man.c

   1 /*      $Id: man.c,v 1.99 2011/02/06 22:05:20 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  int              man_span_alloc(struct man *,
  52                                 const struct tbl_span *);
  53 static  void             man_node_free(struct man_node *);
  54 static  void             man_node_unlink(struct man *,
  55                                 struct man_node *);
  56 static  int              man_ptext(struct man *, int, char *, int);
  57 static  int              man_pmacro(struct man *, int, char *, int);
  58 static  void             man_free1(struct man *);
  59 static  void             man_alloc1(struct man *);
  60 static  int              man_descope(struct man *, int, int);
  61
  62
  63 const struct man_node *
  64 man_node(const struct man *m)
  65 {
  66
  67         assert( ! (MAN_HALT & m->flags));
  68         return(m->first);
  69 }
  70
  71
  72 const struct man_meta *
  73 man_meta(const struct man *m)
  74 {
  75
  76         assert( ! (MAN_HALT & m->flags));
  77         return(&m->meta);
  78 }
  79
  80
  81 void
  82 man_reset(struct man *man)
  83 {
  84
  85         man_free1(man);
  86         man_alloc1(man);
  87 }
  88
  89
  90 void
  91 man_free(struct man *man)
  92 {
  93
  94         man_free1(man);
  95         free(man);
  96 }
  97
  98
  99 struct man *
 100 man_alloc(struct regset *regs, void *data, mandocmsg msg)
 101 {
 102         struct man      *p;
 103
 104         p = mandoc_calloc(1, sizeof(struct man));
 105
 106         man_hash_init();
 107         p->data = data;
 108         p->msg = msg;
 109         p->regs = regs;
 110
 111         man_alloc1(p);
 112         return(p);
 113 }
 114
 115
 116 int
 117 man_endparse(struct man *m)
 118 {
 119
 120         assert( ! (MAN_HALT & m->flags));
 121         if (man_macroend(m))
 122                 return(1);
 123         m->flags |= MAN_HALT;
 124         return(0);
 125 }
 126
 127
 128 int
 129 man_parseln(struct man *m, int ln, char *buf, int offs)
 130 {
 131
 132         m->flags |= MAN_NEWLINE;
 133
 134         assert( ! (MAN_HALT & m->flags));
 135         return(('.' == buf[offs] || '\'' == buf[offs]) ?
 136                         man_pmacro(m, ln, buf, offs) :
 137                         man_ptext(m, ln, buf, offs));
 138 }
 139
 140
 141 static void
 142 man_free1(struct man *man)
 143 {
 144
 145         if (man->first)
 146                 man_node_delete(man, man->first);
 147         if (man->meta.title)
 148                 free(man->meta.title);
 149         if (man->meta.source)
 150                 free(man->meta.source);
 151         if (man->meta.rawdate)
 152                 free(man->meta.rawdate);
 153         if (man->meta.vol)
 154                 free(man->meta.vol);
 155         if (man->meta.msec)
 156                 free(man->meta.msec);
 157 }
 158
 159
 160 static void
 161 man_alloc1(struct man *m)
 162 {
 163
 164         memset(&m->meta, 0, sizeof(struct man_meta));
 165         m->flags = 0;
 166         m->last = mandoc_calloc(1, sizeof(struct man_node));
 167         m->first = m->last;
 168         m->last->type = MAN_ROOT;
 169         m->last->tok = MAN_MAX;
 170         m->next = MAN_NEXT_CHILD;
 171 }
 172
 173
 174 static int
 175 man_node_append(struct man *man, struct man_node *p)
 176 {
 177
 178         assert(man->last);
 179         assert(man->first);
 180         assert(MAN_ROOT != p->type);
 181
 182         switch (man->next) {
 183         case (MAN_NEXT_SIBLING):
 184                 man->last->next = p;
 185                 p->prev = man->last;
 186                 p->parent = man->last->parent;
 187                 break;
 188         case (MAN_NEXT_CHILD):
 189                 man->last->child = p;
 190                 p->parent = man->last;
 191                 break;
 192         default:
 193                 abort();
 194                 /* NOTREACHED */
 195         }
 196
 197         assert(p->parent);
 198         p->parent->nchild++;
 199
 200         if ( ! man_valid_pre(man, p))
 201                 return(0);
 202
 203         switch (p->type) {
 204         case (MAN_HEAD):
 205                 assert(MAN_BLOCK == p->parent->type);
 206                 p->parent->head = p;
 207                 break;
 208         case (MAN_BODY):
 209                 assert(MAN_BLOCK == p->parent->type);
 210                 p->parent->body = p;
 211                 break;
 212         default:
 213                 break;
 214         }
 215
 216         man->last = p;
 217
 218         switch (p->type) {
 219         case (MAN_TBL):
 220                 /* FALLTHROUGH */
 221         case (MAN_TEXT):
 222                 if ( ! man_valid_post(man))
 223                         return(0);
 224                 break;
 225         default:
 226                 break;
 227         }
 228
 229         return(1);
 230 }
 231
 232
 233 static struct man_node *
 234 man_node_alloc(struct man *m, int line, int pos,
 235                 enum man_type type, enum mant tok)
 236 {
 237         struct man_node *p;
 238
 239         p = mandoc_calloc(1, sizeof(struct man_node));
 240         p->line = line;
 241         p->pos = pos;
 242         p->type = type;
 243         p->tok = tok;
 244
 245         if (MAN_NEWLINE & m->flags)
 246                 p->flags |= MAN_LINE;
 247         m->flags &= ~MAN_NEWLINE;
 248         return(p);
 249 }
 250
 251
 252 int
 253 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 254 {
 255         struct man_node *p;
 256
 257         p = man_node_alloc(m, line, pos, MAN_ELEM, tok);
 258         if ( ! man_node_append(m, p))
 259                 return(0);
 260         m->next = MAN_NEXT_CHILD;
 261         return(1);
 262 }
 263
 264
 265 int
 266 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 267 {
 268         struct man_node *p;
 269
 270         p = man_node_alloc(m, line, pos, MAN_HEAD, tok);
 271         if ( ! man_node_append(m, p))
 272                 return(0);
 273         m->next = MAN_NEXT_CHILD;
 274         return(1);
 275 }
 276
 277
 278 int
 279 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 280 {
 281         struct man_node *p;
 282
 283         p = man_node_alloc(m, line, pos, MAN_BODY, tok);
 284         if ( ! man_node_append(m, p))
 285                 return(0);
 286         m->next = MAN_NEXT_CHILD;
 287         return(1);
 288 }
 289
 290
 291 int
 292 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 293 {
 294         struct man_node *p;
 295
 296         p = man_node_alloc(m, line, pos, MAN_BLOCK, tok);
 297         if ( ! man_node_append(m, p))
 298                 return(0);
 299         m->next = MAN_NEXT_CHILD;
 300         return(1);
 301 }
 302
 303 static int
 304 man_span_alloc(struct man *m, const struct tbl_span *span)
 305 {
 306         struct man_node *n;
 307
 308         n = man_node_alloc(m, span->line, 0, MAN_TBL, MAN_MAX);
 309         n->span = span;
 310
 311         if ( ! man_node_append(m, n))
 312                 return(0);
 313
 314         m->next = MAN_NEXT_SIBLING;
 315         return(1);
 316 }
 317
 318 int
 319 man_word_alloc(struct man *m, int line, int pos, const char *word)
 320 {
 321         struct man_node *n;
 322         size_t           sv, len;
 323
 324         len = strlen(word);
 325
 326         n = man_node_alloc(m, line, pos, MAN_TEXT, MAN_MAX);
 327         n->string = mandoc_malloc(len + 1);
 328         sv = strlcpy(n->string, word, len + 1);
 329
 330         /* Prohibit truncation. */
 331         assert(sv < len + 1);
 332
 333         if ( ! man_node_append(m, n))
 334                 return(0);
 335
 336         m->next = MAN_NEXT_SIBLING;
 337         return(1);
 338 }
 339
 340
 341 /*
 342  * Free all of the resources held by a node.  This does NOT unlink a
 343  * node from its context; for that, see man_node_unlink().
 344  */
 345 static void
 346 man_node_free(struct man_node *p)
 347 {
 348
 349         if (p->string)
 350                 free(p->string);
 351         free(p);
 352 }
 353
 354
 355 void
 356 man_node_delete(struct man *m, struct man_node *p)
 357 {
 358
 359         while (p->child)
 360                 man_node_delete(m, p->child);
 361
 362         man_node_unlink(m, p);
 363         man_node_free(p);
 364 }
 365
 366
 367 int
 368 man_addspan(struct man *m, const struct tbl_span *sp)
 369 {
 370
 371         assert( ! (MAN_HALT & m->flags));
 372         if ( ! man_span_alloc(m, sp))
 373                 return(0);
 374         return(man_descope(m, sp->line, 0));
 375 }
 376
 377 static int
 378 man_descope(struct man *m, int line, int offs)
 379 {
 380         /*
 381          * Co-ordinate what happens with having a next-line scope open:
 382          * first close out the element scope (if applicable), then close
 383          * out the block scope (also if applicable).
 384          */
 385
 386         if (MAN_ELINE & m->flags) {
 387                 m->flags &= ~MAN_ELINE;
 388                 if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 389                         return(0);
 390         }
 391
 392         if ( ! (MAN_BLINE & m->flags))
 393                 return(1);
 394         m->flags &= ~MAN_BLINE;
 395
 396         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 397                 return(0);
 398         return(man_body_alloc(m, line, offs, m->last->tok));
 399 }
 400
 401
 402 static int
 403 man_ptext(struct man *m, int line, char *buf, int offs)
 404 {
 405         int              i;
 406
 407         /* Ignore bogus comments. */
 408
 409         if ('\\' == buf[offs] &&
 410                         '.' == buf[offs + 1] &&
 411                         '"' == buf[offs + 2]) {
 412                 man_pmsg(m, line, offs, MANDOCERR_BADCOMMENT);
 413                 return(1);
 414         }
 415
 416         /* Literal free-form text whitespace is preserved. */
 417
 418         if (MAN_LITERAL & m->flags) {
 419                 if ( ! man_word_alloc(m, line, offs, buf + offs))
 420                         return(0);
 421                 return(man_descope(m, line, offs));
 422         }
 423
 424         /* Pump blank lines directly into the backend. */
 425
 426         for (i = offs; ' ' == buf[i]; i++)
 427                 /* Skip leading whitespace. */ ;
 428
 429         if ('\0' == buf[i]) {
 430                 /* Allocate a blank entry. */
 431                 if ( ! man_word_alloc(m, line, offs, ""))
 432                         return(0);
 433                 return(man_descope(m, line, offs));
 434         }
 435
 436         /*
 437          * Warn if the last un-escaped character is whitespace. Then
 438          * strip away the remaining spaces (tabs stay!).
 439          */
 440
 441         i = (int)strlen(buf);
 442         assert(i);
 443
 444         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 445                 if (i > 1 && '\\' != buf[i - 2])
 446                         man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE);
 447
 448                 for (--i; i && ' ' == buf[i]; i--)
 449                         /* Spin back to non-space. */ ;
 450
 451                 /* Jump ahead of escaped whitespace. */
 452                 i += '\\' == buf[i] ? 2 : 1;
 453
 454                 buf[i] = '\0';
 455         }
 456
 457         if ( ! man_word_alloc(m, line, offs, buf + offs))
 458                 return(0);
 459
 460         /*
 461          * End-of-sentence check.  If the last character is an unescaped
 462          * EOS character, then flag the node as being the end of a
 463          * sentence.  The front-end will know how to interpret this.
 464          */
 465
 466         assert(i);
 467         if (mandoc_eos(buf, (size_t)i, 0))
 468                 m->last->flags |= MAN_EOS;
 469
 470         return(man_descope(m, line, offs));
 471 }
 472
 473
 474 static int
 475 man_pmacro(struct man *m, int ln, char *buf, int offs)
 476 {
 477         int              i, j, ppos;
 478         enum mant        tok;
 479         char             mac[5];
 480         struct man_node *n;
 481
 482         /* Comments and empties are quickly ignored. */
 483
 484         offs++;
 485
 486         if ('\0' == buf[offs])
 487                 return(1);
 488
 489         i = offs;
 490
 491         /*
 492          * Skip whitespace between the control character and initial
 493          * text.  "Whitespace" is both spaces and tabs.
 494          */
 495
 496         if (' ' == buf[i] || '\t' == buf[i]) {
 497                 i++;
 498                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 499                         i++;
 500                 if ('\0' == buf[i])
 501                         goto out;
 502         }
 503
 504         ppos = i;
 505
 506         /*
 507          * Copy the first word into a nil-terminated buffer.
 508          * Stop copying when a tab, space, or eoln is encountered.
 509          */
 510
 511         j = 0;
 512         while (j < 4 && '\0' != buf[i] && ' ' != buf[i] && '\t' != buf[i])
 513                 mac[j++] = buf[i++];
 514         mac[j] = '\0';
 515
 516         tok = (j > 0 && j < 4) ? man_hash_find(mac) : MAN_MAX;
 517         if (MAN_MAX == tok) {
 518                 man_vmsg(m, MANDOCERR_MACRO, ln, ppos, "%s", buf + ppos - 1);
 519                 return(1);
 520         }
 521
 522         /* The macro is sane.  Jump to the next word. */
 523
 524         while (buf[i] && ' ' == buf[i])
 525                 i++;
 526
 527         /*
 528          * Trailing whitespace.  Note that tabs are allowed to be passed
 529          * into the parser as "text", so we only warn about spaces here.
 530          */
 531
 532         if ('\0' == buf[i] && ' ' == buf[i - 1])
 533                 man_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE);
 534
 535         /*
 536          * Remove prior ELINE macro, as it's being clobbered by a new
 537          * macro.  Note that NSCOPED macros do not close out ELINE
 538          * macros---they don't print text---so we let those slip by.
 539          */
 540
 541         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 542                         m->flags & MAN_ELINE) {
 543                 n = m->last;
 544                 assert(MAN_TEXT != n->type);
 545
 546                 /* Remove repeated NSCOPED macros causing ELINE. */
 547
 548                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 549                         n = n->parent;
 550
 551                 man_vmsg(m, MANDOCERR_LINESCOPE, n->line, n->pos,
 552                                 "%s", man_macronames[n->tok]);
 553
 554                 man_node_delete(m, n);
 555                 m->flags &= ~MAN_ELINE;
 556         }
 557
 558         /*
 559          * Save the fact that we're in the next-line for a block.  In
 560          * this way, embedded roff instructions can "remember" state
 561          * when they exit.
 562          */
 563
 564         if (MAN_BLINE & m->flags)
 565                 m->flags |= MAN_BPLINE;
 566
 567         /* Call to handler... */
 568
 569         assert(man_macros[tok].fp);
 570         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 571                 goto err;
 572
 573 out:
 574         /*
 575          * We weren't in a block-line scope when entering the
 576          * above-parsed macro, so return.
 577          */
 578
 579         if ( ! (MAN_BPLINE & m->flags)) {
 580                 m->flags &= ~MAN_ILINE;
 581                 return(1);
 582         }
 583         m->flags &= ~MAN_BPLINE;
 584
 585         /*
 586          * If we're in a block scope, then allow this macro to slip by
 587          * without closing scope around it.
 588          */
 589
 590         if (MAN_ILINE & m->flags) {
 591                 m->flags &= ~MAN_ILINE;
 592                 return(1);
 593         }
 594
 595         /*
 596          * If we've opened a new next-line element scope, then return
 597          * now, as the next line will close out the block scope.
 598          */
 599
 600         if (MAN_ELINE & m->flags)
 601                 return(1);
 602
 603         /* Close out the block scope opened in the prior line.  */
 604
 605         assert(MAN_BLINE & m->flags);
 606         m->flags &= ~MAN_BLINE;
 607
 608         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 609                 return(0);
 610         return(man_body_alloc(m, ln, offs, m->last->tok));
 611
 612 err:    /* Error out. */
 613
 614         m->flags |= MAN_HALT;
 615         return(0);
 616 }
 617
 618
 619 int
 620 man_vmsg(struct man *man, enum mandocerr t,
 621                 int ln, int pos, const char *fmt, ...)
 622 {
 623         char             buf[256];
 624         va_list          ap;
 625
 626         va_start(ap, fmt);
 627         vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 628         va_end(ap);
 629         return((*man->msg)(t, man->data, ln, pos, buf));
 630 }
 631
 632
 633 /*
 634  * Unlink a node from its context.  If "m" is provided, the last parse
 635  * point will also be adjusted accordingly.
 636  */
 637 static void
 638 man_node_unlink(struct man *m, struct man_node *n)
 639 {
 640
 641         /* Adjust siblings. */
 642
 643         if (n->prev)
 644                 n->prev->next = n->next;
 645         if (n->next)
 646                 n->next->prev = n->prev;
 647
 648         /* Adjust parent. */
 649
 650         if (n->parent) {
 651                 n->parent->nchild--;
 652                 if (n->parent->child == n)
 653                         n->parent->child = n->prev ? n->prev : n->next;
 654         }
 655
 656         /* Adjust parse point, if applicable. */
 657
 658         if (m && m->last == n) {
 659                 /*XXX: this can occur when bailing from validation. */
 660                 /*assert(NULL == n->next);*/
 661                 if (n->prev) {
 662                         m->last = n->prev;
 663                         m->next = MAN_NEXT_SIBLING;
 664                 } else {
 665                         m->last = n->parent;
 666                         m->next = MAN_NEXT_CHILD;
 667                 }
 668         }
 669
 670         if (m && m->first == n)
 671                 m->first = NULL;
 672 }