man.c

   1 /*      $Id: man.c,v 1.93 2011/01/01 10:51:30 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",           "sp",           "nf",
  40         "fi",           "RE",           "RS",           "DT",
  41         "UC",           "PD",           "AT",           "in",
  42         "ft"
  43         };
  44
  45 const   char * const *man_macronames = __man_macronames;
  46
  47 static  struct man_node *man_node_alloc(int, int,
  48                                 enum man_type, enum mant);
  49 static  int              man_node_append(struct man *,
  50                                 struct man_node *);
  51 static  void             man_node_free(struct man_node *);
  52 static  void             man_node_unlink(struct man *,
  53                                 struct man_node *);
  54 static  int              man_ptext(struct man *, int, char *, int);
  55 static  int              man_pmacro(struct man *, int, char *, int);
  56 static  void             man_free1(struct man *);
  57 static  void             man_alloc1(struct man *);
  58
  59
  60 const struct man_node *
  61 man_node(const struct man *m)
  62 {
  63
  64         return(MAN_HALT & m->flags ? NULL : m->first);
  65 }
  66
  67
  68 const struct man_meta *
  69 man_meta(const struct man *m)
  70 {
  71
  72         return(MAN_HALT & m->flags ? NULL : &m->meta);
  73 }
  74
  75
  76 void
  77 man_reset(struct man *man)
  78 {
  79
  80         man_free1(man);
  81         man_alloc1(man);
  82 }
  83
  84
  85 void
  86 man_free(struct man *man)
  87 {
  88
  89         man_free1(man);
  90         free(man);
  91 }
  92
  93
  94 struct man *
  95 man_alloc(struct regset *regs, void *data, mandocmsg msg)
  96 {
  97         struct man      *p;
  98
  99         p = mandoc_calloc(1, sizeof(struct man));
 100
 101         man_hash_init();
 102         p->data = data;
 103         p->msg = msg;
 104         p->regs = regs;
 105
 106         man_alloc1(p);
 107         return(p);
 108 }
 109
 110
 111 int
 112 man_endparse(struct man *m)
 113 {
 114
 115         if (MAN_HALT & m->flags)
 116                 return(0);
 117         else if (man_macroend(m))
 118                 return(1);
 119         m->flags |= MAN_HALT;
 120         return(0);
 121 }
 122
 123
 124 int
 125 man_parseln(struct man *m, int ln, char *buf, int offs)
 126 {
 127
 128         if (MAN_HALT & m->flags)
 129                 return(0);
 130
 131         return(('.' == buf[offs] || '\'' == buf[offs]) ?
 132                         man_pmacro(m, ln, buf, offs) :
 133                         man_ptext(m, ln, buf, offs));
 134 }
 135
 136
 137 static void
 138 man_free1(struct man *man)
 139 {
 140
 141         if (man->first)
 142                 man_node_delete(man, man->first);
 143         if (man->meta.title)
 144                 free(man->meta.title);
 145         if (man->meta.source)
 146                 free(man->meta.source);
 147         if (man->meta.rawdate)
 148                 free(man->meta.rawdate);
 149         if (man->meta.vol)
 150                 free(man->meta.vol);
 151         if (man->meta.msec)
 152                 free(man->meta.msec);
 153 }
 154
 155
 156 static void
 157 man_alloc1(struct man *m)
 158 {
 159
 160         memset(&m->meta, 0, sizeof(struct man_meta));
 161         m->flags = 0;
 162         m->last = mandoc_calloc(1, sizeof(struct man_node));
 163         m->first = m->last;
 164         m->last->type = MAN_ROOT;
 165         m->last->tok = MAN_MAX;
 166         m->next = MAN_NEXT_CHILD;
 167 }
 168
 169
 170 static int
 171 man_node_append(struct man *man, struct man_node *p)
 172 {
 173
 174         assert(man->last);
 175         assert(man->first);
 176         assert(MAN_ROOT != p->type);
 177
 178         switch (man->next) {
 179         case (MAN_NEXT_SIBLING):
 180                 man->last->next = p;
 181                 p->prev = man->last;
 182                 p->parent = man->last->parent;
 183                 break;
 184         case (MAN_NEXT_CHILD):
 185                 man->last->child = p;
 186                 p->parent = man->last;
 187                 break;
 188         default:
 189                 abort();
 190                 /* NOTREACHED */
 191         }
 192
 193         assert(p->parent);
 194         p->parent->nchild++;
 195
 196         if ( ! man_valid_pre(man, p))
 197                 return(0);
 198
 199         switch (p->type) {
 200         case (MAN_HEAD):
 201                 assert(MAN_BLOCK == p->parent->type);
 202                 p->parent->head = p;
 203                 break;
 204         case (MAN_BODY):
 205                 assert(MAN_BLOCK == p->parent->type);
 206                 p->parent->body = p;
 207                 break;
 208         default:
 209                 break;
 210         }
 211
 212         man->last = p;
 213
 214         switch (p->type) {
 215         case (MAN_TEXT):
 216                 if ( ! man_valid_post(man))
 217                         return(0);
 218                 break;
 219         default:
 220                 break;
 221         }
 222
 223         return(1);
 224 }
 225
 226
 227 static struct man_node *
 228 man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 229 {
 230         struct man_node *p;
 231
 232         p = mandoc_calloc(1, sizeof(struct man_node));
 233         p->line = line;
 234         p->pos = pos;
 235         p->type = type;
 236         p->tok = tok;
 237         return(p);
 238 }
 239
 240
 241 int
 242 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 243 {
 244         struct man_node *p;
 245
 246         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 247         if ( ! man_node_append(m, p))
 248                 return(0);
 249         m->next = MAN_NEXT_CHILD;
 250         return(1);
 251 }
 252
 253
 254 int
 255 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 256 {
 257         struct man_node *p;
 258
 259         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 260         if ( ! man_node_append(m, p))
 261                 return(0);
 262         m->next = MAN_NEXT_CHILD;
 263         return(1);
 264 }
 265
 266
 267 int
 268 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 269 {
 270         struct man_node *p;
 271
 272         p = man_node_alloc(line, pos, MAN_BODY, tok);
 273         if ( ! man_node_append(m, p))
 274                 return(0);
 275         m->next = MAN_NEXT_CHILD;
 276         return(1);
 277 }
 278
 279
 280 int
 281 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 282 {
 283         struct man_node *p;
 284
 285         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 286         if ( ! man_node_append(m, p))
 287                 return(0);
 288         m->next = MAN_NEXT_CHILD;
 289         return(1);
 290 }
 291
 292
 293 int
 294 man_word_alloc(struct man *m, int line, int pos, const char *word)
 295 {
 296         struct man_node *n;
 297         size_t           sv, len;
 298
 299         len = strlen(word);
 300
 301         n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
 302         n->string = mandoc_malloc(len + 1);
 303         sv = strlcpy(n->string, word, len + 1);
 304
 305         /* Prohibit truncation. */
 306         assert(sv < len + 1);
 307
 308         if ( ! man_node_append(m, n))
 309                 return(0);
 310
 311         m->next = MAN_NEXT_SIBLING;
 312         return(1);
 313 }
 314
 315
 316 /*
 317  * Free all of the resources held by a node.  This does NOT unlink a
 318  * node from its context; for that, see man_node_unlink().
 319  */
 320 static void
 321 man_node_free(struct man_node *p)
 322 {
 323
 324         if (p->string)
 325                 free(p->string);
 326         free(p);
 327 }
 328
 329
 330 void
 331 man_node_delete(struct man *m, struct man_node *p)
 332 {
 333
 334         while (p->child)
 335                 man_node_delete(m, p->child);
 336
 337         man_node_unlink(m, p);
 338         man_node_free(p);
 339 }
 340
 341
 342 static int
 343 man_ptext(struct man *m, int line, char *buf, int offs)
 344 {
 345         int              i;
 346
 347         /* Ignore bogus comments. */
 348
 349         if ('\\' == buf[offs] &&
 350                         '.' == buf[offs + 1] &&
 351                         '"' == buf[offs + 2]) {
 352                 man_pmsg(m, line, offs, MANDOCERR_BADCOMMENT);
 353                 return(1);
 354         }
 355
 356         /* Literal free-form text whitespace is preserved. */
 357
 358         if (MAN_LITERAL & m->flags) {
 359                 if ( ! man_word_alloc(m, line, offs, buf + offs))
 360                         return(0);
 361                 goto descope;
 362         }
 363
 364         /* Pump blank lines directly into the backend. */
 365
 366         for (i = offs; ' ' == buf[i]; i++)
 367                 /* Skip leading whitespace. */ ;
 368
 369         if ('\0' == buf[i]) {
 370                 /* Allocate a blank entry. */
 371                 if ( ! man_word_alloc(m, line, offs, ""))
 372                         return(0);
 373                 goto descope;
 374         }
 375
 376         /*
 377          * Warn if the last un-escaped character is whitespace. Then
 378          * strip away the remaining spaces (tabs stay!).
 379          */
 380
 381         i = (int)strlen(buf);
 382         assert(i);
 383
 384         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 385                 if (i > 1 && '\\' != buf[i - 2])
 386                         man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE);
 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, "%s", buf + ppos - 1);
 478                 return(1);
 479         }
 480
 481         /* The macro is sane.  Jump to the next word. */
 482
 483         while (buf[i] && ' ' == buf[i])
 484                 i++;
 485
 486         /*
 487          * Trailing whitespace.  Note that tabs are allowed to be passed
 488          * into the parser as "text", so we only warn about spaces here.
 489          */
 490
 491         if ('\0' == buf[i] && ' ' == buf[i - 1])
 492                 man_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE);
 493
 494         /*
 495          * Remove prior ELINE macro, as it's being clobbered by a new
 496          * macro.  Note that NSCOPED macros do not close out ELINE
 497          * macros---they don't print text---so we let those slip by.
 498          */
 499
 500         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 501                         m->flags & MAN_ELINE) {
 502                 n = m->last;
 503                 assert(MAN_TEXT != n->type);
 504
 505                 /* Remove repeated NSCOPED macros causing ELINE. */
 506
 507                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 508                         n = n->parent;
 509
 510                 man_vmsg(m, MANDOCERR_LINESCOPE, n->line, n->pos,
 511                                 "%s", man_macronames[n->tok]);
 512
 513                 man_node_delete(m, n);
 514                 m->flags &= ~MAN_ELINE;
 515         }
 516
 517         /*
 518          * Save the fact that we're in the next-line for a block.  In
 519          * this way, embedded roff instructions can "remember" state
 520          * when they exit.
 521          */
 522
 523         if (MAN_BLINE & m->flags)
 524                 m->flags |= MAN_BPLINE;
 525
 526         /* Call to handler... */
 527
 528         assert(man_macros[tok].fp);
 529         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 530                 goto err;
 531
 532 out:
 533         /*
 534          * We weren't in a block-line scope when entering the
 535          * above-parsed macro, so return.
 536          */
 537
 538         if ( ! (MAN_BPLINE & m->flags)) {
 539                 m->flags &= ~MAN_ILINE;
 540                 return(1);
 541         }
 542         m->flags &= ~MAN_BPLINE;
 543
 544         /*
 545          * If we're in a block scope, then allow this macro to slip by
 546          * without closing scope around it.
 547          */
 548
 549         if (MAN_ILINE & m->flags) {
 550                 m->flags &= ~MAN_ILINE;
 551                 return(1);
 552         }
 553
 554         /*
 555          * If we've opened a new next-line element scope, then return
 556          * now, as the next line will close out the block scope.
 557          */
 558
 559         if (MAN_ELINE & m->flags)
 560                 return(1);
 561
 562         /* Close out the block scope opened in the prior line.  */
 563
 564         assert(MAN_BLINE & m->flags);
 565         m->flags &= ~MAN_BLINE;
 566
 567         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 568                 return(0);
 569         return(man_body_alloc(m, ln, offs, m->last->tok));
 570
 571 err:    /* Error out. */
 572
 573         m->flags |= MAN_HALT;
 574         return(0);
 575 }
 576
 577
 578 int
 579 man_vmsg(struct man *man, enum mandocerr t,
 580                 int ln, int pos, const char *fmt, ...)
 581 {
 582         char             buf[256];
 583         va_list          ap;
 584
 585         va_start(ap, fmt);
 586         vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 587         va_end(ap);
 588         return((*man->msg)(t, man->data, ln, pos, buf));
 589 }
 590
 591
 592 /*
 593  * Unlink a node from its context.  If "m" is provided, the last parse
 594  * point will also be adjusted accordingly.
 595  */
 596 static void
 597 man_node_unlink(struct man *m, struct man_node *n)
 598 {
 599
 600         /* Adjust siblings. */
 601
 602         if (n->prev)
 603                 n->prev->next = n->next;
 604         if (n->next)
 605                 n->next->prev = n->prev;
 606
 607         /* Adjust parent. */
 608
 609         if (n->parent) {
 610                 n->parent->nchild--;
 611                 if (n->parent->child == n)
 612                         n->parent->child = n->prev ? n->prev : n->next;
 613         }
 614
 615         /* Adjust parse point, if applicable. */
 616
 617         if (m && m->last == n) {
 618                 /*XXX: this can occur when bailing from validation. */
 619                 /*assert(NULL == n->next);*/
 620                 if (n->prev) {
 621                         m->last = n->prev;
 622                         m->next = MAN_NEXT_SIBLING;
 623                 } else {
 624                         m->last = n->parent;
 625                         m->next = MAN_NEXT_CHILD;
 626                 }
 627         }
 628
 629         if (m && m->first == n)
 630                 m->first = NULL;
 631 }