man.c

   1 /*      $Id: man.c,v 1.83 2010/07/18 17:00:26 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010 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 <ctype.h>
  25 #include <stdarg.h>
  26 #include <stdlib.h>
  27 #include <stdio.h>
  28 #include <string.h>
  29
  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",           "i",            "sp",
  41         "nf",           "fi",           "r",            "RE",
  42         "RS",           "DT",           "UC",           "PD",
  43         "Sp",           "Vb",           "Ve",           "AT",
  44         };
  45
  46 const   char * const *man_macronames = __man_macronames;
  47
  48 static  struct man_node *man_node_alloc(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              macrowarn(struct man *, int, const char *, int);
  60
  61
  62 const struct man_node *
  63 man_node(const struct man *m)
  64 {
  65
  66         return(MAN_HALT & m->flags ? NULL : m->first);
  67 }
  68
  69
  70 const struct man_meta *
  71 man_meta(const struct man *m)
  72 {
  73
  74         return(MAN_HALT & m->flags ? NULL : &m->meta);
  75 }
  76
  77
  78 void
  79 man_reset(struct man *man)
  80 {
  81
  82         man_free1(man);
  83         man_alloc1(man);
  84 }
  85
  86
  87 void
  88 man_free(struct man *man)
  89 {
  90
  91         man_free1(man);
  92         free(man);
  93 }
  94
  95
  96 struct man *
  97 man_alloc(struct regset *regs, void *data,
  98                 int pflags, mandocmsg msg)
  99 {
 100         struct man      *p;
 101
 102         p = mandoc_calloc(1, sizeof(struct man));
 103
 104         man_hash_init();
 105         p->data = data;
 106         p->pflags = pflags;
 107         p->msg = msg;
 108         p->regs = regs;
 109
 110         man_alloc1(p);
 111         return(p);
 112 }
 113
 114
 115 int
 116 man_endparse(struct man *m)
 117 {
 118
 119         if (MAN_HALT & m->flags)
 120                 return(0);
 121         else if (man_macroend(m))
 122                 return(1);
 123         m->flags |= MAN_HALT;
 124         return(0);
 125 }
 126
 127
 128 int
 129 man_parseln(struct man *m, int ln, char *buf, int offs)
 130 {
 131
 132         if (MAN_HALT & m->flags)
 133                 return(0);
 134
 135         return(('.' == buf[offs] || '\'' == buf[offs]) ?
 136                         man_pmacro(m, ln, buf, offs) :
 137                         man_ptext(m, ln, buf, offs));
 138 }
 139
 140
 141 static void
 142 man_free1(struct man *man)
 143 {
 144
 145         if (man->first)
 146                 man_node_delete(man, man->first);
 147         if (man->meta.title)
 148                 free(man->meta.title);
 149         if (man->meta.source)
 150                 free(man->meta.source);
 151         if (man->meta.rawdate)
 152                 free(man->meta.rawdate);
 153         if (man->meta.vol)
 154                 free(man->meta.vol);
 155         if (man->meta.msec)
 156                 free(man->meta.msec);
 157 }
 158
 159
 160 static void
 161 man_alloc1(struct man *m)
 162 {
 163
 164         memset(&m->meta, 0, sizeof(struct man_meta));
 165         m->flags = 0;
 166         m->last = mandoc_calloc(1, sizeof(struct man_node));
 167         m->first = m->last;
 168         m->last->type = MAN_ROOT;
 169         m->last->tok = MAN_MAX;
 170         m->next = MAN_NEXT_CHILD;
 171 }
 172
 173
 174 static int
 175 man_node_append(struct man *man, struct man_node *p)
 176 {
 177
 178         assert(man->last);
 179         assert(man->first);
 180         assert(MAN_ROOT != p->type);
 181
 182         switch (man->next) {
 183         case (MAN_NEXT_SIBLING):
 184                 man->last->next = p;
 185                 p->prev = man->last;
 186                 p->parent = man->last->parent;
 187                 break;
 188         case (MAN_NEXT_CHILD):
 189                 man->last->child = p;
 190                 p->parent = man->last;
 191                 break;
 192         default:
 193                 abort();
 194                 /* NOTREACHED */
 195         }
 196
 197         assert(p->parent);
 198         p->parent->nchild++;
 199
 200         if ( ! man_valid_pre(man, p))
 201                 return(0);
 202
 203         switch (p->type) {
 204         case (MAN_HEAD):
 205                 assert(MAN_BLOCK == p->parent->type);
 206                 p->parent->head = p;
 207                 break;
 208         case (MAN_BODY):
 209                 assert(MAN_BLOCK == p->parent->type);
 210                 p->parent->body = p;
 211                 break;
 212         default:
 213                 break;
 214         }
 215
 216         man->last = p;
 217
 218         switch (p->type) {
 219         case (MAN_TEXT):
 220                 if ( ! man_valid_post(man))
 221                         return(0);
 222                 if ( ! man_action_post(man))
 223                         return(0);
 224                 break;
 225         default:
 226                 break;
 227         }
 228
 229         return(1);
 230 }
 231
 232
 233 static struct man_node *
 234 man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 235 {
 236         struct man_node *p;
 237
 238         p = mandoc_calloc(1, sizeof(struct man_node));
 239         p->line = line;
 240         p->pos = pos;
 241         p->type = type;
 242         p->tok = tok;
 243         return(p);
 244 }
 245
 246
 247 int
 248 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 249 {
 250         struct man_node *p;
 251
 252         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 253         if ( ! man_node_append(m, p))
 254                 return(0);
 255         m->next = MAN_NEXT_CHILD;
 256         return(1);
 257 }
 258
 259
 260 int
 261 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 262 {
 263         struct man_node *p;
 264
 265         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 266         if ( ! man_node_append(m, p))
 267                 return(0);
 268         m->next = MAN_NEXT_CHILD;
 269         return(1);
 270 }
 271
 272
 273 int
 274 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 275 {
 276         struct man_node *p;
 277
 278         p = man_node_alloc(line, pos, MAN_BODY, tok);
 279         if ( ! man_node_append(m, p))
 280                 return(0);
 281         m->next = MAN_NEXT_CHILD;
 282         return(1);
 283 }
 284
 285
 286 int
 287 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 288 {
 289         struct man_node *p;
 290
 291         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 292         if ( ! man_node_append(m, p))
 293                 return(0);
 294         m->next = MAN_NEXT_CHILD;
 295         return(1);
 296 }
 297
 298
 299 int
 300 man_word_alloc(struct man *m, int line, int pos, const char *word)
 301 {
 302         struct man_node *n;
 303         size_t           sv, len;
 304
 305         len = strlen(word);
 306
 307         n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
 308         n->string = mandoc_malloc(len + 1);
 309         sv = strlcpy(n->string, word, len + 1);
 310
 311         /* Prohibit truncation. */
 312         assert(sv < len + 1);
 313
 314         if ( ! man_node_append(m, n))
 315                 return(0);
 316
 317         m->next = MAN_NEXT_SIBLING;
 318         return(1);
 319 }
 320
 321
 322 /*
 323  * Free all of the resources held by a node.  This does NOT unlink a
 324  * node from its context; for that, see man_node_unlink().
 325  */
 326 static void
 327 man_node_free(struct man_node *p)
 328 {
 329
 330         if (p->string)
 331                 free(p->string);
 332         free(p);
 333 }
 334
 335
 336 void
 337 man_node_delete(struct man *m, struct man_node *p)
 338 {
 339
 340         while (p->child)
 341                 man_node_delete(m, p->child);
 342
 343         man_node_unlink(m, p);
 344         man_node_free(p);
 345 }
 346
 347
 348 static int
 349 man_ptext(struct man *m, int line, char *buf, int offs)
 350 {
 351         int              i;
 352
 353         /* Ignore bogus comments. */
 354
 355         if ('\\' == buf[offs] &&
 356                         '.' == buf[offs + 1] &&
 357                         '"' == buf[offs + 2])
 358                 return(man_pmsg(m, line, offs, MANDOCERR_BADCOMMENT));
 359
 360         /* Literal free-form text whitespace is preserved. */
 361
 362         if (MAN_LITERAL & m->flags) {
 363                 if ( ! man_word_alloc(m, line, offs, buf + offs))
 364                         return(0);
 365                 goto descope;
 366         }
 367
 368         /* Pump blank lines directly into the backend. */
 369
 370         for (i = offs; ' ' == buf[i]; i++)
 371                 /* Skip leading whitespace. */ ;
 372
 373         if ('\0' == buf[i]) {
 374                 /* Allocate a blank entry. */
 375                 if ( ! man_word_alloc(m, line, offs, ""))
 376                         return(0);
 377                 goto descope;
 378         }
 379
 380         /*
 381          * Warn if the last un-escaped character is whitespace. Then
 382          * strip away the remaining spaces (tabs stay!).
 383          */
 384
 385         i = (int)strlen(buf);
 386         assert(i);
 387
 388         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 389                 if (i > 1 && '\\' != buf[i - 2])
 390                         if ( ! man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE))
 391                                 return(0);
 392
 393                 for (--i; i && ' ' == buf[i]; i--)
 394                         /* Spin back to non-space. */ ;
 395
 396                 /* Jump ahead of escaped whitespace. */
 397                 i += '\\' == buf[i] ? 2 : 1;
 398
 399                 buf[i] = '\0';
 400         }
 401
 402         if ( ! man_word_alloc(m, line, offs, buf + offs))
 403                 return(0);
 404
 405         /*
 406          * End-of-sentence check.  If the last character is an unescaped
 407          * EOS character, then flag the node as being the end of a
 408          * sentence.  The front-end will know how to interpret this.
 409          */
 410
 411         assert(i);
 412         if (mandoc_eos(buf, (size_t)i, 0))
 413                 m->last->flags |= MAN_EOS;
 414
 415 descope:
 416         /*
 417          * Co-ordinate what happens with having a next-line scope open:
 418          * first close out the element scope (if applicable), then close
 419          * out the block scope (also if applicable).
 420          */
 421
 422         if (MAN_ELINE & m->flags) {
 423                 m->flags &= ~MAN_ELINE;
 424                 if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 425                         return(0);
 426         }
 427
 428         if ( ! (MAN_BLINE & m->flags))
 429                 return(1);
 430         m->flags &= ~MAN_BLINE;
 431
 432         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 433                 return(0);
 434         return(man_body_alloc(m, line, offs, m->last->tok));
 435 }
 436
 437
 438 static int
 439 macrowarn(struct man *m, int ln, const char *buf, int offs)
 440 {
 441         int              rc;
 442
 443         rc = man_vmsg(m, MANDOCERR_MACRO, ln, offs,
 444                         "unknown macro: %s%s",
 445                         buf, strlen(buf) > 3 ? "..." : "");
 446
 447         return(MAN_IGN_MACRO & m->pflags ? rc : 0);
 448 }
 449
 450
 451 int
 452 man_pmacro(struct man *m, int ln, char *buf, int offs)
 453 {
 454         int              i, j, ppos;
 455         enum mant        tok;
 456         char             mac[5];
 457         struct man_node *n;
 458
 459         /* Comments and empties are quickly ignored. */
 460
 461         offs++;
 462
 463         if ('\0' == buf[offs])
 464                 return(1);
 465
 466         i = offs;
 467
 468         /*
 469          * Skip whitespace between the control character and initial
 470          * text.  "Whitespace" is both spaces and tabs.
 471          */
 472
 473         if (' ' == buf[i] || '\t' == buf[i]) {
 474                 i++;
 475                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 476                         i++;
 477                 if ('\0' == buf[i])
 478                         goto out;
 479         }
 480
 481         ppos = i;
 482
 483         /* Copy the first word into a nil-terminated buffer. */
 484
 485         for (j = 0; j < 4; j++, i++) {
 486                 if ('\0' == (mac[j] = buf[i]))
 487                         break;
 488                 else if (' ' == buf[i])
 489                         break;
 490
 491                 /* Check for invalid characters. */
 492
 493                 if (isgraph((u_char)buf[i]))
 494                         continue;
 495                 if ( ! man_pmsg(m, ln, i, MANDOCERR_BADCHAR))
 496                         return(0);
 497                 i--;
 498         }
 499
 500         mac[j] = '\0';
 501
 502         if (j == 4 || j < 1) {
 503                 if ( ! macrowarn(m, ln, mac, ppos))
 504                         goto err;
 505                 return(1);
 506         }
 507
 508         if (MAN_MAX == (tok = man_hash_find(mac))) {
 509                 if ( ! macrowarn(m, ln, mac, ppos))
 510                         goto err;
 511                 return(1);
 512         }
 513
 514         /* The macro is sane.  Jump to the next word. */
 515
 516         while (buf[i] && ' ' == buf[i])
 517                 i++;
 518
 519         /*
 520          * Trailing whitespace.  Note that tabs are allowed to be passed
 521          * into the parser as "text", so we only warn about spaces here.
 522          */
 523
 524         if ('\0' == buf[i] && ' ' == buf[i - 1])
 525                 if ( ! man_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE))
 526                         goto err;
 527
 528         /*
 529          * Remove prior ELINE macro, as it's being clobbering by a new
 530          * macro.  Note that NSCOPED macros do not close out ELINE
 531          * macros---they don't print text---so we let those slip by.
 532          */
 533
 534         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 535                         m->flags & MAN_ELINE) {
 536                 assert(MAN_TEXT != m->last->type);
 537
 538                 /*
 539                  * This occurs in the following construction:
 540                  *   .B
 541                  *   .br
 542                  *   .B
 543                  *   .br
 544                  *   I hate man macros.
 545                  * Flat-out disallow this madness.
 546                  */
 547                 if (MAN_NSCOPED & man_macros[m->last->tok].flags) {
 548                         man_pmsg(m, ln, ppos, MANDOCERR_SYNTLINESCOPE);
 549                         return(0);
 550                 }
 551
 552                 n = m->last;
 553
 554                 assert(n);
 555                 assert(NULL == n->child);
 556                 assert(0 == n->nchild);
 557
 558                 if ( ! man_nmsg(m, n, MANDOCERR_LINESCOPE))
 559                         return(0);
 560
 561                 man_node_delete(m, n);
 562                 m->flags &= ~MAN_ELINE;
 563         }
 564
 565         /*
 566          * Save the fact that we're in the next-line for a block.  In
 567          * this way, embedded roff instructions can "remember" state
 568          * when they exit.
 569          */
 570
 571         if (MAN_BLINE & m->flags)
 572                 m->flags |= MAN_BPLINE;
 573
 574         /* Call to handler... */
 575
 576         assert(man_macros[tok].fp);
 577         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 578                 goto err;
 579
 580 out:
 581         /*
 582          * We weren't in a block-line scope when entering the
 583          * above-parsed macro, so return.
 584          */
 585
 586         if ( ! (MAN_BPLINE & m->flags)) {
 587                 m->flags &= ~MAN_ILINE;
 588                 return(1);
 589         }
 590         m->flags &= ~MAN_BPLINE;
 591
 592         /*
 593          * If we're in a block scope, then allow this macro to slip by
 594          * without closing scope around it.
 595          */
 596
 597         if (MAN_ILINE & m->flags) {
 598                 m->flags &= ~MAN_ILINE;
 599                 return(1);
 600         }
 601
 602         /*
 603          * If we've opened a new next-line element scope, then return
 604          * now, as the next line will close out the block scope.
 605          */
 606
 607         if (MAN_ELINE & m->flags)
 608                 return(1);
 609
 610         /* Close out the block scope opened in the prior line.  */
 611
 612         assert(MAN_BLINE & m->flags);
 613         m->flags &= ~MAN_BLINE;
 614
 615         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 616                 return(0);
 617         return(man_body_alloc(m, ln, offs, m->last->tok));
 618
 619 err:    /* Error out. */
 620
 621         m->flags |= MAN_HALT;
 622         return(0);
 623 }
 624
 625
 626 int
 627 man_vmsg(struct man *man, enum mandocerr t,
 628                 int ln, int pos, const char *fmt, ...)
 629 {
 630         char             buf[256];
 631         va_list          ap;
 632
 633         va_start(ap, fmt);
 634         vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 635         va_end(ap);
 636         return((*man->msg)(t, man->data, ln, pos, buf));
 637 }
 638
 639
 640 /*
 641  * Unlink a node from its context.  If "m" is provided, the last parse
 642  * point will also be adjusted accordingly.
 643  */
 644 static void
 645 man_node_unlink(struct man *m, struct man_node *n)
 646 {
 647
 648         /* Adjust siblings. */
 649
 650         if (n->prev)
 651                 n->prev->next = n->next;
 652         if (n->next)
 653                 n->next->prev = n->prev;
 654
 655         /* Adjust parent. */
 656
 657         if (n->parent) {
 658                 n->parent->nchild--;
 659                 if (n->parent->child == n)
 660                         n->parent->child = n->prev ? n->prev : n->next;
 661         }
 662
 663         /* Adjust parse point, if applicable. */
 664
 665         if (m && m->last == n) {
 666                 /*XXX: this can occur when bailing from validation. */
 667                 /*assert(NULL == n->next);*/
 668                 if (n->prev) {
 669                         m->last = n->prev;
 670                         m->next = MAN_NEXT_SIBLING;
 671                 } else {
 672                         m->last = n->parent;
 673                         m->next = MAN_NEXT_CHILD;
 674                 }
 675         }
 676
 677         if (m && m->first == n)
 678                 m->first = NULL;
 679 }