main.c

   1 /*      $Id: main.c,v 1.39 2009/07/24 14:00:59 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        (1 << 0)       /* All-warnings mask. */
  69 #define WARN_WERR        (1 << 2)       /* Warnings->errors. */
  70         int               fflags;
  71 #define IGN_SCOPE        (1 << 0)       /* Ignore scope errors. */
  72 #define NO_IGN_ESCAPE    (1 << 1)       /* Don't ignore bad escapes. */
  73 #define NO_IGN_MACRO     (1 << 2)       /* Don't ignore bad macros. */
  74 #define NO_IGN_CHARS     (1 << 3)       /* Don't ignore bad chars. */
  75 #define IGN_ERRORS       (1 << 4)       /* Ignore failed parse. */
  76         enum intt         inttype;      /* Input parsers... */
  77         struct man       *man;
  78         struct man       *lastman;
  79         struct mdoc      *mdoc;
  80         struct mdoc      *lastmdoc;
  81         enum outt         outtype;      /* Output devices... */
  82         out_mdoc          outmdoc;
  83         out_man           outman;
  84         out_free          outfree;
  85         void             *outdata;
  86 };
  87
  88 extern  void             *ascii_alloc(void);
  89 extern  int               tree_mdoc(void *, const struct mdoc *);
  90 extern  int               tree_man(void *, const struct man *);
  91 extern  int               terminal_mdoc(void *, const struct mdoc *);
  92 extern  int               terminal_man(void *, const struct man *);
  93 extern  void              terminal_free(void *);
  94
  95 static  int               foptions(int *, char *);
  96 static  int               toptions(enum outt *, char *);
  97 static  int               moptions(enum intt *, char *);
  98 static  int               woptions(int *, char *);
  99 static  int               merr(void *, int, int, const char *);
 100 static  int               mwarn(void *, int, int, const char *);
 101 static  int               ffile(struct buf *, struct buf *,
 102                                 const char *, struct curparse *);
 103 static  int               fdesc(struct buf *, struct buf *,
 104                                 struct curparse *);
 105 static  int               pset(const char *, int, struct curparse *,
 106                                 struct man **, struct mdoc **);
 107 static  struct man       *man_init(struct curparse *);
 108 static  struct mdoc      *mdoc_init(struct curparse *);
 109 __dead  static void       version(void);
 110 __dead  static void       usage(void);
 111
 112 extern  char             *__progname;
 113
 114
 115 int
 116 main(int argc, char *argv[])
 117 {
 118         int              c, rc;
 119         struct buf       ln, blk;
 120         struct curparse  curp;
 121
 122         bzero(&curp, sizeof(struct curparse));
 123
 124         curp.inttype = INTT_AUTO;
 125         curp.outtype = OUTT_ASCII;
 126
 127         /* LINTED */
 128         while (-1 != (c = getopt(argc, argv, "f:m:VW:T:")))
 129                 switch (c) {
 130                 case ('f'):
 131                         if ( ! foptions(&curp.fflags, optarg))
 132                                 return(EXIT_FAILURE);
 133                         break;
 134                 case ('m'):
 135                         if ( ! moptions(&curp.inttype, optarg))
 136                                 return(EXIT_FAILURE);
 137                         break;
 138                 case ('T'):
 139                         if ( ! toptions(&curp.outtype, optarg))
 140                                 return(EXIT_FAILURE);
 141                         break;
 142                 case ('W'):
 143                         if ( ! woptions(&curp.wflags, optarg))
 144                                 return(EXIT_FAILURE);
 145                         break;
 146                 case ('V'):
 147                         version();
 148                         /* NOTREACHED */
 149                 default:
 150                         usage();
 151                         /* NOTREACHED */
 152                 }
 153
 154         argc -= optind;
 155         argv += optind;
 156
 157         bzero(&ln, sizeof(struct buf));
 158         bzero(&blk, sizeof(struct buf));
 159
 160         rc = 1;
 161
 162         if (NULL == *argv) {
 163                 curp.file = "<stdin>";
 164                 curp.fd = STDIN_FILENO;
 165
 166                 c = fdesc(&blk, &ln, &curp);
 167                 if ( ! (IGN_ERRORS & curp.fflags))
 168                         rc = 1 == c ? 1 : 0;
 169                 else
 170                         rc = -1 == c ? 0 : 1;
 171         }
 172
 173         while (rc && *argv) {
 174                 c = ffile(&blk, &ln, *argv, &curp);
 175                 if ( ! (IGN_ERRORS & curp.fflags))
 176                         rc = 1 == c ? 1 : 0;
 177                 else
 178                         rc = -1 == c ? 0 : 1;
 179
 180                 argv++;
 181                 if (*argv && rc) {
 182                         if (curp.lastman)
 183                                 if ( ! man_reset(curp.lastman))
 184                                         rc = 0;
 185                         if (curp.lastmdoc)
 186                                 if ( ! mdoc_reset(curp.lastmdoc))
 187                                         rc = 0;
 188                         curp.lastman = NULL;
 189                         curp.lastmdoc = NULL;
 190                 }
 191         }
 192
 193         if (blk.buf)
 194                 free(blk.buf);
 195         if (ln.buf)
 196                 free(ln.buf);
 197         if (curp.outfree)
 198                 (*curp.outfree)(curp.outdata);
 199         if (curp.mdoc)
 200                 mdoc_free(curp.mdoc);
 201         if (curp.man)
 202                 man_free(curp.man);
 203
 204         return(rc ? EXIT_SUCCESS : EXIT_FAILURE);
 205 }
 206
 207
 208 __dead static void
 209 version(void)
 210 {
 211
 212         (void)printf("%s %s\n", __progname, VERSION);
 213         exit(EXIT_SUCCESS);
 214 }
 215
 216
 217 __dead static void
 218 usage(void)
 219 {
 220
 221         (void)fprintf(stderr, "usage: %s [-V] [-foption...] "
 222                         "[-mformat] [-Toutput] [-Werr...]\n",
 223                         __progname);
 224         exit(EXIT_FAILURE);
 225 }
 226
 227
 228 static struct man *
 229 man_init(struct curparse *curp)
 230 {
 231         int              pflags;
 232         struct man      *man;
 233         struct man_cb    mancb;
 234
 235         mancb.man_err = merr;
 236         mancb.man_warn = mwarn;
 237
 238         /* Defaults from mandoc.1. */
 239
 240         pflags = MAN_IGN_MACRO | MAN_IGN_ESCAPE | MAN_IGN_CHARS;
 241
 242         if (curp->fflags & NO_IGN_MACRO)
 243                 pflags &= ~MAN_IGN_MACRO;
 244         if (curp->fflags & NO_IGN_CHARS)
 245                 pflags &= ~MAN_IGN_CHARS;
 246         if (curp->fflags & NO_IGN_ESCAPE)
 247                 pflags &= ~MAN_IGN_ESCAPE;
 248
 249         if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 250                 warnx("memory exhausted");
 251
 252         return(man);
 253 }
 254
 255
 256 static struct mdoc *
 257 mdoc_init(struct curparse *curp)
 258 {
 259         int              pflags;
 260         struct mdoc     *mdoc;
 261         struct mdoc_cb   mdoccb;
 262
 263         mdoccb.mdoc_err = merr;
 264         mdoccb.mdoc_warn = mwarn;
 265
 266         /* Defaults from mandoc.1. */
 267
 268         pflags = MDOC_IGN_MACRO | MDOC_IGN_ESCAPE | MDOC_IGN_CHARS;
 269
 270         if (curp->fflags & IGN_SCOPE)
 271                 pflags |= MDOC_IGN_SCOPE;
 272         if (curp->fflags & NO_IGN_ESCAPE)
 273                 pflags &= ~MDOC_IGN_ESCAPE;
 274         if (curp->fflags & NO_IGN_MACRO)
 275                 pflags &= ~MDOC_IGN_MACRO;
 276         if (curp->fflags & NO_IGN_CHARS)
 277                 pflags &= ~MDOC_IGN_CHARS;
 278
 279         if (NULL == (mdoc = mdoc_alloc(curp, pflags, &mdoccb)))
 280                 warnx("memory exhausted");
 281
 282         return(mdoc);
 283 }
 284
 285
 286 static int
 287 ffile(struct buf *blk, struct buf *ln,
 288                 const char *file, struct curparse *curp)
 289 {
 290         int              c;
 291
 292         curp->file = file;
 293         if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
 294                 warn("%s", curp->file);
 295                 return(-1);
 296         }
 297
 298         c = fdesc(blk, ln, curp);
 299
 300         if (-1 == close(curp->fd))
 301                 warn("%s", curp->file);
 302
 303         return(c);
 304 }
 305
 306
 307 static int
 308 fdesc(struct buf *blk, struct buf *ln, struct curparse *curp)
 309 {
 310         size_t           sz;
 311         ssize_t          ssz;
 312         struct stat      st;
 313         int              j, i, pos, lnn, comment;
 314         struct man      *man;
 315         struct mdoc     *mdoc;
 316
 317         sz = BUFSIZ;
 318         man = NULL;
 319         mdoc = NULL;
 320
 321         /*
 322          * Two buffers: ln and buf.  buf is the input buffer optimised
 323          * here for each file's block size.  ln is a line buffer.  Both
 324          * growable, hence passed in by ptr-ptr.
 325          */
 326
 327         if (-1 == fstat(curp->fd, &st))
 328                 warn("%s", curp->file);
 329         else if ((size_t)st.st_blksize > sz)
 330                 sz = st.st_blksize;
 331
 332         if (sz > blk->sz) {
 333                 blk->buf = realloc(blk->buf, sz);
 334                 if (NULL == blk->buf) {
 335                         warn("realloc");
 336                         return(-1);
 337                 }
 338                 blk->sz = sz;
 339         }
 340
 341         /* Fill buf with file blocksize. */
 342
 343         for (lnn = pos = comment = 0; ; ) {
 344                 if (-1 == (ssz = read(curp->fd, blk->buf, sz))) {
 345                         warn("%s", curp->file);
 346                         return(-1);
 347                 } else if (0 == ssz)
 348                         break;
 349
 350                 /* Parse the read block into partial or full lines. */
 351
 352                 for (i = 0; i < (int)ssz; i++) {
 353                         if (pos >= (int)ln->sz) {
 354                                 ln->sz += 256; /* Step-size. */
 355                                 ln->buf = realloc(ln->buf, ln->sz);
 356                                 if (NULL == ln->buf) {
 357                                         warn("realloc");
 358                                         return(-1);
 359                                 }
 360                         }
 361
 362                         if ('\n' != blk->buf[i]) {
 363                                 if (comment)
 364                                         continue;
 365                                 ln->buf[pos++] = blk->buf[i];
 366
 367                                 /* Handle in-line `\"' comments. */
 368
 369                                 if (1 == pos || '\"' != ln->buf[pos - 1])
 370                                         continue;
 371
 372                                 for (j = pos - 2; j >= 0; j--)
 373                                         if ('\\' != ln->buf[j])
 374                                                 break;
 375
 376                                 if ( ! ((pos - 2 - j) % 2))
 377                                         continue;
 378
 379                                 comment = 1;
 380                                 pos -= 2;
 381                                 continue;
 382                         }
 383
 384                         /* Handle escaped `\\n' newlines. */
 385
 386                         if (pos > 0 && 0 == comment &&
 387                                         '\\' == ln->buf[pos - 1]) {
 388                                 for (j = pos - 1; j >= 0; j--)
 389                                         if ('\\' != ln->buf[j])
 390                                                 break;
 391                                 if ( ! ((pos - j) % 2)) {
 392                                         pos--;
 393                                         lnn++;
 394                                         continue;
 395                                 }
 396                         }
 397
 398                         ln->buf[pos] = 0;
 399                         lnn++;
 400
 401                         /* If unset, assign parser in pset(). */
 402
 403                         if ( ! (man || mdoc) && ! pset(ln->buf,
 404                                                 pos, curp, &man, &mdoc))
 405                                 return(-1);
 406
 407                         pos = comment = 0;
 408
 409                         /* Pass down into parsers. */
 410
 411                         if (man && ! man_parseln(man, lnn, ln->buf))
 412                                 return(0);
 413                         if (mdoc && ! mdoc_parseln(mdoc, lnn, ln->buf))
 414                                 return(0);
 415                 }
 416         }
 417
 418         /* NOTE a parser may not have been assigned, yet. */
 419
 420         if ( ! (man || mdoc)) {
 421                 warnx("%s: not a manual", curp->file);
 422                 return(0);
 423         }
 424
 425         if (mdoc && ! mdoc_endparse(mdoc))
 426                 return(0);
 427         if (man && ! man_endparse(man))
 428                 return(0);
 429
 430         /* If unset, allocate output dev now (if applicable). */
 431
 432         if ( ! (curp->outman && curp->outmdoc)) {
 433                 switch (curp->outtype) {
 434                 case (OUTT_TREE):
 435                         curp->outman = tree_man;
 436                         curp->outmdoc = tree_mdoc;
 437                         break;
 438                 case (OUTT_LINT):
 439                         break;
 440                 default:
 441                         curp->outdata = ascii_alloc();
 442                         curp->outman = terminal_man;
 443                         curp->outmdoc = terminal_mdoc;
 444                         curp->outfree = terminal_free;
 445                         break;
 446                 }
 447         }
 448
 449         /* Execute the out device, if it exists. */
 450
 451         if (man && curp->outman)
 452                 if ( ! (*curp->outman)(curp->outdata, man))
 453                         return(-1);
 454         if (mdoc && curp->outmdoc)
 455                 if ( ! (*curp->outmdoc)(curp->outdata, mdoc))
 456                         return(-1);
 457
 458         return(1);
 459 }
 460
 461
 462 static int
 463 pset(const char *buf, int pos, struct curparse *curp,
 464                 struct man **man, struct mdoc **mdoc)
 465 {
 466         int              i;
 467
 468         /*
 469          * Try to intuit which kind of manual parser should be used.  If
 470          * passed in by command-line (-man, -mdoc), then use that
 471          * explicitly.  If passed as -mandoc, then try to guess from the
 472          * line: either skip dot-lines, use -mdoc when finding `.Dt', or
 473          * default to -man, which is more lenient.
 474          */
 475
 476         if (buf[0] == '.') {
 477                 for (i = 1; buf[i]; i++)
 478                         if (' ' != buf[i] && '\t' != buf[i])
 479                                 break;
 480                 if (0 == buf[i])
 481                         return(1);
 482         }
 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 static int
 560 foptions(int *fflags, char *arg)
 561 {
 562         char            *v, *o;
 563         char            *toks[7];
 564
 565         toks[0] = "ign-scope";
 566         toks[1] = "no-ign-escape";
 567         toks[2] = "no-ign-macro";
 568         toks[3] = "no-ign-chars";
 569         toks[4] = "ign-errors";
 570         toks[5] = "strict";
 571         toks[6] = NULL;
 572
 573         while (*arg) {
 574                 o = arg;
 575                 switch (getsubopt(&arg, toks, &v)) {
 576                 case (0):
 577                         *fflags |= IGN_SCOPE;
 578                         break;
 579                 case (1):
 580                         *fflags |= NO_IGN_ESCAPE;
 581                         break;
 582                 case (2):
 583                         *fflags |= NO_IGN_MACRO;
 584                         break;
 585                 case (3):
 586                         *fflags |= NO_IGN_CHARS;
 587                         break;
 588                 case (4):
 589                         *fflags |= IGN_ERRORS;
 590                         break;
 591                 case (5):
 592                         *fflags |= NO_IGN_ESCAPE |
 593                                    NO_IGN_MACRO | NO_IGN_CHARS;
 594                         break;
 595                 default:
 596                         warnx("bad argument: -f%s", o);
 597                         return(0);
 598                 }
 599         }
 600
 601         return(1);
 602 }
 603
 604
 605 static int
 606 woptions(int *wflags, char *arg)
 607 {
 608         char            *v, *o;
 609         char            *toks[3];
 610
 611         toks[0] = "all";
 612         toks[1] = "error";
 613         toks[2] = NULL;
 614
 615         while (*arg) {
 616                 o = arg;
 617                 switch (getsubopt(&arg, toks, &v)) {
 618                 case (0):
 619                         *wflags |= WARN_WALL;
 620                         break;
 621                 case (1):
 622                         *wflags |= WARN_WERR;
 623                         break;
 624                 default:
 625                         warnx("bad argument: -W%s", o);
 626                         return(0);
 627                 }
 628         }
 629
 630         return(1);
 631 }
 632
 633
 634 /* ARGSUSED */
 635 static int
 636 merr(void *arg, int line, int col, const char *msg)
 637 {
 638         struct curparse *curp;
 639
 640         curp = (struct curparse *)arg;
 641
 642         warnx("%s:%d: error: %s (column %d)",
 643                         curp->file, line, msg, col);
 644
 645         return(0);
 646 }
 647
 648
 649 static int
 650 mwarn(void *arg, int line, int col, const char *msg)
 651 {
 652         struct curparse *curp;
 653
 654         curp = (struct curparse *)arg;
 655
 656         if ( ! (curp->wflags & WARN_WALL))
 657                 return(1);
 658
 659         warnx("%s:%d: warning: %s (column %d)",
 660                         curp->file, line, msg, col);
 661
 662         if ( ! (curp->wflags & WARN_WERR))
 663                 return(1);
 664
 665         warnx("considering warnings as errors");
 666         return(0);
 667 }
 668