mdoc_action.c

   1 /*      $Id: mdoc_action.c,v 1.86 2010/11/30 10:32:05 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
   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 #ifndef OSNAME
  22 #include <sys/utsname.h>
  23 #endif
  24
  25 #include <assert.h>
  26 #include <stdio.h>
  27 #include <stdlib.h>
  28 #include <string.h>
  29 #include <time.h>
  30
  31 #include "mandoc.h"
  32 #include "libmdoc.h"
  33 #include "libmandoc.h"
  34
  35 /*
  36  * FIXME: this file is deprecated.  All future "actions" should be
  37  * pushed into mdoc_validate.c.
  38  */
  39
  40 #define POST_ARGS struct mdoc *m, struct mdoc_node *n
  41 #define PRE_ARGS  struct mdoc *m, struct mdoc_node *n
  42
  43 #define NUMSIZ    32
  44 #define DATESIZ   32
  45
  46 struct  actions {
  47         int     (*pre)(PRE_ARGS);
  48         int     (*post)(POST_ARGS);
  49 };
  50
  51 static  int       concat(struct mdoc *, char *,
  52                         const struct mdoc_node *, size_t);
  53
  54 static  int       post_bl(POST_ARGS);
  55 static  int       post_bl_tagwidth(POST_ARGS);
  56 static  int       post_bl_width(POST_ARGS);
  57 static  int       post_dd(POST_ARGS);
  58 static  int       post_dt(POST_ARGS);
  59 static  int       post_os(POST_ARGS);
  60 static  int       post_prol(POST_ARGS);
  61 static  int       post_std(POST_ARGS);
  62
  63 static  const struct actions mdoc_actions[MDOC_MAX] = {
  64         { NULL, NULL }, /* Ap */
  65         { NULL, post_dd }, /* Dd */
  66         { NULL, post_dt }, /* Dt */
  67         { NULL, post_os }, /* Os */
  68         { NULL, NULL }, /* Sh */
  69         { NULL, NULL }, /* Ss */
  70         { NULL, NULL }, /* Pp */
  71         { NULL, NULL }, /* D1 */
  72         { NULL, NULL }, /* Dl */
  73         { NULL, NULL }, /* Bd */
  74         { NULL, NULL }, /* Ed */
  75         { NULL, post_bl }, /* Bl */
  76         { NULL, NULL }, /* El */
  77         { NULL, NULL }, /* It */
  78         { NULL, NULL }, /* Ad */
  79         { NULL, NULL }, /* An */
  80         { NULL, NULL }, /* Ar */
  81         { NULL, NULL }, /* Cd */
  82         { NULL, NULL }, /* Cm */
  83         { NULL, NULL }, /* Dv */
  84         { NULL, NULL }, /* Er */
  85         { NULL, NULL }, /* Ev */
  86         { NULL, post_std }, /* Ex */
  87         { NULL, NULL }, /* Fa */
  88         { NULL, NULL }, /* Fd */
  89         { NULL, NULL }, /* Fl */
  90         { NULL, NULL }, /* Fn */
  91         { NULL, NULL }, /* Ft */
  92         { NULL, NULL }, /* Ic */
  93         { NULL, NULL }, /* In */
  94         { NULL, NULL }, /* Li */
  95         { NULL, NULL }, /* Nd */
  96         { NULL, NULL }, /* Nm */
  97         { NULL, NULL }, /* Op */
  98         { NULL, NULL }, /* Ot */
  99         { NULL, NULL }, /* Pa */
 100         { NULL, post_std }, /* Rv */
 101         { NULL, NULL }, /* St */
 102         { NULL, NULL }, /* Va */
 103         { NULL, NULL }, /* Vt */
 104         { NULL, NULL }, /* Xr */
 105         { NULL, NULL }, /* %A */
 106         { NULL, NULL }, /* %B */
 107         { NULL, NULL }, /* %D */
 108         { NULL, NULL }, /* %I */
 109         { NULL, NULL }, /* %J */
 110         { NULL, NULL }, /* %N */
 111         { NULL, NULL }, /* %O */
 112         { NULL, NULL }, /* %P */
 113         { NULL, NULL }, /* %R */
 114         { NULL, NULL }, /* %T */
 115         { NULL, NULL }, /* %V */
 116         { NULL, NULL }, /* Ac */
 117         { NULL, NULL }, /* Ao */
 118         { NULL, NULL }, /* Aq */
 119         { NULL, NULL }, /* At */
 120         { NULL, NULL }, /* Bc */
 121         { NULL, NULL }, /* Bf */
 122         { NULL, NULL }, /* Bo */
 123         { NULL, NULL }, /* Bq */
 124         { NULL, NULL }, /* Bsx */
 125         { NULL, NULL }, /* Bx */
 126         { NULL, NULL }, /* Db */
 127         { NULL, NULL }, /* Dc */
 128         { NULL, NULL }, /* Do */
 129         { NULL, NULL }, /* Dq */
 130         { NULL, NULL }, /* Ec */
 131         { NULL, NULL }, /* Ef */
 132         { NULL, NULL }, /* Em */
 133         { NULL, NULL }, /* Eo */
 134         { NULL, NULL }, /* Fx */
 135         { NULL, NULL }, /* Ms */
 136         { NULL, NULL }, /* No */
 137         { NULL, NULL }, /* Ns */
 138         { NULL, NULL }, /* Nx */
 139         { NULL, NULL }, /* Ox */
 140         { NULL, NULL }, /* Pc */
 141         { NULL, NULL }, /* Pf */
 142         { NULL, NULL }, /* Po */
 143         { NULL, NULL }, /* Pq */
 144         { NULL, NULL }, /* Qc */
 145         { NULL, NULL }, /* Ql */
 146         { NULL, NULL }, /* Qo */
 147         { NULL, NULL }, /* Qq */
 148         { NULL, NULL }, /* Re */
 149         { NULL, NULL }, /* Rs */
 150         { NULL, NULL }, /* Sc */
 151         { NULL, NULL }, /* So */
 152         { NULL, NULL }, /* Sq */
 153         { NULL, NULL }, /* Sm */
 154         { NULL, NULL }, /* Sx */
 155         { NULL, NULL }, /* Sy */
 156         { NULL, NULL }, /* Tn */
 157         { NULL, NULL }, /* Ux */
 158         { NULL, NULL }, /* Xc */
 159         { NULL, NULL }, /* Xo */
 160         { NULL, NULL }, /* Fo */
 161         { NULL, NULL }, /* Fc */
 162         { NULL, NULL }, /* Oo */
 163         { NULL, NULL }, /* Oc */
 164         { NULL, NULL }, /* Bk */
 165         { NULL, NULL }, /* Ek */
 166         { NULL, NULL }, /* Bt */
 167         { NULL, NULL }, /* Hf */
 168         { NULL, NULL }, /* Fr */
 169         { NULL, NULL }, /* Ud */
 170         { NULL, NULL }, /* Lb */
 171         { NULL, NULL }, /* Lp */
 172         { NULL, NULL }, /* Lk */
 173         { NULL, NULL }, /* Mt */
 174         { NULL, NULL }, /* Brq */
 175         { NULL, NULL }, /* Bro */
 176         { NULL, NULL }, /* Brc */
 177         { NULL, NULL }, /* %C */
 178         { NULL, NULL }, /* Es */
 179         { NULL, NULL }, /* En */
 180         { NULL, NULL }, /* Dx */
 181         { NULL, NULL }, /* %Q */
 182         { NULL, NULL }, /* br */
 183         { NULL, NULL }, /* sp */
 184         { NULL, NULL }, /* %U */
 185         { NULL, NULL }, /* Ta */
 186 };
 187
 188
 189 int
 190 mdoc_action_pre(struct mdoc *m, struct mdoc_node *n)
 191 {
 192
 193         switch (n->type) {
 194         case (MDOC_ROOT):
 195                 /* FALLTHROUGH */
 196         case (MDOC_TEXT):
 197                 return(1);
 198         default:
 199                 break;
 200         }
 201
 202         if (NULL == mdoc_actions[n->tok].pre)
 203                 return(1);
 204         return((*mdoc_actions[n->tok].pre)(m, n));
 205 }
 206
 207
 208 int
 209 mdoc_action_post(struct mdoc *m)
 210 {
 211
 212         if (MDOC_ACTED & m->last->flags)
 213                 return(1);
 214         m->last->flags |= MDOC_ACTED;
 215
 216         switch (m->last->type) {
 217         case (MDOC_TEXT):
 218                 /* FALLTHROUGH */
 219         case (MDOC_ROOT):
 220                 return(1);
 221         default:
 222                 break;
 223         }
 224
 225         if (NULL == mdoc_actions[m->last->tok].post)
 226                 return(1);
 227         return((*mdoc_actions[m->last->tok].post)(m, m->last));
 228 }
 229
 230
 231 /*
 232  * Concatenate sibling nodes together.  All siblings must be of type
 233  * MDOC_TEXT or an assertion is raised.  Concatenation is separated by a
 234  * single whitespace.
 235  */
 236 static int
 237 concat(struct mdoc *m, char *p, const struct mdoc_node *n, size_t sz)
 238 {
 239
 240         assert(sz);
 241         p[0] = '\0';
 242         for ( ; n; n = n->next) {
 243                 assert(MDOC_TEXT == n->type);
 244                 /*
 245                  * XXX: yes, these can technically be resized, but it's
 246                  * highly unlikely that we're going to get here, so let
 247                  * it slip for now.
 248                  */
 249                 if (strlcat(p, n->string, sz) >= sz) {
 250                         mdoc_nmsg(m, n, MANDOCERR_MEM);
 251                         return(0);
 252                 }
 253                 if (NULL == n->next)
 254                         continue;
 255                 if (strlcat(p, " ", sz) >= sz) {
 256                         mdoc_nmsg(m, n, MANDOCERR_MEM);
 257                         return(0);
 258                 }
 259         }
 260
 261         return(1);
 262 }
 263
 264
 265 /*
 266  * Macros accepting `-std' as an argument have the name of the current
 267  * document (`Nm') filled in as the argument if it's not provided.
 268  */
 269 static int
 270 post_std(POST_ARGS)
 271 {
 272         struct mdoc_node *nn;
 273
 274         if (n->child)
 275                 return(1);
 276         if (NULL == m->meta.name)
 277                 return(1);
 278
 279         nn = n;
 280         m->next = MDOC_NEXT_CHILD;
 281
 282         if ( ! mdoc_word_alloc(m, n->line, n->pos, m->meta.name))
 283                 return(0);
 284         m->last = nn;
 285         return(1);
 286 }
 287
 288 /*
 289  * Parse out the contents of `Dt'.  See in-line documentation for how we
 290  * handle the various fields of this macro.
 291  */
 292 static int
 293 post_dt(POST_ARGS)
 294 {
 295         struct mdoc_node *nn;
 296         const char       *cp;
 297
 298         if (m->meta.title)
 299                 free(m->meta.title);
 300         if (m->meta.vol)
 301                 free(m->meta.vol);
 302         if (m->meta.arch)
 303                 free(m->meta.arch);
 304
 305         m->meta.title = m->meta.vol = m->meta.arch = NULL;
 306         /* Handles: `.Dt'
 307          *   --> title = unknown, volume = local, msec = 0, arch = NULL
 308          */
 309
 310         if (NULL == (nn = n->child)) {
 311                 /* XXX: make these macro values. */
 312                 /* FIXME: warn about missing values. */
 313                 m->meta.title = mandoc_strdup("UNKNOWN");
 314                 m->meta.vol = mandoc_strdup("LOCAL");
 315                 m->meta.msec = mandoc_strdup("1");
 316                 return(post_prol(m, n));
 317         }
 318
 319         /* Handles: `.Dt TITLE'
 320          *   --> title = TITLE, volume = local, msec = 0, arch = NULL
 321          */
 322
 323         m->meta.title = mandoc_strdup
 324                 ('\0' == nn->string[0] ? "UNKNOWN" : nn->string);
 325
 326         if (NULL == (nn = nn->next)) {
 327                 /* FIXME: warn about missing msec. */
 328                 /* XXX: make this a macro value. */
 329                 m->meta.vol = mandoc_strdup("LOCAL");
 330                 m->meta.msec = mandoc_strdup("1");
 331                 return(post_prol(m, n));
 332         }
 333
 334         /* Handles: `.Dt TITLE SEC'
 335          *   --> title = TITLE, volume = SEC is msec ?
 336          *           format(msec) : SEC,
 337          *       msec = SEC is msec ? atoi(msec) : 0,
 338          *       arch = NULL
 339          */
 340
 341         cp = mdoc_a2msec(nn->string);
 342         if (cp) {
 343                 m->meta.vol = mandoc_strdup(cp);
 344                 m->meta.msec = mandoc_strdup(nn->string);
 345         } else if (mdoc_nmsg(m, n, MANDOCERR_BADMSEC)) {
 346                 m->meta.vol = mandoc_strdup(nn->string);
 347                 m->meta.msec = mandoc_strdup(nn->string);
 348         } else
 349                 return(0);
 350
 351         if (NULL == (nn = nn->next))
 352                 return(post_prol(m, n));
 353
 354         /* Handles: `.Dt TITLE SEC VOL'
 355          *   --> title = TITLE, volume = VOL is vol ?
 356          *       format(VOL) :
 357          *           VOL is arch ? format(arch) :
 358          *               VOL
 359          */
 360
 361         cp = mdoc_a2vol(nn->string);
 362         if (cp) {
 363                 free(m->meta.vol);
 364                 m->meta.vol = mandoc_strdup(cp);
 365         } else {
 366                 /* FIXME: warn about bad arch. */
 367                 cp = mdoc_a2arch(nn->string);
 368                 if (NULL == cp) {
 369                         free(m->meta.vol);
 370                         m->meta.vol = mandoc_strdup(nn->string);
 371                 } else
 372                         m->meta.arch = mandoc_strdup(cp);
 373         }
 374
 375         /* Ignore any subsequent parameters... */
 376         /* FIXME: warn about subsequent parameters. */
 377
 378         return(post_prol(m, n));
 379 }
 380
 381
 382 /*
 383  * Set the operating system by way of the `Os' macro.  Note that if an
 384  * argument isn't provided and -DOSNAME="\"foo\"" is provided during
 385  * compilation, this value will be used instead of filling in "sysname
 386  * release" from uname().
 387  */
 388 static int
 389 post_os(POST_ARGS)
 390 {
 391         char              buf[BUFSIZ];
 392 #ifndef OSNAME
 393         struct utsname    utsname;
 394 #endif
 395
 396         if (m->meta.os)
 397                 free(m->meta.os);
 398
 399         if ( ! concat(m, buf, n->child, BUFSIZ))
 400                 return(0);
 401
 402         /* XXX: yes, these can all be dynamically-adjusted buffers, but
 403          * it's really not worth the extra hackery.
 404          */
 405
 406         if ('\0' == buf[0]) {
 407 #ifdef OSNAME
 408                 if (strlcat(buf, OSNAME, BUFSIZ) >= BUFSIZ) {
 409                         mdoc_nmsg(m, n, MANDOCERR_MEM);
 410                         return(0);
 411                 }
 412 #else /*!OSNAME */
 413                 if (-1 == uname(&utsname))
 414                         return(mdoc_nmsg(m, n, MANDOCERR_UTSNAME));
 415
 416                 if (strlcat(buf, utsname.sysname, BUFSIZ) >= BUFSIZ) {
 417                         mdoc_nmsg(m, n, MANDOCERR_MEM);
 418                         return(0);
 419                 }
 420                 if (strlcat(buf, " ", 64) >= BUFSIZ) {
 421                         mdoc_nmsg(m, n, MANDOCERR_MEM);
 422                         return(0);
 423                 }
 424                 if (strlcat(buf, utsname.release, BUFSIZ) >= BUFSIZ) {
 425                         mdoc_nmsg(m, n, MANDOCERR_MEM);
 426                         return(0);
 427                 }
 428 #endif /*!OSNAME*/
 429         }
 430
 431         m->meta.os = mandoc_strdup(buf);
 432         return(post_prol(m, n));
 433 }
 434
 435
 436 /*
 437  * Calculate the -width for a `Bl -tag' list if it hasn't been provided.
 438  * Uses the first head macro.  NOTE AGAIN: this is ONLY if the -width
 439  * argument has NOT been provided.  See post_bl_width() for converting
 440  * the -width string.
 441  */
 442 static int
 443 post_bl_tagwidth(POST_ARGS)
 444 {
 445         struct mdoc_node *nn;
 446         size_t            sz, ssz;
 447         int               i;
 448         char              buf[NUMSIZ];
 449
 450         sz = 10;
 451
 452         for (nn = n->body->child; nn; nn = nn->next) {
 453                 if (MDOC_It != nn->tok)
 454                         continue;
 455
 456                 assert(MDOC_BLOCK == nn->type);
 457                 nn = nn->head->child;
 458
 459                 if (nn == NULL) {
 460                         /* No -width for .Bl and first .It is emtpy */
 461                         if ( ! mdoc_nmsg(m, n, MANDOCERR_NOWIDTHARG))
 462                                 return(0);
 463                         break;
 464                 }
 465
 466                 if (MDOC_TEXT == nn->type) {
 467                         sz = strlen(nn->string) + 1;
 468                         break;
 469                 }
 470
 471                 if (0 != (ssz = mdoc_macro2len(nn->tok)))
 472                         sz = ssz;
 473                 else if ( ! mdoc_nmsg(m, n, MANDOCERR_NOWIDTHARG))
 474                         return(0);
 475
 476                 break;
 477         }
 478
 479         /* Defaults to ten ens. */
 480
 481         snprintf(buf, NUMSIZ, "%zun", sz);
 482
 483         /*
 484          * We have to dynamically add this to the macro's argument list.
 485          * We're guaranteed that a MDOC_Width doesn't already exist.
 486          */
 487
 488         assert(n->args);
 489         i = (int)(n->args->argc)++;
 490
 491         n->args->argv = mandoc_realloc(n->args->argv,
 492                         n->args->argc * sizeof(struct mdoc_argv));
 493
 494         n->args->argv[i].arg = MDOC_Width;
 495         n->args->argv[i].line = n->line;
 496         n->args->argv[i].pos = n->pos;
 497         n->args->argv[i].sz = 1;
 498         n->args->argv[i].value = mandoc_malloc(sizeof(char *));
 499         n->args->argv[i].value[0] = mandoc_strdup(buf);
 500
 501         /* Set our width! */
 502         n->data.Bl->width = n->args->argv[i].value[0];
 503         return(1);
 504 }
 505
 506
 507 /*
 508  * Calculate the real width of a list from the -width string, which may
 509  * contain a macro (with a known default width), a literal string, or a
 510  * scaling width.
 511  */
 512 static int
 513 post_bl_width(POST_ARGS)
 514 {
 515         size_t            width;
 516         int               i;
 517         enum mdoct        tok;
 518         char              buf[NUMSIZ];
 519
 520         /*
 521          * If the value to -width is a macro, then we re-write it to be
 522          * the macro's width as set in share/tmac/mdoc/doc-common.
 523          */
 524
 525         if (0 == strcmp(n->data.Bl->width, "Ds"))
 526                 width = 6;
 527         else if (MDOC_MAX == (tok = mdoc_hash_find(n->data.Bl->width)))
 528                 return(1);
 529         else if (0 == (width = mdoc_macro2len(tok)))
 530                 return(mdoc_nmsg(m, n, MANDOCERR_BADWIDTH));
 531
 532         /* The value already exists: free and reallocate it. */
 533
 534         assert(n->args);
 535
 536         for (i = 0; i < (int)n->args->argc; i++)
 537                 if (MDOC_Width == n->args->argv[i].arg)
 538                         break;
 539
 540         assert(i < (int)n->args->argc);
 541
 542         snprintf(buf, NUMSIZ, "%zun", width);
 543         free(n->args->argv[i].value[0]);
 544         n->args->argv[i].value[0] = mandoc_strdup(buf);
 545
 546         /* Set our width! */
 547         n->data.Bl->width = n->args->argv[i].value[0];
 548         return(1);
 549 }
 550
 551
 552 static int
 553 post_bl(POST_ARGS)
 554 {
 555
 556         if (MDOC_BLOCK != n->type)
 557                 return(1);
 558
 559         /*
 560          * These are fairly complicated, so we've broken them into two
 561          * functions.  post_bl_tagwidth() is called when a -tag is
 562          * specified, but no -width (it must be guessed).  The second
 563          * when a -width is specified (macro indicators must be
 564          * rewritten into real lengths).
 565          */
 566
 567         if (LIST_tag == n->data.Bl->type && NULL == n->data.Bl->width) {
 568                 if ( ! post_bl_tagwidth(m, n))
 569                         return(0);
 570         } else if (NULL != n->data.Bl->width) {
 571                 if ( ! post_bl_width(m, n))
 572                         return(0);
 573         } else
 574                 return(1);
 575
 576         assert(n->data.Bl->width);
 577         return(1);
 578 }
 579
 580 /*
 581  * Parse the date field in `Dd'.
 582  */
 583 static int
 584 post_dd(POST_ARGS)
 585 {
 586         char            buf[DATESIZ];
 587
 588         if (NULL == n->child) {
 589                 m->meta.date = time(NULL);
 590                 return(post_prol(m, n));
 591         }
 592
 593         if ( ! concat(m, buf, n->child, DATESIZ))
 594                 return(0);
 595
 596         m->meta.date = mandoc_a2time
 597                 (MTIME_MDOCDATE | MTIME_CANONICAL, buf);
 598
 599         if (0 == m->meta.date) {
 600                 if ( ! mdoc_nmsg(m, n, MANDOCERR_BADDATE))
 601                         return(0);
 602                 m->meta.date = time(NULL);
 603         }
 604
 605         return(post_prol(m, n));
 606 }
 607
 608
 609 /*
 610  * Remove prologue macros from the document after they're processed.
 611  * The final document uses mdoc_meta for these values and discards the
 612  * originals.
 613  */
 614 static int
 615 post_prol(POST_ARGS)
 616 {
 617
 618         mdoc_node_delete(m, n);
 619         if (m->meta.title && m->meta.date && m->meta.os)
 620                 m->flags |= MDOC_PBODY;
 621         return(1);
 622 }