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 #include <sys/cdefs.h>
  39 __FBSDID("$FreeBSD: src/usr.bin/sed/process.c,v 1.10.2.10 2003/06/13 07:32:08 fanf Exp $");
  40
  41 #ifndef lint
  42 static const char sccsid[] = "@(#)process.c     8.6 (Berkeley) 4/20/94";
  43 #endif
  44
  45 #include <sys/types.h>
  46 #include <sys/stat.h>
  47 #include <sys/ioctl.h>
  48 #include <sys/uio.h>
  49
  50 #include <ctype.h>
  51 #include <err.h>
  52 #include <errno.h>
  53 #include <fcntl.h>
  54 #include <limits.h>
  55 #include <regex.h>
  56 #include <stdio.h>
  57 #include <stdlib.h>
  58 #include <string.h>
  59 #include <unistd.h>
  60
  61 #include "defs.h"
  62 #include "extern.h"
  63
  64 static SPACE HS, PS, SS;
  65 #define pd              PS.deleted
  66 #define ps              PS.space
  67 #define psl             PS.len
  68 #define hs              HS.space
  69 #define hsl             HS.len
  70
  71 static __inline int      applies(struct s_command *);
  72 static void              flush_appends(void);
  73 static void              lputs(char *);
  74 static __inline 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(s) { fwrite(s, sizeof(u_char), psl, stdout); putchar('\n'); }
  90
  91 void
  92 process()
  93 {
  94         struct s_command *cp;
  95         SPACE tspace;
  96         size_t len, oldpsl = 0;
  97         char *p;
  98
  99         p = NULL;
 100
 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;
 110                         }
 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)
 133                                         (void)printf("%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;
 149                                 }
 150                         case 'g':
 151                                 cspace(&PS, hs, hsl, REPLACE);
 152                                 break;
 153                         case 'G':
 154                                 cspace(&PS, "\n", 1, 0);
 155                                 cspace(&PS, hs, hsl, 0);
 156                                 break;
 157                         case 'h':
 158                                 cspace(&HS, ps, psl, REPLACE);
 159                                 break;
 160                         case 'H':
 161                                 cspace(&HS, "\n", 1, 0);
 162                                 cspace(&HS, ps, psl, 0);
 163                                 break;
 164                         case 'i':
 165                                 (void)printf("%s", cp->t);
 166                                 break;
 167                         case 'l':
 168                                 lputs(ps);
 169                                 break;
 170                         case 'n':
 171                                 if (!nflag && !pd)
 172                                         OUT(ps)
 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, 0);
 181                                 if (!mf_fgets(&PS, 0))
 182                                         exit(0);
 183                                 break;
 184                         case 'p':
 185                                 if (pd)
 186                                         break;
 187                                 OUT(ps)
 188                                 break;
 189                         case 'P':
 190                                 if (pd)
 191                                         break;
 192                                 if (psl != 0 &&
 193                                     (p = memchr(ps, '\n', psl)) != NULL) {
 194                                         oldpsl = psl;
 195                                         psl = p - ps;
 196                                 }
 197                                 OUT(ps)
 198                                 if (p != NULL)
 199                                         psl = oldpsl;
 200                                 break;
 201                         case 'q':
 202                                 if (!nflag && !pd)
 203                                         OUT(ps)
 204                                 flush_appends();
 205                                 exit(0);
 206                         case 'r':
 207                                 if (appendx >= appendnum)
 208                                         if ((appends = realloc(appends,
 209                                             sizeof(struct s_appends) *
 210                                             (appendnum *= 2))) == NULL)
 211                                                 err(1, "realloc");
 212                                 appends[appendx].type = AP_FILE;
 213                                 appends[appendx].s = cp->t;
 214                                 appends[appendx].len = strlen(cp->t);
 215                                 appendx++;
 216                                 break;
 217                         case 's':
 218                                 sdone |= substitute(cp);
 219                                 break;
 220                         case 't':
 221                                 if (sdone) {
 222                                         sdone = 0;
 223                                         cp = cp->u.c;
 224                                         goto redirect;
 225                                 }
 226                                 break;
 227                         case 'w':
 228                                 if (pd)
 229                                         break;
 230                                 if (cp->u.fd == -1 && (cp->u.fd = open(cp->t,
 231                                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
 232                                     DEFFILEMODE)) == -1)
 233                                         err(1, "%s", cp->t);
 234                                 if (write(cp->u.fd, ps, psl) != psl ||
 235                                     write(cp->u.fd, "\n", 1) != 1)
 236                                         err(1, "%s", cp->t);
 237                                 break;
 238                         case 'x':
 239                                 if (hs == NULL)
 240                                         cspace(&HS, "", 0, REPLACE);
 241                                 tspace = PS;
 242                                 PS = HS;
 243                                 HS = tspace;
 244                                 break;
 245                         case 'y':
 246                                 if (pd || psl == 0)
 247                                         break;
 248                                 for (p = ps, len = psl; len--; ++p)
 249                                         *p = cp->u.y[(unsigned char)*p];
 250                                 break;
 251                         case ':':
 252                         case '}':
 253                                 break;
 254                         case '=':
 255                                 (void)printf("%lu\n", linenum);
 256                         }
 257                         cp = cp->next;
 258                 } /* for all cp */
 259
 260 new:            if (!nflag && !pd)
 261                         OUT(ps)
 262                 flush_appends();
 263         } /* for all lines */
 264 }
 265
 266 /*
 267  * TRUE if the address passed matches the current program state
 268  * (lastline, linenumber, ps).
 269  */
 270 #define MATCH(a)                                                \
 271         (a)->type == AT_RE ? regexec_e((a)->u.r, ps, 0, 1, psl) :       \
 272             (a)->type == AT_LINE ? linenum == (a)->u.l : lastline()
 273
 274 /*
 275  * Return TRUE if the command applies to the current line.  Sets the inrange
 276  * flag to process ranges.  Interprets the non-select (``!'') flag.
 277  */
 278 static __inline int
 279 applies(cp)
 280         struct s_command *cp;
 281 {
 282         int r;
 283
 284         lastaddr = 0;
 285         if (cp->a1 == NULL && cp->a2 == NULL)
 286                 r = 1;
 287         else if (cp->a2)
 288                 if (cp->inrange) {
 289                         if (MATCH(cp->a2)) {
 290                                 cp->inrange = 0;
 291                                 lastaddr = 1;
 292                         }
 293                         r = 1;
 294                 } else if (MATCH(cp->a1)) {
 295                         /*
 296                          * If the second address is a number less than or
 297                          * equal to the line number first selected, only
 298                          * one line shall be selected.
 299                          *      -- POSIX 1003.2
 300                          */
 301                         if (cp->a2->type == AT_LINE &&
 302                             linenum >= cp->a2->u.l)
 303                                 lastaddr = 1;
 304                         else
 305                                 cp->inrange = 1;
 306                         r = 1;
 307                 } else
 308                         r = 0;
 309         else
 310                 r = MATCH(cp->a1);
 311         return (cp->nonsel ? ! r : r);
 312 }
 313
 314 /*
 315  * substitute --
 316  *      Do substitutions in the pattern space.  Currently, we build a
 317  *      copy of the new pattern space in the substitute space structure
 318  *      and then swap them.
 319  */
 320 static int
 321 substitute(cp)
 322         struct s_command *cp;
 323 {
 324         SPACE tspace;
 325         regex_t *re;
 326         regoff_t re_off, slen;
 327         int lastempty, n;
 328         char *s;
 329
 330         s = ps;
 331         re = cp->u.s->re;
 332         if (re == NULL) {
 333                 if (defpreg != NULL && cp->u.s->maxbref > defpreg->re_nsub) {
 334                         linenum = cp->u.s->linenum;
 335                         errx(1, "%lu: %s: \\%d not defined in the RE",
 336                                         linenum, fname, cp->u.s->maxbref);
 337                 }
 338         }
 339         if (!regexec_e(re, s, 0, 0, psl))
 340                 return (0);
 341
 342         SS.len = 0;                             /* Clean substitute space. */
 343         slen = psl;
 344         n = cp->u.s->n;
 345         lastempty = 1;
 346
 347         switch (n) {
 348         case 0:                                 /* Global */
 349                 do {
 350                         if (lastempty || match[0].rm_so != match[0].rm_eo) {
 351                                 /* Locate start of replaced string. */
 352                                 re_off = match[0].rm_so;
 353                                 /* Copy leading retained string. */
 354                                 cspace(&SS, s, re_off, APPEND);
 355                                 /* Add in regular expression. */
 356                                 regsub(&SS, s, cp->u.s->new);
 357                         }
 358
 359                         /* Move past this match. */
 360                         if (match[0].rm_so != match[0].rm_eo) {
 361                                 s += match[0].rm_eo;
 362                                 slen -= match[0].rm_eo;
 363                                 lastempty = 0;
 364                         } else {
 365                                 if (match[0].rm_so < slen)
 366                                         cspace(&SS, s + match[0].rm_so, 1,
 367                                             APPEND);
 368                                 s += match[0].rm_so + 1;
 369                                 slen -= match[0].rm_so + 1;
 370                                 lastempty = 1;
 371                         }
 372                 } while (slen >= 0 && regexec_e(re, s, REG_NOTBOL, 0, slen));
 373                 /* Copy trailing retained string. */
 374                 if (slen > 0)
 375                         cspace(&SS, s, slen, APPEND);
 376                 break;
 377         default:                                /* Nth occurrence */
 378                 while (--n) {
 379                         if (match[0].rm_eo == match[0].rm_so)
 380                                 match[0].rm_eo = match[0].rm_so + 1;
 381                         s += match[0].rm_eo;
 382                         slen -= match[0].rm_eo;
 383                         if (slen < 0)
 384                                 return (0);
 385                         if (!regexec_e(re, s, REG_NOTBOL, 0, slen))
 386                                 return (0);
 387                 }
 388                 /* FALLTHROUGH */
 389         case 1:                                 /* 1st occurrence */
 390                 /* Locate start of replaced string. */
 391                 re_off = match[0].rm_so + (s - ps);
 392                 /* Copy leading retained string. */
 393                 cspace(&SS, ps, re_off, APPEND);
 394                 /* Add in regular expression. */
 395                 regsub(&SS, s, cp->u.s->new);
 396                 /* Copy trailing retained string. */
 397                 s += match[0].rm_eo;
 398                 slen -= match[0].rm_eo;
 399                 cspace(&SS, s, slen, APPEND);
 400                 break;
 401         }
 402
 403         /*
 404          * Swap the substitute space and the pattern space, and make sure
 405          * that any leftover pointers into stdio memory get lost.
 406          */
 407         tspace = PS;
 408         PS = SS;
 409         SS = tspace;
 410         SS.space = SS.back;
 411
 412         /* Handle the 'p' flag. */
 413         if (cp->u.s->p)
 414                 OUT(ps)
 415
 416         /* Handle the 'w' flag. */
 417         if (cp->u.s->wfile && !pd) {
 418                 if (cp->u.s->wfd == -1 && (cp->u.s->wfd = open(cp->u.s->wfile,
 419                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, DEFFILEMODE)) == -1)
 420                         err(1, "%s", cp->u.s->wfile);
 421                 if (write(cp->u.s->wfd, ps, psl) != psl ||
 422                     write(cp->u.s->wfd, "\n", 1) != 1)
 423                         err(1, "%s", cp->u.s->wfile);
 424         }
 425         return (1);
 426 }
 427
 428 /*
 429  * Flush append requests.  Always called before reading a line,
 430  * therefore it also resets the substitution done (sdone) flag.
 431  */
 432 static void
 433 flush_appends()
 434 {
 435         FILE *f;
 436         int count, i;
 437         char buf[8 * 1024];
 438
 439         for (i = 0; i < appendx; i++)
 440                 switch (appends[i].type) {
 441                 case AP_STRING:
 442                         fwrite(appends[i].s, sizeof(char), appends[i].len,
 443                             stdout);
 444                         break;
 445                 case AP_FILE:
 446                         /*
 447                          * Read files probably shouldn't be cached.  Since
 448                          * it's not an error to read a non-existent file,
 449                          * it's possible that another program is interacting
 450                          * with the sed script through the filesystem.  It
 451                          * would be truly bizarre, but possible.  It's probably
 452                          * not that big a performance win, anyhow.
 453                          */
 454                         if ((f = fopen(appends[i].s, "r")) == NULL)
 455                                 break;
 456                         while ((count = fread(buf, sizeof(char), sizeof(buf), f)))
 457                                 (void)fwrite(buf, sizeof(char), count, stdout);
 458                         (void)fclose(f);
 459                         break;
 460                 }
 461         if (ferror(stdout))
 462                 errx(1, "stdout: %s", strerror(errno ? errno : EIO));
 463         appendx = sdone = 0;
 464 }
 465
 466 static void
 467 lputs(s)
 468         char *s;
 469 {
 470         int count;
 471         const char *escapes;
 472         char *p;
 473         struct winsize win;
 474         static int termwidth = -1;
 475
 476         if (termwidth == -1) {
 477                 if ((p = getenv("COLUMNS")) && *p != '\0')
 478                         termwidth = atoi(p);
 479                 else if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &win) == 0 &&
 480                     win.ws_col > 0)
 481                         termwidth = win.ws_col;
 482                 else
 483                         termwidth = 60;
 484         }
 485
 486         for (count = 0; *s; ++s) {
 487                 if (count + 5 >= termwidth) {
 488                         (void)printf("\\\n");
 489                         count = 0;
 490                 }
 491                 if (isprint((unsigned char)*s) && *s != '\\') {
 492                         (void)putchar(*s);
 493                         count++;
 494                 } else if (*s == '\n') {
 495                         (void)putchar('$');
 496                         (void)putchar('\n');
 497                         count = 0;
 498                 } else {
 499                         escapes = "\\\a\b\f\r\t\v";
 500                         (void)putchar('\\');
 501                         if ((p = strchr(escapes, *s))) {
 502                                 (void)putchar("\\abfrtv"[p - escapes]);
 503                                 count += 2;
 504                         } else {
 505                                 (void)printf("%03o", *(u_char *)s);
 506                                 count += 4;
 507                         }
 508                 }
 509         }
 510         (void)putchar('$');
 511         (void)putchar('\n');
 512         if (ferror(stdout))
 513                 errx(1, "stdout: %s", strerror(errno ? errno : EIO));
 514 }
 515
 516 static __inline int
 517 regexec_e(preg, string, eflags, nomatch, slen)
 518         regex_t *preg;
 519         const char *string;
 520         int eflags, nomatch;
 521         size_t slen;
 522 {
 523         int eval;
 524
 525         if (preg == NULL) {
 526                 if (defpreg == NULL)
 527                         errx(1, "first RE may not be empty");
 528         } else
 529                 defpreg = preg;
 530
 531         /* Set anchors */
 532         match[0].rm_so = 0;
 533         match[0].rm_eo = slen;
 534
 535         eval = regexec(defpreg, string,
 536             nomatch ? 0 : maxnsub + 1, match, eflags | REG_STARTEND);
 537         switch(eval) {
 538         case 0:
 539                 return (1);
 540         case REG_NOMATCH:
 541                 return (0);
 542         }
 543         errx(1, "RE error: %s", strregerror(eval, defpreg));
 544         /* NOTREACHED */
 545 }
 546
 547 /*
 548  * regsub - perform substitutions after a regexp match
 549  * Based on a routine by Henry Spencer
 550  */
 551 static void
 552 regsub(sp, string, src)
 553         SPACE *sp;
 554         char *string, *src;
 555 {
 556         int len, no;
 557         char c, *dst;
 558
 559 #define NEEDSP(reqlen)                                                  \
 560         if (sp->len >= sp->blen - (reqlen) - 1) {                       \
 561                 sp->blen += (reqlen) + 1024;                            \
 562                 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) \
 563                     == NULL)                                            \
 564                         err(1, "realloc");                              \
 565                 dst = sp->space + sp->len;                              \
 566         }
 567
 568         dst = sp->space + sp->len;
 569         while ((c = *src++) != '\0') {
 570                 if (c == '&')
 571                         no = 0;
 572                 else if (c == '\\' && isdigit((unsigned char)*src))
 573                         no = *src++ - '0';
 574                 else
 575                         no = -1;
 576                 if (no < 0) {           /* Ordinary character. */
 577                         if (c == '\\' && (*src == '\\' || *src == '&'))
 578                                 c = *src++;
 579                         NEEDSP(1);
 580                         *dst++ = c;
 581                         ++sp->len;
 582                 } else if (match[no].rm_so != -1 && match[no].rm_eo != -1) {
 583                         len = match[no].rm_eo - match[no].rm_so;
 584                         NEEDSP(len);
 585                         memmove(dst, string + match[no].rm_so, len);
 586                         dst += len;
 587                         sp->len += len;
 588                 }
 589         }
 590         NEEDSP(1);
 591         *dst = '\0';
 592 }
 593
 594 /*
 595  * aspace --
 596  *      Append the source space to the destination space, allocating new
 597  *      space as necessary.
 598  */
 599 void
 600 cspace(sp, p, len, spflag)
 601         SPACE *sp;
 602         const char *p;
 603         size_t len;
 604         enum e_spflag spflag;
 605 {
 606         size_t tlen;
 607
 608         /* Make sure SPACE has enough memory and ramp up quickly. */
 609         tlen = sp->len + len + 1;
 610         if (tlen > sp->blen) {
 611                 sp->blen = tlen + 1024;
 612                 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) ==
 613                     NULL)
 614                         err(1, "realloc");
 615         }
 616
 617         if (spflag == REPLACE)
 618                 sp->len = 0;
 619
 620         memmove(sp->space + sp->len, p, len);
 621
 622         sp->space[sp->len += len] = '\0';
 623 }
 624
 625 /*
 626  * Close all cached opened files and report any errors
 627  */
 628 void
 629 cfclose(cp, end)
 630         struct s_command *cp, *end;
 631 {
 632
 633         for (; cp != end; cp = cp->next)
 634                 switch(cp->code) {
 635                 case 's':
 636                         if (cp->u.s->wfd != -1 && close(cp->u.s->wfd))
 637                                 err(1, "%s", cp->u.s->wfile);
 638                         cp->u.s->wfd = -1;
 639                         break;
 640                 case 'w':
 641                         if (cp->u.fd != -1 && close(cp->u.fd))
 642                                 err(1, "%s", cp->t);
 643                         cp->u.fd = -1;
 644                         break;
 645                 case '{':
 646                         cfclose(cp->u.c, cp->next);
 647                         break;
 648                 }
 649 }