man.c

   1 /*      $Id: man.c,v 1.143 2014/11/19 03:08:17 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
   5  * Copyright (c) 2011 Joerg Sonnenberger <joerg@netbsd.org>
   6  *
   7  * Permission to use, copy, modify, and distribute this software for any
   8  * purpose with or without fee is hereby granted, provided that the above
   9  * copyright notice and this permission notice appear in all copies.
  10  *
  11  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  14  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  15  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  16  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  17  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  18  */
  19 #include "config.h"
  20
  21 #include <sys/types.h>
  22
  23 #include <assert.h>
  24 #include <ctype.h>
  25 #include <stdarg.h>
  26 #include <stdlib.h>
  27 #include <stdio.h>
  28 #include <string.h>
  29
  30 #include "man.h"
  31 #include "mandoc.h"
  32 #include "mandoc_aux.h"
  33 #include "libman.h"
  34 #include "libmandoc.h"
  35
  36 const   char *const __man_macronames[MAN_MAX] = {
  37         "br",           "TH",           "SH",           "SS",
  38         "TP",           "LP",           "PP",           "P",
  39         "IP",           "HP",           "SM",           "SB",
  40         "BI",           "IB",           "BR",           "RB",
  41         "R",            "B",            "I",            "IR",
  42         "RI",           "na",           "sp",           "nf",
  43         "fi",           "RE",           "RS",           "DT",
  44         "UC",           "PD",           "AT",           "in",
  45         "ft",           "OP",           "EX",           "EE",
  46         "UR",           "UE",           "ll"
  47         };
  48
  49 const   char * const *man_macronames = __man_macronames;
  50
  51 static  struct man_node *man_node_alloc(struct man *, int, int,
  52                                 enum man_type, enum mant);
  53 static  int              man_node_append(struct man *,
  54                                 struct man_node *);
  55 static  void             man_node_free(struct man_node *);
  56 static  void             man_node_unlink(struct man *,
  57                                 struct man_node *);
  58 static  int              man_ptext(struct man *, int, char *, int);
  59 static  int              man_pmacro(struct man *, int, char *, int);
  60 static  void             man_free1(struct man *);
  61 static  void             man_alloc1(struct man *);
  62 static  int              man_descope(struct man *, int, int);
  63
  64
  65 const struct man_node *
  66 man_node(const struct man *man)
  67 {
  68
  69         return(man->first);
  70 }
  71
  72 const struct man_meta *
  73 man_meta(const struct man *man)
  74 {
  75
  76         return(&man->meta);
  77 }
  78
  79 void
  80 man_reset(struct man *man)
  81 {
  82
  83         man_free1(man);
  84         man_alloc1(man);
  85 }
  86
  87 void
  88 man_free(struct man *man)
  89 {
  90
  91         man_free1(man);
  92         free(man);
  93 }
  94
  95 struct man *
  96 man_alloc(struct roff *roff, struct mparse *parse, int quick)
  97 {
  98         struct man      *p;
  99
 100         p = mandoc_calloc(1, sizeof(struct man));
 101
 102         man_hash_init();
 103         p->parse = parse;
 104         p->quick = quick;
 105         p->roff = roff;
 106
 107         man_alloc1(p);
 108         return(p);
 109 }
 110
 111 int
 112 man_endparse(struct man *man)
 113 {
 114
 115         return(man_macroend(man));
 116 }
 117
 118 int
 119 man_parseln(struct man *man, int ln, char *buf, int offs)
 120 {
 121
 122         if (man->last->type != MAN_EQN || ln > man->last->line)
 123                 man->flags |= MAN_NEWLINE;
 124
 125         return (roff_getcontrol(man->roff, buf, &offs) ?
 126             man_pmacro(man, ln, buf, offs) :
 127             man_ptext(man, ln, buf, offs));
 128 }
 129
 130 static void
 131 man_free1(struct man *man)
 132 {
 133
 134         if (man->first)
 135                 man_node_delete(man, man->first);
 136         if (man->meta.title)
 137                 free(man->meta.title);
 138         if (man->meta.source)
 139                 free(man->meta.source);
 140         if (man->meta.date)
 141                 free(man->meta.date);
 142         if (man->meta.vol)
 143                 free(man->meta.vol);
 144         if (man->meta.msec)
 145                 free(man->meta.msec);
 146 }
 147
 148 static void
 149 man_alloc1(struct man *man)
 150 {
 151
 152         memset(&man->meta, 0, sizeof(struct man_meta));
 153         man->flags = 0;
 154         man->last = mandoc_calloc(1, sizeof(struct man_node));
 155         man->first = man->last;
 156         man->last->type = MAN_ROOT;
 157         man->last->tok = MAN_MAX;
 158         man->next = MAN_NEXT_CHILD;
 159 }
 160
 161
 162 static int
 163 man_node_append(struct man *man, struct man_node *p)
 164 {
 165
 166         assert(man->last);
 167         assert(man->first);
 168         assert(MAN_ROOT != p->type);
 169
 170         switch (man->next) {
 171         case MAN_NEXT_SIBLING:
 172                 man->last->next = p;
 173                 p->prev = man->last;
 174                 p->parent = man->last->parent;
 175                 break;
 176         case MAN_NEXT_CHILD:
 177                 man->last->child = p;
 178                 p->parent = man->last;
 179                 break;
 180         default:
 181                 abort();
 182                 /* NOTREACHED */
 183         }
 184
 185         assert(p->parent);
 186         p->parent->nchild++;
 187
 188         switch (p->type) {
 189         case MAN_BLOCK:
 190                 if (p->tok == MAN_SH || p->tok == MAN_SS)
 191                         man->flags &= ~MAN_LITERAL;
 192                 break;
 193         case MAN_HEAD:
 194                 assert(MAN_BLOCK == p->parent->type);
 195                 p->parent->head = p;
 196                 break;
 197         case MAN_TAIL:
 198                 assert(MAN_BLOCK == p->parent->type);
 199                 p->parent->tail = p;
 200                 break;
 201         case MAN_BODY:
 202                 assert(MAN_BLOCK == p->parent->type);
 203                 p->parent->body = p;
 204                 break;
 205         default:
 206                 break;
 207         }
 208
 209         man->last = p;
 210
 211         switch (p->type) {
 212         case MAN_TBL:
 213                 /* FALLTHROUGH */
 214         case MAN_TEXT:
 215                 if ( ! man_valid_post(man))
 216                         return(0);
 217                 break;
 218         default:
 219                 break;
 220         }
 221
 222         return(1);
 223 }
 224
 225 static struct man_node *
 226 man_node_alloc(struct man *man, int line, int pos,
 227                 enum man_type type, enum mant tok)
 228 {
 229         struct man_node *p;
 230
 231         p = mandoc_calloc(1, sizeof(struct man_node));
 232         p->line = line;
 233         p->pos = pos;
 234         p->type = type;
 235         p->tok = tok;
 236
 237         if (MAN_NEWLINE & man->flags)
 238                 p->flags |= MAN_LINE;
 239         man->flags &= ~MAN_NEWLINE;
 240         return(p);
 241 }
 242
 243 int
 244 man_elem_alloc(struct man *man, int line, int pos, enum mant tok)
 245 {
 246         struct man_node *p;
 247
 248         p = man_node_alloc(man, line, pos, MAN_ELEM, tok);
 249         if ( ! man_node_append(man, p))
 250                 return(0);
 251         man->next = MAN_NEXT_CHILD;
 252         return(1);
 253 }
 254
 255 int
 256 man_tail_alloc(struct man *man, int line, int pos, enum mant tok)
 257 {
 258         struct man_node *p;
 259
 260         p = man_node_alloc(man, line, pos, MAN_TAIL, tok);
 261         if ( ! man_node_append(man, p))
 262                 return(0);
 263         man->next = MAN_NEXT_CHILD;
 264         return(1);
 265 }
 266
 267 int
 268 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
 269 {
 270         struct man_node *p;
 271
 272         p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
 273         if ( ! man_node_append(man, p))
 274                 return(0);
 275         man->next = MAN_NEXT_CHILD;
 276         return(1);
 277 }
 278
 279 int
 280 man_body_alloc(struct man *man, int line, int pos, enum mant tok)
 281 {
 282         struct man_node *p;
 283
 284         p = man_node_alloc(man, line, pos, MAN_BODY, tok);
 285         if ( ! man_node_append(man, p))
 286                 return(0);
 287         man->next = MAN_NEXT_CHILD;
 288         return(1);
 289 }
 290
 291 int
 292 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
 293 {
 294         struct man_node *p;
 295
 296         p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
 297         if ( ! man_node_append(man, p))
 298                 return(0);
 299         man->next = MAN_NEXT_CHILD;
 300         return(1);
 301 }
 302
 303 int
 304 man_word_alloc(struct man *man, int line, int pos, const char *word)
 305 {
 306         struct man_node *n;
 307
 308         n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
 309         n->string = roff_strdup(man->roff, word);
 310
 311         if ( ! man_node_append(man, n))
 312                 return(0);
 313
 314         man->next = MAN_NEXT_SIBLING;
 315         return(1);
 316 }
 317
 318 void
 319 man_word_append(struct man *man, const char *word)
 320 {
 321         struct man_node *n;
 322         char            *addstr, *newstr;
 323
 324         n = man->last;
 325         addstr = roff_strdup(man->roff, word);
 326         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 327         free(addstr);
 328         free(n->string);
 329         n->string = newstr;
 330         man->next = MAN_NEXT_SIBLING;
 331 }
 332
 333 /*
 334  * Free all of the resources held by a node.  This does NOT unlink a
 335  * node from its context; for that, see man_node_unlink().
 336  */
 337 static void
 338 man_node_free(struct man_node *p)
 339 {
 340
 341         if (p->string)
 342                 free(p->string);
 343         free(p);
 344 }
 345
 346 void
 347 man_node_delete(struct man *man, struct man_node *p)
 348 {
 349
 350         while (p->child)
 351                 man_node_delete(man, p->child);
 352
 353         man_node_unlink(man, p);
 354         man_node_free(p);
 355 }
 356
 357 int
 358 man_addeqn(struct man *man, const struct eqn *ep)
 359 {
 360         struct man_node *n;
 361
 362         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 363         n->eqn = ep;
 364         if (ep->ln > man->last->line)
 365                 n->flags |= MAN_LINE;
 366
 367         if ( ! man_node_append(man, n))
 368                 return(0);
 369
 370         man->next = MAN_NEXT_SIBLING;
 371         return(man_descope(man, ep->ln, ep->pos));
 372 }
 373
 374 int
 375 man_addspan(struct man *man, const struct tbl_span *sp)
 376 {
 377         struct man_node *n;
 378
 379         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 380         n->span = sp;
 381
 382         if ( ! man_node_append(man, n))
 383                 return(0);
 384
 385         man->next = MAN_NEXT_SIBLING;
 386         return(man_descope(man, sp->line, 0));
 387 }
 388
 389 static int
 390 man_descope(struct man *man, int line, int offs)
 391 {
 392         /*
 393          * Co-ordinate what happens with having a next-line scope open:
 394          * first close out the element scope (if applicable), then close
 395          * out the block scope (also if applicable).
 396          */
 397
 398         if (MAN_ELINE & man->flags) {
 399                 man->flags &= ~MAN_ELINE;
 400                 if ( ! man_unscope(man, man->last->parent))
 401                         return(0);
 402         }
 403
 404         if ( ! (MAN_BLINE & man->flags))
 405                 return(1);
 406         man->flags &= ~MAN_BLINE;
 407
 408         if ( ! man_unscope(man, man->last->parent))
 409                 return(0);
 410         return(man_body_alloc(man, line, offs, man->last->tok));
 411 }
 412
 413 static int
 414 man_ptext(struct man *man, int line, char *buf, int offs)
 415 {
 416         int              i;
 417
 418         /* Literal free-form text whitespace is preserved. */
 419
 420         if (MAN_LITERAL & man->flags) {
 421                 if ( ! man_word_alloc(man, line, offs, buf + offs))
 422                         return(0);
 423                 return(man_descope(man, line, offs));
 424         }
 425
 426         for (i = offs; ' ' == buf[i]; i++)
 427                 /* Skip leading whitespace. */ ;
 428
 429         /*
 430          * Blank lines are ignored right after headings
 431          * but add a single vertical space elsewhere.
 432          */
 433
 434         if ('\0' == buf[i]) {
 435                 /* Allocate a blank entry. */
 436                 if (MAN_SH != man->last->tok &&
 437                     MAN_SS != man->last->tok) {
 438                         if ( ! man_elem_alloc(man, line, offs, MAN_sp))
 439                                 return(0);
 440                         man->next = MAN_NEXT_SIBLING;
 441                 }
 442                 return(1);
 443         }
 444
 445         /*
 446          * Warn if the last un-escaped character is whitespace. Then
 447          * strip away the remaining spaces (tabs stay!).
 448          */
 449
 450         i = (int)strlen(buf);
 451         assert(i);
 452
 453         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 454                 if (i > 1 && '\\' != buf[i - 2])
 455                         mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 456                             line, i - 1, NULL);
 457
 458                 for (--i; i && ' ' == buf[i]; i--)
 459                         /* Spin back to non-space. */ ;
 460
 461                 /* Jump ahead of escaped whitespace. */
 462                 i += '\\' == buf[i] ? 2 : 1;
 463
 464                 buf[i] = '\0';
 465         }
 466
 467         if ( ! man_word_alloc(man, line, offs, buf + offs))
 468                 return(0);
 469
 470         /*
 471          * End-of-sentence check.  If the last character is an unescaped
 472          * EOS character, then flag the node as being the end of a
 473          * sentence.  The front-end will know how to interpret this.
 474          */
 475
 476         assert(i);
 477         if (mandoc_eos(buf, (size_t)i))
 478                 man->last->flags |= MAN_EOS;
 479
 480         return(man_descope(man, line, offs));
 481 }
 482
 483 static int
 484 man_pmacro(struct man *man, int ln, char *buf, int offs)
 485 {
 486         struct man_node *n;
 487         const char      *cp;
 488         enum mant        tok;
 489         int              i, ppos;
 490         int              bline;
 491         char             mac[5];
 492
 493         ppos = offs;
 494
 495         /*
 496          * Copy the first word into a nil-terminated buffer.
 497          * Stop when a space, tab, escape, or eoln is encountered.
 498          */
 499
 500         i = 0;
 501         while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
 502                 mac[i++] = buf[offs++];
 503
 504         mac[i] = '\0';
 505
 506         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 507
 508         if (tok == MAN_MAX) {
 509                 mandoc_msg(MANDOCERR_MACRO, man->parse,
 510                     ln, ppos, buf + ppos - 1);
 511                 return(1);
 512         }
 513
 514         /* Skip a leading escape sequence or tab. */
 515
 516         switch (buf[offs]) {
 517         case '\\':
 518                 cp = buf + offs + 1;
 519                 mandoc_escape(&cp, NULL, NULL);
 520                 offs = cp - buf;
 521                 break;
 522         case '\t':
 523                 offs++;
 524                 break;
 525         default:
 526                 break;
 527         }
 528
 529         /* Jump to the next non-whitespace word. */
 530
 531         while (buf[offs] && ' ' == buf[offs])
 532                 offs++;
 533
 534         /*
 535          * Trailing whitespace.  Note that tabs are allowed to be passed
 536          * into the parser as "text", so we only warn about spaces here.
 537          */
 538
 539         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 540                 mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 541                     ln, offs - 1, NULL);
 542
 543         /*
 544          * Remove prior ELINE macro, as it's being clobbered by a new
 545          * macro.  Note that NSCOPED macros do not close out ELINE
 546          * macros---they don't print text---so we let those slip by.
 547          */
 548
 549         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 550                         man->flags & MAN_ELINE) {
 551                 n = man->last;
 552                 assert(MAN_TEXT != n->type);
 553
 554                 /* Remove repeated NSCOPED macros causing ELINE. */
 555
 556                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 557                         n = n->parent;
 558
 559                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 560                     n->pos, "%s breaks %s", man_macronames[tok],
 561                     man_macronames[n->tok]);
 562
 563                 man_node_delete(man, n);
 564                 man->flags &= ~MAN_ELINE;
 565         }
 566
 567         /*
 568          * Remove prior BLINE macro that is being clobbered.
 569          */
 570         if ((man->flags & MAN_BLINE) &&
 571             (MAN_BSCOPE & man_macros[tok].flags)) {
 572                 n = man->last;
 573
 574                 /* Might be a text node like 8 in
 575                  * .TP 8
 576                  * .SH foo
 577                  */
 578                 if (MAN_TEXT == n->type)
 579                         n = n->parent;
 580
 581                 /* Remove element that didn't end BLINE, if any. */
 582                 if ( ! (MAN_BSCOPE & man_macros[n->tok].flags))
 583                         n = n->parent;
 584
 585                 assert(MAN_HEAD == n->type);
 586                 n = n->parent;
 587                 assert(MAN_BLOCK == n->type);
 588                 assert(MAN_SCOPED & man_macros[n->tok].flags);
 589
 590                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 591                     n->pos, "%s breaks %s", man_macronames[tok],
 592                     man_macronames[n->tok]);
 593
 594                 man_node_delete(man, n);
 595                 man->flags &= ~MAN_BLINE;
 596         }
 597
 598         /* Remember whether we are in next-line scope for a block head. */
 599
 600         bline = man->flags & MAN_BLINE;
 601
 602         /* Call to handler... */
 603
 604         assert(man_macros[tok].fp);
 605         if ( ! (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf))
 606                 return(0);
 607
 608         /* In quick mode (for mandocdb), abort after the NAME section. */
 609
 610         if (man->quick && MAN_SH == tok) {
 611                 n = man->last;
 612                 if (MAN_BODY == n->type &&
 613                     strcmp(n->prev->child->string, "NAME"))
 614                         return(2);
 615         }
 616
 617         /*
 618          * If we are in a next-line scope for a block head,
 619          * close it out now and switch to the body,
 620          * unless the next-line scope is allowed to continue.
 621          */
 622
 623         if ( ! bline || man->flags & MAN_ELINE ||
 624             man_macros[tok].flags & MAN_NSCOPED)
 625                 return(1);
 626
 627         assert(MAN_BLINE & man->flags);
 628         man->flags &= ~MAN_BLINE;
 629
 630         if ( ! man_unscope(man, man->last->parent))
 631                 return(0);
 632         return(man_body_alloc(man, ln, ppos, man->last->tok));
 633 }
 634
 635 /*
 636  * Unlink a node from its context.  If "man" is provided, the last parse
 637  * point will also be adjusted accordingly.
 638  */
 639 static void
 640 man_node_unlink(struct man *man, struct man_node *n)
 641 {
 642
 643         /* Adjust siblings. */
 644
 645         if (n->prev)
 646                 n->prev->next = n->next;
 647         if (n->next)
 648                 n->next->prev = n->prev;
 649
 650         /* Adjust parent. */
 651
 652         if (n->parent) {
 653                 n->parent->nchild--;
 654                 if (n->parent->child == n)
 655                         n->parent->child = n->prev ? n->prev : n->next;
 656         }
 657
 658         /* Adjust parse point, if applicable. */
 659
 660         if (man && man->last == n) {
 661                 /*XXX: this can occur when bailing from validation. */
 662                 /*assert(NULL == n->next);*/
 663                 if (n->prev) {
 664                         man->last = n->prev;
 665                         man->next = MAN_NEXT_SIBLING;
 666                 } else {
 667                         man->last = n->parent;
 668                         man->next = MAN_NEXT_CHILD;
 669                 }
 670         }
 671
 672         if (man && man->first == n)
 673                 man->first = NULL;
 674 }
 675
 676 const struct mparse *
 677 man_mparse(const struct man *man)
 678 {
 679
 680         assert(man && man->parse);
 681         return(man->parse);
 682 }
 683
 684 void
 685 man_deroff(char **dest, const struct man_node *n)
 686 {
 687         char    *cp;
 688         size_t   sz;
 689
 690         if (MAN_TEXT != n->type) {
 691                 for (n = n->child; n; n = n->next)
 692                         man_deroff(dest, n);
 693                 return;
 694         }
 695
 696         /* Skip leading whitespace and escape sequences. */
 697
 698         cp = n->string;
 699         while ('\0' != *cp) {
 700                 if ('\\' == *cp) {
 701                         cp++;
 702                         mandoc_escape((const char **)&cp, NULL, NULL);
 703                 } else if (isspace((unsigned char)*cp))
 704                         cp++;
 705                 else
 706                         break;
 707         }
 708
 709         /* Skip trailing whitespace. */
 710
 711         for (sz = strlen(cp); sz; sz--)
 712                 if (0 == isspace((unsigned char)cp[sz-1]))
 713                         break;
 714
 715         /* Skip empty strings. */
 716
 717         if (0 == sz)
 718                 return;
 719
 720         if (NULL == *dest) {
 721                 *dest = mandoc_strndup(cp, sz);
 722                 return;
 723         }
 724
 725         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 726         free(*dest);
 727         *dest = cp;
 728 }