mdoc.c

   1 /*      $Id: mdoc.c,v 1.84 2009/06/17 07:59:47 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 <assert.h>
  18 #include <ctype.h>
  19 #include <stdarg.h>
  20 #include <stdio.h>
  21 #include <stdlib.h>
  22 #include <string.h>
  23
  24 #include "libmdoc.h"
  25
  26 enum    merr {
  27         ENOCALL,
  28         EBODYPROL,
  29         EPROLBODY,
  30         ESPACE,
  31         ETEXTPROL,
  32         ENOBLANK,
  33         EMALLOC
  34 };
  35
  36 const   char *const __mdoc_macronames[MDOC_MAX] = {
  37         "Ap",           "Dd",           "Dt",           "Os",
  38         "Sh",           "Ss",           "Pp",           "D1",
  39         "Dl",           "Bd",           "Ed",           "Bl",
  40         "El",           "It",           "Ad",           "An",
  41         "Ar",           "Cd",           "Cm",           "Dv",
  42         "Er",           "Ev",           "Ex",           "Fa",
  43         "Fd",           "Fl",           "Fn",           "Ft",
  44         "Ic",           "In",           "Li",           "Nd",
  45         "Nm",           "Op",           "Ot",           "Pa",
  46         "Rv",           "St",           "Va",           "Vt",
  47         /* LINTED */
  48         "Xr",           "\%A",          "\%B",          "\%D",
  49         /* LINTED */
  50         "\%I",          "\%J",          "\%N",          "\%O",
  51         /* LINTED */
  52         "\%P",          "\%R",          "\%T",          "\%V",
  53         "Ac",           "Ao",           "Aq",           "At",
  54         "Bc",           "Bf",           "Bo",           "Bq",
  55         "Bsx",          "Bx",           "Db",           "Dc",
  56         "Do",           "Dq",           "Ec",           "Ef",
  57         "Em",           "Eo",           "Fx",           "Ms",
  58         "No",           "Ns",           "Nx",           "Ox",
  59         "Pc",           "Pf",           "Po",           "Pq",
  60         "Qc",           "Ql",           "Qo",           "Qq",
  61         "Re",           "Rs",           "Sc",           "So",
  62         "Sq",           "Sm",           "Sx",           "Sy",
  63         "Tn",           "Ux",           "Xc",           "Xo",
  64         "Fo",           "Fc",           "Oo",           "Oc",
  65         "Bk",           "Ek",           "Bt",           "Hf",
  66         "Fr",           "Ud",           "Lb",           "Lp",
  67         "Lk",           "Mt",           "Brq",          "Bro",
  68         /* LINTED */
  69         "Brc",          "\%C",          "Es",           "En",
  70         /* LINTED */
  71         "Dx",           "\%Q"
  72         };
  73
  74 const   char *const __mdoc_argnames[MDOC_ARG_MAX] = {
  75         "split",                "nosplit",              "ragged",
  76         "unfilled",             "literal",              "file",
  77         "offset",               "bullet",               "dash",
  78         "hyphen",               "item",                 "enum",
  79         "tag",                  "diag",                 "hang",
  80         "ohang",                "inset",                "column",
  81         "width",                "compact",              "std",
  82         "filled",               "words",                "emphasis",
  83         "symbolic",             "nested"
  84         };
  85
  86 const   char * const *mdoc_macronames = __mdoc_macronames;
  87 const   char * const *mdoc_argnames = __mdoc_argnames;
  88
  89 static  void              mdoc_free1(struct mdoc *);
  90 static  int               mdoc_alloc1(struct mdoc *);
  91 static  struct mdoc_node *node_alloc(struct mdoc *, int, int,
  92                                 int, enum mdoc_type);
  93 static  int               node_append(struct mdoc *,
  94                                 struct mdoc_node *);
  95 static  int               parsetext(struct mdoc *, int, char *);
  96 static  int               parsemacro(struct mdoc *, int, char *);
  97 static  int               macrowarn(struct mdoc *, int, const char *);
  98 static  int               perr(struct mdoc *, int, int, enum merr);
  99
 100 const struct mdoc_node *
 101 mdoc_node(const struct mdoc *m)
 102 {
 103
 104         return(MDOC_HALT & m->flags ? NULL : m->first);
 105 }
 106
 107
 108 const struct mdoc_meta *
 109 mdoc_meta(const struct mdoc *m)
 110 {
 111
 112         return(MDOC_HALT & m->flags ? NULL : &m->meta);
 113 }
 114
 115
 116 static void
 117 mdoc_free1(struct mdoc *mdoc)
 118 {
 119
 120         if (mdoc->first)
 121                 mdoc_node_freelist(mdoc->first);
 122         if (mdoc->meta.title)
 123                 free(mdoc->meta.title);
 124         if (mdoc->meta.os)
 125                 free(mdoc->meta.os);
 126         if (mdoc->meta.name)
 127                 free(mdoc->meta.name);
 128         if (mdoc->meta.arch)
 129                 free(mdoc->meta.arch);
 130         if (mdoc->meta.vol)
 131                 free(mdoc->meta.vol);
 132 }
 133
 134
 135 static int
 136 mdoc_alloc1(struct mdoc *mdoc)
 137 {
 138
 139         bzero(&mdoc->meta, sizeof(struct mdoc_meta));
 140         mdoc->flags = 0;
 141         mdoc->lastnamed = mdoc->lastsec = 0;
 142         mdoc->last = calloc(1, sizeof(struct mdoc_node));
 143         if (NULL == mdoc->last)
 144                 return(0);
 145
 146         mdoc->first = mdoc->last;
 147         mdoc->last->type = MDOC_ROOT;
 148         mdoc->next = MDOC_NEXT_CHILD;
 149         return(1);
 150 }
 151
 152
 153 /*
 154  * Free up all resources contributed by a parse:  the node tree,
 155  * meta-data and so on.  Then reallocate the root node for another
 156  * parse.
 157  */
 158 int
 159 mdoc_reset(struct mdoc *mdoc)
 160 {
 161
 162         mdoc_free1(mdoc);
 163         return(mdoc_alloc1(mdoc));
 164 }
 165
 166
 167 /*
 168  * Completely free up all resources.
 169  */
 170 void
 171 mdoc_free(struct mdoc *mdoc)
 172 {
 173
 174         mdoc_free1(mdoc);
 175         if (mdoc->htab)
 176                 mdoc_hash_free(mdoc->htab);
 177         free(mdoc);
 178 }
 179
 180
 181 struct mdoc *
 182 mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb)
 183 {
 184         struct mdoc     *p;
 185
 186         if (NULL == (p = calloc(1, sizeof(struct mdoc))))
 187                 return(NULL);
 188         if (cb)
 189                 (void)memcpy(&p->cb, cb, sizeof(struct mdoc_cb));
 190
 191         p->data = data;
 192         p->pflags = pflags;
 193
 194         if (NULL == (p->htab = mdoc_hash_alloc())) {
 195                 free(p);
 196                 return(NULL);
 197         } else if (mdoc_alloc1(p))
 198                 return(p);
 199
 200         free(p);
 201         return(NULL);
 202 }
 203
 204
 205 /*
 206  * Climb back up the parse tree, validating open scopes.  Mostly calls
 207  * through to macro_end in macro.c.
 208  */
 209 int
 210 mdoc_endparse(struct mdoc *m)
 211 {
 212
 213         if (MDOC_HALT & m->flags)
 214                 return(0);
 215         else if (mdoc_macroend(m))
 216                 return(1);
 217         m->flags |= MDOC_HALT;
 218         return(0);
 219 }
 220
 221
 222 /*
 223  * Main parse routine.  Parses a single line -- really just hands off to
 224  * the macro or text parser.
 225  */
 226 int
 227 mdoc_parseln(struct mdoc *m, int ln, char *buf)
 228 {
 229
 230         if (MDOC_HALT & m->flags)
 231                 return(0);
 232
 233         return('.' == *buf ? parsemacro(m, ln, buf) :
 234                         parsetext(m, ln, buf));
 235 }
 236
 237
 238 int
 239 mdoc_verr(struct mdoc *mdoc, int ln, int pos,
 240                 const char *fmt, ...)
 241 {
 242         char             buf[256];
 243         va_list          ap;
 244
 245         if (NULL == mdoc->cb.mdoc_err)
 246                 return(0);
 247
 248         va_start(ap, fmt);
 249         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 250         va_end(ap);
 251         return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf));
 252 }
 253
 254
 255 int
 256 mdoc_vwarn(struct mdoc *mdoc, int ln, int pos,
 257                 enum mdoc_warn type, const char *fmt, ...)
 258 {
 259         char             buf[256];
 260         va_list          ap;
 261
 262         if (NULL == mdoc->cb.mdoc_warn)
 263                 return(0);
 264
 265         va_start(ap, fmt);
 266         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 267         va_end(ap);
 268         return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, type, buf));
 269 }
 270
 271
 272 int
 273 mdoc_nerr(struct mdoc *mdoc, const struct mdoc_node *node,
 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,
 286                                 node->line, node->pos, buf));
 287 }
 288
 289
 290 int
 291 mdoc_warn(struct mdoc *mdoc, enum mdoc_warn type,
 292                 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, mdoc->last->line,
 304                                 mdoc->last->pos, type, buf));
 305 }
 306
 307
 308 int
 309 mdoc_err(struct mdoc *mdoc, const char *fmt, ...)
 310 {
 311         char             buf[256];
 312         va_list          ap;
 313
 314         if (NULL == mdoc->cb.mdoc_err)
 315                 return(0);
 316
 317         va_start(ap, fmt);
 318         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 319         va_end(ap);
 320         return((*mdoc->cb.mdoc_err)(mdoc->data, mdoc->last->line,
 321                                 mdoc->last->pos, buf));
 322 }
 323
 324
 325 int
 326 mdoc_pwarn(struct mdoc *mdoc, int line, int pos, enum mdoc_warn type,
 327                 const char *fmt, ...)
 328 {
 329         char             buf[256];
 330         va_list          ap;
 331
 332         if (NULL == mdoc->cb.mdoc_warn)
 333                 return(0);
 334
 335         va_start(ap, fmt);
 336         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 337         va_end(ap);
 338         return((*mdoc->cb.mdoc_warn)(mdoc->data,
 339                                 line, pos, type, buf));
 340 }
 341
 342 int
 343 mdoc_perr(struct mdoc *mdoc, int line, int pos, const char *fmt, ...)
 344 {
 345         char             buf[256];
 346         va_list          ap;
 347
 348         if (NULL == mdoc->cb.mdoc_err)
 349                 return(0);
 350
 351         va_start(ap, fmt);
 352         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 353         va_end(ap);
 354         return((*mdoc->cb.mdoc_err)(mdoc->data, line, pos, buf));
 355 }
 356
 357
 358 int
 359 mdoc_macro(struct mdoc *m, int tok,
 360                 int ln, int pp, int *pos, char *buf)
 361 {
 362
 363         /* FIXME - these should happen during validation. */
 364
 365         if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
 366                         SEC_PROLOGUE != m->lastnamed)
 367                 return(perr(m, ln, pp, EPROLBODY));
 368
 369         if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
 370                         SEC_PROLOGUE == m->lastnamed)
 371                 return(perr(m, ln, pp, EBODYPROL));
 372
 373         if (1 != pp && ! (MDOC_CALLABLE & mdoc_macros[tok].flags))
 374                 return(perr(m, ln, pp, ENOCALL));
 375
 376         return((*mdoc_macros[tok].fp)(m, tok, ln, pp, pos, buf));
 377 }
 378
 379
 380 static int
 381 perr(struct mdoc *m, int line, int pos, enum merr type)
 382 {
 383         char            *p;
 384
 385         p = NULL;
 386         switch (type) {
 387         case (ENOCALL):
 388                 p = "not callable";
 389                 break;
 390         case (EPROLBODY):
 391                 p = "macro disallowed in document body";
 392                 break;
 393         case (EBODYPROL):
 394                 p = "macro disallowed in document prologue";
 395                 break;
 396         case (EMALLOC):
 397                 p = "memory exhausted";
 398                 break;
 399         case (ETEXTPROL):
 400                 p = "text disallowed in document prologue";
 401                 break;
 402         case (ENOBLANK):
 403                 p = "blank lines disallowed in non-literal contexts";
 404                 break;
 405         case (ESPACE):
 406                 p = "whitespace disallowed after delimiter";
 407                 break;
 408         }
 409         assert(p);
 410         return(mdoc_perr(m, line, pos, p));
 411 }
 412
 413
 414 static int
 415 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 416 {
 417
 418         assert(mdoc->last);
 419         assert(mdoc->first);
 420         assert(MDOC_ROOT != p->type);
 421
 422         switch (mdoc->next) {
 423         case (MDOC_NEXT_SIBLING):
 424                 mdoc->last->next = p;
 425                 p->prev = mdoc->last;
 426                 p->parent = mdoc->last->parent;
 427                 break;
 428         case (MDOC_NEXT_CHILD):
 429                 mdoc->last->child = p;
 430                 p->parent = mdoc->last;
 431                 break;
 432         default:
 433                 abort();
 434                 /* NOTREACHED */
 435         }
 436
 437         if ( ! mdoc_valid_pre(mdoc, p))
 438                 return(0);
 439         if ( ! mdoc_action_pre(mdoc, p))
 440                 return(0);
 441
 442         switch (p->type) {
 443         case (MDOC_HEAD):
 444                 assert(MDOC_BLOCK == p->parent->type);
 445                 p->parent->head = p;
 446                 break;
 447         case (MDOC_TAIL):
 448                 assert(MDOC_BLOCK == p->parent->type);
 449                 p->parent->tail = p;
 450                 break;
 451         case (MDOC_BODY):
 452                 assert(MDOC_BLOCK == p->parent->type);
 453                 p->parent->body = p;
 454                 break;
 455         default:
 456                 break;
 457         }
 458
 459         mdoc->last = p;
 460
 461         switch (p->type) {
 462         case (MDOC_TEXT):
 463                 if ( ! mdoc_valid_post(mdoc))
 464                         return(0);
 465                 if ( ! mdoc_action_post(mdoc))
 466                         return(0);
 467                 break;
 468         default:
 469                 break;
 470         }
 471
 472         return(1);
 473 }
 474
 475
 476 static struct mdoc_node *
 477 node_alloc(struct mdoc *mdoc, int line,
 478                 int pos, int tok, enum mdoc_type type)
 479 {
 480         struct mdoc_node *p;
 481
 482         if (NULL == (p = calloc(1, sizeof(struct mdoc_node)))) {
 483                 (void)perr(mdoc, (mdoc)->last->line,
 484                                 (mdoc)->last->pos, EMALLOC);
 485                 return(NULL);
 486         }
 487
 488         p->sec = mdoc->lastsec;
 489         p->line = line;
 490         p->pos = pos;
 491         p->tok = tok;
 492         if (MDOC_TEXT != (p->type = type))
 493                 assert(p->tok >= 0);
 494
 495         return(p);
 496 }
 497
 498
 499 int
 500 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 501 {
 502         struct mdoc_node *p;
 503
 504         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 505         if (NULL == p)
 506                 return(0);
 507         return(node_append(mdoc, p));
 508 }
 509
 510
 511 int
 512 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 513 {
 514         struct mdoc_node *p;
 515
 516         assert(mdoc->first);
 517         assert(mdoc->last);
 518
 519         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 520         if (NULL == p)
 521                 return(0);
 522         return(node_append(mdoc, p));
 523 }
 524
 525
 526 int
 527 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 528 {
 529         struct mdoc_node *p;
 530
 531         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 532         if (NULL == p)
 533                 return(0);
 534         return(node_append(mdoc, p));
 535 }
 536
 537
 538 int
 539 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 540                 int tok, struct mdoc_arg *args)
 541 {
 542         struct mdoc_node *p;
 543
 544         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 545         if (NULL == p)
 546                 return(0);
 547         p->args = args;
 548         if (p->args)
 549                 (args->refcnt)++;
 550         return(node_append(mdoc, p));
 551 }
 552
 553
 554 int
 555 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 556                 int tok, struct mdoc_arg *args)
 557 {
 558         struct mdoc_node *p;
 559
 560         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 561         if (NULL == p)
 562                 return(0);
 563         p->args = args;
 564         if (p->args)
 565                 (args->refcnt)++;
 566         return(node_append(mdoc, p));
 567 }
 568
 569
 570 int
 571 mdoc_word_alloc(struct mdoc *mdoc,
 572                 int line, int pos, const char *word)
 573 {
 574         struct mdoc_node *p;
 575
 576         p = node_alloc(mdoc, line, pos, -1, MDOC_TEXT);
 577         if (NULL == p)
 578                 return(0);
 579         if (NULL == (p->string = strdup(word))) {
 580                 (void)perr(mdoc, (mdoc)->last->line,
 581                                 (mdoc)->last->pos, EMALLOC);
 582                 return(0);
 583         }
 584         return(node_append(mdoc, p));
 585 }
 586
 587
 588 void
 589 mdoc_node_free(struct mdoc_node *p)
 590 {
 591
 592         if (p->string)
 593                 free(p->string);
 594         if (p->args)
 595                 mdoc_argv_free(p->args);
 596         free(p);
 597 }
 598
 599
 600 void
 601 mdoc_node_freelist(struct mdoc_node *p)
 602 {
 603
 604         if (p->child)
 605                 mdoc_node_freelist(p->child);
 606         if (p->next)
 607                 mdoc_node_freelist(p->next);
 608
 609         mdoc_node_free(p);
 610 }
 611
 612
 613 /*
 614  * Parse free-form text, that is, a line that does not begin with the
 615  * control character.
 616  */
 617 static int
 618 parsetext(struct mdoc *m, int line, char *buf)
 619 {
 620
 621         if (SEC_PROLOGUE == m->lastnamed)
 622                 return(perr(m, line, 0, ETEXTPROL));
 623
 624         if (0 == buf[0] && ! (MDOC_LITERAL & m->flags))
 625                 return(perr(m, line, 0, ENOBLANK));
 626
 627         if ( ! mdoc_word_alloc(m, line, 0, buf))
 628                 return(0);
 629
 630         m->next = MDOC_NEXT_SIBLING;
 631         return(1);
 632 }
 633
 634
 635 static int
 636 macrowarn(struct mdoc *m, int ln, const char *buf)
 637 {
 638         if ( ! (MDOC_IGN_MACRO & m->pflags))
 639                 return(mdoc_perr(m, ln, 1,
 640                                 "unknown macro: %s%s",
 641                                 buf, strlen(buf) > 3 ? "..." : ""));
 642         return(mdoc_pwarn(m, ln, 1, WARN_SYNTAX,
 643                                 "unknown macro: %s%s",
 644                                 buf, strlen(buf) > 3 ? "..." : ""));
 645 }
 646
 647
 648 /*
 649  * Parse a macro line, that is, a line beginning with the control
 650  * character.
 651  */
 652 int
 653 parsemacro(struct mdoc *m, int ln, char *buf)
 654 {
 655         int               i, c;
 656         char              mac[5];
 657
 658         /* Empty lines are ignored. */
 659
 660         if (0 == buf[1])
 661                 return(1);
 662
 663         if (' ' == buf[1]) {
 664                 i = 2;
 665                 while (buf[i] && ' ' == buf[i])
 666                         i++;
 667                 if (0 == buf[i])
 668                         return(1);
 669                 return(perr(m, ln, 1, ESPACE));
 670         }
 671
 672         /* Copy the first word into a nil-terminated buffer. */
 673
 674         for (i = 1; i < 5; i++) {
 675                 if (0 == (mac[i - 1] = buf[i]))
 676                         break;
 677                 else if (' ' == buf[i])
 678                         break;
 679         }
 680
 681         mac[i - 1] = 0;
 682
 683         if (i == 5 || i <= 2) {
 684                 if ( ! macrowarn(m, ln, mac))
 685                         goto err;
 686                 return(1);
 687         }
 688
 689         if (MDOC_MAX == (c = mdoc_hash_find(m->htab, mac))) {
 690                 if ( ! macrowarn(m, ln, mac))
 691                         goto err;
 692                 return(1);
 693         }
 694
 695         /* The macro is sane.  Jump to the next word. */
 696
 697         while (buf[i] && ' ' == buf[i])
 698                 i++;
 699
 700         /* Begin recursive parse sequence. */
 701
 702         if ( ! mdoc_macro(m, c, ln, 1, &i, buf))
 703                 goto err;
 704
 705         return(1);
 706
 707 err:    /* Error out. */
 708
 709         m->flags |= MDOC_HALT;
 710         return(0);
 711 }