main.c

   1 /*      $Id: main.c,v 1.54 2009/10/31 06:17:19 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 /* Account for FreeBSD and Linux in our declarations. */
  34
  35 #ifdef __linux__
  36 extern  int               getsubopt(char **, char * const *, char **);
  37 extern  size_t            strlcat(char *, const char *, size_t);
  38 # ifndef __dead
  39 #  define __dead __attribute__((__noreturn__))
  40 # endif
  41 #elif defined(__dead2)
  42 # ifndef __dead
  43 #  define __dead __dead2
  44 # endif
  45 #endif
  46
  47 typedef void            (*out_mdoc)(void *, const struct mdoc *);
  48 typedef void            (*out_man)(void *, const struct man *);
  49 typedef void            (*out_free)(void *);
  50
  51 struct  buf {
  52         char             *buf;
  53         size_t            sz;
  54 };
  55
  56 enum    intt {
  57         INTT_AUTO,
  58         INTT_MDOC,
  59         INTT_MAN
  60 };
  61
  62 enum    outt {
  63         OUTT_ASCII = 0,
  64         OUTT_TREE,
  65         OUTT_HTML,
  66         OUTT_LINT
  67 };
  68
  69 struct  curparse {
  70         const char       *file;         /* Current parse. */
  71         int               fd;           /* Current parse. */
  72         int               wflags;
  73 #define WARN_WALL        (1 << 0)       /* All-warnings mask. */
  74 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  75         int               fflags;
  76 #define IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
  77 #define NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
  78 #define NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
  79 #define NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */
  80 #define IGN_ERRORS       (1 << 4)       /* Ignore failed parse. */
  81         enum intt         inttype;      /* Input parsers... */
  82         struct man       *man;
  83         struct man       *lastman;
  84         struct mdoc      *mdoc;
  85         struct mdoc      *lastmdoc;
  86         enum outt         outtype;      /* Output devices... */
  87         out_mdoc          outmdoc;
  88         out_man           outman;
  89         out_free          outfree;
  90         void             *outdata;
  91         char              outopts[BUFSIZ];
  92 };
  93
  94 static  int               foptions(int *, char *);
  95 static  int               toptions(enum outt *, char *);
  96 static  int               moptions(enum intt *, char *);
  97 static  int               woptions(int *, char *);
  98 static  int               merr(void *, int, int, const char *);
  99 static  int               mwarn(void *, int, int, const char *);
 100 static  int               ffile(struct buf *, struct buf *,
 101                                 const char *, struct curparse *);
 102 static  int               fdesc(struct buf *, struct buf *,
 103                                 struct curparse *);
 104 static  int               pset(const char *, int, struct curparse *,
 105                                 struct man **, struct mdoc **);
 106 static  struct man       *man_init(struct curparse *);
 107 static  struct mdoc      *mdoc_init(struct curparse *);
 108 __dead  static void       version(void);
 109 __dead  static void       usage(void);
 110
 111 static  const char       *progname;
 112
 113
 114 int
 115 main(int argc, char *argv[])
 116 {
 117         int              c, rc;
 118         struct buf       ln, blk;
 119         struct curparse  curp;
 120
 121         progname = strrchr(argv[0], '/');
 122         if (progname == NULL)
 123                 progname = argv[0];
 124         else
 125                 ++progname;
 126
 127         memset(&curp, 0, sizeof(struct curparse));
 128
 129         curp.inttype = INTT_AUTO;
 130         curp.outtype = OUTT_ASCII;
 131
 132         /* LINTED */
 133         while (-1 != (c = getopt(argc, argv, "f:m:O:T:VW:")))
 134                 switch (c) {
 135                 case ('f'):
 136                         if ( ! foptions(&curp.fflags, optarg))
 137                                 return(EXIT_FAILURE);
 138                         break;
 139                 case ('m'):
 140                         if ( ! moptions(&curp.inttype, optarg))
 141                                 return(EXIT_FAILURE);
 142                         break;
 143                 case ('O'):
 144                         (void)strlcat(curp.outopts, optarg, BUFSIZ);
 145                         (void)strlcat(curp.outopts, ",", BUFSIZ);
 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         memset(&ln, 0, sizeof(struct buf));
 167         memset(&blk, 0, 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                                 man_reset(curp.lastman);
 193                         if (curp.lastmdoc)
 194                                 mdoc_reset(curp.lastmdoc);
 195                         curp.lastman = NULL;
 196                         curp.lastmdoc = NULL;
 197                 }
 198         }
 199
 200         if (blk.buf)
 201                 free(blk.buf);
 202         if (ln.buf)
 203                 free(ln.buf);
 204         if (curp.outfree)
 205                 (*curp.outfree)(curp.outdata);
 206         if (curp.mdoc)
 207                 mdoc_free(curp.mdoc);
 208         if (curp.man)
 209                 man_free(curp.man);
 210
 211         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 212 }
 213
 214
 215 __dead static void
 216 version(void)
 217 {
 218
 219         (void)printf("%s %s\n", progname, VERSION);
 220         exit(EXIT_SUCCESS);
 221 }
 222
 223
 224 __dead static void
 225 usage(void)
 226 {
 227
 228         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 229                         "[-mformat] [-Ooption] [-Toutput] "
 230                         "[-Werr...]\n", progname);
 231         exit(EXIT_FAILURE);
 232 }
 233
 234
 235 static struct man *
 236 man_init(struct curparse *curp)
 237 {
 238         int              pflags;
 239         struct man_cb    mancb;
 240
 241         mancb.man_err = merr;
 242         mancb.man_warn = mwarn;
 243
 244         /* Defaults from mandoc.1. */
 245
 246         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS;
 247
 248         if (curp->fflags & NO_IGN_MACRO)
 249                 pflags &= ~MAN_IGN_MACRO;
 250         if (curp->fflags & NO_IGN_CHARS)
 251                 pflags &= ~MAN_IGN_CHARS;
 252         if (curp->fflags & NO_IGN_ESCAPE)
 253                 pflags &= ~MAN_IGN_ESCAPE;
 254
 255         return(man_alloc(curp, pflags, &mancb));
 256 }
 257
 258
 259 static struct mdoc *
 260 mdoc_init(struct curparse *curp)
 261 {
 262         int              pflags;
 263         struct mdoc_cb   mdoccb;
 264
 265         mdoccb.mdoc_err = merr;
 266         mdoccb.mdoc_warn = mwarn;
 267
 268         /* Defaults from mandoc.1. */
 269
 270         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 271
 272         if (curp->fflags & IGN_SCOPE)
 273                 pflags |= MDOC_IGN_SCOPE;
 274         if (curp->fflags & NO_IGN_ESCAPE)
 275                 pflags &= ~MDOC_IGN_ESCAPE;
 276         if (curp->fflags & NO_IGN_MACRO)
 277                 pflags &= ~MDOC_IGN_MACRO;
 278         if (curp->fflags & NO_IGN_CHARS)
 279                 pflags &= ~MDOC_IGN_CHARS;
 280
 281         return(mdoc_alloc(curp, pflags, &mdoccb));
 282 }
 283
 284
 285 static int
 286 ffile(struct buf *blk, struct buf *ln,
 287                 const char *file, struct curparse *curp)
 288 {
 289         int              c;
 290
 291         curp->file = file;
 292         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 293                 perror(curp->file);
 294                 return(-1);
 295         }
 296
 297         c = fdesc(blk, ln, curp);
 298
 299         if (-1 == close(curp->fd))
 300                 perror(curp->file);
 301
 302         return(c);
 303 }
 304
 305
 306 static int
 307 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 308 {
 309         size_t           sz;
 310         ssize_t          ssz;
 311         struct stat      st;
 312         int              j, i, pos, lnn, comment;
 313         struct man      *man;
 314         struct mdoc     *mdoc;
 315
 316         sz = BUFSIZ;
 317         man = NULL;
 318         mdoc = NULL;
 319
 320         /*
 321          * Two buffers: ln and buf.  buf is the input buffer optimised
 322          * here for each file's block size.  ln is a line buffer.  Both
 323          * growable, hence passed in by ptr-ptr.
 324          */
 325
 326         if (-1 == fstat(curp->fd, &st))
 327                 perror(curp->file);
 328         else if ((size_t)st.st_blksize > sz)
 329                 sz = st.st_blksize;
 330
 331         if (sz > blk->sz) {
 332                 blk->buf = realloc(blk->buf, sz);
 333                 if (NULL == blk->buf) {
 334                         perror(NULL);
 335                         exit(EXIT_FAILURE);
 336                 }
 337                 blk->sz = sz;
 338         }
 339
 340         /* Fill buf with file blocksize. */
 341
 342         for (lnn = pos = comment = 0; ; ) {
 343                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 344                         perror(curp->file);
 345                         return(-1);
 346                 } else if (0 == ssz)
 347                         break;
 348
 349                 /* Parse the read block into partial or full lines. */
 350
 351                 for (i = 0; i < (int)ssz; i++) {
 352                         if (pos >= (int)ln->sz) {
 353                                 ln->sz += 256; /* Step-size. */
 354                                 ln->buf = realloc(ln->buf, ln->sz);
 355                                 if (NULL == ln->buf) {
 356                                         perror(NULL);
 357                                         return(EXIT_FAILURE);
 358                                 }
 359                         }
 360
 361                         if ('\n' != blk->buf[i]) {
 362                                 if (comment)
 363                                         continue;
 364                                 ln->buf[pos++] = blk->buf[i];
 365
 366                                 /* Handle in-line `\"' comments. */
 367
 368                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 369                                         continue;
 370
 371                                 for (j = pos - 2; j >= 0; j--)
 372                                         if ('\\' != ln->buf[j])
 373                                                 break;
 374
 375                                 if ( ! ((pos - 2 - j) % 2))
 376                                         continue;
 377
 378                                 comment = 1;
 379                                 pos -= 2;
 380                                 continue;
 381                         }
 382
 383                         /* Handle escaped `\\n' newlines. */
 384
 385                         if (pos > 0 && 0 == comment &&
 386                                         '\\' == ln->buf[pos - 1]) {
 387                                 for (j = pos - 1; j >= 0; j--)
 388                                         if ('\\' != ln->buf[j])
 389                                                 break;
 390                                 if ( ! ((pos - j) % 2)) {
 391                                         pos--;
 392                                         lnn++;
 393                                         continue;
 394                                 }
 395                         }
 396
 397                         ln->buf[pos] = 0;
 398                         lnn++;
 399
 400                         /* If unset, assign parser in pset(). */
 401
 402                         if ( ! (man || mdoc) && ! pset(ln->buf,
 403                                                 pos, curp, &man, &mdoc))
 404                                 return(-1);
 405
 406                         pos = comment = 0;
 407
 408                         /* Pass down into parsers. */
 409
 410                         if (man && ! man_parseln(man, lnn, ln->buf))
 411                                 return(0);
 412                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 413                                 return(0);
 414                 }
 415         }
 416
 417         /* NOTE a parser may not have been assigned, yet. */
 418
 419         if ( ! (man || mdoc)) {
 420                 fprintf(stderr, "%s: Not a manual\n", curp->file);
 421                 return(0);
 422         }
 423
 424         if (mdoc && ! mdoc_endparse(mdoc))
 425                 return(0);
 426         if (man && ! man_endparse(man))
 427                 return(0);
 428
 429         /* If unset, allocate output dev now (if applicable). */
 430
 431         if ( ! (curp->outman && curp->outmdoc)) {
 432                 switch (curp->outtype) {
 433                 case (OUTT_HTML):
 434                         curp->outdata = html_alloc(curp->outopts);
 435                         curp->outman = html_man;
 436                         curp->outmdoc = html_mdoc;
 437                         curp->outfree = html_free;
 438                         break;
 439                 case (OUTT_TREE):
 440                         curp->outman = tree_man;
 441                         curp->outmdoc = tree_mdoc;
 442                         break;
 443                 case (OUTT_LINT):
 444                         break;
 445                 default:
 446                         curp->outdata = ascii_alloc();
 447                         curp->outman = terminal_man;
 448                         curp->outmdoc = terminal_mdoc;
 449                         curp->outfree = terminal_free;
 450                         break;
 451                 }
 452         }
 453
 454         /* Execute the out device, if it exists. */
 455
 456         if (man && curp->outman)
 457                 (*curp->outman)(curp->outdata, man);
 458         if (mdoc && curp->outmdoc)
 459                 (*curp->outmdoc)(curp->outdata, mdoc);
 460
 461         return(1);
 462 }
 463
 464
 465 static int
 466 pset(const char *buf, int pos, struct curparse *curp,
 467                 struct man **man, struct mdoc **mdoc)
 468 {
 469         int              i;
 470
 471         /*
 472          * Try to intuit which kind of manual parser should be used.  If
 473          * passed in by command-line (-man, -mdoc), then use that
 474          * explicitly.  If passed as -mandoc, then try to guess from the
 475          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 476          * default to -man, which is more lenient.
 477          */
 478
 479         if (buf[0] == '.') {
 480                 for (i = 1; buf[i]; i++)
 481                         if (' ' != buf[i] && '\t' != buf[i])
 482                                 break;
 483                 if (0 == buf[i])
 484                         return(1);
 485         }
 486
 487         switch (curp->inttype) {
 488         case (INTT_MDOC):
 489                 if (NULL == curp->mdoc)
 490                         curp->mdoc = mdoc_init(curp);
 491                 if (NULL == (*mdoc = curp->mdoc))
 492                         return(0);
 493                 curp->lastmdoc = *mdoc;
 494                 return(1);
 495         case (INTT_MAN):
 496                 if (NULL == curp->man)
 497                         curp->man = man_init(curp);
 498                 if (NULL == (*man = curp->man))
 499                         return(0);
 500                 curp->lastman = *man;
 501                 return(1);
 502         default:
 503                 break;
 504         }
 505
 506         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 507                 if (NULL == curp->mdoc)
 508                         curp->mdoc = mdoc_init(curp);
 509                 if (NULL == (*mdoc = curp->mdoc))
 510                         return(0);
 511                 curp->lastmdoc = *mdoc;
 512                 return(1);
 513         }
 514
 515         if (NULL == curp->man)
 516                 curp->man = man_init(curp);
 517         if (NULL == (*man = curp->man))
 518                 return(0);
 519         curp->lastman = *man;
 520         return(1);
 521 }
 522
 523
 524 static int
 525 moptions(enum intt *tflags, char *arg)
 526 {
 527
 528         if (0 == strcmp(arg, "doc"))
 529                 *tflags = INTT_MDOC;
 530         else if (0 == strcmp(arg, "andoc"))
 531                 *tflags = INTT_AUTO;
 532         else if (0 == strcmp(arg, "an"))
 533                 *tflags = INTT_MAN;
 534         else {
 535                 fprintf(stderr, "%s: Bad argument", arg);
 536                 return(0);
 537         }
 538
 539         return(1);
 540 }
 541
 542
 543 static int
 544 toptions(enum outt *tflags, char *arg)
 545 {
 546
 547         if (0 == strcmp(arg, "ascii"))
 548                 *tflags = OUTT_ASCII;
 549         else if (0 == strcmp(arg, "lint"))
 550                 *tflags = OUTT_LINT;
 551         else if (0 == strcmp(arg, "tree"))
 552                 *tflags = OUTT_TREE;
 553         else if (0 == strcmp(arg, "html"))
 554                 *tflags = OUTT_HTML;
 555         else {
 556                 fprintf(stderr, "%s: Bad argument", arg);
 557                 return(0);
 558         }
 559
 560         return(1);
 561 }
 562
 563
 564 static int
 565 foptions(int *fflags, char *arg)
 566 {
 567         char            *v, *o;
 568         const char      *toks[8];
 569
 570         toks[0] = "ign-scope";
 571         toks[1] = "no-ign-escape";
 572         toks[2] = "no-ign-macro";
 573         toks[3] = "no-ign-chars";
 574         toks[4] = "ign-errors";
 575         toks[5] = "strict";
 576         toks[6] = "ign-escape";
 577         toks[7] = NULL;
 578
 579         while (*arg) {
 580                 o = arg;
 581                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 582                 case (0):
 583                         *fflags |= IGN_SCOPE;
 584                         break;
 585                 case (1):
 586                         *fflags |= NO_IGN_ESCAPE;
 587                         break;
 588                 case (2):
 589                         *fflags |= NO_IGN_MACRO;
 590                         break;
 591                 case (3):
 592                         *fflags |= NO_IGN_CHARS;
 593                         break;
 594                 case (4):
 595                         *fflags |= IGN_ERRORS;
 596                         break;
 597                 case (5):
 598                         *fflags |= NO_IGN_ESCAPE |
 599                                    NO_IGN_MACRO | NO_IGN_CHARS;
 600                         break;
 601                 case (6):
 602                         *fflags &= ~NO_IGN_ESCAPE;
 603                         break;
 604                 default:
 605                         fprintf(stderr, "%s: Bad argument", o);
 606                         return(0);
 607                 }
 608         }
 609
 610         return(1);
 611 }
 612
 613
 614 static int
 615 woptions(int *wflags, char *arg)
 616 {
 617         char            *v, *o;
 618         const char      *toks[3];
 619
 620         toks[0] = "all";
 621         toks[1] = "error";
 622         toks[2] = NULL;
 623
 624         while (*arg) {
 625                 o = arg;
 626                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 627                 case (0):
 628                         *wflags |= WARN_WALL;
 629                         break;
 630                 case (1):
 631                         *wflags |= WARN_WERR;
 632                         break;
 633                 default:
 634                         fprintf(stderr, "%s: Bad argument", o);
 635                         return(0);
 636                 }
 637         }
 638
 639         return(1);
 640 }
 641
 642
 643 /* ARGSUSED */
 644 static int
 645 merr(void *arg, int line, int col, const char *msg)
 646 {
 647         struct curparse *curp;
 648
 649         curp = (struct curparse *)arg;
 650
 651         (void)fprintf(stderr, "%s:%d:%d: error: %s\n",
 652                         curp->file, line, col + 1, msg);
 653
 654         return(0);
 655 }
 656
 657
 658 static int
 659 mwarn(void *arg, int line, int col, const char *msg)
 660 {
 661         struct curparse *curp;
 662
 663         curp = (struct curparse *)arg;
 664
 665         if ( ! (curp->wflags & WARN_WALL))
 666                 return(1);
 667
 668         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 669                         curp->file, line, col + 1, msg);
 670
 671         if ( ! (curp->wflags & WARN_WERR))
 672                 return(1);
 673
 674         return(0);
 675 }
 676