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  * @(#)process.c        8.6 (Berkeley) 4/20/94
  38  * $FreeBSD: src/usr.bin/sed/process.c,v 1.10.2.10 2003/06/13 07:32:08 fanf Exp $
  39  * $DragonFly: src/usr.bin/sed/process.c,v 1.2 2003/06/17 04:29:31 dillon Exp $
  40  */
  41
  42 #include <sys/types.h>
  43 #include <sys/stat.h>
  44 #include <sys/ioctl.h>
  45 #include <sys/uio.h>
  46
  47 #include <ctype.h>
  48 #include <err.h>
  49 #include <errno.h>
  50 #include <fcntl.h>
  51 #include <limits.h>
  52 #include <regex.h>
  53 #include <stdio.h>
  54 #include <stdlib.h>
  55 #include <string.h>
  56 #include <unistd.h>
  57
  58 #include "defs.h"
  59 #include "extern.h"
  60
  61 static SPACE HS, PS, SS;
  62 #define pd              PS.deleted
  63 #define ps              PS.space
  64 #define psl             PS.len
  65 #define hs              HS.space
  66 #define hsl             HS.len
  67
  68 static __inline int      applies(struct s_command *);
  69 static void              flush_appends(void);
  70 static void              lputs(char *);
  71 static __inline int      regexec_e(regex_t *, const char *, int, int, size_t);
  72 static void              regsub(SPACE *, char *, char *);
  73 static int               substitute(struct s_command *);
  74
  75 struct s_appends *appends;      /* Array of pointers to strings to append. */
  76 static int appendx;             /* Index into appends array. */
  77 int appendnum;                  /* Size of appends array. */
  78
  79 static int lastaddr;            /* Set by applies if last address of a range. */
  80 static int sdone;               /* If any substitutes since last line input. */
  81                                 /* Iov structure for 'w' commands. */
  82 static regex_t *defpreg;
  83 size_t maxnsub;
  84 regmatch_t *match;
  85
  86 #define OUT(s) { fwrite(s, sizeof(u_char), psl, stdout); putchar('\n'); }
  87
  88 void
  89 process()
  90 {
  91         struct s_command *cp;
  92         SPACE tspace;
  93         size_t len, oldpsl = 0;
  94         char *p;
  95
  96         p = NULL;
  97
  98         for (linenum = 0; mf_fgets(&PS, REPLACE);) {
  99                 pd = 0;
 100 top:
 101                 cp = prog;
 102 redirect:
 103                 while (cp != NULL) {
 104                         if (!applies(cp)) {
 105                                 cp = cp->next;
 106                                 continue;
 107                         }
 108                         switch (cp->code) {
 109                         case '{':
 110                                 cp = cp->u.c;
 111                                 goto redirect;
 112                         case 'a':
 113                                 if (appendx >= appendnum)
 114                                         if ((appends = realloc(appends,
 115                                             sizeof(struct s_appends) *
 116                                             (appendnum *= 2))) == NULL)
 117                                                 err(1, "realloc");
 118                                 appends[appendx].type = AP_STRING;
 119                                 appends[appendx].s = cp->t;
 120                                 appends[appendx].len = strlen(cp->t);
 121                                 appendx++;
 122                                 break;
 123                         case 'b':
 124                                 cp = cp->u.c;
 125                                 goto redirect;
 126                         case 'c':
 127                                 pd = 1;
 128                                 psl = 0;
 129                                 if (cp->a2 == NULL || lastaddr)
 130                                         (void)printf("%s", cp->t);
 131                                 break;
 132                         case 'd':
 133                                 pd = 1;
 134                                 goto new;
 135                         case 'D':
 136                                 if (pd)
 137                                         goto new;
 138                                 if (psl == 0 ||
 139                                     (p = memchr(ps, '\n', psl)) == 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, "\n", 1, 0);
 152                                 cspace(&PS, hs, hsl, 0);
 153                                 break;
 154                         case 'h':
 155                                 cspace(&HS, ps, psl, REPLACE);
 156                                 break;
 157                         case 'H':
 158                                 cspace(&HS, "\n", 1, 0);
 159                                 cspace(&HS, ps, psl, 0);
 160                                 break;
 161                         case 'i':
 162                                 (void)printf("%s", cp->t);
 163                                 break;
 164                         case 'l':
 165                                 lputs(ps);
 166                                 break;
 167                         case 'n':
 168                                 if (!nflag && !pd)
 169                                         OUT(ps)
 170                                 flush_appends();
 171                                 if (!mf_fgets(&PS, REPLACE))
 172                                         exit(0);
 173                                 pd = 0;
 174                                 break;
 175                         case 'N':
 176                                 flush_appends();
 177                                 cspace(&PS, "\n", 1, 0);
 178                                 if (!mf_fgets(&PS, 0))
 179                                         exit(0);
 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 (psl != 0 &&
 190                                     (p = memchr(ps, '\n', psl)) != NULL) {
 191                                         oldpsl = psl;
 192                                         psl = p - ps;
 193                                 }
 194                                 OUT(ps)
 195                                 if (p != NULL)
 196                                         psl = oldpsl;
 197                                 break;
 198                         case 'q':
 199                                 if (!nflag && !pd)
 200                                         OUT(ps)
 201                                 flush_appends();
 202                                 exit(0);
 203                         case 'r':
 204                                 if (appendx >= appendnum)
 205                                         if ((appends = realloc(appends,
 206                                             sizeof(struct s_appends) *
 207                                             (appendnum *= 2))) == NULL)
 208                                                 err(1, "realloc");
 209                                 appends[appendx].type = AP_FILE;
 210                                 appends[appendx].s = cp->t;
 211                                 appends[appendx].len = strlen(cp->t);
 212                                 appendx++;
 213                                 break;
 214                         case 's':
 215                                 sdone |= substitute(cp);
 216                                 break;
 217                         case 't':
 218                                 if (sdone) {
 219                                         sdone = 0;
 220                                         cp = cp->u.c;
 221                                         goto redirect;
 222                                 }
 223                                 break;
 224                         case 'w':
 225                                 if (pd)
 226                                         break;
 227                                 if (cp->u.fd == -1 && (cp->u.fd = open(cp->t,
 228                                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
 229                                     DEFFILEMODE)) == -1)
 230                                         err(1, "%s", cp->t);
 231                                 if (write(cp->u.fd, ps, psl) != psl ||
 232                                     write(cp->u.fd, "\n", 1) != 1)
 233                                         err(1, "%s", cp->t);
 234                                 break;
 235                         case 'x':
 236                                 if (hs == NULL)
 237                                         cspace(&HS, "", 0, REPLACE);
 238                                 tspace = PS;
 239                                 PS = HS;
 240                                 HS = tspace;
 241                                 break;
 242                         case 'y':
 243                                 if (pd || psl == 0)
 244                                         break;
 245                                 for (p = ps, len = psl; len--; ++p)
 246                                         *p = cp->u.y[(unsigned char)*p];
 247                                 break;
 248                         case ':':
 249                         case '}':
 250                                 break;
 251                         case '=':
 252                                 (void)printf("%lu\n", linenum);
 253                         }
 254                         cp = cp->next;
 255                 } /* for all cp */
 256
 257 new:            if (!nflag && !pd)
 258                         OUT(ps)
 259                 flush_appends();
 260         } /* for all lines */
 261 }
 262
 263 /*
 264  * TRUE if the address passed matches the current program state
 265  * (lastline, linenumber, ps).
 266  */
 267 #define MATCH(a)                                                \
 268         (a)->type == AT_RE ? regexec_e((a)->u.r, ps, 0, 1, psl) :       \
 269             (a)->type == AT_LINE ? linenum == (a)->u.l : lastline()
 270
 271 /*
 272  * Return TRUE if the command applies to the current line.  Sets the inrange
 273  * flag to process ranges.  Interprets the non-select (``!'') flag.
 274  */
 275 static __inline int
 276 applies(cp)
 277         struct s_command *cp;
 278 {
 279         int r;
 280
 281         lastaddr = 0;
 282         if (cp->a1 == NULL && cp->a2 == NULL)
 283                 r = 1;
 284         else if (cp->a2)
 285                 if (cp->inrange) {
 286                         if (MATCH(cp->a2)) {
 287                                 cp->inrange = 0;
 288                                 lastaddr = 1;
 289                         }
 290                         r = 1;
 291                 } else if (MATCH(cp->a1)) {
 292                         /*
 293                          * If the second address is a number less than or
 294                          * equal to the line number first selected, only
 295                          * one line shall be selected.
 296                          *      -- POSIX 1003.2
 297                          */
 298                         if (cp->a2->type == AT_LINE &&
 299                             linenum >= cp->a2->u.l)
 300                                 lastaddr = 1;
 301                         else
 302                                 cp->inrange = 1;
 303                         r = 1;
 304                 } else
 305                         r = 0;
 306         else
 307                 r = MATCH(cp->a1);
 308         return (cp->nonsel ? ! r : r);
 309 }
 310
 311 /*
 312  * substitute --
 313  *      Do substitutions in the pattern space.  Currently, we build a
 314  *      copy of the new pattern space in the substitute space structure
 315  *      and then swap them.
 316  */
 317 static int
 318 substitute(cp)
 319         struct s_command *cp;
 320 {
 321         SPACE tspace;
 322         regex_t *re;
 323         regoff_t re_off, slen;
 324         int lastempty, n;
 325         char *s;
 326
 327         s = ps;
 328         re = cp->u.s->re;
 329         if (re == NULL) {
 330                 if (defpreg != NULL && cp->u.s->maxbref > defpreg->re_nsub) {
 331                         linenum = cp->u.s->linenum;
 332                         errx(1, "%lu: %s: \\%d not defined in the RE",
 333                                         linenum, fname, cp->u.s->maxbref);
 334                 }
 335         }
 336         if (!regexec_e(re, s, 0, 0, psl))
 337                 return (0);
 338
 339         SS.len = 0;                             /* Clean substitute space. */
 340         slen = psl;
 341         n = cp->u.s->n;
 342         lastempty = 1;
 343
 344         switch (n) {
 345         case 0:                                 /* Global */
 346                 do {
 347                         if (lastempty || match[0].rm_so != match[0].rm_eo) {
 348                                 /* Locate start of replaced string. */
 349                                 re_off = match[0].rm_so;
 350                                 /* Copy leading retained string. */
 351                                 cspace(&SS, s, re_off, APPEND);
 352                                 /* Add in regular expression. */
 353                                 regsub(&SS, s, cp->u.s->new);
 354                         }
 355
 356                         /* Move past this match. */
 357                         if (match[0].rm_so != match[0].rm_eo) {
 358                                 s += match[0].rm_eo;
 359                                 slen -= match[0].rm_eo;
 360                                 lastempty = 0;
 361                         } else {
 362                                 if (match[0].rm_so < slen)
 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                         if (match[0].rm_eo == match[0].rm_so)
 377                                 match[0].rm_eo = match[0].rm_so + 1;
 378                         s += match[0].rm_eo;
 379                         slen -= match[0].rm_eo;
 380                         if (slen < 0)
 381                                 return (0);
 382                         if (!regexec_e(re, s, REG_NOTBOL, 0, slen))
 383                                 return (0);
 384                 }
 385                 /* FALLTHROUGH */
 386         case 1:                                 /* 1st occurrence */
 387                 /* Locate start of replaced string. */
 388                 re_off = match[0].rm_so + (s - ps);
 389                 /* Copy leading retained string. */
 390                 cspace(&SS, ps, re_off, APPEND);
 391                 /* Add in regular expression. */
 392                 regsub(&SS, s, cp->u.s->new);
 393                 /* Copy trailing retained string. */
 394                 s += match[0].rm_eo;
 395                 slen -= match[0].rm_eo;
 396                 cspace(&SS, s, slen, APPEND);
 397                 break;
 398         }
 399
 400         /*
 401          * Swap the substitute space and the pattern space, and make sure
 402          * that any leftover pointers into stdio memory get lost.
 403          */
 404         tspace = PS;
 405         PS = SS;
 406         SS = tspace;
 407         SS.space = SS.back;
 408
 409         /* Handle the 'p' flag. */
 410         if (cp->u.s->p)
 411                 OUT(ps)
 412
 413         /* Handle the 'w' flag. */
 414         if (cp->u.s->wfile && !pd) {
 415                 if (cp->u.s->wfd == -1 && (cp->u.s->wfd = open(cp->u.s->wfile,
 416                     O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, DEFFILEMODE)) == -1)
 417                         err(1, "%s", cp->u.s->wfile);
 418                 if (write(cp->u.s->wfd, ps, psl) != psl ||
 419                     write(cp->u.s->wfd, "\n", 1) != 1)
 420                         err(1, "%s", cp->u.s->wfile);
 421         }
 422         return (1);
 423 }
 424
 425 /*
 426  * Flush append requests.  Always called before reading a line,
 427  * therefore it also resets the substitution done (sdone) flag.
 428  */
 429 static void
 430 flush_appends()
 431 {
 432         FILE *f;
 433         int count, i;
 434         char buf[8 * 1024];
 435
 436         for (i = 0; i < appendx; i++)
 437                 switch (appends[i].type) {
 438                 case AP_STRING:
 439                         fwrite(appends[i].s, sizeof(char), appends[i].len,
 440                             stdout);
 441                         break;
 442                 case AP_FILE:
 443                         /*
 444                          * Read files probably shouldn't be cached.  Since
 445                          * it's not an error to read a non-existent file,
 446                          * it's possible that another program is interacting
 447                          * with the sed script through the filesystem.  It
 448                          * would be truly bizarre, but possible.  It's probably
 449                          * not that big a performance win, anyhow.
 450                          */
 451                         if ((f = fopen(appends[i].s, "r")) == NULL)
 452                                 break;
 453                         while ((count = fread(buf, sizeof(char), sizeof(buf), f)))
 454                                 (void)fwrite(buf, sizeof(char), count, stdout);
 455                         (void)fclose(f);
 456                         break;
 457                 }
 458         if (ferror(stdout))
 459                 errx(1, "stdout: %s", strerror(errno ? errno : EIO));
 460         appendx = sdone = 0;
 461 }
 462
 463 static void
 464 lputs(s)
 465         char *s;
 466 {
 467         int count;
 468         const char *escapes;
 469         char *p;
 470         struct winsize win;
 471         static int termwidth = -1;
 472
 473         if (termwidth == -1) {
 474                 if ((p = getenv("COLUMNS")) && *p != '\0')
 475                         termwidth = atoi(p);
 476                 else if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &win) == 0 &&
 477                     win.ws_col > 0)
 478                         termwidth = win.ws_col;
 479                 else
 480                         termwidth = 60;
 481         }
 482
 483         for (count = 0; *s; ++s) {
 484                 if (count + 5 >= termwidth) {
 485                         (void)printf("\\\n");
 486                         count = 0;
 487                 }
 488                 if (isprint((unsigned char)*s) && *s != '\\') {
 489                         (void)putchar(*s);
 490                         count++;
 491                 } else if (*s == '\n') {
 492                         (void)putchar('$');
 493                         (void)putchar('\n');
 494                         count = 0;
 495                 } else {
 496                         escapes = "\\\a\b\f\r\t\v";
 497                         (void)putchar('\\');
 498                         if ((p = strchr(escapes, *s))) {
 499                                 (void)putchar("\\abfrtv"[p - escapes]);
 500                                 count += 2;
 501                         } else {
 502                                 (void)printf("%03o", *(u_char *)s);
 503                                 count += 4;
 504                         }
 505                 }
 506         }
 507         (void)putchar('$');
 508         (void)putchar('\n');
 509         if (ferror(stdout))
 510                 errx(1, "stdout: %s", strerror(errno ? errno : EIO));
 511 }
 512
 513 static __inline int
 514 regexec_e(preg, string, eflags, nomatch, slen)
 515         regex_t *preg;
 516         const char *string;
 517         int eflags, nomatch;
 518         size_t slen;
 519 {
 520         int eval;
 521
 522         if (preg == NULL) {
 523                 if (defpreg == NULL)
 524                         errx(1, "first RE may not be empty");
 525         } else
 526                 defpreg = preg;
 527
 528         /* Set anchors */
 529         match[0].rm_so = 0;
 530         match[0].rm_eo = slen;
 531
 532         eval = regexec(defpreg, string,
 533             nomatch ? 0 : maxnsub + 1, match, eflags | REG_STARTEND);
 534         switch(eval) {
 535         case 0:
 536                 return (1);
 537         case REG_NOMATCH:
 538                 return (0);
 539         }
 540         errx(1, "RE error: %s", strregerror(eval, defpreg));
 541         /* NOTREACHED */
 542 }
 543
 544 /*
 545  * regsub - perform substitutions after a regexp match
 546  * Based on a routine by Henry Spencer
 547  */
 548 static void
 549 regsub(sp, string, src)
 550         SPACE *sp;
 551         char *string, *src;
 552 {
 553         int len, no;
 554         char c, *dst;
 555
 556 #define NEEDSP(reqlen)                                                  \
 557         if (sp->len >= sp->blen - (reqlen) - 1) {                       \
 558                 sp->blen += (reqlen) + 1024;                            \
 559                 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) \
 560                     == NULL)                                            \
 561                         err(1, "realloc");                              \
 562                 dst = sp->space + sp->len;                              \
 563         }
 564
 565         dst = sp->space + sp->len;
 566         while ((c = *src++) != '\0') {
 567                 if (c == '&')
 568                         no = 0;
 569                 else if (c == '\\' && isdigit((unsigned char)*src))
 570                         no = *src++ - '0';
 571                 else
 572                         no = -1;
 573                 if (no < 0) {           /* Ordinary character. */
 574                         if (c == '\\' && (*src == '\\' || *src == '&'))
 575                                 c = *src++;
 576                         NEEDSP(1);
 577                         *dst++ = c;
 578                         ++sp->len;
 579                 } else if (match[no].rm_so != -1 && match[no].rm_eo != -1) {
 580                         len = match[no].rm_eo - match[no].rm_so;
 581                         NEEDSP(len);
 582                         memmove(dst, string + match[no].rm_so, len);
 583                         dst += len;
 584                         sp->len += len;
 585                 }
 586         }
 587         NEEDSP(1);
 588         *dst = '\0';
 589 }
 590
 591 /*
 592  * aspace --
 593  *      Append the source space to the destination space, allocating new
 594  *      space as necessary.
 595  */
 596 void
 597 cspace(sp, p, len, spflag)
 598         SPACE *sp;
 599         const char *p;
 600         size_t len;
 601         enum e_spflag spflag;
 602 {
 603         size_t tlen;
 604
 605         /* Make sure SPACE has enough memory and ramp up quickly. */
 606         tlen = sp->len + len + 1;
 607         if (tlen > sp->blen) {
 608                 sp->blen = tlen + 1024;
 609                 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) ==
 610                     NULL)
 611                         err(1, "realloc");
 612         }
 613
 614         if (spflag == REPLACE)
 615                 sp->len = 0;
 616
 617         memmove(sp->space + sp->len, p, len);
 618
 619         sp->space[sp->len += len] = '\0';
 620 }
 621
 622 /*
 623  * Close all cached opened files and report any errors
 624  */
 625 void
 626 cfclose(cp, end)
 627         struct s_command *cp, *end;
 628 {
 629
 630         for (; cp != end; cp = cp->next)
 631                 switch(cp->code) {
 632                 case 's':
 633                         if (cp->u.s->wfd != -1 && close(cp->u.s->wfd))
 634                                 err(1, "%s", cp->u.s->wfile);
 635                         cp->u.s->wfd = -1;
 636                         break;
 637                 case 'w':
 638                         if (cp->u.fd != -1 && close(cp->u.fd))
 639                                 err(1, "%s", cp->t);
 640                         cp->u.fd = -1;
 641                         break;
 642                 case '{':
 643                         cfclose(cp->u.c, cp->next);
 644                         break;
 645                 }
 646 }