main.c

   1 /*      $Id: main.c,v 1.123 2010/12/15 23:39:40 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010 Ingo Schwarze <schwarze@openbsd.org>
   5  *
   6  * Permission to use, copy, modify, and distribute this software for any
   7  * purpose with or without fee is hereby granted, provided that the above
   8  * copyright notice and this permission notice appear in all copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  17  */
  18 #ifdef HAVE_CONFIG_H
  19 #include "config.h"
  20 #endif
  21
  22 #include <sys/mman.h>
  23 #include <sys/stat.h>
  24
  25 #include <assert.h>
  26 #include <ctype.h>
  27 #include <fcntl.h>
  28 #include <stdio.h>
  29 #include <stdint.h>
  30 #include <stdlib.h>
  31 #include <string.h>
  32 #include <unistd.h>
  33
  34 #include "mandoc.h"
  35 #include "main.h"
  36 #include "mdoc.h"
  37 #include "man.h"
  38 #include "roff.h"
  39
  40 #ifndef MAP_FILE
  41 #define MAP_FILE        0
  42 #endif
  43
  44 #define REPARSE_LIMIT   1000
  45 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  46
  47 /* FIXME: Intel's compiler?  LLVM?  pcc?  */
  48
  49 #if !defined(__GNUC__) || (__GNUC__ < 2)
  50 # if !defined(lint)
  51 #  define __attribute__(x)
  52 # endif
  53 #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */
  54
  55 typedef void            (*out_mdoc)(void *, const struct mdoc *);
  56 typedef void            (*out_man)(void *, const struct man *);
  57 typedef void            (*out_free)(void *);
  58
  59 struct  buf {
  60         char             *buf;
  61         size_t            sz;
  62 };
  63
  64 enum    intt {
  65         INTT_AUTO,
  66         INTT_MDOC,
  67         INTT_MAN
  68 };
  69
  70 enum    outt {
  71         OUTT_ASCII = 0,
  72         OUTT_TREE,
  73         OUTT_HTML,
  74         OUTT_XHTML,
  75         OUTT_LINT,
  76         OUTT_PS,
  77         OUTT_PDF
  78 };
  79
  80 struct  curparse {
  81         const char       *file;         /* Current parse. */
  82         int               fd;           /* Current parse. */
  83         int               line;         /* Line number in the file. */
  84         enum mandoclevel  wlevel;       /* Ignore messages below this. */
  85         int               wstop;        /* Stop after a file with a warning. */
  86         enum intt         inttype;      /* which parser to use */
  87         struct man       *pman;         /* persistent man parser */
  88         struct mdoc      *pmdoc;        /* persistent mdoc parser */
  89         struct man       *man;          /* man parser */
  90         struct mdoc      *mdoc;         /* mdoc parser */
  91         struct roff      *roff;         /* roff parser (!NULL) */
  92         struct regset     regs;         /* roff registers */
  93         int               reparse_count; /* finite interpolation stack */
  94         enum outt         outtype;      /* which output to use */
  95         out_mdoc          outmdoc;      /* mdoc output ptr */
  96         out_man           outman;       /* man output ptr */
  97         out_free          outfree;      /* free output ptr */
  98         void             *outdata;      /* data for output */
  99         char              outopts[BUFSIZ]; /* buf of output opts */
 100 };
 101
 102 static  const char * const      mandoclevels[MANDOCLEVEL_MAX] = {
 103         "SUCCESS",
 104         "RESERVED",
 105         "WARNING",
 106         "ERROR",
 107         "FATAL",
 108         "BADARG",
 109         "SYSERR"
 110 };
 111
 112 static  const enum mandocerr    mandoclimits[MANDOCLEVEL_MAX] = {
 113         MANDOCERR_OK,
 114         MANDOCERR_WARNING,
 115         MANDOCERR_WARNING,
 116         MANDOCERR_ERROR,
 117         MANDOCERR_FATAL,
 118         MANDOCERR_MAX,
 119         MANDOCERR_MAX
 120 };
 121
 122 static  const char * const      mandocerrs[MANDOCERR_MAX] = {
 123         "ok",
 124
 125         "generic warning",
 126
 127         /* related to the prologue */
 128         "no title in document",
 129         "document title should be all caps",
 130         "unknown manual section",
 131         "cannot parse date argument",
 132         "prologue macros out of order",
 133         "duplicate prologue macro",
 134         "macro not allowed in prologue",
 135         "macro not allowed in body",
 136
 137         /* related to document structure */
 138         ".so is fragile, better use ln(1)",
 139         "NAME section must come first",
 140         "bad NAME section contents",
 141         "manual name not yet set",
 142         "sections out of conventional order",
 143         "duplicate section name",
 144         "section not in conventional manual section",
 145
 146         /* related to macros and nesting */
 147         "skipping obsolete macro",
 148         "blocks badly nested",
 149         "child violates parent syntax",
 150         "nested displays are not portable",
 151         "already in literal mode",
 152
 153         /* related to missing macro arguments */
 154         "skipping empty macro",
 155         "missing display type",
 156         "list type must come first",
 157         "tag lists require a width argument",
 158         "missing font type",
 159
 160         /* related to bad macro arguments */
 161         "skipping argument",
 162         "duplicate argument",
 163         "duplicate display type",
 164         "duplicate list type",
 165         "unknown AT&T UNIX version",
 166         "bad Boolean value",
 167         "unknown font",
 168         "unknown standard specifier",
 169         "bad width argument",
 170
 171         /* related to plain text */
 172         "blank line in non-literal context",
 173         "tab in non-literal context",
 174         "end of line whitespace",
 175         "bad comment style",
 176         "unknown escape sequence",
 177         "unterminated quoted string",
 178
 179         "generic error",
 180
 181         "input stack limit exceeded, infinite loop?",
 182         "skipping bad character",
 183         "skipping text before the first section header",
 184         "skipping unknown macro",
 185         "skipping paragraph macro",
 186         "NOT IMPLEMENTED: skipping request",
 187         "line scope broken",
 188         "argument count wrong",
 189         "skipping end of block that is not open",
 190         "missing end of block",
 191         "scope open on exit",
 192         "uname(3) system call failed",
 193         "macro requires line argument(s)",
 194         "macro requires body argument(s)",
 195         "macro requires argument(s)",
 196         "missing list type",
 197         "line argument(s) will be lost",
 198         "body argument(s) will be lost",
 199
 200         "generic fatal error",
 201
 202         "column syntax is inconsistent",
 203         "NOT IMPLEMENTED: .Bd -file",
 204         "line scope broken, syntax violated",
 205         "argument count wrong, violates syntax",
 206         "child violates parent syntax",
 207         "argument count wrong, violates syntax",
 208         "NOT IMPLEMENTED: .so with absolute path or \"..\"",
 209         "no document body",
 210         "no document prologue",
 211         "static buffer exhausted",
 212 };
 213
 214 static  void              parsebuf(struct curparse *, struct buf, int);
 215 static  void              pdesc(struct curparse *);
 216 static  void              fdesc(struct curparse *);
 217 static  void              ffile(const char *, struct curparse *);
 218 static  int               pfile(const char *, struct curparse *);
 219 static  int               moptions(enum intt *, char *);
 220 static  int               mmsg(enum mandocerr, void *,
 221                                 int, int, const char *);
 222 static  void              pset(const char *, int, struct curparse *);
 223 static  int               toptions(struct curparse *, char *);
 224 static  void              usage(void) __attribute__((noreturn));
 225 static  void              version(void) __attribute__((noreturn));
 226 static  int               woptions(struct curparse *, char *);
 227
 228 static  const char       *progname;
 229 static  enum mandoclevel  file_status = MANDOCLEVEL_OK;
 230 static  enum mandoclevel  exit_status = MANDOCLEVEL_OK;
 231
 232 int
 233 main(int argc, char *argv[])
 234 {
 235         int              c;
 236         struct curparse  curp;
 237
 238         progname = strrchr(argv[0], '/');
 239         if (progname == NULL)
 240                 progname = argv[0];
 241         else
 242                 ++progname;
 243
 244         memset(&curp, 0, sizeof(struct curparse));
 245
 246         curp.inttype = INTT_AUTO;
 247         curp.outtype = OUTT_ASCII;
 248         curp.wlevel  = MANDOCLEVEL_FATAL;
 249
 250         /* LINTED */
 251         while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
 252                 switch (c) {
 253                 case ('m'):
 254                         if ( ! moptions(&curp.inttype, optarg))
 255                                 return((int)MANDOCLEVEL_BADARG);
 256                         break;
 257                 case ('O'):
 258                         (void)strlcat(curp.outopts, optarg, BUFSIZ);
 259                         (void)strlcat(curp.outopts, ",", BUFSIZ);
 260                         break;
 261                 case ('T'):
 262                         if ( ! toptions(&curp, optarg))
 263                                 return((int)MANDOCLEVEL_BADARG);
 264                         break;
 265                 case ('W'):
 266                         if ( ! woptions(&curp, optarg))
 267                                 return((int)MANDOCLEVEL_BADARG);
 268                         break;
 269                 case ('V'):
 270                         version();
 271                         /* NOTREACHED */
 272                 default:
 273                         usage();
 274                         /* NOTREACHED */
 275                 }
 276
 277         argc -= optind;
 278         argv += optind;
 279
 280         if (NULL == *argv) {
 281                 curp.file = "<stdin>";
 282                 curp.fd = STDIN_FILENO;
 283
 284                 fdesc(&curp);
 285         }
 286
 287         while (*argv) {
 288                 ffile(*argv, &curp);
 289                 if (MANDOCLEVEL_OK != exit_status && curp.wstop)
 290                         break;
 291                 ++argv;
 292         }
 293
 294         if (curp.outfree)
 295                 (*curp.outfree)(curp.outdata);
 296         if (curp.pmdoc)
 297                 mdoc_free(curp.pmdoc);
 298         if (curp.pman)
 299                 man_free(curp.pman);
 300         if (curp.roff)
 301                 roff_free(curp.roff);
 302
 303         return((int)exit_status);
 304 }
 305
 306
 307 static void
 308 version(void)
 309 {
 310
 311         (void)printf("%s %s\n", progname, VERSION);
 312         exit((int)MANDOCLEVEL_OK);
 313 }
 314
 315
 316 static void
 317 usage(void)
 318 {
 319
 320         (void)fprintf(stderr, "usage: %s "
 321                         "[-V] "
 322                         "[-foption] "
 323                         "[-mformat] "
 324                         "[-Ooption] "
 325                         "[-Toutput] "
 326                         "[-Werr] "
 327                         "[file...]\n",
 328                         progname);
 329
 330         exit((int)MANDOCLEVEL_BADARG);
 331 }
 332
 333 static void
 334 ffile(const char *file, struct curparse *curp)
 335 {
 336
 337         /*
 338          * Called once per input file.  Get the file ready for reading,
 339          * pass it through to the parser-driver, then close it out.
 340          * XXX: don't do anything special as this is only called for
 341          * files; stdin goes directly to fdesc().
 342          */
 343
 344         curp->file = file;
 345
 346         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 347                 perror(curp->file);
 348                 exit_status = MANDOCLEVEL_SYSERR;
 349                 return;
 350         }
 351
 352         fdesc(curp);
 353
 354         if (-1 == close(curp->fd))
 355                 perror(curp->file);
 356 }
 357
 358 static int
 359 pfile(const char *file, struct curparse *curp)
 360 {
 361         const char      *savefile;
 362         int              fd, savefd;
 363
 364         if (-1 == (fd = open(file, O_RDONLY, 0))) {
 365                 perror(file);
 366                 file_status = MANDOCLEVEL_SYSERR;
 367                 return(0);
 368         }
 369
 370         savefile = curp->file;
 371         savefd = curp->fd;
 372
 373         curp->file = file;
 374         curp->fd = fd;
 375
 376         pdesc(curp);
 377
 378         curp->file = savefile;
 379         curp->fd = savefd;
 380
 381         if (-1 == close(fd))
 382                 perror(file);
 383
 384         return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
 385 }
 386
 387
 388 static void
 389 resize_buf(struct buf *buf, size_t initial)
 390 {
 391
 392         buf->sz = buf->sz ? 2 * buf->sz : initial;
 393         buf->buf = realloc(buf->buf, buf->sz);
 394         if (NULL == buf->buf) {
 395                 perror(NULL);
 396                 exit((int)MANDOCLEVEL_SYSERR);
 397         }
 398 }
 399
 400
 401 static int
 402 read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
 403 {
 404         struct stat      st;
 405         size_t           off;
 406         ssize_t          ssz;
 407
 408         if (-1 == fstat(curp->fd, &st)) {
 409                 perror(curp->file);
 410                 return(0);
 411         }
 412
 413         /*
 414          * If we're a regular file, try just reading in the whole entry
 415          * via mmap().  This is faster than reading it into blocks, and
 416          * since each file is only a few bytes to begin with, I'm not
 417          * concerned that this is going to tank any machines.
 418          */
 419
 420         if (S_ISREG(st.st_mode)) {
 421                 if (st.st_size >= (1U << 31)) {
 422                         fprintf(stderr, "%s: input too large\n",
 423                                         curp->file);
 424                         return(0);
 425                 }
 426                 *with_mmap = 1;
 427                 fb->sz = (size_t)st.st_size;
 428                 fb->buf = mmap(NULL, fb->sz, PROT_READ,
 429                                 MAP_FILE|MAP_SHARED, curp->fd, 0);
 430                 if (fb->buf != MAP_FAILED)
 431                         return(1);
 432         }
 433
 434         /*
 435          * If this isn't a regular file (like, say, stdin), then we must
 436          * go the old way and just read things in bit by bit.
 437          */
 438
 439         *with_mmap = 0;
 440         off = 0;
 441         fb->sz = 0;
 442         fb->buf = NULL;
 443         for (;;) {
 444                 if (off == fb->sz) {
 445                         if (fb->sz == (1U << 31)) {
 446                                 fprintf(stderr, "%s: input too large\n",
 447                                                 curp->file);
 448                                 break;
 449                         }
 450                         resize_buf(fb, 65536);
 451                 }
 452                 ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
 453                 if (ssz == 0) {
 454                         fb->sz = off;
 455                         return(1);
 456                 }
 457                 if (ssz == -1) {
 458                         perror(curp->file);
 459                         break;
 460                 }
 461                 off += (size_t)ssz;
 462         }
 463
 464         free(fb->buf);
 465         fb->buf = NULL;
 466         return(0);
 467 }
 468
 469
 470 static void
 471 fdesc(struct curparse *curp)
 472 {
 473
 474         /*
 475          * Called once per file with an opened file descriptor.  All
 476          * pre-file-parse operations (whether stdin or a file) should go
 477          * here.
 478          *
 479          * This calls down into the nested parser, which drills down and
 480          * fully parses a file and all its dependences (i.e., `so').  It
 481          * then runs the cleanup validators and pushes to output.
 482          */
 483
 484         /* Zero the parse type. */
 485
 486         curp->mdoc = NULL;
 487         curp->man = NULL;
 488         file_status = MANDOCLEVEL_OK;
 489
 490         /* Make sure the mandotory roff parser is initialised. */
 491
 492         if (NULL == curp->roff) {
 493                 curp->roff = roff_alloc(&curp->regs, curp, mmsg);
 494                 assert(curp->roff);
 495         }
 496
 497         /* Fully parse the file. */
 498
 499         pdesc(curp);
 500
 501         if (MANDOCLEVEL_FATAL <= file_status)
 502                 goto cleanup;
 503
 504         /* NOTE a parser may not have been assigned, yet. */
 505
 506         if ( ! (curp->man || curp->mdoc)) {
 507                 fprintf(stderr, "%s: Not a manual\n", curp->file);
 508                 file_status = MANDOCLEVEL_FATAL;
 509                 goto cleanup;
 510         }
 511
 512         /* Clean up the parse routine ASTs. */
 513
 514         if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
 515                 assert(MANDOCLEVEL_FATAL <= file_status);
 516                 goto cleanup;
 517         }
 518
 519         if (curp->man && ! man_endparse(curp->man)) {
 520                 assert(MANDOCLEVEL_FATAL <= file_status);
 521                 goto cleanup;
 522         }
 523
 524         assert(curp->roff);
 525         if ( ! roff_endparse(curp->roff)) {
 526                 assert(MANDOCLEVEL_FATAL <= file_status);
 527                 goto cleanup;
 528         }
 529
 530         /*
 531          * With -Wstop and warnings or errors of at least
 532          * the requested level, do not produce output.
 533          */
 534
 535         if (MANDOCLEVEL_OK != file_status && curp->wstop)
 536                 goto cleanup;
 537
 538         /* If unset, allocate output dev now (if applicable). */
 539
 540         if ( ! (curp->outman && curp->outmdoc)) {
 541                 switch (curp->outtype) {
 542                 case (OUTT_XHTML):
 543                         curp->outdata = xhtml_alloc(curp->outopts);
 544                         break;
 545                 case (OUTT_HTML):
 546                         curp->outdata = html_alloc(curp->outopts);
 547                         break;
 548                 case (OUTT_ASCII):
 549                         curp->outdata = ascii_alloc(curp->outopts);
 550                         curp->outfree = ascii_free;
 551                         break;
 552                 case (OUTT_PDF):
 553                         curp->outdata = pdf_alloc(curp->outopts);
 554                         curp->outfree = pspdf_free;
 555                         break;
 556                 case (OUTT_PS):
 557                         curp->outdata = ps_alloc(curp->outopts);
 558                         curp->outfree = pspdf_free;
 559                         break;
 560                 default:
 561                         break;
 562                 }
 563
 564                 switch (curp->outtype) {
 565                 case (OUTT_HTML):
 566                         /* FALLTHROUGH */
 567                 case (OUTT_XHTML):
 568                         curp->outman = html_man;
 569                         curp->outmdoc = html_mdoc;
 570                         curp->outfree = html_free;
 571                         break;
 572                 case (OUTT_TREE):
 573                         curp->outman = tree_man;
 574                         curp->outmdoc = tree_mdoc;
 575                         break;
 576                 case (OUTT_PDF):
 577                         /* FALLTHROUGH */
 578                 case (OUTT_ASCII):
 579                         /* FALLTHROUGH */
 580                 case (OUTT_PS):
 581                         curp->outman = terminal_man;
 582                         curp->outmdoc = terminal_mdoc;
 583                         break;
 584                 default:
 585                         break;
 586                 }
 587         }
 588
 589         /* Execute the out device, if it exists. */
 590
 591         if (curp->man && curp->outman)
 592                 (*curp->outman)(curp->outdata, curp->man);
 593         if (curp->mdoc && curp->outmdoc)
 594                 (*curp->outmdoc)(curp->outdata, curp->mdoc);
 595
 596  cleanup:
 597
 598         memset(&curp->regs, 0, sizeof(struct regset));
 599
 600         /* Reset the current-parse compilers. */
 601
 602         if (curp->mdoc)
 603                 mdoc_reset(curp->mdoc);
 604         if (curp->man)
 605                 man_reset(curp->man);
 606
 607         assert(curp->roff);
 608         roff_reset(curp->roff);
 609
 610         if (exit_status < file_status)
 611                 exit_status = file_status;
 612
 613         return;
 614 }
 615
 616 static void
 617 pdesc(struct curparse *curp)
 618 {
 619         struct buf       blk;
 620         int              with_mmap;
 621
 622         /*
 623          * Run for each opened file; may be called more than once for
 624          * each full parse sequence if the opened file is nested (i.e.,
 625          * from `so').  Simply sucks in the whole file and moves into
 626          * the parse phase for the file.
 627          */
 628
 629         if ( ! read_whole_file(curp, &blk, &with_mmap)) {
 630                 file_status = MANDOCLEVEL_SYSERR;
 631                 return;
 632         }
 633
 634         /* Line number is per-file. */
 635
 636         curp->line = 1;
 637
 638         parsebuf(curp, blk, 1);
 639
 640         if (with_mmap)
 641                 munmap(blk.buf, blk.sz);
 642         else
 643                 free(blk.buf);
 644 }
 645
 646 static void
 647 parsebuf(struct curparse *curp, struct buf blk, int start)
 648 {
 649         struct buf       ln;
 650         enum rofferr     rr;
 651         int              i, of, rc;
 652         int              pos; /* byte number in the ln buffer */
 653         int              lnn; /* line number in the real file */
 654         unsigned char    c;
 655
 656         /*
 657          * Main parse routine for an opened file.  This is called for
 658          * each opened file and simply loops around the full input file,
 659          * possibly nesting (i.e., with `so').
 660          */
 661
 662         memset(&ln, 0, sizeof(struct buf));
 663
 664         lnn = curp->line;
 665         pos = 0;
 666
 667         for (i = 0; i < (int)blk.sz; ) {
 668                 if (0 == pos && '\0' == blk.buf[i])
 669                         break;
 670
 671                 if (start) {
 672                         curp->line = lnn;
 673                         curp->reparse_count = 0;
 674                 }
 675
 676                 while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
 677                         if ('\n' == blk.buf[i]) {
 678                                 ++i;
 679                                 ++lnn;
 680                                 break;
 681                         }
 682
 683                         /*
 684                          * Warn about bogus characters.  If you're using
 685                          * non-ASCII encoding, you're screwing your
 686                          * readers.  Since I'd rather this not happen,
 687                          * I'll be helpful and drop these characters so
 688                          * we don't display gibberish.  Note to manual
 689                          * writers: use special characters.
 690                          */
 691
 692                         c = (unsigned char) blk.buf[i];
 693
 694                         if ( ! (isascii(c) &&
 695                                         (isgraph(c) || isblank(c)))) {
 696                                 mmsg(MANDOCERR_BADCHAR, curp,
 697                                     curp->line, pos, "ignoring byte");
 698                                 i++;
 699                                 continue;
 700                         }
 701
 702                         /* Trailing backslash = a plain char. */
 703
 704                         if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
 705                                 if (pos >= (int)ln.sz)
 706                                         resize_buf(&ln, 256);
 707                                 ln.buf[pos++] = blk.buf[i++];
 708                                 continue;
 709                         }
 710
 711                         /* Found escape & at least one other char. */
 712
 713                         if ('\n' == blk.buf[i + 1]) {
 714                                 i += 2;
 715                                 /* Escaped newlines are skipped over */
 716                                 ++lnn;
 717                                 continue;
 718                         }
 719
 720                         if ('"' == blk.buf[i + 1]) {
 721                                 i += 2;
 722                                 /* Comment, skip to end of line */
 723                                 for (; i < (int)blk.sz; ++i) {
 724                                         if ('\n' == blk.buf[i]) {
 725                                                 ++i;
 726                                                 ++lnn;
 727                                                 break;
 728                                         }
 729                                 }
 730
 731                                 /* Backout trailing whitespaces */
 732                                 for (; pos > 0; --pos) {
 733                                         if (ln.buf[pos - 1] != ' ')
 734                                                 break;
 735                                         if (pos > 2 && ln.buf[pos - 2] == '\\')
 736                                                 break;
 737                                 }
 738                                 break;
 739                         }
 740
 741                         /* Some other escape sequence, copy & cont. */
 742
 743                         if (pos + 1 >= (int)ln.sz)
 744                                 resize_buf(&ln, 256);
 745
 746                         ln.buf[pos++] = blk.buf[i++];
 747                         ln.buf[pos++] = blk.buf[i++];
 748                 }
 749
 750                 if (pos >= (int)ln.sz)
 751                         resize_buf(&ln, 256);
 752
 753                 ln.buf[pos] = '\0';
 754
 755                 /*
 756                  * A significant amount of complexity is contained by
 757                  * the roff preprocessor.  It's line-oriented but can be
 758                  * expressed on one line, so we need at times to
 759                  * readjust our starting point and re-run it.  The roff
 760                  * preprocessor can also readjust the buffers with new
 761                  * data, so we pass them in wholesale.
 762                  */
 763
 764                 of = 0;
 765
 766 rerun:
 767                 rr = roff_parseln
 768                         (curp->roff, curp->line,
 769                          &ln.buf, &ln.sz, of, &of);
 770
 771                 switch (rr) {
 772                 case (ROFF_REPARSE):
 773                         if (REPARSE_LIMIT >= ++curp->reparse_count)
 774                                 parsebuf(curp, ln, 0);
 775                         else
 776                                 mmsg(MANDOCERR_ROFFLOOP, curp,
 777                                     curp->line, pos, NULL);
 778                         pos = 0;
 779                         continue;
 780                 case (ROFF_APPEND):
 781                         pos = strlen(ln.buf);
 782                         continue;
 783                 case (ROFF_RERUN):
 784                         goto rerun;
 785                 case (ROFF_IGN):
 786                         pos = 0;
 787                         continue;
 788                 case (ROFF_ERR):
 789                         assert(MANDOCLEVEL_FATAL <= file_status);
 790                         break;
 791                 case (ROFF_SO):
 792                         if (pfile(ln.buf + of, curp)) {
 793                                 pos = 0;
 794                                 continue;
 795                         } else
 796                                 break;
 797                 case (ROFF_CONT):
 798                         break;
 799                 }
 800
 801                 /*
 802                  * If input parsers have not been allocated, do so now.
 803                  * We keep these instanced betwen parsers, but set them
 804                  * locally per parse routine since we can use different
 805                  * parsers with each one.
 806                  */
 807
 808                 if ( ! (curp->man || curp->mdoc))
 809                         pset(ln.buf + of, pos - of, curp);
 810
 811                 /*
 812                  * Lastly, push down into the parsers themselves.  One
 813                  * of these will have already been set in the pset()
 814                  * routine.
 815                  */
 816
 817                 if (curp->man || curp->mdoc) {
 818                         rc = curp->man ?
 819                                 man_parseln(curp->man,
 820                                         curp->line, ln.buf, of) :
 821                                 mdoc_parseln(curp->mdoc,
 822                                         curp->line, ln.buf, of);
 823
 824                         if ( ! rc) {
 825                                 assert(MANDOCLEVEL_FATAL <= file_status);
 826                                 break;
 827                         }
 828                 }
 829
 830                 /* Temporary buffers typically are not full. */
 831
 832                 if (0 == start && '\0' == blk.buf[i])
 833                         break;
 834
 835                 /* Start the next input line. */
 836
 837                 pos = 0;
 838         }
 839
 840         free(ln.buf);
 841 }
 842
 843 static void
 844 pset(const char *buf, int pos, struct curparse *curp)
 845 {
 846         int              i;
 847
 848         /*
 849          * Try to intuit which kind of manual parser should be used.  If
 850          * passed in by command-line (-man, -mdoc), then use that
 851          * explicitly.  If passed as -mandoc, then try to guess from the
 852          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 853          * default to -man, which is more lenient.
 854          *
 855          * Separate out pmdoc/pman from mdoc/man: the first persists
 856          * through all parsers, while the latter is used per-parse.
 857          */
 858
 859         if ('.' == buf[0] || '\'' == buf[0]) {
 860                 for (i = 1; buf[i]; i++)
 861                         if (' ' != buf[i] && '\t' != buf[i])
 862                                 break;
 863                 if ('\0' == buf[i])
 864                         return;
 865         }
 866
 867         switch (curp->inttype) {
 868         case (INTT_MDOC):
 869                 if (NULL == curp->pmdoc)
 870                         curp->pmdoc = mdoc_alloc
 871                                 (&curp->regs, curp, mmsg);
 872                 assert(curp->pmdoc);
 873                 curp->mdoc = curp->pmdoc;
 874                 return;
 875         case (INTT_MAN):
 876                 if (NULL == curp->pman)
 877                         curp->pman = man_alloc
 878                                 (&curp->regs, curp, mmsg);
 879                 assert(curp->pman);
 880                 curp->man = curp->pman;
 881                 return;
 882         default:
 883                 break;
 884         }
 885
 886         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 887                 if (NULL == curp->pmdoc)
 888                         curp->pmdoc = mdoc_alloc
 889                                 (&curp->regs, curp, mmsg);
 890                 assert(curp->pmdoc);
 891                 curp->mdoc = curp->pmdoc;
 892                 return;
 893         }
 894
 895         if (NULL == curp->pman)
 896                 curp->pman = man_alloc(&curp->regs, curp, mmsg);
 897         assert(curp->pman);
 898         curp->man = curp->pman;
 899 }
 900
 901 static int
 902 moptions(enum intt *tflags, char *arg)
 903 {
 904
 905         if (0 == strcmp(arg, "doc"))
 906                 *tflags = INTT_MDOC;
 907         else if (0 == strcmp(arg, "andoc"))
 908                 *tflags = INTT_AUTO;
 909         else if (0 == strcmp(arg, "an"))
 910                 *tflags = INTT_MAN;
 911         else {
 912                 fprintf(stderr, "%s: Bad argument\n", arg);
 913                 return(0);
 914         }
 915
 916         return(1);
 917 }
 918
 919 static int
 920 toptions(struct curparse *curp, char *arg)
 921 {
 922
 923         if (0 == strcmp(arg, "ascii"))
 924                 curp->outtype = OUTT_ASCII;
 925         else if (0 == strcmp(arg, "lint")) {
 926                 curp->outtype = OUTT_LINT;
 927                 curp->wlevel  = MANDOCLEVEL_WARNING;
 928         }
 929         else if (0 == strcmp(arg, "tree"))
 930                 curp->outtype = OUTT_TREE;
 931         else if (0 == strcmp(arg, "html"))
 932                 curp->outtype = OUTT_HTML;
 933         else if (0 == strcmp(arg, "xhtml"))
 934                 curp->outtype = OUTT_XHTML;
 935         else if (0 == strcmp(arg, "ps"))
 936                 curp->outtype = OUTT_PS;
 937         else if (0 == strcmp(arg, "pdf"))
 938                 curp->outtype = OUTT_PDF;
 939         else {
 940                 fprintf(stderr, "%s: Bad argument\n", arg);
 941                 return(0);
 942         }
 943
 944         return(1);
 945 }
 946
 947 static int
 948 woptions(struct curparse *curp, char *arg)
 949 {
 950         char            *v, *o;
 951         const char      *toks[6];
 952
 953         toks[0] = "stop";
 954         toks[1] = "all";
 955         toks[2] = "warning";
 956         toks[3] = "error";
 957         toks[4] = "fatal";
 958         toks[5] = NULL;
 959
 960         while (*arg) {
 961                 o = arg;
 962                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 963                 case (0):
 964                         curp->wstop = 1;
 965                         break;
 966                 case (1):
 967                         /* FALLTHROUGH */
 968                 case (2):
 969                         curp->wlevel = MANDOCLEVEL_WARNING;
 970                         break;
 971                 case (3):
 972                         curp->wlevel = MANDOCLEVEL_ERROR;
 973                         break;
 974                 case (4):
 975                         curp->wlevel = MANDOCLEVEL_FATAL;
 976                         break;
 977                 default:
 978                         fprintf(stderr, "-W%s: Bad argument\n", o);
 979                         return(0);
 980                 }
 981         }
 982
 983         return(1);
 984 }
 985
 986 static int
 987 mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
 988 {
 989         struct curparse *cp;
 990         enum mandoclevel level;
 991
 992         level = MANDOCLEVEL_FATAL;
 993         while (t < mandoclimits[level])
 994                 /* LINTED */
 995                 level--;
 996
 997         cp = (struct curparse *)arg;
 998         if (level < cp->wlevel)
 999                 return(1);
1000
1001         fprintf(stderr, "%s:%d:%d: %s: %s",
1002             cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1003         if (msg)
1004                 fprintf(stderr, ": %s", msg);
1005         fputc('\n', stderr);
1006
1007         if (file_status < level)
1008                 file_status = level;
1009
1010         return(level < MANDOCLEVEL_FATAL);
1011 }