usr.bin/m4/main.c

   1 /*      $OpenBSD: main.c,v 1.53 2002/04/26 16:15:16 espie Exp $ */
   2 /*      $NetBSD: main.c,v 1.12 1997/02/08 23:54:49 cgd Exp $    */
   3
   4 /*-
   5  * Copyright (c) 1989, 1993
   6  *      The Regents of the University of California.  All rights reserved.
   7  *
   8  * This code is derived from software contributed to Berkeley by
   9  * Ozan Yigit at York University.
  10  *
  11  * Redistribution and use in source and binary forms, with or without
  12  * modification, are permitted provided that the following conditions
  13  * are met:
  14  * 1. Redistributions of source code must retain the above copyright
  15  *    notice, this list of conditions and the following disclaimer.
  16  * 2. Redistributions in binary form must reproduce the above copyright
  17  *    notice, this list of conditions and the following disclaimer in the
  18  *    documentation and/or other materials provided with the distribution.
  19  * 4. Neither the name of the University nor the names of its contributors
  20  *    may be used to endorse or promote products derived from this software
  21  *    without specific prior written permission.
  22  *
  23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  33  * SUCH DAMAGE.
  34  */
  35
  36 #ifndef lint
  37 static char copyright[] =
  38 "@(#) Copyright (c) 1989, 1993\n\
  39         The Regents of the University of California.  All rights reserved.\n";
  40 #endif /* not lint */
  41
  42 #ifndef lint
  43 #if 0
  44 static char sccsid[] = "@(#)main.c      8.1 (Berkeley) 6/6/93";
  45 #else
  46 #if 0
  47 static char rcsid[] = "$OpenBSD: main.c,v 1.53 2002/04/26 16:15:16 espie Exp $";
  48 #endif
  49 #endif
  50 #endif /* not lint */
  51
  52 #include <sys/cdefs.h>
  53 __FBSDID("$FreeBSD$");
  54
  55 /*
  56  * main.c
  57  * Facility: m4 macro processor
  58  * by: oz
  59  */
  60
  61 #include <sys/types.h>
  62 #include <assert.h>
  63 #include <signal.h>
  64 #include <errno.h>
  65 #include <unistd.h>
  66 #include <stdio.h>
  67 #include <ctype.h>
  68 #include <string.h>
  69 #include <stddef.h>
  70 #include <stdlib.h>
  71 #include <err.h>
  72 #include <locale.h>
  73 #include "mdef.h"
  74 #include "stdd.h"
  75 #include "extern.h"
  76 #include "pathnames.h"
  77
  78 ndptr hashtab[HASHSIZE];        /* hash table for macros etc.  */
  79 stae *mstack;                   /* stack of m4 machine         */
  80 char *sstack;                   /* shadow stack, for string space extension */
  81 static size_t STACKMAX;         /* current maximum size of stack */
  82 int sp;                         /* current m4  stack pointer   */
  83 int fp;                         /* m4 call frame pointer       */
  84 struct input_file infile[MAXINP];/* input file stack (0=stdin)  */
  85 char *inname[MAXINP];           /* names of these input files */
  86 int inlineno[MAXINP];           /* current number in each input file */
  87 FILE **outfile;                 /* diversion array(0=bitbucket)*/
  88 int maxout;
  89 FILE *active;                   /* active output file pointer  */
  90 int ilevel = 0;                 /* input file stack pointer    */
  91 int oindex = 0;                 /* diversion index..           */
  92 char null[] = "";               /* as it says.. just a null..  */
  93 const char *m4wraps = "";       /* m4wrap string default..     */
  94 char lquote[MAXCCHARS+1] = {LQUOTE};    /* left quote character  (`)   */
  95 char rquote[MAXCCHARS+1] = {RQUOTE};    /* right quote character (')   */
  96 char scommt[MAXCCHARS+1] = {SCOMMT};    /* start character for comment */
  97 char ecommt[MAXCCHARS+1] = {ECOMMT};    /* end character for comment   */
  98 int synccpp;                    /* Line synchronisation for C preprocessor */
  99
 100 struct keyblk keywrds[] = {     /* m4 keywords to be installed */
 101         { "include",      INCLTYPE },
 102         { "sinclude",     SINCTYPE },
 103         { "define",       DEFITYPE },
 104         { "defn",         DEFNTYPE },
 105         { "divert",       DIVRTYPE | NOARGS },
 106         { "expr",         EXPRTYPE },
 107         { "eval",         EXPRTYPE },
 108         { "substr",       SUBSTYPE },
 109         { "ifelse",       IFELTYPE },
 110         { "ifdef",        IFDFTYPE },
 111         { "len",          LENGTYPE },
 112         { "incr",         INCRTYPE },
 113         { "decr",         DECRTYPE },
 114         { "dnl",          DNLNTYPE | NOARGS },
 115         { "changequote",  CHNQTYPE | NOARGS },
 116         { "changecom",    CHNCTYPE | NOARGS },
 117         { "index",        INDXTYPE },
 118 #ifdef EXTENDED
 119         { "paste",        PASTTYPE },
 120         { "spaste",       SPASTYPE },
 121         /* Newer extensions, needed to handle gnu-m4 scripts */
 122         { "indir",        INDIRTYPE},
 123         { "builtin",      BUILTINTYPE},
 124         { "patsubst",     PATSTYPE},
 125         { "regexp",       REGEXPTYPE},
 126         { "esyscmd",      ESYSCMDTYPE},
 127         { "__file__",     FILENAMETYPE | NOARGS},
 128         { "__line__",     LINETYPE | NOARGS},
 129 #endif
 130         { "popdef",       POPDTYPE },
 131         { "pushdef",      PUSDTYPE },
 132         { "dumpdef",      DUMPTYPE | NOARGS },
 133         { "shift",        SHIFTYPE | NOARGS },
 134         { "translit",     TRNLTYPE },
 135         { "undefine",     UNDFTYPE },
 136         { "undivert",     UNDVTYPE | NOARGS },
 137         { "divnum",       DIVNTYPE | NOARGS },
 138         { "maketemp",     MKTMTYPE },
 139         { "errprint",     ERRPTYPE | NOARGS },
 140         { "m4wrap",       M4WRTYPE | NOARGS },
 141         { "m4exit",       EXITTYPE | NOARGS },
 142         { "syscmd",       SYSCTYPE },
 143         { "sysval",       SYSVTYPE | NOARGS },
 144         { "traceon",      TRACEONTYPE | NOARGS },
 145         { "traceoff",     TRACEOFFTYPE | NOARGS },
 146
 147 #if defined(unix) || defined(__unix__)
 148         { "unix",         SELFTYPE | NOARGS },
 149 #else
 150 #ifdef vms
 151         { "vms",          SELFTYPE | NOARGS },
 152 #endif
 153 #endif
 154 };
 155
 156 #define MAXKEYS (sizeof(keywrds)/sizeof(struct keyblk))
 157
 158 #define MAXRECORD 50
 159 static struct position {
 160         char *name;
 161         unsigned long line;
 162 } quotes[MAXRECORD], paren[MAXRECORD];
 163
 164 static void record(struct position *, int);
 165 static void dump_stack(struct position *, int);
 166
 167 static void macro(void);
 168 static void initkwds(void);
 169 static ndptr inspect(int, char *);
 170 static int do_look_ahead(int, const char *);
 171
 172 static void enlarge_stack(void);
 173
 174 int
 175 main(int argc, char *argv[])
 176 {
 177         int c;
 178         int n;
 179         int rval;
 180         char *p;
 181
 182         setlocale(LC_ALL, "");
 183
 184         traceout = stderr;
 185
 186         if (signal(SIGINT, SIG_IGN) != SIG_IGN)
 187                 signal(SIGINT, onintr);
 188
 189         initkwds();
 190         initspaces();
 191         STACKMAX = INITSTACKMAX;
 192
 193         mstack = (stae *)xalloc(sizeof(stae) * STACKMAX);
 194         sstack = (char *)xalloc(STACKMAX);
 195
 196         maxout = 0;
 197         outfile = NULL;
 198         resizedivs(MAXOUT);
 199
 200         while ((c = getopt(argc, argv, "gst:d:D:U:o:I:")) != -1)
 201                 switch(c) {
 202                 case 'D':               /* define something..*/
 203                         for (p = optarg; *p; p++)
 204                                 if (*p == '=')
 205                                         break;
 206                         if (p == optarg)
 207                                 errx(1, "null variable cannot be defined");
 208                         if (*p)
 209                                 *p++ = EOS;
 210                         dodefine(optarg, p);
 211                         break;
 212                 case 'I':
 213                         addtoincludepath(optarg);
 214                         break;
 215                 case 'U':               /* undefine...       */
 216                         remhash(optarg, TOP);
 217                         break;
 218                 case 'g':
 219                         mimic_gnu = 1;
 220                         break;
 221                 case 'd':
 222                         set_trace_flags(optarg);
 223                         break;
 224                 case 's':
 225                         synccpp = 1;
 226                         break;
 227                 case 't':
 228                         mark_traced(optarg, 1);
 229                         break;
 230                 case 'o':
 231                         trace_file(optarg);
 232                         break;
 233                 case '?':
 234                 default:
 235                         usage();
 236                 }
 237
 238         argc -= optind;
 239         argv += optind;
 240
 241         rval = 0;
 242         active = stdout;                /* default active output     */
 243         bbase[0] = bufbase;
 244         if (!argc) {
 245                 sp = -1;                /* stack pointer initialized */
 246                 fp = 0;                 /* frame pointer initialized */
 247                 set_input(infile+0, stdin, "stdin");
 248                                         /* default input (naturally) */
 249                 if ((inname[0] = strdup("-")) == NULL)
 250                         err(1, NULL);
 251                 inlineno[0] = 1;
 252                 emitline();
 253                 macro();
 254         } else
 255                 for (; argc--; ++argv) {
 256                         p = *argv;
 257                         if (p[0] == '-' && p[1] == EOS)
 258                                 set_input(infile, stdin, "stdin");
 259                         else if (fopen_trypath(infile, p) == NULL) {
 260                                 warn("%s", p);
 261                                 rval = 1;
 262                                 continue;
 263                         }
 264                         sp = -1;
 265                         fp = 0;
 266                         if ((inname[0] = strdup(p)) == NULL)
 267                                 err(1, NULL);
 268                         inlineno[0] = 1;
 269                         emitline();
 270                         macro();
 271                         release_input(infile);
 272                 }
 273
 274         if (*m4wraps) {                 /* anything for rundown ??   */
 275                 ilevel = 0;             /* in case m4wrap includes.. */
 276                 bufbase = bp = buf;     /* use the entire buffer   */
 277                 pbstr(m4wraps);         /* user-defined wrapup act   */
 278                 macro();                /* last will and testament   */
 279         }
 280
 281         if (active != stdout)
 282                 active = stdout;        /* reset output just in case */
 283         for (n = 1; n < maxout; n++)    /* default wrap-up: undivert */
 284                 if (outfile[n] != NULL)
 285                         getdiv(n);
 286                                         /* remove bitbucket if used  */
 287         if (outfile[0] != NULL) {
 288                 (void) fclose(outfile[0]);
 289         }
 290
 291         exit(rval);
 292 }
 293
 294 /*
 295  * Look ahead for `token'.
 296  * (on input `t == token[0]')
 297  * Used for comment and quoting delimiters.
 298  * Returns 1 if `token' present; copied to output.
 299  *         0 if `token' not found; all characters pushed back
 300  */
 301 static int
 302 do_look_ahead(int t, const char *token)
 303 {
 304         int i;
 305
 306         assert((unsigned char)t == (unsigned char)token[0]);
 307
 308         for (i = 1; *++token; i++) {
 309                 t = gpbc();
 310                 if (t == EOF || (unsigned char)t != (unsigned char)*token) {
 311                         putback(t);
 312                         while (--i)
 313                                 putback(*--token);
 314                         return 0;
 315                 }
 316         }
 317         return 1;
 318 }
 319
 320 #define LOOK_AHEAD(t, token) (t != EOF &&               \
 321     (unsigned char)(t)==(unsigned char)(token)[0] &&    \
 322     do_look_ahead(t,token))
 323
 324 /*
 325  * macro - the work horse..
 326  */
 327 static void
 328 macro(void)
 329 {
 330         char token[MAXTOK+1];
 331         int t, l;
 332         ndptr p;
 333         int  nlpar;
 334
 335         cycle {
 336                 t = gpbc();
 337                 if (t == '_' || isalpha(t)) {
 338                         p = inspect(t, token);
 339                         if (p != nil)
 340                                 putback(l = gpbc());
 341                         if (p == nil || (l != LPAREN &&
 342                             (p->type & NEEDARGS) != 0))
 343                                 outputstr(token);
 344                         else {
 345                 /*
 346                  * real thing.. First build a call frame:
 347                  */
 348                                 pushf(fp);      /* previous call frm */
 349                                 pushf(p->type); /* type of the call  */
 350                                 pushf(0);       /* parenthesis level */
 351                                 fp = sp;        /* new frame pointer */
 352                 /*
 353                  * now push the string arguments:
 354                  */
 355                                 pushs1(p->defn);        /* defn string */
 356                                 pushs1(p->name);        /* macro name  */
 357                                 pushs(ep);              /* start next..*/
 358
 359                                 if (l != LPAREN && PARLEV == 0)  {
 360                                     /* no bracks  */
 361                                         chrsave(EOS);
 362
 363                                         if ((uintptr_t)sp == STACKMAX)
 364                                                 errx(1, "internal stack overflow");
 365                                         eval((const char **) mstack+fp+1, 2,
 366                                             CALTYP);
 367
 368                                         ep = PREVEP;    /* flush strspace */
 369                                         sp = PREVSP;    /* previous sp..  */
 370                                         fp = PREVFP;    /* rewind stack...*/
 371                                 }
 372                         }
 373                 } else if (t == EOF) {
 374                         if (sp > -1) {
 375                                 warnx( "unexpected end of input, unclosed parenthesis:");
 376                                 dump_stack(paren, PARLEV);
 377                                 exit(1);
 378                         }
 379                         if (ilevel <= 0)
 380                                 break;                  /* all done thanks.. */
 381                         release_input(infile+ilevel--);
 382                         free(inname[ilevel+1]);
 383                         bufbase = bbase[ilevel];
 384                         emitline();
 385                         continue;
 386                 }
 387         /*
 388          * non-alpha token possibly seen..
 389          * [the order of else if .. stmts is important.]
 390          */
 391                 else if (LOOK_AHEAD(t,lquote)) {        /* strip quotes */
 392                         nlpar = 0;
 393                         record(quotes, nlpar++);
 394                         /*
 395                          * Opening quote: scan forward until matching
 396                          * closing quote has been found.
 397                          */
 398                         do {
 399
 400                                 l = gpbc();
 401                                 if (LOOK_AHEAD(l,rquote)) {
 402                                         if (--nlpar > 0)
 403                                                 outputstr(rquote);
 404                                 } else if (LOOK_AHEAD(l,lquote)) {
 405                                         record(quotes, nlpar++);
 406                                         outputstr(lquote);
 407                                 } else if (l == EOF) {
 408                                         if (nlpar == 1)
 409                                                 warnx("unclosed quote:");
 410                                         else
 411                                                 warnx("%d unclosed quotes:", nlpar);
 412                                         dump_stack(quotes, nlpar);
 413                                         exit(1);
 414                                 } else {
 415                                         if (nlpar > 0) {
 416                                                 if (sp < 0)
 417                                                         putc(l, active);
 418                                                 else
 419                                                         CHRSAVE(l);
 420                                         }
 421                                 }
 422                         }
 423                         while (nlpar != 0);
 424                 }
 425
 426                 else if (sp < 0 && LOOK_AHEAD(t, scommt)) {
 427                         fputs(scommt, active);
 428
 429                         for(;;) {
 430                                 t = gpbc();
 431                                 if (LOOK_AHEAD(t, ecommt)) {
 432                                         fputs(ecommt, active);
 433                                         break;
 434                                 }
 435                                 if (t == EOF)
 436                                         break;
 437                                 putc(t, active);
 438                         }
 439                 }
 440
 441                 else if (sp < 0) {              /* not in a macro at all */
 442                         putc(t, active);        /* output directly..     */
 443                 }
 444
 445                 else switch(t) {
 446
 447                 case LPAREN:
 448                         if (PARLEV > 0)
 449                                 chrsave(t);
 450                         while (isspace(l = gpbc()))
 451                                 ;               /* skip blank, tab, nl.. */
 452                         putback(l);
 453                         record(paren, PARLEV++);
 454                         break;
 455
 456                 case RPAREN:
 457                         if (--PARLEV > 0)
 458                                 chrsave(t);
 459                         else {                  /* end of argument list */
 460                                 chrsave(EOS);
 461
 462                                 if ((uintptr_t)sp == STACKMAX)
 463                                         errx(1, "internal stack overflow");
 464
 465                                 eval((const char **) mstack+fp+1, sp-fp,
 466                                     CALTYP);
 467
 468                                 ep = PREVEP;    /* flush strspace */
 469                                 sp = PREVSP;    /* previous sp..  */
 470                                 fp = PREVFP;    /* rewind stack...*/
 471                         }
 472                         break;
 473
 474                 case COMMA:
 475                         if (PARLEV == 1) {
 476                                 chrsave(EOS);           /* new argument   */
 477                                 while (isspace(l = gpbc()))
 478                                         ;
 479                                 putback(l);
 480                                 pushs(ep);
 481                         } else
 482                                 chrsave(t);
 483                         break;
 484
 485                 default:
 486                         if (LOOK_AHEAD(t, scommt)) {
 487                                 char *pc;
 488                                 for (pc = scommt; *pc; pc++)
 489                                         chrsave(*pc);
 490                                 for(;;) {
 491                                         t = gpbc();
 492                                         if (LOOK_AHEAD(t, ecommt)) {
 493                                                 for (pc = ecommt; *pc; pc++)
 494                                                         chrsave(*pc);
 495                                                 break;
 496                                         }
 497                                         if (t == EOF)
 498                                             break;
 499                                         CHRSAVE(t);
 500                                 }
 501                         } else
 502                                 CHRSAVE(t);             /* stack the char */
 503                         break;
 504                 }
 505         }
 506 }
 507
 508 /*
 509  * output string directly, without pushing it for reparses.
 510  */
 511 void
 512 outputstr(const char *s)
 513 {
 514         if (sp < 0)
 515                 while (*s)
 516                         putc(*s++, active);
 517         else
 518                 while (*s)
 519                         CHRSAVE(*s++);
 520 }
 521
 522 /*
 523  * build an input token..
 524  * consider only those starting with _ or A-Za-z. This is a
 525  * combo with lookup to speed things up.
 526  */
 527 static ndptr
 528 inspect(int c, char *tp)
 529 {
 530         char *name = tp;
 531         char *etp = tp+MAXTOK;
 532         ndptr p;
 533         unsigned int h;
 534
 535         h = *tp++ = c;
 536
 537         while ((isalnum(c = gpbc()) || c == '_') && tp < etp)
 538                 h = (h << 5) + h + (*tp++ = c);
 539         if (c != EOF)
 540                 PUTBACK(c);
 541         *tp = EOS;
 542         /* token is too long, it won't match anything, but it can still
 543          * be output. */
 544         if (tp == ep) {
 545                 outputstr(name);
 546                 while (isalnum(c = gpbc()) || c == '_') {
 547                         if (sp < 0)
 548                                 putc(c, active);
 549                         else
 550                                 CHRSAVE(c);
 551                 }
 552                 *name = EOS;
 553                 return nil;
 554         }
 555
 556         for (p = hashtab[h % HASHSIZE]; p != nil; p = p->nxtptr)
 557                 if (h == p->hv && STREQ(name, p->name))
 558                         break;
 559         return p;
 560 }
 561
 562 /*
 563  * initkwds - initialise m4 keywords as fast as possible.
 564  * This very similar to install, but without certain overheads,
 565  * such as calling lookup. Malloc is not used for storing the
 566  * keyword strings, since we simply use the static pointers
 567  * within keywrds block.
 568  */
 569 static void
 570 initkwds(void)
 571 {
 572         size_t i;
 573         unsigned int h;
 574         ndptr p;
 575
 576         for (i = 0; i < MAXKEYS; i++) {
 577                 h = hash(keywrds[i].knam);
 578                 p = (ndptr) xalloc(sizeof(struct ndblock));
 579                 p->nxtptr = hashtab[h % HASHSIZE];
 580                 hashtab[h % HASHSIZE] = p;
 581                 p->name = xstrdup(keywrds[i].knam);
 582                 p->defn = null;
 583                 p->hv = h;
 584                 p->type = keywrds[i].ktyp & TYPEMASK;
 585                 if ((keywrds[i].ktyp & NOARGS) == 0)
 586                         p->type |= NEEDARGS;
 587         }
 588 }
 589
 590 /* Look up a builtin type, even if overridden by the user */
 591 int
 592 builtin_type(const char *key)
 593 {
 594         int i;
 595
 596         for (i = 0; i != MAXKEYS; i++)
 597                 if (STREQ(keywrds[i].knam, key))
 598                         return keywrds[i].ktyp;
 599         return -1;
 600 }
 601
 602 const char *
 603 builtin_realname(int n)
 604 {
 605         int i;
 606
 607         for (i = 0; i != MAXKEYS; i++)
 608                 if (((keywrds[i].ktyp ^ n) & TYPEMASK) == 0)
 609                         return keywrds[i].knam;
 610         return NULL;
 611 }
 612
 613 static void
 614 record(struct position *t, int lev)
 615 {
 616         if (lev < MAXRECORD) {
 617                 t[lev].name = CURRENT_NAME;
 618                 t[lev].line = CURRENT_LINE;
 619         }
 620 }
 621
 622 static void
 623 dump_stack(struct position *t, int lev)
 624 {
 625         int i;
 626
 627         for (i = 0; i < lev; i++) {
 628                 if (i == MAXRECORD) {
 629                         fprintf(stderr, "   ...\n");
 630                         break;
 631                 }
 632                 fprintf(stderr, "   %s at line %lu\n",
 633                         t[i].name, t[i].line);
 634         }
 635 }
 636
 637
 638 static void
 639 enlarge_stack(void)
 640 {
 641         STACKMAX *= 2;
 642         mstack = realloc(mstack, sizeof(stae) * STACKMAX);
 643         sstack = realloc(sstack, STACKMAX);
 644         if (mstack == NULL || sstack == NULL)
 645                 errx(1, "Evaluation stack overflow (%lu)",
 646                     (unsigned long)STACKMAX);
 647 }
 648
 649 /* Emit preprocessor #line directive if -s option used. */
 650 void
 651 emitline(void)
 652 {
 653
 654         if (synccpp)
 655                 fprintf(active, "#line %d \"%s\"\n", inlineno[ilevel],
 656                         inname[ilevel]);
 657 }