usr.bin/mandoc/main.c

   1 /*      $Id: main.c,v 1.57 2009/11/02 08:29:25 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 #include <sys/stat.h>
  18
  19 #include <assert.h>
  20 #include <fcntl.h>
  21 #include <stdio.h>
  22 #include <stdint.h>
  23 #include <stdlib.h>
  24 #include <string.h>
  25 #include <unistd.h>
  26
  27 #include "mdoc.h"
  28 #include "man.h"
  29 #include "main.h"
  30
  31 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  32
  33 /* FIXME: Intel's compiler?  LLVM?  pcc?  */
  34
  35 #if !defined(__GNUC__) || (__GNUC__ < 2)
  36 # if !defined(lint)
  37 #  define __attribute__(x)
  38 # endif
  39 #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */
  40
  41 #ifdef __linux__
  42 extern  int               getsubopt(char **, char * const *, char **);
  43 extern  size_t            strlcat(char *, const char *, size_t);
  44 #endif
  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_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 IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
  76 #define NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
  77 #define NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
  78 #define NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */
  79 #define 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 static  int               foptions(int *, char *);
  94 static  int               toptions(enum outt *, char *);
  95 static  int               moptions(enum intt *, char *);
  96 static  int               woptions(int *, char *);
  97 static  int               merr(void *, int, int, const char *);
  98 static  int               mwarn(void *, int, int, const char *);
  99 static  int               ffile(struct buf *, struct buf *,
 100                                 const char *, struct curparse *);
 101 static  int               fdesc(struct buf *, struct buf *,
 102                                 struct curparse *);
 103 static  int               pset(const char *, int, struct curparse *,
 104                                 struct man **, struct mdoc **);
 105 static  struct man       *man_init(struct curparse *);
 106 static  struct mdoc      *mdoc_init(struct curparse *);
 107 static  void              version(void) __attribute__((noreturn));
 108 static  void              usage(void) __attribute__((noreturn));
 109
 110 static  const char       *progname;
 111
 112
 113 int
 114 main(int argc, char *argv[])
 115 {
 116         int              c, rc;
 117         struct buf       ln, blk;
 118         struct curparse  curp;
 119
 120         progname = strrchr(argv[0], '/');
 121         if (progname == NULL)
 122                 progname = argv[0];
 123         else
 124                 ++progname;
 125
 126         memset(&curp, 0, sizeof(struct curparse));
 127
 128         curp.inttype = INTT_AUTO;
 129         curp.outtype = OUTT_ASCII;
 130
 131         /* LINTED */
 132         while (-1 != (c = getopt(argc, argv, "f:m:O:T:VW:")))
 133                 switch (c) {
 134                 case ('f'):
 135                         if ( ! foptions(&curp.fflags, optarg))
 136                                 return(EXIT_FAILURE);
 137                         break;
 138                 case ('m'):
 139                         if ( ! moptions(&curp.inttype, optarg))
 140                                 return(EXIT_FAILURE);
 141                         break;
 142                 case ('O'):
 143                         (void)strlcat(curp.outopts, optarg, BUFSIZ);
 144                         (void)strlcat(curp.outopts, ",", BUFSIZ);
 145                         break;
 146                 case ('T'):
 147                         if ( ! toptions(&curp.outtype, optarg))
 148                                 return(EXIT_FAILURE);
 149                         break;
 150                 case ('W'):
 151                         if ( ! woptions(&curp.wflags, optarg))
 152                                 return(EXIT_FAILURE);
 153                         break;
 154                 case ('V'):
 155                         version();
 156                         /* NOTREACHED */
 157                 default:
 158                         usage();
 159                         /* NOTREACHED */
 160                 }
 161
 162         argc -= optind;
 163         argv += optind;
 164
 165         memset(&ln, 0, sizeof(struct buf));
 166         memset(&blk, 0, sizeof(struct buf));
 167
 168         rc = 1;
 169
 170         if (NULL == *argv) {
 171                 curp.file = "<stdin>";
 172                 curp.fd = STDIN_FILENO;
 173
 174                 c = fdesc(&blk, &ln, &curp);
 175                 if ( ! (IGN_ERRORS & curp.fflags))
 176                         rc = 1 == c ? 1 : 0;
 177                 else
 178                         rc = -1 == c ? 0 : 1;
 179         }
 180
 181         while (rc && *argv) {
 182                 c = ffile(&blk, &ln, *argv, &curp);
 183                 if ( ! (IGN_ERRORS & curp.fflags))
 184                         rc = 1 == c ? 1 : 0;
 185                 else
 186                         rc = -1 == c ? 0 : 1;
 187
 188                 argv++;
 189                 if (*argv && rc) {
 190                         if (curp.lastman)
 191                                 man_reset(curp.lastman);
 192                         if (curp.lastmdoc)
 193                                 mdoc_reset(curp.lastmdoc);
 194                         curp.lastman = NULL;
 195                         curp.lastmdoc = NULL;
 196                 }
 197         }
 198
 199         if (blk.buf)
 200                 free(blk.buf);
 201         if (ln.buf)
 202                 free(ln.buf);
 203         if (curp.outfree)
 204                 (*curp.outfree)(curp.outdata);
 205         if (curp.mdoc)
 206                 mdoc_free(curp.mdoc);
 207         if (curp.man)
 208                 man_free(curp.man);
 209
 210         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 211 }
 212
 213
 214 static void
 215 version(void)
 216 {
 217
 218         (void)printf("%s %s\n", progname, VERSION);
 219         exit(EXIT_SUCCESS);
 220 }
 221
 222
 223 static void
 224 usage(void)
 225 {
 226
 227         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 228                         "[-mformat] [-Ooption] [-Toutput] "
 229                         "[-Werr...]\n", progname);
 230         exit(EXIT_FAILURE);
 231 }
 232
 233
 234 static struct man *
 235 man_init(struct curparse *curp)
 236 {
 237         int              pflags;
 238         struct man_cb    mancb;
 239
 240         mancb.man_err = merr;
 241         mancb.man_warn = mwarn;
 242
 243         /* Defaults from mandoc.1. */
 244
 245         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS;
 246
 247         if (curp->fflags & NO_IGN_MACRO)
 248                 pflags &= ~MAN_IGN_MACRO;
 249         if (curp->fflags & NO_IGN_CHARS)
 250                 pflags &= ~MAN_IGN_CHARS;
 251         if (curp->fflags & NO_IGN_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 | MDOC_IGN_CHARS;
 270
 271         if (curp->fflags & IGN_SCOPE)
 272                 pflags |= MDOC_IGN_SCOPE;
 273         if (curp->fflags & NO_IGN_ESCAPE)
 274                 pflags &= ~MDOC_IGN_ESCAPE;
 275         if (curp->fflags & NO_IGN_MACRO)
 276                 pflags &= ~MDOC_IGN_MACRO;
 277         if (curp->fflags & NO_IGN_CHARS)
 278                 pflags &= ~MDOC_IGN_CHARS;
 279
 280         return(mdoc_alloc(curp, pflags, &mdoccb));
 281 }
 282
 283
 284 static int
 285 ffile(struct buf *blk, struct buf *ln,
 286                 const char *file, struct curparse *curp)
 287 {
 288         int              c;
 289
 290         curp->file = file;
 291         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 292                 perror(curp->file);
 293                 return(-1);
 294         }
 295
 296         c = fdesc(blk, ln, curp);
 297
 298         if (-1 == close(curp->fd))
 299                 perror(curp->file);
 300
 301         return(c);
 302 }
 303
 304
 305 static int
 306 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 307 {
 308         size_t           sz;
 309         ssize_t          ssz;
 310         struct stat      st;
 311         int              j, i, pos, lnn, comment;
 312         struct man      *man;
 313         struct mdoc     *mdoc;
 314
 315         sz = BUFSIZ;
 316         man = NULL;
 317         mdoc = NULL;
 318
 319         /*
 320          * Two buffers: ln and buf.  buf is the input buffer optimised
 321          * here for each file's block size.  ln is a line buffer.  Both
 322          * growable, hence passed in by ptr-ptr.
 323          */
 324
 325         if (-1 == fstat(curp->fd, &st))
 326                 perror(curp->file);
 327         else if ((size_t)st.st_blksize > sz)
 328                 sz = st.st_blksize;
 329
 330         if (sz > blk->sz) {
 331                 blk->buf = realloc(blk->buf, sz);
 332                 if (NULL == blk->buf) {
 333                         perror(NULL);
 334                         exit(EXIT_FAILURE);
 335                 }
 336                 blk->sz = sz;
 337         }
 338
 339         /* Fill buf with file blocksize. */
 340
 341         for (lnn = pos = comment = 0; ; ) {
 342                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 343                         perror(curp->file);
 344                         return(-1);
 345                 } else if (0 == ssz)
 346                         break;
 347
 348                 /* Parse the read block into partial or full lines. */
 349
 350                 for (i = 0; i < (int)ssz; i++) {
 351                         if (pos >= (int)ln->sz) {
 352                                 ln->sz += 256; /* Step-size. */
 353                                 ln->buf = realloc(ln->buf, ln->sz);
 354                                 if (NULL == ln->buf) {
 355                                         perror(NULL);
 356                                         return(EXIT_FAILURE);
 357                                 }
 358                         }
 359
 360                         if ('\n' != blk->buf[i]) {
 361                                 if (comment)
 362                                         continue;
 363                                 ln->buf[pos++] = blk->buf[i];
 364
 365                                 /* Handle in-line `\"' comments. */
 366
 367                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 368                                         continue;
 369
 370                                 for (j = pos - 2; j >= 0; j--)
 371                                         if ('\\' != ln->buf[j])
 372                                                 break;
 373
 374                                 if ( ! ((pos - 2 - j) % 2))
 375                                         continue;
 376
 377                                 comment = 1;
 378                                 pos -= 2;
 379                                 continue;
 380                         }
 381
 382                         /* Handle escaped `\\n' newlines. */
 383
 384                         if (pos > 0 && 0 == comment &&
 385                                         '\\' == ln->buf[pos - 1]) {
 386                                 for (j = pos - 1; j >= 0; j--)
 387                                         if ('\\' != ln->buf[j])
 388                                                 break;
 389                                 if ( ! ((pos - j) % 2)) {
 390                                         pos--;
 391                                         lnn++;
 392                                         continue;
 393                                 }
 394                         }
 395
 396                         ln->buf[pos] = 0;
 397                         lnn++;
 398
 399                         /* If unset, assign parser in pset(). */
 400
 401                         if ( ! (man || mdoc) && ! pset(ln->buf,
 402                                                 pos, curp, &man, &mdoc))
 403                                 return(-1);
 404
 405                         pos = comment = 0;
 406
 407                         /* Pass down into parsers. */
 408
 409                         if (man && ! man_parseln(man, lnn, ln->buf))
 410                                 return(0);
 411                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 412                                 return(0);
 413                 }
 414         }
 415
 416         /* NOTE a parser may not have been assigned, yet. */
 417
 418         if ( ! (man || mdoc)) {
 419                 fprintf(stderr, "%s: Not a manual\n", curp->file);
 420                 return(0);
 421         }
 422
 423         if (mdoc && ! mdoc_endparse(mdoc))
 424                 return(0);
 425         if (man && ! man_endparse(man))
 426                 return(0);
 427
 428         /* If unset, allocate output dev now (if applicable). */
 429
 430         if ( ! (curp->outman && curp->outmdoc)) {
 431                 switch (curp->outtype) {
 432                 case (OUTT_HTML):
 433                         curp->outdata = html_alloc(curp->outopts);
 434                         curp->outman = html_man;
 435                         curp->outmdoc = html_mdoc;
 436                         curp->outfree = html_free;
 437                         break;
 438                 case (OUTT_TREE):
 439                         curp->outman = tree_man;
 440                         curp->outmdoc = tree_mdoc;
 441                         break;
 442                 case (OUTT_LINT):
 443                         break;
 444                 default:
 445                         curp->outdata = ascii_alloc();
 446                         curp->outman = terminal_man;
 447                         curp->outmdoc = terminal_mdoc;
 448                         curp->outfree = terminal_free;
 449                         break;
 450                 }
 451         }
 452
 453         /* Execute the out device, if it exists. */
 454
 455         if (man && curp->outman)
 456                 (*curp->outman)(curp->outdata, man);
 457         if (mdoc && curp->outmdoc)
 458                 (*curp->outmdoc)(curp->outdata, mdoc);
 459
 460         return(1);
 461 }
 462
 463
 464 static int
 465 pset(const char *buf, int pos, struct curparse *curp,
 466                 struct man **man, struct mdoc **mdoc)
 467 {
 468         int              i;
 469
 470         /*
 471          * Try to intuit which kind of manual parser should be used.  If
 472          * passed in by command-line (-man, -mdoc), then use that
 473          * explicitly.  If passed as -mandoc, then try to guess from the
 474          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 475          * default to -man, which is more lenient.
 476          */
 477
 478         if (buf[0] == '.') {
 479                 for (i = 1; buf[i]; i++)
 480                         if (' ' != buf[i] && '\t' != buf[i])
 481                                 break;
 482                 if (0 == buf[i])
 483                         return(1);
 484         }
 485
 486         switch (curp->inttype) {
 487         case (INTT_MDOC):
 488                 if (NULL == curp->mdoc)
 489                         curp->mdoc = mdoc_init(curp);
 490                 if (NULL == (*mdoc = curp->mdoc))
 491                         return(0);
 492                 curp->lastmdoc = *mdoc;
 493                 return(1);
 494         case (INTT_MAN):
 495                 if (NULL == curp->man)
 496                         curp->man = man_init(curp);
 497                 if (NULL == (*man = curp->man))
 498                         return(0);
 499                 curp->lastman = *man;
 500                 return(1);
 501         default:
 502                 break;
 503         }
 504
 505         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 506                 if (NULL == curp->mdoc)
 507                         curp->mdoc = mdoc_init(curp);
 508                 if (NULL == (*mdoc = curp->mdoc))
 509                         return(0);
 510                 curp->lastmdoc = *mdoc;
 511                 return(1);
 512         }
 513
 514         if (NULL == curp->man)
 515                 curp->man = man_init(curp);
 516         if (NULL == (*man = curp->man))
 517                 return(0);
 518         curp->lastman = *man;
 519         return(1);
 520 }
 521
 522
 523 static int
 524 moptions(enum intt *tflags, char *arg)
 525 {
 526
 527         if (0 == strcmp(arg, "doc"))
 528                 *tflags = INTT_MDOC;
 529         else if (0 == strcmp(arg, "andoc"))
 530                 *tflags = INTT_AUTO;
 531         else if (0 == strcmp(arg, "an"))
 532                 *tflags = INTT_MAN;
 533         else {
 534                 fprintf(stderr, "%s: Bad argument\n", arg);
 535                 return(0);
 536         }
 537
 538         return(1);
 539 }
 540
 541
 542 static int
 543 toptions(enum outt *tflags, char *arg)
 544 {
 545
 546         if (0 == strcmp(arg, "ascii"))
 547                 *tflags = OUTT_ASCII;
 548         else if (0 == strcmp(arg, "lint"))
 549                 *tflags = OUTT_LINT;
 550         else if (0 == strcmp(arg, "tree"))
 551                 *tflags = OUTT_TREE;
 552         else if (0 == strcmp(arg, "html"))
 553                 *tflags = OUTT_HTML;
 554         else {
 555                 fprintf(stderr, "%s: Bad argument\n", arg);
 556                 return(0);
 557         }
 558
 559         return(1);
 560 }
 561
 562
 563 static int
 564 foptions(int *fflags, char *arg)
 565 {
 566         char            *v, *o;
 567         const char      *toks[8];
 568
 569         toks[0] = "ign-scope";
 570         toks[1] = "no-ign-escape";
 571         toks[2] = "no-ign-macro";
 572         toks[3] = "no-ign-chars";
 573         toks[4] = "ign-errors";
 574         toks[5] = "strict";
 575         toks[6] = "ign-escape";
 576         toks[7] = NULL;
 577
 578         while (*arg) {
 579                 o = arg;
 580                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 581                 case (0):
 582                         *fflags |= IGN_SCOPE;
 583                         break;
 584                 case (1):
 585                         *fflags |= NO_IGN_ESCAPE;
 586                         break;
 587                 case (2):
 588                         *fflags |= NO_IGN_MACRO;
 589                         break;
 590                 case (3):
 591                         *fflags |= NO_IGN_CHARS;
 592                         break;
 593                 case (4):
 594                         *fflags |= IGN_ERRORS;
 595                         break;
 596                 case (5):
 597                         *fflags |= NO_IGN_ESCAPE |
 598                                    NO_IGN_MACRO | NO_IGN_CHARS;
 599                         break;
 600                 case (6):
 601                         *fflags &= ~NO_IGN_ESCAPE;
 602                         break;
 603                 default:
 604                         fprintf(stderr, "%s: Bad argument\n", o);
 605                         return(0);
 606                 }
 607         }
 608
 609         return(1);
 610 }
 611
 612
 613 static int
 614 woptions(int *wflags, char *arg)
 615 {
 616         char            *v, *o;
 617         const char      *toks[3];
 618
 619         toks[0] = "all";
 620         toks[1] = "error";
 621         toks[2] = NULL;
 622
 623         while (*arg) {
 624                 o = arg;
 625                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 626                 case (0):
 627                         *wflags |= WARN_WALL;
 628                         break;
 629                 case (1):
 630                         *wflags |= WARN_WERR;
 631                         break;
 632                 default:
 633                         fprintf(stderr, "%s: Bad argument\n", o);
 634                         return(0);
 635                 }
 636         }
 637
 638         return(1);
 639 }
 640
 641
 642 /* ARGSUSED */
 643 static int
 644 merr(void *arg, int line, int col, const char *msg)
 645 {
 646         struct curparse *curp;
 647
 648         curp = (struct curparse *)arg;
 649
 650         (void)fprintf(stderr, "%s:%d:%d: error: %s\n",
 651                         curp->file, line, col + 1, msg);
 652
 653         return(0);
 654 }
 655
 656
 657 static int
 658 mwarn(void *arg, int line, int col, const char *msg)
 659 {
 660         struct curparse *curp;
 661
 662         curp = (struct curparse *)arg;
 663
 664         if ( ! (curp->wflags & WARN_WALL))
 665                 return(1);
 666
 667         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 668                         curp->file, line, col + 1, msg);
 669
 670         if ( ! (curp->wflags & WARN_WERR))
 671                 return(1);
 672
 673         return(0);
 674 }