mdoc.c

   1 /*      $Id: mdoc.c,v 1.83 2009/06/16 20:22:23 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 #define verr(m, t) perr((m), (m)->last->line, (m)->last->pos, (t))
 101
 102 const struct mdoc_node *
 103 mdoc_node(const struct mdoc *m)
 104 {
 105
 106         return(MDOC_HALT & m->flags ? NULL : m->first);
 107 }
 108
 109
 110 const struct mdoc_meta *
 111 mdoc_meta(const struct mdoc *m)
 112 {
 113
 114         return(MDOC_HALT & m->flags ? NULL : &m->meta);
 115 }
 116
 117
 118 static void
 119 mdoc_free1(struct mdoc *mdoc)
 120 {
 121
 122         if (mdoc->first)
 123                 mdoc_node_freelist(mdoc->first);
 124         if (mdoc->meta.title)
 125                 free(mdoc->meta.title);
 126         if (mdoc->meta.os)
 127                 free(mdoc->meta.os);
 128         if (mdoc->meta.name)
 129                 free(mdoc->meta.name);
 130         if (mdoc->meta.arch)
 131                 free(mdoc->meta.arch);
 132         if (mdoc->meta.vol)
 133                 free(mdoc->meta.vol);
 134 }
 135
 136
 137 static int
 138 mdoc_alloc1(struct mdoc *mdoc)
 139 {
 140
 141         bzero(&mdoc->meta, sizeof(struct mdoc_meta));
 142         mdoc->flags = 0;
 143         mdoc->lastnamed = mdoc->lastsec = 0;
 144         mdoc->last = calloc(1, sizeof(struct mdoc_node));
 145         if (NULL == mdoc->last)
 146                 return(0);
 147
 148         mdoc->first = mdoc->last;
 149         mdoc->last->type = MDOC_ROOT;
 150         mdoc->next = MDOC_NEXT_CHILD;
 151         return(1);
 152 }
 153
 154
 155 /*
 156  * Free up all resources contributed by a parse:  the node tree,
 157  * meta-data and so on.  Then reallocate the root node for another
 158  * parse.
 159  */
 160 int
 161 mdoc_reset(struct mdoc *mdoc)
 162 {
 163
 164         mdoc_free1(mdoc);
 165         return(mdoc_alloc1(mdoc));
 166 }
 167
 168
 169 /*
 170  * Completely free up all resources.
 171  */
 172 void
 173 mdoc_free(struct mdoc *mdoc)
 174 {
 175
 176         mdoc_free1(mdoc);
 177         if (mdoc->htab)
 178                 mdoc_hash_free(mdoc->htab);
 179         free(mdoc);
 180 }
 181
 182
 183 struct mdoc *
 184 mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb)
 185 {
 186         struct mdoc     *p;
 187
 188         if (NULL == (p = calloc(1, sizeof(struct mdoc))))
 189                 return(NULL);
 190         if (cb)
 191                 (void)memcpy(&p->cb, cb, sizeof(struct mdoc_cb));
 192
 193         p->data = data;
 194         p->pflags = pflags;
 195
 196         if (NULL == (p->htab = mdoc_hash_alloc())) {
 197                 free(p);
 198                 return(NULL);
 199         } else if (mdoc_alloc1(p))
 200                 return(p);
 201
 202         free(p);
 203         return(NULL);
 204 }
 205
 206
 207 /*
 208  * Climb back up the parse tree, validating open scopes.  Mostly calls
 209  * through to macro_end in macro.c.
 210  */
 211 int
 212 mdoc_endparse(struct mdoc *m)
 213 {
 214
 215         if (MDOC_HALT & m->flags)
 216                 return(0);
 217         else if (mdoc_macroend(m))
 218                 return(1);
 219         m->flags |= MDOC_HALT;
 220         return(0);
 221 }
 222
 223
 224 /*
 225  * Main parse routine.  Parses a single line -- really just hands off to
 226  * the macro or text parser.
 227  */
 228 int
 229 mdoc_parseln(struct mdoc *m, int ln, char *buf)
 230 {
 231
 232         /* If in error-mode, then we parse no more. */
 233
 234         if (MDOC_HALT & m->flags)
 235                 return(0);
 236
 237         return('.' == *buf ? parsemacro(m, ln, buf) :
 238                         parsetext(m, ln, buf));
 239 }
 240
 241
 242 int
 243 mdoc_verr(struct mdoc *mdoc, int ln, int pos,
 244                 const char *fmt, ...)
 245 {
 246         char             buf[256];
 247         va_list          ap;
 248
 249         if (NULL == mdoc->cb.mdoc_err)
 250                 return(0);
 251
 252         va_start(ap, fmt);
 253         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 254         va_end(ap);
 255         return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf));
 256 }
 257
 258
 259 int
 260 mdoc_vwarn(struct mdoc *mdoc, int ln, int pos,
 261                 enum mdoc_warn type, const char *fmt, ...)
 262 {
 263         char             buf[256];
 264         va_list          ap;
 265
 266         if (NULL == mdoc->cb.mdoc_warn)
 267                 return(0);
 268
 269         va_start(ap, fmt);
 270         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 271         va_end(ap);
 272         return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, type, buf));
 273 }
 274
 275
 276 int
 277 mdoc_nerr(struct mdoc *mdoc, const struct mdoc_node *node, const char *fmt, ...)
 278 {
 279         char             buf[256];
 280         va_list          ap;
 281
 282         if (NULL == mdoc->cb.mdoc_err)
 283                 return(0);
 284
 285         va_start(ap, fmt);
 286         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 287         va_end(ap);
 288         return((*mdoc->cb.mdoc_err)(mdoc->data, node->line, node->pos, buf));
 289 }
 290
 291
 292 int
 293 mdoc_warn(struct mdoc *mdoc, enum mdoc_warn type, const char *fmt, ...)
 294 {
 295         char             buf[256];
 296         va_list          ap;
 297
 298         if (NULL == mdoc->cb.mdoc_warn)
 299                 return(0);
 300
 301         va_start(ap, fmt);
 302         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 303         va_end(ap);
 304         return((*mdoc->cb.mdoc_warn)(mdoc->data, mdoc->last->line,
 305             mdoc->last->pos, type, buf));
 306 }
 307
 308
 309 int
 310 mdoc_err(struct mdoc *mdoc, const char *fmt, ...)
 311 {
 312         char             buf[256];
 313         va_list          ap;
 314
 315         if (NULL == mdoc->cb.mdoc_err)
 316                 return(0);
 317
 318         va_start(ap, fmt);
 319         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 320         va_end(ap);
 321         return((*mdoc->cb.mdoc_err)(mdoc->data, mdoc->last->line,
 322             mdoc->last->pos, buf));
 323 }
 324
 325
 326 int
 327 mdoc_pwarn(struct mdoc *mdoc, int line, int pos, enum mdoc_warn type,
 328                 const char *fmt, ...)
 329 {
 330         char             buf[256];
 331         va_list          ap;
 332
 333         if (NULL == mdoc->cb.mdoc_warn)
 334                 return(0);
 335
 336         va_start(ap, fmt);
 337         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 338         va_end(ap);
 339         return((*mdoc->cb.mdoc_warn)(mdoc->data, 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)verr(mdoc, EMALLOC);
 484                 return(NULL);
 485         }
 486
 487         p->sec = mdoc->lastsec;
 488         p->line = line;
 489         p->pos = pos;
 490         p->tok = tok;
 491         if (MDOC_TEXT != (p->type = type))
 492                 assert(p->tok >= 0);
 493
 494         return(p);
 495 }
 496
 497
 498 int
 499 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 500 {
 501         struct mdoc_node *p;
 502
 503         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 504         if (NULL == p)
 505                 return(0);
 506         return(node_append(mdoc, p));
 507 }
 508
 509
 510 int
 511 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 512 {
 513         struct mdoc_node *p;
 514
 515         assert(mdoc->first);
 516         assert(mdoc->last);
 517
 518         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 519         if (NULL == p)
 520                 return(0);
 521         return(node_append(mdoc, p));
 522 }
 523
 524
 525 int
 526 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 527 {
 528         struct mdoc_node *p;
 529
 530         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 531         if (NULL == p)
 532                 return(0);
 533         return(node_append(mdoc, p));
 534 }
 535
 536
 537 int
 538 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 539                 int tok, struct mdoc_arg *args)
 540 {
 541         struct mdoc_node *p;
 542
 543         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 544         if (NULL == p)
 545                 return(0);
 546         p->args = args;
 547         if (p->args)
 548                 (args->refcnt)++;
 549         return(node_append(mdoc, p));
 550 }
 551
 552
 553 int
 554 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 555                 int tok, struct mdoc_arg *args)
 556 {
 557         struct mdoc_node *p;
 558
 559         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 560         if (NULL == p)
 561                 return(0);
 562         p->args = args;
 563         if (p->args)
 564                 (args->refcnt)++;
 565         return(node_append(mdoc, p));
 566 }
 567
 568
 569 int
 570 mdoc_word_alloc(struct mdoc *mdoc,
 571                 int line, int pos, const char *word)
 572 {
 573         struct mdoc_node *p;
 574
 575         p = node_alloc(mdoc, line, pos, -1, MDOC_TEXT);
 576         if (NULL == p)
 577                 return(0);
 578         if (NULL == (p->string = strdup(word))) {
 579                 (void)verr(mdoc, EMALLOC);
 580                 return(0);
 581         }
 582         return(node_append(mdoc, p));
 583 }
 584
 585
 586 void
 587 mdoc_node_free(struct mdoc_node *p)
 588 {
 589
 590         if (p->string)
 591                 free(p->string);
 592         if (p->args)
 593                 mdoc_argv_free(p->args);
 594         free(p);
 595 }
 596
 597
 598 void
 599 mdoc_node_freelist(struct mdoc_node *p)
 600 {
 601
 602         if (p->child)
 603                 mdoc_node_freelist(p->child);
 604         if (p->next)
 605                 mdoc_node_freelist(p->next);
 606
 607         mdoc_node_free(p);
 608 }
 609
 610
 611 /*
 612  * Parse free-form text, that is, a line that does not begin with the
 613  * control character.
 614  */
 615 static int
 616 parsetext(struct mdoc *m, int line, char *buf)
 617 {
 618
 619         if (SEC_PROLOGUE == m->lastnamed)
 620                 return(perr(m, line, 0, ETEXTPROL));
 621
 622         if (0 == buf[0] && ! (MDOC_LITERAL & m->flags))
 623                 return(perr(m, line, 0, ENOBLANK));
 624
 625         if ( ! mdoc_word_alloc(m, line, 0, buf))
 626                 return(0);
 627
 628         m->next = MDOC_NEXT_SIBLING;
 629         return(1);
 630 }
 631
 632
 633 static int
 634 macrowarn(struct mdoc *m, int ln, const char *buf)
 635 {
 636         if ( ! (MDOC_IGN_MACRO & m->pflags))
 637                 return(mdoc_perr(m, ln, 1,
 638                                 "unknown macro: %s%s",
 639                                 buf, strlen(buf) > 3 ? "..." : ""));
 640         return(mdoc_pwarn(m, ln, 1, WARN_SYNTAX,
 641                                 "unknown macro: %s%s",
 642                                 buf, strlen(buf) > 3 ? "..." : ""));
 643 }
 644
 645
 646 /*
 647  * Parse a macro line, that is, a line beginning with the control
 648  * character.
 649  */
 650 int
 651 parsemacro(struct mdoc *m, int ln, char *buf)
 652 {
 653         int               i, c;
 654         char              mac[5];
 655
 656         /* Empty lines are ignored. */
 657
 658         if (0 == buf[1])
 659                 return(1);
 660
 661         if (' ' == buf[1]) {
 662                 i = 2;
 663                 while (buf[i] && ' ' == buf[i])
 664                         i++;
 665                 if (0 == buf[i])
 666                         return(1);
 667                 return(perr(m, ln, 1, ESPACE));
 668         }
 669
 670         /* Copy the first word into a nil-terminated buffer. */
 671
 672         for (i = 1; i < 5; i++) {
 673                 if (0 == (mac[i - 1] = buf[i]))
 674                         break;
 675                 else if (' ' == buf[i])
 676                         break;
 677         }
 678
 679         mac[i - 1] = 0;
 680
 681         if (i == 5 || i <= 2) {
 682                 if ( ! macrowarn(m, ln, mac))
 683                         goto err;
 684                 return(1);
 685         }
 686
 687         if (MDOC_MAX == (c = mdoc_hash_find(m->htab, mac))) {
 688                 if ( ! macrowarn(m, ln, mac))
 689                         goto err;
 690                 return(1);
 691         }
 692
 693         /* The macro is sane.  Jump to the next word. */
 694
 695         while (buf[i] && ' ' == buf[i])
 696                 i++;
 697
 698         /* Begin recursive parse sequence. */
 699
 700         if ( ! mdoc_macro(m, c, ln, 1, &i, buf))
 701                 goto err;
 702
 703         return(1);
 704
 705 err:    /* Error out. */
 706
 707         m->flags |= MDOC_HALT;
 708         return(0);
 709 }