main.c

   1 /*      $Id: main.c,v 1.34 2009/07/04 11:10:36 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(EXIT_FAILURE);
 136                         break;
 137                 case ('m'):
 138                         if ( ! moptions(&curp.inttype, optarg))
 139                                 return(EXIT_FAILURE);
 140                         break;
 141                 case ('T'):
 142                         if ( ! toptions(&curp.outtype, optarg))
 143                                 return(EXIT_FAILURE);
 144                         break;
 145                 case ('W'):
 146                         if ( ! woptions(&curp.wflags, optarg))
 147                                 return(EXIT_FAILURE);
 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 | MAN_IGN_ESCAPE | MAN_IGN_CHARS;
 236
 237         if (curp->fflags & NO_IGN_MACRO)
 238                 pflags &= ~MAN_IGN_MACRO;
 239         if (curp->fflags & NO_IGN_CHARS)
 240                 pflags &= ~MAN_IGN_CHARS;
 241         if (curp->fflags & NO_IGN_ESCAPE)
 242                 pflags &= ~MAN_IGN_ESCAPE;
 243
 244         if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 245                 warnx("memory exhausted");
 246
 247         return(man);
 248 }
 249
 250
 251 static struct mdoc *
 252 mdoc_init(struct curparse *curp)
 253 {
 254         int              pflags;
 255         struct mdoc     *mdoc;
 256         struct mdoc_cb   mdoccb;
 257
 258         mdoccb.mdoc_err = merr;
 259         mdoccb.mdoc_warn = mdocwarn;
 260
 261         /* Defaults from mandoc.1. */
 262
 263         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 264
 265         if (curp->fflags & IGN_SCOPE)
 266                 pflags |= MDOC_IGN_SCOPE;
 267         if (curp->fflags & NO_IGN_ESCAPE)
 268                 pflags &= ~MDOC_IGN_ESCAPE;
 269         if (curp->fflags & NO_IGN_MACRO)
 270                 pflags &= ~MDOC_IGN_MACRO;
 271         if (curp->fflags & NO_IGN_CHARS)
 272                 pflags &= ~MDOC_IGN_CHARS;
 273
 274         if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb)))
 275                 warnx("memory exhausted");
 276
 277         return(mdoc);
 278 }
 279
 280
 281 static int
 282 ffile(struct buf *blk, struct buf *ln,
 283                 const char *file, struct curparse *curp)
 284 {
 285         int              c;
 286
 287         curp->file = file;
 288         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 289                 warn("%s", curp->file);
 290                 return(0);
 291         }
 292
 293         c = fdesc(blk, ln, curp);
 294
 295         if (-1 == close(curp->fd))
 296                 warn("%s", curp->file);
 297
 298         return(c);
 299 }
 300
 301
 302 static int
 303 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 304 {
 305         size_t           sz;
 306         ssize_t          ssz;
 307         struct stat      st;
 308         int              j, i, pos, lnn, comment;
 309         struct man      *man;
 310         struct mdoc     *mdoc;
 311
 312         sz = BUFSIZ;
 313         man = NULL;
 314         mdoc = NULL;
 315
 316         /*
 317          * Two buffers: ln and buf.  buf is the input buffer optimised
 318          * here for each file's block size.  ln is a line buffer.  Both
 319          * growable, hence passed in by ptr-ptr.
 320          */
 321
 322         if (-1 == fstat(curp->fd, &st))
 323                 warn("%s", curp->file);
 324         else if ((size_t)st.st_blksize > sz)
 325                 sz = st.st_blksize;
 326
 327         if (sz > blk->sz) {
 328                 blk->buf = realloc(blk->buf, sz);
 329                 if (NULL == blk->buf) {
 330                         warn("realloc");
 331                         return(0);
 332                 }
 333                 blk->sz = sz;
 334         }
 335
 336         /* Fill buf with file blocksize. */
 337
 338         for (lnn = pos = comment = 0; ; ) {
 339                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 340                         warn("%s", curp->file);
 341                         return(0);
 342                 } else if (0 == ssz)
 343                         break;
 344
 345                 /* Parse the read block into partial or full lines. */
 346
 347                 for (i = 0; i < (int)ssz; i++) {
 348                         if (pos >= (int)ln->sz) {
 349                                 ln->sz += 256; /* Step-size. */
 350                                 ln->buf = realloc(ln->buf, ln->sz);
 351                                 if (NULL == ln->buf) {
 352                                         warn("realloc");
 353                                         return(0);
 354                                 }
 355                         }
 356
 357                         if ('\n' != blk->buf[i]) {
 358                                 if (comment)
 359                                         continue;
 360                                 ln->buf[pos++] = blk->buf[i];
 361
 362                                 /* Handle in-line `\"' comments. */
 363
 364                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 365                                         continue;
 366
 367                                 for (j = pos - 2; j >= 0; j--)
 368                                         if ('\\' != ln->buf[j])
 369                                                 break;
 370
 371                                 if ( ! ((pos - 2 - j) % 2))
 372                                         continue;
 373
 374                                 comment = 1;
 375                                 pos -= 2;
 376                                 continue;
 377                         }
 378
 379                         /* Handle escaped `\\n' newlines. */
 380
 381                         if (pos > 0 && 0 == comment &&
 382                                         '\\' == ln->buf[pos - 1]) {
 383                                 for (j = pos - 1; j >= 0; j--)
 384                                         if ('\\' != ln->buf[j])
 385                                                 break;
 386                                 if ( ! ((pos - j) % 2)) {
 387                                         pos--;
 388                                         lnn++;
 389                                         continue;
 390                                 }
 391                         }
 392
 393                         ln->buf[pos] = 0;
 394                         lnn++;
 395
 396                         /* If unset, assign parser in pset(). */
 397
 398                         if ( ! (man || mdoc) && ! pset(ln->buf,
 399                                                 pos, curp, &man, &mdoc))
 400                                 return(0);
 401
 402                         pos = comment = 0;
 403
 404                         /* Pass down into parsers. */
 405
 406                         if (man && ! man_parseln(man, lnn, ln->buf))
 407                                 return(0);
 408                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 409                                 return(0);
 410                 }
 411         }
 412
 413         /* NOTE a parser may not have been assigned, yet. */
 414
 415         if ( ! (man || mdoc)) {
 416                 warnx("%s: not a manual", curp->file);
 417                 return(0);
 418         }
 419
 420         if (mdoc && ! mdoc_endparse(mdoc))
 421                 return(0);
 422         if (man && ! man_endparse(man))
 423                 return(0);
 424
 425         /* If unset, allocate output dev now (if applicable). */
 426
 427         if ( ! (curp->outman && curp->outmdoc)) {
 428                 switch (curp->outtype) {
 429                 case (OUTT_TREE):
 430                         curp->outman = tree_man;
 431                         curp->outmdoc = tree_mdoc;
 432                         break;
 433                 case (OUTT_LINT):
 434                         break;
 435                 default:
 436                         curp->outdata = ascii_alloc();
 437                         curp->outman = terminal_man;
 438                         curp->outmdoc = terminal_mdoc;
 439                         curp->outfree = terminal_free;
 440                         break;
 441                 }
 442         }
 443
 444         /* Execute the out device, if it exists. */
 445
 446         if (man && curp->outman)
 447                 if ( ! (*curp->outman)(curp->outdata, man))
 448                         return(0);
 449         if (mdoc && curp->outmdoc)
 450                 if ( ! (*curp->outmdoc)(curp->outdata, mdoc))
 451                         return(0);
 452
 453         return(1);
 454 }
 455
 456
 457 static int
 458 pset(const char *buf, int pos, struct curparse *curp,
 459                 struct man **man, struct mdoc **mdoc)
 460 {
 461         int              i;
 462
 463         /*
 464          * Try to intuit which kind of manual parser should be used.  If
 465          * passed in by command-line (-man, -mdoc), then use that
 466          * explicitly.  If passed as -mandoc, then try to guess from the
 467          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 468          * default to -man, which is more lenient.
 469          */
 470
 471         if (buf[0] == '.') {
 472                 for (i = 1; buf[i]; i++)
 473                         if (' ' != buf[i] && '\t' != buf[i])
 474                                 break;
 475                 if (0 == buf[i])
 476                         return(1);
 477         }
 478
 479         switch (curp->inttype) {
 480         case (INTT_MDOC):
 481                 if (NULL == curp->mdoc)
 482                         curp->mdoc = mdoc_init(curp);
 483                 if (NULL == (*mdoc = curp->mdoc))
 484                         return(0);
 485                 curp->lastmdoc = *mdoc;
 486                 return(1);
 487         case (INTT_MAN):
 488                 if (NULL == curp->man)
 489                         curp->man = man_init(curp);
 490                 if (NULL == (*man = curp->man))
 491                         return(0);
 492                 curp->lastman = *man;
 493                 return(1);
 494         default:
 495                 break;
 496         }
 497
 498         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 499                 if (NULL == curp->mdoc)
 500                         curp->mdoc = mdoc_init(curp);
 501                 if (NULL == (*mdoc = curp->mdoc))
 502                         return(0);
 503                 curp->lastmdoc = *mdoc;
 504                 return(1);
 505         }
 506
 507         if (NULL == curp->man)
 508                 curp->man = man_init(curp);
 509         if (NULL == (*man = curp->man))
 510                 return(0);
 511         curp->lastman = *man;
 512         return(1);
 513 }
 514
 515
 516 static int
 517 moptions(enum intt *tflags, char *arg)
 518 {
 519
 520         if (0 == strcmp(arg, "doc"))
 521                 *tflags = INTT_MDOC;
 522         else if (0 == strcmp(arg, "andoc"))
 523                 *tflags = INTT_AUTO;
 524         else if (0 == strcmp(arg, "an"))
 525                 *tflags = INTT_MAN;
 526         else {
 527                 warnx("bad argument: -m%s", arg);
 528                 return(0);
 529         }
 530
 531         return(1);
 532 }
 533
 534
 535 static int
 536 toptions(enum outt *tflags, char *arg)
 537 {
 538
 539         if (0 == strcmp(arg, "ascii"))
 540                 *tflags = OUTT_ASCII;
 541         else if (0 == strcmp(arg, "lint"))
 542                 *tflags = OUTT_LINT;
 543         else if (0 == strcmp(arg, "tree"))
 544                 *tflags = OUTT_TREE;
 545         else {
 546                 warnx("bad argument: -T%s", arg);
 547                 return(0);
 548         }
 549
 550         return(1);
 551 }
 552
 553
 554 static int
 555 foptions(int *fflags, char *arg)
 556 {
 557         char            *v, *o;
 558         char            *toks[6];
 559
 560         toks[0] = "ign-scope";
 561         toks[1] = "no-ign-escape";
 562         toks[2] = "no-ign-macro";
 563         toks[3] = "no-ign-chars";
 564         toks[4] = "strict";
 565         toks[5] = NULL;
 566
 567         while (*arg) {
 568                 o = arg;
 569                 switch (getsubopt(&arg, toks, &v)) {
 570                 case (0):
 571                         *fflags |= IGN_SCOPE;
 572                         break;
 573                 case (1):
 574                         *fflags |= NO_IGN_ESCAPE;
 575                         break;
 576                 case (2):
 577                         *fflags |= NO_IGN_MACRO;
 578                         break;
 579                 case (3):
 580                         *fflags |= NO_IGN_CHARS;
 581                         break;
 582                 case (4):
 583                         *fflags |= NO_IGN_ESCAPE |
 584                                    NO_IGN_MACRO | NO_IGN_CHARS;
 585                         break;
 586                 default:
 587                         warnx("bad argument: -f%s", o);
 588                         return(0);
 589                 }
 590         }
 591
 592         return(1);
 593 }
 594
 595
 596 static int
 597 woptions(int *wflags, char *arg)
 598 {
 599         char            *v, *o;
 600         char            *toks[5];
 601
 602         toks[0] = "all";
 603         toks[1] = "compat";
 604         toks[2] = "syntax";
 605         toks[3] = "error";
 606         toks[4] = NULL;
 607
 608         while (*arg) {
 609                 o = arg;
 610                 switch (getsubopt(&arg, toks, &v)) {
 611                 case (0):
 612                         *wflags |= WARN_WALL;
 613                         break;
 614                 case (1):
 615                         *wflags |= WARN_WCOMPAT;
 616                         break;
 617                 case (2):
 618                         *wflags |= WARN_WSYNTAX;
 619                         break;
 620                 case (3):
 621                         *wflags |= WARN_WERR;
 622                         break;
 623                 default:
 624                         warnx("bad argument: -W%s", o);
 625                         return(0);
 626                 }
 627         }
 628
 629         return(1);
 630 }
 631
 632
 633 /* ARGSUSED */
 634 static int
 635 merr(void *arg, int line, int col, const char *msg)
 636 {
 637         struct curparse *curp;
 638
 639         curp = (struct curparse *)arg;
 640         warnx("%s:%d: error: %s (column %d)",
 641                         curp->file, line, msg, col);
 642
 643         return(0);
 644 }
 645
 646
 647 static int
 648 mdocwarn(void *arg, int line, int col,
 649                 enum mdoc_warn type, const char *msg)
 650 {
 651         struct curparse *curp;
 652         char            *wtype;
 653
 654         curp = (struct curparse *)arg;
 655         wtype = NULL;
 656
 657         switch (type) {
 658         case (WARN_COMPAT):
 659                 wtype = "compat";
 660                 if (curp->wflags & WARN_WCOMPAT)
 661                         break;
 662                 return(1);
 663         case (WARN_SYNTAX):
 664                 wtype = "syntax";
 665                 if (curp->wflags & WARN_WSYNTAX)
 666                         break;
 667                 return(1);
 668         }
 669
 670         assert(wtype);
 671         warnx("%s:%d: %s warning: %s (column %d)",
 672                         curp->file, line, wtype, msg, col);
 673
 674         if ( ! (curp->wflags & WARN_WERR))
 675                 return(1);
 676
 677         warnx("considering warnings as errors");
 678         return(0);
 679 }
 680
 681
 682 static int
 683 manwarn(void *arg, int line, int col, const char *msg)
 684 {
 685         struct curparse *curp;
 686
 687         curp = (struct curparse *)arg;
 688
 689         if ( ! (curp->wflags & WARN_WSYNTAX))
 690                 return(1);
 691
 692         warnx("%s:%d: syntax warning: %s (column %d)",
 693                         curp->file, line, msg, col);
 694
 695         if ( ! (curp->wflags & WARN_WERR))
 696                 return(1);
 697
 698         warnx("considering warnings as errors");
 699         return(0);
 700 }