main.c

   1 /*      $Id: main.c,v 1.65 2010/05/15 04:46:10 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 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <sys/stat.h>
  22
  23 #include <assert.h>
  24 #include <fcntl.h>
  25 #include <stdio.h>
  26 #include <stdint.h>
  27 #include <stdlib.h>
  28 #include <string.h>
  29 #include <unistd.h>
  30
  31 #include "mdoc.h"
  32 #include "man.h"
  33 #include "main.h"
  34
  35 #define UNCONST(a)      ((void *)(uintptr_t)(const void *)(a))
  36
  37 /* FIXME: Intel's compiler?  LLVM?  pcc?  */
  38
  39 #if !defined(__GNUC__) || (__GNUC__ < 2)
  40 # if !defined(lint)
  41 #  define __attribute__(x)
  42 # endif
  43 #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */
  44
  45 typedef void            (*out_mdoc)(void *, const struct mdoc *);
  46 typedef void            (*out_man)(void *, const struct man *);
  47 typedef void            (*out_free)(void *);
  48
  49 struct  buf {
  50         char             *buf;
  51         size_t            sz;
  52 };
  53
  54 enum    intt {
  55         INTT_AUTO,
  56         INTT_MDOC,
  57         INTT_MAN
  58 };
  59
  60 enum    outt {
  61         OUTT_ASCII = 0,
  62         OUTT_TREE,
  63         OUTT_HTML,
  64         OUTT_XHTML,
  65         OUTT_LINT
  66 };
  67
  68 struct  curparse {
  69         const char       *file;         /* Current parse. */
  70         int               fd;           /* Current parse. */
  71         int               wflags;
  72 #define WARN_WALL        (1 << 0)       /* All-warnings mask. */
  73 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  74         int               fflags;
  75 #define FL_IGN_SCOPE     (1 << 0)       /* Ignore scope errors. */
  76 #define FL_NIGN_ESCAPE   (1 << 1)       /* Don't ignore bad escapes. */
  77 #define FL_NIGN_MACRO    (1 << 2)       /* Don't ignore bad macros. */
  78 #define FL_IGN_ERRORS    (1 << 4)       /* Ignore failed parse. */
  79         enum intt         inttype;      /* Input parsers... */
  80         struct man       *man;
  81         struct mdoc      *mdoc;
  82         enum outt         outtype;      /* Output devices... */
  83         out_mdoc          outmdoc;
  84         out_man           outman;
  85         out_free          outfree;
  86         void             *outdata;
  87         char              outopts[BUFSIZ];
  88 };
  89
  90 #define FL_STRICT         FL_NIGN_ESCAPE | \
  91                           FL_NIGN_MACRO
  92
  93 static  int               foptions(int *, char *);
  94 static  int               toptions(struct curparse *, char *);
  95 static  int               moptions(enum intt *, char *);
  96 static  int               woptions(int *, char *);
  97 static  int               merr(void *, int, int, const char *);
  98 static  int               mwarn(void *, int, int, const char *);
  99 static  void              ffile(struct buf *, struct buf *,
 100                                 const char *, struct curparse *);
 101 static  void              fdesc(struct buf *, struct buf *,
 102                                 struct curparse *);
 103 static  int               pset(const char *, int, struct curparse *,
 104                                 struct man **, struct mdoc **);
 105 static  struct man       *man_init(struct curparse *);
 106 static  struct mdoc      *mdoc_init(struct curparse *);
 107 static  void              version(void) __attribute__((noreturn));
 108 static  void              usage(void) __attribute__((noreturn));
 109
 110 static  const char       *progname;
 111 static  int               with_error, with_warning;
 112
 113 int
 114 main(int argc, char *argv[])
 115 {
 116         int              c;
 117         struct buf       ln, blk;
 118         struct curparse  curp;
 119
 120         progname = strrchr(argv[0], '/');
 121         if (progname == NULL)
 122                 progname = argv[0];
 123         else
 124                 ++progname;
 125
 126         memset(&curp, 0, sizeof(struct curparse));
 127
 128         curp.inttype = INTT_AUTO;
 129         curp.outtype = OUTT_ASCII;
 130
 131         /* LINTED */
 132         while (-1 != (c = getopt(argc, argv, "f:m:O:T:VW:")))
 133                 switch (c) {
 134                 case ('f'):
 135                         if ( ! foptions(&curp.fflags, optarg))
 136                                 return(EXIT_FAILURE);
 137                         break;
 138                 case ('m'):
 139                         if ( ! moptions(&curp.inttype, optarg))
 140                                 return(EXIT_FAILURE);
 141                         break;
 142                 case ('O'):
 143                         (void)strlcat(curp.outopts, optarg, BUFSIZ);
 144                         (void)strlcat(curp.outopts, ",", BUFSIZ);
 145                         break;
 146                 case ('T'):
 147                         if ( ! toptions(&curp, optarg))
 148                                 return(EXIT_FAILURE);
 149                         break;
 150                 case ('W'):
 151                         if ( ! woptions(&curp.wflags, optarg))
 152                                 return(EXIT_FAILURE);
 153                         break;
 154                 case ('V'):
 155                         version();
 156                         /* NOTREACHED */
 157                 default:
 158                         usage();
 159                         /* NOTREACHED */
 160                 }
 161
 162         argc -= optind;
 163         argv += optind;
 164
 165         memset(&ln, 0, sizeof(struct buf));
 166         memset(&blk, 0, sizeof(struct buf));
 167
 168         if (NULL == *argv) {
 169                 curp.file = "<stdin>";
 170                 curp.fd = STDIN_FILENO;
 171
 172                 fdesc(&blk, &ln, &curp);
 173         }
 174
 175         while (*argv) {
 176                 ffile(&blk, &ln, *argv, &curp);
 177
 178                 if (with_error && !(curp.fflags & FL_IGN_ERRORS))
 179                         break;
 180                 ++argv;
 181         }
 182
 183         if (blk.buf)
 184                 free(blk.buf);
 185         if (ln.buf)
 186                 free(ln.buf);
 187         if (curp.outfree)
 188                 (*curp.outfree)(curp.outdata);
 189
 190         return((with_warning || with_error) ? EXIT_FAILURE :  EXIT_SUCCESS );
 191 }
 192
 193
 194 static void
 195 version(void)
 196 {
 197
 198         (void)printf("%s %s\n", progname, VERSION);
 199         exit(EXIT_SUCCESS);
 200 }
 201
 202
 203 static void
 204 usage(void)
 205 {
 206
 207         (void)fprintf(stderr, "usage: %s [-V] [-foption] "
 208                         "[-mformat] [-Ooption] [-Toutput] "
 209                         "[-Werr] [file...]\n", progname);
 210         exit(EXIT_FAILURE);
 211 }
 212
 213
 214 static struct man *
 215 man_init(struct curparse *curp)
 216 {
 217         int              pflags;
 218         struct man_cb    mancb;
 219
 220         mancb.man_err = merr;
 221         mancb.man_warn = mwarn;
 222
 223         /* Defaults from mandoc.1. */
 224
 225         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE;
 226
 227         if (curp->fflags & FL_NIGN_MACRO)
 228                 pflags &= ~MAN_IGN_MACRO;
 229         if (curp->fflags & FL_NIGN_ESCAPE)
 230                 pflags &= ~MAN_IGN_ESCAPE;
 231
 232         return(man_alloc(curp, pflags, &mancb));
 233 }
 234
 235
 236 static struct mdoc *
 237 mdoc_init(struct curparse *curp)
 238 {
 239         int              pflags;
 240         struct mdoc_cb   mdoccb;
 241
 242         mdoccb.mdoc_err = merr;
 243         mdoccb.mdoc_warn = mwarn;
 244
 245         /* Defaults from mandoc.1. */
 246
 247         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE;
 248
 249         if (curp->fflags & FL_IGN_SCOPE)
 250                 pflags |= MDOC_IGN_SCOPE;
 251         if (curp->fflags & FL_NIGN_ESCAPE)
 252                 pflags &= ~MDOC_IGN_ESCAPE;
 253         if (curp->fflags & FL_NIGN_MACRO)
 254                 pflags &= ~MDOC_IGN_MACRO;
 255
 256         return(mdoc_alloc(curp, pflags, &mdoccb));
 257 }
 258
 259
 260 static void
 261 ffile(struct buf *blk, struct buf *ln,
 262                 const char *file, struct curparse *curp)
 263 {
 264
 265         curp->file = file;
 266         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 267                 perror(curp->file);
 268                 with_error = 1;
 269                 return;
 270         }
 271
 272         fdesc(blk, ln, curp);
 273
 274         if (-1 == close(curp->fd))
 275                 perror(curp->file);
 276 }
 277
 278
 279 static void
 280 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 281 {
 282         size_t           sz;
 283         ssize_t          ssz;
 284         struct stat      st;
 285         int              j, i, pos, lnn, comment;
 286         struct man      *man;
 287         struct mdoc     *mdoc;
 288
 289         sz = BUFSIZ;
 290         man = NULL;
 291         mdoc = NULL;
 292
 293         /*
 294          * Two buffers: ln and buf.  buf is the input buffer optimised
 295          * here for each file's block size.  ln is a line buffer.  Both
 296          * growable, hence passed in by ptr-ptr.
 297          */
 298
 299         if (-1 == fstat(curp->fd, &st)) {
 300                 perror(curp->file);
 301                 with_error = 1;
 302                 return;
 303         }
 304         if ((size_t)st.st_blksize > sz)
 305                 sz = st.st_blksize;
 306
 307         if (sz > blk->sz) {
 308                 void *buf = realloc(blk->buf, sz);
 309
 310                 if (NULL == buf) {
 311                         perror(NULL);
 312                         with_error = 1;
 313                         return;
 314                 }
 315                 blk->buf = buf;
 316                 blk->sz = sz;
 317         }
 318
 319         /* Fill buf with file blocksize. */
 320
 321         for (lnn = pos = comment = 0; ; ) {
 322                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 323                         perror(curp->file);
 324                         goto bailout;
 325                 } else if (0 == ssz)
 326                         break;
 327
 328                 /* Parse the read block into partial or full lines. */
 329
 330                 for (i = 0; i < (int)ssz; i++) {
 331                         if (pos >= (int)ln->sz) {
 332                                 ln->sz += 256; /* Step-size. */
 333                                 ln->buf = realloc(ln->buf, ln->sz);
 334                                 if (NULL == ln->buf) {
 335                                         perror(NULL);
 336                                         goto bailout;
 337                                 }
 338                         }
 339
 340                         if ('\n' != blk->buf[i]) {
 341                                 if (comment)
 342                                         continue;
 343                                 ln->buf[pos++] = blk->buf[i];
 344
 345                                 /* Handle in-line `\"' comments. */
 346
 347                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 348                                         continue;
 349
 350                                 for (j = pos - 2; j >= 0; j--)
 351                                         if ('\\' != ln->buf[j])
 352                                                 break;
 353
 354                                 if ( ! ((pos - 2 - j) % 2))
 355                                         continue;
 356
 357                                 comment = 1;
 358                                 pos -= 2;
 359                                 for (; pos > 0; --pos) {
 360                                         if (ln->buf[pos - 1] != ' ')
 361                                                 break;
 362                                         if (pos > 2 && ln->buf[pos - 2] == '\\')
 363                                                 break;
 364                                 }
 365                                 continue;
 366                         }
 367
 368                         /* Handle escaped `\\n' newlines. */
 369
 370                         if (pos > 0 && 0 == comment &&
 371                                         '\\' == ln->buf[pos - 1]) {
 372                                 for (j = pos - 1; j >= 0; j--)
 373                                         if ('\\' != ln->buf[j])
 374                                                 break;
 375                                 if ( ! ((pos - j) % 2)) {
 376                                         pos--;
 377                                         lnn++;
 378                                         continue;
 379                                 }
 380                         }
 381
 382                         ln->buf[pos] = 0;
 383                         lnn++;
 384
 385                         /* If unset, assign parser in pset(). */
 386
 387                         if ( ! (man || mdoc) && ! pset(ln->buf,
 388                                                 pos, curp, &man, &mdoc))
 389                                 goto bailout;
 390
 391                         pos = comment = 0;
 392
 393                         /* Pass down into parsers. */
 394
 395                         if (man && ! man_parseln(man, lnn, ln->buf))
 396                                 goto bailout;
 397                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 398                                 goto bailout;
 399                 }
 400         }
 401
 402         /* NOTE a parser may not have been assigned, yet. */
 403
 404         if ( ! (man || mdoc)) {
 405                 fprintf(stderr, "%s: Not a manual\n", curp->file);
 406                 goto bailout;
 407         }
 408
 409         if (mdoc && ! mdoc_endparse(mdoc))
 410                 goto bailout;
 411         if (man && ! man_endparse(man))
 412                 goto bailout;
 413
 414         /* If unset, allocate output dev now (if applicable). */
 415
 416         if ( ! (curp->outman && curp->outmdoc)) {
 417                 switch (curp->outtype) {
 418                 case (OUTT_XHTML):
 419                         curp->outdata = xhtml_alloc(curp->outopts);
 420                         curp->outman = html_man;
 421                         curp->outmdoc = html_mdoc;
 422                         curp->outfree = html_free;
 423                         break;
 424                 case (OUTT_HTML):
 425                         curp->outdata = html_alloc(curp->outopts);
 426                         curp->outman = html_man;
 427                         curp->outmdoc = html_mdoc;
 428                         curp->outfree = html_free;
 429                         break;
 430                 case (OUTT_TREE):
 431                         curp->outman = tree_man;
 432                         curp->outmdoc = tree_mdoc;
 433                         break;
 434                 case (OUTT_LINT):
 435                         break;
 436                 default:
 437                         curp->outdata = ascii_alloc();
 438                         curp->outman = terminal_man;
 439                         curp->outmdoc = terminal_mdoc;
 440                         curp->outfree = terminal_free;
 441                         break;
 442                 }
 443         }
 444
 445         /* Execute the out device, if it exists. */
 446
 447         if (man && curp->outman)
 448                 (*curp->outman)(curp->outdata, man);
 449         if (mdoc && curp->outmdoc)
 450                 (*curp->outmdoc)(curp->outdata, mdoc);
 451
 452  cleanup:
 453         if (curp->mdoc) {
 454                 mdoc_free(curp->mdoc);
 455                 curp->mdoc = NULL;
 456         }
 457         if (curp->man) {
 458                 man_free(curp->man);
 459                 curp->man = NULL;
 460         }
 461         return;
 462
 463  bailout:
 464         with_error = 1;
 465         goto cleanup;
 466 }
 467
 468
 469 static int
 470 pset(const char *buf, int pos, struct curparse *curp,
 471                 struct man **man, struct mdoc **mdoc)
 472 {
 473         int              i;
 474
 475         /*
 476          * Try to intuit which kind of manual parser should be used.  If
 477          * passed in by command-line (-man, -mdoc), then use that
 478          * explicitly.  If passed as -mandoc, then try to guess from the
 479          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 480          * default to -man, which is more lenient.
 481          */
 482
 483         if (buf[0] == '.') {
 484                 for (i = 1; buf[i]; i++)
 485                         if (' ' != buf[i] && '\t' != buf[i])
 486                                 break;
 487                 if (0 == buf[i])
 488                         return(1);
 489         }
 490
 491         switch (curp->inttype) {
 492         case (INTT_MDOC):
 493                 if (NULL == curp->mdoc)
 494                         curp->mdoc = mdoc_init(curp);
 495                 if (NULL == (*mdoc = curp->mdoc))
 496                         return(0);
 497                 return(1);
 498         case (INTT_MAN):
 499                 if (NULL == curp->man)
 500                         curp->man = man_init(curp);
 501                 if (NULL == (*man = curp->man))
 502                         return(0);
 503                 return(1);
 504         default:
 505                 break;
 506         }
 507
 508         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 509                 if (NULL == curp->mdoc)
 510                         curp->mdoc = mdoc_init(curp);
 511                 if (NULL == (*mdoc = curp->mdoc))
 512                         return(0);
 513                 return(1);
 514         }
 515
 516         if (NULL == curp->man)
 517                 curp->man = man_init(curp);
 518         if (NULL == (*man = curp->man))
 519                 return(0);
 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\n", arg);
 536                 return(0);
 537         }
 538
 539         return(1);
 540 }
 541
 542
 543 static int
 544 toptions(struct curparse *curp, char *arg)
 545 {
 546
 547         if (0 == strcmp(arg, "ascii"))
 548                 curp->outtype = OUTT_ASCII;
 549         else if (0 == strcmp(arg, "lint")) {
 550                 curp->outtype = OUTT_LINT;
 551                 curp->wflags |= WARN_WALL;
 552                 curp->fflags |= FL_STRICT;
 553         }
 554         else if (0 == strcmp(arg, "tree"))
 555                 curp->outtype = OUTT_TREE;
 556         else if (0 == strcmp(arg, "html"))
 557                 curp->outtype = OUTT_HTML;
 558         else if (0 == strcmp(arg, "xhtml"))
 559                 curp->outtype = OUTT_XHTML;
 560         else {
 561                 fprintf(stderr, "%s: Bad argument\n", arg);
 562                 return(0);
 563         }
 564
 565         return(1);
 566 }
 567
 568
 569 static int
 570 foptions(int *fflags, char *arg)
 571 {
 572         char            *v, *o;
 573         const char      *toks[8];
 574
 575         toks[0] = "ign-scope";
 576         toks[1] = "no-ign-escape";
 577         toks[2] = "no-ign-macro";
 578         toks[3] = "ign-errors";
 579         toks[4] = "strict";
 580         toks[5] = "ign-escape";
 581         toks[6] = NULL;
 582
 583         while (*arg) {
 584                 o = arg;
 585                 switch (getsubopt(&arg, UNCONST(toks), &v)) {
 586                 case (0):
 587                         *fflags |= FL_IGN_SCOPE;
 588                         break;
 589                 case (1):
 590                         *fflags |= FL_NIGN_ESCAPE;
 591                         break;
 592                 case (2):
 593                         *fflags |= FL_NIGN_MACRO;
 594                         break;
 595                 case (3):
 596                         *fflags |= FL_IGN_ERRORS;
 597                         break;
 598                 case (4):
 599                         *fflags |= FL_STRICT;
 600                         break;
 601                 case (5):
 602                         *fflags &= ~FL_NIGN_ESCAPE;
 603                         break;
 604                 default:
 605                         fprintf(stderr, "%s: Bad argument\n", 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\n", 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         with_error = 1;
 655
 656         return(0);
 657 }
 658
 659
 660 static int
 661 mwarn(void *arg, int line, int col, const char *msg)
 662 {
 663         struct curparse *curp;
 664
 665         curp = (struct curparse *)arg;
 666
 667         if ( ! (curp->wflags & WARN_WALL))
 668                 return(1);
 669
 670         (void)fprintf(stderr, "%s:%d:%d: warning: %s\n",
 671                         curp->file, line, col + 1, msg);
 672
 673         with_warning = 1;
 674         if (curp->wflags & WARN_WERR) {
 675                 with_error = 1;
 676                 return(0);
 677         }
 678
 679         return(1);
 680 }
 681