main.c

   1 /*      $Id: main.c,v 1.28 2009/06/15 10:36:01 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               fstdin(struct buf *, struct buf *,
 105                                 struct curparse *);
 106 static  int               ffile(struct buf *, struct buf *,
 107                                 const char *, struct curparse *);
 108 static  int               fdesc(struct buf *, struct buf *,
 109                                 struct curparse *);
 110 static  int               pset(const char *, int, struct curparse *,
 111                                 struct man **, struct mdoc **);
 112 static  struct man       *man_init(struct curparse *);
 113 static  struct mdoc      *mdoc_init(struct curparse *);
 114 __dead  static void       version(void);
 115 __dead  static void       usage(void);
 116
 117 extern  char             *__progname;
 118
 119
 120 int
 121 main(int argc, char *argv[])
 122 {
 123         int              c, rc;
 124         struct buf       ln, blk;
 125         struct curparse  curp;
 126
 127         bzero(&curp, 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:VW:T:")))
 134                 switch (c) {
 135                 case ('f'):
 136                         if ( ! foptions(&curp.fflags, optarg))
 137                                 return(0);
 138                         break;
 139                 case ('m'):
 140                         if ( ! moptions(&curp.inttype, optarg))
 141                                 return(0);
 142                         break;
 143                 case ('T'):
 144                         if ( ! toptions(&curp.outtype, optarg))
 145                                 return(0);
 146                         break;
 147                 case ('W'):
 148                         if ( ! woptions(&curp.wflags, optarg))
 149                                 return(0);
 150                         break;
 151                 case ('V'):
 152                         version();
 153                         /* NOTREACHED */
 154                 default:
 155                         usage();
 156                         /* NOTREACHED */
 157                 }
 158
 159         argc -= optind;
 160         argv += optind;
 161
 162         /* Configure buffers. */
 163
 164         bzero(&ln, sizeof(struct buf));
 165         bzero(&blk, sizeof(struct buf));
 166
 167         rc = 1;
 168
 169         if (NULL == *argv)
 170                 if ( ! fstdin(&blk, &ln, &curp))
 171                         rc = 0;
 172
 173         while (rc && *argv) {
 174                 if ( ! ffile(&blk, &ln, *argv, &curp))
 175                         rc = 0;
 176                 argv++;
 177                 if (*argv && rc) {
 178                         if (curp.lastman)
 179                                 if ( ! man_reset(curp.lastman))
 180                                         rc = 0;
 181                         if (curp.lastmdoc)
 182                                 if ( ! mdoc_reset(curp.lastmdoc))
 183                                         rc = 0;
 184                         curp.lastman = NULL;
 185                         curp.lastmdoc = NULL;
 186                 }
 187         }
 188
 189         if (blk.buf)
 190                 free(blk.buf);
 191         if (ln.buf)
 192                 free(ln.buf);
 193
 194         /* TODO: have a curp_free routine. */
 195         if (curp.outfree)
 196                 (*curp.outfree)(curp.outdata);
 197         if (curp.mdoc)
 198                 mdoc_free(curp.mdoc);
 199         if (curp.man)
 200                 man_free(curp.man);
 201
 202         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 203 }
 204
 205
 206 __dead static void
 207 version(void)
 208 {
 209
 210         (void)printf("%s %s\n", __progname, VERSION);
 211         exit(EXIT_SUCCESS);
 212 }
 213
 214
 215 __dead static void
 216 usage(void)
 217 {
 218
 219         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 220                         "[-mformat] [-Toutput] [-Werr...]\n",
 221                         __progname);
 222         exit(EXIT_FAILURE);
 223 }
 224
 225
 226 static struct man *
 227 man_init(struct curparse *curp)
 228 {
 229         int              pflags;
 230         struct man      *man;
 231         struct man_cb    mancb;
 232
 233         mancb.man_err = merr;
 234         mancb.man_warn = manwarn;
 235
 236         /*
 237          * Default behaviour is to ignore unknown macros.  This is
 238          * specified in mandoc.1.
 239          */
 240
 241         pflags = MAN_IGN_MACRO;
 242
 243         /* Override default behaviour... */
 244
 245         if (curp->fflags & NO_IGN_MACRO)
 246                 pflags &= ~MAN_IGN_MACRO;
 247
 248         if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 249                 warnx("memory exhausted");
 250
 251         return(man);
 252 }
 253
 254
 255 static struct mdoc *
 256 mdoc_init(struct curparse *curp)
 257 {
 258         int              pflags;
 259         struct mdoc     *mdoc;
 260         struct mdoc_cb   mdoccb;
 261
 262         mdoccb.mdoc_err = merr;
 263         mdoccb.mdoc_warn = mdocwarn;
 264
 265         /*
 266          * Default behaviour is to ignore unknown macros, escape
 267          * sequences and characters (very liberal).  This is specified
 268          * in mandoc.1.
 269          */
 270
 271         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 272
 273         /* Override default behaviour... */
 274
 275         if (curp->fflags & IGN_SCOPE)
 276                 pflags |= MDOC_IGN_SCOPE;
 277         if (curp->fflags & NO_IGN_ESCAPE)
 278                 pflags &= ~MDOC_IGN_ESCAPE;
 279         if (curp->fflags & NO_IGN_MACRO)
 280                 pflags &= ~MDOC_IGN_MACRO;
 281         if (curp->fflags & NO_IGN_CHARS)
 282                 pflags &= ~MDOC_IGN_CHARS;
 283
 284         if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb)))
 285                 warnx("memory exhausted");
 286
 287         return(mdoc);
 288 }
 289
 290
 291 static int
 292 fstdin(struct buf *blk, struct buf *ln, struct curparse *curp)
 293 {
 294
 295         curp->file = "<stdin>";
 296         curp->fd = STDIN_FILENO;
 297         return(fdesc(blk, ln, curp));
 298 }
 299
 300
 301 static int
 302 ffile(struct buf *blk, struct buf *ln,
 303                 const char *file, struct curparse *curp)
 304 {
 305         int              c;
 306
 307         curp->file = file;
 308         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 309                 warn("%s", curp->file);
 310                 return(0);
 311         }
 312
 313         c = fdesc(blk, ln, curp);
 314
 315         if (-1 == close(curp->fd))
 316                 warn("%s", curp->file);
 317
 318         return(c);
 319 }
 320
 321
 322 static int
 323 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 324 {
 325         size_t           sz;
 326         ssize_t          ssz;
 327         struct stat      st;
 328         int              j, i, pos, lnn;
 329         struct man      *man;
 330         struct mdoc     *mdoc;
 331
 332         sz = BUFSIZ;
 333         man = NULL;
 334         mdoc = NULL;
 335
 336         /*
 337          * Two buffers: ln and buf.  buf is the input buffer optimised
 338          * here for each file's block size.  ln is a line buffer.  Both
 339          * growable, hence passed in by ptr-ptr.
 340          */
 341
 342         if (-1 == fstat(curp->fd, &st))
 343                 warnx("%s", curp->file);
 344         else if ((size_t)st.st_blksize > sz)
 345                 sz = st.st_blksize;
 346
 347         if (sz > blk->sz) {
 348                 blk->buf = realloc(blk->buf, sz);
 349                 if (NULL == blk->buf) {
 350                         warn("realloc");
 351                         return(0);
 352                 }
 353                 blk->sz = sz;
 354         }
 355
 356         /* Fill buf with file blocksize. */
 357
 358         for (lnn = 0, pos = 0; ; ) {
 359                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 360                         warn("%s", curp->file);
 361                         return(0);
 362                 } else if (0 == ssz)
 363                         break;
 364
 365                 /* Parse the read block into partial or full lines. */
 366
 367                 for (i = 0; i < (int)ssz; i++) {
 368                         if (pos >= (int)ln->sz) {
 369                                 ln->sz += 256; /* Step-size. */
 370                                 ln->buf = realloc(ln->buf, ln->sz);
 371                                 if (NULL == ln->buf) {
 372                                         warn("realloc");
 373                                         return(0);
 374                                 }
 375                         }
 376
 377                         if ('\n' != blk->buf[i]) {
 378                                 ln->buf[pos++] = blk->buf[i];
 379                                 continue;
 380                         }
 381
 382                         /* Check for CPP-escaped newline. */
 383
 384                         if (pos > 0 && '\\' == ln->buf[pos - 1]) {
 385                                 for (j = pos - 1; j >= 0; j--)
 386                                         if ('\\' != ln->buf[j])
 387                                                 break;
 388
 389                                 if ( ! ((pos - j) % 2)) {
 390                                         pos--;
 391                                         lnn++;
 392                                         continue;
 393                                 }
 394                         }
 395
 396                         ln->buf[pos] = 0;
 397                         lnn++;
 398
 399                         /*
 400                          * If no manual parser has been assigned, then
 401                          * try to assign one in pset(), which may do
 402                          * nothing at all.  After this, parse the manual
 403                          * line accordingly.
 404                          */
 405
 406                         if ( ! (man || mdoc) && ! pset(ln->buf,
 407                                                 pos, curp, &man, &mdoc))
 408                                 return(0);
 409
 410                         pos = 0;
 411
 412                         if (man && ! man_parseln(man, lnn, ln->buf))
 413                                 return(0);
 414                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 415                                 return(0);
 416                 }
 417         }
 418
 419         /* Note that a parser may not have been assigned, yet. */
 420
 421         if ( ! (man || mdoc)) {
 422                 warnx("%s: not a manual", curp->file);
 423                 return(0);
 424         }
 425
 426         if (mdoc && ! mdoc_endparse(mdoc))
 427                 return(0);
 428         if (man && ! man_endparse(man))
 429                 return(0);
 430
 431         /*
 432          * If an output device hasn't been allocated, see if we should
 433          * do so now.  Note that not all outtypes have functions, so
 434          * this switch statement may be superfluous, but it's
 435          * low-overhead enough not to matter very much.
 436          */
 437
 438         if ( ! (curp->outman && curp->outmdoc)) {
 439                 switch (curp->outtype) {
 440                 case (OUTT_TREE):
 441                         curp->outman = tree_man;
 442                         curp->outmdoc = tree_mdoc;
 443                         break;
 444                 case (OUTT_LINT):
 445                         break;
 446                 default:
 447                         curp->outdata = ascii_alloc();
 448                         curp->outman = terminal_man;
 449                         curp->outmdoc = terminal_mdoc;
 450                         curp->outfree = terminal_free;
 451                         break;
 452                 }
 453         }
 454
 455         /* Execute the out device, if it exists. */
 456
 457         if (man && curp->outman)
 458                 if ( ! (*curp->outman)(curp->outdata, man))
 459                         return(0);
 460         if (mdoc && curp->outmdoc)
 461                 if ( ! (*curp->outmdoc)(curp->outdata, mdoc))
 462                         return(0);
 463
 464         return(1);
 465 }
 466
 467
 468 static int
 469 pset(const char *buf, int pos, struct curparse *curp,
 470                 struct man **man, struct mdoc **mdoc)
 471 {
 472
 473         /*
 474          * Try to intuit which kind of manual parser should be used.  If
 475          * passed in by command-line (-man, -mdoc), then use that
 476          * explicitly.  If passed as -mandoc, then try to guess from the
 477          * line: either skip comments, use -mdoc when finding `.Dt', or
 478          * default to -man, which is more lenient.
 479          */
 480
 481         if (pos >= 3 && 0 == memcmp(buf, ".\\\"", 3))
 482                 return(1);
 483
 484         switch (curp->inttype) {
 485         case (INTT_MDOC):
 486                 if (NULL == curp->mdoc)
 487                         curp->mdoc = mdoc_init(curp);
 488                 if (NULL == (*mdoc = curp->mdoc))
 489                         return(0);
 490                 curp->lastmdoc = *mdoc;
 491                 return(1);
 492         case (INTT_MAN):
 493                 if (NULL == curp->man)
 494                         curp->man = man_init(curp);
 495                 if (NULL == (*man = curp->man))
 496                         return(0);
 497                 curp->lastman = *man;
 498                 return(1);
 499         default:
 500                 break;
 501         }
 502
 503         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 504                 if (NULL == curp->mdoc)
 505                         curp->mdoc = mdoc_init(curp);
 506                 if (NULL == (*mdoc = curp->mdoc))
 507                         return(0);
 508                 curp->lastmdoc = *mdoc;
 509                 return(1);
 510         }
 511
 512         if (NULL == curp->man)
 513                 curp->man = man_init(curp);
 514         if (NULL == (*man = curp->man))
 515                 return(0);
 516         curp->lastman = *man;
 517         return(1);
 518 }
 519
 520
 521 static int
 522 moptions(enum intt *tflags, char *arg)
 523 {
 524
 525         if (0 == strcmp(arg, "doc"))
 526                 *tflags = INTT_MDOC;
 527         else if (0 == strcmp(arg, "andoc"))
 528                 *tflags = INTT_AUTO;
 529         else if (0 == strcmp(arg, "an"))
 530                 *tflags = INTT_MAN;
 531         else {
 532                 warnx("bad argument: -m%s", arg);
 533                 return(0);
 534         }
 535
 536         return(1);
 537 }
 538
 539
 540 static int
 541 toptions(enum outt *tflags, char *arg)
 542 {
 543
 544         if (0 == strcmp(arg, "ascii"))
 545                 *tflags = OUTT_ASCII;
 546         else if (0 == strcmp(arg, "lint"))
 547                 *tflags = OUTT_LINT;
 548         else if (0 == strcmp(arg, "tree"))
 549                 *tflags = OUTT_TREE;
 550         else {
 551                 warnx("bad argument: -T%s", arg);
 552                 return(0);
 553         }
 554
 555         return(1);
 556 }
 557
 558
 559 /*
 560  * Parse out the options for [-fopt...] setting compiler options.  These
 561  * can be comma-delimited or called again.
 562  */
 563 static int
 564 foptions(int *fflags, char *arg)
 565 {
 566         char            *v;
 567         char            *toks[6];
 568
 569         toks[0] = "ign-scope";
 570         toks[1] = "no-ign-escape";
 571         toks[2] = "no-ign-macro";
 572         toks[3] = "no-ign-chars";
 573         toks[4] = "strict";
 574         toks[5] = NULL;
 575
 576         while (*arg)
 577                 switch (getsubopt(&arg, toks, &v)) {
 578                 case (0):
 579                         *fflags |= IGN_SCOPE;
 580                         break;
 581                 case (1):
 582                         *fflags |= NO_IGN_ESCAPE;
 583                         break;
 584                 case (2):
 585                         *fflags |= NO_IGN_MACRO;
 586                         break;
 587                 case (3):
 588                         *fflags |= NO_IGN_CHARS;
 589                         break;
 590                 case (4):
 591                         *fflags |= NO_IGN_ESCAPE |
 592                                    NO_IGN_MACRO | NO_IGN_CHARS;
 593                         break;
 594                 default:
 595                         warnx("bad argument: -f%s", arg);
 596                         return(0);
 597                 }
 598
 599         return(1);
 600 }
 601
 602
 603 /*
 604  * Parse out the options for [-Werr...], which sets warning modes.
 605  * These can be comma-delimited or called again.
 606  */
 607 static int
 608 woptions(int *wflags, char *arg)
 609 {
 610         char            *v;
 611         char            *toks[5];
 612
 613         toks[0] = "all";
 614         toks[1] = "compat";
 615         toks[2] = "syntax";
 616         toks[3] = "error";
 617         toks[4] = NULL;
 618
 619         while (*arg)
 620                 switch (getsubopt(&arg, toks, &v)) {
 621                 case (0):
 622                         *wflags |= WARN_WALL;
 623                         break;
 624                 case (1):
 625                         *wflags |= WARN_WCOMPAT;
 626                         break;
 627                 case (2):
 628                         *wflags |= WARN_WSYNTAX;
 629                         break;
 630                 case (3):
 631                         *wflags |= WARN_WERR;
 632                         break;
 633                 default:
 634                         warnx("bad argument: -W%s", arg);
 635                         return(0);
 636                 }
 637
 638         return(1);
 639 }
 640
 641
 642 /* ARGSUSED */
 643 static int
 644 merr(void *arg, int line, int col, const char *msg)
 645 {
 646         struct curparse *curp;
 647
 648         curp = (struct curparse *)arg;
 649         warnx("%s:%d: error: %s (column %d)",
 650                         curp->file, line, msg, col);
 651
 652         /* Always exit on errors... */
 653         return(0);
 654 }
 655
 656
 657 static int
 658 mdocwarn(void *arg, int line, int col,
 659                 enum mdoc_warn type, const char *msg)
 660 {
 661         struct curparse *curp;
 662         char            *wtype;
 663
 664         curp = (struct curparse *)arg;
 665         wtype = NULL;
 666
 667         switch (type) {
 668         case (WARN_COMPAT):
 669                 wtype = "compat";
 670                 if (curp->wflags & WARN_WCOMPAT)
 671                         break;
 672                 return(1);
 673         case (WARN_SYNTAX):
 674                 wtype = "syntax";
 675                 if (curp->wflags & WARN_WSYNTAX)
 676                         break;
 677                 return(1);
 678         }
 679
 680         assert(wtype);
 681         warnx("%s:%d: %s warning: %s (column %d)",
 682                         curp->file, line, wtype, msg, col);
 683
 684         if ( ! (curp->wflags & WARN_WERR))
 685                 return(1);
 686
 687         /*
 688          * If the -Werror flag is passed in, as in gcc, then all
 689          * warnings are considered as errors.
 690          */
 691
 692         warnx("%s: considering warnings as errors",
 693                         __progname);
 694         return(0);
 695 }
 696
 697
 698 static int
 699 manwarn(void *arg, int line, int col, const char *msg)
 700 {
 701         struct curparse *curp;
 702
 703         curp = (struct curparse *)arg;
 704
 705         if ( ! (curp->wflags & WARN_WSYNTAX))
 706                 return(1);
 707
 708         warnx("%s:%d: syntax warning: %s (column %d)",
 709                         curp->file, line, msg, col);
 710
 711         if ( ! (curp->wflags & WARN_WERR))
 712                 return(1);
 713
 714         /*
 715          * If the -Werror flag is passed in, as in gcc, then all
 716          * warnings are considered as errors.
 717          */
 718
 719         warnx("%s: considering warnings as errors",
 720                         __progname);
 721         return(0);
 722 }