main.c

   1 /*      $Id: main.c,v 1.61 2010/04/12 19:27:22 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 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <sys/stat.h>
  22
  23 #include <assert.h>
  24 #include <fcntl.h>
  25 #include <stdio.h>
  26 #include <stdint.h>
  27 #include <stdlib.h>
  28 #include <string.h>
  29 #include <unistd.h>
  30
  31 #include "mdoc.h"
  32 #include "man.h"
  33 #include "main.h"
  34
  35 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  36
  37 /* FIXME: Intel's compiler?  LLVM?  pcc?  */
  38
  39 #if !defined(__GNUC__) || (__GNUC__ < 2)
  40 # if !defined(lint)
  41 #  define __attribute__(x)
  42 # endif
  43 #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */
  44
  45 typedef void            (*out_mdoc)(void *, const struct mdoc *);
  46 typedef void            (*out_man)(void *, const struct man *);
  47 typedef void            (*out_free)(void *);
  48
  49 struct  buf {
  50         char             *buf;
  51         size_t            sz;
  52 };
  53
  54 enum    intt {
  55         INTT_AUTO,
  56         INTT_MDOC,
  57         INTT_MAN
  58 };
  59
  60 enum    outt {
  61         OUTT_ASCII = 0,
  62         OUTT_TREE,
  63         OUTT_HTML,
  64         OUTT_XHTML,
  65         OUTT_LINT
  66 };
  67
  68 struct  curparse {
  69         const char       *file;         /* Current parse. */
  70         int               fd;           /* Current parse. */
  71         int               wflags;
  72 #define WARN_WALL        (1 << 0)       /* All-warnings mask. */
  73 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  74         int               fflags;
  75 #define FL_IGN_SCOPE     (1 << 0)       /* Ignore scope errors. */
  76 #define FL_NIGN_ESCAPE   (1 << 1)       /* Don't ignore bad escapes. */
  77 #define FL_NIGN_MACRO    (1 << 2)       /* Don't ignore bad macros. */
  78 #define FL_NIGN_CHARS    (1 << 3)       /* Don't ignore bad chars. */
  79 #define FL_IGN_ERRORS    (1 << 4)       /* Ignore failed parse. */
  80         enum intt         inttype;      /* Input parsers... */
  81         struct man       *man;
  82         struct man       *lastman;
  83         struct mdoc      *mdoc;
  84         struct mdoc      *lastmdoc;
  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 #define FL_STRICT         FL_NIGN_ESCAPE | \
  94                           FL_NIGN_MACRO | \
  95                           FL_NIGN_CHARS
  96
  97 static  int               foptions(int *, char *);
  98 static  int               toptions(struct curparse *, char *);
  99 static  int               moptions(enum intt *, char *);
 100 static  int               woptions(int *, char *);
 101 static  int               merr(void *, int, int, const char *);
 102 static  int               mwarn(void *, int, int, const char *);
 103 static  int               ffile(struct buf *, struct buf *,
 104                                 const char *, struct curparse *);
 105 static  int               fdesc(struct buf *, struct buf *,
 106                                 struct curparse *);
 107 static  int               pset(const char *, int, struct curparse *,
 108                                 struct man **, struct mdoc **);
 109 static  struct man       *man_init(struct curparse *);
 110 static  struct mdoc      *mdoc_init(struct curparse *);
 111 static  void              version(void) __attribute__((noreturn));
 112 static  void              usage(void) __attribute__((noreturn));
 113
 114 static  const char       *progname;
 115
 116
 117 int
 118 main(int argc, char *argv[])
 119 {
 120         int              c, rc;
 121         struct buf       ln, blk;
 122         struct curparse  curp;
 123
 124         progname = strrchr(argv[0], '/');
 125         if (progname == NULL)
 126                 progname = argv[0];
 127         else
 128                 ++progname;
 129
 130         memset(&curp, 0, sizeof(struct curparse));
 131
 132         curp.inttype = INTT_AUTO;
 133         curp.outtype = OUTT_ASCII;
 134
 135         /* LINTED */
 136         while (-1 != (c = getopt(argc, argv, "f:m:O:T:VW:")))
 137                 switch (c) {
 138                 case ('f'):
 139                         if ( ! foptions(&curp.fflags, optarg))
 140                                 return(EXIT_FAILURE);
 141                         break;
 142                 case ('m'):
 143                         if ( ! moptions(&curp.inttype, optarg))
 144                                 return(EXIT_FAILURE);
 145                         break;
 146                 case ('O'):
 147                         (void)strlcat(curp.outopts, optarg, BUFSIZ);
 148                         (void)strlcat(curp.outopts, ",", BUFSIZ);
 149                         break;
 150                 case ('T'):
 151                         if ( ! toptions(&curp, optarg))
 152                                 return(EXIT_FAILURE);
 153                         break;
 154                 case ('W'):
 155                         if ( ! woptions(&curp.wflags, optarg))
 156                                 return(EXIT_FAILURE);
 157                         break;
 158                 case ('V'):
 159                         version();
 160                         /* NOTREACHED */
 161                 default:
 162                         usage();
 163                         /* NOTREACHED */
 164                 }
 165
 166         argc -= optind;
 167         argv += optind;
 168
 169         memset(&ln, 0, sizeof(struct buf));
 170         memset(&blk, 0, sizeof(struct buf));
 171
 172         rc = 1;
 173
 174         if (NULL == *argv) {
 175                 curp.file = "<stdin>";
 176                 curp.fd = STDIN_FILENO;
 177
 178                 c = fdesc(&blk, &ln, &curp);
 179                 if ( ! (FL_IGN_ERRORS & curp.fflags))
 180                         rc = 1 == c ? 1 : 0;
 181                 else
 182                         rc = -1 == c ? 0 : 1;
 183         }
 184
 185         while (rc && *argv) {
 186                 c = ffile(&blk, &ln, *argv, &curp);
 187                 if ( ! (FL_IGN_ERRORS & curp.fflags))
 188                         rc = 1 == c ? 1 : 0;
 189                 else
 190                         rc = -1 == c ? 0 : 1;
 191
 192                 argv++;
 193                 if (*argv && rc) {
 194                         if (curp.lastman)
 195                                 man_reset(curp.lastman);
 196                         if (curp.lastmdoc)
 197                                 mdoc_reset(curp.lastmdoc);
 198                         curp.lastman = NULL;
 199                         curp.lastmdoc = NULL;
 200                 }
 201         }
 202
 203         if (blk.buf)
 204                 free(blk.buf);
 205         if (ln.buf)
 206                 free(ln.buf);
 207         if (curp.outfree)
 208                 (*curp.outfree)(curp.outdata);
 209         if (curp.mdoc)
 210                 mdoc_free(curp.mdoc);
 211         if (curp.man)
 212                 man_free(curp.man);
 213
 214         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 215 }
 216
 217
 218 static void
 219 version(void)
 220 {
 221
 222         (void)printf("%s %s\n", progname, VERSION);
 223         exit(EXIT_SUCCESS);
 224 }
 225
 226
 227 static void
 228 usage(void)
 229 {
 230
 231         (void)fprintf(stderr, "usage: %s [-V] [-foption] "
 232                         "[-mformat] [-Ooption] [-Toutput] "
 233                         "[-Werr] [file...]\n", progname);
 234         exit(EXIT_FAILURE);
 235 }
 236
 237
 238 static struct man *
 239 man_init(struct curparse *curp)
 240 {
 241         int              pflags;
 242         struct man_cb    mancb;
 243
 244         mancb.man_err = merr;
 245         mancb.man_warn = mwarn;
 246
 247         /* Defaults from mandoc.1. */
 248
 249         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS;
 250
 251         if (curp->fflags & FL_NIGN_MACRO)
 252                 pflags &= ~MAN_IGN_MACRO;
 253         if (curp->fflags & FL_NIGN_CHARS)
 254                 pflags &= ~MAN_IGN_CHARS;
 255         if (curp->fflags & FL_NIGN_ESCAPE)
 256                 pflags &= ~MAN_IGN_ESCAPE;
 257
 258         return(man_alloc(curp, pflags, &mancb));
 259 }
 260
 261
 262 static struct mdoc *
 263 mdoc_init(struct curparse *curp)
 264 {
 265         int              pflags;
 266         struct mdoc_cb   mdoccb;
 267
 268         mdoccb.mdoc_err = merr;
 269         mdoccb.mdoc_warn = mwarn;
 270
 271         /* Defaults from mandoc.1. */
 272
 273         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 274
 275         if (curp->fflags & FL_IGN_SCOPE)
 276                 pflags |= MDOC_IGN_SCOPE;
 277         if (curp->fflags & FL_NIGN_ESCAPE)
 278                 pflags &= ~MDOC_IGN_ESCAPE;
 279         if (curp->fflags & FL_NIGN_MACRO)
 280                 pflags &= ~MDOC_IGN_MACRO;
 281         if (curp->fflags & FL_NIGN_CHARS)
 282                 pflags &= ~MDOC_IGN_CHARS;
 283
 284         return(mdoc_alloc(curp, pflags, &mdoccb));
 285 }
 286
 287
 288 static int
 289 ffile(struct buf *blk, struct buf *ln,
 290                 const char *file, struct curparse *curp)
 291 {
 292         int              c;
 293
 294         curp->file = file;
 295         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 296                 perror(curp->file);
 297                 return(-1);
 298         }
 299
 300         c = fdesc(blk, ln, curp);
 301
 302         if (-1 == close(curp->fd))
 303                 perror(curp->file);
 304
 305         return(c);
 306 }
 307
 308
 309 static int
 310 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 311 {
 312         size_t           sz;
 313         ssize_t          ssz;
 314         struct stat      st;
 315         int              j, i, pos, lnn, comment;
 316         struct man      *man;
 317         struct mdoc     *mdoc;
 318
 319         sz = BUFSIZ;
 320         man = NULL;
 321         mdoc = NULL;
 322
 323         /*
 324          * Two buffers: ln and buf.  buf is the input buffer optimised
 325          * here for each file's block size.  ln is a line buffer.  Both
 326          * growable, hence passed in by ptr-ptr.
 327          */
 328
 329         if (-1 == fstat(curp->fd, &st))
 330                 perror(curp->file);
 331         else if ((size_t)st.st_blksize > sz)
 332                 sz = st.st_blksize;
 333
 334         if (sz > blk->sz) {
 335                 blk->buf = realloc(blk->buf, sz);
 336                 if (NULL == blk->buf) {
 337                         perror(NULL);
 338                         exit(EXIT_FAILURE);
 339                 }
 340                 blk->sz = sz;
 341         }
 342
 343         /* Fill buf with file blocksize. */
 344
 345         for (lnn = pos = comment = 0; ; ) {
 346                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 347                         perror(curp->file);
 348                         return(-1);
 349                 } else if (0 == ssz)
 350                         break;
 351
 352                 /* Parse the read block into partial or full lines. */
 353
 354                 for (i = 0; i < (int)ssz; i++) {
 355                         if (pos >= (int)ln->sz) {
 356                                 ln->sz += 256; /* Step-size. */
 357                                 ln->buf = realloc(ln->buf, ln->sz);
 358                                 if (NULL == ln->buf) {
 359                                         perror(NULL);
 360                                         return(EXIT_FAILURE);
 361                                 }
 362                         }
 363
 364                         if ('\n' != blk->buf[i]) {
 365                                 if (comment)
 366                                         continue;
 367                                 ln->buf[pos++] = blk->buf[i];
 368
 369                                 /* Handle in-line `\"' comments. */
 370
 371                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 372                                         continue;
 373
 374                                 for (j = pos - 2; j >= 0; j--)
 375                                         if ('\\' != ln->buf[j])
 376                                                 break;
 377
 378                                 if ( ! ((pos - 2 - j) % 2))
 379                                         continue;
 380
 381                                 comment = 1;
 382                                 pos -= 2;
 383                                 for (; pos > 0; --pos) {
 384                                         if (ln->buf[pos] != ' ')
 385                                                 break;
 386                                         if (ln->buf[pos - 1] == '\\')
 387                                                 break;
 388                                 }
 389                                 continue;
 390                         }
 391
 392                         /* Handle escaped `\\n' newlines. */
 393
 394                         if (pos > 0 && 0 == comment &&
 395                                         '\\' == ln->buf[pos - 1]) {
 396                                 for (j = pos - 1; j >= 0; j--)
 397                                         if ('\\' != ln->buf[j])
 398                                                 break;
 399                                 if ( ! ((pos - j) % 2)) {
 400                                         pos--;
 401                                         lnn++;
 402                                         continue;
 403                                 }
 404                         }
 405
 406                         ln->buf[pos] = 0;
 407                         lnn++;
 408
 409                         /* If unset, assign parser in pset(). */
 410
 411                         if ( ! (man || mdoc) && ! pset(ln->buf,
 412                                                 pos, curp, &man, &mdoc))
 413                                 return(-1);
 414
 415                         pos = comment = 0;
 416
 417                         /* Pass down into parsers. */
 418
 419                         if (man && ! man_parseln(man, lnn, ln->buf))
 420                                 return(0);
 421                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 422                                 return(0);
 423                 }
 424         }
 425
 426         /* NOTE a parser may not have been assigned, yet. */
 427
 428         if ( ! (man || mdoc)) {
 429                 fprintf(stderr, "%s: Not a manual\n", curp->file);
 430                 return(0);
 431         }
 432
 433         if (mdoc && ! mdoc_endparse(mdoc))
 434                 return(0);
 435         if (man && ! man_endparse(man))
 436                 return(0);
 437
 438         /* If unset, allocate output dev now (if applicable). */
 439
 440         if ( ! (curp->outman && curp->outmdoc)) {
 441                 switch (curp->outtype) {
 442                 case (OUTT_XHTML):
 443                         curp->outdata = xhtml_alloc(curp->outopts);
 444                         curp->outman = html_man;
 445                         curp->outmdoc = html_mdoc;
 446                         curp->outfree = html_free;
 447                         break;
 448                 case (OUTT_HTML):
 449                         curp->outdata = html_alloc(curp->outopts);
 450                         curp->outman = html_man;
 451                         curp->outmdoc = html_mdoc;
 452                         curp->outfree = html_free;
 453                         break;
 454                 case (OUTT_TREE):
 455                         curp->outman = tree_man;
 456                         curp->outmdoc = tree_mdoc;
 457                         break;
 458                 case (OUTT_LINT):
 459                         break;
 460                 default:
 461                         curp->outdata = ascii_alloc();
 462                         curp->outman = terminal_man;
 463                         curp->outmdoc = terminal_mdoc;
 464                         curp->outfree = terminal_free;
 465                         break;
 466                 }
 467         }
 468
 469         /* Execute the out device, if it exists. */
 470
 471         if (man && curp->outman)
 472                 (*curp->outman)(curp->outdata, man);
 473         if (mdoc && curp->outmdoc)
 474                 (*curp->outmdoc)(curp->outdata, mdoc);
 475
 476         return(1);
 477 }
 478
 479
 480 static int
 481 pset(const char *buf, int pos, struct curparse *curp,
 482                 struct man **man, struct mdoc **mdoc)
 483 {
 484         int              i;
 485
 486         /*
 487          * Try to intuit which kind of manual parser should be used.  If
 488          * passed in by command-line (-man, -mdoc), then use that
 489          * explicitly.  If passed as -mandoc, then try to guess from the
 490          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 491          * default to -man, which is more lenient.
 492          */
 493
 494         if (buf[0] == '.') {
 495                 for (i = 1; buf[i]; i++)
 496                         if (' ' != buf[i] && '\t' != buf[i])
 497                                 break;
 498                 if (0 == buf[i])
 499                         return(1);
 500         }
 501
 502         switch (curp->inttype) {
 503         case (INTT_MDOC):
 504                 if (NULL == curp->mdoc)
 505                         curp->mdoc = mdoc_init(curp);
 506                 if (NULL == (*mdoc = curp->mdoc))
 507                         return(0);
 508                 curp->lastmdoc = *mdoc;
 509                 return(1);
 510         case (INTT_MAN):
 511                 if (NULL == curp->man)
 512                         curp->man = man_init(curp);
 513                 if (NULL == (*man = curp->man))
 514                         return(0);
 515                 curp->lastman = *man;
 516                 return(1);
 517         default:
 518                 break;
 519         }
 520
 521         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 522                 if (NULL == curp->mdoc)
 523                         curp->mdoc = mdoc_init(curp);
 524                 if (NULL == (*mdoc = curp->mdoc))
 525                         return(0);
 526                 curp->lastmdoc = *mdoc;
 527                 return(1);
 528         }
 529
 530         if (NULL == curp->man)
 531                 curp->man = man_init(curp);
 532         if (NULL == (*man = curp->man))
 533                 return(0);
 534         curp->lastman = *man;
 535         return(1);
 536 }
 537
 538
 539 static int
 540 moptions(enum intt *tflags, char *arg)
 541 {
 542
 543         if (0 == strcmp(arg, "doc"))
 544                 *tflags = INTT_MDOC;
 545         else if (0 == strcmp(arg, "andoc"))
 546                 *tflags = INTT_AUTO;
 547         else if (0 == strcmp(arg, "an"))
 548                 *tflags = INTT_MAN;
 549         else {
 550                 fprintf(stderr, "%s: Bad argument\n", arg);
 551                 return(0);
 552         }
 553
 554         return(1);
 555 }
 556
 557
 558 static int
 559 toptions(struct curparse *curp, char *arg)
 560 {
 561
 562         if (0 == strcmp(arg, "ascii"))
 563                 curp->outtype = OUTT_ASCII;
 564         else if (0 == strcmp(arg, "lint")) {
 565                 curp->outtype = OUTT_LINT;
 566                 curp->wflags |= WARN_WALL;
 567                 curp->fflags |= FL_STRICT;
 568         }
 569         else if (0 == strcmp(arg, "tree"))
 570                 curp->outtype = OUTT_TREE;
 571         else if (0 == strcmp(arg, "html"))
 572                 curp->outtype = OUTT_HTML;
 573         else if (0 == strcmp(arg, "xhtml"))
 574                 curp->outtype = OUTT_XHTML;
 575         else {
 576                 fprintf(stderr, "%s: Bad argument\n", arg);
 577                 return(0);
 578         }
 579
 580         return(1);
 581 }
 582
 583
 584 static int
 585 foptions(int *fflags, char *arg)
 586 {
 587         char            *v, *o;
 588         const char      *toks[8];
 589
 590         toks[0] = "ign-scope";
 591         toks[1] = "no-ign-escape";
 592         toks[2] = "no-ign-macro";
 593         toks[3] = "no-ign-chars";
 594         toks[4] = "ign-errors";
 595         toks[5] = "strict";
 596         toks[6] = "ign-escape";
 597         toks[7] = NULL;
 598
 599         while (*arg) {
 600                 o = arg;
 601                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 602                 case (0):
 603                         *fflags |= FL_IGN_SCOPE;
 604                         break;
 605                 case (1):
 606                         *fflags |= FL_NIGN_ESCAPE;
 607                         break;
 608                 case (2):
 609                         *fflags |= FL_NIGN_MACRO;
 610                         break;
 611                 case (3):
 612                         *fflags |= FL_NIGN_CHARS;
 613                         break;
 614                 case (4):
 615                         *fflags |= FL_IGN_ERRORS;
 616                         break;
 617                 case (5):
 618                         *fflags |= FL_STRICT;
 619                         break;
 620                 case (6):
 621                         *fflags &= ~FL_NIGN_ESCAPE;
 622                         break;
 623                 default:
 624                         fprintf(stderr, "%s: Bad argument\n", o);
 625                         return(0);
 626                 }
 627         }
 628
 629         return(1);
 630 }
 631
 632
 633 static int
 634 woptions(int *wflags, char *arg)
 635 {
 636         char            *v, *o;
 637         const char      *toks[3];
 638
 639         toks[0] = "all";
 640         toks[1] = "error";
 641         toks[2] = NULL;
 642
 643         while (*arg) {
 644                 o = arg;
 645                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 646                 case (0):
 647                         *wflags |= WARN_WALL;
 648                         break;
 649                 case (1):
 650                         *wflags |= WARN_WERR;
 651                         break;
 652                 default:
 653                         fprintf(stderr, "%s: Bad argument\n", o);
 654                         return(0);
 655                 }
 656         }
 657
 658         return(1);
 659 }
 660
 661
 662 /* ARGSUSED */
 663 static int
 664 merr(void *arg, int line, int col, const char *msg)
 665 {
 666         struct curparse *curp;
 667
 668         curp = (struct curparse *)arg;
 669
 670         (void)fprintf(stderr, "%s:%d:%d: error: %s\n",
 671                         curp->file, line, col + 1, msg);
 672
 673         return(0);
 674 }
 675
 676
 677 static int
 678 mwarn(void *arg, int line, int col, const char *msg)
 679 {
 680         struct curparse *curp;
 681
 682         curp = (struct curparse *)arg;
 683
 684         if ( ! (curp->wflags & WARN_WALL))
 685                 return(1);
 686
 687         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 688                         curp->file, line, col + 1, msg);
 689
 690         if ( ! (curp->wflags & WARN_WERR))
 691                 return(1);
 692
 693         return(0);
 694 }
 695