main.c

   1 /*      $Id: main.c,v 1.67 2010/05/15 05:50:19 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 read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
 272 {
 273         struct stat      st;
 274         char            *buf;
 275         size_t           sz, off;
 276         ssize_t          ssz;
 277
 278         if (-1 == fstat(curp->fd, &st)) {
 279                 perror(curp->file);
 280                 with_error = 1;
 281                 return(0);
 282         }
 283
 284         /*
 285          * If we're a regular file, try just reading in the whole entry
 286          * via mmap().  This is faster than reading it into blocks, and
 287          * since each file is only a few bytes to begin with, I'm not
 288          * concerned that this is going to tank any machines.
 289          */
 290
 291         if (S_ISREG(st.st_mode)) {
 292                 if (st.st_size >= (1U << 31)) {
 293                         fprintf(stderr, "%s: input too large\n",
 294                                         curp->file);
 295                         with_error = 1;
 296                         return(0);
 297                 }
 298                 *with_mmap = 1;
 299                 fb->sz = st.st_size;
 300                 fb->buf = mmap(NULL, fb->sz, PROT_READ,
 301                                 MAP_FILE, curp->fd, 0);
 302                 if (fb->buf != MAP_FAILED)
 303                         return(1);
 304         }
 305
 306         /*
 307          * If this isn't a regular file (like, say, stdin), then we must
 308          * go the old way and just read things in bit by bit.
 309          */
 310
 311         *with_mmap = 0;
 312         off = 0;
 313         fb->sz = 0;
 314         fb->buf = NULL;
 315         for (;;) {
 316                 if (off == fb->sz) {
 317                         if (fb->sz == (1U << 31)) {
 318                                 fprintf(stderr, "%s: input too large\n",
 319                                                 curp->file);
 320                                 break;
 321                         }
 322                         if (fb->sz == 0)
 323                                 sz = 65536;
 324                         else
 325                                 sz = 2 * fb->sz;
 326                         buf = realloc(fb->buf, sz);
 327                         if (NULL == buf) {
 328                                 perror(NULL);
 329                                 break;
 330                         }
 331                         fb->buf = buf;
 332                         fb->sz = sz;
 333                 }
 334                 ssz = read(curp->fd, fb->buf + off, fb->sz - off);
 335                 if (ssz == 0) {
 336                         fb->sz = off;
 337                         return(1);
 338                 }
 339                 if (ssz == -1) {
 340                         perror(curp->file);
 341                         break;
 342                 }
 343                 off += ssz;
 344         }
 345
 346         free(fb->buf);
 347         fb->buf = NULL;
 348         with_error = 1;
 349         return(0);
 350 }
 351
 352
 353 static void
 354 fdesc(struct curparse *curp)
 355 {
 356         size_t           sz;
 357         struct buf       ln, blk;
 358         int              j, i, pos, lnn, comment, with_mmap;
 359         struct man      *man;
 360         struct mdoc     *mdoc;
 361
 362         sz = BUFSIZ;
 363         man = NULL;
 364         mdoc = NULL;
 365         memset(&ln, 0, sizeof(struct buf));
 366
 367         /*
 368          * Two buffers: ln and buf.  buf is the input buffer optimised
 369          * here for each file's block size.  ln is a line buffer.  Both
 370          * growable, hence passed in by ptr-ptr.
 371          */
 372
 373         if (!read_whole_file(curp, &blk, &with_mmap))
 374                 return;
 375
 376         /* Fill buf with file blocksize. */
 377
 378         for (i = lnn = pos = comment = 0; i < (int)blk.sz; ++i) {
 379                 if (pos >= (int)ln.sz) {
 380                         ln.sz += 256; /* Step-size. */
 381                         ln.buf = realloc(ln.buf, ln.sz);
 382                         if (NULL == ln.buf) {
 383                                 perror(NULL);
 384                                 goto bailout;
 385                         }
 386                 }
 387
 388                 if ('\n' != blk.buf[i]) {
 389                         if (comment)
 390                                 continue;
 391                         ln.buf[pos++] = blk.buf[i];
 392
 393                         /* Handle in-line `\"' comments. */
 394
 395                         if (1 == pos || '\"' != ln.buf[pos - 1])
 396                                 continue;
 397
 398                         for (j = pos - 2; j >= 0; j--)
 399                                 if ('\\' != ln.buf[j])
 400                                         break;
 401
 402                         if ( ! ((pos - 2 - j) % 2))
 403                                 continue;
 404
 405                         comment = 1;
 406                         pos -= 2;
 407                         for (; pos > 0; --pos) {
 408                                 if (ln.buf[pos - 1] != ' ')
 409                                         break;
 410                                 if (pos > 2 && ln.buf[pos - 2] == '\\')
 411                                         break;
 412                         }
 413                         continue;
 414                 }
 415
 416                 /* Handle escaped `\\n' newlines. */
 417
 418                 if (pos > 0 && 0 == comment && '\\' == ln.buf[pos - 1]) {
 419                         for (j = pos - 1; j >= 0; j--)
 420                                 if ('\\' != ln.buf[j])
 421                                         break;
 422                         if ( ! ((pos - j) % 2)) {
 423                                 pos--;
 424                                 lnn++;
 425                                 continue;
 426                         }
 427                 }
 428
 429                 ln.buf[pos] = 0;
 430                 lnn++;
 431
 432                 /* If unset, assign parser in pset(). */
 433
 434                 if ( ! (man || mdoc) && ! pset(ln.buf, pos, curp, &man, &mdoc))
 435                         goto bailout;
 436
 437                 pos = comment = 0;
 438
 439                 /* Pass down into parsers. */
 440
 441                 if (man && ! man_parseln(man, lnn, ln.buf))
 442                         goto bailout;
 443                 if (mdoc && ! mdoc_parseln(mdoc, lnn, ln.buf))
 444                         goto bailout;
 445         }
 446
 447         /* NOTE a parser may not have been assigned, yet. */
 448
 449         if ( ! (man || mdoc)) {
 450                 fprintf(stderr, "%s: Not a manual\n", curp->file);
 451                 goto bailout;
 452         }
 453
 454         if (mdoc && ! mdoc_endparse(mdoc))
 455                 goto bailout;
 456         if (man && ! man_endparse(man))
 457                 goto bailout;
 458
 459         /* If unset, allocate output dev now (if applicable). */
 460
 461         if ( ! (curp->outman && curp->outmdoc)) {
 462                 switch (curp->outtype) {
 463                 case (OUTT_XHTML):
 464                         curp->outdata = xhtml_alloc(curp->outopts);
 465                         curp->outman = html_man;
 466                         curp->outmdoc = html_mdoc;
 467                         curp->outfree = html_free;
 468                         break;
 469                 case (OUTT_HTML):
 470                         curp->outdata = html_alloc(curp->outopts);
 471                         curp->outman = html_man;
 472                         curp->outmdoc = html_mdoc;
 473                         curp->outfree = html_free;
 474                         break;
 475                 case (OUTT_TREE):
 476                         curp->outman = tree_man;
 477                         curp->outmdoc = tree_mdoc;
 478                         break;
 479                 case (OUTT_LINT):
 480                         break;
 481                 default:
 482                         curp->outdata = ascii_alloc();
 483                         curp->outman = terminal_man;
 484                         curp->outmdoc = terminal_mdoc;
 485                         curp->outfree = terminal_free;
 486                         break;
 487                 }
 488         }
 489
 490         /* Execute the out device, if it exists. */
 491
 492         if (man && curp->outman)
 493                 (*curp->outman)(curp->outdata, man);
 494         if (mdoc && curp->outmdoc)
 495                 (*curp->outmdoc)(curp->outdata, mdoc);
 496
 497  cleanup:
 498         if (curp->mdoc) {
 499                 mdoc_free(curp->mdoc);
 500                 curp->mdoc = NULL;
 501         }
 502         if (curp->man) {
 503                 man_free(curp->man);
 504                 curp->man = NULL;
 505         }
 506         if (ln.buf)
 507                 free(ln.buf);
 508         if (with_mmap)
 509                 munmap(blk.buf, blk.sz);
 510         else
 511                 free(blk.buf);
 512         return;
 513
 514  bailout:
 515         with_error = 1;
 516         goto cleanup;
 517 }
 518
 519
 520 static int
 521 pset(const char *buf, int pos, struct curparse *curp,
 522                 struct man **man, struct mdoc **mdoc)
 523 {
 524         int              i;
 525
 526         /*
 527          * Try to intuit which kind of manual parser should be used.  If
 528          * passed in by command-line (-man, -mdoc), then use that
 529          * explicitly.  If passed as -mandoc, then try to guess from the
 530          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 531          * default to -man, which is more lenient.
 532          */
 533
 534         if (buf[0] == '.') {
 535                 for (i = 1; buf[i]; i++)
 536                         if (' ' != buf[i] && '\t' != buf[i])
 537                                 break;
 538                 if (0 == buf[i])
 539                         return(1);
 540         }
 541
 542         switch (curp->inttype) {
 543         case (INTT_MDOC):
 544                 if (NULL == curp->mdoc)
 545                         curp->mdoc = mdoc_init(curp);
 546                 if (NULL == (*mdoc = curp->mdoc))
 547                         return(0);
 548                 return(1);
 549         case (INTT_MAN):
 550                 if (NULL == curp->man)
 551                         curp->man = man_init(curp);
 552                 if (NULL == (*man = curp->man))
 553                         return(0);
 554                 return(1);
 555         default:
 556                 break;
 557         }
 558
 559         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 560                 if (NULL == curp->mdoc)
 561                         curp->mdoc = mdoc_init(curp);
 562                 if (NULL == (*mdoc = curp->mdoc))
 563                         return(0);
 564                 return(1);
 565         }
 566
 567         if (NULL == curp->man)
 568                 curp->man = man_init(curp);
 569         if (NULL == (*man = curp->man))
 570                 return(0);
 571         return(1);
 572 }
 573
 574
 575 static int
 576 moptions(enum intt *tflags, char *arg)
 577 {
 578
 579         if (0 == strcmp(arg, "doc"))
 580                 *tflags = INTT_MDOC;
 581         else if (0 == strcmp(arg, "andoc"))
 582                 *tflags = INTT_AUTO;
 583         else if (0 == strcmp(arg, "an"))
 584                 *tflags = INTT_MAN;
 585         else {
 586                 fprintf(stderr, "%s: Bad argument\n", arg);
 587                 return(0);
 588         }
 589
 590         return(1);
 591 }
 592
 593
 594 static int
 595 toptions(struct curparse *curp, char *arg)
 596 {
 597
 598         if (0 == strcmp(arg, "ascii"))
 599                 curp->outtype = OUTT_ASCII;
 600         else if (0 == strcmp(arg, "lint")) {
 601                 curp->outtype = OUTT_LINT;
 602                 curp->wflags |= WARN_WALL;
 603                 curp->fflags |= FL_STRICT;
 604         }
 605         else if (0 == strcmp(arg, "tree"))
 606                 curp->outtype = OUTT_TREE;
 607         else if (0 == strcmp(arg, "html"))
 608                 curp->outtype = OUTT_HTML;
 609         else if (0 == strcmp(arg, "xhtml"))
 610                 curp->outtype = OUTT_XHTML;
 611         else {
 612                 fprintf(stderr, "%s: Bad argument\n", arg);
 613                 return(0);
 614         }
 615
 616         return(1);
 617 }
 618
 619
 620 static int
 621 foptions(int *fflags, char *arg)
 622 {
 623         char            *v, *o;
 624         const char      *toks[8];
 625
 626         toks[0] = "ign-scope";
 627         toks[1] = "no-ign-escape";
 628         toks[2] = "no-ign-macro";
 629         toks[3] = "ign-errors";
 630         toks[4] = "strict";
 631         toks[5] = "ign-escape";
 632         toks[6] = NULL;
 633
 634         while (*arg) {
 635                 o = arg;
 636                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 637                 case (0):
 638                         *fflags |= FL_IGN_SCOPE;
 639                         break;
 640                 case (1):
 641                         *fflags |= FL_NIGN_ESCAPE;
 642                         break;
 643                 case (2):
 644                         *fflags |= FL_NIGN_MACRO;
 645                         break;
 646                 case (3):
 647                         *fflags |= FL_IGN_ERRORS;
 648                         break;
 649                 case (4):
 650                         *fflags |= FL_STRICT;
 651                         break;
 652                 case (5):
 653                         *fflags &= ~FL_NIGN_ESCAPE;
 654                         break;
 655                 default:
 656                         fprintf(stderr, "%s: Bad argument\n", o);
 657                         return(0);
 658                 }
 659         }
 660
 661         return(1);
 662 }
 663
 664
 665 static int
 666 woptions(int *wflags, char *arg)
 667 {
 668         char            *v, *o;
 669         const char      *toks[3];
 670
 671         toks[0] = "all";
 672         toks[1] = "error";
 673         toks[2] = NULL;
 674
 675         while (*arg) {
 676                 o = arg;
 677                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 678                 case (0):
 679                         *wflags |= WARN_WALL;
 680                         break;
 681                 case (1):
 682                         *wflags |= WARN_WERR;
 683                         break;
 684                 default:
 685                         fprintf(stderr, "%s: Bad argument\n", o);
 686                         return(0);
 687                 }
 688         }
 689
 690         return(1);
 691 }
 692
 693
 694 /* ARGSUSED */
 695 static int
 696 merr(void *arg, int line, int col, const char *msg)
 697 {
 698         struct curparse *curp;
 699
 700         curp = (struct curparse *)arg;
 701
 702         (void)fprintf(stderr, "%s:%d:%d: error: %s\n",
 703                         curp->file, line, col + 1, msg);
 704
 705         with_error = 1;
 706
 707         return(0);
 708 }
 709
 710
 711 static int
 712 mwarn(void *arg, int line, int col, const char *msg)
 713 {
 714         struct curparse *curp;
 715
 716         curp = (struct curparse *)arg;
 717
 718         if ( ! (curp->wflags & WARN_WALL))
 719                 return(1);
 720
 721         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 722                         curp->file, line, col + 1, msg);
 723
 724         with_warning = 1;
 725         if (curp->wflags & WARN_WERR) {
 726                 with_error = 1;
 727                 return(0);
 728         }
 729
 730         return(1);
 731 }
 732