main.c

   1 /*      $Id: main.c,v 1.36 2009/07/06 09:34:29 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 int             (*out_mdoc)(void *, const struct mdoc *);
  44 typedef int             (*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_LINT
  62 };
  63
  64 struct  curparse {
  65         const char       *file;         /* Current parse. */
  66         int               fd;           /* Current parse. */
  67         int               wflags;
  68 #define WARN_WALL        (1 << 0)       /* All-warnings mask. */
  69 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  70         int               fflags;
  71 #define IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
  72 #define NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
  73 #define NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
  74 #define NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */
  75         enum intt         inttype;      /* Input parsers... */
  76         struct man       *man;
  77         struct man       *lastman;
  78         struct mdoc      *mdoc;
  79         struct mdoc      *lastmdoc;
  80         enum outt         outtype;      /* Output devices... */
  81         out_mdoc          outmdoc;
  82         out_man           outman;
  83         out_free          outfree;
  84         void             *outdata;
  85 };
  86
  87 extern  void             *ascii_alloc(void);
  88 extern  int               tree_mdoc(void *, const struct mdoc *);
  89 extern  int               tree_man(void *, const struct man *);
  90 extern  int               terminal_mdoc(void *, const struct mdoc *);
  91 extern  int               terminal_man(void *, const struct man *);
  92 extern  void              terminal_free(void *);
  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               manwarn(void *, int, int, const char *);
 100 static  int               mdocwarn(void *, int, int, const char *);
 101 static  int               ffile(struct buf *, struct buf *,
 102                                 const char *, struct curparse *);
 103 static  int               fdesc(struct buf *, struct buf *,
 104                                 struct curparse *);
 105 static  int               pset(const char *, int, struct curparse *,
 106                                 struct man **, struct mdoc **);
 107 static  struct man       *man_init(struct curparse *);
 108 static  struct mdoc      *mdoc_init(struct curparse *);
 109 __dead  static void       version(void);
 110 __dead  static void       usage(void);
 111
 112 extern  char             *__progname;
 113
 114
 115 int
 116 main(int argc, char *argv[])
 117 {
 118         int              c, rc;
 119         struct buf       ln, blk;
 120         struct curparse  curp;
 121
 122         bzero(&curp, sizeof(struct curparse));
 123
 124         curp.inttype = INTT_AUTO;
 125         curp.outtype = OUTT_ASCII;
 126
 127         /* LINTED */
 128         while (-1 != (c = getopt(argc, argv, "f:m:VW:T:")))
 129                 switch (c) {
 130                 case ('f'):
 131                         if ( ! foptions(&curp.fflags, optarg))
 132                                 return(EXIT_FAILURE);
 133                         break;
 134                 case ('m'):
 135                         if ( ! moptions(&curp.inttype, optarg))
 136                                 return(EXIT_FAILURE);
 137                         break;
 138                 case ('T'):
 139                         if ( ! toptions(&curp.outtype, optarg))
 140                                 return(EXIT_FAILURE);
 141                         break;
 142                 case ('W'):
 143                         if ( ! woptions(&curp.wflags, optarg))
 144                                 return(EXIT_FAILURE);
 145                         break;
 146                 case ('V'):
 147                         version();
 148                         /* NOTREACHED */
 149                 default:
 150                         usage();
 151                         /* NOTREACHED */
 152                 }
 153
 154         argc -= optind;
 155         argv += optind;
 156
 157         bzero(&ln, sizeof(struct buf));
 158         bzero(&blk, sizeof(struct buf));
 159
 160         rc = 1;
 161
 162         if (NULL == *argv) {
 163                 curp.file = "<stdin>";
 164                 curp.fd = STDIN_FILENO;
 165                 if ( ! fdesc(&blk, &ln, &curp))
 166                         rc = 0;
 167         }
 168
 169         while (rc && *argv) {
 170                 if ( ! ffile(&blk, &ln, *argv, &curp))
 171                         rc = 0;
 172                 argv++;
 173                 if (*argv && rc) {
 174                         if (curp.lastman)
 175                                 if ( ! man_reset(curp.lastman))
 176                                         rc = 0;
 177                         if (curp.lastmdoc)
 178                                 if ( ! mdoc_reset(curp.lastmdoc))
 179                                         rc = 0;
 180                         curp.lastman = NULL;
 181                         curp.lastmdoc = NULL;
 182                 }
 183         }
 184
 185         if (blk.buf)
 186                 free(blk.buf);
 187         if (ln.buf)
 188                 free(ln.buf);
 189         if (curp.outfree)
 190                 (*curp.outfree)(curp.outdata);
 191         if (curp.mdoc)
 192                 mdoc_free(curp.mdoc);
 193         if (curp.man)
 194                 man_free(curp.man);
 195
 196         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 197 }
 198
 199
 200 __dead static void
 201 version(void)
 202 {
 203
 204         (void)printf("%s %s\n", __progname, VERSION);
 205         exit(EXIT_SUCCESS);
 206 }
 207
 208
 209 __dead static void
 210 usage(void)
 211 {
 212
 213         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 214                         "[-mformat] [-Toutput] [-Werr...]\n",
 215                         __progname);
 216         exit(EXIT_FAILURE);
 217 }
 218
 219
 220 static struct man *
 221 man_init(struct curparse *curp)
 222 {
 223         int              pflags;
 224         struct man      *man;
 225         struct man_cb    mancb;
 226
 227         mancb.man_err = merr;
 228         mancb.man_warn = manwarn;
 229
 230         /* Defaults from mandoc.1. */
 231
 232         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS;
 233
 234         if (curp->fflags & NO_IGN_MACRO)
 235                 pflags &= ~MAN_IGN_MACRO;
 236         if (curp->fflags & NO_IGN_CHARS)
 237                 pflags &= ~MAN_IGN_CHARS;
 238         if (curp->fflags & NO_IGN_ESCAPE)
 239                 pflags &= ~MAN_IGN_ESCAPE;
 240
 241         if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 242                 warnx("memory exhausted");
 243
 244         return(man);
 245 }
 246
 247
 248 static struct mdoc *
 249 mdoc_init(struct curparse *curp)
 250 {
 251         int              pflags;
 252         struct mdoc     *mdoc;
 253         struct mdoc_cb   mdoccb;
 254
 255         mdoccb.mdoc_err = merr;
 256         mdoccb.mdoc_warn = mdocwarn;
 257
 258         /* Defaults from mandoc.1. */
 259
 260         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 261
 262         if (curp->fflags & IGN_SCOPE)
 263                 pflags |= MDOC_IGN_SCOPE;
 264         if (curp->fflags & NO_IGN_ESCAPE)
 265                 pflags &= ~MDOC_IGN_ESCAPE;
 266         if (curp->fflags & NO_IGN_MACRO)
 267                 pflags &= ~MDOC_IGN_MACRO;
 268         if (curp->fflags & NO_IGN_CHARS)
 269                 pflags &= ~MDOC_IGN_CHARS;
 270
 271         if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb)))
 272                 warnx("memory exhausted");
 273
 274         return(mdoc);
 275 }
 276
 277
 278 static int
 279 ffile(struct buf *blk, struct buf *ln,
 280                 const char *file, struct curparse *curp)
 281 {
 282         int              c;
 283
 284         curp->file = file;
 285         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 286                 warn("%s", curp->file);
 287                 return(0);
 288         }
 289
 290         c = fdesc(blk, ln, curp);
 291
 292         if (-1 == close(curp->fd))
 293                 warn("%s", curp->file);
 294
 295         return(c);
 296 }
 297
 298
 299 static int
 300 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 301 {
 302         size_t           sz;
 303         ssize_t          ssz;
 304         struct stat      st;
 305         int              j, i, pos, lnn, comment;
 306         struct man      *man;
 307         struct mdoc     *mdoc;
 308
 309         sz = BUFSIZ;
 310         man = NULL;
 311         mdoc = NULL;
 312
 313         /*
 314          * Two buffers: ln and buf.  buf is the input buffer optimised
 315          * here for each file's block size.  ln is a line buffer.  Both
 316          * growable, hence passed in by ptr-ptr.
 317          */
 318
 319         if (-1 == fstat(curp->fd, &st))
 320                 warn("%s", curp->file);
 321         else if ((size_t)st.st_blksize > sz)
 322                 sz = st.st_blksize;
 323
 324         if (sz > blk->sz) {
 325                 blk->buf = realloc(blk->buf, sz);
 326                 if (NULL == blk->buf) {
 327                         warn("realloc");
 328                         return(0);
 329                 }
 330                 blk->sz = sz;
 331         }
 332
 333         /* Fill buf with file blocksize. */
 334
 335         for (lnn = pos = comment = 0; ; ) {
 336                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 337                         warn("%s", curp->file);
 338                         return(0);
 339                 } else if (0 == ssz)
 340                         break;
 341
 342                 /* Parse the read block into partial or full lines. */
 343
 344                 for (i = 0; i < (int)ssz; i++) {
 345                         if (pos >= (int)ln->sz) {
 346                                 ln->sz += 256; /* Step-size. */
 347                                 ln->buf = realloc(ln->buf, ln->sz);
 348                                 if (NULL == ln->buf) {
 349                                         warn("realloc");
 350                                         return(0);
 351                                 }
 352                         }
 353
 354                         if ('\n' != blk->buf[i]) {
 355                                 if (comment)
 356                                         continue;
 357                                 ln->buf[pos++] = blk->buf[i];
 358
 359                                 /* Handle in-line `\"' comments. */
 360
 361                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 362                                         continue;
 363
 364                                 for (j = pos - 2; j >= 0; j--)
 365                                         if ('\\' != ln->buf[j])
 366                                                 break;
 367
 368                                 if ( ! ((pos - 2 - j) % 2))
 369                                         continue;
 370
 371                                 comment = 1;
 372                                 pos -= 2;
 373                                 continue;
 374                         }
 375
 376                         /* Handle escaped `\\n' newlines. */
 377
 378                         if (pos > 0 && 0 == comment &&
 379                                         '\\' == ln->buf[pos - 1]) {
 380                                 for (j = pos - 1; j >= 0; j--)
 381                                         if ('\\' != ln->buf[j])
 382                                                 break;
 383                                 if ( ! ((pos - j) % 2)) {
 384                                         pos--;
 385                                         lnn++;
 386                                         continue;
 387                                 }
 388                         }
 389
 390                         ln->buf[pos] = 0;
 391                         lnn++;
 392
 393                         /* If unset, assign parser in pset(). */
 394
 395                         if ( ! (man || mdoc) && ! pset(ln->buf,
 396                                                 pos, curp, &man, &mdoc))
 397                                 return(0);
 398
 399                         pos = comment = 0;
 400
 401                         /* Pass down into parsers. */
 402
 403                         if (man && ! man_parseln(man, lnn, ln->buf))
 404                                 return(0);
 405                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 406                                 return(0);
 407                 }
 408         }
 409
 410         /* NOTE a parser may not have been assigned, yet. */
 411
 412         if ( ! (man || mdoc)) {
 413                 warnx("%s: not a manual", curp->file);
 414                 return(0);
 415         }
 416
 417         if (mdoc && ! mdoc_endparse(mdoc))
 418                 return(0);
 419         if (man && ! man_endparse(man))
 420                 return(0);
 421
 422         /* If unset, allocate output dev now (if applicable). */
 423
 424         if ( ! (curp->outman && curp->outmdoc)) {
 425                 switch (curp->outtype) {
 426                 case (OUTT_TREE):
 427                         curp->outman = tree_man;
 428                         curp->outmdoc = tree_mdoc;
 429                         break;
 430                 case (OUTT_LINT):
 431                         break;
 432                 default:
 433                         curp->outdata = ascii_alloc();
 434                         curp->outman = terminal_man;
 435                         curp->outmdoc = terminal_mdoc;
 436                         curp->outfree = terminal_free;
 437                         break;
 438                 }
 439         }
 440
 441         /* Execute the out device, if it exists. */
 442
 443         if (man && curp->outman)
 444                 if ( ! (*curp->outman)(curp->outdata, man))
 445                         return(0);
 446         if (mdoc && curp->outmdoc)
 447                 if ( ! (*curp->outmdoc)(curp->outdata, mdoc))
 448                         return(0);
 449
 450         return(1);
 451 }
 452
 453
 454 static int
 455 pset(const char *buf, int pos, struct curparse *curp,
 456                 struct man **man, struct mdoc **mdoc)
 457 {
 458         int              i;
 459
 460         /*
 461          * Try to intuit which kind of manual parser should be used.  If
 462          * passed in by command-line (-man, -mdoc), then use that
 463          * explicitly.  If passed as -mandoc, then try to guess from the
 464          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 465          * default to -man, which is more lenient.
 466          */
 467
 468         if (buf[0] == '.') {
 469                 for (i = 1; buf[i]; i++)
 470                         if (' ' != buf[i] && '\t' != buf[i])
 471                                 break;
 472                 if (0 == buf[i])
 473                         return(1);
 474         }
 475
 476         switch (curp->inttype) {
 477         case (INTT_MDOC):
 478                 if (NULL == curp->mdoc)
 479                         curp->mdoc = mdoc_init(curp);
 480                 if (NULL == (*mdoc = curp->mdoc))
 481                         return(0);
 482                 curp->lastmdoc = *mdoc;
 483                 return(1);
 484         case (INTT_MAN):
 485                 if (NULL == curp->man)
 486                         curp->man = man_init(curp);
 487                 if (NULL == (*man = curp->man))
 488                         return(0);
 489                 curp->lastman = *man;
 490                 return(1);
 491         default:
 492                 break;
 493         }
 494
 495         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 496                 if (NULL == curp->mdoc)
 497                         curp->mdoc = mdoc_init(curp);
 498                 if (NULL == (*mdoc = curp->mdoc))
 499                         return(0);
 500                 curp->lastmdoc = *mdoc;
 501                 return(1);
 502         }
 503
 504         if (NULL == curp->man)
 505                 curp->man = man_init(curp);
 506         if (NULL == (*man = curp->man))
 507                 return(0);
 508         curp->lastman = *man;
 509         return(1);
 510 }
 511
 512
 513 static int
 514 moptions(enum intt *tflags, char *arg)
 515 {
 516
 517         if (0 == strcmp(arg, "doc"))
 518                 *tflags = INTT_MDOC;
 519         else if (0 == strcmp(arg, "andoc"))
 520                 *tflags = INTT_AUTO;
 521         else if (0 == strcmp(arg, "an"))
 522                 *tflags = INTT_MAN;
 523         else {
 524                 warnx("bad argument: -m%s", arg);
 525                 return(0);
 526         }
 527
 528         return(1);
 529 }
 530
 531
 532 static int
 533 toptions(enum outt *tflags, char *arg)
 534 {
 535
 536         if (0 == strcmp(arg, "ascii"))
 537                 *tflags = OUTT_ASCII;
 538         else if (0 == strcmp(arg, "lint"))
 539                 *tflags = OUTT_LINT;
 540         else if (0 == strcmp(arg, "tree"))
 541                 *tflags = OUTT_TREE;
 542         else {
 543                 warnx("bad argument: -T%s", arg);
 544                 return(0);
 545         }
 546
 547         return(1);
 548 }
 549
 550
 551 static int
 552 foptions(int *fflags, char *arg)
 553 {
 554         char            *v, *o;
 555         char            *toks[6];
 556
 557         toks[0] = "ign-scope";
 558         toks[1] = "no-ign-escape";
 559         toks[2] = "no-ign-macro";
 560         toks[3] = "no-ign-chars";
 561         toks[4] = "strict";
 562         toks[5] = NULL;
 563
 564         while (*arg) {
 565                 o = arg;
 566                 switch (getsubopt(&arg, toks, &v)) {
 567                 case (0):
 568                         *fflags |= IGN_SCOPE;
 569                         break;
 570                 case (1):
 571                         *fflags |= NO_IGN_ESCAPE;
 572                         break;
 573                 case (2):
 574                         *fflags |= NO_IGN_MACRO;
 575                         break;
 576                 case (3):
 577                         *fflags |= NO_IGN_CHARS;
 578                         break;
 579                 case (4):
 580                         *fflags |= NO_IGN_ESCAPE |
 581                                    NO_IGN_MACRO | NO_IGN_CHARS;
 582                         break;
 583                 default:
 584                         warnx("bad argument: -f%s", o);
 585                         return(0);
 586                 }
 587         }
 588
 589         return(1);
 590 }
 591
 592
 593 static int
 594 woptions(int *wflags, char *arg)
 595 {
 596         char            *v, *o;
 597         char            *toks[3];
 598
 599         toks[0] = "all";
 600         toks[1] = "error";
 601         toks[2] = NULL;
 602
 603         while (*arg) {
 604                 o = arg;
 605                 switch (getsubopt(&arg, toks, &v)) {
 606                 case (0):
 607                         *wflags |= WARN_WALL;
 608                         break;
 609                 case (1):
 610                         *wflags |= WARN_WERR;
 611                         break;
 612                 default:
 613                         warnx("bad argument: -W%s", o);
 614                         return(0);
 615                 }
 616         }
 617
 618         return(1);
 619 }
 620
 621
 622 /* ARGSUSED */
 623 static int
 624 merr(void *arg, int line, int col, const char *msg)
 625 {
 626         struct curparse *curp;
 627
 628         curp = (struct curparse *)arg;
 629
 630         warnx("%s:%d: error: %s (column %d)",
 631                         curp->file, line, msg, col);
 632
 633         return(0);
 634 }
 635
 636
 637 static int
 638 mdocwarn(void *arg, int line, int col, const char *msg)
 639 {
 640         struct curparse *curp;
 641
 642         curp = (struct curparse *)arg;
 643
 644         if ( ! (curp->wflags & WARN_WALL))
 645                 return(1);
 646
 647         warnx("%s:%d: syntax warning: %s (column %d)",
 648                         curp->file, line, msg, col);
 649
 650         if ( ! (curp->wflags & WARN_WERR))
 651                 return(1);
 652
 653         warnx("considering warnings as errors");
 654         return(0);
 655 }
 656
 657
 658 static int
 659 manwarn(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         warnx("%s:%d: syntax warning: %s (column %d)",
 669                         curp->file, line, msg, col);
 670
 671         if ( ! (curp->wflags & WARN_WERR))
 672                 return(1);
 673
 674         warnx("considering warnings as errors");
 675         return(0);
 676 }