man.c

   1 /*      $Id: man.c,v 1.101 2011/02/09 09:18:15 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 #include <sys/types.h>
  22
  23 #include <assert.h>
  24 #include <stdarg.h>
  25 #include <stdlib.h>
  26 #include <stdio.h>
  27 #include <string.h>
  28
  29 #include "mandoc.h"
  30 #include "libman.h"
  31 #include "libmandoc.h"
  32
  33 const   char *const __man_macronames[MAN_MAX] = {
  34         "br",           "TH",           "SH",           "SS",
  35         "TP",           "LP",           "PP",           "P",
  36         "IP",           "HP",           "SM",           "SB",
  37         "BI",           "IB",           "BR",           "RB",
  38         "R",            "B",            "I",            "IR",
  39         "RI",           "na",           "sp",           "nf",
  40         "fi",           "RE",           "RS",           "DT",
  41         "UC",           "PD",           "AT",           "in",
  42         "ft"
  43         };
  44
  45 const   char * const *man_macronames = __man_macronames;
  46
  47 static  struct man_node *man_node_alloc(struct man *, int, int,
  48                                 enum man_type, enum mant);
  49 static  int              man_node_append(struct man *,
  50                                 struct man_node *);
  51 static  void             man_node_free(struct man_node *);
  52 static  void             man_node_unlink(struct man *,
  53                                 struct man_node *);
  54 static  int              man_ptext(struct man *, int, char *, int);
  55 static  int              man_pmacro(struct man *, int, char *, int);
  56 static  void             man_free1(struct man *);
  57 static  void             man_alloc1(struct man *);
  58 static  int              man_descope(struct man *, int, int);
  59
  60
  61 const struct man_node *
  62 man_node(const struct man *m)
  63 {
  64
  65         assert( ! (MAN_HALT & m->flags));
  66         return(m->first);
  67 }
  68
  69
  70 const struct man_meta *
  71 man_meta(const struct man *m)
  72 {
  73
  74         assert( ! (MAN_HALT & m->flags));
  75         return(&m->meta);
  76 }
  77
  78
  79 void
  80 man_reset(struct man *man)
  81 {
  82
  83         man_free1(man);
  84         man_alloc1(man);
  85 }
  86
  87
  88 void
  89 man_free(struct man *man)
  90 {
  91
  92         man_free1(man);
  93         free(man);
  94 }
  95
  96
  97 struct man *
  98 man_alloc(struct regset *regs, void *data, mandocmsg msg)
  99 {
 100         struct man      *p;
 101
 102         p = mandoc_calloc(1, sizeof(struct man));
 103
 104         man_hash_init();
 105         p->data = data;
 106         p->msg = msg;
 107         p->regs = regs;
 108
 109         man_alloc1(p);
 110         return(p);
 111 }
 112
 113
 114 int
 115 man_endparse(struct man *m)
 116 {
 117
 118         assert( ! (MAN_HALT & m->flags));
 119         if (man_macroend(m))
 120                 return(1);
 121         m->flags |= MAN_HALT;
 122         return(0);
 123 }
 124
 125
 126 int
 127 man_parseln(struct man *m, int ln, char *buf, int offs)
 128 {
 129
 130         m->flags |= MAN_NEWLINE;
 131
 132         assert( ! (MAN_HALT & m->flags));
 133         return(('.' == buf[offs] || '\'' == buf[offs]) ?
 134                         man_pmacro(m, ln, buf, offs) :
 135                         man_ptext(m, ln, buf, offs));
 136 }
 137
 138
 139 static void
 140 man_free1(struct man *man)
 141 {
 142
 143         if (man->first)
 144                 man_node_delete(man, man->first);
 145         if (man->meta.title)
 146                 free(man->meta.title);
 147         if (man->meta.source)
 148                 free(man->meta.source);
 149         if (man->meta.rawdate)
 150                 free(man->meta.rawdate);
 151         if (man->meta.vol)
 152                 free(man->meta.vol);
 153         if (man->meta.msec)
 154                 free(man->meta.msec);
 155 }
 156
 157
 158 static void
 159 man_alloc1(struct man *m)
 160 {
 161
 162         memset(&m->meta, 0, sizeof(struct man_meta));
 163         m->flags = 0;
 164         m->last = mandoc_calloc(1, sizeof(struct man_node));
 165         m->first = m->last;
 166         m->last->type = MAN_ROOT;
 167         m->last->tok = MAN_MAX;
 168         m->next = MAN_NEXT_CHILD;
 169 }
 170
 171
 172 static int
 173 man_node_append(struct man *man, struct man_node *p)
 174 {
 175
 176         assert(man->last);
 177         assert(man->first);
 178         assert(MAN_ROOT != p->type);
 179
 180         switch (man->next) {
 181         case (MAN_NEXT_SIBLING):
 182                 man->last->next = p;
 183                 p->prev = man->last;
 184                 p->parent = man->last->parent;
 185                 break;
 186         case (MAN_NEXT_CHILD):
 187                 man->last->child = p;
 188                 p->parent = man->last;
 189                 break;
 190         default:
 191                 abort();
 192                 /* NOTREACHED */
 193         }
 194
 195         assert(p->parent);
 196         p->parent->nchild++;
 197
 198         if ( ! man_valid_pre(man, p))
 199                 return(0);
 200
 201         switch (p->type) {
 202         case (MAN_HEAD):
 203                 assert(MAN_BLOCK == p->parent->type);
 204                 p->parent->head = p;
 205                 break;
 206         case (MAN_BODY):
 207                 assert(MAN_BLOCK == p->parent->type);
 208                 p->parent->body = p;
 209                 break;
 210         default:
 211                 break;
 212         }
 213
 214         man->last = p;
 215
 216         switch (p->type) {
 217         case (MAN_TBL):
 218                 /* FALLTHROUGH */
 219         case (MAN_TEXT):
 220                 if ( ! man_valid_post(man))
 221                         return(0);
 222                 break;
 223         default:
 224                 break;
 225         }
 226
 227         return(1);
 228 }
 229
 230
 231 static struct man_node *
 232 man_node_alloc(struct man *m, int line, int pos,
 233                 enum man_type type, enum mant tok)
 234 {
 235         struct man_node *p;
 236
 237         p = mandoc_calloc(1, sizeof(struct man_node));
 238         p->line = line;
 239         p->pos = pos;
 240         p->type = type;
 241         p->tok = tok;
 242
 243         if (MAN_NEWLINE & m->flags)
 244                 p->flags |= MAN_LINE;
 245         m->flags &= ~MAN_NEWLINE;
 246         return(p);
 247 }
 248
 249
 250 int
 251 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 252 {
 253         struct man_node *p;
 254
 255         p = man_node_alloc(m, line, pos, MAN_ELEM, tok);
 256         if ( ! man_node_append(m, p))
 257                 return(0);
 258         m->next = MAN_NEXT_CHILD;
 259         return(1);
 260 }
 261
 262
 263 int
 264 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 265 {
 266         struct man_node *p;
 267
 268         p = man_node_alloc(m, line, pos, MAN_HEAD, tok);
 269         if ( ! man_node_append(m, p))
 270                 return(0);
 271         m->next = MAN_NEXT_CHILD;
 272         return(1);
 273 }
 274
 275
 276 int
 277 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 278 {
 279         struct man_node *p;
 280
 281         p = man_node_alloc(m, line, pos, MAN_BODY, tok);
 282         if ( ! man_node_append(m, p))
 283                 return(0);
 284         m->next = MAN_NEXT_CHILD;
 285         return(1);
 286 }
 287
 288
 289 int
 290 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 291 {
 292         struct man_node *p;
 293
 294         p = man_node_alloc(m, line, pos, MAN_BLOCK, tok);
 295         if ( ! man_node_append(m, p))
 296                 return(0);
 297         m->next = MAN_NEXT_CHILD;
 298         return(1);
 299 }
 300
 301 int
 302 man_word_alloc(struct man *m, int line, int pos, const char *word)
 303 {
 304         struct man_node *n;
 305         size_t           sv, len;
 306
 307         len = strlen(word);
 308
 309         n = man_node_alloc(m, line, pos, MAN_TEXT, MAN_MAX);
 310         n->string = mandoc_malloc(len + 1);
 311         sv = strlcpy(n->string, word, len + 1);
 312
 313         /* Prohibit truncation. */
 314         assert(sv < len + 1);
 315
 316         if ( ! man_node_append(m, n))
 317                 return(0);
 318
 319         m->next = MAN_NEXT_SIBLING;
 320         return(1);
 321 }
 322
 323
 324 /*
 325  * Free all of the resources held by a node.  This does NOT unlink a
 326  * node from its context; for that, see man_node_unlink().
 327  */
 328 static void
 329 man_node_free(struct man_node *p)
 330 {
 331
 332         if (p->string)
 333                 free(p->string);
 334         free(p);
 335 }
 336
 337
 338 void
 339 man_node_delete(struct man *m, struct man_node *p)
 340 {
 341
 342         while (p->child)
 343                 man_node_delete(m, p->child);
 344
 345         man_node_unlink(m, p);
 346         man_node_free(p);
 347 }
 348
 349 int
 350 man_addeqn(struct man *m, const struct eqn *ep)
 351 {
 352         struct man_node *n;
 353
 354         assert( ! (MAN_HALT & m->flags));
 355
 356         n = man_node_alloc(m, ep->line, ep->pos, MAN_EQN, MAN_MAX);
 357         n->eqn = ep;
 358
 359         if ( ! man_node_append(m, n))
 360                 return(0);
 361
 362         m->next = MAN_NEXT_SIBLING;
 363         return(man_descope(m, ep->line, ep->pos));
 364 }
 365
 366 int
 367 man_addspan(struct man *m, const struct tbl_span *sp)
 368 {
 369         struct man_node *n;
 370
 371         assert( ! (MAN_HALT & m->flags));
 372
 373         n = man_node_alloc(m, sp->line, 0, MAN_TBL, MAN_MAX);
 374         n->span = sp;
 375
 376         if ( ! man_node_append(m, n))
 377                 return(0);
 378
 379         m->next = MAN_NEXT_SIBLING;
 380         return(man_descope(m, sp->line, 0));
 381 }
 382
 383 static int
 384 man_descope(struct man *m, int line, int offs)
 385 {
 386         /*
 387          * Co-ordinate what happens with having a next-line scope open:
 388          * first close out the element scope (if applicable), then close
 389          * out the block scope (also if applicable).
 390          */
 391
 392         if (MAN_ELINE & m->flags) {
 393                 m->flags &= ~MAN_ELINE;
 394                 if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 395                         return(0);
 396         }
 397
 398         if ( ! (MAN_BLINE & m->flags))
 399                 return(1);
 400         m->flags &= ~MAN_BLINE;
 401
 402         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 403                 return(0);
 404         return(man_body_alloc(m, line, offs, m->last->tok));
 405 }
 406
 407
 408 static int
 409 man_ptext(struct man *m, int line, char *buf, int offs)
 410 {
 411         int              i;
 412
 413         /* Ignore bogus comments. */
 414
 415         if ('\\' == buf[offs] &&
 416                         '.' == buf[offs + 1] &&
 417                         '"' == buf[offs + 2]) {
 418                 man_pmsg(m, line, offs, MANDOCERR_BADCOMMENT);
 419                 return(1);
 420         }
 421
 422         /* Literal free-form text whitespace is preserved. */
 423
 424         if (MAN_LITERAL & m->flags) {
 425                 if ( ! man_word_alloc(m, line, offs, buf + offs))
 426                         return(0);
 427                 return(man_descope(m, line, offs));
 428         }
 429
 430         /* Pump blank lines directly into the backend. */
 431
 432         for (i = offs; ' ' == buf[i]; i++)
 433                 /* Skip leading whitespace. */ ;
 434
 435         if ('\0' == buf[i]) {
 436                 /* Allocate a blank entry. */
 437                 if ( ! man_word_alloc(m, line, offs, ""))
 438                         return(0);
 439                 return(man_descope(m, line, offs));
 440         }
 441
 442         /*
 443          * Warn if the last un-escaped character is whitespace. Then
 444          * strip away the remaining spaces (tabs stay!).
 445          */
 446
 447         i = (int)strlen(buf);
 448         assert(i);
 449
 450         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 451                 if (i > 1 && '\\' != buf[i - 2])
 452                         man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE);
 453
 454                 for (--i; i && ' ' == buf[i]; i--)
 455                         /* Spin back to non-space. */ ;
 456
 457                 /* Jump ahead of escaped whitespace. */
 458                 i += '\\' == buf[i] ? 2 : 1;
 459
 460                 buf[i] = '\0';
 461         }
 462
 463         if ( ! man_word_alloc(m, line, offs, buf + offs))
 464                 return(0);
 465
 466         /*
 467          * End-of-sentence check.  If the last character is an unescaped
 468          * EOS character, then flag the node as being the end of a
 469          * sentence.  The front-end will know how to interpret this.
 470          */
 471
 472         assert(i);
 473         if (mandoc_eos(buf, (size_t)i, 0))
 474                 m->last->flags |= MAN_EOS;
 475
 476         return(man_descope(m, line, offs));
 477 }
 478
 479
 480 static int
 481 man_pmacro(struct man *m, int ln, char *buf, int offs)
 482 {
 483         int              i, j, ppos;
 484         enum mant        tok;
 485         char             mac[5];
 486         struct man_node *n;
 487
 488         /* Comments and empties are quickly ignored. */
 489
 490         offs++;
 491
 492         if ('\0' == buf[offs])
 493                 return(1);
 494
 495         i = offs;
 496
 497         /*
 498          * Skip whitespace between the control character and initial
 499          * text.  "Whitespace" is both spaces and tabs.
 500          */
 501
 502         if (' ' == buf[i] || '\t' == buf[i]) {
 503                 i++;
 504                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 505                         i++;
 506                 if ('\0' == buf[i])
 507                         goto out;
 508         }
 509
 510         ppos = i;
 511
 512         /*
 513          * Copy the first word into a nil-terminated buffer.
 514          * Stop copying when a tab, space, or eoln is encountered.
 515          */
 516
 517         j = 0;
 518         while (j < 4 && '\0' != buf[i] && ' ' != buf[i] && '\t' != buf[i])
 519                 mac[j++] = buf[i++];
 520         mac[j] = '\0';
 521
 522         tok = (j > 0 && j < 4) ? man_hash_find(mac) : MAN_MAX;
 523         if (MAN_MAX == tok) {
 524                 man_vmsg(m, MANDOCERR_MACRO, ln, ppos, "%s", buf + ppos - 1);
 525                 return(1);
 526         }
 527
 528         /* The macro is sane.  Jump to the next word. */
 529
 530         while (buf[i] && ' ' == buf[i])
 531                 i++;
 532
 533         /*
 534          * Trailing whitespace.  Note that tabs are allowed to be passed
 535          * into the parser as "text", so we only warn about spaces here.
 536          */
 537
 538         if ('\0' == buf[i] && ' ' == buf[i - 1])
 539                 man_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE);
 540
 541         /*
 542          * Remove prior ELINE macro, as it's being clobbered by a new
 543          * macro.  Note that NSCOPED macros do not close out ELINE
 544          * macros---they don't print text---so we let those slip by.
 545          */
 546
 547         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 548                         m->flags & MAN_ELINE) {
 549                 n = m->last;
 550                 assert(MAN_TEXT != n->type);
 551
 552                 /* Remove repeated NSCOPED macros causing ELINE. */
 553
 554                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 555                         n = n->parent;
 556
 557                 man_vmsg(m, MANDOCERR_LINESCOPE, n->line, n->pos,
 558                                 "%s", man_macronames[n->tok]);
 559
 560                 man_node_delete(m, n);
 561                 m->flags &= ~MAN_ELINE;
 562         }
 563
 564         /*
 565          * Save the fact that we're in the next-line for a block.  In
 566          * this way, embedded roff instructions can "remember" state
 567          * when they exit.
 568          */
 569
 570         if (MAN_BLINE & m->flags)
 571                 m->flags |= MAN_BPLINE;
 572
 573         /* Call to handler... */
 574
 575         assert(man_macros[tok].fp);
 576         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 577                 goto err;
 578
 579 out:
 580         /*
 581          * We weren't in a block-line scope when entering the
 582          * above-parsed macro, so return.
 583          */
 584
 585         if ( ! (MAN_BPLINE & m->flags)) {
 586                 m->flags &= ~MAN_ILINE;
 587                 return(1);
 588         }
 589         m->flags &= ~MAN_BPLINE;
 590
 591         /*
 592          * If we're in a block scope, then allow this macro to slip by
 593          * without closing scope around it.
 594          */
 595
 596         if (MAN_ILINE & m->flags) {
 597                 m->flags &= ~MAN_ILINE;
 598                 return(1);
 599         }
 600
 601         /*
 602          * If we've opened a new next-line element scope, then return
 603          * now, as the next line will close out the block scope.
 604          */
 605
 606         if (MAN_ELINE & m->flags)
 607                 return(1);
 608
 609         /* Close out the block scope opened in the prior line.  */
 610
 611         assert(MAN_BLINE & m->flags);
 612         m->flags &= ~MAN_BLINE;
 613
 614         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 615                 return(0);
 616         return(man_body_alloc(m, ln, offs, m->last->tok));
 617
 618 err:    /* Error out. */
 619
 620         m->flags |= MAN_HALT;
 621         return(0);
 622 }
 623
 624
 625 int
 626 man_vmsg(struct man *man, enum mandocerr t,
 627                 int ln, int pos, const char *fmt, ...)
 628 {
 629         char             buf[256];
 630         va_list          ap;
 631
 632         va_start(ap, fmt);
 633         vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 634         va_end(ap);
 635         return((*man->msg)(t, man->data, ln, pos, buf));
 636 }
 637
 638
 639 /*
 640  * Unlink a node from its context.  If "m" is provided, the last parse
 641  * point will also be adjusted accordingly.
 642  */
 643 static void
 644 man_node_unlink(struct man *m, struct man_node *n)
 645 {
 646
 647         /* Adjust siblings. */
 648
 649         if (n->prev)
 650                 n->prev->next = n->next;
 651         if (n->next)
 652                 n->next->prev = n->prev;
 653
 654         /* Adjust parent. */
 655
 656         if (n->parent) {
 657                 n->parent->nchild--;
 658                 if (n->parent->child == n)
 659                         n->parent->child = n->prev ? n->prev : n->next;
 660         }
 661
 662         /* Adjust parse point, if applicable. */
 663
 664         if (m && m->last == n) {
 665                 /*XXX: this can occur when bailing from validation. */
 666                 /*assert(NULL == n->next);*/
 667                 if (n->prev) {
 668                         m->last = n->prev;
 669                         m->next = MAN_NEXT_SIBLING;
 670                 } else {
 671                         m->last = n->parent;
 672                         m->next = MAN_NEXT_CHILD;
 673                 }
 674         }
 675
 676         if (m && m->first == n)
 677                 m->first = NULL;
 678 }