usr.bin/m4/main.c

   1 /*      $OpenBSD: main.c,v 1.81 2012/04/12 17:00:11 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. 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 #if defined(unix) || defined(__unix__)
 137         { "unix",         SELFTYPE | NOARGS },
 138 #else
 139 #ifdef vms
 140         { "vms",          SELFTYPE | NOARGS },
 141 #endif
 142 #endif
 143 };
 144
 145 #define MAXKEYS (sizeof(keywrds)/sizeof(struct keyblk))
 146
 147 #define MAXRECORD 50
 148 static struct position {
 149         char *name;
 150         unsigned long line;
 151 } quotes[MAXRECORD], paren[MAXRECORD];
 152
 153 static void record(struct position *, int);
 154 static void dump_stack(struct position *, int);
 155
 156 static void macro(void);
 157 static void initkwds(void);
 158 static ndptr inspect(int, char *);
 159 static int do_look_ahead(int, const char *);
 160 static void reallyoutputstr(const char *);
 161 static void reallyputchar(int);
 162
 163 static void enlarge_stack(void);
 164
 165 int main(int, char *[]);
 166
 167 int exit_code = 0;
 168
 169 int
 170 main(int argc, char *argv[])
 171 {
 172         int c;
 173         int n;
 174         char *p;
 175
 176         if (signal(SIGINT, SIG_IGN) != SIG_IGN)
 177                 signal(SIGINT, onintr);
 178
 179         init_macros();
 180         initspaces();
 181         STACKMAX = INITSTACKMAX;
 182
 183         mstack = (stae *)xalloc(sizeof(stae) * STACKMAX, NULL);
 184         sstack = (char *)xalloc(STACKMAX, NULL);
 185
 186         maxout = 0;
 187         outfile = NULL;
 188         resizedivs(MAXOUT);
 189
 190         while ((c = getopt(argc, argv, "gst:d:D:U:o:I:P")) != -1)
 191                 switch(c) {
 192
 193                 case 'D':               /* define something..*/
 194                         for (p = optarg; *p; p++)
 195                                 if (*p == '=')
 196                                         break;
 197                         if (*p)
 198                                 *p++ = EOS;
 199                         dodefine(optarg, p);
 200                         break;
 201                 case 'I':
 202                         addtoincludepath(optarg);
 203                         break;
 204                 case 'P':
 205                         prefix_builtins = 1;
 206                         break;
 207                 case 'U':               /* undefine...       */
 208                         macro_popdef(optarg);
 209                         break;
 210                 case 'g':
 211                         mimic_gnu = 1;
 212                         break;
 213                 case 'd':
 214                         set_trace_flags(optarg);
 215                         break;
 216                 case 's':
 217                         synch_lines = 1;
 218                         break;
 219                 case 't':
 220                         mark_traced(optarg, 1);
 221                         break;
 222                 case 'o':
 223                         trace_file(optarg);
 224                         break;
 225                 case '?':
 226                         usage();
 227                 }
 228
 229         argc -= optind;
 230         argv += optind;
 231
 232         initkwds();
 233         if (mimic_gnu)
 234                 setup_builtin("format", FORMATTYPE);
 235
 236         active = stdout;                /* default active output     */
 237         bbase[0] = bufbase;
 238         if (!argc) {
 239                 sp = -1;                /* stack pointer initialized */
 240                 fp = 0;                 /* frame pointer initialized */
 241                 set_input(infile+0, stdin, "stdin");
 242                                         /* default input (naturally) */
 243                 macro();
 244         } else
 245                 for (; argc--; ++argv) {
 246                         p = *argv;
 247                         if (p[0] == '-' && p[1] == EOS)
 248                                 set_input(infile, stdin, "stdin");
 249                         else if (fopen_trypath(infile, p) == NULL)
 250                                 err(1, "%s", p);
 251                         sp = -1;
 252                         fp = 0;
 253                         macro();
 254                         release_input(infile);
 255                 }
 256
 257         if (wrapindex) {
 258                 int i;
 259
 260                 ilevel = 0;             /* in case m4wrap includes.. */
 261                 bufbase = bp = buf;     /* use the entire buffer   */
 262                 if (mimic_gnu) {
 263                         while (wrapindex != 0) {
 264                                 for (i = 0; i < wrapindex; i++)
 265                                         pbstr(m4wraps[i]);
 266                                 wrapindex =0;
 267                                 macro();
 268                         }
 269                 } else {
 270                         for (i = 0; i < wrapindex; i++) {
 271                                 pbstr(m4wraps[i]);
 272                                 macro();
 273                         }
 274                 }
 275         }
 276
 277         if (active != stdout)
 278                 active = stdout;        /* reset output just in case */
 279         for (n = 1; n < maxout; n++)    /* default wrap-up: undivert */
 280                 if (outfile[n] != NULL)
 281                         getdiv(n);
 282                                         /* remove bitbucket if used  */
 283         if (outfile[0] != NULL) {
 284                 (void) fclose(outfile[0]);
 285         }
 286
 287         return exit_code;
 288 }
 289
 290 /*
 291  * Look ahead for `token'.
 292  * (on input `t == token[0]')
 293  * Used for comment and quoting delimiters.
 294  * Returns 1 if `token' present; copied to output.
 295  *         0 if `token' not found; all characters pushed back
 296  */
 297 static int
 298 do_look_ahead(int t, const char *token)
 299 {
 300         int i;
 301
 302         assert((unsigned char)t == (unsigned char)token[0]);
 303
 304         for (i = 1; *++token; i++) {
 305                 t = gpbc();
 306                 if (t == EOF || (unsigned char)t != (unsigned char)*token) {
 307                         pushback(t);
 308                         while (--i)
 309                                 pushback(*--token);
 310                         return 0;
 311                 }
 312         }
 313         return 1;
 314 }
 315
 316 #define LOOK_AHEAD(t, token) (t != EOF &&               \
 317     (unsigned char)(t)==(unsigned char)(token)[0] &&    \
 318     do_look_ahead(t,token))
 319
 320 /*
 321  * macro - the work horse..
 322  */
 323 static void
 324 macro(void)
 325 {
 326         char token[MAXTOK+1];
 327         int t, l;
 328         ndptr p;
 329         int  nlpar;
 330
 331         cycle {
 332                 t = gpbc();
 333
 334                 if (LOOK_AHEAD(t,lquote)) {     /* strip quotes */
 335                         nlpar = 0;
 336                         record(quotes, nlpar++);
 337                         /*
 338                          * Opening quote: scan forward until matching
 339                          * closing quote has been found.
 340                          */
 341                         do {
 342
 343                                 l = gpbc();
 344                                 if (LOOK_AHEAD(l,rquote)) {
 345                                         if (--nlpar > 0)
 346                                                 outputstr(rquote);
 347                                 } else if (LOOK_AHEAD(l,lquote)) {
 348                                         record(quotes, nlpar++);
 349                                         outputstr(lquote);
 350                                 } else if (l == EOF) {
 351                                         if (nlpar == 1)
 352                                                 warnx("unclosed quote:");
 353                                         else
 354                                                 warnx("%d unclosed quotes:", nlpar);
 355                                         dump_stack(quotes, nlpar);
 356                                         exit(1);
 357                                 } else {
 358                                         if (nlpar > 0) {
 359                                                 if (sp < 0)
 360                                                         reallyputchar(l);
 361                                                 else
 362                                                         CHRSAVE(l);
 363                                         }
 364                                 }
 365                         }
 366                         while (nlpar != 0);
 367                 } else if (sp < 0 && LOOK_AHEAD(t, scommt)) {
 368                         reallyoutputstr(scommt);
 369
 370                         for(;;) {
 371                                 t = gpbc();
 372                                 if (LOOK_AHEAD(t, ecommt)) {
 373                                         reallyoutputstr(ecommt);
 374                                         break;
 375                                 }
 376                                 if (t == EOF)
 377                                         break;
 378                                 reallyputchar(t);
 379                         }
 380                 } else if (t == '_' || isalpha(t)) {
 381                         p = inspect(t, token);
 382                         if (p != NULL)
 383                                 pushback(l = gpbc());
 384                         if (p == NULL || (l != LPAREN &&
 385                             (macro_getdef(p)->type & NEEDARGS) != 0))
 386                                 outputstr(token);
 387                         else {
 388                 /*
 389                  * real thing.. First build a call frame:
 390                  */
 391                                 pushf(fp);      /* previous call frm */
 392                                 pushf(macro_getdef(p)->type); /* type of the call  */
 393                                 pushf(is_traced(p));
 394                                 pushf(0);       /* parenthesis level */
 395                                 fp = sp;        /* new frame pointer */
 396                 /*
 397                  * now push the string arguments:
 398                  */
 399                                 pushs1(macro_getdef(p)->defn);  /* defn string */
 400                                 pushs1((char *)macro_name(p));  /* macro name  */
 401                                 pushs(ep);                      /* start next..*/
 402
 403                                 if (l != LPAREN && PARLEV == 0) {
 404                                     /* no bracks  */
 405                                         chrsave(EOS);
 406
 407                                         if (sp == (int)STACKMAX)
 408                                                 errx(1, "internal stack overflow");
 409                                         eval((const char **) mstack+fp+1, 2,
 410                                             CALTYP, TRACESTATUS);
 411
 412                                         ep = PREVEP;    /* flush strspace */
 413                                         sp = PREVSP;    /* previous sp..  */
 414                                         fp = PREVFP;    /* rewind stack...*/
 415                                 }
 416                         }
 417                 } else if (t == EOF) {
 418                         if (sp > -1 && ilevel <= 0) {
 419                                 warnx( "unexpected end of input, unclosed parenthesis:");
 420                                 dump_stack(paren, PARLEV);
 421                                 exit(1);
 422                         }
 423                         if (ilevel <= 0)
 424                                 break;                  /* all done thanks.. */
 425                         release_input(infile+ilevel--);
 426                         emit_synchline();
 427                         bufbase = bbase[ilevel];
 428                         continue;
 429                 } else if (sp < 0) {            /* not in a macro at all */
 430                         reallyputchar(t);       /* output directly..     */
 431                 }
 432
 433                 else switch(t) {
 434
 435                 case LPAREN:
 436                         if (PARLEV > 0)
 437                                 chrsave(t);
 438                         while (isspace(l = gpbc())) /* skip blank, tab, nl.. */
 439                                 if (PARLEV > 0)
 440                                         chrsave(l);
 441                         pushback(l);
 442                         record(paren, PARLEV++);
 443                         break;
 444
 445                 case RPAREN:
 446                         if (--PARLEV > 0)
 447                                 chrsave(t);
 448                         else {                  /* end of argument list */
 449                                 chrsave(EOS);
 450
 451                                 if (sp == (int)STACKMAX)
 452                                         errx(1, "internal stack overflow");
 453
 454                                 eval((const char **) mstack+fp+1, sp-fp,
 455                                     CALTYP, TRACESTATUS);
 456
 457                                 ep = PREVEP;    /* flush strspace */
 458                                 sp = PREVSP;    /* previous sp..  */
 459                                 fp = PREVFP;    /* rewind stack...*/
 460                         }
 461                         break;
 462
 463                 case COMMA:
 464                         if (PARLEV == 1) {
 465                                 chrsave(EOS);           /* new argument   */
 466                                 while (isspace(l = gpbc()))
 467                                         ;
 468                                 pushback(l);
 469                                 pushs(ep);
 470                         } else
 471                                 chrsave(t);
 472                         break;
 473
 474                 default:
 475                         if (LOOK_AHEAD(t, scommt)) {
 476                                 char *cp;
 477                                 for (cp = scommt; *cp; cp++)
 478                                         chrsave(*cp);
 479                                 for(;;) {
 480                                         t = gpbc();
 481                                         if (LOOK_AHEAD(t, ecommt)) {
 482                                                 for (cp = ecommt; *cp; cp++)
 483                                                         chrsave(*cp);
 484                                                 break;
 485                                         }
 486                                         if (t == EOF)
 487                                             break;
 488                                         CHRSAVE(t);
 489                                 }
 490                         } else
 491                                 CHRSAVE(t);             /* stack the char */
 492                         break;
 493                 }
 494         }
 495 }
 496
 497 /*
 498  * output string directly, without pushing it for reparses.
 499  */
 500 void
 501 outputstr(const char *s)
 502 {
 503         if (sp < 0)
 504                 reallyoutputstr(s);
 505         else
 506                 while (*s)
 507                         CHRSAVE(*s++);
 508 }
 509
 510 void
 511 reallyoutputstr(const char *s)
 512 {
 513         if (synch_lines) {
 514                 while (*s) {
 515                         fputc(*s, active);
 516                         if (*s++ == '\n') {
 517                                 infile[ilevel].synch_lineno++;
 518                                 if (infile[ilevel].synch_lineno !=
 519                                     infile[ilevel].lineno)
 520                                         do_emit_synchline();
 521                         }
 522                 }
 523         } else
 524                 fputs(s, active);
 525 }
 526
 527 void
 528 reallyputchar(int c)
 529 {
 530         putc(c, active);
 531         if (synch_lines && c == '\n') {
 532                 infile[ilevel].synch_lineno++;
 533                 if (infile[ilevel].synch_lineno != infile[ilevel].lineno)
 534                         do_emit_synchline();
 535         }
 536 }
 537
 538 /*
 539  * build an input token..
 540  * consider only those starting with _ or A-Za-z.
 541  */
 542 static ndptr
 543 inspect(int c, char *tp)
 544 {
 545         char *name = tp;
 546         char *etp = tp+MAXTOK;
 547         ndptr p;
 548
 549         *tp++ = c;
 550
 551         while ((isalnum(c = gpbc()) || c == '_') && tp < etp)
 552                 *tp++ = c;
 553         if (c != EOF)
 554                 PUSHBACK(c);
 555         *tp = EOS;
 556         /* token is too long, it won't match anything, but it can still
 557          * be output. */
 558         if (tp == ep) {
 559                 outputstr(name);
 560                 while (isalnum(c = gpbc()) || c == '_') {
 561                         if (sp < 0)
 562                                 reallyputchar(c);
 563                         else
 564                                 CHRSAVE(c);
 565                 }
 566                 *name = EOS;
 567                 return NULL;
 568         }
 569
 570         p = ohash_find(&macros, ohash_qlookupi(&macros, name, (const char **)&tp));
 571         if (p == NULL)
 572                 return NULL;
 573         if (macro_getdef(p) == NULL)
 574                 return NULL;
 575         return p;
 576 }
 577
 578 /*
 579  * initkwds - initialise m4 keywords as fast as possible.
 580  * This very similar to install, but without certain overheads,
 581  * such as calling lookup. Malloc is not used for storing the
 582  * keyword strings, since we simply use the static pointers
 583  * within keywrds block.
 584  */
 585 static void
 586 initkwds(void)
 587 {
 588         unsigned int type;
 589         int i;
 590
 591         for (i = 0; i < (int)MAXKEYS; i++) {
 592                 type = keywrds[i].ktyp & TYPEMASK;
 593                 if ((keywrds[i].ktyp & NOARGS) == 0)
 594                         type |= NEEDARGS;
 595                 setup_builtin(keywrds[i].knam, type);
 596         }
 597 }
 598
 599 static void
 600 record(struct position *t, int lev)
 601 {
 602         if (lev < MAXRECORD) {
 603                 t[lev].name = CURRENT_NAME;
 604                 t[lev].line = CURRENT_LINE;
 605         }
 606 }
 607
 608 static void
 609 dump_stack(struct position *t, int lev)
 610 {
 611         int i;
 612
 613         for (i = 0; i < lev; i++) {
 614                 if (i == MAXRECORD) {
 615                         fprintf(stderr, "   ...\n");
 616                         break;
 617                 }
 618                 fprintf(stderr, "   %s at line %lu\n",
 619                         t[i].name, t[i].line);
 620         }
 621 }
 622
 623
 624 static void
 625 enlarge_stack(void)
 626 {
 627         STACKMAX += STACKMAX/2;
 628         mstack = xrealloc(mstack, sizeof(stae) * STACKMAX,
 629             "Evaluation stack overflow (%lu)",
 630             (unsigned long)STACKMAX);
 631         sstack = xrealloc(sstack, STACKMAX,
 632             "Evaluation stack overflow (%lu)",
 633             (unsigned long)STACKMAX);
 634 }