main.c

   1 /*      $Id: main.c,v 1.62 2010/05/09 21:19:42 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_IGN_ERRORS    (1 << 4)       /* Ignore failed parse. */
  79         enum intt         inttype;      /* Input parsers... */
  80         struct man       *man;
  81         struct man       *lastman;
  82         struct mdoc      *mdoc;
  83         struct mdoc      *lastmdoc;
  84         enum outt         outtype;      /* Output devices... */
  85         out_mdoc          outmdoc;
  86         out_man           outman;
  87         out_free          outfree;
  88         void             *outdata;
  89         char              outopts[BUFSIZ];
  90 };
  91
  92 #define FL_STRICT         FL_NIGN_ESCAPE | \
  93                           FL_NIGN_MACRO
  94
  95 static  int               foptions(int *, char *);
  96 static  int               toptions(struct curparse *, char *);
  97 static  int               moptions(enum intt *, char *);
  98 static  int               woptions(int *, char *);
  99 static  int               merr(void *, int, int, const char *);
 100 static  int               mwarn(void *, int, int, const char *);
 101 static  int               ffile(struct buf *, struct buf *,
 102                                 const char *, struct curparse *);
 103 static  int               fdesc(struct buf *, struct buf *,
 104                                 struct curparse *);
 105 static  int               pset(const char *, int, struct curparse *,
 106                                 struct man **, struct mdoc **);
 107 static  struct man       *man_init(struct curparse *);
 108 static  struct mdoc      *mdoc_init(struct curparse *);
 109 static  void              version(void) __attribute__((noreturn));
 110 static  void              usage(void) __attribute__((noreturn));
 111
 112 static  const char       *progname;
 113
 114
 115 int
 116 main(int argc, char *argv[])
 117 {
 118         int              c, rc;
 119         struct buf       ln, blk;
 120         struct curparse  curp;
 121
 122         progname = strrchr(argv[0], '/');
 123         if (progname == NULL)
 124                 progname = argv[0];
 125         else
 126                 ++progname;
 127
 128         memset(&curp, 0, sizeof(struct curparse));
 129
 130         curp.inttype = INTT_AUTO;
 131         curp.outtype = OUTT_ASCII;
 132
 133         /* LINTED */
 134         while (-1 != (c = getopt(argc, argv, "f:m:O:T:VW:")))
 135                 switch (c) {
 136                 case ('f'):
 137                         if ( ! foptions(&curp.fflags, optarg))
 138                                 return(EXIT_FAILURE);
 139                         break;
 140                 case ('m'):
 141                         if ( ! moptions(&curp.inttype, optarg))
 142                                 return(EXIT_FAILURE);
 143                         break;
 144                 case ('O'):
 145                         (void)strlcat(curp.outopts, optarg, BUFSIZ);
 146                         (void)strlcat(curp.outopts, ",", BUFSIZ);
 147                         break;
 148                 case ('T'):
 149                         if ( ! toptions(&curp, optarg))
 150                                 return(EXIT_FAILURE);
 151                         break;
 152                 case ('W'):
 153                         if ( ! woptions(&curp.wflags, optarg))
 154                                 return(EXIT_FAILURE);
 155                         break;
 156                 case ('V'):
 157                         version();
 158                         /* NOTREACHED */
 159                 default:
 160                         usage();
 161                         /* NOTREACHED */
 162                 }
 163
 164         argc -= optind;
 165         argv += optind;
 166
 167         memset(&ln, 0, sizeof(struct buf));
 168         memset(&blk, 0, sizeof(struct buf));
 169
 170         rc = 1;
 171
 172         if (NULL == *argv) {
 173                 curp.file = "<stdin>";
 174                 curp.fd = STDIN_FILENO;
 175
 176                 c = fdesc(&blk, &ln, &curp);
 177                 if ( ! (FL_IGN_ERRORS & curp.fflags))
 178                         rc = 1 == c ? 1 : 0;
 179                 else
 180                         rc = -1 == c ? 0 : 1;
 181         }
 182
 183         while (rc && *argv) {
 184                 c = ffile(&blk, &ln, *argv, &curp);
 185                 if ( ! (FL_IGN_ERRORS & curp.fflags))
 186                         rc = 1 == c ? 1 : 0;
 187                 else
 188                         rc = -1 == c ? 0 : 1;
 189
 190                 argv++;
 191                 if (*argv && rc) {
 192                         if (curp.lastman)
 193                                 man_reset(curp.lastman);
 194                         if (curp.lastmdoc)
 195                                 mdoc_reset(curp.lastmdoc);
 196                         curp.lastman = NULL;
 197                         curp.lastmdoc = NULL;
 198                 }
 199         }
 200
 201         if (blk.buf)
 202                 free(blk.buf);
 203         if (ln.buf)
 204                 free(ln.buf);
 205         if (curp.outfree)
 206                 (*curp.outfree)(curp.outdata);
 207         if (curp.mdoc)
 208                 mdoc_free(curp.mdoc);
 209         if (curp.man)
 210                 man_free(curp.man);
 211
 212         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 213 }
 214
 215
 216 static void
 217 version(void)
 218 {
 219
 220         (void)printf("%s %s\n", progname, VERSION);
 221         exit(EXIT_SUCCESS);
 222 }
 223
 224
 225 static void
 226 usage(void)
 227 {
 228
 229         (void)fprintf(stderr, "usage: %s [-V] [-foption] "
 230                         "[-mformat] [-Ooption] [-Toutput] "
 231                         "[-Werr] [file...]\n", progname);
 232         exit(EXIT_FAILURE);
 233 }
 234
 235
 236 static struct man *
 237 man_init(struct curparse *curp)
 238 {
 239         int              pflags;
 240         struct man_cb    mancb;
 241
 242         mancb.man_err = merr;
 243         mancb.man_warn = mwarn;
 244
 245         /* Defaults from mandoc.1. */
 246
 247         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE;
 248
 249         if (curp->fflags & FL_NIGN_MACRO)
 250                 pflags &= ~MAN_IGN_MACRO;
 251         if (curp->fflags & FL_NIGN_ESCAPE)
 252                 pflags &= ~MAN_IGN_ESCAPE;
 253
 254         return(man_alloc(curp, pflags, &mancb));
 255 }
 256
 257
 258 static struct mdoc *
 259 mdoc_init(struct curparse *curp)
 260 {
 261         int              pflags;
 262         struct mdoc_cb   mdoccb;
 263
 264         mdoccb.mdoc_err = merr;
 265         mdoccb.mdoc_warn = mwarn;
 266
 267         /* Defaults from mandoc.1. */
 268
 269         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE;
 270
 271         if (curp->fflags & FL_IGN_SCOPE)
 272                 pflags |= MDOC_IGN_SCOPE;
 273         if (curp->fflags & FL_NIGN_ESCAPE)
 274                 pflags &= ~MDOC_IGN_ESCAPE;
 275         if (curp->fflags & FL_NIGN_MACRO)
 276                 pflags &= ~MDOC_IGN_MACRO;
 277
 278         return(mdoc_alloc(curp, pflags, &mdoccb));
 279 }
 280
 281
 282 static int
 283 ffile(struct buf *blk, struct buf *ln,
 284                 const char *file, struct curparse *curp)
 285 {
 286         int              c;
 287
 288         curp->file = file;
 289         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 290                 perror(curp->file);
 291                 return(-1);
 292         }
 293
 294         c = fdesc(blk, ln, curp);
 295
 296         if (-1 == close(curp->fd))
 297                 perror(curp->file);
 298
 299         return(c);
 300 }
 301
 302
 303 static int
 304 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 305 {
 306         size_t           sz;
 307         ssize_t          ssz;
 308         struct stat      st;
 309         int              j, i, pos, lnn, comment;
 310         struct man      *man;
 311         struct mdoc     *mdoc;
 312
 313         sz = BUFSIZ;
 314         man = NULL;
 315         mdoc = NULL;
 316
 317         /*
 318          * Two buffers: ln and buf.  buf is the input buffer optimised
 319          * here for each file's block size.  ln is a line buffer.  Both
 320          * growable, hence passed in by ptr-ptr.
 321          */
 322
 323         if (-1 == fstat(curp->fd, &st))
 324                 perror(curp->file);
 325         else if ((size_t)st.st_blksize > sz)
 326                 sz = st.st_blksize;
 327
 328         if (sz > blk->sz) {
 329                 blk->buf = realloc(blk->buf, sz);
 330                 if (NULL == blk->buf) {
 331                         perror(NULL);
 332                         exit(EXIT_FAILURE);
 333                 }
 334                 blk->sz = sz;
 335         }
 336
 337         /* Fill buf with file blocksize. */
 338
 339         for (lnn = pos = comment = 0; ; ) {
 340                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 341                         perror(curp->file);
 342                         return(-1);
 343                 } else if (0 == ssz)
 344                         break;
 345
 346                 /* Parse the read block into partial or full lines. */
 347
 348                 for (i = 0; i < (int)ssz; i++) {
 349                         if (pos >= (int)ln->sz) {
 350                                 ln->sz += 256; /* Step-size. */
 351                                 ln->buf = realloc(ln->buf, ln->sz);
 352                                 if (NULL == ln->buf) {
 353                                         perror(NULL);
 354                                         return(EXIT_FAILURE);
 355                                 }
 356                         }
 357
 358                         if ('\n' != blk->buf[i]) {
 359                                 if (comment)
 360                                         continue;
 361                                 ln->buf[pos++] = blk->buf[i];
 362
 363                                 /* Handle in-line `\"' comments. */
 364
 365                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 366                                         continue;
 367
 368                                 for (j = pos - 2; j >= 0; j--)
 369                                         if ('\\' != ln->buf[j])
 370                                                 break;
 371
 372                                 if ( ! ((pos - 2 - j) % 2))
 373                                         continue;
 374
 375                                 comment = 1;
 376                                 pos -= 2;
 377                                 for (; pos > 0; --pos) {
 378                                         if (ln->buf[pos] != ' ')
 379                                                 break;
 380                                         if (ln->buf[pos - 1] == '\\')
 381                                                 break;
 382                                 }
 383                                 continue;
 384                         }
 385
 386                         /* Handle escaped `\\n' newlines. */
 387
 388                         if (pos > 0 && 0 == comment &&
 389                                         '\\' == ln->buf[pos - 1]) {
 390                                 for (j = pos - 1; j >= 0; j--)
 391                                         if ('\\' != ln->buf[j])
 392                                                 break;
 393                                 if ( ! ((pos - j) % 2)) {
 394                                         pos--;
 395                                         lnn++;
 396                                         continue;
 397                                 }
 398                         }
 399
 400                         ln->buf[pos] = 0;
 401                         lnn++;
 402
 403                         /* If unset, assign parser in pset(). */
 404
 405                         if ( ! (man || mdoc) && ! pset(ln->buf,
 406                                                 pos, curp, &man, &mdoc))
 407                                 return(-1);
 408
 409                         pos = comment = 0;
 410
 411                         /* Pass down into parsers. */
 412
 413                         if (man && ! man_parseln(man, lnn, ln->buf))
 414                                 return(0);
 415                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 416                                 return(0);
 417                 }
 418         }
 419
 420         /* NOTE a parser may not have been assigned, yet. */
 421
 422         if ( ! (man || mdoc)) {
 423                 fprintf(stderr, "%s: Not a manual\n", curp->file);
 424                 return(0);
 425         }
 426
 427         if (mdoc && ! mdoc_endparse(mdoc))
 428                 return(0);
 429         if (man && ! man_endparse(man))
 430                 return(0);
 431
 432         /* If unset, allocate output dev now (if applicable). */
 433
 434         if ( ! (curp->outman && curp->outmdoc)) {
 435                 switch (curp->outtype) {
 436                 case (OUTT_XHTML):
 437                         curp->outdata = xhtml_alloc(curp->outopts);
 438                         curp->outman = html_man;
 439                         curp->outmdoc = html_mdoc;
 440                         curp->outfree = html_free;
 441                         break;
 442                 case (OUTT_HTML):
 443                         curp->outdata = html_alloc(curp->outopts);
 444                         curp->outman = html_man;
 445                         curp->outmdoc = html_mdoc;
 446                         curp->outfree = html_free;
 447                         break;
 448                 case (OUTT_TREE):
 449                         curp->outman = tree_man;
 450                         curp->outmdoc = tree_mdoc;
 451                         break;
 452                 case (OUTT_LINT):
 453                         break;
 454                 default:
 455                         curp->outdata = ascii_alloc();
 456                         curp->outman = terminal_man;
 457                         curp->outmdoc = terminal_mdoc;
 458                         curp->outfree = terminal_free;
 459                         break;
 460                 }
 461         }
 462
 463         /* Execute the out device, if it exists. */
 464
 465         if (man && curp->outman)
 466                 (*curp->outman)(curp->outdata, man);
 467         if (mdoc && curp->outmdoc)
 468                 (*curp->outmdoc)(curp->outdata, mdoc);
 469
 470         return(1);
 471 }
 472
 473
 474 static int
 475 pset(const char *buf, int pos, struct curparse *curp,
 476                 struct man **man, struct mdoc **mdoc)
 477 {
 478         int              i;
 479
 480         /*
 481          * Try to intuit which kind of manual parser should be used.  If
 482          * passed in by command-line (-man, -mdoc), then use that
 483          * explicitly.  If passed as -mandoc, then try to guess from the
 484          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 485          * default to -man, which is more lenient.
 486          */
 487
 488         if (buf[0] == '.') {
 489                 for (i = 1; buf[i]; i++)
 490                         if (' ' != buf[i] && '\t' != buf[i])
 491                                 break;
 492                 if (0 == buf[i])
 493                         return(1);
 494         }
 495
 496         switch (curp->inttype) {
 497         case (INTT_MDOC):
 498                 if (NULL == curp->mdoc)
 499                         curp->mdoc = mdoc_init(curp);
 500                 if (NULL == (*mdoc = curp->mdoc))
 501                         return(0);
 502                 curp->lastmdoc = *mdoc;
 503                 return(1);
 504         case (INTT_MAN):
 505                 if (NULL == curp->man)
 506                         curp->man = man_init(curp);
 507                 if (NULL == (*man = curp->man))
 508                         return(0);
 509                 curp->lastman = *man;
 510                 return(1);
 511         default:
 512                 break;
 513         }
 514
 515         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 516                 if (NULL == curp->mdoc)
 517                         curp->mdoc = mdoc_init(curp);
 518                 if (NULL == (*mdoc = curp->mdoc))
 519                         return(0);
 520                 curp->lastmdoc = *mdoc;
 521                 return(1);
 522         }
 523
 524         if (NULL == curp->man)
 525                 curp->man = man_init(curp);
 526         if (NULL == (*man = curp->man))
 527                 return(0);
 528         curp->lastman = *man;
 529         return(1);
 530 }
 531
 532
 533 static int
 534 moptions(enum intt *tflags, char *arg)
 535 {
 536
 537         if (0 == strcmp(arg, "doc"))
 538                 *tflags = INTT_MDOC;
 539         else if (0 == strcmp(arg, "andoc"))
 540                 *tflags = INTT_AUTO;
 541         else if (0 == strcmp(arg, "an"))
 542                 *tflags = INTT_MAN;
 543         else {
 544                 fprintf(stderr, "%s: Bad argument\n", arg);
 545                 return(0);
 546         }
 547
 548         return(1);
 549 }
 550
 551
 552 static int
 553 toptions(struct curparse *curp, char *arg)
 554 {
 555
 556         if (0 == strcmp(arg, "ascii"))
 557                 curp->outtype = OUTT_ASCII;
 558         else if (0 == strcmp(arg, "lint")) {
 559                 curp->outtype = OUTT_LINT;
 560                 curp->wflags |= WARN_WALL;
 561                 curp->fflags |= FL_STRICT;
 562         }
 563         else if (0 == strcmp(arg, "tree"))
 564                 curp->outtype = OUTT_TREE;
 565         else if (0 == strcmp(arg, "html"))
 566                 curp->outtype = OUTT_HTML;
 567         else if (0 == strcmp(arg, "xhtml"))
 568                 curp->outtype = OUTT_XHTML;
 569         else {
 570                 fprintf(stderr, "%s: Bad argument\n", arg);
 571                 return(0);
 572         }
 573
 574         return(1);
 575 }
 576
 577
 578 static int
 579 foptions(int *fflags, char *arg)
 580 {
 581         char            *v, *o;
 582         const char      *toks[8];
 583
 584         toks[0] = "ign-scope";
 585         toks[1] = "no-ign-escape";
 586         toks[2] = "no-ign-macro";
 587         toks[3] = "ign-errors";
 588         toks[4] = "strict";
 589         toks[5] = "ign-escape";
 590         toks[6] = NULL;
 591
 592         while (*arg) {
 593                 o = arg;
 594                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 595                 case (0):
 596                         *fflags |= FL_IGN_SCOPE;
 597                         break;
 598                 case (1):
 599                         *fflags |= FL_NIGN_ESCAPE;
 600                         break;
 601                 case (2):
 602                         *fflags |= FL_NIGN_MACRO;
 603                         break;
 604                 case (3):
 605                         *fflags |= FL_IGN_ERRORS;
 606                         break;
 607                 case (4):
 608                         *fflags |= FL_STRICT;
 609                         break;
 610                 case (5):
 611                         *fflags &= ~FL_NIGN_ESCAPE;
 612                         break;
 613                 default:
 614                         fprintf(stderr, "%s: Bad argument\n", o);
 615                         return(0);
 616                 }
 617         }
 618
 619         return(1);
 620 }
 621
 622
 623 static int
 624 woptions(int *wflags, char *arg)
 625 {
 626         char            *v, *o;
 627         const char      *toks[3];
 628
 629         toks[0] = "all";
 630         toks[1] = "error";
 631         toks[2] = NULL;
 632
 633         while (*arg) {
 634                 o = arg;
 635                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 636                 case (0):
 637                         *wflags |= WARN_WALL;
 638                         break;
 639                 case (1):
 640                         *wflags |= WARN_WERR;
 641                         break;
 642                 default:
 643                         fprintf(stderr, "%s: Bad argument\n", o);
 644                         return(0);
 645                 }
 646         }
 647
 648         return(1);
 649 }
 650
 651
 652 /* ARGSUSED */
 653 static int
 654 merr(void *arg, int line, int col, const char *msg)
 655 {
 656         struct curparse *curp;
 657
 658         curp = (struct curparse *)arg;
 659
 660         (void)fprintf(stderr, "%s:%d:%d: error: %s\n",
 661                         curp->file, line, col + 1, msg);
 662
 663         return(0);
 664 }
 665
 666
 667 static int
 668 mwarn(void *arg, int line, int col, const char *msg)
 669 {
 670         struct curparse *curp;
 671
 672         curp = (struct curparse *)arg;
 673
 674         if ( ! (curp->wflags & WARN_WALL))
 675                 return(1);
 676
 677         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 678                         curp->file, line, col + 1, msg);
 679
 680         if ( ! (curp->wflags & WARN_WERR))
 681                 return(1);
 682
 683         return(0);
 684 }
 685