contrib/mdocml/man.c

   1 /*      $Id: man.c,v 1.137 2014/08/01 21:24:17 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 #ifdef HAVE_CONFIG_H
  20 #include "config.h"
  21 #endif
  22
  23 #include <sys/types.h>
  24
  25 #include <assert.h>
  26 #include <ctype.h>
  27 #include <stdarg.h>
  28 #include <stdlib.h>
  29 #include <stdio.h>
  30 #include <string.h>
  31
  32 #include "man.h"
  33 #include "mandoc.h"
  34 #include "mandoc_aux.h"
  35 #include "libman.h"
  36 #include "libmandoc.h"
  37
  38 const   char *const __man_macronames[MAN_MAX] = {
  39         "br",           "TH",           "SH",           "SS",
  40         "TP",           "LP",           "PP",           "P",
  41         "IP",           "HP",           "SM",           "SB",
  42         "BI",           "IB",           "BR",           "RB",
  43         "R",            "B",            "I",            "IR",
  44         "RI",           "na",           "sp",           "nf",
  45         "fi",           "RE",           "RS",           "DT",
  46         "UC",           "PD",           "AT",           "in",
  47         "ft",           "OP",           "EX",           "EE",
  48         "UR",           "UE",           "ll"
  49         };
  50
  51 const   char * const *man_macronames = __man_macronames;
  52
  53 static  struct man_node *man_node_alloc(struct man *, int, int,
  54                                 enum man_type, enum mant);
  55 static  int              man_node_append(struct man *,
  56                                 struct man_node *);
  57 static  void             man_node_free(struct man_node *);
  58 static  void             man_node_unlink(struct man *,
  59                                 struct man_node *);
  60 static  int              man_ptext(struct man *, int, char *, int);
  61 static  int              man_pmacro(struct man *, int, char *, int);
  62 static  void             man_free1(struct man *);
  63 static  void             man_alloc1(struct man *);
  64 static  int              man_descope(struct man *, int, int);
  65
  66
  67 const struct man_node *
  68 man_node(const struct man *man)
  69 {
  70
  71         return(man->first);
  72 }
  73
  74 const struct man_meta *
  75 man_meta(const struct man *man)
  76 {
  77
  78         return(&man->meta);
  79 }
  80
  81 void
  82 man_reset(struct man *man)
  83 {
  84
  85         man_free1(man);
  86         man_alloc1(man);
  87 }
  88
  89 void
  90 man_free(struct man *man)
  91 {
  92
  93         man_free1(man);
  94         free(man);
  95 }
  96
  97 struct man *
  98 man_alloc(struct roff *roff, struct mparse *parse, int quick)
  99 {
 100         struct man      *p;
 101
 102         p = mandoc_calloc(1, sizeof(struct man));
 103
 104         man_hash_init();
 105         p->parse = parse;
 106         p->quick = quick;
 107         p->roff = roff;
 108
 109         man_alloc1(p);
 110         return(p);
 111 }
 112
 113 int
 114 man_endparse(struct man *man)
 115 {
 116
 117         return(man_macroend(man));
 118 }
 119
 120 int
 121 man_parseln(struct man *man, int ln, char *buf, int offs)
 122 {
 123
 124         man->flags |= MAN_NEWLINE;
 125
 126         return (roff_getcontrol(man->roff, buf, &offs) ?
 127             man_pmacro(man, ln, buf, offs) :
 128             man_ptext(man, ln, buf, offs));
 129 }
 130
 131 static void
 132 man_free1(struct man *man)
 133 {
 134
 135         if (man->first)
 136                 man_node_delete(man, man->first);
 137         if (man->meta.title)
 138                 free(man->meta.title);
 139         if (man->meta.source)
 140                 free(man->meta.source);
 141         if (man->meta.date)
 142                 free(man->meta.date);
 143         if (man->meta.vol)
 144                 free(man->meta.vol);
 145         if (man->meta.msec)
 146                 free(man->meta.msec);
 147 }
 148
 149 static void
 150 man_alloc1(struct man *man)
 151 {
 152
 153         memset(&man->meta, 0, sizeof(struct man_meta));
 154         man->flags = 0;
 155         man->last = mandoc_calloc(1, sizeof(struct man_node));
 156         man->first = man->last;
 157         man->last->type = MAN_ROOT;
 158         man->last->tok = MAN_MAX;
 159         man->next = MAN_NEXT_CHILD;
 160 }
 161
 162
 163 static int
 164 man_node_append(struct man *man, struct man_node *p)
 165 {
 166
 167         assert(man->last);
 168         assert(man->first);
 169         assert(MAN_ROOT != p->type);
 170
 171         switch (man->next) {
 172         case MAN_NEXT_SIBLING:
 173                 man->last->next = p;
 174                 p->prev = man->last;
 175                 p->parent = man->last->parent;
 176                 break;
 177         case MAN_NEXT_CHILD:
 178                 man->last->child = p;
 179                 p->parent = man->last;
 180                 break;
 181         default:
 182                 abort();
 183                 /* NOTREACHED */
 184         }
 185
 186         assert(p->parent);
 187         p->parent->nchild++;
 188
 189         switch (p->type) {
 190         case MAN_BLOCK:
 191                 if (p->tok == MAN_SH || p->tok == MAN_SS)
 192                         man->flags &= ~MAN_LITERAL;
 193                 break;
 194         case MAN_HEAD:
 195                 assert(MAN_BLOCK == p->parent->type);
 196                 p->parent->head = p;
 197                 break;
 198         case MAN_TAIL:
 199                 assert(MAN_BLOCK == p->parent->type);
 200                 p->parent->tail = p;
 201                 break;
 202         case MAN_BODY:
 203                 assert(MAN_BLOCK == p->parent->type);
 204                 p->parent->body = p;
 205                 break;
 206         default:
 207                 break;
 208         }
 209
 210         man->last = p;
 211
 212         switch (p->type) {
 213         case MAN_TBL:
 214                 /* FALLTHROUGH */
 215         case MAN_TEXT:
 216                 if ( ! man_valid_post(man))
 217                         return(0);
 218                 break;
 219         default:
 220                 break;
 221         }
 222
 223         return(1);
 224 }
 225
 226 static struct man_node *
 227 man_node_alloc(struct man *man, int line, int pos,
 228                 enum man_type type, enum mant tok)
 229 {
 230         struct man_node *p;
 231
 232         p = mandoc_calloc(1, sizeof(struct man_node));
 233         p->line = line;
 234         p->pos = pos;
 235         p->type = type;
 236         p->tok = tok;
 237
 238         if (MAN_NEWLINE & man->flags)
 239                 p->flags |= MAN_LINE;
 240         man->flags &= ~MAN_NEWLINE;
 241         return(p);
 242 }
 243
 244 int
 245 man_elem_alloc(struct man *man, int line, int pos, enum mant tok)
 246 {
 247         struct man_node *p;
 248
 249         p = man_node_alloc(man, line, pos, MAN_ELEM, tok);
 250         if ( ! man_node_append(man, p))
 251                 return(0);
 252         man->next = MAN_NEXT_CHILD;
 253         return(1);
 254 }
 255
 256 int
 257 man_tail_alloc(struct man *man, int line, int pos, enum mant tok)
 258 {
 259         struct man_node *p;
 260
 261         p = man_node_alloc(man, line, pos, MAN_TAIL, tok);
 262         if ( ! man_node_append(man, p))
 263                 return(0);
 264         man->next = MAN_NEXT_CHILD;
 265         return(1);
 266 }
 267
 268 int
 269 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
 270 {
 271         struct man_node *p;
 272
 273         p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
 274         if ( ! man_node_append(man, p))
 275                 return(0);
 276         man->next = MAN_NEXT_CHILD;
 277         return(1);
 278 }
 279
 280 int
 281 man_body_alloc(struct man *man, int line, int pos, enum mant tok)
 282 {
 283         struct man_node *p;
 284
 285         p = man_node_alloc(man, line, pos, MAN_BODY, tok);
 286         if ( ! man_node_append(man, p))
 287                 return(0);
 288         man->next = MAN_NEXT_CHILD;
 289         return(1);
 290 }
 291
 292 int
 293 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
 294 {
 295         struct man_node *p;
 296
 297         p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
 298         if ( ! man_node_append(man, p))
 299                 return(0);
 300         man->next = MAN_NEXT_CHILD;
 301         return(1);
 302 }
 303
 304 int
 305 man_word_alloc(struct man *man, int line, int pos, const char *word)
 306 {
 307         struct man_node *n;
 308
 309         n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
 310         n->string = roff_strdup(man->roff, word);
 311
 312         if ( ! man_node_append(man, n))
 313                 return(0);
 314
 315         man->next = MAN_NEXT_SIBLING;
 316         return(1);
 317 }
 318
 319 /*
 320  * Free all of the resources held by a node.  This does NOT unlink a
 321  * node from its context; for that, see man_node_unlink().
 322  */
 323 static void
 324 man_node_free(struct man_node *p)
 325 {
 326
 327         if (p->string)
 328                 free(p->string);
 329         free(p);
 330 }
 331
 332 void
 333 man_node_delete(struct man *man, struct man_node *p)
 334 {
 335
 336         while (p->child)
 337                 man_node_delete(man, p->child);
 338
 339         man_node_unlink(man, p);
 340         man_node_free(p);
 341 }
 342
 343 int
 344 man_addeqn(struct man *man, const struct eqn *ep)
 345 {
 346         struct man_node *n;
 347
 348         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 349         n->eqn = ep;
 350
 351         if ( ! man_node_append(man, n))
 352                 return(0);
 353
 354         man->next = MAN_NEXT_SIBLING;
 355         return(man_descope(man, ep->ln, ep->pos));
 356 }
 357
 358 int
 359 man_addspan(struct man *man, const struct tbl_span *sp)
 360 {
 361         struct man_node *n;
 362
 363         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 364         n->span = sp;
 365
 366         if ( ! man_node_append(man, n))
 367                 return(0);
 368
 369         man->next = MAN_NEXT_SIBLING;
 370         return(man_descope(man, sp->line, 0));
 371 }
 372
 373 static int
 374 man_descope(struct man *man, int line, int offs)
 375 {
 376         /*
 377          * Co-ordinate what happens with having a next-line scope open:
 378          * first close out the element scope (if applicable), then close
 379          * out the block scope (also if applicable).
 380          */
 381
 382         if (MAN_ELINE & man->flags) {
 383                 man->flags &= ~MAN_ELINE;
 384                 if ( ! man_unscope(man, man->last->parent))
 385                         return(0);
 386         }
 387
 388         if ( ! (MAN_BLINE & man->flags))
 389                 return(1);
 390         man->flags &= ~MAN_BLINE;
 391
 392         if ( ! man_unscope(man, man->last->parent))
 393                 return(0);
 394         return(man_body_alloc(man, line, offs, man->last->tok));
 395 }
 396
 397 static int
 398 man_ptext(struct man *man, int line, char *buf, int offs)
 399 {
 400         int              i;
 401
 402         /* Literal free-form text whitespace is preserved. */
 403
 404         if (MAN_LITERAL & man->flags) {
 405                 if ( ! man_word_alloc(man, line, offs, buf + offs))
 406                         return(0);
 407                 return(man_descope(man, line, offs));
 408         }
 409
 410         for (i = offs; ' ' == buf[i]; i++)
 411                 /* Skip leading whitespace. */ ;
 412
 413         /*
 414          * Blank lines are ignored right after headings
 415          * but add a single vertical space elsewhere.
 416          */
 417
 418         if ('\0' == buf[i]) {
 419                 /* Allocate a blank entry. */
 420                 if (MAN_SH != man->last->tok &&
 421                     MAN_SS != man->last->tok) {
 422                         if ( ! man_elem_alloc(man, line, offs, MAN_sp))
 423                                 return(0);
 424                         man->next = MAN_NEXT_SIBLING;
 425                 }
 426                 return(1);
 427         }
 428
 429         /*
 430          * Warn if the last un-escaped character is whitespace. Then
 431          * strip away the remaining spaces (tabs stay!).
 432          */
 433
 434         i = (int)strlen(buf);
 435         assert(i);
 436
 437         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 438                 if (i > 1 && '\\' != buf[i - 2])
 439                         mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 440                             line, i - 1, NULL);
 441
 442                 for (--i; i && ' ' == buf[i]; i--)
 443                         /* Spin back to non-space. */ ;
 444
 445                 /* Jump ahead of escaped whitespace. */
 446                 i += '\\' == buf[i] ? 2 : 1;
 447
 448                 buf[i] = '\0';
 449         }
 450
 451         if ( ! man_word_alloc(man, line, offs, buf + offs))
 452                 return(0);
 453
 454         /*
 455          * End-of-sentence check.  If the last character is an unescaped
 456          * EOS character, then flag the node as being the end of a
 457          * sentence.  The front-end will know how to interpret this.
 458          */
 459
 460         assert(i);
 461         if (mandoc_eos(buf, (size_t)i))
 462                 man->last->flags |= MAN_EOS;
 463
 464         return(man_descope(man, line, offs));
 465 }
 466
 467 static int
 468 man_pmacro(struct man *man, int ln, char *buf, int offs)
 469 {
 470         char             mac[5];
 471         struct man_node *n;
 472         enum mant        tok;
 473         int              i, ppos;
 474         int              bline;
 475
 476         if ('"' == buf[offs]) {
 477                 mandoc_msg(MANDOCERR_COMMENT_BAD, man->parse,
 478                     ln, offs, NULL);
 479                 return(1);
 480         } else if ('\0' == buf[offs])
 481                 return(1);
 482
 483         ppos = offs;
 484
 485         /*
 486          * Copy the first word into a nil-terminated buffer.
 487          * Stop copying when a tab, space, or eoln is encountered.
 488          */
 489
 490         i = 0;
 491         while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
 492             '\t' != buf[offs])
 493                 mac[i++] = buf[offs++];
 494
 495         mac[i] = '\0';
 496
 497         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 498
 499         if (MAN_MAX == tok) {
 500                 mandoc_msg(MANDOCERR_MACRO, man->parse,
 501                     ln, ppos, buf + ppos - 1);
 502                 return(1);
 503         }
 504
 505         /* The macro is sane.  Jump to the next word. */
 506
 507         while (buf[offs] && ' ' == buf[offs])
 508                 offs++;
 509
 510         /*
 511          * Trailing whitespace.  Note that tabs are allowed to be passed
 512          * into the parser as "text", so we only warn about spaces here.
 513          */
 514
 515         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 516                 mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 517                     ln, offs - 1, NULL);
 518
 519         /*
 520          * Remove prior ELINE macro, as it's being clobbered by a new
 521          * macro.  Note that NSCOPED macros do not close out ELINE
 522          * macros---they don't print text---so we let those slip by.
 523          */
 524
 525         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 526                         man->flags & MAN_ELINE) {
 527                 n = man->last;
 528                 assert(MAN_TEXT != n->type);
 529
 530                 /* Remove repeated NSCOPED macros causing ELINE. */
 531
 532                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 533                         n = n->parent;
 534
 535                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 536                     n->pos, "%s breaks %s", man_macronames[tok],
 537                     man_macronames[n->tok]);
 538
 539                 man_node_delete(man, n);
 540                 man->flags &= ~MAN_ELINE;
 541         }
 542
 543         /*
 544          * Remove prior BLINE macro that is being clobbered.
 545          */
 546         if ((man->flags & MAN_BLINE) &&
 547             (MAN_BSCOPE & man_macros[tok].flags)) {
 548                 n = man->last;
 549
 550                 /* Might be a text node like 8 in
 551                  * .TP 8
 552                  * .SH foo
 553                  */
 554                 if (MAN_TEXT == n->type)
 555                         n = n->parent;
 556
 557                 /* Remove element that didn't end BLINE, if any. */
 558                 if ( ! (MAN_BSCOPE & man_macros[n->tok].flags))
 559                         n = n->parent;
 560
 561                 assert(MAN_HEAD == n->type);
 562                 n = n->parent;
 563                 assert(MAN_BLOCK == n->type);
 564                 assert(MAN_SCOPED & man_macros[n->tok].flags);
 565
 566                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 567                     n->pos, "%s breaks %s", man_macronames[tok],
 568                     man_macronames[n->tok]);
 569
 570                 man_node_delete(man, n);
 571                 man->flags &= ~MAN_BLINE;
 572         }
 573
 574         /* Remember whether we are in next-line scope for a block head. */
 575
 576         bline = man->flags & MAN_BLINE;
 577
 578         /* Call to handler... */
 579
 580         assert(man_macros[tok].fp);
 581         if ( ! (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf))
 582                 return(0);
 583
 584         /* In quick mode (for mandocdb), abort after the NAME section. */
 585
 586         if (man->quick && MAN_SH == tok) {
 587                 n = man->last;
 588                 if (MAN_BODY == n->type &&
 589                     strcmp(n->prev->child->string, "NAME"))
 590                         return(2);
 591         }
 592
 593         /*
 594          * If we are in a next-line scope for a block head,
 595          * close it out now and switch to the body,
 596          * unless the next-line scope is allowed to continue.
 597          */
 598
 599         if ( ! bline || man->flags & MAN_ELINE ||
 600             man_macros[tok].flags & MAN_NSCOPED)
 601                 return(1);
 602
 603         assert(MAN_BLINE & man->flags);
 604         man->flags &= ~MAN_BLINE;
 605
 606         if ( ! man_unscope(man, man->last->parent))
 607                 return(0);
 608         return(man_body_alloc(man, ln, ppos, man->last->tok));
 609 }
 610
 611 /*
 612  * Unlink a node from its context.  If "man" is provided, the last parse
 613  * point will also be adjusted accordingly.
 614  */
 615 static void
 616 man_node_unlink(struct man *man, struct man_node *n)
 617 {
 618
 619         /* Adjust siblings. */
 620
 621         if (n->prev)
 622                 n->prev->next = n->next;
 623         if (n->next)
 624                 n->next->prev = n->prev;
 625
 626         /* Adjust parent. */
 627
 628         if (n->parent) {
 629                 n->parent->nchild--;
 630                 if (n->parent->child == n)
 631                         n->parent->child = n->prev ? n->prev : n->next;
 632         }
 633
 634         /* Adjust parse point, if applicable. */
 635
 636         if (man && man->last == n) {
 637                 /*XXX: this can occur when bailing from validation. */
 638                 /*assert(NULL == n->next);*/
 639                 if (n->prev) {
 640                         man->last = n->prev;
 641                         man->next = MAN_NEXT_SIBLING;
 642                 } else {
 643                         man->last = n->parent;
 644                         man->next = MAN_NEXT_CHILD;
 645                 }
 646         }
 647
 648         if (man && man->first == n)
 649                 man->first = NULL;
 650 }
 651
 652 const struct mparse *
 653 man_mparse(const struct man *man)
 654 {
 655
 656         assert(man && man->parse);
 657         return(man->parse);
 658 }
 659
 660 void
 661 man_deroff(char **dest, const struct man_node *n)
 662 {
 663         char    *cp;
 664         size_t   sz;
 665
 666         if (MAN_TEXT != n->type) {
 667                 for (n = n->child; n; n = n->next)
 668                         man_deroff(dest, n);
 669                 return;
 670         }
 671
 672         /* Skip leading whitespace and escape sequences. */
 673
 674         cp = n->string;
 675         while ('\0' != *cp) {
 676                 if ('\\' == *cp) {
 677                         cp++;
 678                         mandoc_escape((const char **)&cp, NULL, NULL);
 679                 } else if (isspace((unsigned char)*cp))
 680                         cp++;
 681                 else
 682                         break;
 683         }
 684
 685         /* Skip trailing whitespace. */
 686
 687         for (sz = strlen(cp); sz; sz--)
 688                 if (0 == isspace((unsigned char)cp[sz-1]))
 689                         break;
 690
 691         /* Skip empty strings. */
 692
 693         if (0 == sz)
 694                 return;
 695
 696         if (NULL == *dest) {
 697                 *dest = mandoc_strndup(cp, sz);
 698                 return;
 699         }
 700
 701         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 702         free(*dest);
 703         *dest = cp;
 704 }