man.c

   1 /*      $Id: man.c,v 1.124 2014/01/06 00:53:33 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 #ifdef HAVE_CONFIG_H
  20 #include "config.h"
  21 #endif
  22
  23 #include <sys/types.h>
  24
  25 #include <assert.h>
  26 #include <stdarg.h>
  27 #include <stdlib.h>
  28 #include <stdio.h>
  29 #include <string.h>
  30
  31 #include "man.h"
  32 #include "mandoc.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"
  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         assert( ! (MAN_HALT & man->flags));
  70         return(man->first);
  71 }
  72
  73
  74 const struct man_meta *
  75 man_meta(const struct man *man)
  76 {
  77
  78         assert( ! (MAN_HALT & man->flags));
  79         return(&man->meta);
  80 }
  81
  82
  83 void
  84 man_reset(struct man *man)
  85 {
  86
  87         man_free1(man);
  88         man_alloc1(man);
  89 }
  90
  91
  92 void
  93 man_free(struct man *man)
  94 {
  95
  96         man_free1(man);
  97         free(man);
  98 }
  99
 100
 101 struct man *
 102 man_alloc(struct roff *roff, struct mparse *parse, int quick)
 103 {
 104         struct man      *p;
 105
 106         p = mandoc_calloc(1, sizeof(struct man));
 107
 108         man_hash_init();
 109         p->parse = parse;
 110         p->quick = quick;
 111         p->roff = roff;
 112
 113         man_alloc1(p);
 114         return(p);
 115 }
 116
 117
 118 int
 119 man_endparse(struct man *man)
 120 {
 121
 122         assert( ! (MAN_HALT & man->flags));
 123         if (man_macroend(man))
 124                 return(1);
 125         man->flags |= MAN_HALT;
 126         return(0);
 127 }
 128
 129
 130 int
 131 man_parseln(struct man *man, int ln, char *buf, int offs)
 132 {
 133
 134         man->flags |= MAN_NEWLINE;
 135
 136         assert( ! (MAN_HALT & man->flags));
 137
 138         return (roff_getcontrol(man->roff, buf, &offs) ?
 139                         man_pmacro(man, ln, buf, offs) :
 140                         man_ptext(man, ln, buf, offs));
 141 }
 142
 143
 144 static void
 145 man_free1(struct man *man)
 146 {
 147
 148         if (man->first)
 149                 man_node_delete(man, man->first);
 150         if (man->meta.title)
 151                 free(man->meta.title);
 152         if (man->meta.source)
 153                 free(man->meta.source);
 154         if (man->meta.date)
 155                 free(man->meta.date);
 156         if (man->meta.vol)
 157                 free(man->meta.vol);
 158         if (man->meta.msec)
 159                 free(man->meta.msec);
 160 }
 161
 162
 163 static void
 164 man_alloc1(struct man *man)
 165 {
 166
 167         memset(&man->meta, 0, sizeof(struct man_meta));
 168         man->flags = 0;
 169         man->last = mandoc_calloc(1, sizeof(struct man_node));
 170         man->first = man->last;
 171         man->last->type = MAN_ROOT;
 172         man->last->tok = MAN_MAX;
 173         man->next = MAN_NEXT_CHILD;
 174 }
 175
 176
 177 static int
 178 man_node_append(struct man *man, struct man_node *p)
 179 {
 180
 181         assert(man->last);
 182         assert(man->first);
 183         assert(MAN_ROOT != p->type);
 184
 185         switch (man->next) {
 186         case (MAN_NEXT_SIBLING):
 187                 man->last->next = p;
 188                 p->prev = man->last;
 189                 p->parent = man->last->parent;
 190                 break;
 191         case (MAN_NEXT_CHILD):
 192                 man->last->child = p;
 193                 p->parent = man->last;
 194                 break;
 195         default:
 196                 abort();
 197                 /* NOTREACHED */
 198         }
 199
 200         assert(p->parent);
 201         p->parent->nchild++;
 202
 203         if ( ! man_valid_pre(man, p))
 204                 return(0);
 205
 206         switch (p->type) {
 207         case (MAN_HEAD):
 208                 assert(MAN_BLOCK == p->parent->type);
 209                 p->parent->head = p;
 210                 break;
 211         case (MAN_TAIL):
 212                 assert(MAN_BLOCK == p->parent->type);
 213                 p->parent->tail = p;
 214                 break;
 215         case (MAN_BODY):
 216                 assert(MAN_BLOCK == p->parent->type);
 217                 p->parent->body = p;
 218                 break;
 219         default:
 220                 break;
 221         }
 222
 223         man->last = p;
 224
 225         switch (p->type) {
 226         case (MAN_TBL):
 227                 /* FALLTHROUGH */
 228         case (MAN_TEXT):
 229                 if ( ! man_valid_post(man))
 230                         return(0);
 231                 break;
 232         default:
 233                 break;
 234         }
 235
 236         return(1);
 237 }
 238
 239
 240 static struct man_node *
 241 man_node_alloc(struct man *man, int line, int pos,
 242                 enum man_type type, enum mant tok)
 243 {
 244         struct man_node *p;
 245
 246         p = mandoc_calloc(1, sizeof(struct man_node));
 247         p->line = line;
 248         p->pos = pos;
 249         p->type = type;
 250         p->tok = tok;
 251
 252         if (MAN_NEWLINE & man->flags)
 253                 p->flags |= MAN_LINE;
 254         man->flags &= ~MAN_NEWLINE;
 255         return(p);
 256 }
 257
 258
 259 int
 260 man_elem_alloc(struct man *man, int line, int pos, enum mant tok)
 261 {
 262         struct man_node *p;
 263
 264         p = man_node_alloc(man, line, pos, MAN_ELEM, tok);
 265         if ( ! man_node_append(man, p))
 266                 return(0);
 267         man->next = MAN_NEXT_CHILD;
 268         return(1);
 269 }
 270
 271
 272 int
 273 man_tail_alloc(struct man *man, int line, int pos, enum mant tok)
 274 {
 275         struct man_node *p;
 276
 277         p = man_node_alloc(man, line, pos, MAN_TAIL, tok);
 278         if ( ! man_node_append(man, p))
 279                 return(0);
 280         man->next = MAN_NEXT_CHILD;
 281         return(1);
 282 }
 283
 284
 285 int
 286 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
 287 {
 288         struct man_node *p;
 289
 290         p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
 291         if ( ! man_node_append(man, p))
 292                 return(0);
 293         man->next = MAN_NEXT_CHILD;
 294         return(1);
 295 }
 296
 297
 298 int
 299 man_body_alloc(struct man *man, int line, int pos, enum mant tok)
 300 {
 301         struct man_node *p;
 302
 303         p = man_node_alloc(man, line, pos, MAN_BODY, tok);
 304         if ( ! man_node_append(man, p))
 305                 return(0);
 306         man->next = MAN_NEXT_CHILD;
 307         return(1);
 308 }
 309
 310
 311 int
 312 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
 313 {
 314         struct man_node *p;
 315
 316         p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
 317         if ( ! man_node_append(man, p))
 318                 return(0);
 319         man->next = MAN_NEXT_CHILD;
 320         return(1);
 321 }
 322
 323 int
 324 man_word_alloc(struct man *man, int line, int pos, const char *word)
 325 {
 326         struct man_node *n;
 327
 328         n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
 329         n->string = roff_strdup(man->roff, word);
 330
 331         if ( ! man_node_append(man, n))
 332                 return(0);
 333
 334         man->next = MAN_NEXT_SIBLING;
 335         return(1);
 336 }
 337
 338
 339 /*
 340  * Free all of the resources held by a node.  This does NOT unlink a
 341  * node from its context; for that, see man_node_unlink().
 342  */
 343 static void
 344 man_node_free(struct man_node *p)
 345 {
 346
 347         if (p->string)
 348                 free(p->string);
 349         free(p);
 350 }
 351
 352
 353 void
 354 man_node_delete(struct man *man, struct man_node *p)
 355 {
 356
 357         while (p->child)
 358                 man_node_delete(man, p->child);
 359
 360         man_node_unlink(man, p);
 361         man_node_free(p);
 362 }
 363
 364 int
 365 man_addeqn(struct man *man, const struct eqn *ep)
 366 {
 367         struct man_node *n;
 368
 369         assert( ! (MAN_HALT & man->flags));
 370
 371         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 372         n->eqn = ep;
 373
 374         if ( ! man_node_append(man, n))
 375                 return(0);
 376
 377         man->next = MAN_NEXT_SIBLING;
 378         return(man_descope(man, ep->ln, ep->pos));
 379 }
 380
 381 int
 382 man_addspan(struct man *man, const struct tbl_span *sp)
 383 {
 384         struct man_node *n;
 385
 386         assert( ! (MAN_HALT & man->flags));
 387
 388         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 389         n->span = sp;
 390
 391         if ( ! man_node_append(man, n))
 392                 return(0);
 393
 394         man->next = MAN_NEXT_SIBLING;
 395         return(man_descope(man, sp->line, 0));
 396 }
 397
 398 static int
 399 man_descope(struct man *man, int line, int offs)
 400 {
 401         /*
 402          * Co-ordinate what happens with having a next-line scope open:
 403          * first close out the element scope (if applicable), then close
 404          * out the block scope (also if applicable).
 405          */
 406
 407         if (MAN_ELINE & man->flags) {
 408                 man->flags &= ~MAN_ELINE;
 409                 if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 410                         return(0);
 411         }
 412
 413         if ( ! (MAN_BLINE & man->flags))
 414                 return(1);
 415         man->flags &= ~MAN_BLINE;
 416
 417         if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 418                 return(0);
 419         return(man_body_alloc(man, line, offs, man->last->tok));
 420 }
 421
 422 static int
 423 man_ptext(struct man *man, int line, char *buf, int offs)
 424 {
 425         int              i;
 426
 427         /* Literal free-form text whitespace is preserved. */
 428
 429         if (MAN_LITERAL & man->flags) {
 430                 if ( ! man_word_alloc(man, line, offs, buf + offs))
 431                         return(0);
 432                 return(man_descope(man, line, offs));
 433         }
 434
 435         for (i = offs; ' ' == buf[i]; i++)
 436                 /* Skip leading whitespace. */ ;
 437
 438         /*
 439          * Blank lines are ignored right after headings
 440          * but add a single vertical space elsewhere.
 441          */
 442
 443         if ('\0' == buf[i]) {
 444                 /* Allocate a blank entry. */
 445                 if (MAN_SH != man->last->tok &&
 446                     MAN_SS != man->last->tok) {
 447                         if ( ! man_elem_alloc(man, line, offs, MAN_sp))
 448                                 return(0);
 449                         man->next = MAN_NEXT_SIBLING;
 450                 }
 451                 return(1);
 452         }
 453
 454         /*
 455          * Warn if the last un-escaped character is whitespace. Then
 456          * strip away the remaining spaces (tabs stay!).
 457          */
 458
 459         i = (int)strlen(buf);
 460         assert(i);
 461
 462         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 463                 if (i > 1 && '\\' != buf[i - 2])
 464                         man_pmsg(man, line, i - 1, MANDOCERR_EOLNSPACE);
 465
 466                 for (--i; i && ' ' == buf[i]; i--)
 467                         /* Spin back to non-space. */ ;
 468
 469                 /* Jump ahead of escaped whitespace. */
 470                 i += '\\' == buf[i] ? 2 : 1;
 471
 472                 buf[i] = '\0';
 473         }
 474
 475         if ( ! man_word_alloc(man, line, offs, buf + offs))
 476                 return(0);
 477
 478         /*
 479          * End-of-sentence check.  If the last character is an unescaped
 480          * EOS character, then flag the node as being the end of a
 481          * sentence.  The front-end will know how to interpret this.
 482          */
 483
 484         assert(i);
 485         if (mandoc_eos(buf, (size_t)i))
 486                 man->last->flags |= MAN_EOS;
 487
 488         return(man_descope(man, line, offs));
 489 }
 490
 491 static int
 492 man_pmacro(struct man *man, int ln, char *buf, int offs)
 493 {
 494         int              i, ppos;
 495         enum mant        tok;
 496         char             mac[5];
 497         struct man_node *n;
 498
 499         if ('"' == buf[offs]) {
 500                 man_pmsg(man, ln, offs, MANDOCERR_BADCOMMENT);
 501                 return(1);
 502         } else if ('\0' == buf[offs])
 503                 return(1);
 504
 505         ppos = offs;
 506
 507         /*
 508          * Copy the first word into a nil-terminated buffer.
 509          * Stop copying when a tab, space, or eoln is encountered.
 510          */
 511
 512         i = 0;
 513         while (i < 4 && '\0' != buf[offs] &&
 514                         ' ' != buf[offs] && '\t' != buf[offs])
 515                 mac[i++] = buf[offs++];
 516
 517         mac[i] = '\0';
 518
 519         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 520
 521         if (MAN_MAX == tok) {
 522                 mandoc_vmsg(MANDOCERR_MACRO, man->parse, ln,
 523                                 ppos, "%s", buf + ppos - 1);
 524                 return(1);
 525         }
 526
 527         /* The macro is sane.  Jump to the next word. */
 528
 529         while (buf[offs] && ' ' == buf[offs])
 530                 offs++;
 531
 532         /*
 533          * Trailing whitespace.  Note that tabs are allowed to be passed
 534          * into the parser as "text", so we only warn about spaces here.
 535          */
 536
 537         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 538                 man_pmsg(man, ln, offs - 1, MANDOCERR_EOLNSPACE);
 539
 540         /*
 541          * Remove prior ELINE macro, as it's being clobbered by a new
 542          * macro.  Note that NSCOPED macros do not close out ELINE
 543          * macros---they don't print text---so we let those slip by.
 544          */
 545
 546         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 547                         man->flags & MAN_ELINE) {
 548                 n = man->last;
 549                 assert(MAN_TEXT != n->type);
 550
 551                 /* Remove repeated NSCOPED macros causing ELINE. */
 552
 553                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 554                         n = n->parent;
 555
 556                 mandoc_vmsg(MANDOCERR_LINESCOPE, man->parse, n->line,
 557                     n->pos, "%s breaks %s", man_macronames[tok],
 558                     man_macronames[n->tok]);
 559
 560                 man_node_delete(man, n);
 561                 man->flags &= ~MAN_ELINE;
 562         }
 563
 564         /*
 565          * Remove prior BLINE macro that is being clobbered.
 566          */
 567         if ((man->flags & MAN_BLINE) &&
 568             (MAN_BSCOPE & man_macros[tok].flags)) {
 569                 n = man->last;
 570
 571                 /* Might be a text node like 8 in
 572                  * .TP 8
 573                  * .SH foo
 574                  */
 575                 if (MAN_TEXT == n->type)
 576                         n = n->parent;
 577
 578                 /* Remove element that didn't end BLINE, if any. */
 579                 if ( ! (MAN_BSCOPE & man_macros[n->tok].flags))
 580                         n = n->parent;
 581
 582                 assert(MAN_HEAD == n->type);
 583                 n = n->parent;
 584                 assert(MAN_BLOCK == n->type);
 585                 assert(MAN_SCOPED & man_macros[n->tok].flags);
 586
 587                 mandoc_vmsg(MANDOCERR_LINESCOPE, man->parse, n->line,
 588                     n->pos, "%s breaks %s", man_macronames[tok],
 589                     man_macronames[n->tok]);
 590
 591                 man_node_delete(man, n);
 592                 man->flags &= ~MAN_BLINE;
 593         }
 594
 595         /*
 596          * Save the fact that we're in the next-line for a block.  In
 597          * this way, embedded roff instructions can "remember" state
 598          * when they exit.
 599          */
 600
 601         if (MAN_BLINE & man->flags)
 602                 man->flags |= MAN_BPLINE;
 603
 604         /* Call to handler... */
 605
 606         assert(man_macros[tok].fp);
 607         if ( ! (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf))
 608                 goto err;
 609
 610         /* In quick mode (for mandocdb), abort after the NAME section. */
 611
 612         if (man->quick && MAN_SH == tok &&
 613             strcmp(man->last->prev->child->string, "NAME"))
 614                 return(2);
 615
 616         /*
 617          * We weren't in a block-line scope when entering the
 618          * above-parsed macro, so return.
 619          */
 620
 621         if ( ! (MAN_BPLINE & man->flags)) {
 622                 man->flags &= ~MAN_ILINE;
 623                 return(1);
 624         }
 625         man->flags &= ~MAN_BPLINE;
 626
 627         /*
 628          * If we're in a block scope, then allow this macro to slip by
 629          * without closing scope around it.
 630          */
 631
 632         if (MAN_ILINE & man->flags) {
 633                 man->flags &= ~MAN_ILINE;
 634                 return(1);
 635         }
 636
 637         /*
 638          * If we've opened a new next-line element scope, then return
 639          * now, as the next line will close out the block scope.
 640          */
 641
 642         if (MAN_ELINE & man->flags)
 643                 return(1);
 644
 645         /* Close out the block scope opened in the prior line.  */
 646
 647         assert(MAN_BLINE & man->flags);
 648         man->flags &= ~MAN_BLINE;
 649
 650         if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 651                 return(0);
 652         return(man_body_alloc(man, ln, ppos, man->last->tok));
 653
 654 err:    /* Error out. */
 655
 656         man->flags |= MAN_HALT;
 657         return(0);
 658 }
 659
 660 /*
 661  * Unlink a node from its context.  If "man" is provided, the last parse
 662  * point will also be adjusted accordingly.
 663  */
 664 static void
 665 man_node_unlink(struct man *man, struct man_node *n)
 666 {
 667
 668         /* Adjust siblings. */
 669
 670         if (n->prev)
 671                 n->prev->next = n->next;
 672         if (n->next)
 673                 n->next->prev = n->prev;
 674
 675         /* Adjust parent. */
 676
 677         if (n->parent) {
 678                 n->parent->nchild--;
 679                 if (n->parent->child == n)
 680                         n->parent->child = n->prev ? n->prev : n->next;
 681         }
 682
 683         /* Adjust parse point, if applicable. */
 684
 685         if (man && man->last == n) {
 686                 /*XXX: this can occur when bailing from validation. */
 687                 /*assert(NULL == n->next);*/
 688                 if (n->prev) {
 689                         man->last = n->prev;
 690                         man->next = MAN_NEXT_SIBLING;
 691                 } else {
 692                         man->last = n->parent;
 693                         man->next = MAN_NEXT_CHILD;
 694                 }
 695         }
 696
 697         if (man && man->first == n)
 698                 man->first = NULL;
 699 }
 700
 701 const struct mparse *
 702 man_mparse(const struct man *man)
 703 {
 704
 705         assert(man && man->parse);
 706         return(man->parse);
 707 }