main.c

   1 /*      $Id: main.c,v 1.25 2009/04/12 19:45:26 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@openbsd.org>
   4  *
   5  * Permission to use, copy, modify, and distribute this software for any
   6  * purpose with or without fee is hereby granted, provided that the above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  16  */
  17 #include <sys/stat.h>
  18
  19 #include <assert.h>
  20 #include <err.h>
  21 #include <fcntl.h>
  22 #include <stdio.h>
  23 #include <stdlib.h>
  24 #include <string.h>
  25 #include <unistd.h>
  26
  27 #include "mdoc.h"
  28 #include "man.h"
  29
  30 /* Account for FreeBSD and Linux in our declarations. */
  31
  32 #ifdef __linux__
  33 extern  int               getsubopt(char **, char * const *, char **);
  34 # ifndef __dead
  35 #  define __dead __attribute__((__noreturn__))
  36 # endif
  37 #elif defined(__dead2)
  38 # ifndef __dead
  39 #  define __dead __dead2
  40 # endif
  41 #endif
  42
  43 typedef int             (*out_mdoc)(void *, const struct mdoc *);
  44 typedef int             (*out_man)(void *, const struct man *);
  45 typedef void            (*out_free)(void *);
  46
  47 struct  buf {
  48         char             *buf;
  49         size_t            sz;
  50 };
  51
  52 enum    intt {
  53         INTT_AUTO,
  54         INTT_MDOC,
  55         INTT_MAN
  56 };
  57
  58 enum    outt {
  59         OUTT_ASCII = 0,
  60         OUTT_TREE,
  61         OUTT_LINT
  62 };
  63
  64 struct  curparse {
  65         const char       *file;         /* Current parse. */
  66         int               fd;           /* Current parse. */
  67         int               wflags;
  68 #define WARN_WALL         0x03          /* All-warnings mask. */
  69 #define WARN_WCOMPAT     (1 << 0)       /* Compatibility warnings. */
  70 #define WARN_WSYNTAX     (1 << 1)       /* Syntax warnings. */
  71 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  72         int               fflags;
  73 #define IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
  74 #define NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
  75 #define NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
  76 #define NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */
  77         enum intt         inttype;      /* Input parsers. */
  78         struct man       *man;
  79         struct man       *lastman;
  80         struct mdoc      *mdoc;
  81         struct mdoc      *lastmdoc;
  82         enum outt         outtype;      /* Output devices. */
  83         out_mdoc          outmdoc;
  84         out_man           outman;
  85         out_free          outfree;
  86         void             *outdata;
  87 };
  88
  89 extern  void             *ascii_alloc(void);
  90 extern  int               tree_mdoc(void *, const struct mdoc *);
  91 extern  int               tree_man(void *, const struct man *);
  92 extern  int               terminal_mdoc(void *, const struct mdoc *);
  93 extern  int               terminal_man(void *, const struct man *);
  94 extern  void              terminal_free(void *);
  95
  96 static  int               foptions(int *, char *);
  97 static  int               toptions(enum outt *, char *);
  98 static  int               moptions(enum intt *, char *);
  99 static  int               woptions(int *, char *);
 100 static  int               merr(void *, int, int, const char *);
 101 static  int               manwarn(void *, int, int, const char *);
 102 static  int               mdocwarn(void *, int, int,
 103                                 enum mdoc_warn, const char *);
 104 static  int               fstdin(struct buf *, struct buf *,
 105                                 struct curparse *);
 106 static  int               ffile(struct buf *, struct buf *,
 107                                 const char *, struct curparse *);
 108 static  int               fdesc(struct buf *, struct buf *,
 109                                 struct curparse *);
 110 static  int               pset(const char *, int, struct curparse *,
 111                                 struct man **, struct mdoc **);
 112 static  struct man       *man_init(struct curparse *);
 113 static  struct mdoc      *mdoc_init(struct curparse *);
 114 __dead  static void       version(void);
 115 __dead  static void       usage(void);
 116
 117 extern  char             *__progname;
 118
 119
 120 int
 121 main(int argc, char *argv[])
 122 {
 123         int              c, rc;
 124         struct buf       ln, blk;
 125         struct curparse  curp;
 126
 127         bzero(&curp, sizeof(struct curparse));
 128
 129         curp.inttype = INTT_AUTO;
 130         curp.outtype = OUTT_ASCII;
 131
 132         /* LINTED */
 133         while (-1 != (c = getopt(argc, argv, "f:m:VW:T:")))
 134                 switch (c) {
 135                 case ('f'):
 136                         if ( ! foptions(&curp.fflags, optarg))
 137                                 return(0);
 138                         break;
 139                 case ('m'):
 140                         if ( ! moptions(&curp.inttype, optarg))
 141                                 return(0);
 142                         break;
 143                 case ('T'):
 144                         if ( ! toptions(&curp.outtype, optarg))
 145                                 return(0);
 146                         break;
 147                 case ('W'):
 148                         if ( ! woptions(&curp.wflags, optarg))
 149                                 return(0);
 150                         break;
 151                 case ('V'):
 152                         version();
 153                         /* NOTREACHED */
 154                 default:
 155                         usage();
 156                         /* NOTREACHED */
 157                 }
 158
 159         argc -= optind;
 160         argv += optind;
 161
 162         /* Configure buffers. */
 163
 164         bzero(&ln, sizeof(struct buf));
 165         bzero(&blk, sizeof(struct buf));
 166
 167         rc = 1;
 168
 169         if (NULL == *argv)
 170                 if ( ! fstdin(&blk, &ln, &curp))
 171                         rc = 0;
 172
 173         while (rc && *argv) {
 174                 if ( ! ffile(&blk, &ln, *argv, &curp))
 175                         rc = 0;
 176                 argv++;
 177                 if (*argv && rc) {
 178                         if (curp.lastman)
 179                                 if ( ! man_reset(curp.lastman))
 180                                         rc = 0;
 181                         if (curp.lastmdoc)
 182                                 if ( ! mdoc_reset(curp.lastmdoc))
 183                                         rc = 0;
 184                         curp.lastman = NULL;
 185                         curp.lastmdoc = NULL;
 186                 }
 187         }
 188
 189         if (blk.buf)
 190                 free(blk.buf);
 191         if (ln.buf)
 192                 free(ln.buf);
 193         if (curp.outfree)
 194                 (*curp.outfree)(curp.outdata);
 195         if (curp.mdoc)
 196                 mdoc_free(curp.mdoc);
 197         if (curp.man)
 198                 man_free(curp.man);
 199
 200         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 201 }
 202
 203
 204 __dead static void
 205 version(void)
 206 {
 207
 208         (void)printf("%s %s\n", __progname, VERSION);
 209         exit(EXIT_SUCCESS);
 210 }
 211
 212
 213 __dead static void
 214 usage(void)
 215 {
 216
 217         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 218                         "[-mformat] [-Toutput] [-Werr...]\n",
 219                         __progname);
 220         exit(EXIT_FAILURE);
 221 }
 222
 223
 224 static struct man *
 225 man_init(struct curparse *curp)
 226 {
 227         int              pflags;
 228         struct man      *man;
 229         struct man_cb    mancb;
 230
 231         mancb.man_err = merr;
 232         mancb.man_warn = manwarn;
 233
 234         pflags = MAN_IGN_MACRO; /* XXX */
 235
 236         if (curp->fflags & NO_IGN_MACRO)
 237                 pflags &= ~MAN_IGN_MACRO;
 238
 239         if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 240                 warnx("memory exhausted");
 241
 242         return(man);
 243 }
 244
 245
 246 static struct mdoc *
 247 mdoc_init(struct curparse *curp)
 248 {
 249         int              pflags;
 250         struct mdoc     *mdoc;
 251         struct mdoc_cb   mdoccb;
 252
 253         mdoccb.mdoc_msg = NULL;
 254         mdoccb.mdoc_err = merr;
 255         mdoccb.mdoc_warn = mdocwarn;
 256
 257         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 258
 259         if (curp->fflags & IGN_SCOPE)
 260                 pflags |= MDOC_IGN_SCOPE;
 261         if (curp->fflags & NO_IGN_ESCAPE)
 262                 pflags &= ~MDOC_IGN_ESCAPE;
 263         if (curp->fflags & NO_IGN_MACRO)
 264                 pflags &= ~MDOC_IGN_MACRO;
 265         if (curp->fflags & NO_IGN_CHARS)
 266                 pflags &= ~MDOC_IGN_CHARS;
 267
 268         if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb)))
 269                 warnx("memory exhausted");
 270
 271         return(mdoc);
 272 }
 273
 274
 275 static int
 276 fstdin(struct buf *blk, struct buf *ln, struct curparse *curp)
 277 {
 278
 279         curp->file = "<stdin>";
 280         curp->fd = STDIN_FILENO;
 281         return(fdesc(blk, ln, curp));
 282 }
 283
 284
 285 static int
 286 ffile(struct buf *blk, struct buf *ln,
 287                 const char *file, struct curparse *curp)
 288 {
 289         int              c;
 290
 291         curp->file = file;
 292         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 293                 warn("%s", curp->file);
 294                 return(0);
 295         }
 296
 297         c = fdesc(blk, ln, curp);
 298
 299         if (-1 == close(curp->fd))
 300                 warn("%s", curp->file);
 301
 302         return(c);
 303 }
 304
 305
 306 static int
 307 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 308 {
 309         size_t           sz;
 310         ssize_t          ssz;
 311         struct stat      st;
 312         int              j, i, pos, lnn;
 313         struct man      *man;
 314         struct mdoc     *mdoc;
 315
 316         sz = BUFSIZ;
 317         man = NULL;
 318         mdoc = NULL;
 319
 320         /*
 321          * Two buffers: ln and buf.  buf is the input buffer optimised
 322          * here for each file's block size.  ln is a line buffer.  Both
 323          * growable, hence passed in by ptr-ptr.
 324          */
 325
 326         if (-1 == fstat(curp->fd, &st))
 327                 warnx("%s", curp->file);
 328         else if ((size_t)st.st_blksize > sz)
 329                 sz = st.st_blksize;
 330
 331         if (sz > blk->sz) {
 332                 blk->buf = realloc(blk->buf, sz);
 333                 if (NULL == blk->buf) {
 334                         warn("realloc");
 335                         return(0);
 336                 }
 337                 blk->sz = sz;
 338         }
 339
 340         /* Fill buf with file blocksize. */
 341
 342         for (lnn = 0, pos = 0; ; ) {
 343                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 344                         warn("%s", curp->file);
 345                         return(0);
 346                 } else if (0 == ssz)
 347                         break;
 348
 349                 /* Parse the read block into partial or full lines. */
 350
 351                 for (i = 0; i < (int)ssz; i++) {
 352                         if (pos >= (int)ln->sz) {
 353                                 ln->sz += 256; /* Step-size. */
 354                                 ln->buf = realloc(ln->buf, ln->sz);
 355                                 if (NULL == ln->buf) {
 356                                         warn("realloc");
 357                                         return(0);
 358                                 }
 359                         }
 360
 361                         if ('\n' != blk->buf[i]) {
 362                                 ln->buf[pos++] = blk->buf[i];
 363                                 continue;
 364                         }
 365
 366                         /* Check for CPP-escaped newline.  */
 367
 368                         if (pos > 0 && '\\' == ln->buf[pos - 1]) {
 369                                 for (j = pos - 1; j >= 0; j--)
 370                                         if ('\\' != ln->buf[j])
 371                                                 break;
 372
 373                                 if ( ! ((pos - j) % 2)) {
 374                                         pos--;
 375                                         lnn++;
 376                                         continue;
 377                                 }
 378                         }
 379
 380                         ln->buf[pos] = 0;
 381                         lnn++;
 382
 383                         /*
 384                          * If no manual parser has been assigned, then
 385                          * try to assign one in pset(), which may do
 386                          * nothing at all.  After this, parse the manual
 387                          * line accordingly.
 388                          */
 389
 390                         if ( ! (man || mdoc) && ! pset(ln->buf,
 391                                                 pos, curp, &man, &mdoc))
 392                                 return(0);
 393
 394                         pos = 0;
 395
 396                         if (man && ! man_parseln(man, lnn, ln->buf))
 397                                 return(0);
 398                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 399                                 return(0);
 400                 }
 401         }
 402
 403         /* Note that a parser may not have been assigned, yet. */
 404
 405         if ( ! (man || mdoc)) {
 406                 warnx("%s: not a manual", curp->file);
 407                 return(0);
 408         }
 409
 410         if (mdoc && ! mdoc_endparse(mdoc))
 411                 return(0);
 412         if (man && ! man_endparse(man))
 413                 return(0);
 414
 415         /*
 416          * If an output device hasn't been allocated, see if we should
 417          * do so now.  Note that not all outtypes have functions, so
 418          * this switch statement may be superfluous, but it's
 419          * low-overhead enough not to matter very much.
 420          */
 421
 422         if ( ! (curp->outman && curp->outmdoc)) {
 423                 switch (curp->outtype) {
 424                 case (OUTT_TREE):
 425                         curp->outman = tree_man;
 426                         curp->outmdoc = tree_mdoc;
 427                         break;
 428                 case (OUTT_LINT):
 429                         break;
 430                 default:
 431                         curp->outdata = ascii_alloc();
 432                         curp->outman = terminal_man;
 433                         curp->outmdoc = terminal_mdoc;
 434                         curp->outfree = terminal_free;
 435                         break;
 436                 }
 437         }
 438
 439         /* Execute the out device, if it exists. */
 440
 441         if (man && curp->outman)
 442                 if ( ! (*curp->outman)(curp->outdata, man))
 443                         return(0);
 444         if (mdoc && curp->outmdoc)
 445                 if ( ! (*curp->outmdoc)(curp->outdata, mdoc))
 446                         return(0);
 447
 448         return(1);
 449 }
 450
 451
 452 static int
 453 pset(const char *buf, int pos, struct curparse *curp,
 454                 struct man **man, struct mdoc **mdoc)
 455 {
 456
 457         /*
 458          * Try to intuit which kind of manual parser should be used.  If
 459          * passed in by command-line (-man, -mdoc), then use that
 460          * explicitly.  If passed as -mandoc, then try to guess from the
 461          * line: either skip comments, use -mdoc when finding `.Dt', or
 462          * default to -man, which is more lenient.
 463          */
 464
 465         if (pos >= 3 && 0 == memcmp(buf, ".\\\"", 3))
 466                 return(1);
 467
 468         switch (curp->inttype) {
 469         case (INTT_MDOC):
 470                 if (NULL == curp->mdoc)
 471                         curp->mdoc = mdoc_init(curp);
 472                 if (NULL == (*mdoc = curp->mdoc))
 473                         return(0);
 474                 curp->lastmdoc = *mdoc;
 475                 return(1);
 476         case (INTT_MAN):
 477                 if (NULL == curp->man)
 478                         curp->man = man_init(curp);
 479                 if (NULL == (*man = curp->man))
 480                         return(0);
 481                 curp->lastman = *man;
 482                 return(1);
 483         default:
 484                 break;
 485         }
 486
 487         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 488                 if (NULL == curp->mdoc)
 489                         curp->mdoc = mdoc_init(curp);
 490                 if (NULL == (*mdoc = curp->mdoc))
 491                         return(0);
 492                 curp->lastmdoc = *mdoc;
 493                 return(1);
 494         }
 495
 496         if (NULL == curp->man)
 497                 curp->man = man_init(curp);
 498         if (NULL == (*man = curp->man))
 499                 return(0);
 500         curp->lastman = *man;
 501         return(1);
 502 }
 503
 504
 505 static int
 506 moptions(enum intt *tflags, char *arg)
 507 {
 508
 509         if (0 == strcmp(arg, "doc"))
 510                 *tflags = INTT_MDOC;
 511         else if (0 == strcmp(arg, "andoc"))
 512                 *tflags = INTT_AUTO;
 513         else if (0 == strcmp(arg, "an"))
 514                 *tflags = INTT_MAN;
 515         else {
 516                 warnx("bad argument: -m%s", arg);
 517                 return(0);
 518         }
 519
 520         return(1);
 521 }
 522
 523
 524 static int
 525 toptions(enum outt *tflags, char *arg)
 526 {
 527
 528         if (0 == strcmp(arg, "ascii"))
 529                 *tflags = OUTT_ASCII;
 530         else if (0 == strcmp(arg, "lint"))
 531                 *tflags = OUTT_LINT;
 532         else if (0 == strcmp(arg, "tree"))
 533                 *tflags = OUTT_TREE;
 534         else {
 535                 warnx("bad argument: -T%s", arg);
 536                 return(0);
 537         }
 538
 539         return(1);
 540 }
 541
 542
 543 /*
 544  * Parse out the options for [-fopt...] setting compiler options.  These
 545  * can be comma-delimited or called again.
 546  */
 547 static int
 548 foptions(int *fflags, char *arg)
 549 {
 550         char            *v;
 551         char            *toks[6];
 552
 553         toks[0] = "ign-scope";
 554         toks[1] = "no-ign-escape";
 555         toks[2] = "no-ign-macro";
 556         toks[3] = "no-ign-chars";
 557         toks[4] = "strict";
 558         toks[5] = NULL;
 559
 560         while (*arg)
 561                 switch (getsubopt(&arg, toks, &v)) {
 562                 case (0):
 563                         *fflags |= IGN_SCOPE;
 564                         break;
 565                 case (1):
 566                         *fflags |= NO_IGN_ESCAPE;
 567                         break;
 568                 case (2):
 569                         *fflags |= NO_IGN_MACRO;
 570                         break;
 571                 case (3):
 572                         *fflags |= NO_IGN_CHARS;
 573                         break;
 574                 case (4):
 575                         *fflags |= NO_IGN_ESCAPE |
 576                                    NO_IGN_MACRO | NO_IGN_CHARS;
 577                         break;
 578                 default:
 579                         warnx("bad argument: -f%s", arg);
 580                         return(0);
 581                 }
 582
 583         return(1);
 584 }
 585
 586
 587 /*
 588  * Parse out the options for [-Werr...], which sets warning modes.
 589  * These can be comma-delimited or called again.
 590  */
 591 static int
 592 woptions(int *wflags, char *arg)
 593 {
 594         char            *v;
 595         char            *toks[5];
 596
 597         toks[0] = "all";
 598         toks[1] = "compat";
 599         toks[2] = "syntax";
 600         toks[3] = "error";
 601         toks[4] = NULL;
 602
 603         while (*arg)
 604                 switch (getsubopt(&arg, toks, &v)) {
 605                 case (0):
 606                         *wflags |= WARN_WALL;
 607                         break;
 608                 case (1):
 609                         *wflags |= WARN_WCOMPAT;
 610                         break;
 611                 case (2):
 612                         *wflags |= WARN_WSYNTAX;
 613                         break;
 614                 case (3):
 615                         *wflags |= WARN_WERR;
 616                         break;
 617                 default:
 618                         warnx("bad argument: -W%s", arg);
 619                         return(0);
 620                 }
 621
 622         return(1);
 623 }
 624
 625
 626 /* ARGSUSED */
 627 static int
 628 merr(void *arg, int line, int col, const char *msg)
 629 {
 630         struct curparse *curp;
 631
 632         curp = (struct curparse *)arg;
 633
 634         warnx("%s:%d: error: %s (column %d)",
 635                         curp->file, line, msg, col);
 636         return(0);
 637 }
 638
 639
 640 static int
 641 mdocwarn(void *arg, int line, int col,
 642                 enum mdoc_warn type, const char *msg)
 643 {
 644         struct curparse *curp;
 645         char            *wtype;
 646
 647         curp = (struct curparse *)arg;
 648         wtype = NULL;
 649
 650         switch (type) {
 651         case (WARN_COMPAT):
 652                 wtype = "compat";
 653                 if (curp->wflags & WARN_WCOMPAT)
 654                         break;
 655                 return(1);
 656         case (WARN_SYNTAX):
 657                 wtype = "syntax";
 658                 if (curp->wflags & WARN_WSYNTAX)
 659                         break;
 660                 return(1);
 661         }
 662
 663         assert(wtype);
 664         warnx("%s:%d: %s warning: %s (column %d)",
 665                         curp->file, line, wtype, msg, col);
 666
 667         if ( ! (curp->wflags & WARN_WERR))
 668                 return(1);
 669
 670         warnx("%s: considering warnings as errors",
 671                         __progname);
 672         return(0);
 673 }
 674
 675
 676 static int
 677 manwarn(void *arg, int line, int col, const char *msg)
 678 {
 679         struct curparse *curp;
 680
 681         curp = (struct curparse *)arg;
 682
 683         if ( ! (curp->wflags & WARN_WSYNTAX))
 684                 return(1);
 685
 686         warnx("%s:%d: syntax warning: %s (column %d)",
 687                         curp->file, line, msg, col);
 688
 689         if ( ! (curp->wflags & WARN_WERR))
 690                 return(1);
 691
 692         warnx("%s: considering warnings as errors",
 693                         __progname);
 694         return(0);
 695 }