man.c

   1 /*      $Id: man.c,v 1.74 2010/05/17 22:11:42 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
   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(void *data, int pflags, mandocmsg msg)
  98 {
  99         struct man      *p;
 100
 101         p = mandoc_calloc(1, sizeof(struct man));
 102
 103         man_hash_init();
 104         p->data = data;
 105         p->pflags = pflags;
 106         p->msg = msg;
 107
 108         man_alloc1(p);
 109         return(p);
 110 }
 111
 112
 113 int
 114 man_endparse(struct man *m)
 115 {
 116
 117         if (MAN_HALT & m->flags)
 118                 return(0);
 119         else 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         if (MAN_HALT & m->flags)
 131                 return(0);
 132
 133         return(('.' == buf[offs] || '\'' == buf[offs]) ?
 134                         man_pmacro(m, ln, buf, offs) :
 135                         man_ptext(m, ln, buf, offs));
 136 }
 137
 138
 139 static void
 140 man_free1(struct man *man)
 141 {
 142
 143         if (man->first)
 144                 man_node_delete(man, man->first);
 145         if (man->meta.title)
 146                 free(man->meta.title);
 147         if (man->meta.source)
 148                 free(man->meta.source);
 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                 if ( ! man_action_post(man))
 219                         return(0);
 220                 break;
 221         default:
 222                 break;
 223         }
 224
 225         return(1);
 226 }
 227
 228
 229 static struct man_node *
 230 man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 231 {
 232         struct man_node *p;
 233
 234         p = mandoc_calloc(1, sizeof(struct man_node));
 235         p->line = line;
 236         p->pos = pos;
 237         p->type = type;
 238         p->tok = tok;
 239         return(p);
 240 }
 241
 242
 243 int
 244 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 245 {
 246         struct man_node *p;
 247
 248         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 249         if ( ! man_node_append(m, p))
 250                 return(0);
 251         m->next = MAN_NEXT_CHILD;
 252         return(1);
 253 }
 254
 255
 256 int
 257 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 258 {
 259         struct man_node *p;
 260
 261         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 262         if ( ! man_node_append(m, p))
 263                 return(0);
 264         m->next = MAN_NEXT_CHILD;
 265         return(1);
 266 }
 267
 268
 269 int
 270 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 271 {
 272         struct man_node *p;
 273
 274         p = man_node_alloc(line, pos, MAN_BODY, tok);
 275         if ( ! man_node_append(m, p))
 276                 return(0);
 277         m->next = MAN_NEXT_CHILD;
 278         return(1);
 279 }
 280
 281
 282 int
 283 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 284 {
 285         struct man_node *p;
 286
 287         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 288         if ( ! man_node_append(m, p))
 289                 return(0);
 290         m->next = MAN_NEXT_CHILD;
 291         return(1);
 292 }
 293
 294
 295 int
 296 man_word_alloc(struct man *m, int line, int pos, const char *word)
 297 {
 298         struct man_node *n;
 299         size_t           sv, len;
 300
 301         len = strlen(word);
 302
 303         n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
 304         n->string = mandoc_malloc(len + 1);
 305         sv = strlcpy(n->string, word, len + 1);
 306
 307         /* Prohibit truncation. */
 308         assert(sv < len + 1);
 309
 310         if ( ! man_node_append(m, n))
 311                 return(0);
 312
 313         m->next = MAN_NEXT_SIBLING;
 314         return(1);
 315 }
 316
 317
 318 /*
 319  * Free all of the resources held by a node.  This does NOT unlink a
 320  * node from its context; for that, see man_node_unlink().
 321  */
 322 static void
 323 man_node_free(struct man_node *p)
 324 {
 325
 326         if (p->string)
 327                 free(p->string);
 328         free(p);
 329 }
 330
 331
 332 void
 333 man_node_delete(struct man *m, struct man_node *p)
 334 {
 335
 336         while (p->child)
 337                 man_node_delete(m, p->child);
 338
 339         man_node_unlink(m, p);
 340         man_node_free(p);
 341 }
 342
 343
 344 static int
 345 man_ptext(struct man *m, int line, char *buf, int offs)
 346 {
 347         int              i;
 348
 349         /* Ignore bogus comments. */
 350
 351         if ('\\' == buf[offs] &&
 352                         '.' == buf[offs + 1] &&
 353                         '"' == buf[offs + 2])
 354                 return(man_pmsg(m, line, offs, MANDOCERR_BADCOMMENT));
 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                         if ( ! man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE))
 387                                 return(0);
 388
 389                 for (--i; i && ' ' == buf[i]; i--)
 390                         /* Spin back to non-space. */ ;
 391
 392                 /* Jump ahead of escaped whitespace. */
 393                 i += '\\' == buf[i] ? 2 : 1;
 394
 395                 buf[i] = '\0';
 396         }
 397
 398         if ( ! man_word_alloc(m, line, offs, buf + offs))
 399                 return(0);
 400
 401         /*
 402          * End-of-sentence check.  If the last character is an unescaped
 403          * EOS character, then flag the node as being the end of a
 404          * sentence.  The front-end will know how to interpret this.
 405          */
 406
 407         assert(i);
 408         if (mandoc_eos(buf, (size_t)i))
 409                 m->last->flags |= MAN_EOS;
 410
 411 descope:
 412         /*
 413          * Co-ordinate what happens with having a next-line scope open:
 414          * first close out the element scope (if applicable), then close
 415          * out the block scope (also if applicable).
 416          */
 417
 418         if (MAN_ELINE & m->flags) {
 419                 m->flags &= ~MAN_ELINE;
 420                 if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 421                         return(0);
 422         }
 423
 424         if ( ! (MAN_BLINE & m->flags))
 425                 return(1);
 426         m->flags &= ~MAN_BLINE;
 427
 428         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 429                 return(0);
 430         return(man_body_alloc(m, line, offs, m->last->tok));
 431 }
 432
 433
 434 static int
 435 macrowarn(struct man *m, int ln, const char *buf, int offs)
 436 {
 437         int              rc;
 438
 439         rc = man_vmsg(m, MANDOCERR_MACRO, ln, offs,
 440                         "unknown macro: %s%s",
 441                         buf, strlen(buf) > 3 ? "..." : "");
 442
 443         return(MAN_IGN_MACRO & m->pflags ? rc : 0);
 444 }
 445
 446
 447 int
 448 man_pmacro(struct man *m, int ln, char *buf, int offs)
 449 {
 450         int              i, j, ppos;
 451         enum mant        tok;
 452         char             mac[5];
 453         struct man_node *n;
 454
 455         /* Comments and empties are quickly ignored. */
 456
 457         offs++;
 458
 459         if ('\0' == buf[offs])
 460                 return(1);
 461
 462         i = offs;
 463
 464         /*
 465          * Skip whitespace between the control character and initial
 466          * text.  "Whitespace" is both spaces and tabs.
 467          */
 468
 469         if (' ' == buf[i] || '\t' == buf[i]) {
 470                 i++;
 471                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 472                         i++;
 473                 if ('\0' == buf[i])
 474                         goto out;
 475         }
 476
 477         ppos = i;
 478
 479         /* Copy the first word into a nil-terminated buffer. */
 480
 481         for (j = 0; j < 4; j++, i++) {
 482                 if ('\0' == (mac[j] = buf[i]))
 483                         break;
 484                 else if (' ' == buf[i])
 485                         break;
 486
 487                 /* Check for invalid characters. */
 488
 489                 if (isgraph((u_char)buf[i]))
 490                         continue;
 491                 if ( ! man_pmsg(m, ln, i, MANDOCERR_BADCHAR))
 492                         return(0);
 493                 i--;
 494         }
 495
 496         mac[j] = '\0';
 497
 498         if (j == 4 || j < 1) {
 499                 if ( ! macrowarn(m, ln, mac, ppos))
 500                         goto err;
 501                 return(1);
 502         }
 503
 504         if (MAN_MAX == (tok = man_hash_find(mac))) {
 505                 if ( ! macrowarn(m, ln, mac, ppos))
 506                         goto err;
 507                 return(1);
 508         }
 509
 510         /* The macro is sane.  Jump to the next word. */
 511
 512         while (buf[i] && ' ' == buf[i])
 513                 i++;
 514
 515         /*
 516          * Trailing whitespace.  Note that tabs are allowed to be passed
 517          * into the parser as "text", so we only warn about spaces here.
 518          */
 519
 520         if ('\0' == buf[i] && ' ' == buf[i - 1])
 521                 if ( ! man_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE))
 522                         goto err;
 523
 524         /*
 525          * Remove prior ELINE macro, as it's being clobbering by a new
 526          * macro.  Note that NSCOPED macros do not close out ELINE
 527          * macros---they don't print text---so we let those slip by.
 528          */
 529
 530         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 531                         m->flags & MAN_ELINE) {
 532                 assert(MAN_TEXT != m->last->type);
 533
 534                 /*
 535                  * This occurs in the following construction:
 536                  *   .B
 537                  *   .br
 538                  *   .B
 539                  *   .br
 540                  *   I hate man macros.
 541                  * Flat-out disallow this madness.
 542                  */
 543                 if (MAN_NSCOPED & man_macros[m->last->tok].flags) {
 544                         man_pmsg(m, ln, ppos, MANDOCERR_SYNTLINESCOPE);
 545                         return(0);
 546                 }
 547
 548                 n = m->last;
 549
 550                 assert(n);
 551                 assert(NULL == n->child);
 552                 assert(0 == n->nchild);
 553
 554                 if ( ! man_nmsg(m, n, MANDOCERR_LINESCOPE))
 555                         return(0);
 556
 557                 man_node_delete(m, n);
 558                 m->flags &= ~MAN_ELINE;
 559         }
 560
 561         /*
 562          * Save the fact that we're in the next-line for a block.  In
 563          * this way, embedded roff instructions can "remember" state
 564          * when they exit.
 565          */
 566
 567         if (MAN_BLINE & m->flags)
 568                 m->flags |= MAN_BPLINE;
 569
 570         /* Call to handler... */
 571
 572         assert(man_macros[tok].fp);
 573         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 574                 goto err;
 575
 576 out:
 577         /*
 578          * We weren't in a block-line scope when entering the
 579          * above-parsed macro, so return.
 580          */
 581
 582         if ( ! (MAN_BPLINE & m->flags)) {
 583                 m->flags &= ~MAN_ILINE;
 584                 return(1);
 585         }
 586         m->flags &= ~MAN_BPLINE;
 587
 588         /*
 589          * If we're in a block scope, then allow this macro to slip by
 590          * without closing scope around it.
 591          */
 592
 593         if (MAN_ILINE & m->flags) {
 594                 m->flags &= ~MAN_ILINE;
 595                 return(1);
 596         }
 597
 598         /*
 599          * If we've opened a new next-line element scope, then return
 600          * now, as the next line will close out the block scope.
 601          */
 602
 603         if (MAN_ELINE & m->flags)
 604                 return(1);
 605
 606         /* Close out the block scope opened in the prior line.  */
 607
 608         assert(MAN_BLINE & m->flags);
 609         m->flags &= ~MAN_BLINE;
 610
 611         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 612                 return(0);
 613         return(man_body_alloc(m, ln, offs, m->last->tok));
 614
 615 err:    /* Error out. */
 616
 617         m->flags |= MAN_HALT;
 618         return(0);
 619 }
 620
 621
 622 int
 623 man_vmsg(struct man *man, enum mandocerr t,
 624                 int ln, int pos, const char *fmt, ...)
 625 {
 626         char             buf[256];
 627         va_list          ap;
 628
 629         va_start(ap, fmt);
 630         vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 631         va_end(ap);
 632         return((*man->msg)(t, man->data, ln, pos, buf));
 633 }
 634
 635
 636 /*
 637  * Unlink a node from its context.  If "m" is provided, the last parse
 638  * point will also be adjusted accordingly.
 639  */
 640 static void
 641 man_node_unlink(struct man *m, struct man_node *n)
 642 {
 643
 644         /* Adjust siblings. */
 645
 646         if (n->prev)
 647                 n->prev->next = n->next;
 648         if (n->next)
 649                 n->next->prev = n->prev;
 650
 651         /* Adjust parent. */
 652
 653         if (n->parent) {
 654                 n->parent->nchild--;
 655                 if (n->parent->child == n)
 656                         n->parent->child = n->prev ? n->prev : n->next;
 657         }
 658
 659         /* Adjust parse point, if applicable. */
 660
 661         if (m && m->last == n) {
 662                 /*XXX: this can occur when bailing from validation. */
 663                 /*assert(NULL == n->next);*/
 664                 if (n->prev) {
 665                         m->last = n->prev;
 666                         m->next = MAN_NEXT_SIBLING;
 667                 } else {
 668                         m->last = n->parent;
 669                         m->next = MAN_NEXT_CHILD;
 670                 }
 671         }
 672
 673         if (m && m->first == n)
 674                 m->first = NULL;
 675 }