main.c

   1 /*      $Id: main.c,v 1.45 2009/10/13 10:21:24 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 <stdint.h>
  24 #include <stdlib.h>
  25 #include <string.h>
  26 #include <unistd.h>
  27
  28 #include "mdoc.h"
  29 #include "man.h"
  30
  31 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  32
  33 /* Account for FreeBSD and Linux in our declarations. */
  34
  35 #ifdef __linux__
  36 extern  int               getsubopt(char **, char * const *, char **);
  37 # ifndef __dead
  38 #  define __dead __attribute__((__noreturn__))
  39 # endif
  40 #elif defined(__dead2)
  41 # ifndef __dead
  42 #  define __dead __dead2
  43 # endif
  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;
  91 };
  92
  93 extern  void             *html_alloc(char *);
  94 extern  void              html_mdoc(void *, const struct mdoc *);
  95 extern  void              html_man(void *, const struct man *);
  96 extern  void              html_free(void *);
  97 extern  void             *ascii_alloc(void);
  98 extern  void              tree_mdoc(void *, const struct mdoc *);
  99 extern  void              tree_man(void *, const struct man *);
 100 extern  void              terminal_mdoc(void *, const struct mdoc *);
 101 extern  void              terminal_man(void *, const struct man *);
 102 extern  void              terminal_free(void *);
 103
 104 static  int               foptions(int *, char *);
 105 static  int               toptions(enum outt *, char *);
 106 static  int               moptions(enum intt *, char *);
 107 static  int               woptions(int *, char *);
 108 static  int               merr(void *, int, int, const char *);
 109 static  int               mwarn(void *, int, int, const char *);
 110 static  int               ffile(struct buf *, struct buf *,
 111                                 const char *, struct curparse *);
 112 static  int               fdesc(struct buf *, struct buf *,
 113                                 struct curparse *);
 114 static  int               pset(const char *, int, struct curparse *,
 115                                 struct man **, struct mdoc **);
 116 static  struct man       *man_init(struct curparse *);
 117 static  struct mdoc      *mdoc_init(struct curparse *);
 118 __dead  static void       version(void);
 119 __dead  static void       usage(void);
 120
 121 extern  char             *__progname;
 122
 123
 124 int
 125 main(int argc, char *argv[])
 126 {
 127         int              c, rc;
 128         struct buf       ln, blk;
 129         struct curparse  curp;
 130
 131         bzero(&curp, sizeof(struct curparse));
 132
 133         curp.inttype = INTT_AUTO;
 134         curp.outtype = OUTT_ASCII;
 135
 136         /* LINTED */
 137         while (-1 != (c = getopt(argc, argv, "f:m:o:T:VW:")))
 138                 switch (c) {
 139                 case ('f'):
 140                         if ( ! foptions(&curp.fflags, optarg))
 141                                 return(EXIT_FAILURE);
 142                         break;
 143                 case ('m'):
 144                         if ( ! moptions(&curp.inttype, optarg))
 145                                 return(EXIT_FAILURE);
 146                         break;
 147                 case ('o'):
 148                         curp.outopts = optarg;
 149                         break;
 150                 case ('T'):
 151                         if ( ! toptions(&curp.outtype, optarg))
 152                                 return(EXIT_FAILURE);
 153                         break;
 154                 case ('W'):
 155                         if ( ! woptions(&curp.wflags, optarg))
 156                                 return(EXIT_FAILURE);
 157                         break;
 158                 case ('V'):
 159                         version();
 160                         /* NOTREACHED */
 161                 default:
 162                         usage();
 163                         /* NOTREACHED */
 164                 }
 165
 166         argc -= optind;
 167         argv += optind;
 168
 169         bzero(&ln, sizeof(struct buf));
 170         bzero(&blk, sizeof(struct buf));
 171
 172         rc = 1;
 173
 174         if (NULL == *argv) {
 175                 curp.file = "<stdin>";
 176                 curp.fd = STDIN_FILENO;
 177
 178                 c = fdesc(&blk, &ln, &curp);
 179                 if ( ! (IGN_ERRORS & curp.fflags))
 180                         rc = 1 == c ? 1 : 0;
 181                 else
 182                         rc = -1 == c ? 0 : 1;
 183         }
 184
 185         while (rc && *argv) {
 186                 c = ffile(&blk, &ln, *argv, &curp);
 187                 if ( ! (IGN_ERRORS & curp.fflags))
 188                         rc = 1 == c ? 1 : 0;
 189                 else
 190                         rc = -1 == c ? 0 : 1;
 191
 192                 argv++;
 193                 if (*argv && rc) {
 194                         if (curp.lastman)
 195                                 if ( ! man_reset(curp.lastman))
 196                                         rc = 0;
 197                         if (curp.lastmdoc)
 198                                 if ( ! mdoc_reset(curp.lastmdoc))
 199                                         rc = 0;
 200                         curp.lastman = NULL;
 201                         curp.lastmdoc = NULL;
 202                 }
 203         }
 204
 205         if (blk.buf)
 206                 free(blk.buf);
 207         if (ln.buf)
 208                 free(ln.buf);
 209         if (curp.outfree)
 210                 (*curp.outfree)(curp.outdata);
 211         if (curp.mdoc)
 212                 mdoc_free(curp.mdoc);
 213         if (curp.man)
 214                 man_free(curp.man);
 215
 216         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 217 }
 218
 219
 220 __dead static void
 221 version(void)
 222 {
 223
 224         (void)printf("%s %s\n", __progname, VERSION);
 225         exit(EXIT_SUCCESS);
 226 }
 227
 228
 229 __dead static void
 230 usage(void)
 231 {
 232
 233         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 234                         "[-mformat] [-Toutput] [-Werr...]\n",
 235                         __progname);
 236         exit(EXIT_FAILURE);
 237 }
 238
 239
 240 static struct man *
 241 man_init(struct curparse *curp)
 242 {
 243         int              pflags;
 244         struct man      *man;
 245         struct man_cb    mancb;
 246
 247         mancb.man_err = merr;
 248         mancb.man_warn = mwarn;
 249
 250         /* Defaults from mandoc.1. */
 251
 252         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS;
 253
 254         if (curp->fflags & NO_IGN_MACRO)
 255                 pflags &= ~MAN_IGN_MACRO;
 256         if (curp->fflags & NO_IGN_CHARS)
 257                 pflags &= ~MAN_IGN_CHARS;
 258         if (curp->fflags & NO_IGN_ESCAPE)
 259                 pflags &= ~MAN_IGN_ESCAPE;
 260
 261         if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 262                 warnx("memory exhausted");
 263
 264         return(man);
 265 }
 266
 267
 268 static struct mdoc *
 269 mdoc_init(struct curparse *curp)
 270 {
 271         int              pflags;
 272         struct mdoc     *mdoc;
 273         struct mdoc_cb   mdoccb;
 274
 275         mdoccb.mdoc_err = merr;
 276         mdoccb.mdoc_warn = mwarn;
 277
 278         /* Defaults from mandoc.1. */
 279
 280         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 281
 282         if (curp->fflags & IGN_SCOPE)
 283                 pflags |= MDOC_IGN_SCOPE;
 284         if (curp->fflags & NO_IGN_ESCAPE)
 285                 pflags &= ~MDOC_IGN_ESCAPE;
 286         if (curp->fflags & NO_IGN_MACRO)
 287                 pflags &= ~MDOC_IGN_MACRO;
 288         if (curp->fflags & NO_IGN_CHARS)
 289                 pflags &= ~MDOC_IGN_CHARS;
 290
 291         if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb)))
 292                 warnx("memory exhausted");
 293
 294         return(mdoc);
 295 }
 296
 297
 298 static int
 299 ffile(struct buf *blk, struct buf *ln,
 300                 const char *file, struct curparse *curp)
 301 {
 302         int              c;
 303
 304         curp->file = file;
 305         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 306                 warn("%s", curp->file);
 307                 return(-1);
 308         }
 309
 310         c = fdesc(blk, ln, curp);
 311
 312         if (-1 == close(curp->fd))
 313                 warn("%s", curp->file);
 314
 315         return(c);
 316 }
 317
 318
 319 static int
 320 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 321 {
 322         size_t           sz;
 323         ssize_t          ssz;
 324         struct stat      st;
 325         int              j, i, pos, lnn, comment;
 326         struct man      *man;
 327         struct mdoc     *mdoc;
 328
 329         sz = BUFSIZ;
 330         man = NULL;
 331         mdoc = NULL;
 332
 333         /*
 334          * Two buffers: ln and buf.  buf is the input buffer optimised
 335          * here for each file's block size.  ln is a line buffer.  Both
 336          * growable, hence passed in by ptr-ptr.
 337          */
 338
 339         if (-1 == fstat(curp->fd, &st))
 340                 warn("%s", curp->file);
 341         else if ((size_t)st.st_blksize > sz)
 342                 sz = st.st_blksize;
 343
 344         if (sz > blk->sz) {
 345                 blk->buf = realloc(blk->buf, sz);
 346                 if (NULL == blk->buf) {
 347                         warn("realloc");
 348                         return(-1);
 349                 }
 350                 blk->sz = sz;
 351         }
 352
 353         /* Fill buf with file blocksize. */
 354
 355         for (lnn = pos = comment = 0; ; ) {
 356                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 357                         warn("%s", curp->file);
 358                         return(-1);
 359                 } else if (0 == ssz)
 360                         break;
 361
 362                 /* Parse the read block into partial or full lines. */
 363
 364                 for (i = 0; i < (int)ssz; i++) {
 365                         if (pos >= (int)ln->sz) {
 366                                 ln->sz += 256; /* Step-size. */
 367                                 ln->buf = realloc(ln->buf, ln->sz);
 368                                 if (NULL == ln->buf) {
 369                                         warn("realloc");
 370                                         return(-1);
 371                                 }
 372                         }
 373
 374                         if ('\n' != blk->buf[i]) {
 375                                 if (comment)
 376                                         continue;
 377                                 ln->buf[pos++] = blk->buf[i];
 378
 379                                 /* Handle in-line `\"' comments. */
 380
 381                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 382                                         continue;
 383
 384                                 for (j = pos - 2; j >= 0; j--)
 385                                         if ('\\' != ln->buf[j])
 386                                                 break;
 387
 388                                 if ( ! ((pos - 2 - j) % 2))
 389                                         continue;
 390
 391                                 comment = 1;
 392                                 pos -= 2;
 393                                 continue;
 394                         }
 395
 396                         /* Handle escaped `\\n' newlines. */
 397
 398                         if (pos > 0 && 0 == comment &&
 399                                         '\\' == ln->buf[pos - 1]) {
 400                                 for (j = pos - 1; j >= 0; j--)
 401                                         if ('\\' != ln->buf[j])
 402                                                 break;
 403                                 if ( ! ((pos - j) % 2)) {
 404                                         pos--;
 405                                         lnn++;
 406                                         continue;
 407                                 }
 408                         }
 409
 410                         ln->buf[pos] = 0;
 411                         lnn++;
 412
 413                         /* If unset, assign parser in pset(). */
 414
 415                         if ( ! (man || mdoc) && ! pset(ln->buf,
 416                                                 pos, curp, &man, &mdoc))
 417                                 return(-1);
 418
 419                         pos = comment = 0;
 420
 421                         /* Pass down into parsers. */
 422
 423                         if (man && ! man_parseln(man, lnn, ln->buf))
 424                                 return(0);
 425                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 426                                 return(0);
 427                 }
 428         }
 429
 430         /* NOTE a parser may not have been assigned, yet. */
 431
 432         if ( ! (man || mdoc)) {
 433                 (void)fprintf(stderr, "%s: not a manual\n",
 434                                 curp->file);
 435                 return(0);
 436         }
 437
 438         if (mdoc && ! mdoc_endparse(mdoc))
 439                 return(0);
 440         if (man && ! man_endparse(man))
 441                 return(0);
 442
 443         /* If unset, allocate output dev now (if applicable). */
 444
 445         if ( ! (curp->outman && curp->outmdoc)) {
 446                 switch (curp->outtype) {
 447                 case (OUTT_HTML):
 448                         curp->outdata = html_alloc(curp->outopts);
 449                         curp->outman = html_man;
 450                         curp->outmdoc = html_mdoc;
 451                         curp->outfree = html_free;
 452                         break;
 453                 case (OUTT_TREE):
 454                         curp->outman = tree_man;
 455                         curp->outmdoc = tree_mdoc;
 456                         break;
 457                 case (OUTT_LINT):
 458                         break;
 459                 default:
 460                         curp->outdata = ascii_alloc();
 461                         curp->outman = terminal_man;
 462                         curp->outmdoc = terminal_mdoc;
 463                         curp->outfree = terminal_free;
 464                         break;
 465                 }
 466         }
 467
 468         /* Execute the out device, if it exists. */
 469
 470         if (man && curp->outman)
 471                 (*curp->outman)(curp->outdata, man);
 472         if (mdoc && curp->outmdoc)
 473                 (*curp->outmdoc)(curp->outdata, mdoc);
 474
 475         return(1);
 476 }
 477
 478
 479 static int
 480 pset(const char *buf, int pos, struct curparse *curp,
 481                 struct man **man, struct mdoc **mdoc)
 482 {
 483         int              i;
 484
 485         /*
 486          * Try to intuit which kind of manual parser should be used.  If
 487          * passed in by command-line (-man, -mdoc), then use that
 488          * explicitly.  If passed as -mandoc, then try to guess from the
 489          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 490          * default to -man, which is more lenient.
 491          */
 492
 493         if (buf[0] == '.') {
 494                 for (i = 1; buf[i]; i++)
 495                         if (' ' != buf[i] && '\t' != buf[i])
 496                                 break;
 497                 if (0 == buf[i])
 498                         return(1);
 499         }
 500
 501         switch (curp->inttype) {
 502         case (INTT_MDOC):
 503                 if (NULL == curp->mdoc)
 504                         curp->mdoc = mdoc_init(curp);
 505                 if (NULL == (*mdoc = curp->mdoc))
 506                         return(0);
 507                 curp->lastmdoc = *mdoc;
 508                 return(1);
 509         case (INTT_MAN):
 510                 if (NULL == curp->man)
 511                         curp->man = man_init(curp);
 512                 if (NULL == (*man = curp->man))
 513                         return(0);
 514                 curp->lastman = *man;
 515                 return(1);
 516         default:
 517                 break;
 518         }
 519
 520         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 521                 if (NULL == curp->mdoc)
 522                         curp->mdoc = mdoc_init(curp);
 523                 if (NULL == (*mdoc = curp->mdoc))
 524                         return(0);
 525                 curp->lastmdoc = *mdoc;
 526                 return(1);
 527         }
 528
 529         if (NULL == curp->man)
 530                 curp->man = man_init(curp);
 531         if (NULL == (*man = curp->man))
 532                 return(0);
 533         curp->lastman = *man;
 534         return(1);
 535 }
 536
 537
 538 static int
 539 moptions(enum intt *tflags, char *arg)
 540 {
 541
 542         if (0 == strcmp(arg, "doc"))
 543                 *tflags = INTT_MDOC;
 544         else if (0 == strcmp(arg, "andoc"))
 545                 *tflags = INTT_AUTO;
 546         else if (0 == strcmp(arg, "an"))
 547                 *tflags = INTT_MAN;
 548         else {
 549                 warnx("bad argument: -m%s", arg);
 550                 return(0);
 551         }
 552
 553         return(1);
 554 }
 555
 556
 557 static int
 558 toptions(enum outt *tflags, char *arg)
 559 {
 560
 561         if (0 == strcmp(arg, "ascii"))
 562                 *tflags = OUTT_ASCII;
 563         else if (0 == strcmp(arg, "lint"))
 564                 *tflags = OUTT_LINT;
 565         else if (0 == strcmp(arg, "tree"))
 566                 *tflags = OUTT_TREE;
 567         else if (0 == strcmp(arg, "html"))
 568                 *tflags = OUTT_HTML;
 569         else {
 570                 warnx("bad argument: -T%s", 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[7];
 583
 584         toks[0] = "ign-scope";
 585         toks[1] = "no-ign-escape";
 586         toks[2] = "no-ign-macro";
 587         toks[3] = "no-ign-chars";
 588         toks[4] = "ign-errors";
 589         toks[5] = "strict";
 590         toks[6] = NULL;
 591
 592         while (*arg) {
 593                 o = arg;
 594                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 595                 case (0):
 596                         *fflags |= IGN_SCOPE;
 597                         break;
 598                 case (1):
 599                         *fflags |= NO_IGN_ESCAPE;
 600                         break;
 601                 case (2):
 602                         *fflags |= NO_IGN_MACRO;
 603                         break;
 604                 case (3):
 605                         *fflags |= NO_IGN_CHARS;
 606                         break;
 607                 case (4):
 608                         *fflags |= IGN_ERRORS;
 609                         break;
 610                 case (5):
 611                         *fflags |= NO_IGN_ESCAPE |
 612                                    NO_IGN_MACRO | NO_IGN_CHARS;
 613                         break;
 614                 default:
 615                         warnx("bad argument: -f%s", o);
 616                         return(0);
 617                 }
 618         }
 619
 620         return(1);
 621 }
 622
 623
 624 static int
 625 woptions(int *wflags, char *arg)
 626 {
 627         char            *v, *o;
 628         const char      *toks[3];
 629
 630         toks[0] = "all";
 631         toks[1] = "error";
 632         toks[2] = NULL;
 633
 634         while (*arg) {
 635                 o = arg;
 636                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 637                 case (0):
 638                         *wflags |= WARN_WALL;
 639                         break;
 640                 case (1):
 641                         *wflags |= WARN_WERR;
 642                         break;
 643                 default:
 644                         warnx("bad argument: -W%s", o);
 645                         return(0);
 646                 }
 647         }
 648
 649         return(1);
 650 }
 651
 652
 653 /* ARGSUSED */
 654 static int
 655 merr(void *arg, int line, int col, const char *msg)
 656 {
 657         struct curparse *curp;
 658
 659         curp = (struct curparse *)arg;
 660
 661         (void)fprintf(stderr, "%s:%d:%d: error: %s\n",
 662                         curp->file, line, col + 1, msg);
 663
 664         return(0);
 665 }
 666
 667
 668 static int
 669 mwarn(void *arg, int line, int col, const char *msg)
 670 {
 671         struct curparse *curp;
 672
 673         curp = (struct curparse *)arg;
 674
 675         if ( ! (curp->wflags & WARN_WALL))
 676                 return(1);
 677
 678         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 679                         curp->file, line, col + 1, msg);
 680
 681         if ( ! (curp->wflags & WARN_WERR))
 682                 return(1);
 683
 684         return(0);
 685 }
 686