man.c

   1 /*      $Id: man.c,v 1.109 2011/07/21 10:24:35 kristaps 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"
  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 *m)
  64 {
  65
  66         assert( ! (MAN_HALT & m->flags));
  67         return(m->first);
  68 }
  69
  70
  71 const struct man_meta *
  72 man_meta(const struct man *m)
  73 {
  74
  75         assert( ! (MAN_HALT & m->flags));
  76         return(&m->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 *m)
 116 {
 117
 118         assert( ! (MAN_HALT & m->flags));
 119         if (man_macroend(m))
 120                 return(1);
 121         m->flags |= MAN_HALT;
 122         return(0);
 123 }
 124
 125
 126 int
 127 man_parseln(struct man *m, int ln, char *buf, int offs)
 128 {
 129
 130         m->flags |= MAN_NEWLINE;
 131
 132         assert( ! (MAN_HALT & m->flags));
 133
 134         return (mandoc_getcontrol(buf, &offs) ?
 135                         man_pmacro(m, ln, buf, offs) :
 136                         man_ptext(m, 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 *m)
 161 {
 162
 163         memset(&m->meta, 0, sizeof(struct man_meta));
 164         m->flags = 0;
 165         m->last = mandoc_calloc(1, sizeof(struct man_node));
 166         m->first = m->last;
 167         m->last->type = MAN_ROOT;
 168         m->last->tok = MAN_MAX;
 169         m->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 *m, 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 & m->flags)
 249                 p->flags |= MAN_LINE;
 250         m->flags &= ~MAN_NEWLINE;
 251         return(p);
 252 }
 253
 254
 255 int
 256 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 257 {
 258         struct man_node *p;
 259
 260         p = man_node_alloc(m, line, pos, MAN_ELEM, tok);
 261         if ( ! man_node_append(m, p))
 262                 return(0);
 263         m->next = MAN_NEXT_CHILD;
 264         return(1);
 265 }
 266
 267
 268 int
 269 man_tail_alloc(struct man *m, int line, int pos, enum mant tok)
 270 {
 271         struct man_node *p;
 272
 273         p = man_node_alloc(m, line, pos, MAN_TAIL, tok);
 274         if ( ! man_node_append(m, p))
 275                 return(0);
 276         m->next = MAN_NEXT_CHILD;
 277         return(1);
 278 }
 279
 280
 281 int
 282 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 283 {
 284         struct man_node *p;
 285
 286         p = man_node_alloc(m, line, pos, MAN_HEAD, tok);
 287         if ( ! man_node_append(m, p))
 288                 return(0);
 289         m->next = MAN_NEXT_CHILD;
 290         return(1);
 291 }
 292
 293
 294 int
 295 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 296 {
 297         struct man_node *p;
 298
 299         p = man_node_alloc(m, line, pos, MAN_BODY, tok);
 300         if ( ! man_node_append(m, p))
 301                 return(0);
 302         m->next = MAN_NEXT_CHILD;
 303         return(1);
 304 }
 305
 306
 307 int
 308 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 309 {
 310         struct man_node *p;
 311
 312         p = man_node_alloc(m, line, pos, MAN_BLOCK, tok);
 313         if ( ! man_node_append(m, p))
 314                 return(0);
 315         m->next = MAN_NEXT_CHILD;
 316         return(1);
 317 }
 318
 319 int
 320 man_word_alloc(struct man *m, int line, int pos, const char *word)
 321 {
 322         struct man_node *n;
 323         size_t           sv, len;
 324
 325         len = strlen(word);
 326
 327         n = man_node_alloc(m, line, pos, MAN_TEXT, MAN_MAX);
 328         n->string = mandoc_malloc(len + 1);
 329         sv = strlcpy(n->string, word, len + 1);
 330
 331         /* Prohibit truncation. */
 332         assert(sv < len + 1);
 333
 334         if ( ! man_node_append(m, n))
 335                 return(0);
 336
 337         m->next = MAN_NEXT_SIBLING;
 338         return(1);
 339 }
 340
 341
 342 /*
 343  * Free all of the resources held by a node.  This does NOT unlink a
 344  * node from its context; for that, see man_node_unlink().
 345  */
 346 static void
 347 man_node_free(struct man_node *p)
 348 {
 349
 350         if (p->string)
 351                 free(p->string);
 352         free(p);
 353 }
 354
 355
 356 void
 357 man_node_delete(struct man *m, struct man_node *p)
 358 {
 359
 360         while (p->child)
 361                 man_node_delete(m, p->child);
 362
 363         man_node_unlink(m, p);
 364         man_node_free(p);
 365 }
 366
 367 int
 368 man_addeqn(struct man *m, const struct eqn *ep)
 369 {
 370         struct man_node *n;
 371
 372         assert( ! (MAN_HALT & m->flags));
 373
 374         n = man_node_alloc(m, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 375         n->eqn = ep;
 376
 377         if ( ! man_node_append(m, n))
 378                 return(0);
 379
 380         m->next = MAN_NEXT_SIBLING;
 381         return(man_descope(m, ep->ln, ep->pos));
 382 }
 383
 384 int
 385 man_addspan(struct man *m, const struct tbl_span *sp)
 386 {
 387         struct man_node *n;
 388
 389         assert( ! (MAN_HALT & m->flags));
 390
 391         n = man_node_alloc(m, sp->line, 0, MAN_TBL, MAN_MAX);
 392         n->span = sp;
 393
 394         if ( ! man_node_append(m, n))
 395                 return(0);
 396
 397         m->next = MAN_NEXT_SIBLING;
 398         return(man_descope(m, sp->line, 0));
 399 }
 400
 401 static int
 402 man_descope(struct man *m, int line, int offs)
 403 {
 404         /*
 405          * Co-ordinate what happens with having a next-line scope open:
 406          * first close out the element scope (if applicable), then close
 407          * out the block scope (also if applicable).
 408          */
 409
 410         if (MAN_ELINE & m->flags) {
 411                 m->flags &= ~MAN_ELINE;
 412                 if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 413                         return(0);
 414         }
 415
 416         if ( ! (MAN_BLINE & m->flags))
 417                 return(1);
 418         m->flags &= ~MAN_BLINE;
 419
 420         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 421                 return(0);
 422         return(man_body_alloc(m, line, offs, m->last->tok));
 423 }
 424
 425 static int
 426 man_ptext(struct man *m, int line, char *buf, int offs)
 427 {
 428         int              i;
 429
 430         /* Literal free-form text whitespace is preserved. */
 431
 432         if (MAN_LITERAL & m->flags) {
 433                 if ( ! man_word_alloc(m, line, offs, buf + offs))
 434                         return(0);
 435                 return(man_descope(m, line, offs));
 436         }
 437
 438         /* Pump blank lines directly into the backend. */
 439
 440         for (i = offs; ' ' == buf[i]; i++)
 441                 /* Skip leading whitespace. */ ;
 442
 443         if ('\0' == buf[i]) {
 444                 /* Allocate a blank entry. */
 445                 if ( ! man_word_alloc(m, line, offs, ""))
 446                         return(0);
 447                 return(man_descope(m, line, offs));
 448         }
 449
 450         /*
 451          * Warn if the last un-escaped character is whitespace. Then
 452          * strip away the remaining spaces (tabs stay!).
 453          */
 454
 455         i = (int)strlen(buf);
 456         assert(i);
 457
 458         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 459                 if (i > 1 && '\\' != buf[i - 2])
 460                         man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE);
 461
 462                 for (--i; i && ' ' == buf[i]; i--)
 463                         /* Spin back to non-space. */ ;
 464
 465                 /* Jump ahead of escaped whitespace. */
 466                 i += '\\' == buf[i] ? 2 : 1;
 467
 468                 buf[i] = '\0';
 469         }
 470
 471         if ( ! man_word_alloc(m, line, offs, buf + offs))
 472                 return(0);
 473
 474         /*
 475          * End-of-sentence check.  If the last character is an unescaped
 476          * EOS character, then flag the node as being the end of a
 477          * sentence.  The front-end will know how to interpret this.
 478          */
 479
 480         assert(i);
 481         if (mandoc_eos(buf, (size_t)i, 0))
 482                 m->last->flags |= MAN_EOS;
 483
 484         return(man_descope(m, line, offs));
 485 }
 486
 487 static int
 488 man_pmacro(struct man *m, int ln, char *buf, int offs)
 489 {
 490         int              i, ppos;
 491         enum mant        tok;
 492         char             mac[5];
 493         struct man_node *n;
 494
 495         if ('"' == buf[offs]) {
 496                 man_pmsg(m, ln, offs, MANDOCERR_BADCOMMENT);
 497                 return(1);
 498         } else if ('\0' == buf[offs])
 499                 return(1);
 500
 501         ppos = offs;
 502
 503         /*
 504          * Copy the first word into a nil-terminated buffer.
 505          * Stop copying when a tab, space, or eoln is encountered.
 506          */
 507
 508         i = 0;
 509         while (i < 4 && '\0' != buf[offs] &&
 510                         ' ' != buf[offs] && '\t' != buf[offs])
 511                 mac[i++] = buf[offs++];
 512
 513         mac[i] = '\0';
 514
 515         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 516
 517         if (MAN_MAX == tok) {
 518                 mandoc_vmsg(MANDOCERR_MACRO, m->parse, ln,
 519                                 ppos, "%s", buf + ppos - 1);
 520                 return(1);
 521         }
 522
 523         /* The macro is sane.  Jump to the next word. */
 524
 525         while (buf[offs] && ' ' == buf[offs])
 526                 offs++;
 527
 528         /*
 529          * Trailing whitespace.  Note that tabs are allowed to be passed
 530          * into the parser as "text", so we only warn about spaces here.
 531          */
 532
 533         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 534                 man_pmsg(m, ln, offs - 1, MANDOCERR_EOLNSPACE);
 535
 536         /*
 537          * Remove prior ELINE macro, as it's being clobbered by a new
 538          * macro.  Note that NSCOPED macros do not close out ELINE
 539          * macros---they don't print text---so we let those slip by.
 540          */
 541
 542         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 543                         m->flags & MAN_ELINE) {
 544                 n = m->last;
 545                 assert(MAN_TEXT != n->type);
 546
 547                 /* Remove repeated NSCOPED macros causing ELINE. */
 548
 549                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 550                         n = n->parent;
 551
 552                 mandoc_vmsg(MANDOCERR_LINESCOPE, m->parse, n->line,
 553                                 n->pos, "%s", man_macronames[n->tok]);
 554
 555                 man_node_delete(m, n);
 556                 m->flags &= ~MAN_ELINE;
 557         }
 558
 559         /*
 560          * Save the fact that we're in the next-line for a block.  In
 561          * this way, embedded roff instructions can "remember" state
 562          * when they exit.
 563          */
 564
 565         if (MAN_BLINE & m->flags)
 566                 m->flags |= MAN_BPLINE;
 567
 568         /* Call to handler... */
 569
 570         assert(man_macros[tok].fp);
 571         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &offs, buf))
 572                 goto err;
 573
 574         /*
 575          * We weren't in a block-line scope when entering the
 576          * above-parsed macro, so return.
 577          */
 578
 579         if ( ! (MAN_BPLINE & m->flags)) {
 580                 m->flags &= ~MAN_ILINE;
 581                 return(1);
 582         }
 583         m->flags &= ~MAN_BPLINE;
 584
 585         /*
 586          * If we're in a block scope, then allow this macro to slip by
 587          * without closing scope around it.
 588          */
 589
 590         if (MAN_ILINE & m->flags) {
 591                 m->flags &= ~MAN_ILINE;
 592                 return(1);
 593         }
 594
 595         /*
 596          * If we've opened a new next-line element scope, then return
 597          * now, as the next line will close out the block scope.
 598          */
 599
 600         if (MAN_ELINE & m->flags)
 601                 return(1);
 602
 603         /* Close out the block scope opened in the prior line.  */
 604
 605         assert(MAN_BLINE & m->flags);
 606         m->flags &= ~MAN_BLINE;
 607
 608         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 609                 return(0);
 610         return(man_body_alloc(m, ln, ppos, m->last->tok));
 611
 612 err:    /* Error out. */
 613
 614         m->flags |= MAN_HALT;
 615         return(0);
 616 }
 617
 618 /*
 619  * Unlink a node from its context.  If "m" is provided, the last parse
 620  * point will also be adjusted accordingly.
 621  */
 622 static void
 623 man_node_unlink(struct man *m, struct man_node *n)
 624 {
 625
 626         /* Adjust siblings. */
 627
 628         if (n->prev)
 629                 n->prev->next = n->next;
 630         if (n->next)
 631                 n->next->prev = n->prev;
 632
 633         /* Adjust parent. */
 634
 635         if (n->parent) {
 636                 n->parent->nchild--;
 637                 if (n->parent->child == n)
 638                         n->parent->child = n->prev ? n->prev : n->next;
 639         }
 640
 641         /* Adjust parse point, if applicable. */
 642
 643         if (m && m->last == n) {
 644                 /*XXX: this can occur when bailing from validation. */
 645                 /*assert(NULL == n->next);*/
 646                 if (n->prev) {
 647                         m->last = n->prev;
 648                         m->next = MAN_NEXT_SIBLING;
 649                 } else {
 650                         m->last = n->parent;
 651                         m->next = MAN_NEXT_CHILD;
 652                 }
 653         }
 654
 655         if (m && m->first == n)
 656                 m->first = NULL;
 657 }