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  * 3. All advertising materials mentioning features or use of this software
  18  *    must display the following acknowledgement:
  19  *      This product includes software developed by the University of
  20  *      California, Berkeley and its contributors.
  21  * 4. Neither the name of the University nor the names of its contributors
  22  *    may be used to endorse or promote products derived from this software
  23  *    without specific prior written permission.
  24  *
  25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  35  * SUCH DAMAGE.
  36  */
  37
  38 #ifndef lint
  39 #if 0
  40 static char sccsid[] = "@(#)process.c   8.6 (Berkeley) 4/20/94";
  41 #endif
  42 static const char rcsid[] =
  43   "$FreeBSD$";
  44 #endif /* not lint */
  45
  46 #include <sys/types.h>
  47 #include <sys/stat.h>
  48 #include <sys/ioctl.h>
  49 #include <sys/uio.h>
  50
  51 #include <ctype.h>
  52 #include <err.h>
  53 #include <errno.h>
  54 #include <fcntl.h>
  55 #include <limits.h>
  56 #include <regex.h>
  57 #include <stdio.h>
  58 #include <stdlib.h>
  59 #include <string.h>
  60 #include <unistd.h>
  61
  62 #include "defs.h"
  63 #include "extern.h"
  64
  65 static SPACE HS, PS, SS;
  66 #define pd              PS.deleted
  67 #define ps              PS.space
  68 #define psl             PS.len
  69 #define hs              HS.space
  70 #define hsl             HS.len
  71
  72 static inline int        applies __P((struct s_command *));
  73 static void              flush_appends __P((void));
  74 static void              lputs __P((char *));
  75 static inline int        regexec_e __P((regex_t *, const char *, int, int, size_t));
  76 static void              regsub __P((SPACE *, char *, char *));
  77 static int               substitute __P((struct s_command *));
  78
  79 struct s_appends *appends;      /* Array of pointers to strings to append. */
  80 static int appendx;             /* Index into appends array. */
  81 int appendnum;                  /* Size of appends array. */
  82
  83 static int lastaddr;            /* Set by applies if last address of a range. */
  84 static int sdone;               /* If any substitutes since last line input. */
  85                                 /* Iov structure for 'w' commands. */
  86 static regex_t *defpreg;
  87 size_t maxnsub;
  88 regmatch_t *match;
  89
  90 #define OUT(s) { fwrite(s, sizeof(u_char), psl, stdout); }
  91
  92 void
  93 process()
  94 {
  95         struct s_command *cp;
  96         SPACE tspace;
  97         size_t len, oldpsl = 0;
  98         char *p;
  99
 100         for (linenum = 0; mf_fgets(&PS, REPLACE);) {
 101                 pd = 0;
 102 top:
 103                 cp = prog;
 104 redirect:
 105                 while (cp != NULL) {
 106                         if (!applies(cp)) {
 107                                 cp = cp->next;
 108                                 continue;
 109                         }
 110                         switch (cp->code) {
 111                         case '{':
 112                                 cp = cp->u.c;
 113                                 goto redirect;
 114                         case 'a':
 115                                 if (appendx >= appendnum)
 116                                         appends = xrealloc(appends,
 117                                             sizeof(struct s_appends) *
 118                                             (appendnum *= 2));
 119                                 appends[appendx].type = AP_STRING;
 120                                 appends[appendx].s = cp->t;
 121                                 appends[appendx].len = strlen(cp->t);
 122                                 appendx++;
 123                                 break;
 124                         case 'b':
 125                                 cp = cp->u.c;
 126                                 goto redirect;
 127                         case 'c':
 128                                 pd = 1;
 129                                 psl = 0;
 130                                 if (cp->a2 == NULL || lastaddr)
 131                                         (void)printf("%s", cp->t);
 132                                 break;
 133                         case 'd':
 134                                 pd = 1;
 135                                 goto new;
 136                         case 'D':
 137                                 if (pd)
 138                                         goto new;
 139                                 if ((p = memchr(ps, '\n', psl - 1)) == NULL) {
 140                                         pd = 1;
 141                                         goto new;
 142                                 } else {
 143                                         psl -= (p + 1) - ps;
 144                                         memmove(ps, p + 1, psl);
 145                                         goto top;
 146                                 }
 147                         case 'g':
 148                                 cspace(&PS, hs, hsl, REPLACE);
 149                                 break;
 150                         case 'G':
 151                                 cspace(&PS, hs, hsl, 0);
 152                                 break;
 153                         case 'h':
 154                                 cspace(&HS, ps, psl, REPLACE);
 155                                 break;
 156                         case 'H':
 157                                 cspace(&HS, ps, psl, 0);
 158                                 break;
 159                         case 'i':
 160                                 (void)printf("%s", cp->t);
 161                                 break;
 162                         case 'l':
 163                                 lputs(ps);
 164                                 break;
 165                         case 'n':
 166                                 if (!nflag && !pd)
 167                                         OUT(ps)
 168                                 flush_appends();
 169                                 if (!mf_fgets(&PS, REPLACE))
 170                                         exit(0);
 171                                 pd = 0;
 172                                 break;
 173                         case 'N':
 174                                 flush_appends();
 175                                 if (!mf_fgets(&PS, 0)) {
 176                                         if (!nflag && !pd)
 177                                                 OUT(ps)
 178                                         exit(0);
 179                                 }
 180                                 break;
 181                         case 'p':
 182                                 if (pd)
 183                                         break;
 184                                 OUT(ps)
 185                                 break;
 186                         case 'P':
 187                                 if (pd)
 188                                         break;
 189                                 if ((p = memchr(ps, '\n', psl - 1)) != NULL) {
 190                                         oldpsl = psl;
 191                                         psl = (p + 1) - ps;
 192                                 }
 193                                 OUT(ps)
 194                                 if (p != NULL)
 195                                         psl = oldpsl;
 196                                 break;
 197                         case 'q':
 198                                 if (!nflag && !pd)
 199                                         OUT(ps)
 200                                 flush_appends();
 201                                 exit(0);
 202                         case 'r':
 203                                 if (appendx >= appendnum)
 204                                         appends = xrealloc(appends,
 205                                             sizeof(struct s_appends) *
 206                                             (appendnum *= 2));
 207                                 appends[appendx].type = AP_FILE;
 208                                 appends[appendx].s = cp->t;
 209                                 appends[appendx].len = strlen(cp->t);
 210                                 appendx++;
 211                                 break;
 212                         case 's':
 213                                 sdone |= substitute(cp);
 214                                 break;
 215                         case 't':
 216                                 if (sdone) {
 217                                         sdone = 0;
 218                                         cp = cp->u.c;
 219                                         goto redirect;
 220                                 }
 221                                 break;
 222                         case 'w':
 223                                 if (pd)
 224                                         break;
 225                                 if (cp->u.fd == -1 && (cp->u.fd = open(cp->t,
 226                                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
 227                                     DEFFILEMODE)) == -1)
 228                                         err(1, "%s", cp->t);
 229                                 if (write(cp->u.fd, ps, psl) != psl)
 230                                         err(1, "%s", cp->t);
 231                                 break;
 232                         case 'x':
 233                                 if (hs == NULL)
 234                                         cspace(&HS, "\n", 1, REPLACE);
 235                                 tspace = PS;
 236                                 PS = HS;
 237                                 HS = tspace;
 238                                 break;
 239                         case 'y':
 240                                 if (pd)
 241                                         break;
 242                                 for (p = ps, len = psl; --len; ++p)
 243                                         *p = cp->u.y[(unsigned char)*p];
 244                                 break;
 245                         case ':':
 246                         case '}':
 247                                 break;
 248                         case '=':
 249                                 (void)printf("%lu\n", linenum);
 250                         }
 251                         cp = cp->next;
 252                 } /* for all cp */
 253
 254 new:            if (!nflag && !pd)
 255                         OUT(ps)
 256                 flush_appends();
 257         } /* for all lines */
 258 }
 259
 260 /*
 261  * TRUE if the address passed matches the current program state
 262  * (lastline, linenumber, ps).
 263  */
 264 #define MATCH(a)                                                \
 265         (a)->type == AT_RE ? regexec_e((a)->u.r, ps, 0, 1, psl) :       \
 266             (a)->type == AT_LINE ? linenum == (a)->u.l : lastline
 267
 268 /*
 269  * Return TRUE if the command applies to the current line.  Sets the inrange
 270  * flag to process ranges.  Interprets the non-select (``!'') flag.
 271  */
 272 static inline int
 273 applies(cp)
 274         struct s_command *cp;
 275 {
 276         int r;
 277
 278         lastaddr = 0;
 279         if (cp->a1 == NULL && cp->a2 == NULL)
 280                 r = 1;
 281         else if (cp->a2)
 282                 if (cp->inrange) {
 283                         if (MATCH(cp->a2)) {
 284                                 cp->inrange = 0;
 285                                 lastaddr = 1;
 286                         }
 287                         r = 1;
 288                 } else if (MATCH(cp->a1)) {
 289                         /*
 290                          * If the second address is a number less than or
 291                          * equal to the line number first selected, only
 292                          * one line shall be selected.
 293                          *      -- POSIX 1003.2
 294                          */
 295                         if (cp->a2->type == AT_LINE &&
 296                             linenum >= cp->a2->u.l)
 297                                 lastaddr = 1;
 298                         else
 299                                 cp->inrange = 1;
 300                         r = 1;
 301                 } else
 302                         r = 0;
 303         else
 304                 r = MATCH(cp->a1);
 305         return (cp->nonsel ? ! r : r);
 306 }
 307
 308 /*
 309  * substitute --
 310  *      Do substitutions in the pattern space.  Currently, we build a
 311  *      copy of the new pattern space in the substitute space structure
 312  *      and then swap them.
 313  */
 314 static int
 315 substitute(cp)
 316         struct s_command *cp;
 317 {
 318         SPACE tspace;
 319         regex_t *re;
 320         size_t re_off, slen;
 321         int lastempty, n;
 322         char *s;
 323
 324         s = ps;
 325         re = cp->u.s->re;
 326         if (re == NULL) {
 327                 if (defpreg != NULL && cp->u.s->maxbref > defpreg->re_nsub) {
 328                         linenum = cp->u.s->linenum;
 329                         errx(1, "%lu: %s: \\%d not defined in the RE",
 330                                         linenum, fname, cp->u.s->maxbref);
 331                 }
 332         }
 333         if (!regexec_e(re, s, 0, 0, psl))
 334                 return (0);
 335
 336         SS.len = 0;                             /* Clean substitute space. */
 337         slen = psl;
 338         n = cp->u.s->n;
 339         lastempty = 1;
 340
 341         switch (n) {
 342         case 0:                                 /* Global */
 343                 do {
 344                         if (lastempty || match[0].rm_so != match[0].rm_eo) {
 345                                 /* Locate start of replaced string. */
 346                                 re_off = match[0].rm_so;
 347                                 /* Copy leading retained string. */
 348                                 cspace(&SS, s, re_off, APPEND);
 349                                 /* Add in regular expression. */
 350                                 regsub(&SS, s, cp->u.s->new);
 351                         }
 352
 353                         /* Move past this match. */
 354                         if (match[0].rm_so != match[0].rm_eo) {
 355                                 s += match[0].rm_eo;
 356                                 slen -= match[0].rm_eo;
 357                                 lastempty = 0;
 358                         } else {
 359                                 if (match[0].rm_so == 0)
 360                                         cspace(&SS, s, match[0].rm_so + 1,
 361                                             APPEND);
 362                                 else
 363                                         cspace(&SS, s + match[0].rm_so, 1,
 364                                             APPEND);
 365                                 s += match[0].rm_so + 1;
 366                                 slen -= match[0].rm_so + 1;
 367                                 lastempty = 1;
 368                         }
 369                 } while (slen > 0 && regexec_e(re, s, REG_NOTBOL, 0, slen));
 370                 /* Copy trailing retained string. */
 371                 if (slen > 0)
 372                         cspace(&SS, s, slen, APPEND);
 373                 break;
 374         default:                                /* Nth occurrence */
 375                 while (--n) {
 376                         s += match[0].rm_eo;
 377                         slen -= match[0].rm_eo;
 378                         if (!regexec_e(re, s, REG_NOTBOL, 0, slen))
 379                                 return (0);
 380                 }
 381                 /* FALLTHROUGH */
 382         case 1:                                 /* 1st occurrence */
 383                 /* Locate start of replaced string. */
 384                 re_off = match[0].rm_so + (s - ps);
 385                 /* Copy leading retained string. */
 386                 cspace(&SS, ps, re_off, APPEND);
 387                 /* Add in regular expression. */
 388                 regsub(&SS, s, cp->u.s->new);
 389                 /* Copy trailing retained string. */
 390                 s += match[0].rm_eo;
 391                 slen -= match[0].rm_eo;
 392                 cspace(&SS, s, slen, APPEND);
 393                 break;
 394         }
 395
 396         /*
 397          * Swap the substitute space and the pattern space, and make sure
 398          * that any leftover pointers into stdio memory get lost.
 399          */
 400         tspace = PS;
 401         PS = SS;
 402         SS = tspace;
 403         SS.space = SS.back;
 404
 405         /* Handle the 'p' flag. */
 406         if (cp->u.s->p)
 407                 OUT(ps)
 408
 409         /* Handle the 'w' flag. */
 410         if (cp->u.s->wfile && !pd) {
 411                 if (cp->u.s->wfd == -1 && (cp->u.s->wfd = open(cp->u.s->wfile,
 412                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, DEFFILEMODE)) == -1)
 413                         err(1, "%s", cp->u.s->wfile);
 414                 if (write(cp->u.s->wfd, ps, psl) != psl)
 415                         err(1, "%s", cp->u.s->wfile);
 416         }
 417         return (1);
 418 }
 419
 420 /*
 421  * Flush append requests.  Always called before reading a line,
 422  * therefore it also resets the substitution done (sdone) flag.
 423  */
 424 static void
 425 flush_appends()
 426 {
 427         FILE *f;
 428         int count, i;
 429         char buf[8 * 1024];
 430
 431         for (i = 0; i < appendx; i++)
 432                 switch (appends[i].type) {
 433                 case AP_STRING:
 434                         fwrite(appends[i].s, sizeof(char), appends[i].len,
 435                             stdout);
 436                         break;
 437                 case AP_FILE:
 438                         /*
 439                          * Read files probably shouldn't be cached.  Since
 440                          * it's not an error to read a non-existent file,
 441                          * it's possible that another program is interacting
 442                          * with the sed script through the file system.  It
 443                          * would be truly bizarre, but possible.  It's probably
 444                          * not that big a performance win, anyhow.
 445                          */
 446                         if ((f = fopen(appends[i].s, "r")) == NULL)
 447                                 break;
 448                         while ((count = fread(buf, sizeof(char), sizeof(buf), f)))
 449                                 (void)fwrite(buf, sizeof(char), count, stdout);
 450                         (void)fclose(f);
 451                         break;
 452                 }
 453         if (ferror(stdout))
 454                 errx(1, "stdout: %s", strerror(errno ? errno : EIO));
 455         appendx = sdone = 0;
 456 }
 457
 458 static void
 459 lputs(s)
 460         register char *s;
 461 {
 462         register int count;
 463         register char *escapes, *p;
 464         struct winsize win;
 465         static int termwidth = -1;
 466
 467         if (termwidth == -1) {
 468                 if ((p = getenv("COLUMNS")))
 469                         termwidth = atoi(p);
 470                 else if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &win) == 0 &&
 471                     win.ws_col > 0)
 472                         termwidth = win.ws_col;
 473                 else
 474                         termwidth = 60;
 475         }
 476
 477         for (count = 0; *s; ++s) {
 478                 if (count >= termwidth) {
 479                         (void)printf("\\\n");
 480                         count = 0;
 481                 }
 482                 if (isprint((unsigned char)*s) && *s != '\\') {
 483                         (void)putchar(*s);
 484                         count++;
 485                 } else {
 486                         escapes = "\\\a\b\f\n\r\t\v";
 487                         (void)putchar('\\');
 488                         if ((p = strchr(escapes, *s))) {
 489                                 (void)putchar("\\abfnrtv"[p - escapes]);
 490                                 count += 2;
 491                         } else {
 492                                 (void)printf("%03o", *(u_char *)s);
 493                                 count += 4;
 494                         }
 495                 }
 496         }
 497         (void)putchar('$');
 498         (void)putchar('\n');
 499         if (ferror(stdout))
 500                 errx(1, "stdout: %s", strerror(errno ? errno : EIO));
 501 }
 502
 503 static inline int
 504 regexec_e(preg, string, eflags, nomatch, slen)
 505         regex_t *preg;
 506         const char *string;
 507         int eflags, nomatch;
 508         size_t slen;
 509 {
 510         int eval;
 511
 512         if (preg == NULL) {
 513                 if (defpreg == NULL)
 514                         errx(1, "first RE may not be empty");
 515         } else
 516                 defpreg = preg;
 517
 518         /* Set anchors, discounting trailing newline (if any). */
 519         if (slen > 0 && string[slen - 1] == '\n')
 520                 slen--;
 521         match[0].rm_so = 0;
 522         match[0].rm_eo = slen;
 523
 524         eval = regexec(defpreg, string,
 525             nomatch ? 0 : maxnsub + 1, match, eflags | REG_STARTEND);
 526         switch(eval) {
 527         case 0:
 528                 return (1);
 529         case REG_NOMATCH:
 530                 return (0);
 531         }
 532         errx(1, "RE error: %s", strregerror(eval, defpreg));
 533         /* NOTREACHED */
 534 }
 535
 536 /*
 537  * regsub - perform substitutions after a regexp match
 538  * Based on a routine by Henry Spencer
 539  */
 540 static void
 541 regsub(sp, string, src)
 542         SPACE *sp;
 543         char *string, *src;
 544 {
 545         register int len, no;
 546         register char c, *dst;
 547
 548 #define NEEDSP(reqlen)                                                  \
 549         if (sp->len >= sp->blen - (reqlen) - 1) {                       \
 550                 sp->blen += (reqlen) + 1024;                            \
 551                 sp->space = sp->back = xrealloc(sp->back, sp->blen);    \
 552                 dst = sp->space + sp->len;                              \
 553         }
 554
 555         dst = sp->space + sp->len;
 556         while ((c = *src++) != '\0') {
 557                 if (c == '&')
 558                         no = 0;
 559                 else if (c == '\\' && isdigit((unsigned char)*src))
 560                         no = *src++ - '0';
 561                 else
 562                         no = -1;
 563                 if (no < 0) {           /* Ordinary character. */
 564                         if (c == '\\' && (*src == '\\' || *src == '&'))
 565                                 c = *src++;
 566                         NEEDSP(1);
 567                         *dst++ = c;
 568                         ++sp->len;
 569                 } else if (match[no].rm_so != -1 && match[no].rm_eo != -1) {
 570                         len = match[no].rm_eo - match[no].rm_so;
 571                         NEEDSP(len);
 572                         memmove(dst, string + match[no].rm_so, len);
 573                         dst += len;
 574                         sp->len += len;
 575                 }
 576         }
 577         NEEDSP(1);
 578         *dst = '\0';
 579 }
 580
 581 /*
 582  * aspace --
 583  *      Append the source space to the destination space, allocating new
 584  *      space as necessary.
 585  */
 586 void
 587 cspace(sp, p, len, spflag)
 588         SPACE *sp;
 589         char *p;
 590         size_t len;
 591         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                 sp->blen = tlen + 1024;
 599                 sp->space = sp->back = xrealloc(sp->back, sp->blen);
 600         }
 601
 602         if (spflag == REPLACE)
 603                 sp->len = 0;
 604
 605         memmove(sp->space + sp->len, p, len);
 606
 607         sp->space[sp->len += len] = '\0';
 608 }
 609
 610 /*
 611  * Close all cached opened files and report any errors
 612  */
 613 void
 614 cfclose(cp, end)
 615         register struct s_command *cp, *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 }