usr.bin/sed/compile.c

   1 /*-
   2  * Copyright (c) 1992 Diomidis Spinellis.
   3  * Copyright (c) 1992, 1993
   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. Neither the name of the University nor the names of its contributors
  18  *    may be used to endorse or promote products derived from this software
  19  *    without specific prior written permission.
  20  *
  21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  31  * SUCH DAMAGE.
  32  *
  33  * @(#)compile.c        8.1 (Berkeley) 6/6/93
  34  * $FreeBSD: head/usr.bin/sed/compile.c 289677 2015-10-21 05:37:09Z eadler $
  35  */
  36
  37 #include <sys/types.h>
  38 #include <sys/stat.h>
  39
  40 #include <ctype.h>
  41 #include <err.h>
  42 #include <errno.h>
  43 #include <fcntl.h>
  44 #include <limits.h>
  45 #include <regex.h>
  46 #include <stdio.h>
  47 #include <stdlib.h>
  48 #include <string.h>
  49 #include <wchar.h>
  50
  51 #include "defs.h"
  52 #include "extern.h"
  53
  54 #define LHSZ    128
  55 #define LHMASK  (LHSZ - 1)
  56 static struct labhash {
  57         struct  labhash *lh_next;
  58         u_int   lh_hash;
  59         struct  s_command *lh_cmd;
  60         int     lh_ref;
  61 } *labels[LHSZ];
  62
  63 static char      *compile_addr(char *, struct s_addr *);
  64 static char      *compile_ccl(char **, char *);
  65 static char      *compile_delimited(char *, char *, int);
  66 static char      *compile_flags(char *, struct s_subst *);
  67 static regex_t   *compile_re(char *, int);
  68 static char      *compile_subst(char *, struct s_subst *);
  69 static char      *compile_text(void);
  70 static char      *compile_tr(char *, struct s_tr **);
  71 static struct s_command
  72                 **compile_stream(struct s_command **);
  73 static char      *duptoeol(char *, const char *);
  74 static void       enterlabel(struct s_command *);
  75 static struct s_command
  76                  *findlabel(char *);
  77 static void       fixuplabel(struct s_command *, struct s_command *);
  78 static void       uselabel(void);
  79
  80 /*
  81  * Command specification.  This is used to drive the command parser.
  82  */
  83 struct s_format {
  84         char code;                              /* Command code */
  85         int naddr;                              /* Number of address args */
  86         enum e_args args;                       /* Argument type */
  87 };
  88
  89 static struct s_format cmd_fmts[] = {
  90         {'{', 2, GROUP},
  91         {'}', 0, ENDGROUP},
  92         {'a', 1, TEXT},
  93         {'b', 2, BRANCH},
  94         {'c', 2, TEXT},
  95         {'d', 2, EMPTY},
  96         {'D', 2, EMPTY},
  97         {'g', 2, EMPTY},
  98         {'G', 2, EMPTY},
  99         {'h', 2, EMPTY},
 100         {'H', 2, EMPTY},
 101         {'i', 1, TEXT},
 102         {'l', 2, EMPTY},
 103         {'n', 2, EMPTY},
 104         {'N', 2, EMPTY},
 105         {'p', 2, EMPTY},
 106         {'P', 2, EMPTY},
 107         {'q', 1, EMPTY},
 108         {'r', 1, RFILE},
 109         {'s', 2, SUBST},
 110         {'t', 2, BRANCH},
 111         {'w', 2, WFILE},
 112         {'x', 2, EMPTY},
 113         {'y', 2, TR},
 114         {'!', 2, NONSEL},
 115         {':', 0, LABEL},
 116         {'#', 0, COMMENT},
 117         {'=', 1, EMPTY},
 118         {'\0', 0, COMMENT},
 119 };
 120
 121 /* The compiled program. */
 122 struct s_command *prog;
 123
 124 /*
 125  * Compile the program into prog.
 126  * Initialise appends.
 127  */
 128 void
 129 compile(void)
 130 {
 131         *compile_stream(&prog) = NULL;
 132         fixuplabel(prog, NULL);
 133         uselabel();
 134         if (appendnum == 0)
 135                 appends = NULL;
 136         else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) ==
 137             NULL)
 138                 err(1, "malloc");
 139         if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL)
 140                 err(1, "malloc");
 141 }
 142
 143 #define EATSPACE() do {                                                 \
 144         if (p)                                                          \
 145                 while (*p && isspace((unsigned char)*p))                \
 146                         p++;                                            \
 147         } while (0)
 148
 149 static struct s_command **
 150 compile_stream(struct s_command **link)
 151 {
 152         char *p;
 153         static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */
 154         struct s_command *cmd, *cmd2, *stack;
 155         struct s_format *fp;
 156         char re[_POSIX2_LINE_MAX + 1];
 157         int naddr;                              /* Number of addresses */
 158
 159         stack = NULL;
 160         for (;;) {
 161                 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) {
 162                         if (stack != NULL)
 163                                 errx(1, "%lu: %s: unexpected EOF (pending }'s)",
 164                                                         linenum, fname);
 165                         return (link);
 166                 }
 167
 168 semicolon:      EATSPACE();
 169                 if (p) {
 170                         if (*p == '#' || *p == '\0')
 171                                 continue;
 172                         else if (*p == ';') {
 173                                 p++;
 174                                 goto semicolon;
 175                         }
 176                 }
 177                 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL)
 178                         err(1, "malloc");
 179                 link = &cmd->next;
 180                 cmd->startline = cmd->nonsel = 0;
 181                 /* First parse the addresses */
 182                 naddr = 0;
 183
 184 /* Valid characters to start an address */
 185 #define addrchar(c)     (strchr("0123456789/\\$", (c)))
 186                 if (addrchar(*p)) {
 187                         naddr++;
 188                         if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL)
 189                                 err(1, "malloc");
 190                         p = compile_addr(p, cmd->a1);
 191                         EATSPACE();                             /* EXTENSION */
 192                         if (*p == ',') {
 193                                 p++;
 194                                 EATSPACE();                     /* EXTENSION */
 195                                 naddr++;
 196                                 if ((cmd->a2 = malloc(sizeof(struct s_addr)))
 197                                     == NULL)
 198                                         err(1, "malloc");
 199                                 p = compile_addr(p, cmd->a2);
 200                                 EATSPACE();
 201                         } else
 202                                 cmd->a2 = 0;
 203                 } else
 204                         cmd->a1 = cmd->a2 = 0;
 205
 206 nonsel:         /* Now parse the command */
 207                 if (!*p)
 208                         errx(1, "%lu: %s: command expected", linenum, fname);
 209                 cmd->code = *p;
 210                 for (fp = cmd_fmts; fp->code; fp++)
 211                         if (fp->code == *p)
 212                                 break;
 213                 if (!fp->code)
 214                         errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p);
 215                 if (naddr > fp->naddr)
 216                         errx(1,
 217                                 "%lu: %s: command %c expects up to %d address(es), found %d",
 218                                 linenum, fname, *p, fp->naddr, naddr);
 219                 switch (fp->args) {
 220                 case NONSEL:                    /* ! */
 221                         p++;
 222                         EATSPACE();
 223                         cmd->nonsel = 1;
 224                         goto nonsel;
 225                 case GROUP:                     /* { */
 226                         p++;
 227                         EATSPACE();
 228                         cmd->next = stack;
 229                         stack = cmd;
 230                         link = &cmd->u.c;
 231                         if (*p)
 232                                 goto semicolon;
 233                         break;
 234                 case ENDGROUP:
 235                         /*
 236                          * Short-circuit command processing, since end of
 237                          * group is really just a noop.
 238                          */
 239                         cmd->nonsel = 1;
 240                         if (stack == NULL)
 241                                 errx(1, "%lu: %s: unexpected }", linenum, fname);
 242                         cmd2 = stack;
 243                         stack = cmd2->next;
 244                         cmd2->next = cmd;
 245                         /*FALLTHROUGH*/
 246                 case EMPTY:             /* d D g G h H l n N p P q x = \0 */
 247                         p++;
 248                         EATSPACE();
 249                         if (*p == ';') {
 250                                 p++;
 251                                 link = &cmd->next;
 252                                 goto semicolon;
 253                         }
 254                         if (*p)
 255                                 errx(1, "%lu: %s: extra characters at the end of %c command",
 256                                                 linenum, fname, cmd->code);
 257                         break;
 258                 case TEXT:                      /* a c i */
 259                         p++;
 260                         EATSPACE();
 261                         if (*p != '\\')
 262                                 errx(1,
 263 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code);
 264                         p++;
 265                         EATSPACE();
 266                         if (*p)
 267                                 errx(1,
 268                                 "%lu: %s: extra characters after \\ at the end of %c command",
 269                                 linenum, fname, cmd->code);
 270                         cmd->t = compile_text();
 271                         break;
 272                 case COMMENT:                   /* \0 # */
 273                         break;
 274                 case WFILE:                     /* w */
 275                         p++;
 276                         EATSPACE();
 277                         if (*p == '\0')
 278                                 errx(1, "%lu: %s: filename expected", linenum, fname);
 279                         cmd->t = duptoeol(p, "w command");
 280                         if (aflag)
 281                                 cmd->u.fd = -1;
 282                         else if ((cmd->u.fd = open(p,
 283                             O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
 284                             DEFFILEMODE)) == -1)
 285                                 err(1, "%s", p);
 286                         break;
 287                 case RFILE:                     /* r */
 288                         p++;
 289                         EATSPACE();
 290                         if (*p == '\0')
 291                                 errx(1, "%lu: %s: filename expected", linenum, fname);
 292                         else
 293                                 cmd->t = duptoeol(p, "read command");
 294                         break;
 295                 case BRANCH:                    /* b t */
 296                         p++;
 297                         EATSPACE();
 298                         if (*p == '\0')
 299                                 cmd->t = NULL;
 300                         else
 301                                 cmd->t = duptoeol(p, "branch");
 302                         break;
 303                 case LABEL:                     /* : */
 304                         p++;
 305                         EATSPACE();
 306                         cmd->t = duptoeol(p, "label");
 307                         if (strlen(p) == 0)
 308                                 errx(1, "%lu: %s: empty label", linenum, fname);
 309                         enterlabel(cmd);
 310                         break;
 311                 case SUBST:                     /* s */
 312                         p++;
 313                         if (*p == '\0' || *p == '\\')
 314                                 errx(1,
 315 "%lu: %s: substitute pattern can not be delimited by newline or backslash",
 316                                         linenum, fname);
 317                         if ((cmd->u.s = calloc(1, sizeof(struct s_subst))) == NULL)
 318                                 err(1, "malloc");
 319                         p = compile_delimited(p, re, 0);
 320                         if (p == NULL)
 321                                 errx(1,
 322                                 "%lu: %s: unterminated substitute pattern", linenum, fname);
 323
 324                         /* Compile RE with no case sensitivity temporarily */
 325                         if (*re == '\0')
 326                                 cmd->u.s->re = NULL;
 327                         else
 328                                 cmd->u.s->re = compile_re(re, 0);
 329                         --p;
 330                         p = compile_subst(p, cmd->u.s);
 331                         p = compile_flags(p, cmd->u.s);
 332
 333                         /* Recompile RE with case sensitivity from "I" flag if any */
 334                         if (*re == '\0')
 335                                 cmd->u.s->re = NULL;
 336                         else
 337                                 cmd->u.s->re = compile_re(re, cmd->u.s->icase);
 338                         EATSPACE();
 339                         if (*p == ';') {
 340                                 p++;
 341                                 link = &cmd->next;
 342                                 goto semicolon;
 343                         }
 344                         break;
 345                 case TR:                        /* y */
 346                         p++;
 347                         p = compile_tr(p, &cmd->u.y);
 348                         EATSPACE();
 349                         if (*p == ';') {
 350                                 p++;
 351                                 link = &cmd->next;
 352                                 goto semicolon;
 353                         }
 354                         if (*p)
 355                                 errx(1,
 356 "%lu: %s: extra text at the end of a transform command", linenum, fname);
 357                         break;
 358                 }
 359         }
 360 }
 361
 362 /*
 363  * Get a delimited string.  P points to the delimiter of the string; d points
 364  * to a buffer area.  Newline and delimiter escapes are processed; other
 365  * escapes are ignored.
 366  *
 367  * Returns a pointer to the first character after the final delimiter or NULL
 368  * in the case of a non-terminated string.  The character array d is filled
 369  * with the processed string.
 370  */
 371 static char *
 372 compile_delimited(char *p, char *d, int is_tr)
 373 {
 374         char c;
 375
 376         c = *p++;
 377         if (c == '\0')
 378                 return (NULL);
 379         else if (c == '\\')
 380                 errx(1, "%lu: %s: \\ can not be used as a string delimiter",
 381                                 linenum, fname);
 382         else if (c == '\n')
 383                 errx(1, "%lu: %s: newline can not be used as a string delimiter",
 384                                 linenum, fname);
 385         while (*p) {
 386                 if (*p == '[' && *p != c) {
 387                         if ((d = compile_ccl(&p, d)) == NULL)
 388                                 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname);
 389                         continue;
 390                 } else if (*p == '\\' && p[1] == '[') {
 391                         *d++ = *p++;
 392                 } else if (*p == '\\' && p[1] == c)
 393                         p++;
 394                 else if (*p == '\\' && p[1] == 'n') {
 395                         *d++ = '\n';
 396                         p += 2;
 397                         continue;
 398                 } else if (*p == '\\' && p[1] == '\\') {
 399                         if (is_tr)
 400                                 p++;
 401                         else
 402                                 *d++ = *p++;
 403                 } else if (*p == c) {
 404                         *d = '\0';
 405                         return (p + 1);
 406                 }
 407                 *d++ = *p++;
 408         }
 409         return (NULL);
 410 }
 411
 412
 413 /* compile_ccl: expand a POSIX character class */
 414 static char *
 415 compile_ccl(char **sp, char *t)
 416 {
 417         int c, d;
 418         char *s = *sp;
 419
 420         *t++ = *s++;
 421         if (*s == '^')
 422                 *t++ = *s++;
 423         if (*s == ']')
 424                 *t++ = *s++;
 425         for (; *s && (*t = *s) != ']'; s++, t++)
 426                 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
 427                         *++t = *++s, t++, s++;
 428                         for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
 429                                 if ((c = *s) == '\0')
 430                                         return NULL;
 431                 }
 432         return (*s == ']') ? *sp = ++s, ++t : NULL;
 433 }
 434
 435 /*
 436  * Compiles the regular expression in RE and returns a pointer to the compiled
 437  * regular expression.
 438  * Cflags are passed to regcomp.
 439  */
 440 static regex_t *
 441 compile_re(char *re, int case_insensitive)
 442 {
 443         regex_t *rep;
 444         int eval, flags;
 445
 446
 447         flags = rflags;
 448         if (case_insensitive)
 449                 flags |= REG_ICASE;
 450         if ((rep = malloc(sizeof(regex_t))) == NULL)
 451                 err(1, "malloc");
 452         if ((eval = regcomp(rep, re, flags)) != 0)
 453                 errx(1, "%lu: %s: RE error: %s",
 454                                 linenum, fname, strregerror(eval, rep));
 455         if (maxnsub < rep->re_nsub)
 456                 maxnsub = rep->re_nsub;
 457         return (rep);
 458 }
 459
 460 /*
 461  * Compile the substitution string of a regular expression and set res to
 462  * point to a saved copy of it.  Nsub is the number of parenthesized regular
 463  * expressions.
 464  */
 465 static char *
 466 compile_subst(char *p, struct s_subst *s)
 467 {
 468         static char lbuf[_POSIX2_LINE_MAX + 1];
 469         int asize, size;
 470         u_char ref;
 471         char c, *text, *op, *sp;
 472         int more = 1, sawesc = 0;
 473
 474         c = *p++;                       /* Terminator character */
 475         if (c == '\0')
 476                 return (NULL);
 477
 478         s->maxbref = 0;
 479         s->linenum = linenum;
 480         asize = 2 * _POSIX2_LINE_MAX + 1;
 481         if ((text = malloc(asize)) == NULL)
 482                 err(1, "malloc");
 483         size = 0;
 484         do {
 485                 op = sp = text + size;
 486                 for (; *p; p++) {
 487                         if (*p == '\\' || sawesc) {
 488                                 /*
 489                                  * If this is a continuation from the last
 490                                  * buffer, we won't have a character to
 491                                  * skip over.
 492                                  */
 493                                 if (sawesc)
 494                                         sawesc = 0;
 495                                 else
 496                                         p++;
 497
 498                                 if (*p == '\0') {
 499                                         /*
 500                                          * This escaped character is continued
 501                                          * in the next part of the line.  Note
 502                                          * this fact, then cause the loop to
 503                                          * exit w/ normal EOL case and reenter
 504                                          * above with the new buffer.
 505                                          */
 506                                         sawesc = 1;
 507                                         p--;
 508                                         continue;
 509                                 } else if (strchr("123456789", *p) != NULL) {
 510                                         *sp++ = '\\';
 511                                         ref = *p - '0';
 512                                         if (s->re != NULL &&
 513                                             ref > s->re->re_nsub)
 514                                                 errx(1, "%lu: %s: \\%c not defined in the RE",
 515                                                                 linenum, fname, *p);
 516                                         if (s->maxbref < ref)
 517                                                 s->maxbref = ref;
 518                                 } else if (*p == '&' || *p == '\\')
 519                                         *sp++ = '\\';
 520                         } else if (*p == c) {
 521                                 if (*++p == '\0' && more) {
 522                                         if (cu_fgets(lbuf, sizeof(lbuf), &more))
 523                                                 p = lbuf;
 524                                 }
 525                                 *sp++ = '\0';
 526                                 size += sp - op;
 527                                 if ((s->new = realloc(text, size)) == NULL)
 528                                         err(1, "realloc");
 529                                 return (p);
 530                         } else if (*p == '\n') {
 531                                 errx(1,
 532 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname);
 533                                 /* NOTREACHED */
 534                         }
 535                         *sp++ = *p;
 536                 }
 537                 size += sp - op;
 538                 if (asize - size < _POSIX2_LINE_MAX + 1) {
 539                         asize *= 2;
 540                         if ((text = realloc(text, asize)) == NULL)
 541                                 err(1, "realloc");
 542                 }
 543         } while (cu_fgets(p = lbuf, sizeof(lbuf), &more));
 544         errx(1, "%lu: %s: unterminated substitute in regular expression",
 545                         linenum, fname);
 546         /* NOTREACHED */
 547 }
 548
 549 /*
 550  * Compile the flags of the s command
 551  */
 552 static char *
 553 compile_flags(char *p, struct s_subst *s)
 554 {
 555         int gn;                 /* True if we have seen g or n */
 556         unsigned long nval;
 557         char wfile[_POSIX2_LINE_MAX + 1], *q, *eq;
 558
 559         s->n = 1;                               /* Default */
 560         s->p = 0;
 561         s->wfile = NULL;
 562         s->wfd = -1;
 563         s->icase = 0;
 564         for (gn = 0;;) {
 565                 EATSPACE();                     /* EXTENSION */
 566                 switch (*p) {
 567                 case 'g':
 568                         if (gn)
 569                                 errx(1,
 570 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
 571                         gn = 1;
 572                         s->n = 0;
 573                         break;
 574                 case '\0':
 575                 case '\n':
 576                 case ';':
 577                         return (p);
 578                 case 'p':
 579                         s->p = 1;
 580                         break;
 581                 case 'i':
 582                 case 'I':
 583                         s->icase = 1;
 584                         break;
 585                 case '1': case '2': case '3':
 586                 case '4': case '5': case '6':
 587                 case '7': case '8': case '9':
 588                         if (gn)
 589                                 errx(1,
 590 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
 591                         gn = 1;
 592                         errno = 0;
 593                         nval = strtol(p, &p, 10);
 594                         if (errno == ERANGE || nval > INT_MAX)
 595                                 errx(1,
 596 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname);
 597                         s->n = nval;
 598                         p--;
 599                         break;
 600                 case 'w':
 601                         p++;
 602 #ifdef HISTORIC_PRACTICE
 603                         if (*p != ' ') {
 604                                 warnx("%lu: %s: space missing before w wfile", linenum, fname);
 605                                 return (p);
 606                         }
 607 #endif
 608                         EATSPACE();
 609                         q = wfile;
 610                         eq = wfile + sizeof(wfile) - 1;
 611                         while (*p) {
 612                                 if (*p == '\n')
 613                                         break;
 614                                 if (q >= eq)
 615                                         err(1, "wfile too long");
 616                                 *q++ = *p++;
 617                         }
 618                         *q = '\0';
 619                         if (q == wfile)
 620                                 errx(1, "%lu: %s: no wfile specified", linenum, fname);
 621                         s->wfile = strdup(wfile);
 622                         if (!aflag && (s->wfd = open(wfile,
 623                             O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
 624                             DEFFILEMODE)) == -1)
 625                                 err(1, "%s", wfile);
 626                         return (p);
 627                 default:
 628                         errx(1, "%lu: %s: bad flag in substitute command: '%c'",
 629                                         linenum, fname, *p);
 630                         break;
 631                 }
 632                 p++;
 633         }
 634 }
 635
 636 /*
 637  * Compile a translation set of strings into a lookup table.
 638  */
 639 static char *
 640 compile_tr(char *p, struct s_tr **py)
 641 {
 642         struct s_tr *y;
 643         int i;
 644         const char *op, *np;
 645         char old[_POSIX2_LINE_MAX + 1];
 646         char new[_POSIX2_LINE_MAX + 1];
 647         size_t oclen, oldlen, nclen, newlen;
 648         mbstate_t mbs1, mbs2;
 649
 650         if ((*py = y = malloc(sizeof(*y))) == NULL)
 651                 err(1, NULL);
 652         y->multis = NULL;
 653         y->nmultis = 0;
 654
 655         if (*p == '\0' || *p == '\\')
 656                 errx(1,
 657         "%lu: %s: transform pattern can not be delimited by newline or backslash",
 658                         linenum, fname);
 659         p = compile_delimited(p, old, 1);
 660         if (p == NULL)
 661                 errx(1, "%lu: %s: unterminated transform source string",
 662                                 linenum, fname);
 663         p = compile_delimited(p - 1, new, 1);
 664         if (p == NULL)
 665                 errx(1, "%lu: %s: unterminated transform target string",
 666                                 linenum, fname);
 667         EATSPACE();
 668         op = old;
 669         oldlen = mbsrtowcs(NULL, &op, 0, NULL);
 670         if (oldlen == (size_t)-1)
 671                 err(1, NULL);
 672         np = new;
 673         newlen = mbsrtowcs(NULL, &np, 0, NULL);
 674         if (newlen == (size_t)-1)
 675                 err(1, NULL);
 676         if (newlen != oldlen)
 677                 errx(1, "%lu: %s: transform strings are not the same length",
 678                                 linenum, fname);
 679         if (MB_CUR_MAX == 1) {
 680                 /*
 681                  * The single-byte encoding case is easy: generate a
 682                  * lookup table.
 683                  */
 684                 for (i = 0; i <= UCHAR_MAX; i++)
 685                         y->bytetab[i] = (char)i;
 686                 for (; *op; op++, np++)
 687                         y->bytetab[(u_char)*op] = *np;
 688         } else {
 689                 /*
 690                  * Multi-byte encoding case: generate a lookup table as
 691                  * above, but only for single-byte characters. The first
 692                  * bytes of multi-byte characters have their lookup table
 693                  * entries set to 0, which causes do_tr() to search through
 694                  * an auxiliary vector of multi-byte mappings.
 695                  */
 696                 memset(&mbs1, 0, sizeof(mbs1));
 697                 memset(&mbs2, 0, sizeof(mbs2));
 698                 for (i = 0; i <= UCHAR_MAX; i++)
 699                         y->bytetab[i] = (btowc(i) != WEOF) ? i : 0;
 700                 while (*op != '\0') {
 701                         oclen = mbrlen(op, MB_LEN_MAX, &mbs1);
 702                         if (oclen == (size_t)-1 || oclen == (size_t)-2)
 703                                 errc(1, EILSEQ, NULL);
 704                         nclen = mbrlen(np, MB_LEN_MAX, &mbs2);
 705                         if (nclen == (size_t)-1 || nclen == (size_t)-2)
 706                                 errc(1, EILSEQ, NULL);
 707                         if (oclen == 1 && nclen == 1)
 708                                 y->bytetab[(u_char)*op] = *np;
 709                         else {
 710                                 y->bytetab[(u_char)*op] = 0;
 711                                 y->multis = realloc(y->multis,
 712                                     (y->nmultis + 1) * sizeof(*y->multis));
 713                                 if (y->multis == NULL)
 714                                         err(1, NULL);
 715                                 i = y->nmultis++;
 716                                 y->multis[i].fromlen = oclen;
 717                                 memcpy(y->multis[i].from, op, oclen);
 718                                 y->multis[i].tolen = nclen;
 719                                 memcpy(y->multis[i].to, np, nclen);
 720                         }
 721                         op += oclen;
 722                         np += nclen;
 723                 }
 724         }
 725         return (p);
 726 }
 727
 728 /*
 729  * Compile the text following an a or i command.
 730  */
 731 static char *
 732 compile_text(void)
 733 {
 734         int asize, esc_nl, size;
 735         char *text, *p, *op, *s;
 736         char lbuf[_POSIX2_LINE_MAX + 1];
 737
 738         asize = 2 * _POSIX2_LINE_MAX + 1;
 739         if ((text = malloc(asize)) == NULL)
 740                 err(1, "malloc");
 741         size = 0;
 742         while (cu_fgets(lbuf, sizeof(lbuf), NULL)) {
 743                 op = s = text + size;
 744                 p = lbuf;
 745                 EATSPACE();
 746                 for (esc_nl = 0; *p != '\0'; p++) {
 747                         if (*p == '\\' && p[1] != '\0' && *++p == '\n')
 748                                 esc_nl = 1;
 749                         *s++ = *p;
 750                 }
 751                 size += s - op;
 752                 if (!esc_nl) {
 753                         *s = '\0';
 754                         break;
 755                 }
 756                 if (asize - size < _POSIX2_LINE_MAX + 1) {
 757                         asize *= 2;
 758                         if ((text = realloc(text, asize)) == NULL)
 759                                 err(1, "realloc");
 760                 }
 761         }
 762         text[size] = '\0';
 763         if ((p = realloc(text, size + 1)) == NULL)
 764                 err(1, "realloc");
 765         return (p);
 766 }
 767
 768 /*
 769  * Get an address and return a pointer to the first character after
 770  * it.  Fill the structure pointed to according to the address.
 771  */
 772 static char *
 773 compile_addr(char *p, struct s_addr *a)
 774 {
 775         char *end, re[_POSIX2_LINE_MAX + 1];
 776         int icase;
 777
 778         icase = 0;
 779
 780         a->type = 0;
 781         switch (*p) {
 782         case '\\':                              /* Context address */
 783                 ++p;
 784                 /* FALLTHROUGH */
 785         case '/':                               /* Context address */
 786                 p = compile_delimited(p, re, 0);
 787                 if (p == NULL)
 788                         errx(1, "%lu: %s: unterminated regular expression", linenum, fname);
 789                 /* Check for case insensitive regexp flag */
 790                 if (*p == 'I') {
 791                         icase = 1;
 792                         p++;
 793                 }
 794                 if (*re == '\0')
 795                         a->u.r = NULL;
 796                 else
 797                         a->u.r = compile_re(re, icase);
 798                 a->type = AT_RE;
 799                 return (p);
 800
 801         case '$':                               /* Last line */
 802                 a->type = AT_LAST;
 803                 return (p + 1);
 804
 805         case '+':                               /* Relative line number */
 806                 a->type = AT_RELLINE;
 807                 p++;
 808                 /* FALLTHROUGH */
 809                                                 /* Line number */
 810         case '0': case '1': case '2': case '3': case '4':
 811         case '5': case '6': case '7': case '8': case '9':
 812                 if (a->type == 0)
 813                         a->type = AT_LINE;
 814                 a->u.l = strtol(p, &end, 10);
 815                 return (end);
 816         default:
 817                 errx(1, "%lu: %s: expected context address", linenum, fname);
 818                 return (NULL);
 819         }
 820 }
 821
 822 /*
 823  * duptoeol --
 824  *      Return a copy of all the characters up to \n or \0.
 825  */
 826 static char *
 827 duptoeol(char *s, const char *ctype)
 828 {
 829         size_t len;
 830         int ws;
 831         char *p, *start;
 832
 833         ws = 0;
 834         for (start = s; *s != '\0' && *s != '\n'; ++s)
 835                 ws = isspace((unsigned char)*s);
 836         *s = '\0';
 837         if (ws)
 838                 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype);
 839         len = s - start + 1;
 840         if ((p = malloc(len)) == NULL)
 841                 err(1, "malloc");
 842         return (memmove(p, start, len));
 843 }
 844
 845 /*
 846  * Convert goto label names to addresses, and count a and r commands, in
 847  * the given subset of the script.  Free the memory used by labels in b
 848  * and t commands (but not by :).
 849  *
 850  * TODO: Remove } nodes
 851  */
 852 static void
 853 fixuplabel(struct s_command *cp, struct s_command *end)
 854 {
 855
 856         for (; cp != end; cp = cp->next)
 857                 switch (cp->code) {
 858                 case 'a':
 859                 case 'r':
 860                         appendnum++;
 861                         break;
 862                 case 'b':
 863                 case 't':
 864                         /* Resolve branch target. */
 865                         if (cp->t == NULL) {
 866                                 cp->u.c = NULL;
 867                                 break;
 868                         }
 869                         if ((cp->u.c = findlabel(cp->t)) == NULL)
 870                                 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t);
 871                         free(cp->t);
 872                         break;
 873                 case '{':
 874                         /* Do interior commands. */
 875                         fixuplabel(cp->u.c, cp->next);
 876                         break;
 877                 }
 878 }
 879
 880 /*
 881  * Associate the given command label for later lookup.
 882  */
 883 static void
 884 enterlabel(struct s_command *cp)
 885 {
 886         struct labhash **lhp, *lh;
 887         u_char *p;
 888         u_int h, c;
 889
 890         for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
 891                 h = (h << 5) + h + c;
 892         lhp = &labels[h & LHMASK];
 893         for (lh = *lhp; lh != NULL; lh = lh->lh_next)
 894                 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
 895                         errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t);
 896         if ((lh = malloc(sizeof *lh)) == NULL)
 897                 err(1, "malloc");
 898         lh->lh_next = *lhp;
 899         lh->lh_hash = h;
 900         lh->lh_cmd = cp;
 901         lh->lh_ref = 0;
 902         *lhp = lh;
 903 }
 904
 905 /*
 906  * Find the label contained in the command l in the command linked
 907  * list cp.  L is excluded from the search.  Return NULL if not found.
 908  */
 909 static struct s_command *
 910 findlabel(char *name)
 911 {
 912         struct labhash *lh;
 913         u_char *p;
 914         u_int h, c;
 915
 916         for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
 917                 h = (h << 5) + h + c;
 918         for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
 919                 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
 920                         lh->lh_ref = 1;
 921                         return (lh->lh_cmd);
 922                 }
 923         }
 924         return (NULL);
 925 }
 926
 927 /*
 928  * Warn about any unused labels.  As a side effect, release the label hash
 929  * table space.
 930  */
 931 static void
 932 uselabel(void)
 933 {
 934         struct labhash *lh, *next;
 935         int i;
 936
 937         for (i = 0; i < LHSZ; i++) {
 938                 for (lh = labels[i]; lh != NULL; lh = next) {
 939                         next = lh->lh_next;
 940                         if (!lh->lh_ref)
 941                                 warnx("%lu: %s: unused label '%s'",
 942                                     linenum, fname, lh->lh_cmd->t);
 943                         free(lh);
 944                 }
 945         }
 946 }