mdoc.c

   1 /* $Id: mdoc.c,v 1.50 2009/03/02 17:14:46 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008 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
   7  * above copyright notice and this permission notice appear in all
   8  * copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL
  11  * WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED
  12  * WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE
  13  * AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL
  14  * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR
  15  * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
  16  * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
  17  * PERFORMANCE OF THIS SOFTWARE.
  18  */
  19 #include <assert.h>
  20 #include <ctype.h>
  21 #include <err.h>
  22 #include <stdarg.h>
  23 #include <stdlib.h>
  24 #include <stdio.h>
  25 #include <string.h>
  26
  27 #include "private.h"
  28
  29 /*
  30  * Main caller in the libmdoc library.  This begins the parsing routine,
  31  * handles allocation of data, and so forth.  Most of the "work" is done
  32  * in macro.c and validate.c.
  33  */
  34
  35 static  struct mdoc_arg  *argdup(size_t, const struct mdoc_arg *);
  36 static  void              argfree(size_t, struct mdoc_arg *);
  37 static  void              argcpy(struct mdoc_arg *,
  38                                 const struct mdoc_arg *);
  39
  40 static  struct mdoc_node *mdoc_node_alloc(const struct mdoc *);
  41 static  int               mdoc_node_append(struct mdoc *,
  42                                 struct mdoc_node *);
  43 static  void              mdoc_elem_free(struct mdoc_elem *);
  44 static  void              mdoc_text_free(struct mdoc_text *);
  45
  46
  47 const   char *const __mdoc_macronames[MDOC_MAX] = {
  48         "\\\"",         "Dd",           "Dt",           "Os",
  49         "Sh",           "Ss",           "Pp",           "D1",
  50         "Dl",           "Bd",           "Ed",           "Bl",
  51         "El",           "It",           "Ad",           "An",
  52         "Ar",           "Cd",           "Cm",           "Dv",
  53         "Er",           "Ev",           "Ex",           "Fa",
  54         "Fd",           "Fl",           "Fn",           "Ft",
  55         "Ic",           "In",           "Li",           "Nd",
  56         "Nm",           "Op",           "Ot",           "Pa",
  57         "Rv",           "St",           "Va",           "Vt",
  58         /* LINTED */
  59         "Xr",           "\%A",          "\%B",          "\%D",
  60         /* LINTED */
  61         "\%I",          "\%J",          "\%N",          "\%O",
  62         /* LINTED */
  63         "\%P",          "\%R",          "\%T",          "\%V",
  64         "Ac",           "Ao",           "Aq",           "At",
  65         "Bc",           "Bf",           "Bo",           "Bq",
  66         "Bsx",          "Bx",           "Db",           "Dc",
  67         "Do",           "Dq",           "Ec",           "Ef",
  68         "Em",           "Eo",           "Fx",           "Ms",
  69         "No",           "Ns",           "Nx",           "Ox",
  70         "Pc",           "Pf",           "Po",           "Pq",
  71         "Qc",           "Ql",           "Qo",           "Qq",
  72         "Re",           "Rs",           "Sc",           "So",
  73         "Sq",           "Sm",           "Sx",           "Sy",
  74         "Tn",           "Ux",           "Xc",           "Xo",
  75         "Fo",           "Fc",           "Oo",           "Oc",
  76         "Bk",           "Ek",           "Bt",           "Hf",
  77         "Fr",           "Ud",
  78         };
  79
  80 const   char *const __mdoc_argnames[MDOC_ARG_MAX] = {
  81         "split",                "nosplit",              "ragged",
  82         "unfilled",             "literal",              "file",
  83         "offset",               "bullet",               "dash",
  84         "hyphen",               "item",                 "enum",
  85         "tag",                  "diag",                 "hang",
  86         "ohang",                "inset",                "column",
  87         "width",                "compact",              "std",
  88         "p1003.1-88",           "p1003.1-90",           "p1003.1-96",
  89         "p1003.1-2001",         "p1003.1-2004",         "p1003.1",
  90         "p1003.1b",             "p1003.1b-93",          "p1003.1c-95",
  91         "p1003.1g-2000",        "p1003.2-92",           "p1387.2-95",
  92         "p1003.2",              "p1387.2",              "isoC-90",
  93         "isoC-amd1",            "isoC-tcor1",           "isoC-tcor2",
  94         "isoC-99",              "ansiC",                "ansiC-89",
  95         "ansiC-99",             "ieee754",              "iso8802-3",
  96         "xpg3",                 "xpg4",                 "xpg4.2",
  97         "xpg4.3",               "xbd5",                 "xcu5",
  98         "xsh5",                 "xns5",                 "xns5.2d2.0",
  99         "xcurses4.2",           "susv2",                "susv3",
 100         "svid4",                "filled",               "words",
 101         "emphasis",             "symbolic",
 102         };
 103
 104 const   char * const *mdoc_macronames = __mdoc_macronames;
 105 const   char * const *mdoc_argnames = __mdoc_argnames;
 106
 107
 108 const struct mdoc_node *
 109 mdoc_node(const struct mdoc *mdoc)
 110 {
 111
 112         return(mdoc->first);
 113 }
 114
 115
 116 const struct mdoc_meta *
 117 mdoc_meta(const struct mdoc *mdoc)
 118 {
 119
 120         return(&mdoc->meta);
 121 }
 122
 123
 124 void
 125 mdoc_free(struct mdoc *mdoc)
 126 {
 127
 128         if (mdoc->first)
 129                 mdoc_node_freelist(mdoc->first);
 130         if (mdoc->htab)
 131                 mdoc_tokhash_free(mdoc->htab);
 132         if (mdoc->meta.title)
 133                 free(mdoc->meta.title);
 134         if (mdoc->meta.os)
 135                 free(mdoc->meta.os);
 136         if (mdoc->meta.name)
 137                 free(mdoc->meta.name);
 138
 139         free(mdoc);
 140 }
 141
 142
 143 struct mdoc *
 144 mdoc_alloc(void *data, const struct mdoc_cb *cb)
 145 {
 146         struct mdoc     *p;
 147
 148         p = xcalloc(1, sizeof(struct mdoc));
 149
 150         p->data = data;
 151         if (cb)
 152                 (void)memcpy(&p->cb, cb, sizeof(struct mdoc_cb));
 153
 154         p->last = xcalloc(1, sizeof(struct mdoc_node));
 155         p->last->type = MDOC_ROOT;
 156         p->first = p->last;
 157
 158         p->next = MDOC_NEXT_CHILD;
 159         p->htab = mdoc_tokhash_alloc();
 160
 161         return(p);
 162 }
 163
 164
 165 int
 166 mdoc_endparse(struct mdoc *mdoc)
 167 {
 168
 169         if (MDOC_HALT & mdoc->flags)
 170                 return(0);
 171         if (NULL == mdoc->first)
 172                 return(1);
 173
 174         assert(mdoc->last);
 175         if ( ! macro_end(mdoc)) {
 176                 mdoc->flags |= MDOC_HALT;
 177                 return(0);
 178         }
 179         return(1);
 180 }
 181
 182
 183 /*
 184  * Main line-parsing routine.  If the line is a macro-line (started with
 185  * a '.' control character), then pass along to the parser, which parses
 186  * subsequent macros until the end of line.  If normal text, simply
 187  * append the entire line to the chain.
 188  */
 189 int
 190 mdoc_parseln(struct mdoc *mdoc, int line, char *buf)
 191 {
 192         int               c, i;
 193         char              tmp[5];
 194
 195         if (MDOC_HALT & mdoc->flags)
 196                 return(0);
 197
 198         mdoc->linetok = 0;
 199
 200         if ('.' != *buf) {
 201                 /*
 202                  * Free-form text.  Not allowed in the prologue.
 203                  */
 204                 if (SEC_PROLOGUE == mdoc->lastnamed)
 205                         return(mdoc_perr(mdoc, line, 0,
 206                                         "no text in prologue"));
 207
 208                 if ( ! mdoc_word_alloc(mdoc, line, 0, buf))
 209                         return(0);
 210                 mdoc->next = MDOC_NEXT_SIBLING;
 211                 return(1);
 212         }
 213
 214         /*
 215          * Control-character detected.  Begin the parsing sequence.
 216          */
 217
 218         if (buf[1] && '\\' == buf[1])
 219                 if (buf[2] && '\"' == buf[2])
 220                         return(1);
 221
 222         i = 1;
 223         while (buf[i] && ! isspace((int)buf[i]) && i < (int)sizeof(tmp))
 224                 i++;
 225
 226         if (i == (int)sizeof(tmp)) {
 227                 mdoc->flags |= MDOC_HALT;
 228                 return(mdoc_perr(mdoc, line, 1, "unknown macro"));
 229         } else if (i <= 2) {
 230                 mdoc->flags |= MDOC_HALT;
 231                 return(mdoc_perr(mdoc, line, 1, "unknown macro"));
 232         }
 233
 234         i--;
 235
 236         (void)memcpy(tmp, buf + 1, (size_t)i);
 237         tmp[i++] = 0;
 238
 239         if (MDOC_MAX == (c = mdoc_find(mdoc, tmp))) {
 240                 mdoc->flags |= MDOC_HALT;
 241                 return(mdoc_perr(mdoc, line, 1, "unknown macro"));
 242         }
 243
 244         while (buf[i] && isspace((int)buf[i]))
 245                 i++;
 246
 247         if ( ! mdoc_macro(mdoc, c, line, 1, &i, buf)) {
 248                 mdoc->flags |= MDOC_HALT;
 249                 return(0);
 250         }
 251
 252         return(1);
 253 }
 254
 255
 256 void
 257 mdoc_vmsg(struct mdoc *mdoc, int ln, int pos, const char *fmt, ...)
 258 {
 259         char              buf[256];
 260         va_list           ap;
 261
 262         if (NULL == mdoc->cb.mdoc_msg)
 263                 return;
 264
 265         va_start(ap, fmt);
 266         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 267         va_end(ap);
 268         (*mdoc->cb.mdoc_msg)(mdoc->data, ln, pos, buf);
 269 }
 270
 271
 272 int
 273 mdoc_verr(struct mdoc *mdoc, int ln, int pos,
 274                 const char *fmt, ...)
 275 {
 276         char             buf[256];
 277         va_list          ap;
 278
 279         if (NULL == mdoc->cb.mdoc_err)
 280                 return(0);
 281
 282         va_start(ap, fmt);
 283         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 284         va_end(ap);
 285         return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf));
 286 }
 287
 288
 289 int
 290 mdoc_vwarn(struct mdoc *mdoc, int ln, int pos,
 291                 enum mdoc_warn type, const char *fmt, ...)
 292 {
 293         char             buf[256];
 294         va_list          ap;
 295
 296         if (NULL == mdoc->cb.mdoc_warn)
 297                 return(0);
 298
 299         va_start(ap, fmt);
 300         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 301         va_end(ap);
 302         return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, type, buf));
 303 }
 304
 305
 306 int
 307 mdoc_macro(struct mdoc *mdoc, int tok,
 308                 int ln, int ppos, int *pos, char *buf)
 309 {
 310
 311         assert(mdoc_macros[tok].fp);
 312
 313         if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
 314                         SEC_PROLOGUE != mdoc->lastnamed)
 315                 return(mdoc_perr(mdoc, ln, ppos, "macro disallowed in document body"));
 316         if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
 317                         SEC_PROLOGUE == mdoc->lastnamed)
 318                 return(mdoc_perr(mdoc, ln, ppos, "macro disallowed in document prologue"));
 319         if (1 != ppos && ! (MDOC_CALLABLE & mdoc_macros[tok].flags))
 320                 return(mdoc_perr(mdoc, ln, ppos, "macro not callable"));
 321         return((*mdoc_macros[tok].fp)(mdoc, tok, ln, ppos, pos, buf));
 322 }
 323
 324
 325 static int
 326 mdoc_node_append(struct mdoc *mdoc, struct mdoc_node *p)
 327 {
 328         const char       *nn, *nt, *on, *ot, *act;
 329
 330         assert(mdoc->last);
 331         assert(mdoc->first);
 332         assert(MDOC_ROOT != p->type);
 333
 334         /* See if we exceed the suggest line-max. */
 335
 336         switch (p->type) {
 337         case (MDOC_TEXT):
 338                 /* FALLTHROUGH */
 339         case (MDOC_ELEM):
 340                 /* FALLTHROUGH */
 341         case (MDOC_BLOCK):
 342                 mdoc->linetok++;
 343                 break;
 344         default:
 345                 break;
 346         }
 347
 348         /* This sort-of works (re-opening of text macros...). */
 349         if (mdoc->linetok > MDOC_LINEARG_SOFTMAX)
 350                 if ( ! mdoc_nwarn(mdoc, p, WARN_COMPAT,
 351                                         "suggested %d tokens per line exceeded (has %d)",
 352                                         MDOC_LINEARG_SOFTMAX, mdoc->linetok))
 353                         return(0);
 354
 355         if (MDOC_TEXT == mdoc->last->type)
 356                 on = "<text>";
 357         else if (MDOC_ROOT == mdoc->last->type)
 358                 on = "<root>";
 359         else
 360                 on = mdoc_macronames[mdoc->last->tok];
 361
 362         if (MDOC_TEXT == p->type)
 363                 nn = "<text>";
 364         else if (MDOC_ROOT == p->type)
 365                 nn = "<root>";
 366         else
 367                 nn = mdoc_macronames[p->tok];
 368
 369         ot = mdoc_type2a(mdoc->last->type);
 370         nt = mdoc_type2a(p->type);
 371
 372         switch (mdoc->next) {
 373         case (MDOC_NEXT_SIBLING):
 374                 mdoc->last->next = p;
 375                 p->prev = mdoc->last;
 376                 p->parent = mdoc->last->parent;
 377                 act = "sibling";
 378                 break;
 379         case (MDOC_NEXT_CHILD):
 380                 mdoc->last->child = p;
 381                 p->parent = mdoc->last;
 382                 act = "child";
 383                 break;
 384         default:
 385                 abort();
 386                 /* NOTREACHED */
 387         }
 388
 389         if ( ! mdoc_valid_pre(mdoc, p))
 390                 return(0);
 391
 392         switch (p->type) {
 393         case (MDOC_HEAD):
 394                 assert(MDOC_BLOCK == p->parent->type);
 395                 p->parent->data.block.head = p;
 396                 break;
 397         case (MDOC_TAIL):
 398                 assert(MDOC_BLOCK == p->parent->type);
 399                 p->parent->data.block.tail = p;
 400                 break;
 401         case (MDOC_BODY):
 402                 assert(MDOC_BLOCK == p->parent->type);
 403                 p->parent->data.block.body = p;
 404                 break;
 405         default:
 406                 break;
 407         }
 408
 409         mdoc->last = p;
 410         mdoc_msg(mdoc, "parse: %s `%s' %s of %s `%s'",
 411                         nt, nn, act, ot, on);
 412         return(1);
 413 }
 414
 415
 416 static struct mdoc_node *
 417 mdoc_node_alloc(const struct mdoc *mdoc)
 418 {
 419         struct mdoc_node *p;
 420
 421         p = xcalloc(1, sizeof(struct mdoc_node));
 422         p->sec = mdoc->lastsec;
 423
 424         return(p);
 425 }
 426
 427
 428 int
 429 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 430 {
 431         struct mdoc_node *p;
 432
 433         assert(mdoc->first);
 434         assert(mdoc->last);
 435
 436         p = mdoc_node_alloc(mdoc);
 437
 438         p->line = line;
 439         p->pos = pos;
 440         p->type = MDOC_TAIL;
 441         p->tok = tok;
 442
 443         return(mdoc_node_append(mdoc, p));
 444 }
 445
 446
 447 int
 448 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 449 {
 450         struct mdoc_node *p;
 451
 452         assert(mdoc->first);
 453         assert(mdoc->last);
 454
 455         p = mdoc_node_alloc(mdoc);
 456
 457         p->line = line;
 458         p->pos = pos;
 459         p->type = MDOC_HEAD;
 460         p->tok = tok;
 461
 462         return(mdoc_node_append(mdoc, p));
 463 }
 464
 465
 466 int
 467 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 468 {
 469         struct mdoc_node *p;
 470
 471         assert(mdoc->first);
 472         assert(mdoc->last);
 473
 474         p = mdoc_node_alloc(mdoc);
 475
 476         p->line = line;
 477         p->pos = pos;
 478         p->type = MDOC_BODY;
 479         p->tok = tok;
 480
 481         return(mdoc_node_append(mdoc, p));
 482 }
 483
 484
 485 int
 486 mdoc_root_alloc(struct mdoc *mdoc)
 487 {
 488         struct mdoc_node *p;
 489
 490         p = mdoc_node_alloc(mdoc);
 491
 492         p->type = MDOC_ROOT;
 493
 494         return(mdoc_node_append(mdoc, p));
 495 }
 496
 497
 498 int
 499 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 500                 int tok, size_t argsz, const struct mdoc_arg *args)
 501 {
 502         struct mdoc_node *p;
 503
 504         p = mdoc_node_alloc(mdoc);
 505
 506         p->pos = pos;
 507         p->line = line;
 508         p->type = MDOC_BLOCK;
 509         p->tok = tok;
 510         p->data.block.argc = argsz;
 511         p->data.block.argv = argdup(argsz, args);
 512
 513         return(mdoc_node_append(mdoc, p));
 514 }
 515
 516
 517 int
 518 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 519                 int tok, size_t argsz, const struct mdoc_arg *args)
 520 {
 521         struct mdoc_node *p;
 522
 523         p = mdoc_node_alloc(mdoc);
 524
 525         p->line = line;
 526         p->pos = pos;
 527         p->type = MDOC_ELEM;
 528         p->tok = tok;
 529         p->data.elem.argc = argsz;
 530         p->data.elem.argv = argdup(argsz, args);
 531
 532         return(mdoc_node_append(mdoc, p));
 533 }
 534
 535
 536 int
 537 mdoc_word_alloc(struct mdoc *mdoc,
 538                 int line, int pos, const char *word)
 539 {
 540         struct mdoc_node *p;
 541
 542         p = mdoc_node_alloc(mdoc);
 543
 544         p->line = line;
 545         p->pos = pos;
 546         p->type = MDOC_TEXT;
 547         p->data.text.string = xstrdup(word);
 548
 549         return(mdoc_node_append(mdoc, p));
 550 }
 551
 552
 553 static void
 554 argfree(size_t sz, struct mdoc_arg *p)
 555 {
 556         int              i, j;
 557
 558         if (0 == sz)
 559                 return;
 560
 561         assert(p);
 562         /* LINTED */
 563         for (i = 0; i < (int)sz; i++)
 564                 if (p[i].sz > 0) {
 565                         assert(p[i].value);
 566                         /* LINTED */
 567                         for (j = 0; j < (int)p[i].sz; j++)
 568                                 free(p[i].value[j]);
 569                         free(p[i].value);
 570                 }
 571         free(p);
 572 }
 573
 574
 575 static void
 576 mdoc_elem_free(struct mdoc_elem *p)
 577 {
 578
 579         argfree(p->argc, p->argv);
 580 }
 581
 582
 583 static void
 584 mdoc_block_free(struct mdoc_block *p)
 585 {
 586
 587         argfree(p->argc, p->argv);
 588 }
 589
 590
 591 static void
 592 mdoc_text_free(struct mdoc_text *p)
 593 {
 594
 595         if (p->string)
 596                 free(p->string);
 597 }
 598
 599
 600 void
 601 mdoc_node_free(struct mdoc_node *p)
 602 {
 603
 604         switch (p->type) {
 605         case (MDOC_TEXT):
 606                 mdoc_text_free(&p->data.text);
 607                 break;
 608         case (MDOC_ELEM):
 609                 mdoc_elem_free(&p->data.elem);
 610                 break;
 611         case (MDOC_BLOCK):
 612                 mdoc_block_free(&p->data.block);
 613                 break;
 614         default:
 615                 break;
 616         }
 617
 618         free(p);
 619 }
 620
 621
 622 void
 623 mdoc_node_freelist(struct mdoc_node *p)
 624 {
 625
 626         if (p->child)
 627                 mdoc_node_freelist(p->child);
 628         if (p->next)
 629                 mdoc_node_freelist(p->next);
 630
 631         mdoc_node_free(p);
 632 }
 633
 634
 635 int
 636 mdoc_find(const struct mdoc *mdoc, const char *key)
 637 {
 638
 639         return(mdoc_tokhash_find(mdoc->htab, key));
 640 }
 641
 642
 643 static void
 644 argcpy(struct mdoc_arg *dst, const struct mdoc_arg *src)
 645 {
 646         int              i;
 647
 648         dst->line = src->line;
 649         dst->pos = src->pos;
 650         dst->arg = src->arg;
 651         if (0 == (dst->sz = src->sz))
 652                 return;
 653         dst->value = xcalloc(dst->sz, sizeof(char *));
 654         for (i = 0; i < (int)dst->sz; i++)
 655                 dst->value[i] = xstrdup(src->value[i]);
 656 }
 657
 658
 659 static struct mdoc_arg *
 660 argdup(size_t argsz, const struct mdoc_arg *args)
 661 {
 662         struct mdoc_arg *pp;
 663         int              i;
 664
 665         if (0 == argsz)
 666                 return(NULL);
 667
 668         pp = xcalloc((size_t)argsz, sizeof(struct mdoc_arg));
 669         for (i = 0; i < (int)argsz; i++)
 670                 argcpy(&pp[i], &args[i]);
 671
 672         return(pp);
 673 }
 674
 675
 676 /* FIXME: deprecate. */
 677 char *
 678 mdoc_node2a(struct mdoc_node *node)
 679 {
 680         static char      buf[64];
 681
 682         assert(node);
 683
 684         buf[0] = 0;
 685         (void)xstrlcat(buf, mdoc_type2a(node->type), 64);
 686         if (MDOC_ROOT == node->type)
 687                 return(buf);
 688         (void)xstrlcat(buf, " `", 64);
 689         if (MDOC_TEXT == node->type)
 690                 (void)xstrlcat(buf, node->data.text.string, 64);
 691         else
 692                 (void)xstrlcat(buf, mdoc_macronames[node->tok], 64);
 693         (void)xstrlcat(buf, "'", 64);
 694
 695         return(buf);
 696 }
 697
 698