man.c

   1 /*      $Id: man.c,v 1.88 2010/11/30 15:36:28 kristaps 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 <stdarg.h>
  25 #include <stdlib.h>
  26 #include <stdio.h>
  27 #include <string.h>
  28
  29 #include "mandoc.h"
  30 #include "libman.h"
  31 #include "libmandoc.h"
  32
  33 const   char *const __man_macronames[MAN_MAX] = {
  34         "br",           "TH",           "SH",           "SS",
  35         "TP",           "LP",           "PP",           "P",
  36         "IP",           "HP",           "SM",           "SB",
  37         "BI",           "IB",           "BR",           "RB",
  38         "R",            "B",            "I",            "IR",
  39         "RI",           "na",           "i",            "sp",
  40         "nf",           "fi",           "r",            "RE",
  41         "RS",           "DT",           "UC",           "PD",
  42         "Sp",           "Vb",           "Ve",           "AT",
  43         "in"
  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
  60
  61 const struct man_node *
  62 man_node(const struct man *m)
  63 {
  64
  65         return(MAN_HALT & m->flags ? NULL : m->first);
  66 }
  67
  68
  69 const struct man_meta *
  70 man_meta(const struct man *m)
  71 {
  72
  73         return(MAN_HALT & m->flags ? NULL : &m->meta);
  74 }
  75
  76
  77 void
  78 man_reset(struct man *man)
  79 {
  80
  81         man_free1(man);
  82         man_alloc1(man);
  83 }
  84
  85
  86 void
  87 man_free(struct man *man)
  88 {
  89
  90         man_free1(man);
  91         free(man);
  92 }
  93
  94
  95 struct man *
  96 man_alloc(struct regset *regs, void *data, mandocmsg msg)
  97 {
  98         struct man      *p;
  99
 100         p = mandoc_calloc(1, sizeof(struct man));
 101
 102         man_hash_init();
 103         p->data = data;
 104         p->msg = msg;
 105         p->regs = regs;
 106
 107         man_alloc1(p);
 108         return(p);
 109 }
 110
 111
 112 int
 113 man_endparse(struct man *m)
 114 {
 115
 116         if (MAN_HALT & m->flags)
 117                 return(0);
 118         else if (man_macroend(m))
 119                 return(1);
 120         m->flags |= MAN_HALT;
 121         return(0);
 122 }
 123
 124
 125 int
 126 man_parseln(struct man *m, int ln, char *buf, int offs)
 127 {
 128
 129         if (MAN_HALT & m->flags)
 130                 return(0);
 131
 132         return(('.' == buf[offs] || '\'' == buf[offs]) ?
 133                         man_pmacro(m, ln, buf, offs) :
 134                         man_ptext(m, ln, buf, offs));
 135 }
 136
 137
 138 static void
 139 man_free1(struct man *man)
 140 {
 141
 142         if (man->first)
 143                 man_node_delete(man, man->first);
 144         if (man->meta.title)
 145                 free(man->meta.title);
 146         if (man->meta.source)
 147                 free(man->meta.source);
 148         if (man->meta.rawdate)
 149                 free(man->meta.rawdate);
 150         if (man->meta.vol)
 151                 free(man->meta.vol);
 152         if (man->meta.msec)
 153                 free(man->meta.msec);
 154 }
 155
 156
 157 static void
 158 man_alloc1(struct man *m)
 159 {
 160
 161         memset(&m->meta, 0, sizeof(struct man_meta));
 162         m->flags = 0;
 163         m->last = mandoc_calloc(1, sizeof(struct man_node));
 164         m->first = m->last;
 165         m->last->type = MAN_ROOT;
 166         m->last->tok = MAN_MAX;
 167         m->next = MAN_NEXT_CHILD;
 168 }
 169
 170
 171 static int
 172 man_node_append(struct man *man, struct man_node *p)
 173 {
 174
 175         assert(man->last);
 176         assert(man->first);
 177         assert(MAN_ROOT != p->type);
 178
 179         switch (man->next) {
 180         case (MAN_NEXT_SIBLING):
 181                 man->last->next = p;
 182                 p->prev = man->last;
 183                 p->parent = man->last->parent;
 184                 break;
 185         case (MAN_NEXT_CHILD):
 186                 man->last->child = p;
 187                 p->parent = man->last;
 188                 break;
 189         default:
 190                 abort();
 191                 /* NOTREACHED */
 192         }
 193
 194         assert(p->parent);
 195         p->parent->nchild++;
 196
 197         if ( ! man_valid_pre(man, p))
 198                 return(0);
 199
 200         switch (p->type) {
 201         case (MAN_HEAD):
 202                 assert(MAN_BLOCK == p->parent->type);
 203                 p->parent->head = p;
 204                 break;
 205         case (MAN_BODY):
 206                 assert(MAN_BLOCK == p->parent->type);
 207                 p->parent->body = p;
 208                 break;
 209         default:
 210                 break;
 211         }
 212
 213         man->last = p;
 214
 215         switch (p->type) {
 216         case (MAN_TEXT):
 217                 if ( ! man_valid_post(man))
 218                         return(0);
 219                 break;
 220         default:
 221                 break;
 222         }
 223
 224         return(1);
 225 }
 226
 227
 228 static struct man_node *
 229 man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 230 {
 231         struct man_node *p;
 232
 233         p = mandoc_calloc(1, sizeof(struct man_node));
 234         p->line = line;
 235         p->pos = pos;
 236         p->type = type;
 237         p->tok = tok;
 238         return(p);
 239 }
 240
 241
 242 int
 243 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 244 {
 245         struct man_node *p;
 246
 247         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 248         if ( ! man_node_append(m, p))
 249                 return(0);
 250         m->next = MAN_NEXT_CHILD;
 251         return(1);
 252 }
 253
 254
 255 int
 256 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 257 {
 258         struct man_node *p;
 259
 260         p = man_node_alloc(line, pos, MAN_HEAD, 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_body_alloc(struct man *m, int line, int pos, enum mant tok)
 270 {
 271         struct man_node *p;
 272
 273         p = man_node_alloc(line, pos, MAN_BODY, 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_block_alloc(struct man *m, int line, int pos, enum mant tok)
 283 {
 284         struct man_node *p;
 285
 286         p = man_node_alloc(line, pos, MAN_BLOCK, 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_word_alloc(struct man *m, int line, int pos, const char *word)
 296 {
 297         struct man_node *n;
 298         size_t           sv, len;
 299
 300         len = strlen(word);
 301
 302         n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
 303         n->string = mandoc_malloc(len + 1);
 304         sv = strlcpy(n->string, word, len + 1);
 305
 306         /* Prohibit truncation. */
 307         assert(sv < len + 1);
 308
 309         if ( ! man_node_append(m, n))
 310                 return(0);
 311
 312         m->next = MAN_NEXT_SIBLING;
 313         return(1);
 314 }
 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
 331 void
 332 man_node_delete(struct man *m, struct man_node *p)
 333 {
 334
 335         while (p->child)
 336                 man_node_delete(m, p->child);
 337
 338         man_node_unlink(m, p);
 339         man_node_free(p);
 340 }
 341
 342
 343 static int
 344 man_ptext(struct man *m, int line, char *buf, int offs)
 345 {
 346         int              i;
 347
 348         /* Ignore bogus comments. */
 349
 350         if ('\\' == buf[offs] &&
 351                         '.' == buf[offs + 1] &&
 352                         '"' == buf[offs + 2])
 353                 return(man_pmsg(m, line, offs, MANDOCERR_BADCOMMENT));
 354
 355         /* Literal free-form text whitespace is preserved. */
 356
 357         if (MAN_LITERAL & m->flags) {
 358                 if ( ! man_word_alloc(m, line, offs, buf + offs))
 359                         return(0);
 360                 goto descope;
 361         }
 362
 363         /* Pump blank lines directly into the backend. */
 364
 365         for (i = offs; ' ' == buf[i]; i++)
 366                 /* Skip leading whitespace. */ ;
 367
 368         if ('\0' == buf[i]) {
 369                 /* Allocate a blank entry. */
 370                 if ( ! man_word_alloc(m, line, offs, ""))
 371                         return(0);
 372                 goto descope;
 373         }
 374
 375         /*
 376          * Warn if the last un-escaped character is whitespace. Then
 377          * strip away the remaining spaces (tabs stay!).
 378          */
 379
 380         i = (int)strlen(buf);
 381         assert(i);
 382
 383         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 384                 if (i > 1 && '\\' != buf[i - 2])
 385                         if ( ! man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE))
 386                                 return(0);
 387
 388                 for (--i; i && ' ' == buf[i]; i--)
 389                         /* Spin back to non-space. */ ;
 390
 391                 /* Jump ahead of escaped whitespace. */
 392                 i += '\\' == buf[i] ? 2 : 1;
 393
 394                 buf[i] = '\0';
 395         }
 396
 397         if ( ! man_word_alloc(m, line, offs, buf + offs))
 398                 return(0);
 399
 400         /*
 401          * End-of-sentence check.  If the last character is an unescaped
 402          * EOS character, then flag the node as being the end of a
 403          * sentence.  The front-end will know how to interpret this.
 404          */
 405
 406         assert(i);
 407         if (mandoc_eos(buf, (size_t)i, 0))
 408                 m->last->flags |= MAN_EOS;
 409
 410 descope:
 411         /*
 412          * Co-ordinate what happens with having a next-line scope open:
 413          * first close out the element scope (if applicable), then close
 414          * out the block scope (also if applicable).
 415          */
 416
 417         if (MAN_ELINE & m->flags) {
 418                 m->flags &= ~MAN_ELINE;
 419                 if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 420                         return(0);
 421         }
 422
 423         if ( ! (MAN_BLINE & m->flags))
 424                 return(1);
 425         m->flags &= ~MAN_BLINE;
 426
 427         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 428                 return(0);
 429         return(man_body_alloc(m, line, offs, m->last->tok));
 430 }
 431
 432
 433 int
 434 man_pmacro(struct man *m, int ln, char *buf, int offs)
 435 {
 436         int              i, j, ppos;
 437         enum mant        tok;
 438         char             mac[5];
 439         struct man_node *n;
 440
 441         /* Comments and empties are quickly ignored. */
 442
 443         offs++;
 444
 445         if ('\0' == buf[offs])
 446                 return(1);
 447
 448         i = offs;
 449
 450         /*
 451          * Skip whitespace between the control character and initial
 452          * text.  "Whitespace" is both spaces and tabs.
 453          */
 454
 455         if (' ' == buf[i] || '\t' == buf[i]) {
 456                 i++;
 457                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 458                         i++;
 459                 if ('\0' == buf[i])
 460                         goto out;
 461         }
 462
 463         ppos = i;
 464
 465         /*
 466          * Copy the first word into a nil-terminated buffer.
 467          * Stop copying when a tab, space, or eoln is encountered.
 468          */
 469
 470         j = 0;
 471         while (j < 4 && '\0' != buf[i] && ' ' != buf[i] && '\t' != buf[i])
 472                 mac[j++] = buf[i++];
 473         mac[j] = '\0';
 474
 475         tok = (j > 0 && j < 4) ? man_hash_find(mac) : MAN_MAX;
 476         if (MAN_MAX == tok) {
 477                 man_vmsg(m, MANDOCERR_MACRO, ln, ppos,
 478                     "unknown macro: %s%s",
 479                     buf, strlen(buf) > 3 ? "..." : "");
 480                 return(1);
 481         }
 482
 483         /* The macro is sane.  Jump to the next word. */
 484
 485         while (buf[i] && ' ' == buf[i])
 486                 i++;
 487
 488         /*
 489          * Trailing whitespace.  Note that tabs are allowed to be passed
 490          * into the parser as "text", so we only warn about spaces here.
 491          */
 492
 493         if ('\0' == buf[i] && ' ' == buf[i - 1])
 494                 if ( ! man_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE))
 495                         goto err;
 496
 497         /*
 498          * Remove prior ELINE macro, as it's being clobbering by a new
 499          * macro.  Note that NSCOPED macros do not close out ELINE
 500          * macros---they don't print text---so we let those slip by.
 501          */
 502
 503         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 504                         m->flags & MAN_ELINE) {
 505                 assert(MAN_TEXT != m->last->type);
 506
 507                 /*
 508                  * This occurs in the following construction:
 509                  *   .B
 510                  *   .br
 511                  *   .B
 512                  *   .br
 513                  *   I hate man macros.
 514                  * Flat-out disallow this madness.
 515                  */
 516                 if (MAN_NSCOPED & man_macros[m->last->tok].flags) {
 517                         man_pmsg(m, ln, ppos, MANDOCERR_SYNTLINESCOPE);
 518                         return(0);
 519                 }
 520
 521                 n = m->last;
 522
 523                 assert(n);
 524                 assert(NULL == n->child);
 525                 assert(0 == n->nchild);
 526
 527                 if ( ! man_nmsg(m, n, MANDOCERR_LINESCOPE))
 528                         return(0);
 529
 530                 man_node_delete(m, n);
 531                 m->flags &= ~MAN_ELINE;
 532         }
 533
 534         /*
 535          * Save the fact that we're in the next-line for a block.  In
 536          * this way, embedded roff instructions can "remember" state
 537          * when they exit.
 538          */
 539
 540         if (MAN_BLINE & m->flags)
 541                 m->flags |= MAN_BPLINE;
 542
 543         /* Call to handler... */
 544
 545         assert(man_macros[tok].fp);
 546         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 547                 goto err;
 548
 549 out:
 550         /*
 551          * We weren't in a block-line scope when entering the
 552          * above-parsed macro, so return.
 553          */
 554
 555         if ( ! (MAN_BPLINE & m->flags)) {
 556                 m->flags &= ~MAN_ILINE;
 557                 return(1);
 558         }
 559         m->flags &= ~MAN_BPLINE;
 560
 561         /*
 562          * If we're in a block scope, then allow this macro to slip by
 563          * without closing scope around it.
 564          */
 565
 566         if (MAN_ILINE & m->flags) {
 567                 m->flags &= ~MAN_ILINE;
 568                 return(1);
 569         }
 570
 571         /*
 572          * If we've opened a new next-line element scope, then return
 573          * now, as the next line will close out the block scope.
 574          */
 575
 576         if (MAN_ELINE & m->flags)
 577                 return(1);
 578
 579         /* Close out the block scope opened in the prior line.  */
 580
 581         assert(MAN_BLINE & m->flags);
 582         m->flags &= ~MAN_BLINE;
 583
 584         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 585                 return(0);
 586         return(man_body_alloc(m, ln, offs, m->last->tok));
 587
 588 err:    /* Error out. */
 589
 590         m->flags |= MAN_HALT;
 591         return(0);
 592 }
 593
 594
 595 int
 596 man_vmsg(struct man *man, enum mandocerr t,
 597                 int ln, int pos, const char *fmt, ...)
 598 {
 599         char             buf[256];
 600         va_list          ap;
 601
 602         va_start(ap, fmt);
 603         vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 604         va_end(ap);
 605         return((*man->msg)(t, man->data, ln, pos, buf));
 606 }
 607
 608
 609 /*
 610  * Unlink a node from its context.  If "m" is provided, the last parse
 611  * point will also be adjusted accordingly.
 612  */
 613 static void
 614 man_node_unlink(struct man *m, 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 (m && m->last == n) {
 635                 /*XXX: this can occur when bailing from validation. */
 636                 /*assert(NULL == n->next);*/
 637                 if (n->prev) {
 638                         m->last = n->prev;
 639                         m->next = MAN_NEXT_SIBLING;
 640                 } else {
 641                         m->last = n->parent;
 642                         m->next = MAN_NEXT_CHILD;
 643                 }
 644         }
 645
 646         if (m && m->first == n)
 647                 m->first = NULL;
 648 }