main.c

   1 /*      $Id: main.c,v 1.31 2009/06/18 10:32:00 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 | 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
 242         if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 243                 warnx("memory exhausted");
 244
 245         return(man);
 246 }
 247
 248
 249 static struct mdoc *
 250 mdoc_init(struct curparse *curp)
 251 {
 252         int              pflags;
 253         struct mdoc     *mdoc;
 254         struct mdoc_cb   mdoccb;
 255
 256         mdoccb.mdoc_err = merr;
 257         mdoccb.mdoc_warn = mdocwarn;
 258
 259         /* Defaults from mandoc.1. */
 260
 261         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 262
 263         if (curp->fflags & IGN_SCOPE)
 264                 pflags |= MDOC_IGN_SCOPE;
 265         if (curp->fflags & NO_IGN_ESCAPE)
 266                 pflags &= ~MDOC_IGN_ESCAPE;
 267         if (curp->fflags & NO_IGN_MACRO)
 268                 pflags &= ~MDOC_IGN_MACRO;
 269         if (curp->fflags & NO_IGN_CHARS)
 270                 pflags &= ~MDOC_IGN_CHARS;
 271
 272         if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb)))
 273                 warnx("memory exhausted");
 274
 275         return(mdoc);
 276 }
 277
 278
 279 static int
 280 ffile(struct buf *blk, struct buf *ln,
 281                 const char *file, struct curparse *curp)
 282 {
 283         int              c;
 284
 285         curp->file = file;
 286         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 287                 warn("%s", curp->file);
 288                 return(0);
 289         }
 290
 291         c = fdesc(blk, ln, curp);
 292
 293         if (-1 == close(curp->fd))
 294                 warn("%s", curp->file);
 295
 296         return(c);
 297 }
 298
 299
 300 static int
 301 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 302 {
 303         size_t           sz;
 304         ssize_t          ssz;
 305         struct stat      st;
 306         int              j, i, pos, lnn, comment;
 307         struct man      *man;
 308         struct mdoc     *mdoc;
 309
 310         sz = BUFSIZ;
 311         man = NULL;
 312         mdoc = NULL;
 313
 314         /*
 315          * Two buffers: ln and buf.  buf is the input buffer optimised
 316          * here for each file's block size.  ln is a line buffer.  Both
 317          * growable, hence passed in by ptr-ptr.
 318          */
 319
 320         if (-1 == fstat(curp->fd, &st))
 321                 warnx("%s", curp->file);
 322         else if ((size_t)st.st_blksize > sz)
 323                 sz = st.st_blksize;
 324
 325         if (sz > blk->sz) {
 326                 blk->buf = realloc(blk->buf, sz);
 327                 if (NULL == blk->buf) {
 328                         warn("realloc");
 329                         return(0);
 330                 }
 331                 blk->sz = sz;
 332         }
 333
 334         /* Fill buf with file blocksize. */
 335
 336         for (lnn = pos = comment = 0; ; ) {
 337                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 338                         warn("%s", curp->file);
 339                         return(0);
 340                 } else if (0 == ssz)
 341                         break;
 342
 343                 /* Parse the read block into partial or full lines. */
 344
 345                 for (i = 0; i < (int)ssz; i++) {
 346                         if (pos >= (int)ln->sz) {
 347                                 ln->sz += 256; /* Step-size. */
 348                                 ln->buf = realloc(ln->buf, ln->sz);
 349                                 if (NULL == ln->buf) {
 350                                         warn("realloc");
 351                                         return(0);
 352                                 }
 353                         }
 354
 355                         if ('\n' != blk->buf[i]) {
 356                                 if (comment)
 357                                         continue;
 358                                 ln->buf[pos++] = blk->buf[i];
 359
 360                                 /* Handle in-line `\"' comments. */
 361
 362                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 363                                         continue;
 364
 365                                 for (j = pos - 2; j >= 0; j--)
 366                                         if ('\\' != ln->buf[j])
 367                                                 break;
 368
 369                                 if ( ! ((pos - 2 - j) % 2))
 370                                         continue;
 371
 372                                 comment = 1;
 373                                 pos -= 2;
 374                                 continue;
 375                         }
 376
 377                         /* Handle escaped `\\n' newlines. */
 378
 379                         if (pos > 0 && 0 == comment &&
 380                                         '\\' == ln->buf[pos - 1]) {
 381                                 for (j = pos - 1; j >= 0; j--)
 382                                         if ('\\' != ln->buf[j])
 383                                                 break;
 384                                 if ( ! ((pos - j) % 2)) {
 385                                         pos--;
 386                                         lnn++;
 387                                         continue;
 388                                 }
 389                         }
 390
 391                         ln->buf[pos] = 0;
 392                         lnn++;
 393
 394                         /* If unset, assign parser in pset(). */
 395
 396                         if ( ! (man || mdoc) && ! pset(ln->buf,
 397                                                 pos, curp, &man, &mdoc))
 398                                 return(0);
 399
 400                         pos = comment = 0;
 401
 402                         /* Pass down into parsers. */
 403
 404                         if (man && ! man_parseln(man, lnn, ln->buf))
 405                                 return(0);
 406                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 407                                 return(0);
 408                 }
 409         }
 410
 411         /* NOTE a parser may not have been assigned, yet. */
 412
 413         if ( ! (man || mdoc)) {
 414                 warnx("%s: not a manual", curp->file);
 415                 return(0);
 416         }
 417
 418         if (mdoc && ! mdoc_endparse(mdoc))
 419                 return(0);
 420         if (man && ! man_endparse(man))
 421                 return(0);
 422
 423         /* If unset, allocate output dev now (if applicable). */
 424
 425         if ( ! (curp->outman && curp->outmdoc)) {
 426                 switch (curp->outtype) {
 427                 case (OUTT_TREE):
 428                         curp->outman = tree_man;
 429                         curp->outmdoc = tree_mdoc;
 430                         break;
 431                 case (OUTT_LINT):
 432                         break;
 433                 default:
 434                         curp->outdata = ascii_alloc();
 435                         curp->outman = terminal_man;
 436                         curp->outmdoc = terminal_mdoc;
 437                         curp->outfree = terminal_free;
 438                         break;
 439                 }
 440         }
 441
 442         /* Execute the out device, if it exists. */
 443
 444         if (man && curp->outman)
 445                 if ( ! (*curp->outman)(curp->outdata, man))
 446                         return(0);
 447         if (mdoc && curp->outmdoc)
 448                 if ( ! (*curp->outmdoc)(curp->outdata, mdoc))
 449                         return(0);
 450
 451         return(1);
 452 }
 453
 454
 455 static int
 456 pset(const char *buf, int pos, struct curparse *curp,
 457                 struct man **man, struct mdoc **mdoc)
 458 {
 459         int              i;
 460
 461         /*
 462          * Try to intuit which kind of manual parser should be used.  If
 463          * passed in by command-line (-man, -mdoc), then use that
 464          * explicitly.  If passed as -mandoc, then try to guess from the
 465          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 466          * default to -man, which is more lenient.
 467          */
 468
 469         if (buf[0] == '.') {
 470                 for (i = 1; buf[i]; i++)
 471                         if (' ' != buf[i] && '\t' != buf[i])
 472                                 break;
 473                 if (0 == buf[i])
 474                         return(1);
 475         }
 476
 477         switch (curp->inttype) {
 478         case (INTT_MDOC):
 479                 if (NULL == curp->mdoc)
 480                         curp->mdoc = mdoc_init(curp);
 481                 if (NULL == (*mdoc = curp->mdoc))
 482                         return(0);
 483                 curp->lastmdoc = *mdoc;
 484                 return(1);
 485         case (INTT_MAN):
 486                 if (NULL == curp->man)
 487                         curp->man = man_init(curp);
 488                 if (NULL == (*man = curp->man))
 489                         return(0);
 490                 curp->lastman = *man;
 491                 return(1);
 492         default:
 493                 break;
 494         }
 495
 496         if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3))  {
 497                 if (NULL == curp->mdoc)
 498                         curp->mdoc = mdoc_init(curp);
 499                 if (NULL == (*mdoc = curp->mdoc))
 500                         return(0);
 501                 curp->lastmdoc = *mdoc;
 502                 return(1);
 503         }
 504
 505         if (NULL == curp->man)
 506                 curp->man = man_init(curp);
 507         if (NULL == (*man = curp->man))
 508                 return(0);
 509         curp->lastman = *man;
 510         return(1);
 511 }
 512
 513
 514 static int
 515 moptions(enum intt *tflags, char *arg)
 516 {
 517
 518         if (0 == strcmp(arg, "doc"))
 519                 *tflags = INTT_MDOC;
 520         else if (0 == strcmp(arg, "andoc"))
 521                 *tflags = INTT_AUTO;
 522         else if (0 == strcmp(arg, "an"))
 523                 *tflags = INTT_MAN;
 524         else {
 525                 warnx("bad argument: -m%s", arg);
 526                 return(0);
 527         }
 528
 529         return(1);
 530 }
 531
 532
 533 static int
 534 toptions(enum outt *tflags, char *arg)
 535 {
 536
 537         if (0 == strcmp(arg, "ascii"))
 538                 *tflags = OUTT_ASCII;
 539         else if (0 == strcmp(arg, "lint"))
 540                 *tflags = OUTT_LINT;
 541         else if (0 == strcmp(arg, "tree"))
 542                 *tflags = OUTT_TREE;
 543         else {
 544                 warnx("bad argument: -T%s", arg);
 545                 return(0);
 546         }
 547
 548         return(1);
 549 }
 550
 551
 552 static int
 553 foptions(int *fflags, char *arg)
 554 {
 555         char            *v;
 556         char            *toks[6];
 557
 558         toks[0] = "ign-scope";
 559         toks[1] = "no-ign-escape";
 560         toks[2] = "no-ign-macro";
 561         toks[3] = "no-ign-chars";
 562         toks[4] = "strict";
 563         toks[5] = NULL;
 564
 565         while (*arg)
 566                 switch (getsubopt(&arg, toks, &v)) {
 567                 case (0):
 568                         *fflags |= IGN_SCOPE;
 569                         break;
 570                 case (1):
 571                         *fflags |= NO_IGN_ESCAPE;
 572                         break;
 573                 case (2):
 574                         *fflags |= NO_IGN_MACRO;
 575                         break;
 576                 case (3):
 577                         *fflags |= NO_IGN_CHARS;
 578                         break;
 579                 case (4):
 580                         *fflags |= NO_IGN_ESCAPE |
 581                                    NO_IGN_MACRO | NO_IGN_CHARS;
 582                         break;
 583                 default:
 584                         warnx("bad argument: -f%s", arg);
 585                         return(0);
 586                 }
 587
 588         return(1);
 589 }
 590
 591
 592 static int
 593 woptions(int *wflags, char *arg)
 594 {
 595         char            *v;
 596         char            *toks[5];
 597
 598         toks[0] = "all";
 599         toks[1] = "compat";
 600         toks[2] = "syntax";
 601         toks[3] = "error";
 602         toks[4] = NULL;
 603
 604         while (*arg)
 605                 switch (getsubopt(&arg, toks, &v)) {
 606                 case (0):
 607                         *wflags |= WARN_WALL;
 608                         break;
 609                 case (1):
 610                         *wflags |= WARN_WCOMPAT;
 611                         break;
 612                 case (2):
 613                         *wflags |= WARN_WSYNTAX;
 614                         break;
 615                 case (3):
 616                         *wflags |= WARN_WERR;
 617                         break;
 618                 default:
 619                         warnx("bad argument: -W%s", arg);
 620                         return(0);
 621                 }
 622
 623         return(1);
 624 }
 625
 626
 627 /* ARGSUSED */
 628 static int
 629 merr(void *arg, int line, int col, const char *msg)
 630 {
 631         struct curparse *curp;
 632
 633         curp = (struct curparse *)arg;
 634         warnx("%s:%d: error: %s (column %d)",
 635                         curp->file, line, msg, col);
 636
 637         return(0);
 638 }
 639
 640
 641 static int
 642 mdocwarn(void *arg, int line, int col,
 643                 enum mdoc_warn type, const char *msg)
 644 {
 645         struct curparse *curp;
 646         char            *wtype;
 647
 648         curp = (struct curparse *)arg;
 649         wtype = NULL;
 650
 651         switch (type) {
 652         case (WARN_COMPAT):
 653                 wtype = "compat";
 654                 if (curp->wflags & WARN_WCOMPAT)
 655                         break;
 656                 return(1);
 657         case (WARN_SYNTAX):
 658                 wtype = "syntax";
 659                 if (curp->wflags & WARN_WSYNTAX)
 660                         break;
 661                 return(1);
 662         }
 663
 664         assert(wtype);
 665         warnx("%s:%d: %s warning: %s (column %d)",
 666                         curp->file, line, wtype, msg, col);
 667
 668         if ( ! (curp->wflags & WARN_WERR))
 669                 return(1);
 670
 671         warnx("considering warnings as errors");
 672         return(0);
 673 }
 674
 675
 676 static int
 677 manwarn(void *arg, int line, int col, const char *msg)
 678 {
 679         struct curparse *curp;
 680
 681         curp = (struct curparse *)arg;
 682
 683         if ( ! (curp->wflags & WARN_WSYNTAX))
 684                 return(1);
 685
 686         warnx("%s:%d: syntax warning: %s (column %d)",
 687                         curp->file, line, msg, col);
 688
 689         if ( ! (curp->wflags & WARN_WERR))
 690                 return(1);
 691
 692         warnx("considering warnings as errors");
 693         return(0);
 694 }