contrib/mdocml/man.c

   1 /*      $Id: man.c,v 1.119 2012/11/17 00:26:33 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  *
   5  * Permission to use, copy, modify, and distribute this software for any
   6  * purpose with or without fee is hereby granted, provided that the above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  16  */
  17 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <sys/types.h>
  22
  23 #include <assert.h>
  24 #include <stdarg.h>
  25 #include <stdlib.h>
  26 #include <stdio.h>
  27 #include <string.h>
  28
  29 #include "man.h"
  30 #include "mandoc.h"
  31 #include "libman.h"
  32 #include "libmandoc.h"
  33
  34 const   char *const __man_macronames[MAN_MAX] = {
  35         "br",           "TH",           "SH",           "SS",
  36         "TP",           "LP",           "PP",           "P",
  37         "IP",           "HP",           "SM",           "SB",
  38         "BI",           "IB",           "BR",           "RB",
  39         "R",            "B",            "I",            "IR",
  40         "RI",           "na",           "sp",           "nf",
  41         "fi",           "RE",           "RS",           "DT",
  42         "UC",           "PD",           "AT",           "in",
  43         "ft",           "OP",           "EX",           "EE"
  44         };
  45
  46 const   char * const *man_macronames = __man_macronames;
  47
  48 static  struct man_node *man_node_alloc(struct man *, int, int,
  49                                 enum man_type, enum mant);
  50 static  int              man_node_append(struct man *,
  51                                 struct man_node *);
  52 static  void             man_node_free(struct man_node *);
  53 static  void             man_node_unlink(struct man *,
  54                                 struct man_node *);
  55 static  int              man_ptext(struct man *, int, char *, int);
  56 static  int              man_pmacro(struct man *, int, char *, int);
  57 static  void             man_free1(struct man *);
  58 static  void             man_alloc1(struct man *);
  59 static  int              man_descope(struct man *, int, int);
  60
  61
  62 const struct man_node *
  63 man_node(const struct man *man)
  64 {
  65
  66         assert( ! (MAN_HALT & man->flags));
  67         return(man->first);
  68 }
  69
  70
  71 const struct man_meta *
  72 man_meta(const struct man *man)
  73 {
  74
  75         assert( ! (MAN_HALT & man->flags));
  76         return(&man->meta);
  77 }
  78
  79
  80 void
  81 man_reset(struct man *man)
  82 {
  83
  84         man_free1(man);
  85         man_alloc1(man);
  86 }
  87
  88
  89 void
  90 man_free(struct man *man)
  91 {
  92
  93         man_free1(man);
  94         free(man);
  95 }
  96
  97
  98 struct man *
  99 man_alloc(struct roff *roff, struct mparse *parse)
 100 {
 101         struct man      *p;
 102
 103         p = mandoc_calloc(1, sizeof(struct man));
 104
 105         man_hash_init();
 106         p->parse = parse;
 107         p->roff = roff;
 108
 109         man_alloc1(p);
 110         return(p);
 111 }
 112
 113
 114 int
 115 man_endparse(struct man *man)
 116 {
 117
 118         assert( ! (MAN_HALT & man->flags));
 119         if (man_macroend(man))
 120                 return(1);
 121         man->flags |= MAN_HALT;
 122         return(0);
 123 }
 124
 125
 126 int
 127 man_parseln(struct man *man, int ln, char *buf, int offs)
 128 {
 129
 130         man->flags |= MAN_NEWLINE;
 131
 132         assert( ! (MAN_HALT & man->flags));
 133
 134         return (roff_getcontrol(man->roff, buf, &offs) ?
 135                         man_pmacro(man, ln, buf, offs) :
 136                         man_ptext(man, ln, buf, offs));
 137 }
 138
 139
 140 static void
 141 man_free1(struct man *man)
 142 {
 143
 144         if (man->first)
 145                 man_node_delete(man, man->first);
 146         if (man->meta.title)
 147                 free(man->meta.title);
 148         if (man->meta.source)
 149                 free(man->meta.source);
 150         if (man->meta.date)
 151                 free(man->meta.date);
 152         if (man->meta.vol)
 153                 free(man->meta.vol);
 154         if (man->meta.msec)
 155                 free(man->meta.msec);
 156 }
 157
 158
 159 static void
 160 man_alloc1(struct man *man)
 161 {
 162
 163         memset(&man->meta, 0, sizeof(struct man_meta));
 164         man->flags = 0;
 165         man->last = mandoc_calloc(1, sizeof(struct man_node));
 166         man->first = man->last;
 167         man->last->type = MAN_ROOT;
 168         man->last->tok = MAN_MAX;
 169         man->next = MAN_NEXT_CHILD;
 170 }
 171
 172
 173 static int
 174 man_node_append(struct man *man, struct man_node *p)
 175 {
 176
 177         assert(man->last);
 178         assert(man->first);
 179         assert(MAN_ROOT != p->type);
 180
 181         switch (man->next) {
 182         case (MAN_NEXT_SIBLING):
 183                 man->last->next = p;
 184                 p->prev = man->last;
 185                 p->parent = man->last->parent;
 186                 break;
 187         case (MAN_NEXT_CHILD):
 188                 man->last->child = p;
 189                 p->parent = man->last;
 190                 break;
 191         default:
 192                 abort();
 193                 /* NOTREACHED */
 194         }
 195
 196         assert(p->parent);
 197         p->parent->nchild++;
 198
 199         if ( ! man_valid_pre(man, p))
 200                 return(0);
 201
 202         switch (p->type) {
 203         case (MAN_HEAD):
 204                 assert(MAN_BLOCK == p->parent->type);
 205                 p->parent->head = p;
 206                 break;
 207         case (MAN_TAIL):
 208                 assert(MAN_BLOCK == p->parent->type);
 209                 p->parent->tail = p;
 210                 break;
 211         case (MAN_BODY):
 212                 assert(MAN_BLOCK == p->parent->type);
 213                 p->parent->body = p;
 214                 break;
 215         default:
 216                 break;
 217         }
 218
 219         man->last = p;
 220
 221         switch (p->type) {
 222         case (MAN_TBL):
 223                 /* FALLTHROUGH */
 224         case (MAN_TEXT):
 225                 if ( ! man_valid_post(man))
 226                         return(0);
 227                 break;
 228         default:
 229                 break;
 230         }
 231
 232         return(1);
 233 }
 234
 235
 236 static struct man_node *
 237 man_node_alloc(struct man *man, int line, int pos,
 238                 enum man_type type, enum mant tok)
 239 {
 240         struct man_node *p;
 241
 242         p = mandoc_calloc(1, sizeof(struct man_node));
 243         p->line = line;
 244         p->pos = pos;
 245         p->type = type;
 246         p->tok = tok;
 247
 248         if (MAN_NEWLINE & man->flags)
 249                 p->flags |= MAN_LINE;
 250         man->flags &= ~MAN_NEWLINE;
 251         return(p);
 252 }
 253
 254
 255 int
 256 man_elem_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_ELEM, tok);
 261         if ( ! man_node_append(man, p))
 262                 return(0);
 263         man->next = MAN_NEXT_CHILD;
 264         return(1);
 265 }
 266
 267
 268 int
 269 man_tail_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_TAIL, tok);
 274         if ( ! man_node_append(man, p))
 275                 return(0);
 276         man->next = MAN_NEXT_CHILD;
 277         return(1);
 278 }
 279
 280
 281 int
 282 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
 283 {
 284         struct man_node *p;
 285
 286         p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
 287         if ( ! man_node_append(man, p))
 288                 return(0);
 289         man->next = MAN_NEXT_CHILD;
 290         return(1);
 291 }
 292
 293
 294 int
 295 man_body_alloc(struct man *man, int line, int pos, enum mant tok)
 296 {
 297         struct man_node *p;
 298
 299         p = man_node_alloc(man, line, pos, MAN_BODY, tok);
 300         if ( ! man_node_append(man, p))
 301                 return(0);
 302         man->next = MAN_NEXT_CHILD;
 303         return(1);
 304 }
 305
 306
 307 int
 308 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
 309 {
 310         struct man_node *p;
 311
 312         p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
 313         if ( ! man_node_append(man, p))
 314                 return(0);
 315         man->next = MAN_NEXT_CHILD;
 316         return(1);
 317 }
 318
 319 int
 320 man_word_alloc(struct man *man, int line, int pos, const char *word)
 321 {
 322         struct man_node *n;
 323
 324         n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
 325         n->string = roff_strdup(man->roff, word);
 326
 327         if ( ! man_node_append(man, n))
 328                 return(0);
 329
 330         man->next = MAN_NEXT_SIBLING;
 331         return(1);
 332 }
 333
 334
 335 /*
 336  * Free all of the resources held by a node.  This does NOT unlink a
 337  * node from its context; for that, see man_node_unlink().
 338  */
 339 static void
 340 man_node_free(struct man_node *p)
 341 {
 342
 343         if (p->string)
 344                 free(p->string);
 345         free(p);
 346 }
 347
 348
 349 void
 350 man_node_delete(struct man *man, struct man_node *p)
 351 {
 352
 353         while (p->child)
 354                 man_node_delete(man, p->child);
 355
 356         man_node_unlink(man, p);
 357         man_node_free(p);
 358 }
 359
 360 int
 361 man_addeqn(struct man *man, const struct eqn *ep)
 362 {
 363         struct man_node *n;
 364
 365         assert( ! (MAN_HALT & man->flags));
 366
 367         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 368         n->eqn = ep;
 369
 370         if ( ! man_node_append(man, n))
 371                 return(0);
 372
 373         man->next = MAN_NEXT_SIBLING;
 374         return(man_descope(man, ep->ln, ep->pos));
 375 }
 376
 377 int
 378 man_addspan(struct man *man, const struct tbl_span *sp)
 379 {
 380         struct man_node *n;
 381
 382         assert( ! (MAN_HALT & man->flags));
 383
 384         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 385         n->span = sp;
 386
 387         if ( ! man_node_append(man, n))
 388                 return(0);
 389
 390         man->next = MAN_NEXT_SIBLING;
 391         return(man_descope(man, sp->line, 0));
 392 }
 393
 394 static int
 395 man_descope(struct man *man, int line, int offs)
 396 {
 397         /*
 398          * Co-ordinate what happens with having a next-line scope open:
 399          * first close out the element scope (if applicable), then close
 400          * out the block scope (also if applicable).
 401          */
 402
 403         if (MAN_ELINE & man->flags) {
 404                 man->flags &= ~MAN_ELINE;
 405                 if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 406                         return(0);
 407         }
 408
 409         if ( ! (MAN_BLINE & man->flags))
 410                 return(1);
 411         man->flags &= ~MAN_BLINE;
 412
 413         if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 414                 return(0);
 415         return(man_body_alloc(man, line, offs, man->last->tok));
 416 }
 417
 418 static int
 419 man_ptext(struct man *man, int line, char *buf, int offs)
 420 {
 421         int              i;
 422
 423         /* Literal free-form text whitespace is preserved. */
 424
 425         if (MAN_LITERAL & man->flags) {
 426                 if ( ! man_word_alloc(man, line, offs, buf + offs))
 427                         return(0);
 428                 return(man_descope(man, line, offs));
 429         }
 430
 431         /* Pump blank lines directly into the backend. */
 432
 433         for (i = offs; ' ' == buf[i]; i++)
 434                 /* Skip leading whitespace. */ ;
 435
 436         if ('\0' == buf[i]) {
 437                 /* Allocate a blank entry. */
 438                 if ( ! man_elem_alloc(man, line, offs, MAN_sp))
 439                         return(0);
 440                 man->next = MAN_NEXT_SIBLING;
 441                 return(1);
 442         }
 443
 444         /*
 445          * Warn if the last un-escaped character is whitespace. Then
 446          * strip away the remaining spaces (tabs stay!).
 447          */
 448
 449         i = (int)strlen(buf);
 450         assert(i);
 451
 452         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 453                 if (i > 1 && '\\' != buf[i - 2])
 454                         man_pmsg(man, line, i - 1, MANDOCERR_EOLNSPACE);
 455
 456                 for (--i; i && ' ' == buf[i]; i--)
 457                         /* Spin back to non-space. */ ;
 458
 459                 /* Jump ahead of escaped whitespace. */
 460                 i += '\\' == buf[i] ? 2 : 1;
 461
 462                 buf[i] = '\0';
 463         }
 464
 465         if ( ! man_word_alloc(man, line, offs, buf + offs))
 466                 return(0);
 467
 468         /*
 469          * End-of-sentence check.  If the last character is an unescaped
 470          * EOS character, then flag the node as being the end of a
 471          * sentence.  The front-end will know how to interpret this.
 472          */
 473
 474         assert(i);
 475         if (mandoc_eos(buf, (size_t)i, 0))
 476                 man->last->flags |= MAN_EOS;
 477
 478         return(man_descope(man, line, offs));
 479 }
 480
 481 static int
 482 man_pmacro(struct man *man, int ln, char *buf, int offs)
 483 {
 484         int              i, ppos;
 485         enum mant        tok;
 486         char             mac[5];
 487         struct man_node *n;
 488
 489         if ('"' == buf[offs]) {
 490                 man_pmsg(man, ln, offs, MANDOCERR_BADCOMMENT);
 491                 return(1);
 492         } else if ('\0' == buf[offs])
 493                 return(1);
 494
 495         ppos = offs;
 496
 497         /*
 498          * Copy the first word into a nil-terminated buffer.
 499          * Stop copying when a tab, space, or eoln is encountered.
 500          */
 501
 502         i = 0;
 503         while (i < 4 && '\0' != buf[offs] &&
 504                         ' ' != buf[offs] && '\t' != buf[offs])
 505                 mac[i++] = buf[offs++];
 506
 507         mac[i] = '\0';
 508
 509         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 510
 511         if (MAN_MAX == tok) {
 512                 mandoc_vmsg(MANDOCERR_MACRO, man->parse, ln,
 513                                 ppos, "%s", buf + ppos - 1);
 514                 return(1);
 515         }
 516
 517         /* The macro is sane.  Jump to the next word. */
 518
 519         while (buf[offs] && ' ' == buf[offs])
 520                 offs++;
 521
 522         /*
 523          * Trailing whitespace.  Note that tabs are allowed to be passed
 524          * into the parser as "text", so we only warn about spaces here.
 525          */
 526
 527         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 528                 man_pmsg(man, ln, offs - 1, MANDOCERR_EOLNSPACE);
 529
 530         /*
 531          * Remove prior ELINE macro, as it's being clobbered by a new
 532          * macro.  Note that NSCOPED macros do not close out ELINE
 533          * macros---they don't print text---so we let those slip by.
 534          */
 535
 536         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 537                         man->flags & MAN_ELINE) {
 538                 n = man->last;
 539                 assert(MAN_TEXT != n->type);
 540
 541                 /* Remove repeated NSCOPED macros causing ELINE. */
 542
 543                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 544                         n = n->parent;
 545
 546                 mandoc_vmsg(MANDOCERR_LINESCOPE, man->parse, n->line,
 547                     n->pos, "%s breaks %s", man_macronames[tok],
 548                     man_macronames[n->tok]);
 549
 550                 man_node_delete(man, n);
 551                 man->flags &= ~MAN_ELINE;
 552         }
 553
 554         /*
 555          * Remove prior BLINE macro that is being clobbered.
 556          */
 557         if ((man->flags & MAN_BLINE) &&
 558             (MAN_BSCOPE & man_macros[tok].flags)) {
 559                 n = man->last;
 560
 561                 /* Might be a text node like 8 in
 562                  * .TP 8
 563                  * .SH foo
 564                  */
 565                 if (MAN_TEXT == n->type)
 566                         n = n->parent;
 567
 568                 /* Remove element that didn't end BLINE, if any. */
 569                 if ( ! (MAN_BSCOPE & man_macros[n->tok].flags))
 570                         n = n->parent;
 571
 572                 assert(MAN_HEAD == n->type);
 573                 n = n->parent;
 574                 assert(MAN_BLOCK == n->type);
 575                 assert(MAN_SCOPED & man_macros[n->tok].flags);
 576
 577                 mandoc_vmsg(MANDOCERR_LINESCOPE, man->parse, n->line,
 578                     n->pos, "%s breaks %s", man_macronames[tok],
 579                     man_macronames[n->tok]);
 580
 581                 man_node_delete(man, n);
 582                 man->flags &= ~MAN_BLINE;
 583         }
 584
 585         /*
 586          * Save the fact that we're in the next-line for a block.  In
 587          * this way, embedded roff instructions can "remember" state
 588          * when they exit.
 589          */
 590
 591         if (MAN_BLINE & man->flags)
 592                 man->flags |= MAN_BPLINE;
 593
 594         /* Call to handler... */
 595
 596         assert(man_macros[tok].fp);
 597         if ( ! (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf))
 598                 goto err;
 599
 600         /*
 601          * We weren't in a block-line scope when entering the
 602          * above-parsed macro, so return.
 603          */
 604
 605         if ( ! (MAN_BPLINE & man->flags)) {
 606                 man->flags &= ~MAN_ILINE;
 607                 return(1);
 608         }
 609         man->flags &= ~MAN_BPLINE;
 610
 611         /*
 612          * If we're in a block scope, then allow this macro to slip by
 613          * without closing scope around it.
 614          */
 615
 616         if (MAN_ILINE & man->flags) {
 617                 man->flags &= ~MAN_ILINE;
 618                 return(1);
 619         }
 620
 621         /*
 622          * If we've opened a new next-line element scope, then return
 623          * now, as the next line will close out the block scope.
 624          */
 625
 626         if (MAN_ELINE & man->flags)
 627                 return(1);
 628
 629         /* Close out the block scope opened in the prior line.  */
 630
 631         assert(MAN_BLINE & man->flags);
 632         man->flags &= ~MAN_BLINE;
 633
 634         if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 635                 return(0);
 636         return(man_body_alloc(man, ln, ppos, man->last->tok));
 637
 638 err:    /* Error out. */
 639
 640         man->flags |= MAN_HALT;
 641         return(0);
 642 }
 643
 644 /*
 645  * Unlink a node from its context.  If "man" is provided, the last parse
 646  * point will also be adjusted accordingly.
 647  */
 648 static void
 649 man_node_unlink(struct man *man, struct man_node *n)
 650 {
 651
 652         /* Adjust siblings. */
 653
 654         if (n->prev)
 655                 n->prev->next = n->next;
 656         if (n->next)
 657                 n->next->prev = n->prev;
 658
 659         /* Adjust parent. */
 660
 661         if (n->parent) {
 662                 n->parent->nchild--;
 663                 if (n->parent->child == n)
 664                         n->parent->child = n->prev ? n->prev : n->next;
 665         }
 666
 667         /* Adjust parse point, if applicable. */
 668
 669         if (man && man->last == n) {
 670                 /*XXX: this can occur when bailing from validation. */
 671                 /*assert(NULL == n->next);*/
 672                 if (n->prev) {
 673                         man->last = n->prev;
 674                         man->next = MAN_NEXT_SIBLING;
 675                 } else {
 676                         man->last = n->parent;
 677                         man->next = MAN_NEXT_CHILD;
 678                 }
 679         }
 680
 681         if (man && man->first == n)
 682                 man->first = NULL;
 683 }
 684
 685 const struct mparse *
 686 man_mparse(const struct man *man)
 687 {
 688
 689         assert(man && man->parse);
 690         return(man->parse);
 691 }