, so just close the range and
 320                                          * return false.
 321                                          */
 322                                         cp->startline = 0;
 323                                         r = 0;
 324                                 } else
 325                                         r = 1;
 326                         }
 327                 } else if (cp->a1 && MATCH(cp->a1)) {
 328                         /*
 329                          * If the second address is a number less than or
 330                          * equal to the line number first selected, only
 331                          * one line shall be selected.
 332                          *      -- POSIX 1003.2
 333                          * Likewise if the relative second line address is zero.
 334                          */
 335                         if ((cp->a2->type == AT_LINE &&
 336                             linenum >= cp->a2->u.l) ||
 337                             (cp->a2->type == AT_RELLINE && cp->a2->u.l == 0))
 338                                 lastaddr = 1;
 339                         else {
 340                                 cp->startline = linenum;
 341                         }
 342                         r = 1;
 343                 } else
 344                         r = 0;
 345         else
 346                 r = MATCH(cp->a1);
 347         return (cp->nonsel ? ! r : r);
 348 }
 349
 350 /*
 351  * Reset the sed processor to its initial state.
 352  */
 353 void
 354 resetstate(void)
 355 {
 356         struct s_command *cp;
 357
 358         /*
 359          * Reset all in-range markers.
 360          */
 361         for (cp = prog; cp; cp = cp->code == '{' ? cp->u.c : cp->next)
 362                 if (cp->a2)
 363                         cp->startline = 0;
 364
 365         /*
 366          * Clear out the hold space.
 367          */
 368         cspace(&HS, "", 0, REPLACE);
 369 }
 370
 371 /*
 372  * substitute --
 373  *      Do substitutions in the pattern space.  Currently, we build a
 374  *      copy of the new pattern space in the substitute space structure
 375  *      and then swap them.
 376  */
 377 static int
 378 substitute(struct s_command *cp)
 379 {
 380         SPACE tspace;
 381         regex_t *re;
 382         regoff_t re_off, slen;
 383         int lastempty, n;
 384         char *s;
 385
 386         s = ps;
 387         re = cp->u.s->re;
 388         if (re == NULL) {
 389                 if (defpreg != NULL && cp->u.s->maxbref > defpreg->re_nsub) {
 390                         linenum = cp->u.s->linenum;
 391                         errx(1, "%lu: %s: \\%u not defined in the RE",
 392                                         linenum, fname, cp->u.s->maxbref);
 393                 }
 394         }
 395         if (!regexec_e(re, s, 0, 0, psl))
 396                 return (0);
 397
 398         SS.len = 0;                             /* Clean substitute space. */
 399         slen = psl;
 400         n = cp->u.s->n;
 401         lastempty = 1;
 402
 403         switch (n) {
 404         case 0:                                 /* Global */
 405                 do {
 406                         if (lastempty || match[0].rm_so != match[0].rm_eo) {
 407                                 /* Locate start of replaced string. */
 408                                 re_off = match[0].rm_so;
 409                                 /* Copy leading retained string. */
 410                                 cspace(&SS, s, re_off, APPEND);
 411                                 /* Add in regular expression. */
 412                                 regsub(&SS, s, cp->u.s->new);
 413                         }
 414
 415                         /* Move past this match. */
 416                         if (match[0].rm_so != match[0].rm_eo) {
 417                                 s += match[0].rm_eo;
 418                                 slen -= match[0].rm_eo;
 419                                 lastempty = 0;
 420                         } else {
 421                                 if (match[0].rm_so < slen)
 422                                         cspace(&SS, s + match[0].rm_so, 1,
 423                                             APPEND);
 424                                 s += match[0].rm_so + 1;
 425                                 slen -= match[0].rm_so + 1;
 426                                 lastempty = 1;
 427                         }
 428                 } while (slen >= 0 && regexec_e(re, s, REG_NOTBOL, 0, slen));
 429                 /* Copy trailing retained string. */
 430                 if (slen > 0)
 431                         cspace(&SS, s, slen, APPEND);
 432                 break;
 433         default:                                /* Nth occurrence */
 434                 while (--n) {
 435                         if (match[0].rm_eo == match[0].rm_so)
 436                                 match[0].rm_eo = match[0].rm_so + 1;
 437                         s += match[0].rm_eo;
 438                         slen -= match[0].rm_eo;
 439                         if (slen < 0)
 440                                 return (0);
 441                         if (!regexec_e(re, s, REG_NOTBOL, 0, slen))
 442                                 return (0);
 443                 }
 444                 /* FALLTHROUGH */
 445         case 1:                                 /* 1st occurrence */
 446                 /* Locate start of replaced string. */
 447                 re_off = match[0].rm_so + (s - ps);
 448                 /* Copy leading retained string. */
 449                 cspace(&SS, ps, re_off, APPEND);
 450                 /* Add in regular expression. */
 451                 regsub(&SS, s, cp->u.s->new);
 452                 /* Copy trailing retained string. */
 453                 s += match[0].rm_eo;
 454                 slen -= match[0].rm_eo;
 455                 cspace(&SS, s, slen, APPEND);
 456                 break;
 457         }
 458
 459         /*
 460          * Swap the substitute space and the pattern space, and make sure
 461          * that any leftover pointers into stdio memory get lost.
 462          */
 463         tspace = PS;
 464         PS = SS;
 465         psanl = tspace.append_newline;
 466         SS = tspace;
 467         SS.space = SS.back;
 468
 469         /* Handle the 'p' flag. */
 470         if (cp->u.s->p)
 471                 OUT();
 472
 473         /* Handle the 'w' flag. */
 474         if (cp->u.s->wfile && !pd) {
 475                 if (cp->u.s->wfd == -1 && (cp->u.s->wfd = open(cp->u.s->wfile,
 476                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, DEFFILEMODE)) == -1)
 477                         err(1, "%s", cp->u.s->wfile);
 478                 if (write(cp->u.s->wfd, ps, psl) != (ssize_t)psl ||
 479                     write(cp->u.s->wfd, "\n", 1) != 1)
 480                         err(1, "%s", cp->u.s->wfile);
 481         }
 482         return (1);
 483 }
 484
 485 /*
 486  * do_tr --
 487  *      Perform translation ('y' command) in the pattern space.
 488  */
 489 static void
 490 do_tr(struct s_tr *y)
 491 {
 492         SPACE tmp;
 493         char c, *p;
 494         size_t clen, left;
 495         int i;
 496
 497         if (MB_CUR_MAX == 1) {
 498                 /*
 499                  * Single-byte encoding: perform in-place translation
 500                  * of the pattern space.
 501                  */
 502                 for (p = ps; p < &ps[psl]; p++)
 503                         *p = y->bytetab[(u_char)*p];
 504         } else {
 505                 /*
 506                  * Multi-byte encoding: perform translation into the
 507                  * translation space, then swap the translation and
 508                  * pattern spaces.
 509                  */
 510                 /* Clean translation space. */
 511                 YS.len = 0;
 512                 for (p = ps, left = psl; left > 0; p += clen, left -= clen) {
 513                         if ((c = y->bytetab[(u_char)*p]) != '\0') {
 514                                 cspace(&YS, &c, 1, APPEND);
 515                                 clen = 1;
 516                                 continue;
 517                         }
 518                         for (i = 0; i < y->nmultis; i++)
 519                                 if (left >= y->multis[i].fromlen &&
 520                                     memcmp(p, y->multis[i].from,
 521                                     y->multis[i].fromlen) == 0)
 522                                         break;
 523                         if (i < y->nmultis) {
 524                                 cspace(&YS, y->multis[i].to,
 525                                     y->multis[i].tolen, APPEND);
 526                                 clen = y->multis[i].fromlen;
 527                         } else {
 528                                 cspace(&YS, p, 1, APPEND);
 529                                 clen = 1;
 530                         }
 531                 }
 532                 /* Swap the translation space and the pattern space. */
 533                 tmp = PS;
 534                 PS = YS;
 535                 psanl = tmp.append_newline;
 536                 YS = tmp;
 537                 YS.space = YS.back;
 538         }
 539 }
 540
 541 /*
 542  * Flush append requests.  Always called before reading a line,
 543  * therefore it also resets the substitution done (sdone) flag.
 544  */
 545 static void
 546 flush_appends(void)
 547 {
 548         FILE *f;
 549         int count, i;
 550         char buf[8 * 1024];
 551
 552         for (i = 0; i < appendx; i++)
 553                 switch (appends[i].type) {
 554                 case AP_STRING:
 555                         fwrite(appends[i].s, sizeof(char), appends[i].len,
 556                             outfile);
 557                         break;
 558                 case AP_FILE:
 559                         /*
 560                          * Read files probably shouldn't be cached.  Since
 561                          * it's not an error to read a non-existent file,
 562                          * it's possible that another program is interacting
 563                          * with the sed script through the filesystem.  It
 564                          * would be truly bizarre, but possible.  It's probably
 565                          * not that big a performance win, anyhow.
 566                          */
 567                         if ((f = fopen(appends[i].s, "r")) == NULL)
 568                                 break;
 569                         while ((count = fread(buf, sizeof(char), sizeof(buf), f)))
 570                                 (void)fwrite(buf, sizeof(char), count, outfile);
 571                         (void)fclose(f);
 572                         break;
 573                 }
 574         if (ferror(outfile))
 575                 errx(1, "%s: %s", outfname, strerror(errno ? errno : EIO));
 576         appendx = sdone = 0;
 577 }
 578
 579 static void
 580 lputs(char *s, size_t len)
 581 {
 582         static const char escapes[] = "\\\a\b\f\r\t\v";
 583         int c, col, width;
 584         const char *p;
 585         struct winsize win;
 586         static int termwidth = -1;
 587         size_t clen, i;
 588         wchar_t wc;
 589         mbstate_t mbs;
 590
 591         if (outfile != stdout)
 592                 termwidth = 60;
 593         if (termwidth == -1) {
 594                 if ((p = getenv("COLUMNS")) && *p != '\0')
 595                         termwidth = atoi(p);
 596                 else if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &win) == 0 &&
 597                     win.ws_col > 0)
 598                         termwidth = win.ws_col;
 599                 else
 600                         termwidth = 60;
 601         }
 602         if (termwidth <= 0)
 603                 termwidth = 1;
 604
 605         memset(&mbs, 0, sizeof(mbs));
 606         col = 0;
 607         while (len != 0) {
 608                 clen = mbrtowc(&wc, s, len, &mbs);
 609                 if (clen == 0)
 610                         clen = 1;
 611                 if (clen == (size_t)-1 || clen == (size_t)-2) {
 612                         wc = (unsigned char)*s;
 613                         clen = 1;
 614                         memset(&mbs, 0, sizeof(mbs));
 615                 }
 616                 if (wc == '\n') {
 617                         if (col + 1 >= termwidth)
 618                                 fprintf(outfile, "\\\n");
 619                         fputc('$', outfile);
 620                         fputc('\n', outfile);
 621                         col = 0;
 622                 } else if (iswprint(wc)) {
 623                         width = wcwidth(wc);
 624                         if (col + width >= termwidth) {
 625                                 fprintf(outfile, "\\\n");
 626                                 col = 0;
 627                         }
 628                         fwrite(s, 1, clen, outfile);
 629                         col += width;
 630                 } else if (wc != L'\0' && (c = wctob(wc)) != EOF &&
 631                     (p = strchr(escapes, c)) != NULL) {
 632                         if (col + 2 >= termwidth) {
 633                                 fprintf(outfile, "\\\n");
 634                                 col = 0;
 635                         }
 636                         fprintf(outfile, "\\%c", "\\abfrtv"[p - escapes]);
 637                         col += 2;
 638                 } else {
 639                         if (col + 4 * clen >= (unsigned)termwidth) {
 640                                 fprintf(outfile, "\\\n");
 641                                 col = 0;
 642                         }
 643                         for (i = 0; i < clen; i++)
 644                                 fprintf(outfile, "\\%03o",
 645                                     (int)(unsigned char)s[i]);
 646                         col += 4 * clen;
 647                 }
 648                 s += clen;
 649                 len -= clen;
 650         }
 651         if (col + 1 >= termwidth)
 652                 fprintf(outfile, "\\\n");
 653         (void)fputc('$', outfile);
 654         (void)fputc('\n', outfile);
 655         if (ferror(outfile))
 656                 errx(1, "%s: %s", outfname, strerror(errno ? errno : EIO));
 657 }
 658
 659 static int
 660 regexec_e(regex_t *preg, const char *string, int eflags, int nomatch,
 661         size_t slen)
 662 {
 663         int eval;
 664
 665         if (preg == NULL) {
 666                 if (defpreg == NULL)
 667                         errx(1, "first RE may not be empty");
 668         } else
 669                 defpreg = preg;
 670
 671         /* Set anchors */
 672         match[0].rm_so = 0;
 673         match[0].rm_eo = slen;
 674
 675         eval = regexec(defpreg, string,
 676             nomatch ? 0 : maxnsub + 1, match, eflags | REG_STARTEND);
 677         switch(eval) {
 678         case 0:
 679                 return (1);
 680         case REG_NOMATCH:
 681                 return (0);
 682         }
 683         errx(1, "RE error: %s", strregerror(eval, defpreg));
 684         /* NOTREACHED */
 685 }
 686
 687 /*
 688  * regsub - perform substitutions after a regexp match
 689  * Based on a routine by Henry Spencer
 690  */
 691 static void
 692 regsub(SPACE *sp, char *string, char *src)
 693 {
 694         int len, no;
 695         char c, *dst;
 696
 697 #define NEEDSP(reqlen)                                                  \
 698         /* XXX What is the +1 for? */                                   \
 699         if (sp->len + (reqlen) + 1 >= sp->blen) {                       \
 700                 sp->blen += (reqlen) + 1024;                            \
 701                 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) \
 702                     == NULL)                                            \
 703                         err(1, "realloc");                              \
 704                 dst = sp->space + sp->len;                              \
 705         }
 706
 707         dst = sp->space + sp->len;
 708         while ((c = *src++) != '\0') {
 709                 if (c == '&')
 710                         no = 0;
 711                 else if (c == '\\' && isdigit((unsigned char)*src))
 712                         no = *src++ - '0';
 713                 else
 714                         no = -1;
 715                 if (no < 0) {           /* Ordinary character. */
 716                         if (c == '\\' && (*src == '\\' || *src == '&'))
 717                                 c = *src++;
 718                         NEEDSP(1);
 719                         *dst++ = c;
 720                         ++sp->len;
 721                 } else if (match[no].rm_so != -1 && match[no].rm_eo != -1) {
 722                         len = match[no].rm_eo - match[no].rm_so;
 723                         NEEDSP(len);
 724                         memmove(dst, string + match[no].rm_so, len);
 725                         dst += len;
 726                         sp->len += len;
 727                 }
 728         }
 729         NEEDSP(1);
 730         *dst = '\0';
 731 }
 732
 733 /*
 734  * cspace --
 735  *      Concatenate space: append the source space to the destination space,
 736  *      allocating new space as necessary.
 737  */
 738 void
 739 cspace(SPACE *sp, const char *p, size_t len, enum e_spflag spflag)
 740 {
 741         size_t tlen;
 742
 743         /* Make sure SPACE has enough memory and ramp up quickly. */
 744         tlen = sp->len + len + 1;
 745         if (tlen > sp->blen) {
 746                 sp->blen = tlen + 1024;
 747                 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) ==
 748                     NULL)
 749                         err(1, "realloc");
 750         }
 751
 752         if (spflag == REPLACE)
 753                 sp->len = 0;
 754
 755         memmove(sp->space + sp->len, p, len);
 756
 757         sp->space[sp->len += len] = '\0';
 758 }
 759
 760 /*
 761  * Close all cached opened files and report any errors
 762  */
 763 void
 764 cfclose(struct s_command *cp, struct s_command *end)
 765 {
 766
 767         for (; cp != end; cp = cp->next)
 768                 switch(cp->code) {
 769                 case 's':
 770                         if (cp->u.s->wfd != -1 && close(cp->u.s->wfd))
 771                                 err(1, "%s", cp->u.s->wfile);
 772                         cp->u.s->wfd = -1;
 773                         break;
 774                 case 'w':
 775                         if (cp->u.fd != -1 && close(cp->u.fd))
 776                                 err(1, "%s", cp->t);
 777                         cp->u.fd = -1;
 778                         break;
 779                 case '{':
 780                         cfclose(cp->u.c, cp->next);
 781                         break;
 782                 }
 783 }