main.c

   1 /*      $Id: main.c,v 1.30 2009/06/16 20:22:23 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         0x03          /* All-warnings mask. */
  69 #define WARN_WCOMPAT     (1 << 0)       /* Compatibility warnings. */
  70 #define WARN_WSYNTAX     (1 << 1)       /* Syntax warnings. */
  71 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  72         int               fflags;
  73 #define IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
  74 #define NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
  75 #define NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
  76 #define NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */
  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 };
  88
  89 extern  void             *ascii_alloc(void);
  90 extern  int               tree_mdoc(void *, const struct mdoc *);
  91 extern  int               tree_man(void *, const struct man *);
  92 extern  int               terminal_mdoc(void *, const struct mdoc *);
  93 extern  int               terminal_man(void *, const struct man *);
  94 extern  void              terminal_free(void *);
  95
  96 static  int               foptions(int *, char *);
  97 static  int               toptions(enum outt *, char *);
  98 static  int               moptions(enum intt *, char *);
  99 static  int               woptions(int *, char *);
 100 static  int               merr(void *, int, int, const char *);
 101 static  int               manwarn(void *, int, int, const char *);
 102 static  int               mdocwarn(void *, int, int,
 103                                 enum mdoc_warn, const char *);
 104 static  int               ffile(struct buf *, struct buf *,
 105                                 const char *, struct curparse *);
 106 static  int               fdesc(struct buf *, struct buf *,
 107                                 struct curparse *);
 108 static  int               pset(const char *, int, struct curparse *,
 109                                 struct man **, struct mdoc **);
 110 static  struct man       *man_init(struct curparse *);
 111 static  struct mdoc      *mdoc_init(struct curparse *);
 112 __dead  static void       version(void);
 113 __dead  static void       usage(void);
 114
 115 extern  char             *__progname;
 116
 117
 118 int
 119 main(int argc, char *argv[])
 120 {
 121         int              c, rc;
 122         struct buf       ln, blk;
 123         struct curparse  curp;
 124
 125         bzero(&curp, sizeof(struct curparse));
 126
 127         curp.inttype = INTT_AUTO;
 128         curp.outtype = OUTT_ASCII;
 129
 130         /* LINTED */
 131         while (-1 != (c = getopt(argc, argv, "f:m:VW:T:")))
 132                 switch (c) {
 133                 case ('f'):
 134                         if ( ! foptions(&curp.fflags, optarg))
 135                                 return(0);
 136                         break;
 137                 case ('m'):
 138                         if ( ! moptions(&curp.inttype, optarg))
 139                                 return(0);
 140                         break;
 141                 case ('T'):
 142                         if ( ! toptions(&curp.outtype, optarg))
 143                                 return(0);
 144                         break;
 145                 case ('W'):
 146                         if ( ! woptions(&curp.wflags, optarg))
 147                                 return(0);
 148                         break;
 149                 case ('V'):
 150                         version();
 151                         /* NOTREACHED */
 152                 default:
 153                         usage();
 154                         /* NOTREACHED */
 155                 }
 156
 157         argc -= optind;
 158         argv += optind;
 159
 160         bzero(&ln, sizeof(struct buf));
 161         bzero(&blk, sizeof(struct buf));
 162
 163         rc = 1;
 164
 165         if (NULL == *argv) {
 166                 curp.file = "<stdin>";
 167                 curp.fd = STDIN_FILENO;
 168                 if ( ! fdesc(&blk, &ln, &curp))
 169                         rc = 0;
 170         }
 171
 172         while (rc && *argv) {
 173                 if ( ! ffile(&blk, &ln, *argv, &curp))
 174                         rc = 0;
 175                 argv++;
 176                 if (*argv && rc) {
 177                         if (curp.lastman)
 178                                 if ( ! man_reset(curp.lastman))
 179                                         rc = 0;
 180                         if (curp.lastmdoc)
 181                                 if ( ! mdoc_reset(curp.lastmdoc))
 182                                         rc = 0;
 183                         curp.lastman = NULL;
 184                         curp.lastmdoc = NULL;
 185                 }
 186         }
 187
 188         if (blk.buf)
 189                 free(blk.buf);
 190         if (ln.buf)
 191                 free(ln.buf);
 192         if (curp.outfree)
 193                 (*curp.outfree)(curp.outdata);
 194         if (curp.mdoc)
 195                 mdoc_free(curp.mdoc);
 196         if (curp.man)
 197                 man_free(curp.man);
 198
 199         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 200 }
 201
 202
 203 __dead static void
 204 version(void)
 205 {
 206
 207         (void)printf("%s %s\n", __progname, VERSION);
 208         exit(EXIT_SUCCESS);
 209 }
 210
 211
 212 __dead static void
 213 usage(void)
 214 {
 215
 216         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 217                         "[-mformat] [-Toutput] [-Werr...]\n",
 218                         __progname);
 219         exit(EXIT_FAILURE);
 220 }
 221
 222
 223 static struct man *
 224 man_init(struct curparse *curp)
 225 {
 226         int              pflags;
 227         struct man      *man;
 228         struct man_cb    mancb;
 229
 230         mancb.man_err = merr;
 231         mancb.man_warn = manwarn;
 232
 233         /* Defaults from mandoc.1. */
 234
 235         pflags = MAN_IGN_MACRO;
 236
 237         if (curp->fflags & NO_IGN_MACRO)
 238                 pflags &= ~MAN_IGN_MACRO;
 239
 240         if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 241                 warnx("memory exhausted");
 242
 243         return(man);
 244 }
 245
 246
 247 static struct mdoc *
 248 mdoc_init(struct curparse *curp)
 249 {
 250         int              pflags;
 251         struct mdoc     *mdoc;
 252         struct mdoc_cb   mdoccb;
 253
 254         mdoccb.mdoc_err = merr;
 255         mdoccb.mdoc_warn = mdocwarn;
 256
 257         /* Defaults from mandoc.1. */
 258
 259         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 260
 261         if (curp->fflags & IGN_SCOPE)
 262                 pflags |= MDOC_IGN_SCOPE;
 263         if (curp->fflags & NO_IGN_ESCAPE)
 264                 pflags &= ~MDOC_IGN_ESCAPE;
 265         if (curp->fflags & NO_IGN_MACRO)
 266                 pflags &= ~MDOC_IGN_MACRO;
 267         if (curp->fflags & NO_IGN_CHARS)
 268                 pflags &= ~MDOC_IGN_CHARS;
 269
 270         if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb)))
 271                 warnx("memory exhausted");
 272
 273         return(mdoc);
 274 }
 275
 276
 277 static int
 278 ffile(struct buf *blk, struct buf *ln,
 279                 const char *file, struct curparse *curp)
 280 {
 281         int              c;
 282
 283         curp->file = file;
 284         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 285                 warn("%s", curp->file);
 286                 return(0);
 287         }
 288
 289         c = fdesc(blk, ln, curp);
 290
 291         if (-1 == close(curp->fd))
 292                 warn("%s", curp->file);
 293
 294         return(c);
 295 }
 296
 297
 298 static int
 299 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 300 {
 301         size_t           sz;
 302         ssize_t          ssz;
 303         struct stat      st;
 304         int              j, i, pos, lnn, comment;
 305         struct man      *man;
 306         struct mdoc     *mdoc;
 307
 308         sz = BUFSIZ;
 309         man = NULL;
 310         mdoc = NULL;
 311
 312         /*
 313          * Two buffers: ln and buf.  buf is the input buffer optimised
 314          * here for each file's block size.  ln is a line buffer.  Both
 315          * growable, hence passed in by ptr-ptr.
 316          */
 317
 318         if (-1 == fstat(curp->fd, &st))
 319                 warnx("%s", curp->file);
 320         else if ((size_t)st.st_blksize > sz)
 321                 sz = st.st_blksize;
 322
 323         if (sz > blk->sz) {
 324                 blk->buf = realloc(blk->buf, sz);
 325                 if (NULL == blk->buf) {
 326                         warn("realloc");
 327                         return(0);
 328                 }
 329                 blk->sz = sz;
 330         }
 331
 332         /* Fill buf with file blocksize. */
 333
 334         for (lnn = pos = comment = 0; ; ) {
 335                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 336                         warn("%s", curp->file);
 337                         return(0);
 338                 } else if (0 == ssz)
 339                         break;
 340
 341                 /* Parse the read block into partial or full lines. */
 342
 343                 for (i = 0; i < (int)ssz; i++) {
 344                         if (pos >= (int)ln->sz) {
 345                                 ln->sz += 256; /* Step-size. */
 346                                 ln->buf = realloc(ln->buf, ln->sz);
 347                                 if (NULL == ln->buf) {
 348                                         warn("realloc");
 349                                         return(0);
 350                                 }
 351                         }
 352
 353                         if ('\n' != blk->buf[i]) {
 354                                 if (comment)
 355                                         continue;
 356                                 ln->buf[pos++] = blk->buf[i];
 357
 358                                 /* Handle in-line `\"' comments. */
 359
 360                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 361                                         continue;
 362
 363                                 for (j = pos - 2; j >= 0; j--)
 364                                         if ('\\' != ln->buf[j])
 365                                                 break;
 366
 367                                 if ( ! ((pos - 2 - j) % 2))
 368                                         continue;
 369
 370                                 comment = 1;
 371                                 pos -= 2;
 372                                 continue;
 373                         }
 374
 375                         /* Handle escaped `\\n' newlines. */
 376
 377                         if (pos > 0 && 0 == comment &&
 378                                         '\\' == ln->buf[pos - 1]) {
 379                                 for (j = pos - 1; j >= 0; j--)
 380                                         if ('\\' != ln->buf[j])
 381                                                 break;
 382                                 if ( ! ((pos - j) % 2)) {
 383                                         pos--;
 384                                         lnn++;
 385                                         continue;
 386                                 }
 387                         }
 388
 389                         ln->buf[pos] = 0;
 390                         lnn++;
 391
 392                         /* If unset, assign parser in pset(). */
 393
 394                         if ( ! (man || mdoc) && ! pset(ln->buf,
 395                                                 pos, curp, &man, &mdoc))
 396                                 return(0);
 397
 398                         pos = comment = 0;
 399
 400                         /* Pass down into parsers. */
 401
 402                         if (man && ! man_parseln(man, lnn, ln->buf))
 403                                 return(0);
 404                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 405                                 return(0);
 406                 }
 407         }
 408
 409         /* NOTE a parser may not have been assigned, yet. */
 410
 411         if ( ! (man || mdoc)) {
 412                 warnx("%s: not a manual", curp->file);
 413                 return(0);
 414         }
 415
 416         if (mdoc && ! mdoc_endparse(mdoc))
 417                 return(0);
 418         if (man && ! man_endparse(man))
 419                 return(0);
 420
 421         /* If unset, allocate output dev now (if applicable). */
 422
 423         if ( ! (curp->outman && curp->outmdoc)) {
 424                 switch (curp->outtype) {
 425                 case (OUTT_TREE):
 426                         curp->outman = tree_man;
 427                         curp->outmdoc = tree_mdoc;
 428                         break;
 429                 case (OUTT_LINT):
 430                         break;
 431                 default:
 432                         curp->outdata = ascii_alloc();
 433                         curp->outman = terminal_man;
 434                         curp->outmdoc = terminal_mdoc;
 435                         curp->outfree = terminal_free;
 436                         break;
 437                 }
 438         }
 439
 440         /* Execute the out device, if it exists. */
 441
 442         if (man && curp->outman)
 443                 if ( ! (*curp->outman)(curp->outdata, man))
 444                         return(0);
 445         if (mdoc && curp->outmdoc)
 446                 if ( ! (*curp->outmdoc)(curp->outdata, mdoc))
 447                         return(0);
 448
 449         return(1);
 450 }
 451
 452
 453 static int
 454 pset(const char *buf, int pos, struct curparse *curp,
 455                 struct man **man, struct mdoc **mdoc)
 456 {
 457         int              i;
 458
 459         /*
 460          * Try to intuit which kind of manual parser should be used.  If
 461          * passed in by command-line (-man, -mdoc), then use that
 462          * explicitly.  If passed as -mandoc, then try to guess from the
 463          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 464          * default to -man, which is more lenient.
 465          */
 466
 467         if (buf[0] == '.') {
 468                 for (i = 1; buf[i]; i++)
 469                         if (' ' != buf[i] && '\t' != buf[i])
 470                                 break;
 471                 if (0 == buf[i])
 472                         return(1);
 473         }
 474
 475         switch (curp->inttype) {
 476         case (INTT_MDOC):
 477                 if (NULL == curp->mdoc)
 478                         curp->mdoc = mdoc_init(curp);
 479                 if (NULL == (*mdoc = curp->mdoc))
 480                         return(0);
 481                 curp->lastmdoc = *mdoc;
 482                 return(1);
 483         case (INTT_MAN):
 484                 if (NULL == curp->man)
 485                         curp->man = man_init(curp);
 486                 if (NULL == (*man = curp->man))
 487                         return(0);
 488                 curp->lastman = *man;
 489                 return(1);
 490         default:
 491                 break;
 492         }
 493
 494         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 495                 if (NULL == curp->mdoc)
 496                         curp->mdoc = mdoc_init(curp);
 497                 if (NULL == (*mdoc = curp->mdoc))
 498                         return(0);
 499                 curp->lastmdoc = *mdoc;
 500                 return(1);
 501         }
 502
 503         if (NULL == curp->man)
 504                 curp->man = man_init(curp);
 505         if (NULL == (*man = curp->man))
 506                 return(0);
 507         curp->lastman = *man;
 508         return(1);
 509 }
 510
 511
 512 static int
 513 moptions(enum intt *tflags, char *arg)
 514 {
 515
 516         if (0 == strcmp(arg, "doc"))
 517                 *tflags = INTT_MDOC;
 518         else if (0 == strcmp(arg, "andoc"))
 519                 *tflags = INTT_AUTO;
 520         else if (0 == strcmp(arg, "an"))
 521                 *tflags = INTT_MAN;
 522         else {
 523                 warnx("bad argument: -m%s", arg);
 524                 return(0);
 525         }
 526
 527         return(1);
 528 }
 529
 530
 531 static int
 532 toptions(enum outt *tflags, char *arg)
 533 {
 534
 535         if (0 == strcmp(arg, "ascii"))
 536                 *tflags = OUTT_ASCII;
 537         else if (0 == strcmp(arg, "lint"))
 538                 *tflags = OUTT_LINT;
 539         else if (0 == strcmp(arg, "tree"))
 540                 *tflags = OUTT_TREE;
 541         else {
 542                 warnx("bad argument: -T%s", arg);
 543                 return(0);
 544         }
 545
 546         return(1);
 547 }
 548
 549
 550 static int
 551 foptions(int *fflags, char *arg)
 552 {
 553         char            *v;
 554         char            *toks[6];
 555
 556         toks[0] = "ign-scope";
 557         toks[1] = "no-ign-escape";
 558         toks[2] = "no-ign-macro";
 559         toks[3] = "no-ign-chars";
 560         toks[4] = "strict";
 561         toks[5] = NULL;
 562
 563         while (*arg)
 564                 switch (getsubopt(&arg, toks, &v)) {
 565                 case (0):
 566                         *fflags |= IGN_SCOPE;
 567                         break;
 568                 case (1):
 569                         *fflags |= NO_IGN_ESCAPE;
 570                         break;
 571                 case (2):
 572                         *fflags |= NO_IGN_MACRO;
 573                         break;
 574                 case (3):
 575                         *fflags |= NO_IGN_CHARS;
 576                         break;
 577                 case (4):
 578                         *fflags |= NO_IGN_ESCAPE |
 579                                    NO_IGN_MACRO | NO_IGN_CHARS;
 580                         break;
 581                 default:
 582                         warnx("bad argument: -f%s", arg);
 583                         return(0);
 584                 }
 585
 586         return(1);
 587 }
 588
 589
 590 static int
 591 woptions(int *wflags, char *arg)
 592 {
 593         char            *v;
 594         char            *toks[5];
 595
 596         toks[0] = "all";
 597         toks[1] = "compat";
 598         toks[2] = "syntax";
 599         toks[3] = "error";
 600         toks[4] = NULL;
 601
 602         while (*arg)
 603                 switch (getsubopt(&arg, toks, &v)) {
 604                 case (0):
 605                         *wflags |= WARN_WALL;
 606                         break;
 607                 case (1):
 608                         *wflags |= WARN_WCOMPAT;
 609                         break;
 610                 case (2):
 611                         *wflags |= WARN_WSYNTAX;
 612                         break;
 613                 case (3):
 614                         *wflags |= WARN_WERR;
 615                         break;
 616                 default:
 617                         warnx("bad argument: -W%s", arg);
 618                         return(0);
 619                 }
 620
 621         return(1);
 622 }
 623
 624
 625 /* ARGSUSED */
 626 static int
 627 merr(void *arg, int line, int col, const char *msg)
 628 {
 629         struct curparse *curp;
 630
 631         curp = (struct curparse *)arg;
 632         warnx("%s:%d: error: %s (column %d)",
 633                         curp->file, line, msg, col);
 634
 635         return(0);
 636 }
 637
 638
 639 static int
 640 mdocwarn(void *arg, int line, int col,
 641                 enum mdoc_warn type, const char *msg)
 642 {
 643         struct curparse *curp;
 644         char            *wtype;
 645
 646         curp = (struct curparse *)arg;
 647         wtype = NULL;
 648
 649         switch (type) {
 650         case (WARN_COMPAT):
 651                 wtype = "compat";
 652                 if (curp->wflags & WARN_WCOMPAT)
 653                         break;
 654                 return(1);
 655         case (WARN_SYNTAX):
 656                 wtype = "syntax";
 657                 if (curp->wflags & WARN_WSYNTAX)
 658                         break;
 659                 return(1);
 660         }
 661
 662         assert(wtype);
 663         warnx("%s:%d: %s warning: %s (column %d)",
 664                         curp->file, line, wtype, msg, col);
 665
 666         if ( ! (curp->wflags & WARN_WERR))
 667                 return(1);
 668
 669         warnx("considering warnings as errors");
 670         return(0);
 671 }
 672
 673
 674 static int
 675 manwarn(void *arg, int line, int col, const char *msg)
 676 {
 677         struct curparse *curp;
 678
 679         curp = (struct curparse *)arg;
 680
 681         if ( ! (curp->wflags & WARN_WSYNTAX))
 682                 return(1);
 683
 684         warnx("%s:%d: syntax warning: %s (column %d)",
 685                         curp->file, line, msg, col);
 686
 687         if ( ! (curp->wflags & WARN_WERR))
 688                 return(1);
 689
 690         warnx("considering warnings as errors");
 691         return(0);
 692 }