usr.bin/mandoc/main.c

   1 /*      $Id: main.c,v 1.59 2010/01/29 14:39:38 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
   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 <fcntl.h>
  21 #include <stdio.h>
  22 #include <stdint.h>
  23 #include <stdlib.h>
  24 #include <string.h>
  25 #include <unistd.h>
  26
  27 #include "mdoc.h"
  28 #include "man.h"
  29 #include "main.h"
  30
  31 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  32
  33 /* FIXME: Intel's compiler?  LLVM?  pcc?  */
  34
  35 #if !defined(__GNUC__) || (__GNUC__ < 2)
  36 # if !defined(lint)
  37 #  define __attribute__(x)
  38 # endif
  39 #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */
  40
  41 typedef void            (*out_mdoc)(void *, const struct mdoc *);
  42 typedef void            (*out_man)(void *, const struct man *);
  43 typedef void            (*out_free)(void *);
  44
  45 struct  buf {
  46         char             *buf;
  47         size_t            sz;
  48 };
  49
  50 enum    intt {
  51         INTT_AUTO,
  52         INTT_MDOC,
  53         INTT_MAN
  54 };
  55
  56 enum    outt {
  57         OUTT_ASCII = 0,
  58         OUTT_TREE,
  59         OUTT_HTML,
  60         OUTT_XHTML,
  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        (1 << 0)       /* All-warnings mask. */
  69 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  70         int               fflags;
  71 #define IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
  72 #define NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
  73 #define NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
  74 #define NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */
  75 #define IGN_ERRORS       (1 << 4)       /* Ignore failed parse. */
  76         enum intt         inttype;      /* Input parsers... */
  77         struct man       *man;
  78         struct man       *lastman;
  79         struct mdoc      *mdoc;
  80         struct mdoc      *lastmdoc;
  81         enum outt         outtype;      /* Output devices... */
  82         out_mdoc          outmdoc;
  83         out_man           outman;
  84         out_free          outfree;
  85         void             *outdata;
  86         char              outopts[BUFSIZ];
  87 };
  88
  89 static  int               foptions(int *, char *);
  90 static  int               toptions(enum outt *, char *);
  91 static  int               moptions(enum intt *, char *);
  92 static  int               woptions(int *, char *);
  93 static  int               merr(void *, int, int, const char *);
  94 static  int               mwarn(void *, int, int, const char *);
  95 static  int               ffile(struct buf *, struct buf *,
  96                                 const char *, struct curparse *);
  97 static  int               fdesc(struct buf *, struct buf *,
  98                                 struct curparse *);
  99 static  int               pset(const char *, int, struct curparse *,
 100                                 struct man **, struct mdoc **);
 101 static  struct man       *man_init(struct curparse *);
 102 static  struct mdoc      *mdoc_init(struct curparse *);
 103 static  void              version(void) __attribute__((noreturn));
 104 static  void              usage(void) __attribute__((noreturn));
 105
 106 static  const char       *progname;
 107
 108
 109 int
 110 main(int argc, char *argv[])
 111 {
 112         int              c, rc;
 113         struct buf       ln, blk;
 114         struct curparse  curp;
 115
 116         progname = strrchr(argv[0], '/');
 117         if (progname == NULL)
 118                 progname = argv[0];
 119         else
 120                 ++progname;
 121
 122         memset(&curp, 0, sizeof(struct curparse));
 123
 124         curp.inttype = INTT_AUTO;
 125         curp.outtype = OUTT_ASCII;
 126
 127         /* LINTED */
 128         while (-1 != (c = getopt(argc, argv, "f:m:O:T:VW:")))
 129                 switch (c) {
 130                 case ('f'):
 131                         if ( ! foptions(&curp.fflags, optarg))
 132                                 return(EXIT_FAILURE);
 133                         break;
 134                 case ('m'):
 135                         if ( ! moptions(&curp.inttype, optarg))
 136                                 return(EXIT_FAILURE);
 137                         break;
 138                 case ('O'):
 139                         (void)strlcat(curp.outopts, optarg, BUFSIZ);
 140                         (void)strlcat(curp.outopts, ",", BUFSIZ);
 141                         break;
 142                 case ('T'):
 143                         if ( ! toptions(&curp.outtype, optarg))
 144                                 return(EXIT_FAILURE);
 145                         break;
 146                 case ('W'):
 147                         if ( ! woptions(&curp.wflags, optarg))
 148                                 return(EXIT_FAILURE);
 149                         break;
 150                 case ('V'):
 151                         version();
 152                         /* NOTREACHED */
 153                 default:
 154                         usage();
 155                         /* NOTREACHED */
 156                 }
 157
 158         argc -= optind;
 159         argv += optind;
 160
 161         memset(&ln, 0, sizeof(struct buf));
 162         memset(&blk, 0, sizeof(struct buf));
 163
 164         rc = 1;
 165
 166         if (NULL == *argv) {
 167                 curp.file = "<stdin>";
 168                 curp.fd = STDIN_FILENO;
 169
 170                 c = fdesc(&blk, &ln, &curp);
 171                 if ( ! (IGN_ERRORS & curp.fflags))
 172                         rc = 1 == c ? 1 : 0;
 173                 else
 174                         rc = -1 == c ? 0 : 1;
 175         }
 176
 177         while (rc && *argv) {
 178                 c = ffile(&blk, &ln, *argv, &curp);
 179                 if ( ! (IGN_ERRORS & curp.fflags))
 180                         rc = 1 == c ? 1 : 0;
 181                 else
 182                         rc = -1 == c ? 0 : 1;
 183
 184                 argv++;
 185                 if (*argv && rc) {
 186                         if (curp.lastman)
 187                                 man_reset(curp.lastman);
 188                         if (curp.lastmdoc)
 189                                 mdoc_reset(curp.lastmdoc);
 190                         curp.lastman = NULL;
 191                         curp.lastmdoc = NULL;
 192                 }
 193         }
 194
 195         if (blk.buf)
 196                 free(blk.buf);
 197         if (ln.buf)
 198                 free(ln.buf);
 199         if (curp.outfree)
 200                 (*curp.outfree)(curp.outdata);
 201         if (curp.mdoc)
 202                 mdoc_free(curp.mdoc);
 203         if (curp.man)
 204                 man_free(curp.man);
 205
 206         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 207 }
 208
 209
 210 static void
 211 version(void)
 212 {
 213
 214         (void)printf("%s %s\n", progname, VERSION);
 215         exit(EXIT_SUCCESS);
 216 }
 217
 218
 219 static void
 220 usage(void)
 221 {
 222
 223         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 224                         "[-mformat] [-Ooption] [-Toutput] "
 225                         "[-Werr...]\n", progname);
 226         exit(EXIT_FAILURE);
 227 }
 228
 229
 230 static struct man *
 231 man_init(struct curparse *curp)
 232 {
 233         int              pflags;
 234         struct man_cb    mancb;
 235
 236         mancb.man_err = merr;
 237         mancb.man_warn = mwarn;
 238
 239         /* Defaults from mandoc.1. */
 240
 241         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS;
 242
 243         if (curp->fflags & NO_IGN_MACRO)
 244                 pflags &= ~MAN_IGN_MACRO;
 245         if (curp->fflags & NO_IGN_CHARS)
 246                 pflags &= ~MAN_IGN_CHARS;
 247         if (curp->fflags & NO_IGN_ESCAPE)
 248                 pflags &= ~MAN_IGN_ESCAPE;
 249
 250         return(man_alloc(curp, pflags, &mancb));
 251 }
 252
 253
 254 static struct mdoc *
 255 mdoc_init(struct curparse *curp)
 256 {
 257         int              pflags;
 258         struct mdoc_cb   mdoccb;
 259
 260         mdoccb.mdoc_err = merr;
 261         mdoccb.mdoc_warn = mwarn;
 262
 263         /* Defaults from mandoc.1. */
 264
 265         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 266
 267         if (curp->fflags & IGN_SCOPE)
 268                 pflags |= MDOC_IGN_SCOPE;
 269         if (curp->fflags & NO_IGN_ESCAPE)
 270                 pflags &= ~MDOC_IGN_ESCAPE;
 271         if (curp->fflags & NO_IGN_MACRO)
 272                 pflags &= ~MDOC_IGN_MACRO;
 273         if (curp->fflags & NO_IGN_CHARS)
 274                 pflags &= ~MDOC_IGN_CHARS;
 275
 276         return(mdoc_alloc(curp, pflags, &mdoccb));
 277 }
 278
 279
 280 static int
 281 ffile(struct buf *blk, struct buf *ln,
 282                 const char *file, struct curparse *curp)
 283 {
 284         int              c;
 285
 286         curp->file = file;
 287         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 288                 perror(curp->file);
 289                 return(-1);
 290         }
 291
 292         c = fdesc(blk, ln, curp);
 293
 294         if (-1 == close(curp->fd))
 295                 perror(curp->file);
 296
 297         return(c);
 298 }
 299
 300
 301 static int
 302 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 303 {
 304         size_t           sz;
 305         ssize_t          ssz;
 306         struct stat      st;
 307         int              j, i, pos, lnn, comment;
 308         struct man      *man;
 309         struct mdoc     *mdoc;
 310
 311         sz = BUFSIZ;
 312         man = NULL;
 313         mdoc = NULL;
 314
 315         /*
 316          * Two buffers: ln and buf.  buf is the input buffer optimised
 317          * here for each file's block size.  ln is a line buffer.  Both
 318          * growable, hence passed in by ptr-ptr.
 319          */
 320
 321         if (-1 == fstat(curp->fd, &st))
 322                 perror(curp->file);
 323         else if ((size_t)st.st_blksize > sz)
 324                 sz = st.st_blksize;
 325
 326         if (sz > blk->sz) {
 327                 blk->buf = realloc(blk->buf, sz);
 328                 if (NULL == blk->buf) {
 329                         perror(NULL);
 330                         exit(EXIT_FAILURE);
 331                 }
 332                 blk->sz = sz;
 333         }
 334
 335         /* Fill buf with file blocksize. */
 336
 337         for (lnn = pos = comment = 0; ; ) {
 338                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 339                         perror(curp->file);
 340                         return(-1);
 341                 } else if (0 == ssz)
 342                         break;
 343
 344                 /* Parse the read block into partial or full lines. */
 345
 346                 for (i = 0; i < (int)ssz; i++) {
 347                         if (pos >= (int)ln->sz) {
 348                                 ln->sz += 256; /* Step-size. */
 349                                 ln->buf = realloc(ln->buf, ln->sz);
 350                                 if (NULL == ln->buf) {
 351                                         perror(NULL);
 352                                         return(EXIT_FAILURE);
 353                                 }
 354                         }
 355
 356                         if ('\n' != blk->buf[i]) {
 357                                 if (comment)
 358                                         continue;
 359                                 ln->buf[pos++] = blk->buf[i];
 360
 361                                 /* Handle in-line `\"' comments. */
 362
 363                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 364                                         continue;
 365
 366                                 for (j = pos - 2; j >= 0; j--)
 367                                         if ('\\' != ln->buf[j])
 368                                                 break;
 369
 370                                 if ( ! ((pos - 2 - j) % 2))
 371                                         continue;
 372
 373                                 comment = 1;
 374                                 pos -= 2;
 375                                 continue;
 376                         }
 377
 378                         /* Handle escaped `\\n' newlines. */
 379
 380                         if (pos > 0 && 0 == comment &&
 381                                         '\\' == ln->buf[pos - 1]) {
 382                                 for (j = pos - 1; j >= 0; j--)
 383                                         if ('\\' != ln->buf[j])
 384                                                 break;
 385                                 if ( ! ((pos - j) % 2)) {
 386                                         pos--;
 387                                         lnn++;
 388                                         continue;
 389                                 }
 390                         }
 391
 392                         ln->buf[pos] = 0;
 393                         lnn++;
 394
 395                         /* If unset, assign parser in pset(). */
 396
 397                         if ( ! (man || mdoc) && ! pset(ln->buf,
 398                                                 pos, curp, &man, &mdoc))
 399                                 return(-1);
 400
 401                         pos = comment = 0;
 402
 403                         /* Pass down into parsers. */
 404
 405                         if (man && ! man_parseln(man, lnn, ln->buf))
 406                                 return(0);
 407                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 408                                 return(0);
 409                 }
 410         }
 411
 412         /* NOTE a parser may not have been assigned, yet. */
 413
 414         if ( ! (man || mdoc)) {
 415                 fprintf(stderr, "%s: Not a manual\n", curp->file);
 416                 return(0);
 417         }
 418
 419         if (mdoc && ! mdoc_endparse(mdoc))
 420                 return(0);
 421         if (man && ! man_endparse(man))
 422                 return(0);
 423
 424         /* If unset, allocate output dev now (if applicable). */
 425
 426         if ( ! (curp->outman && curp->outmdoc)) {
 427                 switch (curp->outtype) {
 428                 case (OUTT_XHTML):
 429                         curp->outdata = xhtml_alloc(curp->outopts);
 430                         curp->outman = html_man;
 431                         curp->outmdoc = html_mdoc;
 432                         curp->outfree = html_free;
 433                         break;
 434                 case (OUTT_HTML):
 435                         curp->outdata = html_alloc(curp->outopts);
 436                         curp->outman = html_man;
 437                         curp->outmdoc = html_mdoc;
 438                         curp->outfree = html_free;
 439                         break;
 440                 case (OUTT_TREE):
 441                         curp->outman = tree_man;
 442                         curp->outmdoc = tree_mdoc;
 443                         break;
 444                 case (OUTT_LINT):
 445                         break;
 446                 default:
 447                         curp->outdata = ascii_alloc();
 448                         curp->outman = terminal_man;
 449                         curp->outmdoc = terminal_mdoc;
 450                         curp->outfree = terminal_free;
 451                         break;
 452                 }
 453         }
 454
 455         /* Execute the out device, if it exists. */
 456
 457         if (man && curp->outman)
 458                 (*curp->outman)(curp->outdata, man);
 459         if (mdoc && curp->outmdoc)
 460                 (*curp->outmdoc)(curp->outdata, mdoc);
 461
 462         return(1);
 463 }
 464
 465
 466 static int
 467 pset(const char *buf, int pos, struct curparse *curp,
 468                 struct man **man, struct mdoc **mdoc)
 469 {
 470         int              i;
 471
 472         /*
 473          * Try to intuit which kind of manual parser should be used.  If
 474          * passed in by command-line (-man, -mdoc), then use that
 475          * explicitly.  If passed as -mandoc, then try to guess from the
 476          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 477          * default to -man, which is more lenient.
 478          */
 479
 480         if (buf[0] == '.') {
 481                 for (i = 1; buf[i]; i++)
 482                         if (' ' != buf[i] && '\t' != buf[i])
 483                                 break;
 484                 if (0 == buf[i])
 485                         return(1);
 486         }
 487
 488         switch (curp->inttype) {
 489         case (INTT_MDOC):
 490                 if (NULL == curp->mdoc)
 491                         curp->mdoc = mdoc_init(curp);
 492                 if (NULL == (*mdoc = curp->mdoc))
 493                         return(0);
 494                 curp->lastmdoc = *mdoc;
 495                 return(1);
 496         case (INTT_MAN):
 497                 if (NULL == curp->man)
 498                         curp->man = man_init(curp);
 499                 if (NULL == (*man = curp->man))
 500                         return(0);
 501                 curp->lastman = *man;
 502                 return(1);
 503         default:
 504                 break;
 505         }
 506
 507         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 508                 if (NULL == curp->mdoc)
 509                         curp->mdoc = mdoc_init(curp);
 510                 if (NULL == (*mdoc = curp->mdoc))
 511                         return(0);
 512                 curp->lastmdoc = *mdoc;
 513                 return(1);
 514         }
 515
 516         if (NULL == curp->man)
 517                 curp->man = man_init(curp);
 518         if (NULL == (*man = curp->man))
 519                 return(0);
 520         curp->lastman = *man;
 521         return(1);
 522 }
 523
 524
 525 static int
 526 moptions(enum intt *tflags, char *arg)
 527 {
 528
 529         if (0 == strcmp(arg, "doc"))
 530                 *tflags = INTT_MDOC;
 531         else if (0 == strcmp(arg, "andoc"))
 532                 *tflags = INTT_AUTO;
 533         else if (0 == strcmp(arg, "an"))
 534                 *tflags = INTT_MAN;
 535         else {
 536                 fprintf(stderr, "%s: Bad argument\n", arg);
 537                 return(0);
 538         }
 539
 540         return(1);
 541 }
 542
 543
 544 static int
 545 toptions(enum outt *tflags, char *arg)
 546 {
 547
 548         if (0 == strcmp(arg, "ascii"))
 549                 *tflags = OUTT_ASCII;
 550         else if (0 == strcmp(arg, "lint"))
 551                 *tflags = OUTT_LINT;
 552         else if (0 == strcmp(arg, "tree"))
 553                 *tflags = OUTT_TREE;
 554         else if (0 == strcmp(arg, "html"))
 555                 *tflags = OUTT_HTML;
 556         else if (0 == strcmp(arg, "xhtml"))
 557                 *tflags = OUTT_XHTML;
 558         else {
 559                 fprintf(stderr, "%s: Bad argument\n", arg);
 560                 return(0);
 561         }
 562
 563         return(1);
 564 }
 565
 566
 567 static int
 568 foptions(int *fflags, char *arg)
 569 {
 570         char            *v, *o;
 571         const char      *toks[8];
 572
 573         toks[0] = "ign-scope";
 574         toks[1] = "no-ign-escape";
 575         toks[2] = "no-ign-macro";
 576         toks[3] = "no-ign-chars";
 577         toks[4] = "ign-errors";
 578         toks[5] = "strict";
 579         toks[6] = "ign-escape";
 580         toks[7] = NULL;
 581
 582         while (*arg) {
 583                 o = arg;
 584                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 585                 case (0):
 586                         *fflags |= IGN_SCOPE;
 587                         break;
 588                 case (1):
 589                         *fflags |= NO_IGN_ESCAPE;
 590                         break;
 591                 case (2):
 592                         *fflags |= NO_IGN_MACRO;
 593                         break;
 594                 case (3):
 595                         *fflags |= NO_IGN_CHARS;
 596                         break;
 597                 case (4):
 598                         *fflags |= IGN_ERRORS;
 599                         break;
 600                 case (5):
 601                         *fflags |= NO_IGN_ESCAPE |
 602                                    NO_IGN_MACRO | NO_IGN_CHARS;
 603                         break;
 604                 case (6):
 605                         *fflags &= ~NO_IGN_ESCAPE;
 606                         break;
 607                 default:
 608                         fprintf(stderr, "%s: Bad argument\n", o);
 609                         return(0);
 610                 }
 611         }
 612
 613         return(1);
 614 }
 615
 616
 617 static int
 618 woptions(int *wflags, char *arg)
 619 {
 620         char            *v, *o;
 621         const char      *toks[3];
 622
 623         toks[0] = "all";
 624         toks[1] = "error";
 625         toks[2] = NULL;
 626
 627         while (*arg) {
 628                 o = arg;
 629                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 630                 case (0):
 631                         *wflags |= WARN_WALL;
 632                         break;
 633                 case (1):
 634                         *wflags |= WARN_WERR;
 635                         break;
 636                 default:
 637                         fprintf(stderr, "%s: Bad argument\n", o);
 638                         return(0);
 639                 }
 640         }
 641
 642         return(1);
 643 }
 644
 645
 646 /* ARGSUSED */
 647 static int
 648 merr(void *arg, int line, int col, const char *msg)
 649 {
 650         struct curparse *curp;
 651
 652         curp = (struct curparse *)arg;
 653
 654         (void)fprintf(stderr, "%s:%d:%d: error: %s\n",
 655                         curp->file, line, col + 1, msg);
 656
 657         return(0);
 658 }
 659
 660
 661 static int
 662 mwarn(void *arg, int line, int col, const char *msg)
 663 {
 664         struct curparse *curp;
 665
 666         curp = (struct curparse *)arg;
 667
 668         if ( ! (curp->wflags & WARN_WALL))
 669                 return(1);
 670
 671         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 672                         curp->file, line, col + 1, msg);
 673
 674         if ( ! (curp->wflags & WARN_WERR))
 675                 return(1);
 676
 677         return(0);
 678 }