main.c

   1 /*      $Id: main.c,v 1.44 2009/09/21 13:06:13 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 <err.h>
  21 #include <fcntl.h>
  22 #include <stdio.h>
  23 #include <stdlib.h>
  24 #include <string.h>
  25 #include <unistd.h>
  26
  27 #include "mdoc.h"
  28 #include "man.h"
  29
  30 /* Account for FreeBSD and Linux in our declarations. */
  31
  32 #ifdef __linux__
  33 extern  int               getsubopt(char **, char * const *, char **);
  34 # ifndef __dead
  35 #  define __dead __attribute__((__noreturn__))
  36 # endif
  37 #elif defined(__dead2)
  38 # ifndef __dead
  39 #  define __dead __dead2
  40 # endif
  41 #endif
  42
  43 typedef void            (*out_mdoc)(void *, const struct mdoc *);
  44 typedef void            (*out_man)(void *, const struct man *);
  45 typedef void            (*out_free)(void *);
  46
  47 struct  buf {
  48         char             *buf;
  49         size_t            sz;
  50 };
  51
  52 enum    intt {
  53         INTT_AUTO,
  54         INTT_MDOC,
  55         INTT_MAN
  56 };
  57
  58 enum    outt {
  59         OUTT_ASCII = 0,
  60         OUTT_TREE,
  61         OUTT_HTML,
  62         OUTT_LINT
  63 };
  64
  65 struct  curparse {
  66         const char       *file;         /* Current parse. */
  67         int               fd;           /* Current parse. */
  68         int               wflags;
  69 #define WARN_WALL        (1 << 0)       /* All-warnings mask. */
  70 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  71         int               fflags;
  72 #define IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
  73 #define NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
  74 #define NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
  75 #define NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */
  76 #define IGN_ERRORS       (1 << 4)       /* Ignore failed parse. */
  77         enum intt         inttype;      /* Input parsers... */
  78         struct man       *man;
  79         struct man       *lastman;
  80         struct mdoc      *mdoc;
  81         struct mdoc      *lastmdoc;
  82         enum outt         outtype;      /* Output devices... */
  83         out_mdoc          outmdoc;
  84         out_man           outman;
  85         out_free          outfree;
  86         void             *outdata;
  87         char             *outopts;
  88 };
  89
  90 extern  void             *html_alloc(char *);
  91 extern  void              html_mdoc(void *, const struct mdoc *);
  92 extern  void              html_man(void *, const struct man *);
  93 extern  void              html_free(void *);
  94 extern  void             *ascii_alloc(void);
  95 extern  void              tree_mdoc(void *, const struct mdoc *);
  96 extern  void              tree_man(void *, const struct man *);
  97 extern  void              terminal_mdoc(void *, const struct mdoc *);
  98 extern  void              terminal_man(void *, const struct man *);
  99 extern  void              terminal_free(void *);
 100
 101 static  int               foptions(int *, char *);
 102 static  int               toptions(enum outt *, char *);
 103 static  int               moptions(enum intt *, char *);
 104 static  int               woptions(int *, char *);
 105 static  int               merr(void *, int, int, const char *);
 106 static  int               mwarn(void *, int, int, const char *);
 107 static  int               ffile(struct buf *, struct buf *,
 108                                 const char *, struct curparse *);
 109 static  int               fdesc(struct buf *, struct buf *,
 110                                 struct curparse *);
 111 static  int               pset(const char *, int, struct curparse *,
 112                                 struct man **, struct mdoc **);
 113 static  struct man       *man_init(struct curparse *);
 114 static  struct mdoc      *mdoc_init(struct curparse *);
 115 __dead  static void       version(void);
 116 __dead  static void       usage(void);
 117
 118 extern  char             *__progname;
 119
 120
 121 int
 122 main(int argc, char *argv[])
 123 {
 124         int              c, rc;
 125         struct buf       ln, blk;
 126         struct curparse  curp;
 127
 128         bzero(&curp, 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                         curp.outopts = optarg;
 146                         break;
 147                 case ('T'):
 148                         if ( ! toptions(&curp.outtype, optarg))
 149                                 return(EXIT_FAILURE);
 150                         break;
 151                 case ('W'):
 152                         if ( ! woptions(&curp.wflags, optarg))
 153                                 return(EXIT_FAILURE);
 154                         break;
 155                 case ('V'):
 156                         version();
 157                         /* NOTREACHED */
 158                 default:
 159                         usage();
 160                         /* NOTREACHED */
 161                 }
 162
 163         argc -= optind;
 164         argv += optind;
 165
 166         bzero(&ln, sizeof(struct buf));
 167         bzero(&blk, sizeof(struct buf));
 168
 169         rc = 1;
 170
 171         if (NULL == *argv) {
 172                 curp.file = "<stdin>";
 173                 curp.fd = STDIN_FILENO;
 174
 175                 c = fdesc(&blk, &ln, &curp);
 176                 if ( ! (IGN_ERRORS & curp.fflags))
 177                         rc = 1 == c ? 1 : 0;
 178                 else
 179                         rc = -1 == c ? 0 : 1;
 180         }
 181
 182         while (rc && *argv) {
 183                 c = ffile(&blk, &ln, *argv, &curp);
 184                 if ( ! (IGN_ERRORS & curp.fflags))
 185                         rc = 1 == c ? 1 : 0;
 186                 else
 187                         rc = -1 == c ? 0 : 1;
 188
 189                 argv++;
 190                 if (*argv && rc) {
 191                         if (curp.lastman)
 192                                 if ( ! man_reset(curp.lastman))
 193                                         rc = 0;
 194                         if (curp.lastmdoc)
 195                                 if ( ! mdoc_reset(curp.lastmdoc))
 196                                         rc = 0;
 197                         curp.lastman = NULL;
 198                         curp.lastmdoc = NULL;
 199                 }
 200         }
 201
 202         if (blk.buf)
 203                 free(blk.buf);
 204         if (ln.buf)
 205                 free(ln.buf);
 206         if (curp.outfree)
 207                 (*curp.outfree)(curp.outdata);
 208         if (curp.mdoc)
 209                 mdoc_free(curp.mdoc);
 210         if (curp.man)
 211                 man_free(curp.man);
 212
 213         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 214 }
 215
 216
 217 __dead static void
 218 version(void)
 219 {
 220
 221         (void)printf("%s %s\n", __progname, VERSION);
 222         exit(EXIT_SUCCESS);
 223 }
 224
 225
 226 __dead static void
 227 usage(void)
 228 {
 229
 230         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 231                         "[-mformat] [-Toutput] [-Werr...]\n",
 232                         __progname);
 233         exit(EXIT_FAILURE);
 234 }
 235
 236
 237 static struct man *
 238 man_init(struct curparse *curp)
 239 {
 240         int              pflags;
 241         struct man      *man;
 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 & NO_IGN_MACRO)
 252                 pflags &= ~MAN_IGN_MACRO;
 253         if (curp->fflags & NO_IGN_CHARS)
 254                 pflags &= ~MAN_IGN_CHARS;
 255         if (curp->fflags & NO_IGN_ESCAPE)
 256                 pflags &= ~MAN_IGN_ESCAPE;
 257
 258         if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 259                 warnx("memory exhausted");
 260
 261         return(man);
 262 }
 263
 264
 265 static struct mdoc *
 266 mdoc_init(struct curparse *curp)
 267 {
 268         int              pflags;
 269         struct mdoc     *mdoc;
 270         struct mdoc_cb   mdoccb;
 271
 272         mdoccb.mdoc_err = merr;
 273         mdoccb.mdoc_warn = mwarn;
 274
 275         /* Defaults from mandoc.1. */
 276
 277         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 278
 279         if (curp->fflags & IGN_SCOPE)
 280                 pflags |= MDOC_IGN_SCOPE;
 281         if (curp->fflags & NO_IGN_ESCAPE)
 282                 pflags &= ~MDOC_IGN_ESCAPE;
 283         if (curp->fflags & NO_IGN_MACRO)
 284                 pflags &= ~MDOC_IGN_MACRO;
 285         if (curp->fflags & NO_IGN_CHARS)
 286                 pflags &= ~MDOC_IGN_CHARS;
 287
 288         if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb)))
 289                 warnx("memory exhausted");
 290
 291         return(mdoc);
 292 }
 293
 294
 295 static int
 296 ffile(struct buf *blk, struct buf *ln,
 297                 const char *file, struct curparse *curp)
 298 {
 299         int              c;
 300
 301         curp->file = file;
 302         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 303                 warn("%s", curp->file);
 304                 return(-1);
 305         }
 306
 307         c = fdesc(blk, ln, curp);
 308
 309         if (-1 == close(curp->fd))
 310                 warn("%s", curp->file);
 311
 312         return(c);
 313 }
 314
 315
 316 static int
 317 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 318 {
 319         size_t           sz;
 320         ssize_t          ssz;
 321         struct stat      st;
 322         int              j, i, pos, lnn, comment;
 323         struct man      *man;
 324         struct mdoc     *mdoc;
 325
 326         sz = BUFSIZ;
 327         man = NULL;
 328         mdoc = NULL;
 329
 330         /*
 331          * Two buffers: ln and buf.  buf is the input buffer optimised
 332          * here for each file's block size.  ln is a line buffer.  Both
 333          * growable, hence passed in by ptr-ptr.
 334          */
 335
 336         if (-1 == fstat(curp->fd, &st))
 337                 warn("%s", curp->file);
 338         else if ((size_t)st.st_blksize > sz)
 339                 sz = st.st_blksize;
 340
 341         if (sz > blk->sz) {
 342                 blk->buf = realloc(blk->buf, sz);
 343                 if (NULL == blk->buf) {
 344                         warn("realloc");
 345                         return(-1);
 346                 }
 347                 blk->sz = sz;
 348         }
 349
 350         /* Fill buf with file blocksize. */
 351
 352         for (lnn = pos = comment = 0; ; ) {
 353                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 354                         warn("%s", curp->file);
 355                         return(-1);
 356                 } else if (0 == ssz)
 357                         break;
 358
 359                 /* Parse the read block into partial or full lines. */
 360
 361                 for (i = 0; i < (int)ssz; i++) {
 362                         if (pos >= (int)ln->sz) {
 363                                 ln->sz += 256; /* Step-size. */
 364                                 ln->buf = realloc(ln->buf, ln->sz);
 365                                 if (NULL == ln->buf) {
 366                                         warn("realloc");
 367                                         return(-1);
 368                                 }
 369                         }
 370
 371                         if ('\n' != blk->buf[i]) {
 372                                 if (comment)
 373                                         continue;
 374                                 ln->buf[pos++] = blk->buf[i];
 375
 376                                 /* Handle in-line `\"' comments. */
 377
 378                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 379                                         continue;
 380
 381                                 for (j = pos - 2; j >= 0; j--)
 382                                         if ('\\' != ln->buf[j])
 383                                                 break;
 384
 385                                 if ( ! ((pos - 2 - j) % 2))
 386                                         continue;
 387
 388                                 comment = 1;
 389                                 pos -= 2;
 390                                 continue;
 391                         }
 392
 393                         /* Handle escaped `\\n' newlines. */
 394
 395                         if (pos > 0 && 0 == comment &&
 396                                         '\\' == ln->buf[pos - 1]) {
 397                                 for (j = pos - 1; j >= 0; j--)
 398                                         if ('\\' != ln->buf[j])
 399                                                 break;
 400                                 if ( ! ((pos - j) % 2)) {
 401                                         pos--;
 402                                         lnn++;
 403                                         continue;
 404                                 }
 405                         }
 406
 407                         ln->buf[pos] = 0;
 408                         lnn++;
 409
 410                         /* If unset, assign parser in pset(). */
 411
 412                         if ( ! (man || mdoc) && ! pset(ln->buf,
 413                                                 pos, curp, &man, &mdoc))
 414                                 return(-1);
 415
 416                         pos = comment = 0;
 417
 418                         /* Pass down into parsers. */
 419
 420                         if (man && ! man_parseln(man, lnn, ln->buf))
 421                                 return(0);
 422                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 423                                 return(0);
 424                 }
 425         }
 426
 427         /* NOTE a parser may not have been assigned, yet. */
 428
 429         if ( ! (man || mdoc)) {
 430                 (void)fprintf(stderr, "%s: not a manual\n",
 431                                 curp->file);
 432                 return(0);
 433         }
 434
 435         if (mdoc && ! mdoc_endparse(mdoc))
 436                 return(0);
 437         if (man && ! man_endparse(man))
 438                 return(0);
 439
 440         /* If unset, allocate output dev now (if applicable). */
 441
 442         if ( ! (curp->outman && curp->outmdoc)) {
 443                 switch (curp->outtype) {
 444                 case (OUTT_HTML):
 445                         curp->outdata = html_alloc(curp->outopts);
 446                         curp->outman = html_man;
 447                         curp->outmdoc = html_mdoc;
 448                         curp->outfree = html_free;
 449                         break;
 450                 case (OUTT_TREE):
 451                         curp->outman = tree_man;
 452                         curp->outmdoc = tree_mdoc;
 453                         break;
 454                 case (OUTT_LINT):
 455                         break;
 456                 default:
 457                         curp->outdata = ascii_alloc();
 458                         curp->outman = terminal_man;
 459                         curp->outmdoc = terminal_mdoc;
 460                         curp->outfree = terminal_free;
 461                         break;
 462                 }
 463         }
 464
 465         /* Execute the out device, if it exists. */
 466
 467         if (man && curp->outman)
 468                 (*curp->outman)(curp->outdata, man);
 469         if (mdoc && curp->outmdoc)
 470                 (*curp->outmdoc)(curp->outdata, mdoc);
 471
 472         return(1);
 473 }
 474
 475
 476 static int
 477 pset(const char *buf, int pos, struct curparse *curp,
 478                 struct man **man, struct mdoc **mdoc)
 479 {
 480         int              i;
 481
 482         /*
 483          * Try to intuit which kind of manual parser should be used.  If
 484          * passed in by command-line (-man, -mdoc), then use that
 485          * explicitly.  If passed as -mandoc, then try to guess from the
 486          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 487          * default to -man, which is more lenient.
 488          */
 489
 490         if (buf[0] == '.') {
 491                 for (i = 1; buf[i]; i++)
 492                         if (' ' != buf[i] && '\t' != buf[i])
 493                                 break;
 494                 if (0 == buf[i])
 495                         return(1);
 496         }
 497
 498         switch (curp->inttype) {
 499         case (INTT_MDOC):
 500                 if (NULL == curp->mdoc)
 501                         curp->mdoc = mdoc_init(curp);
 502                 if (NULL == (*mdoc = curp->mdoc))
 503                         return(0);
 504                 curp->lastmdoc = *mdoc;
 505                 return(1);
 506         case (INTT_MAN):
 507                 if (NULL == curp->man)
 508                         curp->man = man_init(curp);
 509                 if (NULL == (*man = curp->man))
 510                         return(0);
 511                 curp->lastman = *man;
 512                 return(1);
 513         default:
 514                 break;
 515         }
 516
 517         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 518                 if (NULL == curp->mdoc)
 519                         curp->mdoc = mdoc_init(curp);
 520                 if (NULL == (*mdoc = curp->mdoc))
 521                         return(0);
 522                 curp->lastmdoc = *mdoc;
 523                 return(1);
 524         }
 525
 526         if (NULL == curp->man)
 527                 curp->man = man_init(curp);
 528         if (NULL == (*man = curp->man))
 529                 return(0);
 530         curp->lastman = *man;
 531         return(1);
 532 }
 533
 534
 535 static int
 536 moptions(enum intt *tflags, char *arg)
 537 {
 538
 539         if (0 == strcmp(arg, "doc"))
 540                 *tflags = INTT_MDOC;
 541         else if (0 == strcmp(arg, "andoc"))
 542                 *tflags = INTT_AUTO;
 543         else if (0 == strcmp(arg, "an"))
 544                 *tflags = INTT_MAN;
 545         else {
 546                 warnx("bad argument: -m%s", arg);
 547                 return(0);
 548         }
 549
 550         return(1);
 551 }
 552
 553
 554 static int
 555 toptions(enum outt *tflags, char *arg)
 556 {
 557
 558         if (0 == strcmp(arg, "ascii"))
 559                 *tflags = OUTT_ASCII;
 560         else if (0 == strcmp(arg, "lint"))
 561                 *tflags = OUTT_LINT;
 562         else if (0 == strcmp(arg, "tree"))
 563                 *tflags = OUTT_TREE;
 564         else if (0 == strcmp(arg, "html"))
 565                 *tflags = OUTT_HTML;
 566         else {
 567                 warnx("bad argument: -T%s", arg);
 568                 return(0);
 569         }
 570
 571         return(1);
 572 }
 573
 574
 575 static int
 576 foptions(int *fflags, char *arg)
 577 {
 578         char            *v, *o;
 579         char            *toks[7];
 580
 581         toks[0] = "ign-scope";
 582         toks[1] = "no-ign-escape";
 583         toks[2] = "no-ign-macro";
 584         toks[3] = "no-ign-chars";
 585         toks[4] = "ign-errors";
 586         toks[5] = "strict";
 587         toks[6] = NULL;
 588
 589         while (*arg) {
 590                 o = arg;
 591                 switch (getsubopt(&arg, toks, &v)) {
 592                 case (0):
 593                         *fflags |= IGN_SCOPE;
 594                         break;
 595                 case (1):
 596                         *fflags |= NO_IGN_ESCAPE;
 597                         break;
 598                 case (2):
 599                         *fflags |= NO_IGN_MACRO;
 600                         break;
 601                 case (3):
 602                         *fflags |= NO_IGN_CHARS;
 603                         break;
 604                 case (4):
 605                         *fflags |= IGN_ERRORS;
 606                         break;
 607                 case (5):
 608                         *fflags |= NO_IGN_ESCAPE |
 609                                    NO_IGN_MACRO | NO_IGN_CHARS;
 610                         break;
 611                 default:
 612                         warnx("bad argument: -f%s", o);
 613                         return(0);
 614                 }
 615         }
 616
 617         return(1);
 618 }
 619
 620
 621 static int
 622 woptions(int *wflags, char *arg)
 623 {
 624         char            *v, *o;
 625         char            *toks[3];
 626
 627         toks[0] = "all";
 628         toks[1] = "error";
 629         toks[2] = NULL;
 630
 631         while (*arg) {
 632                 o = arg;
 633                 switch (getsubopt(&arg, toks, &v)) {
 634                 case (0):
 635                         *wflags |= WARN_WALL;
 636                         break;
 637                 case (1):
 638                         *wflags |= WARN_WERR;
 639                         break;
 640                 default:
 641                         warnx("bad argument: -W%s", o);
 642                         return(0);
 643                 }
 644         }
 645
 646         return(1);
 647 }
 648
 649
 650 /* ARGSUSED */
 651 static int
 652 merr(void *arg, int line, int col, const char *msg)
 653 {
 654         struct curparse *curp;
 655
 656         curp = (struct curparse *)arg;
 657
 658         (void)fprintf(stderr, "%s:%d:%d: error: %s\n",
 659                         curp->file, line, col + 1, msg);
 660
 661         return(0);
 662 }
 663
 664
 665 static int
 666 mwarn(void *arg, int line, int col, const char *msg)
 667 {
 668         struct curparse *curp;
 669
 670         curp = (struct curparse *)arg;
 671
 672         if ( ! (curp->wflags & WARN_WALL))
 673                 return(1);
 674
 675         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 676                         curp->file, line, col + 1, msg);
 677
 678         if ( ! (curp->wflags & WARN_WERR))
 679                 return(1);
 680
 681         return(0);
 682 }
 683