usr.bin/m4/main.c

   1 /*      $OpenBSD: main.c,v 1.86 2015/11/03 16:21:47 deraadt 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. 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 /*
  37  * main.c
  38  * Facility: m4 macro processor
  39  * by: oz
  40  */
  41 #include <sys/cdefs.h>
  42 __FBSDID("$FreeBSD$");
  43
  44 #include <assert.h>
  45 #include <signal.h>
  46 #include <err.h>
  47 #include <errno.h>
  48 #include <unistd.h>
  49 #include <stdio.h>
  50 #include <ctype.h>
  51 #include <string.h>
  52 #include <stddef.h>
  53 #include <stdint.h>
  54 #include <stdlib.h>
  55 #include <ohash.h>
  56 #include "mdef.h"
  57 #include "stdd.h"
  58 #include "extern.h"
  59 #include "pathnames.h"
  60
  61 stae *mstack;                   /* stack of m4 machine         */
  62 char *sstack;                   /* shadow stack, for string space extension */
  63 static size_t STACKMAX;         /* current maximum size of stack */
  64 int sp;                         /* current m4  stack pointer   */
  65 int fp;                         /* m4 call frame pointer       */
  66 struct input_file infile[MAXINP];/* input file stack (0=stdin)  */
  67 FILE **outfile;                 /* diversion array(0=bitbucket)*/
  68 int maxout;
  69 FILE *active;                   /* active output file pointer  */
  70 int ilevel = 0;                 /* input file stack pointer    */
  71 int oindex = 0;                 /* diversion index..           */
  72 const char *null = "";                /* as it says.. just a null..  */
  73 char **m4wraps = NULL;          /* m4wraps array.              */
  74 int maxwraps = 0;               /* size of m4wraps array       */
  75 int wrapindex = 0;              /* current offset in m4wraps   */
  76 char lquote[MAXCCHARS+1] = {LQUOTE};    /* left quote character  (`)   */
  77 char rquote[MAXCCHARS+1] = {RQUOTE};    /* right quote character (')   */
  78 char scommt[MAXCCHARS+1] = {SCOMMT};    /* start character for comment */
  79 char ecommt[MAXCCHARS+1] = {ECOMMT};    /* end character for comment   */
  80 int  synch_lines = 0;           /* line synchronisation for C preprocessor */
  81 int  prefix_builtins = 0;       /* -P option to prefix builtin keywords */
  82
  83 struct keyblk {
  84         const char    *knam;          /* keyword name */
  85         int     ktyp;           /* keyword type */
  86 };
  87
  88 static struct keyblk keywrds[] = {      /* m4 keywords to be installed */
  89         { "include",      INCLTYPE },
  90         { "sinclude",     SINCTYPE },
  91         { "define",       DEFITYPE },
  92         { "defn",         DEFNTYPE },
  93         { "divert",       DIVRTYPE | NOARGS },
  94         { "expr",         EXPRTYPE },
  95         { "eval",         EXPRTYPE },
  96         { "substr",       SUBSTYPE },
  97         { "ifelse",       IFELTYPE },
  98         { "ifdef",        IFDFTYPE },
  99         { "len",          LENGTYPE },
 100         { "incr",         INCRTYPE },
 101         { "decr",         DECRTYPE },
 102         { "dnl",          DNLNTYPE | NOARGS },
 103         { "changequote",  CHNQTYPE | NOARGS },
 104         { "changecom",    CHNCTYPE | NOARGS },
 105         { "index",        INDXTYPE },
 106 #ifdef EXTENDED
 107         { "paste",        PASTTYPE },
 108         { "spaste",       SPASTYPE },
 109         /* Newer extensions, needed to handle gnu-m4 scripts */
 110         { "indir",        INDIRTYPE},
 111         { "builtin",      BUILTINTYPE},
 112         { "patsubst",     PATSTYPE},
 113         { "regexp",       REGEXPTYPE},
 114         { "esyscmd",      ESYSCMDTYPE},
 115         { "__file__",     FILENAMETYPE | NOARGS},
 116         { "__line__",     LINETYPE | NOARGS},
 117 #endif
 118         { "popdef",       POPDTYPE },
 119         { "pushdef",      PUSDTYPE },
 120         { "dumpdef",      DUMPTYPE | NOARGS },
 121         { "shift",        SHIFTYPE | NOARGS },
 122         { "translit",     TRNLTYPE },
 123         { "undefine",     UNDFTYPE },
 124         { "undivert",     UNDVTYPE | NOARGS },
 125         { "divnum",       DIVNTYPE | NOARGS },
 126         { "maketemp",     MKTMTYPE },
 127         { "mkstemp",      MKTMTYPE },
 128         { "errprint",     ERRPTYPE | NOARGS },
 129         { "m4wrap",       M4WRTYPE | NOARGS },
 130         { "m4exit",       EXITTYPE | NOARGS },
 131         { "syscmd",       SYSCTYPE },
 132         { "sysval",       SYSVTYPE | NOARGS },
 133         { "traceon",      TRACEONTYPE | NOARGS },
 134         { "traceoff",     TRACEOFFTYPE | NOARGS },
 135
 136         { "unix",         SELFTYPE | NOARGS },
 137 };
 138
 139 #define MAXKEYS (sizeof(keywrds)/sizeof(struct keyblk))
 140
 141 extern int optind;
 142 extern char *optarg;
 143
 144 #define MAXRECORD 50
 145 static struct position {
 146         char *name;
 147         unsigned long line;
 148 } quotes[MAXRECORD], paren[MAXRECORD];
 149
 150 static void record(struct position *, int);
 151 static void dump_stack(struct position *, int);
 152
 153 static void macro(void);
 154 static void initkwds(void);
 155 static ndptr inspect(int, char *);
 156 static int do_look_ahead(int, const char *);
 157 static void reallyoutputstr(const char *);
 158 static void reallyputchar(int);
 159
 160 static void enlarge_stack(void);
 161
 162 int main(int, char *[]);
 163
 164 int exit_code = 0;
 165
 166 int
 167 main(int argc, char *argv[])
 168 {
 169         int c;
 170         int n;
 171         char *p;
 172
 173         if (signal(SIGINT, SIG_IGN) != SIG_IGN)
 174                 signal(SIGINT, onintr);
 175
 176         init_macros();
 177         initspaces();
 178         STACKMAX = INITSTACKMAX;
 179
 180         mstack = xreallocarray(NULL, STACKMAX, sizeof(stae), NULL);
 181         sstack = xalloc(STACKMAX, NULL);
 182
 183         maxout = 0;
 184         outfile = NULL;
 185         resizedivs(MAXOUT);
 186
 187         while ((c = getopt(argc, argv, "gst:d:D:U:o:I:P")) != -1)
 188                 switch(c) {
 189
 190                 case 'D':               /* define something..*/
 191                         for (p = optarg; *p; p++)
 192                                 if (*p == '=')
 193                                         break;
 194                         if (*p)
 195                                 *p++ = EOS;
 196                         dodefine(optarg, p);
 197                         break;
 198                 case 'I':
 199                         addtoincludepath(optarg);
 200                         break;
 201                 case 'P':
 202                         prefix_builtins = 1;
 203                         break;
 204                 case 'U':               /* undefine...       */
 205                         macro_popdef(optarg);
 206                         break;
 207                 case 'g':
 208                         mimic_gnu = 1;
 209                         break;
 210                 case 'd':
 211                         set_trace_flags(optarg);
 212                         break;
 213                 case 's':
 214                         synch_lines = 1;
 215                         break;
 216                 case 't':
 217                         mark_traced(optarg, 1);
 218                         break;
 219                 case 'o':
 220                         trace_file(optarg);
 221                         break;
 222                 case '?':
 223                         usage();
 224                 }
 225
 226         argc -= optind;
 227         argv += optind;
 228
 229         initkwds();
 230         if (mimic_gnu)
 231                 setup_builtin("format", FORMATTYPE);
 232
 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                 macro();
 241         } else
 242                 for (; argc--; ++argv) {
 243                         p = *argv;
 244                         if (p[0] == '-' && p[1] == EOS)
 245                                 set_input(infile, stdin, "stdin");
 246                         else if (fopen_trypath(infile, p) == NULL)
 247                                 err(1, "%s", p);
 248                         sp = -1;
 249                         fp = 0;
 250                         macro();
 251                         release_input(infile);
 252                 }
 253
 254         if (wrapindex) {
 255                 int i;
 256
 257                 ilevel = 0;             /* in case m4wrap includes.. */
 258                 bufbase = bp = buf;     /* use the entire buffer   */
 259                 if (mimic_gnu) {
 260                         while (wrapindex != 0) {
 261                                 for (i = 0; i < wrapindex; i++)
 262                                         pbstr(m4wraps[i]);
 263                                 wrapindex =0;
 264                                 macro();
 265                         }
 266                 } else {
 267                         for (i = 0; i < wrapindex; i++) {
 268                                 pbstr(m4wraps[i]);
 269                                 macro();
 270                         }
 271                 }
 272         }
 273
 274         if (active != stdout)
 275                 active = stdout;        /* reset output just in case */
 276         for (n = 1; n < maxout; n++)    /* default wrap-up: undivert */
 277                 if (outfile[n] != NULL)
 278                         getdiv(n);
 279                                         /* remove bitbucket if used  */
 280         if (outfile[0] != NULL) {
 281                 (void) fclose(outfile[0]);
 282         }
 283
 284         return exit_code;
 285 }
 286
 287 /*
 288  * Look ahead for `token'.
 289  * (on input `t == token[0]')
 290  * Used for comment and quoting delimiters.
 291  * Returns 1 if `token' present; copied to output.
 292  *         0 if `token' not found; all characters pushed back
 293  */
 294 static int
 295 do_look_ahead(int t, const char *token)
 296 {
 297         int i;
 298
 299         assert((unsigned char)t == (unsigned char)token[0]);
 300
 301         for (i = 1; *++token; i++) {
 302                 t = gpbc();
 303                 if (t == EOF || (unsigned char)t != (unsigned char)*token) {
 304                         pushback(t);
 305                         while (--i)
 306                                 pushback(*--token);
 307                         return 0;
 308                 }
 309         }
 310         return 1;
 311 }
 312
 313 #define LOOK_AHEAD(t, token) (t != EOF &&               \
 314     (unsigned char)(t)==(unsigned char)(token)[0] &&    \
 315     do_look_ahead(t,token))
 316
 317 /*
 318  * macro - the work horse..
 319  */
 320 static void
 321 macro(void)
 322 {
 323         char token[MAXTOK+1];
 324         int t, l;
 325         ndptr p;
 326         int  nlpar;
 327
 328         cycle {
 329                 t = gpbc();
 330
 331                 if (LOOK_AHEAD(t,lquote)) {     /* strip quotes */
 332                         nlpar = 0;
 333                         record(quotes, nlpar++);
 334                         /*
 335                          * Opening quote: scan forward until matching
 336                          * closing quote has been found.
 337                          */
 338                         do {
 339
 340                                 l = gpbc();
 341                                 if (LOOK_AHEAD(l,rquote)) {
 342                                         if (--nlpar > 0)
 343                                                 outputstr(rquote);
 344                                 } else if (LOOK_AHEAD(l,lquote)) {
 345                                         record(quotes, nlpar++);
 346                                         outputstr(lquote);
 347                                 } else if (l == EOF) {
 348                                         if (nlpar == 1)
 349                                                 warnx("unclosed quote:");
 350                                         else
 351                                                 warnx("%d unclosed quotes:", nlpar);
 352                                         dump_stack(quotes, nlpar);
 353                                         exit(1);
 354                                 } else {
 355                                         if (nlpar > 0) {
 356                                                 if (sp < 0)
 357                                                         reallyputchar(l);
 358                                                 else
 359                                                         CHRSAVE(l);
 360                                         }
 361                                 }
 362                         }
 363                         while (nlpar != 0);
 364                 } else if (sp < 0 && LOOK_AHEAD(t, scommt)) {
 365                         reallyoutputstr(scommt);
 366
 367                         for(;;) {
 368                                 t = gpbc();
 369                                 if (LOOK_AHEAD(t, ecommt)) {
 370                                         reallyoutputstr(ecommt);
 371                                         break;
 372                                 }
 373                                 if (t == EOF)
 374                                         break;
 375                                 reallyputchar(t);
 376                         }
 377                 } else if (t == '_' || isalpha(t)) {
 378                         p = inspect(t, token);
 379                         if (p != NULL)
 380                                 pushback(l = gpbc());
 381                         if (p == NULL || (l != LPAREN &&
 382                             (macro_getdef(p)->type & NEEDARGS) != 0))
 383                                 outputstr(token);
 384                         else {
 385                 /*
 386                  * real thing.. First build a call frame:
 387                  */
 388                                 pushf(fp);      /* previous call frm */
 389                                 pushf(macro_getdef(p)->type); /* type of the call  */
 390                                 pushf(is_traced(p));
 391                                 pushf(0);       /* parenthesis level */
 392                                 fp = sp;        /* new frame pointer */
 393                 /*
 394                  * now push the string arguments:
 395                  */
 396                                 pushdef(p);                     /* defn string */
 397                                 pushs1((char *)macro_name(p));  /* macro name  */
 398                                 pushs(ep);                      /* start next..*/
 399
 400                                 if (l != LPAREN && PARLEV == 0) {
 401                                     /* no bracks  */
 402                                         chrsave(EOS);
 403
 404                                         if (sp == (int)STACKMAX)
 405                                                 errx(1, "internal stack overflow");
 406                                         eval((const char **) mstack+fp+1, 2,
 407                                             CALTYP, TRACESTATUS);
 408
 409                                         ep = PREVEP;    /* flush strspace */
 410                                         sp = PREVSP;    /* previous sp..  */
 411                                         fp = PREVFP;    /* rewind stack...*/
 412                                 }
 413                         }
 414                 } else if (t == EOF) {
 415                         if (!mimic_gnu /* you can puke right there */
 416                             && sp > -1 && ilevel <= 0) {
 417                                 warnx( "unexpected end of input, unclosed parenthesis:");
 418                                 dump_stack(paren, PARLEV);
 419                                 exit(1);
 420                         }
 421                         if (ilevel <= 0)
 422                                 break;                  /* all done thanks.. */
 423                         release_input(infile+ilevel--);
 424                         emit_synchline();
 425                         bufbase = bbase[ilevel];
 426                         continue;
 427                 } else if (sp < 0) {            /* not in a macro at all */
 428                         reallyputchar(t);       /* output directly..     */
 429                 }
 430
 431                 else switch(t) {
 432
 433                 case LPAREN:
 434                         if (PARLEV > 0)
 435                                 chrsave(t);
 436                         while (isspace(l = gpbc())) /* skip blank, tab, nl.. */
 437                                 if (PARLEV > 0)
 438                                         chrsave(l);
 439                         pushback(l);
 440                         record(paren, PARLEV++);
 441                         break;
 442
 443                 case RPAREN:
 444                         if (--PARLEV > 0)
 445                                 chrsave(t);
 446                         else {                  /* end of argument list */
 447                                 chrsave(EOS);
 448
 449                                 if (sp == (int)STACKMAX)
 450                                         errx(1, "internal stack overflow");
 451
 452                                 eval((const char **) mstack+fp+1, sp-fp,
 453                                     CALTYP, TRACESTATUS);
 454
 455                                 ep = PREVEP;    /* flush strspace */
 456                                 sp = PREVSP;    /* previous sp..  */
 457                                 fp = PREVFP;    /* rewind stack...*/
 458                         }
 459                         break;
 460
 461                 case COMMA:
 462                         if (PARLEV == 1) {
 463                                 chrsave(EOS);           /* new argument   */
 464                                 while (isspace(l = gpbc()))
 465                                         ;
 466                                 pushback(l);
 467                                 pushs(ep);
 468                         } else
 469                                 chrsave(t);
 470                         break;
 471
 472                 default:
 473                         if (LOOK_AHEAD(t, scommt)) {
 474                                 char *p;
 475                                 for (p = scommt; *p; p++)
 476                                         chrsave(*p);
 477                                 for(;;) {
 478                                         t = gpbc();
 479                                         if (LOOK_AHEAD(t, ecommt)) {
 480                                                 for (p = ecommt; *p; p++)
 481                                                         chrsave(*p);
 482                                                 break;
 483                                         }
 484                                         if (t == EOF)
 485                                             break;
 486                                         CHRSAVE(t);
 487                                 }
 488                         } else
 489                                 CHRSAVE(t);             /* stack the char */
 490                         break;
 491                 }
 492         }
 493 }
 494
 495 /*
 496  * output string directly, without pushing it for reparses.
 497  */
 498 void
 499 outputstr(const char *s)
 500 {
 501         if (sp < 0)
 502                 reallyoutputstr(s);
 503         else
 504                 while (*s)
 505                         CHRSAVE(*s++);
 506 }
 507
 508 void
 509 reallyoutputstr(const char *s)
 510 {
 511         if (synch_lines) {
 512                 while (*s) {
 513                         fputc(*s, active);
 514                         if (*s++ == '\n') {
 515                                 infile[ilevel].synch_lineno++;
 516                                 if (infile[ilevel].synch_lineno !=
 517                                     infile[ilevel].lineno)
 518                                         do_emit_synchline();
 519                         }
 520                 }
 521         } else
 522                 fputs(s, active);
 523 }
 524
 525 void
 526 reallyputchar(int c)
 527 {
 528         putc(c, active);
 529         if (synch_lines && c == '\n') {
 530                 infile[ilevel].synch_lineno++;
 531                 if (infile[ilevel].synch_lineno != infile[ilevel].lineno)
 532                         do_emit_synchline();
 533         }
 534 }
 535
 536 /*
 537  * build an input token..
 538  * consider only those starting with _ or A-Za-z.
 539  */
 540 static ndptr
 541 inspect(int c, char *tp)
 542 {
 543         char *name = tp;
 544         char *etp = tp+MAXTOK;
 545         ndptr p;
 546
 547         *tp++ = c;
 548
 549         while ((isalnum(c = gpbc()) || c == '_') && tp < etp)
 550                 *tp++ = c;
 551         if (c != EOF)
 552                 PUSHBACK(c);
 553         *tp = EOS;
 554         /* token is too long, it won't match anything, but it can still
 555          * be output. */
 556         if (tp == ep) {
 557                 outputstr(name);
 558                 while (isalnum(c = gpbc()) || c == '_') {
 559                         if (sp < 0)
 560                                 reallyputchar(c);
 561                         else
 562                                 CHRSAVE(c);
 563                 }
 564                 *name = EOS;
 565                 return NULL;
 566         }
 567
 568         p = ohash_find(&macros, ohash_qlookupi(&macros, name, (const char **)&tp));
 569         if (p == NULL)
 570                 return NULL;
 571         if (macro_getdef(p) == NULL)
 572                 return NULL;
 573         return p;
 574 }
 575
 576 /*
 577  * initkwds - initialise m4 keywords as fast as possible.
 578  * This very similar to install, but without certain overheads,
 579  * such as calling lookup. Malloc is not used for storing the
 580  * keyword strings, since we simply use the static pointers
 581  * within keywrds block.
 582  */
 583 static void
 584 initkwds(void)
 585 {
 586         unsigned int type;
 587         int i;
 588
 589         for (i = 0; i < (int)MAXKEYS; i++) {
 590                 type = keywrds[i].ktyp & TYPEMASK;
 591                 if ((keywrds[i].ktyp & NOARGS) == 0)
 592                         type |= NEEDARGS;
 593                 setup_builtin(keywrds[i].knam, type);
 594         }
 595 }
 596
 597 static void
 598 record(struct position *t, int lev)
 599 {
 600         if (lev < MAXRECORD) {
 601                 t[lev].name = CURRENT_NAME;
 602                 t[lev].line = CURRENT_LINE;
 603         }
 604 }
 605
 606 static void
 607 dump_stack(struct position *t, int lev)
 608 {
 609         int i;
 610
 611         for (i = 0; i < lev; i++) {
 612                 if (i == MAXRECORD) {
 613                         fprintf(stderr, "   ...\n");
 614                         break;
 615                 }
 616                 fprintf(stderr, "   %s at line %lu\n",
 617                         t[i].name, t[i].line);
 618         }
 619 }
 620
 621
 622 static void
 623 enlarge_stack(void)
 624 {
 625         STACKMAX += STACKMAX/2;
 626         mstack = xreallocarray(mstack, STACKMAX, sizeof(stae),
 627             "Evaluation stack overflow (%lu)",
 628             (unsigned long)STACKMAX);
 629         sstack = xrealloc(sstack, STACKMAX,
 630             "Evaluation stack overflow (%lu)",
 631             (unsigned long)STACKMAX);
 632 }