mdoc.c

   1 /* $Id: mdoc.c,v 1.51 2009/03/05 13:12:12 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((u_char)buf[i]) &&
 224                         i < (int)sizeof(tmp))
 225                 i++;
 226
 227         if (i == (int)sizeof(tmp)) {
 228                 mdoc->flags |= MDOC_HALT;
 229                 return(mdoc_perr(mdoc, line, 1, "unknown macro"));
 230         } else if (i <= 2) {
 231                 mdoc->flags |= MDOC_HALT;
 232                 return(mdoc_perr(mdoc, line, 1, "unknown macro"));
 233         }
 234
 235         i--;
 236
 237         (void)memcpy(tmp, buf + 1, (size_t)i);
 238         tmp[i++] = 0;
 239
 240         if (MDOC_MAX == (c = mdoc_find(mdoc, tmp))) {
 241                 mdoc->flags |= MDOC_HALT;
 242                 return(mdoc_perr(mdoc, line, 1, "unknown macro"));
 243         }
 244
 245         while (buf[i] && isspace((u_char)buf[i]))
 246                 i++;
 247
 248         if ( ! mdoc_macro(mdoc, c, line, 1, &i, buf)) {
 249                 mdoc->flags |= MDOC_HALT;
 250                 return(0);
 251         }
 252
 253         return(1);
 254 }
 255
 256
 257 void
 258 mdoc_vmsg(struct mdoc *mdoc, int ln, int pos, const char *fmt, ...)
 259 {
 260         char              buf[256];
 261         va_list           ap;
 262
 263         if (NULL == mdoc->cb.mdoc_msg)
 264                 return;
 265
 266         va_start(ap, fmt);
 267         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 268         va_end(ap);
 269         (*mdoc->cb.mdoc_msg)(mdoc->data, ln, pos, buf);
 270 }
 271
 272
 273 int
 274 mdoc_verr(struct mdoc *mdoc, int ln, int pos,
 275                 const char *fmt, ...)
 276 {
 277         char             buf[256];
 278         va_list          ap;
 279
 280         if (NULL == mdoc->cb.mdoc_err)
 281                 return(0);
 282
 283         va_start(ap, fmt);
 284         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 285         va_end(ap);
 286         return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf));
 287 }
 288
 289
 290 int
 291 mdoc_vwarn(struct mdoc *mdoc, int ln, int pos,
 292                 enum mdoc_warn type, const char *fmt, ...)
 293 {
 294         char             buf[256];
 295         va_list          ap;
 296
 297         if (NULL == mdoc->cb.mdoc_warn)
 298                 return(0);
 299
 300         va_start(ap, fmt);
 301         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 302         va_end(ap);
 303         return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, type, buf));
 304 }
 305
 306
 307 int
 308 mdoc_macro(struct mdoc *mdoc, int tok,
 309                 int ln, int ppos, int *pos, char *buf)
 310 {
 311
 312         assert(mdoc_macros[tok].fp);
 313
 314         if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
 315                         SEC_PROLOGUE != mdoc->lastnamed)
 316                 return(mdoc_perr(mdoc, ln, ppos, "macro disallowed in document body"));
 317         if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
 318                         SEC_PROLOGUE == mdoc->lastnamed)
 319                 return(mdoc_perr(mdoc, ln, ppos, "macro disallowed in document prologue"));
 320         if (1 != ppos && ! (MDOC_CALLABLE & mdoc_macros[tok].flags))
 321                 return(mdoc_perr(mdoc, ln, ppos, "macro not callable"));
 322         return((*mdoc_macros[tok].fp)(mdoc, tok, ln, ppos, pos, buf));
 323 }
 324
 325
 326 static int
 327 mdoc_node_append(struct mdoc *mdoc, struct mdoc_node *p)
 328 {
 329         const char       *nn, *nt, *on, *ot, *act;
 330
 331         assert(mdoc->last);
 332         assert(mdoc->first);
 333         assert(MDOC_ROOT != p->type);
 334
 335         /* See if we exceed the suggest line-max. */
 336
 337         switch (p->type) {
 338         case (MDOC_TEXT):
 339                 /* FALLTHROUGH */
 340         case (MDOC_ELEM):
 341                 /* FALLTHROUGH */
 342         case (MDOC_BLOCK):
 343                 mdoc->linetok++;
 344                 break;
 345         default:
 346                 break;
 347         }
 348
 349         /* This sort-of works (re-opening of text macros...). */
 350         if (mdoc->linetok > MDOC_LINEARG_SOFTMAX)
 351                 if ( ! mdoc_nwarn(mdoc, p, WARN_COMPAT,
 352                                         "suggested %d tokens per line exceeded (has %d)",
 353                                         MDOC_LINEARG_SOFTMAX, mdoc->linetok))
 354                         return(0);
 355
 356         if (MDOC_TEXT == mdoc->last->type)
 357                 on = "<text>";
 358         else if (MDOC_ROOT == mdoc->last->type)
 359                 on = "<root>";
 360         else
 361                 on = mdoc_macronames[mdoc->last->tok];
 362
 363         if (MDOC_TEXT == p->type)
 364                 nn = "<text>";
 365         else if (MDOC_ROOT == p->type)
 366                 nn = "<root>";
 367         else
 368                 nn = mdoc_macronames[p->tok];
 369
 370         ot = mdoc_type2a(mdoc->last->type);
 371         nt = mdoc_type2a(p->type);
 372
 373         switch (mdoc->next) {
 374         case (MDOC_NEXT_SIBLING):
 375                 mdoc->last->next = p;
 376                 p->prev = mdoc->last;
 377                 p->parent = mdoc->last->parent;
 378                 act = "sibling";
 379                 break;
 380         case (MDOC_NEXT_CHILD):
 381                 mdoc->last->child = p;
 382                 p->parent = mdoc->last;
 383                 act = "child";
 384                 break;
 385         default:
 386                 abort();
 387                 /* NOTREACHED */
 388         }
 389
 390         if ( ! mdoc_valid_pre(mdoc, p))
 391                 return(0);
 392
 393         switch (p->type) {
 394         case (MDOC_HEAD):
 395                 assert(MDOC_BLOCK == p->parent->type);
 396                 p->parent->data.block.head = p;
 397                 break;
 398         case (MDOC_TAIL):
 399                 assert(MDOC_BLOCK == p->parent->type);
 400                 p->parent->data.block.tail = p;
 401                 break;
 402         case (MDOC_BODY):
 403                 assert(MDOC_BLOCK == p->parent->type);
 404                 p->parent->data.block.body = p;
 405                 break;
 406         default:
 407                 break;
 408         }
 409
 410         mdoc->last = p;
 411         mdoc_msg(mdoc, "parse: %s `%s' %s of %s `%s'",
 412                         nt, nn, act, ot, on);
 413         return(1);
 414 }
 415
 416
 417 static struct mdoc_node *
 418 mdoc_node_alloc(const struct mdoc *mdoc)
 419 {
 420         struct mdoc_node *p;
 421
 422         p = xcalloc(1, sizeof(struct mdoc_node));
 423         p->sec = mdoc->lastsec;
 424
 425         return(p);
 426 }
 427
 428
 429 int
 430 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 431 {
 432         struct mdoc_node *p;
 433
 434         assert(mdoc->first);
 435         assert(mdoc->last);
 436
 437         p = mdoc_node_alloc(mdoc);
 438
 439         p->line = line;
 440         p->pos = pos;
 441         p->type = MDOC_TAIL;
 442         p->tok = tok;
 443
 444         return(mdoc_node_append(mdoc, p));
 445 }
 446
 447
 448 int
 449 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 450 {
 451         struct mdoc_node *p;
 452
 453         assert(mdoc->first);
 454         assert(mdoc->last);
 455
 456         p = mdoc_node_alloc(mdoc);
 457
 458         p->line = line;
 459         p->pos = pos;
 460         p->type = MDOC_HEAD;
 461         p->tok = tok;
 462
 463         return(mdoc_node_append(mdoc, p));
 464 }
 465
 466
 467 int
 468 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 469 {
 470         struct mdoc_node *p;
 471
 472         assert(mdoc->first);
 473         assert(mdoc->last);
 474
 475         p = mdoc_node_alloc(mdoc);
 476
 477         p->line = line;
 478         p->pos = pos;
 479         p->type = MDOC_BODY;
 480         p->tok = tok;
 481
 482         return(mdoc_node_append(mdoc, p));
 483 }
 484
 485
 486 int
 487 mdoc_root_alloc(struct mdoc *mdoc)
 488 {
 489         struct mdoc_node *p;
 490
 491         p = mdoc_node_alloc(mdoc);
 492
 493         p->type = MDOC_ROOT;
 494
 495         return(mdoc_node_append(mdoc, p));
 496 }
 497
 498
 499 int
 500 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 501                 int tok, size_t argsz, const struct mdoc_arg *args)
 502 {
 503         struct mdoc_node *p;
 504
 505         p = mdoc_node_alloc(mdoc);
 506
 507         p->pos = pos;
 508         p->line = line;
 509         p->type = MDOC_BLOCK;
 510         p->tok = tok;
 511         p->data.block.argc = argsz;
 512         p->data.block.argv = argdup(argsz, args);
 513
 514         return(mdoc_node_append(mdoc, p));
 515 }
 516
 517
 518 int
 519 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 520                 int tok, size_t argsz, const struct mdoc_arg *args)
 521 {
 522         struct mdoc_node *p;
 523
 524         p = mdoc_node_alloc(mdoc);
 525
 526         p->line = line;
 527         p->pos = pos;
 528         p->type = MDOC_ELEM;
 529         p->tok = tok;
 530         p->data.elem.argc = argsz;
 531         p->data.elem.argv = argdup(argsz, args);
 532
 533         return(mdoc_node_append(mdoc, p));
 534 }
 535
 536
 537 int
 538 mdoc_word_alloc(struct mdoc *mdoc,
 539                 int line, int pos, const char *word)
 540 {
 541         struct mdoc_node *p;
 542
 543         p = mdoc_node_alloc(mdoc);
 544
 545         p->line = line;
 546         p->pos = pos;
 547         p->type = MDOC_TEXT;
 548         p->data.text.string = xstrdup(word);
 549
 550         return(mdoc_node_append(mdoc, p));
 551 }
 552
 553
 554 static void
 555 argfree(size_t sz, struct mdoc_arg *p)
 556 {
 557         int              i, j;
 558
 559         if (0 == sz)
 560                 return;
 561
 562         assert(p);
 563         /* LINTED */
 564         for (i = 0; i < (int)sz; i++)
 565                 if (p[i].sz > 0) {
 566                         assert(p[i].value);
 567                         /* LINTED */
 568                         for (j = 0; j < (int)p[i].sz; j++)
 569                                 free(p[i].value[j]);
 570                         free(p[i].value);
 571                 }
 572         free(p);
 573 }
 574
 575
 576 static void
 577 mdoc_elem_free(struct mdoc_elem *p)
 578 {
 579
 580         argfree(p->argc, p->argv);
 581 }
 582
 583
 584 static void
 585 mdoc_block_free(struct mdoc_block *p)
 586 {
 587
 588         argfree(p->argc, p->argv);
 589 }
 590
 591
 592 static void
 593 mdoc_text_free(struct mdoc_text *p)
 594 {
 595
 596         if (p->string)
 597                 free(p->string);
 598 }
 599
 600
 601 void
 602 mdoc_node_free(struct mdoc_node *p)
 603 {
 604
 605         switch (p->type) {
 606         case (MDOC_TEXT):
 607                 mdoc_text_free(&p->data.text);
 608                 break;
 609         case (MDOC_ELEM):
 610                 mdoc_elem_free(&p->data.elem);
 611                 break;
 612         case (MDOC_BLOCK):
 613                 mdoc_block_free(&p->data.block);
 614                 break;
 615         default:
 616                 break;
 617         }
 618
 619         free(p);
 620 }
 621
 622
 623 void
 624 mdoc_node_freelist(struct mdoc_node *p)
 625 {
 626
 627         if (p->child)
 628                 mdoc_node_freelist(p->child);
 629         if (p->next)
 630                 mdoc_node_freelist(p->next);
 631
 632         mdoc_node_free(p);
 633 }
 634
 635
 636 int
 637 mdoc_find(const struct mdoc *mdoc, const char *key)
 638 {
 639
 640         return(mdoc_tokhash_find(mdoc->htab, key));
 641 }
 642
 643
 644 static void
 645 argcpy(struct mdoc_arg *dst, const struct mdoc_arg *src)
 646 {
 647         int              i;
 648
 649         dst->line = src->line;
 650         dst->pos = src->pos;
 651         dst->arg = src->arg;
 652         if (0 == (dst->sz = src->sz))
 653                 return;
 654         dst->value = xcalloc(dst->sz, sizeof(char *));
 655         for (i = 0; i < (int)dst->sz; i++)
 656                 dst->value[i] = xstrdup(src->value[i]);
 657 }
 658
 659
 660 static struct mdoc_arg *
 661 argdup(size_t argsz, const struct mdoc_arg *args)
 662 {
 663         struct mdoc_arg *pp;
 664         int              i;
 665
 666         if (0 == argsz)
 667                 return(NULL);
 668
 669         pp = xcalloc((size_t)argsz, sizeof(struct mdoc_arg));
 670         for (i = 0; i < (int)argsz; i++)
 671                 argcpy(&pp[i], &args[i]);
 672
 673         return(pp);
 674 }
 675
 676
 677 /* FIXME: deprecate. */
 678 char *
 679 mdoc_node2a(struct mdoc_node *node)
 680 {
 681         static char      buf[64];
 682
 683         assert(node);
 684
 685         buf[0] = 0;
 686         (void)xstrlcat(buf, mdoc_type2a(node->type), 64);
 687         if (MDOC_ROOT == node->type)
 688                 return(buf);
 689         (void)xstrlcat(buf, " `", 64);
 690         if (MDOC_TEXT == node->type)
 691                 (void)xstrlcat(buf, node->data.text.string, 64);
 692         else
 693                 (void)xstrlcat(buf, mdoc_macronames[node->tok], 64);
 694         (void)xstrlcat(buf, "'", 64);
 695
 696         return(buf);
 697 }
 698
 699