main.c

   1 /*      $Id: main.c,v 1.68 2010/05/15 09:46:31 joerg 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 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <sys/mman.h>
  22 #include <sys/stat.h>
  23
  24 #include <assert.h>
  25 #include <fcntl.h>
  26 #include <stdio.h>
  27 #include <stdint.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30 #include <unistd.h>
  31
  32 #include "mdoc.h"
  33 #include "man.h"
  34 #include "main.h"
  35
  36 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  37
  38 /* FIXME: Intel's compiler?  LLVM?  pcc?  */
  39
  40 #if !defined(__GNUC__) || (__GNUC__ < 2)
  41 # if !defined(lint)
  42 #  define __attribute__(x)
  43 # endif
  44 #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */
  45
  46 typedef void            (*out_mdoc)(void *, const struct mdoc *);
  47 typedef void            (*out_man)(void *, const struct man *);
  48 typedef void            (*out_free)(void *);
  49
  50 struct  buf {
  51         char             *buf;
  52         size_t            sz;
  53 };
  54
  55 enum    intt {
  56         INTT_AUTO,
  57         INTT_MDOC,
  58         INTT_MAN
  59 };
  60
  61 enum    outt {
  62         OUTT_ASCII = 0,
  63         OUTT_TREE,
  64         OUTT_HTML,
  65         OUTT_XHTML,
  66         OUTT_LINT
  67 };
  68
  69 struct  curparse {
  70         const char       *file;         /* Current parse. */
  71         int               fd;           /* Current parse. */
  72         int               wflags;
  73 #define WARN_WALL        (1 << 0)       /* All-warnings mask. */
  74 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  75         int               fflags;
  76 #define FL_IGN_SCOPE     (1 << 0)       /* Ignore scope errors. */
  77 #define FL_NIGN_ESCAPE   (1 << 1)       /* Don't ignore bad escapes. */
  78 #define FL_NIGN_MACRO    (1 << 2)       /* Don't ignore bad macros. */
  79 #define FL_IGN_ERRORS    (1 << 4)       /* Ignore failed parse. */
  80 #define FL_STRICT         FL_NIGN_ESCAPE | \
  81                           FL_NIGN_MACRO
  82         enum intt         inttype;      /* Input parsers... */
  83         struct man       *man;
  84         struct mdoc      *mdoc;
  85         enum outt         outtype;      /* Output devices... */
  86         out_mdoc          outmdoc;
  87         out_man           outman;
  88         out_free          outfree;
  89         void             *outdata;
  90         char              outopts[BUFSIZ];
  91 };
  92
  93 static  void              fdesc(struct curparse *);
  94 static  void              ffile(const char *, struct curparse *);
  95 static  int               foptions(int *, char *);
  96 static  struct man       *man_init(struct curparse *);
  97 static  struct mdoc      *mdoc_init(struct curparse *);
  98 static  int               merr(void *, int, int, const char *);
  99 static  int               moptions(enum intt *, char *);
 100 static  int               mwarn(void *, int, int, const char *);
 101 static  int               pset(const char *, int, struct curparse *,
 102                                 struct man **, struct mdoc **);
 103 static  int               toptions(struct curparse *, char *);
 104 static  void              usage(void) __attribute__((noreturn));
 105 static  void              version(void) __attribute__((noreturn));
 106 static  int               woptions(int *, char *);
 107
 108 static  const char       *progname;
 109 static  int               with_error;
 110 static  int               with_warning;
 111
 112 int
 113 main(int argc, char *argv[])
 114 {
 115         int              c;
 116         struct curparse  curp;
 117
 118         progname = strrchr(argv[0], '/');
 119         if (progname == NULL)
 120                 progname = argv[0];
 121         else
 122                 ++progname;
 123
 124         memset(&curp, 0, sizeof(struct curparse));
 125
 126         curp.inttype = INTT_AUTO;
 127         curp.outtype = OUTT_ASCII;
 128
 129         /* LINTED */
 130         while (-1 != (c = getopt(argc, argv, "f:m:O:T:VW:")))
 131                 switch (c) {
 132                 case ('f'):
 133                         if ( ! foptions(&curp.fflags, optarg))
 134                                 return(EXIT_FAILURE);
 135                         break;
 136                 case ('m'):
 137                         if ( ! moptions(&curp.inttype, optarg))
 138                                 return(EXIT_FAILURE);
 139                         break;
 140                 case ('O'):
 141                         (void)strlcat(curp.outopts, optarg, BUFSIZ);
 142                         (void)strlcat(curp.outopts, ",", BUFSIZ);
 143                         break;
 144                 case ('T'):
 145                         if ( ! toptions(&curp, optarg))
 146                                 return(EXIT_FAILURE);
 147                         break;
 148                 case ('W'):
 149                         if ( ! woptions(&curp.wflags, optarg))
 150                                 return(EXIT_FAILURE);
 151                         break;
 152                 case ('V'):
 153                         version();
 154                         /* NOTREACHED */
 155                 default:
 156                         usage();
 157                         /* NOTREACHED */
 158                 }
 159
 160         argc -= optind;
 161         argv += optind;
 162
 163         if (NULL == *argv) {
 164                 curp.file = "<stdin>";
 165                 curp.fd = STDIN_FILENO;
 166
 167                 fdesc(&curp);
 168         }
 169
 170         while (*argv) {
 171                 ffile(*argv, &curp);
 172
 173                 if (with_error && !(curp.fflags & FL_IGN_ERRORS))
 174                         break;
 175                 ++argv;
 176         }
 177
 178         if (curp.outfree)
 179                 (*curp.outfree)(curp.outdata);
 180
 181         return((with_warning || with_error) ?
 182                         EXIT_FAILURE :  EXIT_SUCCESS);
 183 }
 184
 185
 186 static void
 187 version(void)
 188 {
 189
 190         (void)printf("%s %s\n", progname, VERSION);
 191         exit(EXIT_SUCCESS);
 192 }
 193
 194
 195 static void
 196 usage(void)
 197 {
 198
 199         (void)fprintf(stderr, "usage: %s [-V] [-foption] "
 200                         "[-mformat] [-Ooption] [-Toutput] "
 201                         "[-Werr] [file...]\n", progname);
 202         exit(EXIT_FAILURE);
 203 }
 204
 205
 206 static struct man *
 207 man_init(struct curparse *curp)
 208 {
 209         int              pflags;
 210         struct man_cb    mancb;
 211
 212         mancb.man_err = merr;
 213         mancb.man_warn = mwarn;
 214
 215         /* Defaults from mandoc.1. */
 216
 217         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE;
 218
 219         if (curp->fflags & FL_NIGN_MACRO)
 220                 pflags &= ~MAN_IGN_MACRO;
 221         if (curp->fflags & FL_NIGN_ESCAPE)
 222                 pflags &= ~MAN_IGN_ESCAPE;
 223
 224         return(man_alloc(curp, pflags, &mancb));
 225 }
 226
 227
 228 static struct mdoc *
 229 mdoc_init(struct curparse *curp)
 230 {
 231         int              pflags;
 232         struct mdoc_cb   mdoccb;
 233
 234         mdoccb.mdoc_err = merr;
 235         mdoccb.mdoc_warn = mwarn;
 236
 237         /* Defaults from mandoc.1. */
 238
 239         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE;
 240
 241         if (curp->fflags & FL_IGN_SCOPE)
 242                 pflags |= MDOC_IGN_SCOPE;
 243         if (curp->fflags & FL_NIGN_ESCAPE)
 244                 pflags &= ~MDOC_IGN_ESCAPE;
 245         if (curp->fflags & FL_NIGN_MACRO)
 246                 pflags &= ~MDOC_IGN_MACRO;
 247
 248         return(mdoc_alloc(curp, pflags, &mdoccb));
 249 }
 250
 251
 252 static void
 253 ffile(const char *file, struct curparse *curp)
 254 {
 255
 256         curp->file = file;
 257         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 258                 perror(curp->file);
 259                 with_error = 1;
 260                 return;
 261         }
 262
 263         fdesc(curp);
 264
 265         if (-1 == close(curp->fd))
 266                 perror(curp->file);
 267 }
 268
 269
 270 static int
 271 resize_buf(struct buf *buf, size_t initial)
 272 {
 273         void *tmp;
 274         size_t sz;
 275
 276         if (buf->sz == 0)
 277                 sz = initial;
 278         else
 279                 sz = 2 * buf->sz;
 280         tmp = realloc(buf->buf, sz);
 281         if (NULL == tmp) {
 282                 perror(NULL);
 283                 return(0);
 284         }
 285         buf->buf = tmp;
 286         buf->sz = sz;
 287         return(1);
 288 }
 289
 290
 291 static int
 292 read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
 293 {
 294         struct stat      st;
 295         size_t           off;
 296         ssize_t          ssz;
 297
 298         if (-1 == fstat(curp->fd, &st)) {
 299                 perror(curp->file);
 300                 with_error = 1;
 301                 return(0);
 302         }
 303
 304         /*
 305          * If we're a regular file, try just reading in the whole entry
 306          * via mmap().  This is faster than reading it into blocks, and
 307          * since each file is only a few bytes to begin with, I'm not
 308          * concerned that this is going to tank any machines.
 309          */
 310
 311         if (S_ISREG(st.st_mode)) {
 312                 if (st.st_size >= (1U << 31)) {
 313                         fprintf(stderr, "%s: input too large\n",
 314                                         curp->file);
 315                         with_error = 1;
 316                         return(0);
 317                 }
 318                 *with_mmap = 1;
 319                 fb->sz = st.st_size;
 320                 fb->buf = mmap(NULL, fb->sz, PROT_READ,
 321                                 MAP_FILE, curp->fd, 0);
 322                 if (fb->buf != MAP_FAILED)
 323                         return(1);
 324         }
 325
 326         /*
 327          * If this isn't a regular file (like, say, stdin), then we must
 328          * go the old way and just read things in bit by bit.
 329          */
 330
 331         *with_mmap = 0;
 332         off = 0;
 333         fb->sz = 0;
 334         fb->buf = NULL;
 335         for (;;) {
 336                 if (off == fb->sz) {
 337                         if (fb->sz == (1U << 31)) {
 338                                 fprintf(stderr, "%s: input too large\n",
 339                                                 curp->file);
 340                                 break;
 341                         }
 342                         if (! resize_buf(fb, 65536))
 343                                 break;
 344                 }
 345                 ssz = read(curp->fd, fb->buf + off, fb->sz - off);
 346                 if (ssz == 0) {
 347                         fb->sz = off;
 348                         return(1);
 349                 }
 350                 if (ssz == -1) {
 351                         perror(curp->file);
 352                         break;
 353                 }
 354                 off += ssz;
 355         }
 356
 357         free(fb->buf);
 358         fb->buf = NULL;
 359         with_error = 1;
 360         return(0);
 361 }
 362
 363
 364 static void
 365 fdesc(struct curparse *curp)
 366 {
 367         struct buf       ln, blk;
 368         int              j, i, pos, lnn, comment, with_mmap;
 369         struct man      *man;
 370         struct mdoc     *mdoc;
 371
 372         man = NULL;
 373         mdoc = NULL;
 374         memset(&ln, 0, sizeof(struct buf));
 375
 376         /*
 377          * Two buffers: ln and buf.  buf is the input buffer optimised
 378          * here for each file's block size.  ln is a line buffer.  Both
 379          * growable, hence passed in by ptr-ptr.
 380          */
 381
 382         if (!read_whole_file(curp, &blk, &with_mmap))
 383                 return;
 384
 385         /* Fill buf with file blocksize. */
 386
 387         for (i = lnn = pos = comment = 0; i < (int)blk.sz; ++i) {
 388                 if (pos >= (int)ln.sz) {
 389                         if (! resize_buf(&ln, 256))
 390                                 goto bailout;
 391                 }
 392
 393                 if ('\n' != blk.buf[i]) {
 394                         if (comment)
 395                                 continue;
 396                         ln.buf[pos++] = blk.buf[i];
 397
 398                         /* Handle in-line `\"' comments. */
 399
 400                         if (1 == pos || '\"' != ln.buf[pos - 1])
 401                                 continue;
 402
 403                         for (j = pos - 2; j >= 0; j--)
 404                                 if ('\\' != ln.buf[j])
 405                                         break;
 406
 407                         if ( ! ((pos - 2 - j) % 2))
 408                                 continue;
 409
 410                         comment = 1;
 411                         pos -= 2;
 412                         for (; pos > 0; --pos) {
 413                                 if (ln.buf[pos - 1] != ' ')
 414                                         break;
 415                                 if (pos > 2 && ln.buf[pos - 2] == '\\')
 416                                         break;
 417                         }
 418                         continue;
 419                 }
 420
 421                 /* Handle escaped `\\n' newlines. */
 422
 423                 if (pos > 0 && 0 == comment && '\\' == ln.buf[pos - 1]) {
 424                         for (j = pos - 1; j >= 0; j--)
 425                                 if ('\\' != ln.buf[j])
 426                                         break;
 427                         if ( ! ((pos - j) % 2)) {
 428                                 pos--;
 429                                 lnn++;
 430                                 continue;
 431                         }
 432                 }
 433
 434                 ln.buf[pos] = 0;
 435                 lnn++;
 436
 437                 /* If unset, assign parser in pset(). */
 438
 439                 if ( ! (man || mdoc) && ! pset(ln.buf, pos, curp, &man, &mdoc))
 440                         goto bailout;
 441
 442                 pos = comment = 0;
 443
 444                 /* Pass down into parsers. */
 445
 446                 if (man && ! man_parseln(man, lnn, ln.buf))
 447                         goto bailout;
 448                 if (mdoc && ! mdoc_parseln(mdoc, lnn, ln.buf))
 449                         goto bailout;
 450         }
 451
 452         /* NOTE a parser may not have been assigned, yet. */
 453
 454         if ( ! (man || mdoc)) {
 455                 fprintf(stderr, "%s: Not a manual\n", curp->file);
 456                 goto bailout;
 457         }
 458
 459         if (mdoc && ! mdoc_endparse(mdoc))
 460                 goto bailout;
 461         if (man && ! man_endparse(man))
 462                 goto bailout;
 463
 464         /* If unset, allocate output dev now (if applicable). */
 465
 466         if ( ! (curp->outman && curp->outmdoc)) {
 467                 switch (curp->outtype) {
 468                 case (OUTT_XHTML):
 469                         curp->outdata = xhtml_alloc(curp->outopts);
 470                         curp->outman = html_man;
 471                         curp->outmdoc = html_mdoc;
 472                         curp->outfree = html_free;
 473                         break;
 474                 case (OUTT_HTML):
 475                         curp->outdata = html_alloc(curp->outopts);
 476                         curp->outman = html_man;
 477                         curp->outmdoc = html_mdoc;
 478                         curp->outfree = html_free;
 479                         break;
 480                 case (OUTT_TREE):
 481                         curp->outman = tree_man;
 482                         curp->outmdoc = tree_mdoc;
 483                         break;
 484                 case (OUTT_LINT):
 485                         break;
 486                 default:
 487                         curp->outdata = ascii_alloc();
 488                         curp->outman = terminal_man;
 489                         curp->outmdoc = terminal_mdoc;
 490                         curp->outfree = terminal_free;
 491                         break;
 492                 }
 493         }
 494
 495         /* Execute the out device, if it exists. */
 496
 497         if (man && curp->outman)
 498                 (*curp->outman)(curp->outdata, man);
 499         if (mdoc && curp->outmdoc)
 500                 (*curp->outmdoc)(curp->outdata, mdoc);
 501
 502  cleanup:
 503         if (curp->mdoc) {
 504                 mdoc_free(curp->mdoc);
 505                 curp->mdoc = NULL;
 506         }
 507         if (curp->man) {
 508                 man_free(curp->man);
 509                 curp->man = NULL;
 510         }
 511         if (ln.buf)
 512                 free(ln.buf);
 513         if (with_mmap)
 514                 munmap(blk.buf, blk.sz);
 515         else
 516                 free(blk.buf);
 517         return;
 518
 519  bailout:
 520         with_error = 1;
 521         goto cleanup;
 522 }
 523
 524
 525 static int
 526 pset(const char *buf, int pos, struct curparse *curp,
 527                 struct man **man, struct mdoc **mdoc)
 528 {
 529         int              i;
 530
 531         /*
 532          * Try to intuit which kind of manual parser should be used.  If
 533          * passed in by command-line (-man, -mdoc), then use that
 534          * explicitly.  If passed as -mandoc, then try to guess from the
 535          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 536          * default to -man, which is more lenient.
 537          */
 538
 539         if (buf[0] == '.') {
 540                 for (i = 1; buf[i]; i++)
 541                         if (' ' != buf[i] && '\t' != buf[i])
 542                                 break;
 543                 if (0 == buf[i])
 544                         return(1);
 545         }
 546
 547         switch (curp->inttype) {
 548         case (INTT_MDOC):
 549                 if (NULL == curp->mdoc)
 550                         curp->mdoc = mdoc_init(curp);
 551                 if (NULL == (*mdoc = curp->mdoc))
 552                         return(0);
 553                 return(1);
 554         case (INTT_MAN):
 555                 if (NULL == curp->man)
 556                         curp->man = man_init(curp);
 557                 if (NULL == (*man = curp->man))
 558                         return(0);
 559                 return(1);
 560         default:
 561                 break;
 562         }
 563
 564         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 565                 if (NULL == curp->mdoc)
 566                         curp->mdoc = mdoc_init(curp);
 567                 if (NULL == (*mdoc = curp->mdoc))
 568                         return(0);
 569                 return(1);
 570         }
 571
 572         if (NULL == curp->man)
 573                 curp->man = man_init(curp);
 574         if (NULL == (*man = curp->man))
 575                 return(0);
 576         return(1);
 577 }
 578
 579
 580 static int
 581 moptions(enum intt *tflags, char *arg)
 582 {
 583
 584         if (0 == strcmp(arg, "doc"))
 585                 *tflags = INTT_MDOC;
 586         else if (0 == strcmp(arg, "andoc"))
 587                 *tflags = INTT_AUTO;
 588         else if (0 == strcmp(arg, "an"))
 589                 *tflags = INTT_MAN;
 590         else {
 591                 fprintf(stderr, "%s: Bad argument\n", arg);
 592                 return(0);
 593         }
 594
 595         return(1);
 596 }
 597
 598
 599 static int
 600 toptions(struct curparse *curp, char *arg)
 601 {
 602
 603         if (0 == strcmp(arg, "ascii"))
 604                 curp->outtype = OUTT_ASCII;
 605         else if (0 == strcmp(arg, "lint")) {
 606                 curp->outtype = OUTT_LINT;
 607                 curp->wflags |= WARN_WALL;
 608                 curp->fflags |= FL_STRICT;
 609         }
 610         else if (0 == strcmp(arg, "tree"))
 611                 curp->outtype = OUTT_TREE;
 612         else if (0 == strcmp(arg, "html"))
 613                 curp->outtype = OUTT_HTML;
 614         else if (0 == strcmp(arg, "xhtml"))
 615                 curp->outtype = OUTT_XHTML;
 616         else {
 617                 fprintf(stderr, "%s: Bad argument\n", arg);
 618                 return(0);
 619         }
 620
 621         return(1);
 622 }
 623
 624
 625 static int
 626 foptions(int *fflags, char *arg)
 627 {
 628         char            *v, *o;
 629         const char      *toks[8];
 630
 631         toks[0] = "ign-scope";
 632         toks[1] = "no-ign-escape";
 633         toks[2] = "no-ign-macro";
 634         toks[3] = "ign-errors";
 635         toks[4] = "strict";
 636         toks[5] = "ign-escape";
 637         toks[6] = NULL;
 638
 639         while (*arg) {
 640                 o = arg;
 641                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 642                 case (0):
 643                         *fflags |= FL_IGN_SCOPE;
 644                         break;
 645                 case (1):
 646                         *fflags |= FL_NIGN_ESCAPE;
 647                         break;
 648                 case (2):
 649                         *fflags |= FL_NIGN_MACRO;
 650                         break;
 651                 case (3):
 652                         *fflags |= FL_IGN_ERRORS;
 653                         break;
 654                 case (4):
 655                         *fflags |= FL_STRICT;
 656                         break;
 657                 case (5):
 658                         *fflags &= ~FL_NIGN_ESCAPE;
 659                         break;
 660                 default:
 661                         fprintf(stderr, "%s: Bad argument\n", o);
 662                         return(0);
 663                 }
 664         }
 665
 666         return(1);
 667 }
 668
 669
 670 static int
 671 woptions(int *wflags, char *arg)
 672 {
 673         char            *v, *o;
 674         const char      *toks[3];
 675
 676         toks[0] = "all";
 677         toks[1] = "error";
 678         toks[2] = NULL;
 679
 680         while (*arg) {
 681                 o = arg;
 682                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 683                 case (0):
 684                         *wflags |= WARN_WALL;
 685                         break;
 686                 case (1):
 687                         *wflags |= WARN_WERR;
 688                         break;
 689                 default:
 690                         fprintf(stderr, "%s: Bad argument\n", o);
 691                         return(0);
 692                 }
 693         }
 694
 695         return(1);
 696 }
 697
 698
 699 /* ARGSUSED */
 700 static int
 701 merr(void *arg, int line, int col, const char *msg)
 702 {
 703         struct curparse *curp;
 704
 705         curp = (struct curparse *)arg;
 706
 707         (void)fprintf(stderr, "%s:%d:%d: error: %s\n",
 708                         curp->file, line, col + 1, msg);
 709
 710         with_error = 1;
 711
 712         return(0);
 713 }
 714
 715
 716 static int
 717 mwarn(void *arg, int line, int col, const char *msg)
 718 {
 719         struct curparse *curp;
 720
 721         curp = (struct curparse *)arg;
 722
 723         if ( ! (curp->wflags & WARN_WALL))
 724                 return(1);
 725
 726         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 727                         curp->file, line, col + 1, msg);
 728
 729         with_warning = 1;
 730         if (curp->wflags & WARN_WERR) {
 731                 with_error = 1;
 732                 return(0);
 733         }
 734
 735         return(1);
 736 }
 737