man.c

   1 /*      $Id: man.c,v 1.141 2014/10/20 15:50:24 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 /*
 319  * Free all of the resources held by a node.  This does NOT unlink a
 320  * node from its context; for that, see man_node_unlink().
 321  */
 322 static void
 323 man_node_free(struct man_node *p)
 324 {
 325
 326         if (p->string)
 327                 free(p->string);
 328         free(p);
 329 }
 330
 331 void
 332 man_node_delete(struct man *man, struct man_node *p)
 333 {
 334
 335         while (p->child)
 336                 man_node_delete(man, p->child);
 337
 338         man_node_unlink(man, p);
 339         man_node_free(p);
 340 }
 341
 342 int
 343 man_addeqn(struct man *man, const struct eqn *ep)
 344 {
 345         struct man_node *n;
 346
 347         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 348         n->eqn = ep;
 349         if (ep->ln > man->last->line)
 350                 n->flags |= MAN_LINE;
 351
 352         if ( ! man_node_append(man, n))
 353                 return(0);
 354
 355         man->next = MAN_NEXT_SIBLING;
 356         return(man_descope(man, ep->ln, ep->pos));
 357 }
 358
 359 int
 360 man_addspan(struct man *man, const struct tbl_span *sp)
 361 {
 362         struct man_node *n;
 363
 364         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 365         n->span = sp;
 366
 367         if ( ! man_node_append(man, n))
 368                 return(0);
 369
 370         man->next = MAN_NEXT_SIBLING;
 371         return(man_descope(man, sp->line, 0));
 372 }
 373
 374 static int
 375 man_descope(struct man *man, int line, int offs)
 376 {
 377         /*
 378          * Co-ordinate what happens with having a next-line scope open:
 379          * first close out the element scope (if applicable), then close
 380          * out the block scope (also if applicable).
 381          */
 382
 383         if (MAN_ELINE & man->flags) {
 384                 man->flags &= ~MAN_ELINE;
 385                 if ( ! man_unscope(man, man->last->parent))
 386                         return(0);
 387         }
 388
 389         if ( ! (MAN_BLINE & man->flags))
 390                 return(1);
 391         man->flags &= ~MAN_BLINE;
 392
 393         if ( ! man_unscope(man, man->last->parent))
 394                 return(0);
 395         return(man_body_alloc(man, line, offs, man->last->tok));
 396 }
 397
 398 static int
 399 man_ptext(struct man *man, int line, char *buf, int offs)
 400 {
 401         int              i;
 402
 403         /* Literal free-form text whitespace is preserved. */
 404
 405         if (MAN_LITERAL & man->flags) {
 406                 if ( ! man_word_alloc(man, line, offs, buf + offs))
 407                         return(0);
 408                 return(man_descope(man, line, offs));
 409         }
 410
 411         for (i = offs; ' ' == buf[i]; i++)
 412                 /* Skip leading whitespace. */ ;
 413
 414         /*
 415          * Blank lines are ignored right after headings
 416          * but add a single vertical space elsewhere.
 417          */
 418
 419         if ('\0' == buf[i]) {
 420                 /* Allocate a blank entry. */
 421                 if (MAN_SH != man->last->tok &&
 422                     MAN_SS != man->last->tok) {
 423                         if ( ! man_elem_alloc(man, line, offs, MAN_sp))
 424                                 return(0);
 425                         man->next = MAN_NEXT_SIBLING;
 426                 }
 427                 return(1);
 428         }
 429
 430         /*
 431          * Warn if the last un-escaped character is whitespace. Then
 432          * strip away the remaining spaces (tabs stay!).
 433          */
 434
 435         i = (int)strlen(buf);
 436         assert(i);
 437
 438         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 439                 if (i > 1 && '\\' != buf[i - 2])
 440                         mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 441                             line, i - 1, NULL);
 442
 443                 for (--i; i && ' ' == buf[i]; i--)
 444                         /* Spin back to non-space. */ ;
 445
 446                 /* Jump ahead of escaped whitespace. */
 447                 i += '\\' == buf[i] ? 2 : 1;
 448
 449                 buf[i] = '\0';
 450         }
 451
 452         if ( ! man_word_alloc(man, line, offs, buf + offs))
 453                 return(0);
 454
 455         /*
 456          * End-of-sentence check.  If the last character is an unescaped
 457          * EOS character, then flag the node as being the end of a
 458          * sentence.  The front-end will know how to interpret this.
 459          */
 460
 461         assert(i);
 462         if (mandoc_eos(buf, (size_t)i))
 463                 man->last->flags |= MAN_EOS;
 464
 465         return(man_descope(man, line, offs));
 466 }
 467
 468 static int
 469 man_pmacro(struct man *man, int ln, char *buf, int offs)
 470 {
 471         char             mac[5];
 472         struct man_node *n;
 473         enum mant        tok;
 474         int              i, ppos;
 475         int              bline;
 476
 477         ppos = offs;
 478
 479         /*
 480          * Copy the first word into a nil-terminated buffer.
 481          * Stop copying when a tab, space, or eoln is encountered.
 482          */
 483
 484         i = 0;
 485         while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
 486             '\t' != buf[offs])
 487                 mac[i++] = buf[offs++];
 488
 489         mac[i] = '\0';
 490
 491         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 492
 493         if (MAN_MAX == tok) {
 494                 mandoc_msg(MANDOCERR_MACRO, man->parse,
 495                     ln, ppos, buf + ppos - 1);
 496                 return(1);
 497         }
 498
 499         /* The macro is sane.  Jump to the next word. */
 500
 501         while (buf[offs] && ' ' == buf[offs])
 502                 offs++;
 503
 504         /*
 505          * Trailing whitespace.  Note that tabs are allowed to be passed
 506          * into the parser as "text", so we only warn about spaces here.
 507          */
 508
 509         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 510                 mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 511                     ln, offs - 1, NULL);
 512
 513         /*
 514          * Remove prior ELINE macro, as it's being clobbered by a new
 515          * macro.  Note that NSCOPED macros do not close out ELINE
 516          * macros---they don't print text---so we let those slip by.
 517          */
 518
 519         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 520                         man->flags & MAN_ELINE) {
 521                 n = man->last;
 522                 assert(MAN_TEXT != n->type);
 523
 524                 /* Remove repeated NSCOPED macros causing ELINE. */
 525
 526                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 527                         n = n->parent;
 528
 529                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 530                     n->pos, "%s breaks %s", man_macronames[tok],
 531                     man_macronames[n->tok]);
 532
 533                 man_node_delete(man, n);
 534                 man->flags &= ~MAN_ELINE;
 535         }
 536
 537         /*
 538          * Remove prior BLINE macro that is being clobbered.
 539          */
 540         if ((man->flags & MAN_BLINE) &&
 541             (MAN_BSCOPE & man_macros[tok].flags)) {
 542                 n = man->last;
 543
 544                 /* Might be a text node like 8 in
 545                  * .TP 8
 546                  * .SH foo
 547                  */
 548                 if (MAN_TEXT == n->type)
 549                         n = n->parent;
 550
 551                 /* Remove element that didn't end BLINE, if any. */
 552                 if ( ! (MAN_BSCOPE & man_macros[n->tok].flags))
 553                         n = n->parent;
 554
 555                 assert(MAN_HEAD == n->type);
 556                 n = n->parent;
 557                 assert(MAN_BLOCK == n->type);
 558                 assert(MAN_SCOPED & man_macros[n->tok].flags);
 559
 560                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 561                     n->pos, "%s breaks %s", man_macronames[tok],
 562                     man_macronames[n->tok]);
 563
 564                 man_node_delete(man, n);
 565                 man->flags &= ~MAN_BLINE;
 566         }
 567
 568         /* Remember whether we are in next-line scope for a block head. */
 569
 570         bline = man->flags & MAN_BLINE;
 571
 572         /* Call to handler... */
 573
 574         assert(man_macros[tok].fp);
 575         if ( ! (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf))
 576                 return(0);
 577
 578         /* In quick mode (for mandocdb), abort after the NAME section. */
 579
 580         if (man->quick && MAN_SH == tok) {
 581                 n = man->last;
 582                 if (MAN_BODY == n->type &&
 583                     strcmp(n->prev->child->string, "NAME"))
 584                         return(2);
 585         }
 586
 587         /*
 588          * If we are in a next-line scope for a block head,
 589          * close it out now and switch to the body,
 590          * unless the next-line scope is allowed to continue.
 591          */
 592
 593         if ( ! bline || man->flags & MAN_ELINE ||
 594             man_macros[tok].flags & MAN_NSCOPED)
 595                 return(1);
 596
 597         assert(MAN_BLINE & man->flags);
 598         man->flags &= ~MAN_BLINE;
 599
 600         if ( ! man_unscope(man, man->last->parent))
 601                 return(0);
 602         return(man_body_alloc(man, ln, ppos, man->last->tok));
 603 }
 604
 605 /*
 606  * Unlink a node from its context.  If "man" is provided, the last parse
 607  * point will also be adjusted accordingly.
 608  */
 609 static void
 610 man_node_unlink(struct man *man, struct man_node *n)
 611 {
 612
 613         /* Adjust siblings. */
 614
 615         if (n->prev)
 616                 n->prev->next = n->next;
 617         if (n->next)
 618                 n->next->prev = n->prev;
 619
 620         /* Adjust parent. */
 621
 622         if (n->parent) {
 623                 n->parent->nchild--;
 624                 if (n->parent->child == n)
 625                         n->parent->child = n->prev ? n->prev : n->next;
 626         }
 627
 628         /* Adjust parse point, if applicable. */
 629
 630         if (man && man->last == n) {
 631                 /*XXX: this can occur when bailing from validation. */
 632                 /*assert(NULL == n->next);*/
 633                 if (n->prev) {
 634                         man->last = n->prev;
 635                         man->next = MAN_NEXT_SIBLING;
 636                 } else {
 637                         man->last = n->parent;
 638                         man->next = MAN_NEXT_CHILD;
 639                 }
 640         }
 641
 642         if (man && man->first == n)
 643                 man->first = NULL;
 644 }
 645
 646 const struct mparse *
 647 man_mparse(const struct man *man)
 648 {
 649
 650         assert(man && man->parse);
 651         return(man->parse);
 652 }
 653
 654 void
 655 man_deroff(char **dest, const struct man_node *n)
 656 {
 657         char    *cp;
 658         size_t   sz;
 659
 660         if (MAN_TEXT != n->type) {
 661                 for (n = n->child; n; n = n->next)
 662                         man_deroff(dest, n);
 663                 return;
 664         }
 665
 666         /* Skip leading whitespace and escape sequences. */
 667
 668         cp = n->string;
 669         while ('\0' != *cp) {
 670                 if ('\\' == *cp) {
 671                         cp++;
 672                         mandoc_escape((const char **)&cp, NULL, NULL);
 673                 } else if (isspace((unsigned char)*cp))
 674                         cp++;
 675                 else
 676                         break;
 677         }
 678
 679         /* Skip trailing whitespace. */
 680
 681         for (sz = strlen(cp); sz; sz--)
 682                 if (0 == isspace((unsigned char)cp[sz-1]))
 683                         break;
 684
 685         /* Skip empty strings. */
 686
 687         if (0 == sz)
 688                 return;
 689
 690         if (NULL == *dest) {
 691                 *dest = mandoc_strndup(cp, sz);
 692                 return;
 693         }
 694
 695         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 696         free(*dest);
 697         *dest = cp;
 698 }