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("$FreeBSD$");
  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         for (linenum = 0; mf_fgets(&PS, REPLACE);) {
 100                 pd = 0;
 101 top:
 102                 cp = prog;
 103 redirect:
 104                 while (cp != NULL) {
 105                         if (!applies(cp)) {
 106                                 cp = cp->next;
 107                                 continue;
 108                         }
 109                         switch (cp->code) {
 110                         case '{':
 111                                 cp = cp->u.c;
 112                                 goto redirect;
 113                         case 'a':
 114                                 if (appendx >= appendnum)
 115                                         if ((appends = realloc(appends,
 116                                             sizeof(struct s_appends) *
 117                                             (appendnum *= 2))) == NULL)
 118                                                 err(1, "realloc");
 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 (psl == 0 ||
 140                                     (p = memchr(ps, '\n', psl - 1)) == NULL) {
 141                                         pd = 1;
 142                                         goto new;
 143                                 } else {
 144                                         psl -= (p + 1) - ps;
 145                                         memmove(ps, p + 1, psl);
 146                                         goto top;
 147                                 }
 148                         case 'g':
 149                                 cspace(&PS, hs, hsl, REPLACE);
 150                                 break;
 151                         case 'G':
 152                                 cspace(&PS, "\n", 1, 0);
 153                                 cspace(&PS, hs, hsl, 0);
 154                                 break;
 155                         case 'h':
 156                                 cspace(&HS, ps, psl, REPLACE);
 157                                 break;
 158                         case 'H':
 159                                 cspace(&HS, "\n", 1, 0);
 160                                 cspace(&HS, ps, psl, 0);
 161                                 break;
 162                         case 'i':
 163                                 (void)printf("%s", cp->t);
 164                                 break;
 165                         case 'l':
 166                                 lputs(ps);
 167                                 break;
 168                         case 'n':
 169                                 if (!nflag && !pd)
 170                                         OUT(ps)
 171                                 flush_appends();
 172                                 if (!mf_fgets(&PS, REPLACE))
 173                                         exit(0);
 174                                 pd = 0;
 175                                 break;
 176                         case 'N':
 177                                 flush_appends();
 178                                 cspace(&PS, "\n", 1, 0);
 179                                 if (!mf_fgets(&PS, 0)) {
 180                                         if (!nflag && !pd)
 181                                                 OUT(ps)
 182                                         exit(0);
 183                                 }
 184                                 break;
 185                         case 'p':
 186                                 if (pd)
 187                                         break;
 188                                 OUT(ps)
 189                                 break;
 190                         case 'P':
 191                                 if (pd)
 192                                         break;
 193                                 if (psl != 0 &&
 194                                     (p = memchr(ps, '\n', psl - 1)) != NULL) {
 195                                         oldpsl = psl;
 196                                         psl = p - ps;
 197                                 }
 198                                 OUT(ps)
 199                                 if (p != NULL)
 200                                         psl = oldpsl;
 201                                 break;
 202                         case 'q':
 203                                 if (!nflag && !pd)
 204                                         OUT(ps)
 205                                 flush_appends();
 206                                 exit(0);
 207                         case 'r':
 208                                 if (appendx >= appendnum)
 209                                         if ((appends = realloc(appends,
 210                                             sizeof(struct s_appends) *
 211                                             (appendnum *= 2))) == NULL)
 212                                                 err(1, "realloc");
 213                                 appends[appendx].type = AP_FILE;
 214                                 appends[appendx].s = cp->t;
 215                                 appends[appendx].len = strlen(cp->t);
 216                                 appendx++;
 217                                 break;
 218                         case 's':
 219                                 sdone |= substitute(cp);
 220                                 break;
 221                         case 't':
 222                                 if (sdone) {
 223                                         sdone = 0;
 224                                         cp = cp->u.c;
 225                                         goto redirect;
 226                                 }
 227                                 break;
 228                         case 'w':
 229                                 if (pd)
 230                                         break;
 231                                 if (cp->u.fd == -1 && (cp->u.fd = open(cp->t,
 232                                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
 233                                     DEFFILEMODE)) == -1)
 234                                         err(1, "%s", cp->t);
 235                                 if (write(cp->u.fd, ps, psl) != psl ||
 236                                     write(cp->u.fd, "\n", 1) != 1)
 237                                         err(1, "%s", cp->t);
 238                                 break;
 239                         case 'x':
 240                                 tspace = PS;
 241                                 PS = HS;
 242                                 HS = tspace;
 243                                 break;
 244                         case 'y':
 245                                 if (pd || psl == 0)
 246                                         break;
 247                                 for (p = ps, len = psl; len--; ++p)
 248                                         *p = cp->u.y[(unsigned char)*p];
 249                                 break;
 250                         case ':':
 251                         case '}':
 252                                 break;
 253                         case '=':
 254                                 (void)printf("%lu\n", linenum);
 255                         }
 256                         cp = cp->next;
 257                 } /* for all cp */
 258
 259 new:            if (!nflag && !pd)
 260                         OUT(ps)
 261                 flush_appends();
 262         } /* for all lines */
 263 }
 264
 265 /*
 266  * TRUE if the address passed matches the current program state
 267  * (lastline, linenumber, ps).
 268  */
 269 #define MATCH(a)                                                \
 270         (a)->type == AT_RE ? regexec_e((a)->u.r, ps, 0, 1, psl) :       \
 271             (a)->type == AT_LINE ? linenum == (a)->u.l : lastline
 272
 273 /*
 274  * Return TRUE if the command applies to the current line.  Sets the inrange
 275  * flag to process ranges.  Interprets the non-select (``!'') flag.
 276  */
 277 static inline int
 278 applies(cp)
 279         struct s_command *cp;
 280 {
 281         int r;
 282
 283         lastaddr = 0;
 284         if (cp->a1 == NULL && cp->a2 == NULL)
 285                 r = 1;
 286         else if (cp->a2)
 287                 if (cp->inrange) {
 288                         if (MATCH(cp->a2)) {
 289                                 cp->inrange = 0;
 290                                 lastaddr = 1;
 291                         }
 292                         r = 1;
 293                 } else if (MATCH(cp->a1)) {
 294                         /*
 295                          * If the second address is a number less than or
 296                          * equal to the line number first selected, only
 297                          * one line shall be selected.
 298                          *      -- POSIX 1003.2
 299                          */
 300                         if (cp->a2->type == AT_LINE &&
 301                             linenum >= cp->a2->u.l)
 302                                 lastaddr = 1;
 303                         else
 304                                 cp->inrange = 1;
 305                         r = 1;
 306                 } else
 307                         r = 0;
 308         else
 309                 r = MATCH(cp->a1);
 310         return (cp->nonsel ? ! r : r);
 311 }
 312
 313 /*
 314  * substitute --
 315  *      Do substitutions in the pattern space.  Currently, we build a
 316  *      copy of the new pattern space in the substitute space structure
 317  *      and then swap them.
 318  */
 319 static int
 320 substitute(cp)
 321         struct s_command *cp;
 322 {
 323         SPACE tspace;
 324         regex_t *re;
 325         size_t re_off, slen;
 326         int lastempty, n;
 327         char *s;
 328
 329         s = ps;
 330         re = cp->u.s->re;
 331         if (re == NULL) {
 332                 if (defpreg != NULL && cp->u.s->maxbref > defpreg->re_nsub) {
 333                         linenum = cp->u.s->linenum;
 334                         errx(1, "%lu: %s: \\%d not defined in the RE",
 335                                         linenum, fname, cp->u.s->maxbref);
 336                 }
 337         }
 338         if (!regexec_e(re, s, 0, 0, psl))
 339                 return (0);
 340
 341         SS.len = 0;                             /* Clean substitute space. */
 342         slen = psl;
 343         n = cp->u.s->n;
 344         lastempty = 1;
 345
 346         switch (n) {
 347         case 0:                                 /* Global */
 348                 do {
 349                         if (lastempty || match[0].rm_so != match[0].rm_eo) {
 350                                 /* Locate start of replaced string. */
 351                                 re_off = match[0].rm_so;
 352                                 /* Copy leading retained string. */
 353                                 cspace(&SS, s, re_off, APPEND);
 354                                 /* Add in regular expression. */
 355                                 regsub(&SS, s, cp->u.s->new);
 356                         }
 357
 358                         /* Move past this match. */
 359                         if (match[0].rm_so != match[0].rm_eo) {
 360                                 s += match[0].rm_eo;
 361                                 slen -= match[0].rm_eo;
 362                                 lastempty = 0;
 363                         } else if (match[0].rm_so == slen) {
 364                                 s += match[0].rm_so;
 365                                 slen = 0;
 366                         } else {
 367                                 if (match[0].rm_so == 0)
 368                                         cspace(&SS, s, match[0].rm_so + 1,
 369                                             APPEND);
 370                                 else
 371                                         cspace(&SS, s + match[0].rm_so, 1,
 372                                             APPEND);
 373                                 s += match[0].rm_so + 1;
 374                                 slen -= match[0].rm_so + 1;
 375                                 lastempty = 1;
 376                         }
 377                 } while (slen > 0 && regexec_e(re, s, REG_NOTBOL, 0, slen));
 378                 /* Copy trailing retained string. */
 379                 if (slen > 0)
 380                         cspace(&SS, s, slen, APPEND);
 381                 break;
 382         default:                                /* Nth occurrence */
 383                 while (--n) {
 384                         s += match[0].rm_eo;
 385                         slen -= match[0].rm_eo;
 386                         if (!regexec_e(re, s, REG_NOTBOL, 0, slen))
 387                                 return (0);
 388                 }
 389                 /* FALLTHROUGH */
 390         case 1:                                 /* 1st occurrence */
 391                 /* Locate start of replaced string. */
 392                 re_off = match[0].rm_so + (s - ps);
 393                 /* Copy leading retained string. */
 394                 cspace(&SS, ps, re_off, APPEND);
 395                 /* Add in regular expression. */
 396                 regsub(&SS, s, cp->u.s->new);
 397                 /* Copy trailing retained string. */
 398                 s += match[0].rm_eo;
 399                 slen -= match[0].rm_eo;
 400                 cspace(&SS, s, slen, APPEND);
 401                 break;
 402         }
 403
 404         /*
 405          * Swap the substitute space and the pattern space, and make sure
 406          * that any leftover pointers into stdio memory get lost.
 407          */
 408         tspace = PS;
 409         PS = SS;
 410         SS = tspace;
 411         SS.space = SS.back;
 412
 413         /* Handle the 'p' flag. */
 414         if (cp->u.s->p)
 415                 OUT(ps)
 416
 417         /* Handle the 'w' flag. */
 418         if (cp->u.s->wfile && !pd) {
 419                 if (cp->u.s->wfd == -1 && (cp->u.s->wfd = open(cp->u.s->wfile,
 420                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, DEFFILEMODE)) == -1)
 421                         err(1, "%s", cp->u.s->wfile);
 422                 if (write(cp->u.s->wfd, ps, psl) != psl ||
 423                     write(cp->u.s->wfd, "\n", 1) != 1)
 424                         err(1, "%s", cp->u.s->wfile);
 425         }
 426         return (1);
 427 }
 428
 429 /*
 430  * Flush append requests.  Always called before reading a line,
 431  * therefore it also resets the substitution done (sdone) flag.
 432  */
 433 static void
 434 flush_appends()
 435 {
 436         FILE *f;
 437         int count, i;
 438         char buf[8 * 1024];
 439
 440         for (i = 0; i < appendx; i++)
 441                 switch (appends[i].type) {
 442                 case AP_STRING:
 443                         fwrite(appends[i].s, sizeof(char), appends[i].len,
 444                             stdout);
 445                         break;
 446                 case AP_FILE:
 447                         /*
 448                          * Read files probably shouldn't be cached.  Since
 449                          * it's not an error to read a non-existent file,
 450                          * it's possible that another program is interacting
 451                          * with the sed script through the filesystem.  It
 452                          * would be truly bizarre, but possible.  It's probably
 453                          * not that big a performance win, anyhow.
 454                          */
 455                         if ((f = fopen(appends[i].s, "r")) == NULL)
 456                                 break;
 457                         while ((count = fread(buf, sizeof(char), sizeof(buf), f)))
 458                                 (void)fwrite(buf, sizeof(char), count, stdout);
 459                         (void)fclose(f);
 460                         break;
 461                 }
 462         if (ferror(stdout))
 463                 errx(1, "stdout: %s", strerror(errno ? errno : EIO));
 464         appendx = sdone = 0;
 465 }
 466
 467 static void
 468 lputs(s)
 469         char *s;
 470 {
 471         int count;
 472         char *escapes, *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         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 }