man.c

   1 /*      $Id: man.c,v 1.138 2014/08/10 23:54:41 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         man->flags |= MAN_NEWLINE;
 123
 124         return (roff_getcontrol(man->roff, buf, &offs) ?
 125             man_pmacro(man, ln, buf, offs) :
 126             man_ptext(man, ln, buf, offs));
 127 }
 128
 129 static void
 130 man_free1(struct man *man)
 131 {
 132
 133         if (man->first)
 134                 man_node_delete(man, man->first);
 135         if (man->meta.title)
 136                 free(man->meta.title);
 137         if (man->meta.source)
 138                 free(man->meta.source);
 139         if (man->meta.date)
 140                 free(man->meta.date);
 141         if (man->meta.vol)
 142                 free(man->meta.vol);
 143         if (man->meta.msec)
 144                 free(man->meta.msec);
 145 }
 146
 147 static void
 148 man_alloc1(struct man *man)
 149 {
 150
 151         memset(&man->meta, 0, sizeof(struct man_meta));
 152         man->flags = 0;
 153         man->last = mandoc_calloc(1, sizeof(struct man_node));
 154         man->first = man->last;
 155         man->last->type = MAN_ROOT;
 156         man->last->tok = MAN_MAX;
 157         man->next = MAN_NEXT_CHILD;
 158 }
 159
 160
 161 static int
 162 man_node_append(struct man *man, struct man_node *p)
 163 {
 164
 165         assert(man->last);
 166         assert(man->first);
 167         assert(MAN_ROOT != p->type);
 168
 169         switch (man->next) {
 170         case MAN_NEXT_SIBLING:
 171                 man->last->next = p;
 172                 p->prev = man->last;
 173                 p->parent = man->last->parent;
 174                 break;
 175         case MAN_NEXT_CHILD:
 176                 man->last->child = p;
 177                 p->parent = man->last;
 178                 break;
 179         default:
 180                 abort();
 181                 /* NOTREACHED */
 182         }
 183
 184         assert(p->parent);
 185         p->parent->nchild++;
 186
 187         switch (p->type) {
 188         case MAN_BLOCK:
 189                 if (p->tok == MAN_SH || p->tok == MAN_SS)
 190                         man->flags &= ~MAN_LITERAL;
 191                 break;
 192         case MAN_HEAD:
 193                 assert(MAN_BLOCK == p->parent->type);
 194                 p->parent->head = p;
 195                 break;
 196         case MAN_TAIL:
 197                 assert(MAN_BLOCK == p->parent->type);
 198                 p->parent->tail = p;
 199                 break;
 200         case MAN_BODY:
 201                 assert(MAN_BLOCK == p->parent->type);
 202                 p->parent->body = p;
 203                 break;
 204         default:
 205                 break;
 206         }
 207
 208         man->last = p;
 209
 210         switch (p->type) {
 211         case MAN_TBL:
 212                 /* FALLTHROUGH */
 213         case MAN_TEXT:
 214                 if ( ! man_valid_post(man))
 215                         return(0);
 216                 break;
 217         default:
 218                 break;
 219         }
 220
 221         return(1);
 222 }
 223
 224 static struct man_node *
 225 man_node_alloc(struct man *man, int line, int pos,
 226                 enum man_type type, enum mant tok)
 227 {
 228         struct man_node *p;
 229
 230         p = mandoc_calloc(1, sizeof(struct man_node));
 231         p->line = line;
 232         p->pos = pos;
 233         p->type = type;
 234         p->tok = tok;
 235
 236         if (MAN_NEWLINE & man->flags)
 237                 p->flags |= MAN_LINE;
 238         man->flags &= ~MAN_NEWLINE;
 239         return(p);
 240 }
 241
 242 int
 243 man_elem_alloc(struct man *man, int line, int pos, enum mant tok)
 244 {
 245         struct man_node *p;
 246
 247         p = man_node_alloc(man, line, pos, MAN_ELEM, tok);
 248         if ( ! man_node_append(man, p))
 249                 return(0);
 250         man->next = MAN_NEXT_CHILD;
 251         return(1);
 252 }
 253
 254 int
 255 man_tail_alloc(struct man *man, int line, int pos, enum mant tok)
 256 {
 257         struct man_node *p;
 258
 259         p = man_node_alloc(man, line, pos, MAN_TAIL, tok);
 260         if ( ! man_node_append(man, p))
 261                 return(0);
 262         man->next = MAN_NEXT_CHILD;
 263         return(1);
 264 }
 265
 266 int
 267 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
 268 {
 269         struct man_node *p;
 270
 271         p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
 272         if ( ! man_node_append(man, p))
 273                 return(0);
 274         man->next = MAN_NEXT_CHILD;
 275         return(1);
 276 }
 277
 278 int
 279 man_body_alloc(struct man *man, int line, int pos, enum mant tok)
 280 {
 281         struct man_node *p;
 282
 283         p = man_node_alloc(man, line, pos, MAN_BODY, tok);
 284         if ( ! man_node_append(man, p))
 285                 return(0);
 286         man->next = MAN_NEXT_CHILD;
 287         return(1);
 288 }
 289
 290 int
 291 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
 292 {
 293         struct man_node *p;
 294
 295         p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
 296         if ( ! man_node_append(man, p))
 297                 return(0);
 298         man->next = MAN_NEXT_CHILD;
 299         return(1);
 300 }
 301
 302 int
 303 man_word_alloc(struct man *man, int line, int pos, const char *word)
 304 {
 305         struct man_node *n;
 306
 307         n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
 308         n->string = roff_strdup(man->roff, word);
 309
 310         if ( ! man_node_append(man, n))
 311                 return(0);
 312
 313         man->next = MAN_NEXT_SIBLING;
 314         return(1);
 315 }
 316
 317 /*
 318  * Free all of the resources held by a node.  This does NOT unlink a
 319  * node from its context; for that, see man_node_unlink().
 320  */
 321 static void
 322 man_node_free(struct man_node *p)
 323 {
 324
 325         if (p->string)
 326                 free(p->string);
 327         free(p);
 328 }
 329
 330 void
 331 man_node_delete(struct man *man, struct man_node *p)
 332 {
 333
 334         while (p->child)
 335                 man_node_delete(man, p->child);
 336
 337         man_node_unlink(man, p);
 338         man_node_free(p);
 339 }
 340
 341 int
 342 man_addeqn(struct man *man, const struct eqn *ep)
 343 {
 344         struct man_node *n;
 345
 346         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 347         n->eqn = ep;
 348
 349         if ( ! man_node_append(man, n))
 350                 return(0);
 351
 352         man->next = MAN_NEXT_SIBLING;
 353         return(man_descope(man, ep->ln, ep->pos));
 354 }
 355
 356 int
 357 man_addspan(struct man *man, const struct tbl_span *sp)
 358 {
 359         struct man_node *n;
 360
 361         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 362         n->span = sp;
 363
 364         if ( ! man_node_append(man, n))
 365                 return(0);
 366
 367         man->next = MAN_NEXT_SIBLING;
 368         return(man_descope(man, sp->line, 0));
 369 }
 370
 371 static int
 372 man_descope(struct man *man, int line, int offs)
 373 {
 374         /*
 375          * Co-ordinate what happens with having a next-line scope open:
 376          * first close out the element scope (if applicable), then close
 377          * out the block scope (also if applicable).
 378          */
 379
 380         if (MAN_ELINE & man->flags) {
 381                 man->flags &= ~MAN_ELINE;
 382                 if ( ! man_unscope(man, man->last->parent))
 383                         return(0);
 384         }
 385
 386         if ( ! (MAN_BLINE & man->flags))
 387                 return(1);
 388         man->flags &= ~MAN_BLINE;
 389
 390         if ( ! man_unscope(man, man->last->parent))
 391                 return(0);
 392         return(man_body_alloc(man, line, offs, man->last->tok));
 393 }
 394
 395 static int
 396 man_ptext(struct man *man, int line, char *buf, int offs)
 397 {
 398         int              i;
 399
 400         /* Literal free-form text whitespace is preserved. */
 401
 402         if (MAN_LITERAL & man->flags) {
 403                 if ( ! man_word_alloc(man, line, offs, buf + offs))
 404                         return(0);
 405                 return(man_descope(man, line, offs));
 406         }
 407
 408         for (i = offs; ' ' == buf[i]; i++)
 409                 /* Skip leading whitespace. */ ;
 410
 411         /*
 412          * Blank lines are ignored right after headings
 413          * but add a single vertical space elsewhere.
 414          */
 415
 416         if ('\0' == buf[i]) {
 417                 /* Allocate a blank entry. */
 418                 if (MAN_SH != man->last->tok &&
 419                     MAN_SS != man->last->tok) {
 420                         if ( ! man_elem_alloc(man, line, offs, MAN_sp))
 421                                 return(0);
 422                         man->next = MAN_NEXT_SIBLING;
 423                 }
 424                 return(1);
 425         }
 426
 427         /*
 428          * Warn if the last un-escaped character is whitespace. Then
 429          * strip away the remaining spaces (tabs stay!).
 430          */
 431
 432         i = (int)strlen(buf);
 433         assert(i);
 434
 435         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 436                 if (i > 1 && '\\' != buf[i - 2])
 437                         mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 438                             line, i - 1, NULL);
 439
 440                 for (--i; i && ' ' == buf[i]; i--)
 441                         /* Spin back to non-space. */ ;
 442
 443                 /* Jump ahead of escaped whitespace. */
 444                 i += '\\' == buf[i] ? 2 : 1;
 445
 446                 buf[i] = '\0';
 447         }
 448
 449         if ( ! man_word_alloc(man, line, offs, buf + offs))
 450                 return(0);
 451
 452         /*
 453          * End-of-sentence check.  If the last character is an unescaped
 454          * EOS character, then flag the node as being the end of a
 455          * sentence.  The front-end will know how to interpret this.
 456          */
 457
 458         assert(i);
 459         if (mandoc_eos(buf, (size_t)i))
 460                 man->last->flags |= MAN_EOS;
 461
 462         return(man_descope(man, line, offs));
 463 }
 464
 465 static int
 466 man_pmacro(struct man *man, int ln, char *buf, int offs)
 467 {
 468         char             mac[5];
 469         struct man_node *n;
 470         enum mant        tok;
 471         int              i, ppos;
 472         int              bline;
 473
 474         if ('"' == buf[offs]) {
 475                 mandoc_msg(MANDOCERR_COMMENT_BAD, man->parse,
 476                     ln, offs, NULL);
 477                 return(1);
 478         } else if ('\0' == buf[offs])
 479                 return(1);
 480
 481         ppos = offs;
 482
 483         /*
 484          * Copy the first word into a nil-terminated buffer.
 485          * Stop copying when a tab, space, or eoln is encountered.
 486          */
 487
 488         i = 0;
 489         while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
 490             '\t' != buf[offs])
 491                 mac[i++] = buf[offs++];
 492
 493         mac[i] = '\0';
 494
 495         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 496
 497         if (MAN_MAX == tok) {
 498                 mandoc_msg(MANDOCERR_MACRO, man->parse,
 499                     ln, ppos, buf + ppos - 1);
 500                 return(1);
 501         }
 502
 503         /* The macro is sane.  Jump to the next word. */
 504
 505         while (buf[offs] && ' ' == buf[offs])
 506                 offs++;
 507
 508         /*
 509          * Trailing whitespace.  Note that tabs are allowed to be passed
 510          * into the parser as "text", so we only warn about spaces here.
 511          */
 512
 513         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 514                 mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 515                     ln, offs - 1, NULL);
 516
 517         /*
 518          * Remove prior ELINE macro, as it's being clobbered by a new
 519          * macro.  Note that NSCOPED macros do not close out ELINE
 520          * macros---they don't print text---so we let those slip by.
 521          */
 522
 523         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 524                         man->flags & MAN_ELINE) {
 525                 n = man->last;
 526                 assert(MAN_TEXT != n->type);
 527
 528                 /* Remove repeated NSCOPED macros causing ELINE. */
 529
 530                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 531                         n = n->parent;
 532
 533                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 534                     n->pos, "%s breaks %s", man_macronames[tok],
 535                     man_macronames[n->tok]);
 536
 537                 man_node_delete(man, n);
 538                 man->flags &= ~MAN_ELINE;
 539         }
 540
 541         /*
 542          * Remove prior BLINE macro that is being clobbered.
 543          */
 544         if ((man->flags & MAN_BLINE) &&
 545             (MAN_BSCOPE & man_macros[tok].flags)) {
 546                 n = man->last;
 547
 548                 /* Might be a text node like 8 in
 549                  * .TP 8
 550                  * .SH foo
 551                  */
 552                 if (MAN_TEXT == n->type)
 553                         n = n->parent;
 554
 555                 /* Remove element that didn't end BLINE, if any. */
 556                 if ( ! (MAN_BSCOPE & man_macros[n->tok].flags))
 557                         n = n->parent;
 558
 559                 assert(MAN_HEAD == n->type);
 560                 n = n->parent;
 561                 assert(MAN_BLOCK == n->type);
 562                 assert(MAN_SCOPED & man_macros[n->tok].flags);
 563
 564                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 565                     n->pos, "%s breaks %s", man_macronames[tok],
 566                     man_macronames[n->tok]);
 567
 568                 man_node_delete(man, n);
 569                 man->flags &= ~MAN_BLINE;
 570         }
 571
 572         /* Remember whether we are in next-line scope for a block head. */
 573
 574         bline = man->flags & MAN_BLINE;
 575
 576         /* Call to handler... */
 577
 578         assert(man_macros[tok].fp);
 579         if ( ! (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf))
 580                 return(0);
 581
 582         /* In quick mode (for mandocdb), abort after the NAME section. */
 583
 584         if (man->quick && MAN_SH == tok) {
 585                 n = man->last;
 586                 if (MAN_BODY == n->type &&
 587                     strcmp(n->prev->child->string, "NAME"))
 588                         return(2);
 589         }
 590
 591         /*
 592          * If we are in a next-line scope for a block head,
 593          * close it out now and switch to the body,
 594          * unless the next-line scope is allowed to continue.
 595          */
 596
 597         if ( ! bline || man->flags & MAN_ELINE ||
 598             man_macros[tok].flags & MAN_NSCOPED)
 599                 return(1);
 600
 601         assert(MAN_BLINE & man->flags);
 602         man->flags &= ~MAN_BLINE;
 603
 604         if ( ! man_unscope(man, man->last->parent))
 605                 return(0);
 606         return(man_body_alloc(man, ln, ppos, man->last->tok));
 607 }
 608
 609 /*
 610  * Unlink a node from its context.  If "man" is provided, the last parse
 611  * point will also be adjusted accordingly.
 612  */
 613 static void
 614 man_node_unlink(struct man *man, struct man_node *n)
 615 {
 616
 617         /* Adjust siblings. */
 618
 619         if (n->prev)
 620                 n->prev->next = n->next;
 621         if (n->next)
 622                 n->next->prev = n->prev;
 623
 624         /* Adjust parent. */
 625
 626         if (n->parent) {
 627                 n->parent->nchild--;
 628                 if (n->parent->child == n)
 629                         n->parent->child = n->prev ? n->prev : n->next;
 630         }
 631
 632         /* Adjust parse point, if applicable. */
 633
 634         if (man && man->last == n) {
 635                 /*XXX: this can occur when bailing from validation. */
 636                 /*assert(NULL == n->next);*/
 637                 if (n->prev) {
 638                         man->last = n->prev;
 639                         man->next = MAN_NEXT_SIBLING;
 640                 } else {
 641                         man->last = n->parent;
 642                         man->next = MAN_NEXT_CHILD;
 643                 }
 644         }
 645
 646         if (man && man->first == n)
 647                 man->first = NULL;
 648 }
 649
 650 const struct mparse *
 651 man_mparse(const struct man *man)
 652 {
 653
 654         assert(man && man->parse);
 655         return(man->parse);
 656 }
 657
 658 void
 659 man_deroff(char **dest, const struct man_node *n)
 660 {
 661         char    *cp;
 662         size_t   sz;
 663
 664         if (MAN_TEXT != n->type) {
 665                 for (n = n->child; n; n = n->next)
 666                         man_deroff(dest, n);
 667                 return;
 668         }
 669
 670         /* Skip leading whitespace and escape sequences. */
 671
 672         cp = n->string;
 673         while ('\0' != *cp) {
 674                 if ('\\' == *cp) {
 675                         cp++;
 676                         mandoc_escape((const char **)&cp, NULL, NULL);
 677                 } else if (isspace((unsigned char)*cp))
 678                         cp++;
 679                 else
 680                         break;
 681         }
 682
 683         /* Skip trailing whitespace. */
 684
 685         for (sz = strlen(cp); sz; sz--)
 686                 if (0 == isspace((unsigned char)cp[sz-1]))
 687                         break;
 688
 689         /* Skip empty strings. */
 690
 691         if (0 == sz)
 692                 return;
 693
 694         if (NULL == *dest) {
 695                 *dest = mandoc_strndup(cp, sz);
 696                 return;
 697         }
 698
 699         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 700         free(*dest);
 701         *dest = cp;
 702 }