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  * 3. All advertising materials mentioning features or use of this software
  20  *    must display the following acknowledgement:
  21  *      This product includes software developed by the University of
  22  *      California, Berkeley and its contributors.
  23  * 4. Neither the name of the University nor the names of its contributors
  24  *    may be used to endorse or promote products derived from this software
  25  *    without specific prior written permission.
  26  *
  27  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  28  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  29  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  30  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  31  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  32  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  33  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  34  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  35  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  36  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  37  * SUCH DAMAGE.
  38  *
  39  * @(#) Copyright (c) 1989, 1993 The Regents of the University of California.  All rights reserved.
  40  * @(#)main.c   8.1 (Berkeley) 6/6/93
  41  * $OpenBSD: main.c,v 1.53 2002/04/26 16:15:16 espie Exp $
  42  * $FreeBSD: src/usr.bin/m4/main.c,v 1.26 2004/08/16 14:18:22 tjr Exp $
  43  * $DragonFly: src/usr.bin/m4/main.c,v 1.3 2006/12/27 21:29:02 pavalos Exp $
  44  */
  45
  46 /*
  47  * main.c
  48  * Facility: m4 macro processor
  49  * by: oz
  50  */
  51
  52 #include <sys/types.h>
  53 #include <assert.h>
  54 #include <signal.h>
  55 #include <errno.h>
  56 #include <unistd.h>
  57 #include <stdio.h>
  58 #include <ctype.h>
  59 #include <string.h>
  60 #include <stddef.h>
  61 #include <stdlib.h>
  62 #include <err.h>
  63 #include <locale.h>
  64 #include "mdef.h"
  65 #include "stdd.h"
  66 #include "extern.h"
  67 #include "pathnames.h"
  68
  69 ndptr hashtab[HASHSIZE];        /* hash table for macros etc.  */
  70 stae *mstack;                   /* stack of m4 machine         */
  71 char *sstack;                   /* shadow stack, for string space extension */
  72 static size_t STACKMAX;         /* current maximum size of stack */
  73 int sp;                         /* current m4  stack pointer   */
  74 int fp;                         /* m4 call frame pointer       */
  75 struct input_file infile[MAXINP];/* input file stack (0=stdin)  */
  76 char *inname[MAXINP];           /* names of these input files */
  77 int inlineno[MAXINP];           /* current number in each input file */
  78 FILE **outfile;                 /* diversion array(0=bitbucket)*/
  79 int maxout;
  80 FILE *active;                   /* active output file pointer  */
  81 int ilevel = 0;                 /* input file stack pointer    */
  82 int oindex = 0;                 /* diversion index..           */
  83 char null[] = "";               /* as it says.. just a null..  */
  84 const char *m4wraps = "";       /* m4wrap string default..     */
  85 char lquote[MAXCCHARS+1] = {LQUOTE};    /* left quote character  (`)   */
  86 char rquote[MAXCCHARS+1] = {RQUOTE};    /* right quote character (')   */
  87 char scommt[MAXCCHARS+1] = {SCOMMT};    /* start character for comment */
  88 char ecommt[MAXCCHARS+1] = {ECOMMT};    /* end character for comment   */
  89 int synccpp;                    /* Line synchronisation for C preprocessor */
  90
  91 struct keyblk keywrds[] = {     /* m4 keywords to be installed */
  92         { "include",      INCLTYPE },
  93         { "sinclude",     SINCTYPE },
  94         { "define",       DEFITYPE },
  95         { "defn",         DEFNTYPE },
  96         { "divert",       DIVRTYPE | NOARGS },
  97         { "expr",         EXPRTYPE },
  98         { "eval",         EXPRTYPE },
  99         { "substr",       SUBSTYPE },
 100         { "ifelse",       IFELTYPE },
 101         { "ifdef",        IFDFTYPE },
 102         { "len",          LENGTYPE },
 103         { "incr",         INCRTYPE },
 104         { "decr",         DECRTYPE },
 105         { "dnl",          DNLNTYPE | NOARGS },
 106         { "changequote",  CHNQTYPE | NOARGS },
 107         { "changecom",    CHNCTYPE | NOARGS },
 108         { "index",        INDXTYPE },
 109 #ifdef EXTENDED
 110         { "paste",        PASTTYPE },
 111         { "spaste",       SPASTYPE },
 112         /* Newer extensions, needed to handle gnu-m4 scripts */
 113         { "indir",        INDIRTYPE},
 114         { "builtin",      BUILTINTYPE},
 115         { "patsubst",     PATSTYPE},
 116         { "regexp",       REGEXPTYPE},
 117         { "esyscmd",      ESYSCMDTYPE},
 118         { "__file__",     FILENAMETYPE | NOARGS},
 119         { "__line__",     LINETYPE | NOARGS},
 120 #endif
 121         { "popdef",       POPDTYPE },
 122         { "pushdef",      PUSDTYPE },
 123         { "dumpdef",      DUMPTYPE | NOARGS },
 124         { "shift",        SHIFTYPE | NOARGS },
 125         { "translit",     TRNLTYPE },
 126         { "undefine",     UNDFTYPE },
 127         { "undivert",     UNDVTYPE | NOARGS },
 128         { "divnum",       DIVNTYPE | NOARGS },
 129         { "maketemp",     MKTMTYPE },
 130         { "errprint",     ERRPTYPE | NOARGS },
 131         { "m4wrap",       M4WRTYPE | NOARGS },
 132         { "m4exit",       EXITTYPE | NOARGS },
 133         { "syscmd",       SYSCTYPE },
 134         { "sysval",       SYSVTYPE | NOARGS },
 135         { "traceon",      TRACEONTYPE | NOARGS },
 136         { "traceoff",     TRACEOFFTYPE | NOARGS },
 137
 138 #if defined(unix) || defined(__unix__)
 139         { "unix",         SELFTYPE | NOARGS },
 140 #else
 141 #ifdef vms
 142         { "vms",          SELFTYPE | NOARGS },
 143 #endif
 144 #endif
 145 };
 146
 147 #define MAXKEYS (sizeof(keywrds)/sizeof(struct keyblk))
 148
 149 #define MAXRECORD 50
 150 static struct position {
 151         char *name;
 152         unsigned long line;
 153 } quotes[MAXRECORD], paren[MAXRECORD];
 154
 155 static void record(struct position *, int);
 156 static void dump_stack(struct position *, int);
 157
 158 static void macro(void);
 159 static void initkwds(void);
 160 static ndptr inspect(int, char *);
 161 static int do_look_ahead(int, const char *);
 162
 163 static void enlarge_stack(void);
 164
 165 int
 166 main(int argc, char *argv[])
 167 {
 168         int c;
 169         int n;
 170         int rval;
 171         char *p;
 172
 173         setlocale(LC_ALL, "");
 174
 175         traceout = stderr;
 176
 177         if (signal(SIGINT, SIG_IGN) != SIG_IGN)
 178                 signal(SIGINT, onintr);
 179
 180         initkwds();
 181         initspaces();
 182         STACKMAX = INITSTACKMAX;
 183
 184         mstack = (stae *)xalloc(sizeof(stae) * STACKMAX);
 185         sstack = (char *)xalloc(STACKMAX);
 186
 187         maxout = 0;
 188         outfile = NULL;
 189         resizedivs(MAXOUT);
 190
 191         while ((c = getopt(argc, argv, "gst:d:D:U:o:I:")) != -1)
 192                 switch(c) {
 193                 case 'D':               /* define something..*/
 194                         for (p = optarg; *p; p++)
 195                                 if (*p == '=')
 196                                         break;
 197                         if (p == optarg)
 198                                 errx(1, "null variable cannot be defined");
 199                         if (*p)
 200                                 *p++ = EOS;
 201                         dodefine(optarg, p);
 202                         break;
 203                 case 'I':
 204                         addtoincludepath(optarg);
 205                         break;
 206                 case 'U':               /* undefine...       */
 207                         remhash(optarg, TOP);
 208                         break;
 209                 case 'g':
 210                         mimic_gnu = 1;
 211                         break;
 212                 case 'd':
 213                         set_trace_flags(optarg);
 214                         break;
 215                 case 's':
 216                         synccpp = 1;
 217                         break;
 218                 case 't':
 219                         mark_traced(optarg, 1);
 220                         break;
 221                 case 'o':
 222                         trace_file(optarg);
 223                         break;
 224                 case '?':
 225                 default:
 226                         usage();
 227                 }
 228
 229         argc -= optind;
 230         argv += optind;
 231
 232         rval = 0;
 233         active = stdout;                /* default active output     */
 234         bbase[0] = bufbase;
 235         if (!argc) {
 236                 sp = -1;                /* stack pointer initialized */
 237                 fp = 0;                 /* frame pointer initialized */
 238                 set_input(infile+0, stdin, "stdin");
 239                                         /* default input (naturally) */
 240                 if ((inname[0] = strdup("-")) == NULL)
 241                         err(1, NULL);
 242                 inlineno[0] = 1;
 243                 emitline();
 244                 macro();
 245         } else
 246                 for (; argc--; ++argv) {
 247                         p = *argv;
 248                         if (p[0] == '-' && p[1] == EOS)
 249                                 set_input(infile, stdin, "stdin");
 250                         else if (fopen_trypath(infile, p) == NULL) {
 251                                 warn("%s", p);
 252                                 rval = 1;
 253                                 continue;
 254                         }
 255                         sp = -1;
 256                         fp = 0;
 257                         if ((inname[0] = strdup(p)) == NULL)
 258                                 err(1, NULL);
 259                         inlineno[0] = 1;
 260                         emitline();
 261                         macro();
 262                         release_input(infile);
 263                 }
 264
 265         if (*m4wraps) {                 /* anything for rundown ??   */
 266                 ilevel = 0;             /* in case m4wrap includes.. */
 267                 bufbase = bp = buf;     /* use the entire buffer   */
 268                 pbstr(m4wraps);         /* user-defined wrapup act   */
 269                 macro();                /* last will and testament   */
 270         }
 271
 272         if (active != stdout)
 273                 active = stdout;        /* reset output just in case */
 274         for (n = 1; n < maxout; n++)    /* default wrap-up: undivert */
 275                 if (outfile[n] != NULL)
 276                         getdiv(n);
 277                                         /* remove bitbucket if used  */
 278         if (outfile[0] != NULL) {
 279                 (void) fclose(outfile[0]);
 280         }
 281
 282         exit(rval);
 283 }
 284
 285 /*
 286  * Look ahead for `token'.
 287  * (on input `t == token[0]')
 288  * Used for comment and quoting delimiters.
 289  * Returns 1 if `token' present; copied to output.
 290  *         0 if `token' not found; all characters pushed back
 291  */
 292 static int
 293 do_look_ahead(int t, const char *token)
 294 {
 295         int i;
 296
 297         assert((unsigned char)t == (unsigned char)token[0]);
 298
 299         for (i = 1; *++token; i++) {
 300                 t = gpbc();
 301                 if (t == EOF || (unsigned char)t != (unsigned char)*token) {
 302                         putback(t);
 303                         while (--i)
 304                                 putback(*--token);
 305                         return 0;
 306                 }
 307         }
 308         return 1;
 309 }
 310
 311 #define LOOK_AHEAD(t, token) (t != EOF &&               \
 312     (unsigned char)(t)==(unsigned char)(token)[0] &&    \
 313     do_look_ahead(t,token))
 314
 315 /*
 316  * macro - the work horse..
 317  */
 318 static void
 319 macro(void)
 320 {
 321         char token[MAXTOK+1];
 322         int t, l;
 323         ndptr p;
 324         int  nlpar;
 325
 326         cycle {
 327                 t = gpbc();
 328                 if (t == '_' || isalpha(t)) {
 329                         p = inspect(t, token);
 330                         if (p != nil)
 331                                 putback(l = gpbc());
 332                         if (p == nil || (l != LPAREN &&
 333                             (p->type & NEEDARGS) != 0))
 334                                 outputstr(token);
 335                         else {
 336                 /*
 337                  * real thing.. First build a call frame:
 338                  */
 339                                 pushf(fp);      /* previous call frm */
 340                                 pushf(p->type); /* type of the call  */
 341                                 pushf(0);       /* parenthesis level */
 342                                 fp = sp;        /* new frame pointer */
 343                 /*
 344                  * now push the string arguments:
 345                  */
 346                                 pushs1(p->defn);        /* defn string */
 347                                 pushs1(p->name);        /* macro name  */
 348                                 pushs(ep);              /* start next..*/
 349
 350                                 if (l != LPAREN && PARLEV == 0)  {
 351                                     /* no bracks  */
 352                                         chrsave(EOS);
 353
 354                                         if ((uintptr_t)sp == STACKMAX)
 355                                                 errx(1, "internal stack overflow");
 356                                         eval((const char **) mstack+fp+1, 2,
 357                                             CALTYP);
 358
 359                                         ep = PREVEP;    /* flush strspace */
 360                                         sp = PREVSP;    /* previous sp..  */
 361                                         fp = PREVFP;    /* rewind stack...*/
 362                                 }
 363                         }
 364                 } else if (t == EOF) {
 365                         if (sp > -1) {
 366                                 warnx( "unexpected end of input, unclosed parenthesis:");
 367                                 dump_stack(paren, PARLEV);
 368                                 exit(1);
 369                         }
 370                         if (ilevel <= 0)
 371                                 break;                  /* all done thanks.. */
 372                         release_input(infile+ilevel--);
 373                         free(inname[ilevel+1]);
 374                         bufbase = bbase[ilevel];
 375                         emitline();
 376                         continue;
 377                 }
 378         /*
 379          * non-alpha token possibly seen..
 380          * [the order of else if .. stmts is important.]
 381          */
 382                 else if (LOOK_AHEAD(t,lquote)) {        /* strip quotes */
 383                         nlpar = 0;
 384                         record(quotes, nlpar++);
 385                         /*
 386                          * Opening quote: scan forward until matching
 387                          * closing quote has been found.
 388                          */
 389                         do {
 390
 391                                 l = gpbc();
 392                                 if (LOOK_AHEAD(l,rquote)) {
 393                                         if (--nlpar > 0)
 394                                                 outputstr(rquote);
 395                                 } else if (LOOK_AHEAD(l,lquote)) {
 396                                         record(quotes, nlpar++);
 397                                         outputstr(lquote);
 398                                 } else if (l == EOF) {
 399                                         if (nlpar == 1)
 400                                                 warnx("unclosed quote:");
 401                                         else
 402                                                 warnx("%d unclosed quotes:", nlpar);
 403                                         dump_stack(quotes, nlpar);
 404                                         exit(1);
 405                                 } else {
 406                                         if (nlpar > 0) {
 407                                                 if (sp < 0)
 408                                                         putc(l, active);
 409                                                 else
 410                                                         CHRSAVE(l);
 411                                         }
 412                                 }
 413                         }
 414                         while (nlpar != 0);
 415                 }
 416
 417                 else if (sp < 0 && LOOK_AHEAD(t, scommt)) {
 418                         fputs(scommt, active);
 419
 420                         for(;;) {
 421                                 t = gpbc();
 422                                 if (LOOK_AHEAD(t, ecommt)) {
 423                                         fputs(ecommt, active);
 424                                         break;
 425                                 }
 426                                 if (t == EOF)
 427                                         break;
 428                                 putc(t, active);
 429                         }
 430                 }
 431
 432                 else if (sp < 0) {              /* not in a macro at all */
 433                         putc(t, active);        /* output directly..     */
 434                 }
 435
 436                 else switch(t) {
 437
 438                 case LPAREN:
 439                         if (PARLEV > 0)
 440                                 chrsave(t);
 441                         while (isspace(l = gpbc()))
 442                                 ;               /* skip blank, tab, nl.. */
 443                         putback(l);
 444                         record(paren, PARLEV++);
 445                         break;
 446
 447                 case RPAREN:
 448                         if (--PARLEV > 0)
 449                                 chrsave(t);
 450                         else {                  /* end of argument list */
 451                                 chrsave(EOS);
 452
 453                                 if ((uintptr_t)sp == STACKMAX)
 454                                         errx(1, "internal stack overflow");
 455
 456                                 eval((const char **) mstack+fp+1, sp-fp,
 457                                     CALTYP);
 458
 459                                 ep = PREVEP;    /* flush strspace */
 460                                 sp = PREVSP;    /* previous sp..  */
 461                                 fp = PREVFP;    /* rewind stack...*/
 462                         }
 463                         break;
 464
 465                 case COMMA:
 466                         if (PARLEV == 1) {
 467                                 chrsave(EOS);           /* new argument   */
 468                                 while (isspace(l = gpbc()))
 469                                         ;
 470                                 putback(l);
 471                                 pushs(ep);
 472                         } else
 473                                 chrsave(t);
 474                         break;
 475
 476                 default:
 477                         if (LOOK_AHEAD(t, scommt)) {
 478                                 char *pc;
 479                                 for (pc = scommt; *pc; pc++)
 480                                         chrsave(*pc);
 481                                 for(;;) {
 482                                         t = gpbc();
 483                                         if (LOOK_AHEAD(t, ecommt)) {
 484                                                 for (pc = ecommt; *pc; pc++)
 485                                                         chrsave(*pc);
 486                                                 break;
 487                                         }
 488                                         if (t == EOF)
 489                                             break;
 490                                         CHRSAVE(t);
 491                                 }
 492                         } else
 493                                 CHRSAVE(t);             /* stack the char */
 494                         break;
 495                 }
 496         }
 497 }
 498
 499 /*
 500  * output string directly, without pushing it for reparses.
 501  */
 502 void
 503 outputstr(const char *s)
 504 {
 505         if (sp < 0)
 506                 while (*s)
 507                         putc(*s++, active);
 508         else
 509                 while (*s)
 510                         CHRSAVE(*s++);
 511 }
 512
 513 /*
 514  * build an input token..
 515  * consider only those starting with _ or A-Za-z. This is a
 516  * combo with lookup to speed things up.
 517  */
 518 static ndptr
 519 inspect(int c, char *tp)
 520 {
 521         char *name = tp;
 522         char *etp = tp+MAXTOK;
 523         ndptr p;
 524         unsigned int h;
 525
 526         h = *tp++ = c;
 527
 528         while ((isalnum(c = gpbc()) || c == '_') && tp < etp)
 529                 h = (h << 5) + h + (*tp++ = c);
 530         if (c != EOF)
 531                 PUTBACK(c);
 532         *tp = EOS;
 533         /* token is too long, it won't match anything, but it can still
 534          * be output. */
 535         if (tp == ep) {
 536                 outputstr(name);
 537                 while (isalnum(c = gpbc()) || c == '_') {
 538                         if (sp < 0)
 539                                 putc(c, active);
 540                         else
 541                                 CHRSAVE(c);
 542                 }
 543                 *name = EOS;
 544                 return nil;
 545         }
 546
 547         for (p = hashtab[h % HASHSIZE]; p != nil; p = p->nxtptr)
 548                 if (h == p->hv && STREQ(name, p->name))
 549                         break;
 550         return p;
 551 }
 552
 553 /*
 554  * initkwds - initialise m4 keywords as fast as possible.
 555  * This very similar to install, but without certain overheads,
 556  * such as calling lookup. Malloc is not used for storing the
 557  * keyword strings, since we simply use the static pointers
 558  * within keywrds block.
 559  */
 560 static void
 561 initkwds(void)
 562 {
 563         size_t i;
 564         unsigned int h;
 565         ndptr p;
 566
 567         for (i = 0; i < MAXKEYS; i++) {
 568                 h = hash(keywrds[i].knam);
 569                 p = (ndptr) xalloc(sizeof(struct ndblock));
 570                 p->nxtptr = hashtab[h % HASHSIZE];
 571                 hashtab[h % HASHSIZE] = p;
 572                 p->name = xstrdup(keywrds[i].knam);
 573                 p->defn = null;
 574                 p->hv = h;
 575                 p->type = keywrds[i].ktyp & TYPEMASK;
 576                 if ((keywrds[i].ktyp & NOARGS) == 0)
 577                         p->type |= NEEDARGS;
 578         }
 579 }
 580
 581 /* Look up a builtin type, even if overridden by the user */
 582 int
 583 builtin_type(const char *key)
 584 {
 585         int i;
 586
 587         for (i = 0; i != MAXKEYS; i++)
 588                 if (STREQ(keywrds[i].knam, key))
 589                         return keywrds[i].ktyp;
 590         return -1;
 591 }
 592
 593 const char *
 594 builtin_realname(int n)
 595 {
 596         int i;
 597
 598         for (i = 0; i != MAXKEYS; i++)
 599                 if (((keywrds[i].ktyp ^ n) & TYPEMASK) == 0)
 600                         return keywrds[i].knam;
 601         return NULL;
 602 }
 603
 604 static void
 605 record(struct position *t, int lev)
 606 {
 607         if (lev < MAXRECORD) {
 608                 t[lev].name = CURRENT_NAME;
 609                 t[lev].line = CURRENT_LINE;
 610         }
 611 }
 612
 613 static void
 614 dump_stack(struct position *t, int lev)
 615 {
 616         int i;
 617
 618         for (i = 0; i < lev; i++) {
 619                 if (i == MAXRECORD) {
 620                         fprintf(stderr, "   ...\n");
 621                         break;
 622                 }
 623                 fprintf(stderr, "   %s at line %lu\n",
 624                         t[i].name, t[i].line);
 625         }
 626 }
 627
 628
 629 static void
 630 enlarge_stack(void)
 631 {
 632         STACKMAX *= 2;
 633         mstack = realloc(mstack, sizeof(stae) * STACKMAX);
 634         sstack = realloc(sstack, STACKMAX);
 635         if (mstack == NULL || sstack == NULL)
 636                 errx(1, "Evaluation stack overflow (%lu)",
 637                     (unsigned long)STACKMAX);
 638 }
 639
 640 /* Emit preprocessor #line directive if -s option used. */
 641 void
 642 emitline(void)
 643 {
 644
 645         if (synccpp)
 646                 fprintf(active, "#line %d \"%s\"\n", inlineno[ilevel],
 647                         inname[ilevel]);
 648 }